commit 6de505049cba68082f9a10dbb2699eecb3224673 Author: ModelHub XC Date: Fri Aug 28 19:30:43 2026 +0800 初始化项目，由ModelHub XC社区提供模型 Model: Neelectric/Llama-3.1-8B-Instruct_SFT_MoTv00.01 Source: Original Platform diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000..52373fe --- /dev/null +++ b/.gitattributes @@ -0,0 +1,36 @@ +*.7z filter=lfs diff=lfs merge=lfs -text +*.arrow filter=lfs diff=lfs merge=lfs -text +*.bin filter=lfs diff=lfs merge=lfs -text +*.bz2 filter=lfs diff=lfs merge=lfs -text +*.ckpt filter=lfs diff=lfs merge=lfs -text +*.ftz filter=lfs diff=lfs merge=lfs -text +*.gz filter=lfs diff=lfs merge=lfs -text +*.h5 filter=lfs diff=lfs merge=lfs -text +*.joblib filter=lfs diff=lfs merge=lfs -text +*.lfs.* filter=lfs diff=lfs merge=lfs -text +*.mlmodel filter=lfs diff=lfs merge=lfs -text +*.model filter=lfs diff=lfs merge=lfs -text +*.msgpack filter=lfs diff=lfs merge=lfs -text +*.npy filter=lfs diff=lfs merge=lfs -text +*.npz filter=lfs diff=lfs merge=lfs -text +*.onnx filter=lfs diff=lfs merge=lfs -text +*.ot filter=lfs diff=lfs merge=lfs -text +*.parquet filter=lfs diff=lfs merge=lfs -text +*.pb filter=lfs diff=lfs merge=lfs -text +*.pickle filter=lfs diff=lfs merge=lfs -text +*.pkl filter=lfs diff=lfs merge=lfs -text +*.pt filter=lfs diff=lfs merge=lfs -text +*.pth filter=lfs diff=lfs merge=lfs -text +*.rar filter=lfs diff=lfs merge=lfs -text +*.safetensors filter=lfs diff=lfs merge=lfs -text +saved_model/**/* filter=lfs diff=lfs merge=lfs -text +*.tar.* filter=lfs diff=lfs merge=lfs -text +*.tar filter=lfs diff=lfs merge=lfs -text +*.tflite filter=lfs diff=lfs merge=lfs -text +*.tgz filter=lfs diff=lfs merge=lfs -text +*.wasm filter=lfs diff=lfs merge=lfs -text +*.xz filter=lfs diff=lfs merge=lfs -text +*.zip filter=lfs diff=lfs merge=lfs -text +*.zst filter=lfs diff=lfs merge=lfs -text +*tfevents* filter=lfs diff=lfs merge=lfs -text +tokenizer.json filter=lfs diff=lfs merge=lfs -text diff --git a/README.md b/README.md new file mode 100644 index 0000000..830a809 --- /dev/null +++ b/README.md @@ -0,0 +1,59 @@ +--- +base_model: meta-llama/Llama-3.1-8B-Instruct +datasets: Neelectric/MoT_all_Llama3_4096toks +library_name: transformers +model_name: Llama-3.1-8B-Instruct_SFT_MoTv00.01 +tags: +- generated_from_trainer +- trl +- sft +- open-r1 +licence: license +--- + +# Model Card for Llama-3.1-8B-Instruct_SFT_MoTv00.01 + +This model is a fine-tuned version of [meta-llama/Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct) on the [Neelectric/MoT_all_Llama3_4096toks](https://huggingface.co/datasets/Neelectric/MoT_all_Llama3_4096toks) dataset. +It has been trained using [TRL](https://github.com/huggingface/trl). + +## Quick start + +```python +from transformers import pipeline + +question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?" +generator = pipeline("text-generation", model="Neelectric/Llama-3.1-8B-Instruct_SFT_MoTv00.01", device="cuda") +output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0] +print(output["generated_text"]) +``` + +## Training procedure + +[Visualize in Weights & Biases](https://wandb.ai/neelectric/open-r1_mot/runs/93mzvipq) + + +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..752f1f9 --- /dev/null +++ b/all_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 4.923132181932828e+19, + "train_loss": 0.6614166504737585, + "train_runtime": 43606.5901, + "train_samples": 194029, + "train_samples_per_second": 13.349, + "train_steps_per_second": 0.834 +} \ 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..40ba811 --- /dev/null +++ b/model-00001-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50fccf0b310b731b6d940ee29f48962504c9c2cc1a535d1f6c435915695691ac +size 4976698672 diff --git a/model-00002-of-00004.safetensors b/model-00002-of-00004.safetensors new file mode 100644 index 0000000..1304805 --- /dev/null +++ b/model-00002-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c768fcd0665d9203bf89345ea21b1af4bea59e98921798eb37a05029764f6f9f +size 4999802720 diff --git a/model-00003-of-00004.safetensors b/model-00003-of-00004.safetensors new file mode 100644 index 0000000..f83106c --- /dev/null +++ b/model-00003-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c99e23e9a5f9e11c513e81c41993de8b289592c46f191754ff2b4ba3c0441e7 +size 4915916176 diff --git a/model-00004-of-00004.safetensors b/model-00004-of-00004.safetensors new file mode 100644 index 0000000..f79832b --- /dev/null +++ b/model-00004-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3cb128c829a8743da6d36480c13ca591b5bc4ef0fbf7e4ec263bbf36222b5886 +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..752f1f9 --- /dev/null +++ b/train_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 4.923132181932828e+19, + "train_loss": 0.6614166504737585, + "train_runtime": 43606.5901, + "train_samples": 194029, + "train_samples_per_second": 13.349, + "train_steps_per_second": 0.834 +} \ No newline at end of file diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000..4ebf2a2 --- /dev/null +++ b/trainer_state.json @@ -0,0 +1,327472 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 3.0, + "eval_steps": 500, + "global_step": 36381, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 8.246062505153789e-05, + "grad_norm": 6.654606342315674, + "learning_rate": 0.0, + "loss": 1.3137, + "mean_token_accuracy": 0.6519534587860107, + "num_tokens": 26300.0, + "step": 1 + }, + { + "epoch": 0.00016492125010307578, + "grad_norm": 7.604219913482666, + "learning_rate": 2.748007694421544e-10, + "loss": 1.3109, + "mean_token_accuracy": 0.6504651308059692, + "num_tokens": 48008.0, + "step": 2 + }, + { + "epoch": 0.0002473818751546137, + "grad_norm": 6.584382057189941, + "learning_rate": 5.496015388843088e-10, + "loss": 1.3101, + "mean_token_accuracy": 0.635266900062561, + "num_tokens": 73753.0, + "step": 3 + }, + { + "epoch": 0.00032984250020615157, + "grad_norm": 6.414369106292725, + "learning_rate": 8.244023083264633e-10, + "loss": 1.1224, + "mean_token_accuracy": 0.682575523853302, + "num_tokens": 99173.0, + "step": 4 + }, + { + "epoch": 0.00041230312525768946, + "grad_norm": 6.313216686248779, + "learning_rate": 1.0992030777686176e-09, + "loss": 1.4129, + "mean_token_accuracy": 0.624323308467865, + "num_tokens": 130021.0, + "step": 5 + }, + { + "epoch": 0.0004947637503092274, + "grad_norm": 6.205374717712402, + "learning_rate": 1.3740038472107722e-09, + "loss": 1.2714, + "mean_token_accuracy": 0.6612653136253357, + "num_tokens": 159498.0, + "step": 6 + }, + { + "epoch": 0.0005772243753607652, + "grad_norm": 6.796016693115234, + "learning_rate": 1.6488046166529266e-09, + "loss": 1.2378, + "mean_token_accuracy": 0.6609314680099487, + "num_tokens": 184546.0, + "step": 7 + }, + { + "epoch": 0.0006596850004123031, + "grad_norm": 5.76704740524292, + "learning_rate": 1.923605386095081e-09, + "loss": 1.2415, + "mean_token_accuracy": 0.6621792316436768, + "num_tokens": 216446.0, + "step": 8 + }, + { + "epoch": 0.000742145625463841, + "grad_norm": 5.283108711242676, + "learning_rate": 2.198406155537235e-09, + "loss": 1.1371, + "mean_token_accuracy": 0.6906483769416809, + "num_tokens": 254137.0, + "step": 9 + }, + { + "epoch": 0.0008246062505153789, + "grad_norm": 6.035084247589111, + "learning_rate": 2.47320692497939e-09, + "loss": 1.2611, + "mean_token_accuracy": 0.6734949946403503, + "num_tokens": 284487.0, + "step": 10 + }, + { + "epoch": 0.0009070668755669168, + "grad_norm": 7.093228816986084, + "learning_rate": 2.7480076944215444e-09, + "loss": 1.3873, + "mean_token_accuracy": 0.6417458057403564, + "num_tokens": 309431.0, + "step": 11 + }, + { + "epoch": 0.0009895275006184547, + "grad_norm": 6.320603847503662, + "learning_rate": 3.0228084638636986e-09, + "loss": 1.3073, + "mean_token_accuracy": 0.6613972783088684, + "num_tokens": 337886.0, + "step": 12 + }, + { + "epoch": 0.0010719881256699925, + "grad_norm": 6.034179210662842, + "learning_rate": 3.297609233305853e-09, + "loss": 1.309, + "mean_token_accuracy": 0.6560201644897461, + "num_tokens": 367985.0, + "step": 13 + }, + { + "epoch": 0.0011544487507215305, + "grad_norm": 4.564627647399902, + "learning_rate": 3.572410002748008e-09, + "loss": 1.0211, + "mean_token_accuracy": 0.7244976758956909, + "num_tokens": 406957.0, + "step": 14 + }, + { + "epoch": 0.0012369093757730683, + "grad_norm": 6.395641803741455, + "learning_rate": 3.847210772190162e-09, + "loss": 1.0941, + "mean_token_accuracy": 0.7010120153427124, + "num_tokens": 434289.0, + "step": 15 + }, + { + "epoch": 0.0013193700008246063, + "grad_norm": 5.404370307922363, + "learning_rate": 4.122011541632316e-09, + "loss": 1.1876, + "mean_token_accuracy": 0.6685316562652588, + "num_tokens": 470251.0, + "step": 16 + }, + { + "epoch": 0.001401830625876144, + "grad_norm": 5.367228984832764, + "learning_rate": 4.39681231107447e-09, + "loss": 1.1799, + "mean_token_accuracy": 0.6752609014511108, + "num_tokens": 504501.0, + "step": 17 + }, + { + "epoch": 0.001484291250927682, + "grad_norm": 5.84118127822876, + "learning_rate": 4.671613080516625e-09, + "loss": 1.2716, + "mean_token_accuracy": 0.6661390662193298, + "num_tokens": 537334.0, + "step": 18 + }, + { + "epoch": 0.0015667518759792198, + "grad_norm": 6.231723785400391, + "learning_rate": 4.94641384995878e-09, + "loss": 1.3288, + "mean_token_accuracy": 0.6504815816879272, + "num_tokens": 566206.0, + "step": 19 + }, + { + "epoch": 0.0016492125010307578, + "grad_norm": 6.759753704071045, + "learning_rate": 5.221214619400934e-09, + "loss": 1.3772, + "mean_token_accuracy": 0.6391499042510986, + "num_tokens": 593220.0, + "step": 20 + }, + { + "epoch": 0.0017316731260822956, + "grad_norm": 5.558818340301514, + "learning_rate": 5.496015388843089e-09, + "loss": 1.1649, + "mean_token_accuracy": 0.6823024153709412, + "num_tokens": 625614.0, + "step": 21 + }, + { + "epoch": 0.0018141337511338336, + "grad_norm": 5.951631546020508, + "learning_rate": 5.770816158285243e-09, + "loss": 1.2658, + "mean_token_accuracy": 0.6583248376846313, + "num_tokens": 656459.0, + "step": 22 + }, + { + "epoch": 0.0018965943761853714, + "grad_norm": 5.905514717102051, + "learning_rate": 6.045616927727397e-09, + "loss": 1.2706, + "mean_token_accuracy": 0.6625591516494751, + "num_tokens": 687430.0, + "step": 23 + }, + { + "epoch": 0.0019790550012369094, + "grad_norm": 5.903946876525879, + "learning_rate": 6.320417697169552e-09, + "loss": 1.2729, + "mean_token_accuracy": 0.6645016074180603, + "num_tokens": 718362.0, + "step": 24 + }, + { + "epoch": 0.002061515626288447, + "grad_norm": 5.484348297119141, + "learning_rate": 6.595218466611706e-09, + "loss": 1.1918, + "mean_token_accuracy": 0.6961411237716675, + "num_tokens": 752388.0, + "step": 25 + }, + { + "epoch": 0.002143976251339985, + "grad_norm": 6.3132219314575195, + "learning_rate": 6.870019236053861e-09, + "loss": 1.2745, + "mean_token_accuracy": 0.6701524257659912, + "num_tokens": 780451.0, + "step": 26 + }, + { + "epoch": 0.002226436876391523, + "grad_norm": 6.4636077880859375, + "learning_rate": 7.144820005496016e-09, + "loss": 1.2975, + "mean_token_accuracy": 0.6539273858070374, + "num_tokens": 808019.0, + "step": 27 + }, + { + "epoch": 0.002308897501443061, + "grad_norm": 5.862984657287598, + "learning_rate": 7.41962077493817e-09, + "loss": 1.3516, + "mean_token_accuracy": 0.6415241956710815, + "num_tokens": 841519.0, + "step": 28 + }, + { + "epoch": 0.0023913581264945988, + "grad_norm": 5.5022077560424805, + "learning_rate": 7.694421544380324e-09, + "loss": 1.1937, + "mean_token_accuracy": 0.6725273132324219, + "num_tokens": 875087.0, + "step": 29 + }, + { + "epoch": 0.0024738187515461365, + "grad_norm": 6.791399955749512, + "learning_rate": 7.969222313822477e-09, + "loss": 1.4942, + "mean_token_accuracy": 0.6086106300354004, + "num_tokens": 903825.0, + "step": 30 + }, + { + "epoch": 0.0025562793765976748, + "grad_norm": 5.57703971862793, + "learning_rate": 8.244023083264632e-09, + "loss": 1.0544, + "mean_token_accuracy": 0.7042329907417297, + "num_tokens": 936418.0, + "step": 31 + }, + { + "epoch": 0.0026387400016492125, + "grad_norm": 5.852244853973389, + "learning_rate": 8.518823852706787e-09, + "loss": 1.2848, + "mean_token_accuracy": 0.6415903568267822, + "num_tokens": 968483.0, + "step": 32 + }, + { + "epoch": 0.0027212006267007503, + "grad_norm": 5.7766594886779785, + "learning_rate": 8.79362462214894e-09, + "loss": 1.279, + "mean_token_accuracy": 0.6564512252807617, + "num_tokens": 999172.0, + "step": 33 + }, + { + "epoch": 0.002803661251752288, + "grad_norm": 6.990233898162842, + "learning_rate": 9.068425391591096e-09, + "loss": 1.3752, + "mean_token_accuracy": 0.6317729949951172, + "num_tokens": 1025240.0, + "step": 34 + }, + { + "epoch": 0.0028861218768038263, + "grad_norm": 6.128726005554199, + "learning_rate": 9.34322616103325e-09, + "loss": 1.3392, + "mean_token_accuracy": 0.6430271863937378, + "num_tokens": 1056101.0, + "step": 35 + }, + { + "epoch": 0.002968582501855364, + "grad_norm": 5.962752819061279, + "learning_rate": 9.618026930475404e-09, + "loss": 1.4211, + "mean_token_accuracy": 0.6237756013870239, + "num_tokens": 1087867.0, + "step": 36 + }, + { + "epoch": 0.003051043126906902, + "grad_norm": 6.1746110916137695, + "learning_rate": 9.89282769991756e-09, + "loss": 1.2534, + "mean_token_accuracy": 0.6600938439369202, + "num_tokens": 1115447.0, + "step": 37 + }, + { + "epoch": 0.0031335037519584397, + "grad_norm": 5.779659271240234, + "learning_rate": 1.0167628469359714e-08, + "loss": 1.1662, + "mean_token_accuracy": 0.6911199688911438, + "num_tokens": 1147649.0, + "step": 38 + }, + { + "epoch": 0.003215964377009978, + "grad_norm": 5.969843864440918, + "learning_rate": 1.0442429238801868e-08, + "loss": 1.1905, + "mean_token_accuracy": 0.682762622833252, + "num_tokens": 1176786.0, + "step": 39 + }, + { + "epoch": 0.0032984250020615157, + "grad_norm": 5.713975429534912, + "learning_rate": 1.0717230008244023e-08, + "loss": 1.0769, + "mean_token_accuracy": 0.6998977661132812, + "num_tokens": 1207705.0, + "step": 40 + }, + { + "epoch": 0.0033808856271130535, + "grad_norm": 6.944198131561279, + "learning_rate": 1.0992030777686178e-08, + "loss": 1.373, + "mean_token_accuracy": 0.6433942317962646, + "num_tokens": 1232344.0, + "step": 41 + }, + { + "epoch": 0.0034633462521645913, + "grad_norm": 5.985361576080322, + "learning_rate": 1.1266831547128331e-08, + "loss": 1.189, + "mean_token_accuracy": 0.6749066114425659, + "num_tokens": 1262173.0, + "step": 42 + }, + { + "epoch": 0.0035458068772161295, + "grad_norm": 5.8308634757995605, + "learning_rate": 1.1541632316570486e-08, + "loss": 1.3584, + "mean_token_accuracy": 0.6450214982032776, + "num_tokens": 1294870.0, + "step": 43 + }, + { + "epoch": 0.0036282675022676673, + "grad_norm": 6.4569902420043945, + "learning_rate": 1.1816433086012641e-08, + "loss": 1.1706, + "mean_token_accuracy": 0.6841771602630615, + "num_tokens": 1321651.0, + "step": 44 + }, + { + "epoch": 0.003710728127319205, + "grad_norm": 5.517654895782471, + "learning_rate": 1.2091233855454794e-08, + "loss": 1.079, + "mean_token_accuracy": 0.7040948271751404, + "num_tokens": 1353707.0, + "step": 45 + }, + { + "epoch": 0.003793188752370743, + "grad_norm": 5.883145332336426, + "learning_rate": 1.236603462489695e-08, + "loss": 1.2927, + "mean_token_accuracy": 0.6588669419288635, + "num_tokens": 1385817.0, + "step": 46 + }, + { + "epoch": 0.003875649377422281, + "grad_norm": 5.286581516265869, + "learning_rate": 1.2640835394339104e-08, + "loss": 1.0588, + "mean_token_accuracy": 0.7039430141448975, + "num_tokens": 1420363.0, + "step": 47 + }, + { + "epoch": 0.003958110002473819, + "grad_norm": 5.743771553039551, + "learning_rate": 1.2915636163781258e-08, + "loss": 1.3667, + "mean_token_accuracy": 0.6428502798080444, + "num_tokens": 1454407.0, + "step": 48 + }, + { + "epoch": 0.004040570627525357, + "grad_norm": 5.754266262054443, + "learning_rate": 1.3190436933223413e-08, + "loss": 1.1959, + "mean_token_accuracy": 0.6773254871368408, + "num_tokens": 1484787.0, + "step": 49 + }, + { + "epoch": 0.004123031252576894, + "grad_norm": 5.313749313354492, + "learning_rate": 1.3465237702665568e-08, + "loss": 1.2425, + "mean_token_accuracy": 0.6649414300918579, + "num_tokens": 1521230.0, + "step": 50 + }, + { + "epoch": 0.004205491877628432, + "grad_norm": 6.171838760375977, + "learning_rate": 1.3740038472107721e-08, + "loss": 1.2478, + "mean_token_accuracy": 0.6720819473266602, + "num_tokens": 1550100.0, + "step": 51 + }, + { + "epoch": 0.00428795250267997, + "grad_norm": 5.228869438171387, + "learning_rate": 1.4014839241549876e-08, + "loss": 1.1212, + "mean_token_accuracy": 0.6976279020309448, + "num_tokens": 1583512.0, + "step": 52 + }, + { + "epoch": 0.004370413127731509, + "grad_norm": 6.39807653427124, + "learning_rate": 1.4289640010992031e-08, + "loss": 1.0891, + "mean_token_accuracy": 0.6783063411712646, + "num_tokens": 1609101.0, + "step": 53 + }, + { + "epoch": 0.004452873752783046, + "grad_norm": 6.670544624328613, + "learning_rate": 1.4564440780434185e-08, + "loss": 1.4114, + "mean_token_accuracy": 0.6280949115753174, + "num_tokens": 1638601.0, + "step": 54 + }, + { + "epoch": 0.004535334377834584, + "grad_norm": 6.133373260498047, + "learning_rate": 1.483924154987634e-08, + "loss": 1.417, + "mean_token_accuracy": 0.6313337683677673, + "num_tokens": 1670025.0, + "step": 55 + }, + { + "epoch": 0.004617795002886122, + "grad_norm": 5.351467132568359, + "learning_rate": 1.5114042319318493e-08, + "loss": 1.2493, + "mean_token_accuracy": 0.6697174906730652, + "num_tokens": 1706909.0, + "step": 56 + }, + { + "epoch": 0.00470025562793766, + "grad_norm": 5.932159423828125, + "learning_rate": 1.5388843088760648e-08, + "loss": 1.2537, + "mean_token_accuracy": 0.6626352667808533, + "num_tokens": 1738165.0, + "step": 57 + }, + { + "epoch": 0.0047827162529891975, + "grad_norm": 6.338653564453125, + "learning_rate": 1.56636438582028e-08, + "loss": 1.3048, + "mean_token_accuracy": 0.6594691276550293, + "num_tokens": 1765974.0, + "step": 58 + }, + { + "epoch": 0.004865176878040735, + "grad_norm": 5.5872344970703125, + "learning_rate": 1.5938444627644955e-08, + "loss": 1.3266, + "mean_token_accuracy": 0.6476937532424927, + "num_tokens": 1801148.0, + "step": 59 + }, + { + "epoch": 0.004947637503092273, + "grad_norm": 5.969048023223877, + "learning_rate": 1.621324539708711e-08, + "loss": 1.1334, + "mean_token_accuracy": 0.6832501888275146, + "num_tokens": 1830957.0, + "step": 60 + }, + { + "epoch": 0.005030098128143812, + "grad_norm": 7.203317642211914, + "learning_rate": 1.6488046166529265e-08, + "loss": 1.2549, + "mean_token_accuracy": 0.6532508134841919, + "num_tokens": 1854308.0, + "step": 61 + }, + { + "epoch": 0.0051125587531953495, + "grad_norm": 5.867611885070801, + "learning_rate": 1.676284693597142e-08, + "loss": 1.2342, + "mean_token_accuracy": 0.6837673187255859, + "num_tokens": 1886283.0, + "step": 62 + }, + { + "epoch": 0.005195019378246887, + "grad_norm": 5.889924049377441, + "learning_rate": 1.7037647705413575e-08, + "loss": 1.244, + "mean_token_accuracy": 0.657524824142456, + "num_tokens": 1916459.0, + "step": 63 + }, + { + "epoch": 0.005277480003298425, + "grad_norm": 6.187511920928955, + "learning_rate": 1.7312448474855727e-08, + "loss": 1.3306, + "mean_token_accuracy": 0.6431701183319092, + "num_tokens": 1947601.0, + "step": 64 + }, + { + "epoch": 0.005359940628349963, + "grad_norm": 6.723060607910156, + "learning_rate": 1.758724924429788e-08, + "loss": 1.5038, + "mean_token_accuracy": 0.6044639348983765, + "num_tokens": 1975752.0, + "step": 65 + }, + { + "epoch": 0.005442401253401501, + "grad_norm": 7.605066776275635, + "learning_rate": 1.7862050013740037e-08, + "loss": 1.1863, + "mean_token_accuracy": 0.6861426830291748, + "num_tokens": 1996855.0, + "step": 66 + }, + { + "epoch": 0.0055248618784530384, + "grad_norm": 6.475542068481445, + "learning_rate": 1.813685078318219e-08, + "loss": 1.3312, + "mean_token_accuracy": 0.6513146162033081, + "num_tokens": 2025126.0, + "step": 67 + }, + { + "epoch": 0.005607322503504576, + "grad_norm": 7.206477165222168, + "learning_rate": 1.8411651552624347e-08, + "loss": 1.3712, + "mean_token_accuracy": 0.636810302734375, + "num_tokens": 2048763.0, + "step": 68 + }, + { + "epoch": 0.005689783128556115, + "grad_norm": 5.685336589813232, + "learning_rate": 1.86864523220665e-08, + "loss": 1.1837, + "mean_token_accuracy": 0.6688089966773987, + "num_tokens": 2082503.0, + "step": 69 + }, + { + "epoch": 0.005772243753607653, + "grad_norm": 5.396636009216309, + "learning_rate": 1.8961253091508653e-08, + "loss": 1.1134, + "mean_token_accuracy": 0.6937447190284729, + "num_tokens": 2114397.0, + "step": 70 + }, + { + "epoch": 0.0058547043786591904, + "grad_norm": 4.933896064758301, + "learning_rate": 1.9236053860950808e-08, + "loss": 1.1873, + "mean_token_accuracy": 0.6840764284133911, + "num_tokens": 2155850.0, + "step": 71 + }, + { + "epoch": 0.005937165003710728, + "grad_norm": 5.631091594696045, + "learning_rate": 1.9510854630392963e-08, + "loss": 1.3825, + "mean_token_accuracy": 0.6411492228507996, + "num_tokens": 2191136.0, + "step": 72 + }, + { + "epoch": 0.006019625628762266, + "grad_norm": 6.937118053436279, + "learning_rate": 1.978565539983512e-08, + "loss": 1.4861, + "mean_token_accuracy": 0.6134428977966309, + "num_tokens": 2218330.0, + "step": 73 + }, + { + "epoch": 0.006102086253813804, + "grad_norm": 6.121829032897949, + "learning_rate": 2.0060456169277273e-08, + "loss": 1.2521, + "mean_token_accuracy": 0.6642712950706482, + "num_tokens": 2247241.0, + "step": 74 + }, + { + "epoch": 0.006184546878865342, + "grad_norm": 6.343318462371826, + "learning_rate": 2.033525693871943e-08, + "loss": 1.263, + "mean_token_accuracy": 0.6698535084724426, + "num_tokens": 2273781.0, + "step": 75 + }, + { + "epoch": 0.006267007503916879, + "grad_norm": 6.011372089385986, + "learning_rate": 2.061005770816158e-08, + "loss": 1.4239, + "mean_token_accuracy": 0.6156130433082581, + "num_tokens": 2306315.0, + "step": 76 + }, + { + "epoch": 0.006349468128968417, + "grad_norm": 7.0069193840026855, + "learning_rate": 2.0884858477603735e-08, + "loss": 1.534, + "mean_token_accuracy": 0.6011335849761963, + "num_tokens": 2333388.0, + "step": 77 + }, + { + "epoch": 0.006431928754019956, + "grad_norm": 7.314938068389893, + "learning_rate": 2.115965924704589e-08, + "loss": 1.6428, + "mean_token_accuracy": 0.5810621976852417, + "num_tokens": 2360584.0, + "step": 78 + }, + { + "epoch": 0.006514389379071494, + "grad_norm": 6.434562683105469, + "learning_rate": 2.1434460016488045e-08, + "loss": 1.2393, + "mean_token_accuracy": 0.6681070327758789, + "num_tokens": 2387976.0, + "step": 79 + }, + { + "epoch": 0.006596850004123031, + "grad_norm": 7.042853832244873, + "learning_rate": 2.17092607859302e-08, + "loss": 1.3068, + "mean_token_accuracy": 0.6432262659072876, + "num_tokens": 2412237.0, + "step": 80 + }, + { + "epoch": 0.006679310629174569, + "grad_norm": 5.069262981414795, + "learning_rate": 2.1984061555372355e-08, + "loss": 1.175, + "mean_token_accuracy": 0.6856743097305298, + "num_tokens": 2448864.0, + "step": 81 + }, + { + "epoch": 0.006761771254226107, + "grad_norm": 6.486860275268555, + "learning_rate": 2.2258862324814507e-08, + "loss": 1.4362, + "mean_token_accuracy": 0.622995138168335, + "num_tokens": 2477499.0, + "step": 82 + }, + { + "epoch": 0.006844231879277645, + "grad_norm": 5.3121137619018555, + "learning_rate": 2.2533663094256662e-08, + "loss": 1.1812, + "mean_token_accuracy": 0.6867948174476624, + "num_tokens": 2512313.0, + "step": 83 + }, + { + "epoch": 0.0069266925043291825, + "grad_norm": 5.730700492858887, + "learning_rate": 2.2808463863698817e-08, + "loss": 1.1859, + "mean_token_accuracy": 0.679646372795105, + "num_tokens": 2542392.0, + "step": 84 + }, + { + "epoch": 0.00700915312938072, + "grad_norm": 5.313466548919678, + "learning_rate": 2.3083264633140972e-08, + "loss": 1.1654, + "mean_token_accuracy": 0.6908590197563171, + "num_tokens": 2576633.0, + "step": 85 + }, + { + "epoch": 0.007091613754432259, + "grad_norm": 6.5687174797058105, + "learning_rate": 2.3358065402583127e-08, + "loss": 1.3131, + "mean_token_accuracy": 0.6543537378311157, + "num_tokens": 2603218.0, + "step": 86 + }, + { + "epoch": 0.007174074379483797, + "grad_norm": 5.670802116394043, + "learning_rate": 2.3632866172025282e-08, + "loss": 1.2336, + "mean_token_accuracy": 0.6773097515106201, + "num_tokens": 2635605.0, + "step": 87 + }, + { + "epoch": 0.0072565350045353345, + "grad_norm": 6.671061992645264, + "learning_rate": 2.3907666941467434e-08, + "loss": 1.3879, + "mean_token_accuracy": 0.6372206211090088, + "num_tokens": 2662108.0, + "step": 88 + }, + { + "epoch": 0.007338995629586872, + "grad_norm": 5.788033485412598, + "learning_rate": 2.418246771090959e-08, + "loss": 1.0822, + "mean_token_accuracy": 0.7066649794578552, + "num_tokens": 2690539.0, + "step": 89 + }, + { + "epoch": 0.00742145625463841, + "grad_norm": 5.770917892456055, + "learning_rate": 2.4457268480351744e-08, + "loss": 1.3131, + "mean_token_accuracy": 0.6571993231773376, + "num_tokens": 2722965.0, + "step": 90 + }, + { + "epoch": 0.007503916879689948, + "grad_norm": 5.975864887237549, + "learning_rate": 2.47320692497939e-08, + "loss": 1.3716, + "mean_token_accuracy": 0.6347829103469849, + "num_tokens": 2754264.0, + "step": 91 + }, + { + "epoch": 0.007586377504741486, + "grad_norm": 5.286114692687988, + "learning_rate": 2.5006870019236054e-08, + "loss": 1.1693, + "mean_token_accuracy": 0.6818931102752686, + "num_tokens": 2787861.0, + "step": 92 + }, + { + "epoch": 0.007668838129793023, + "grad_norm": 6.72628927230835, + "learning_rate": 2.528167078867821e-08, + "loss": 1.3427, + "mean_token_accuracy": 0.6396744847297668, + "num_tokens": 2814459.0, + "step": 93 + }, + { + "epoch": 0.007751298754844562, + "grad_norm": 6.088047504425049, + "learning_rate": 2.555647155812036e-08, + "loss": 1.39, + "mean_token_accuracy": 0.6335146427154541, + "num_tokens": 2846007.0, + "step": 94 + }, + { + "epoch": 0.0078337593798961, + "grad_norm": 8.082930564880371, + "learning_rate": 2.5831272327562516e-08, + "loss": 1.4481, + "mean_token_accuracy": 0.6209391355514526, + "num_tokens": 2865986.0, + "step": 95 + }, + { + "epoch": 0.007916220004947638, + "grad_norm": 5.738827228546143, + "learning_rate": 2.610607309700467e-08, + "loss": 1.2513, + "mean_token_accuracy": 0.660571813583374, + "num_tokens": 2895813.0, + "step": 96 + }, + { + "epoch": 0.007998680629999175, + "grad_norm": 5.186507701873779, + "learning_rate": 2.6380873866446826e-08, + "loss": 1.2102, + "mean_token_accuracy": 0.6837412714958191, + "num_tokens": 2930311.0, + "step": 97 + }, + { + "epoch": 0.008081141255050713, + "grad_norm": 5.574655055999756, + "learning_rate": 2.665567463588898e-08, + "loss": 1.1734, + "mean_token_accuracy": 0.682221531867981, + "num_tokens": 2960345.0, + "step": 98 + }, + { + "epoch": 0.008163601880102251, + "grad_norm": 4.888953685760498, + "learning_rate": 2.6930475405331136e-08, + "loss": 1.1235, + "mean_token_accuracy": 0.6935495138168335, + "num_tokens": 2997256.0, + "step": 99 + }, + { + "epoch": 0.008246062505153789, + "grad_norm": 6.029450416564941, + "learning_rate": 2.7205276174773287e-08, + "loss": 1.2211, + "mean_token_accuracy": 0.6599943041801453, + "num_tokens": 3023854.0, + "step": 100 + }, + { + "epoch": 0.008328523130205327, + "grad_norm": 5.477640151977539, + "learning_rate": 2.7480076944215442e-08, + "loss": 1.0331, + "mean_token_accuracy": 0.7184940576553345, + "num_tokens": 3053619.0, + "step": 101 + }, + { + "epoch": 0.008410983755256864, + "grad_norm": 5.947192668914795, + "learning_rate": 2.7754877713657597e-08, + "loss": 1.2352, + "mean_token_accuracy": 0.6741504669189453, + "num_tokens": 3082031.0, + "step": 102 + }, + { + "epoch": 0.008493444380308402, + "grad_norm": 6.559183120727539, + "learning_rate": 2.8029678483099752e-08, + "loss": 1.3299, + "mean_token_accuracy": 0.6372818946838379, + "num_tokens": 3108443.0, + "step": 103 + }, + { + "epoch": 0.00857590500535994, + "grad_norm": 5.9085845947265625, + "learning_rate": 2.8304479252541907e-08, + "loss": 1.3113, + "mean_token_accuracy": 0.6493954658508301, + "num_tokens": 3138862.0, + "step": 104 + }, + { + "epoch": 0.008658365630411478, + "grad_norm": 5.937504291534424, + "learning_rate": 2.8579280021984062e-08, + "loss": 1.3773, + "mean_token_accuracy": 0.6436346769332886, + "num_tokens": 3169851.0, + "step": 105 + }, + { + "epoch": 0.008740826255463017, + "grad_norm": 6.015381336212158, + "learning_rate": 2.8854080791426214e-08, + "loss": 1.1909, + "mean_token_accuracy": 0.6684638261795044, + "num_tokens": 3197297.0, + "step": 106 + }, + { + "epoch": 0.008823286880514555, + "grad_norm": 6.416136741638184, + "learning_rate": 2.912888156086837e-08, + "loss": 1.3082, + "mean_token_accuracy": 0.6513243913650513, + "num_tokens": 3222998.0, + "step": 107 + }, + { + "epoch": 0.008905747505566093, + "grad_norm": 6.039608001708984, + "learning_rate": 2.9403682330310524e-08, + "loss": 1.3311, + "mean_token_accuracy": 0.6485172510147095, + "num_tokens": 3252525.0, + "step": 108 + }, + { + "epoch": 0.00898820813061763, + "grad_norm": 7.083721160888672, + "learning_rate": 2.967848309975268e-08, + "loss": 1.3992, + "mean_token_accuracy": 0.6359808444976807, + "num_tokens": 3277376.0, + "step": 109 + }, + { + "epoch": 0.009070668755669168, + "grad_norm": 5.922708988189697, + "learning_rate": 2.9953283869194834e-08, + "loss": 1.2394, + "mean_token_accuracy": 0.6662954092025757, + "num_tokens": 3307384.0, + "step": 110 + }, + { + "epoch": 0.009153129380720706, + "grad_norm": 5.325632095336914, + "learning_rate": 3.0228084638636986e-08, + "loss": 1.3062, + "mean_token_accuracy": 0.6446139216423035, + "num_tokens": 3343221.0, + "step": 111 + }, + { + "epoch": 0.009235590005772244, + "grad_norm": 5.16700553894043, + "learning_rate": 3.0502885408079144e-08, + "loss": 1.0572, + "mean_token_accuracy": 0.7184382677078247, + "num_tokens": 3375354.0, + "step": 112 + }, + { + "epoch": 0.009318050630823782, + "grad_norm": 6.018697738647461, + "learning_rate": 3.0777686177521296e-08, + "loss": 1.1335, + "mean_token_accuracy": 0.6817761063575745, + "num_tokens": 3402348.0, + "step": 113 + }, + { + "epoch": 0.00940051125587532, + "grad_norm": 5.848097801208496, + "learning_rate": 3.1052486946963454e-08, + "loss": 1.2661, + "mean_token_accuracy": 0.6647899150848389, + "num_tokens": 3432734.0, + "step": 114 + }, + { + "epoch": 0.009482971880926857, + "grad_norm": 5.89310884475708, + "learning_rate": 3.13272877164056e-08, + "loss": 1.311, + "mean_token_accuracy": 0.6435703039169312, + "num_tokens": 3462246.0, + "step": 115 + }, + { + "epoch": 0.009565432505978395, + "grad_norm": 5.622847557067871, + "learning_rate": 3.160208848584776e-08, + "loss": 1.193, + "mean_token_accuracy": 0.6750844717025757, + "num_tokens": 3492802.0, + "step": 116 + }, + { + "epoch": 0.009647893131029933, + "grad_norm": 5.635550022125244, + "learning_rate": 3.187688925528991e-08, + "loss": 1.4613, + "mean_token_accuracy": 0.6156536340713501, + "num_tokens": 3527168.0, + "step": 117 + }, + { + "epoch": 0.00973035375608147, + "grad_norm": 5.599702835083008, + "learning_rate": 3.215169002473207e-08, + "loss": 1.3216, + "mean_token_accuracy": 0.6567103266716003, + "num_tokens": 3559832.0, + "step": 118 + }, + { + "epoch": 0.009812814381133008, + "grad_norm": 6.269641876220703, + "learning_rate": 3.242649079417422e-08, + "loss": 1.1958, + "mean_token_accuracy": 0.671033501625061, + "num_tokens": 3586270.0, + "step": 119 + }, + { + "epoch": 0.009895275006184546, + "grad_norm": 5.825554370880127, + "learning_rate": 3.270129156361638e-08, + "loss": 1.1765, + "mean_token_accuracy": 0.6766960620880127, + "num_tokens": 3616005.0, + "step": 120 + }, + { + "epoch": 0.009977735631236084, + "grad_norm": 5.768960475921631, + "learning_rate": 3.297609233305853e-08, + "loss": 1.0529, + "mean_token_accuracy": 0.6943415403366089, + "num_tokens": 3643200.0, + "step": 121 + }, + { + "epoch": 0.010060196256287623, + "grad_norm": 5.692115783691406, + "learning_rate": 3.325089310250068e-08, + "loss": 1.3552, + "mean_token_accuracy": 0.6404517889022827, + "num_tokens": 3677278.0, + "step": 122 + }, + { + "epoch": 0.010142656881339161, + "grad_norm": 5.656862258911133, + "learning_rate": 3.352569387194284e-08, + "loss": 1.3034, + "mean_token_accuracy": 0.643051266670227, + "num_tokens": 3709505.0, + "step": 123 + }, + { + "epoch": 0.010225117506390699, + "grad_norm": 5.10575008392334, + "learning_rate": 3.380049464138499e-08, + "loss": 1.2902, + "mean_token_accuracy": 0.6482587456703186, + "num_tokens": 3748325.0, + "step": 124 + }, + { + "epoch": 0.010307578131442237, + "grad_norm": 5.597771644592285, + "learning_rate": 3.407529541082715e-08, + "loss": 1.1189, + "mean_token_accuracy": 0.6876203417778015, + "num_tokens": 3777328.0, + "step": 125 + }, + { + "epoch": 0.010390038756493775, + "grad_norm": 5.270596981048584, + "learning_rate": 3.43500961802693e-08, + "loss": 1.2067, + "mean_token_accuracy": 0.6742527484893799, + "num_tokens": 3810081.0, + "step": 126 + }, + { + "epoch": 0.010472499381545312, + "grad_norm": 5.677281856536865, + "learning_rate": 3.462489694971145e-08, + "loss": 1.3196, + "mean_token_accuracy": 0.6542202234268188, + "num_tokens": 3839108.0, + "step": 127 + }, + { + "epoch": 0.01055496000659685, + "grad_norm": 6.0510125160217285, + "learning_rate": 3.489969771915361e-08, + "loss": 1.4742, + "mean_token_accuracy": 0.6098109483718872, + "num_tokens": 3870018.0, + "step": 128 + }, + { + "epoch": 0.010637420631648388, + "grad_norm": 5.390179634094238, + "learning_rate": 3.517449848859576e-08, + "loss": 1.1973, + "mean_token_accuracy": 0.6720077395439148, + "num_tokens": 3899583.0, + "step": 129 + }, + { + "epoch": 0.010719881256699926, + "grad_norm": 4.678900718688965, + "learning_rate": 3.544929925803792e-08, + "loss": 1.0746, + "mean_token_accuracy": 0.698418378829956, + "num_tokens": 3937203.0, + "step": 130 + }, + { + "epoch": 0.010802341881751464, + "grad_norm": 4.594918727874756, + "learning_rate": 3.572410002748007e-08, + "loss": 1.1702, + "mean_token_accuracy": 0.6728537082672119, + "num_tokens": 3976237.0, + "step": 131 + }, + { + "epoch": 0.010884802506803001, + "grad_norm": 4.6312255859375, + "learning_rate": 3.599890079692223e-08, + "loss": 1.074, + "mean_token_accuracy": 0.7068182229995728, + "num_tokens": 4014616.0, + "step": 132 + }, + { + "epoch": 0.010967263131854539, + "grad_norm": 4.833487033843994, + "learning_rate": 3.627370156636438e-08, + "loss": 1.0661, + "mean_token_accuracy": 0.7072105407714844, + "num_tokens": 4047517.0, + "step": 133 + }, + { + "epoch": 0.011049723756906077, + "grad_norm": 5.477456569671631, + "learning_rate": 3.6548502335806535e-08, + "loss": 1.3452, + "mean_token_accuracy": 0.6375916004180908, + "num_tokens": 4079562.0, + "step": 134 + }, + { + "epoch": 0.011132184381957615, + "grad_norm": 6.798102378845215, + "learning_rate": 3.682330310524869e-08, + "loss": 1.3713, + "mean_token_accuracy": 0.6332886219024658, + "num_tokens": 4103074.0, + "step": 135 + }, + { + "epoch": 0.011214645007009152, + "grad_norm": 4.6200103759765625, + "learning_rate": 3.7098103874690845e-08, + "loss": 1.1158, + "mean_token_accuracy": 0.6927310824394226, + "num_tokens": 4139708.0, + "step": 136 + }, + { + "epoch": 0.01129710563206069, + "grad_norm": 6.142546653747559, + "learning_rate": 3.7372904644133e-08, + "loss": 1.2513, + "mean_token_accuracy": 0.6578351259231567, + "num_tokens": 4164566.0, + "step": 137 + }, + { + "epoch": 0.01137956625711223, + "grad_norm": 5.669223308563232, + "learning_rate": 3.7647705413575155e-08, + "loss": 1.4462, + "mean_token_accuracy": 0.6098023653030396, + "num_tokens": 4197008.0, + "step": 138 + }, + { + "epoch": 0.011462026882163768, + "grad_norm": 4.950489044189453, + "learning_rate": 3.7922506183017307e-08, + "loss": 1.0784, + "mean_token_accuracy": 0.710374653339386, + "num_tokens": 4230391.0, + "step": 139 + }, + { + "epoch": 0.011544487507215305, + "grad_norm": 7.047382831573486, + "learning_rate": 3.8197306952459465e-08, + "loss": 1.3895, + "mean_token_accuracy": 0.6311147212982178, + "num_tokens": 4252354.0, + "step": 140 + }, + { + "epoch": 0.011626948132266843, + "grad_norm": 6.123524188995361, + "learning_rate": 3.8472107721901617e-08, + "loss": 1.2456, + "mean_token_accuracy": 0.6655297875404358, + "num_tokens": 4278307.0, + "step": 141 + }, + { + "epoch": 0.011709408757318381, + "grad_norm": 4.898184776306152, + "learning_rate": 3.8746908491343775e-08, + "loss": 1.1879, + "mean_token_accuracy": 0.6837910413742065, + "num_tokens": 4311182.0, + "step": 142 + }, + { + "epoch": 0.011791869382369919, + "grad_norm": 5.183446884155273, + "learning_rate": 3.902170926078593e-08, + "loss": 1.2639, + "mean_token_accuracy": 0.6650547981262207, + "num_tokens": 4346473.0, + "step": 143 + }, + { + "epoch": 0.011874330007421456, + "grad_norm": 4.748002529144287, + "learning_rate": 3.9296510030228085e-08, + "loss": 1.1265, + "mean_token_accuracy": 0.6964375972747803, + "num_tokens": 4382913.0, + "step": 144 + }, + { + "epoch": 0.011956790632472994, + "grad_norm": 5.348448753356934, + "learning_rate": 3.957131079967024e-08, + "loss": 1.278, + "mean_token_accuracy": 0.6547508239746094, + "num_tokens": 4414339.0, + "step": 145 + }, + { + "epoch": 0.012039251257524532, + "grad_norm": 6.6946587562561035, + "learning_rate": 3.984611156911239e-08, + "loss": 1.3954, + "mean_token_accuracy": 0.6334571242332458, + "num_tokens": 4438683.0, + "step": 146 + }, + { + "epoch": 0.01212171188257607, + "grad_norm": 6.089223384857178, + "learning_rate": 4.012091233855455e-08, + "loss": 1.3438, + "mean_token_accuracy": 0.6489163637161255, + "num_tokens": 4465949.0, + "step": 147 + }, + { + "epoch": 0.012204172507627608, + "grad_norm": 5.879213333129883, + "learning_rate": 4.03957131079967e-08, + "loss": 1.2147, + "mean_token_accuracy": 0.6612224578857422, + "num_tokens": 4493987.0, + "step": 148 + }, + { + "epoch": 0.012286633132679145, + "grad_norm": 5.042893886566162, + "learning_rate": 4.067051387743886e-08, + "loss": 1.1666, + "mean_token_accuracy": 0.6846092343330383, + "num_tokens": 4527762.0, + "step": 149 + }, + { + "epoch": 0.012369093757730683, + "grad_norm": 7.939912796020508, + "learning_rate": 4.094531464688101e-08, + "loss": 1.1535, + "mean_token_accuracy": 0.6937382221221924, + "num_tokens": 4546348.0, + "step": 150 + }, + { + "epoch": 0.012451554382782221, + "grad_norm": 5.6081767082214355, + "learning_rate": 4.122011541632316e-08, + "loss": 1.2498, + "mean_token_accuracy": 0.6595978140830994, + "num_tokens": 4575997.0, + "step": 151 + }, + { + "epoch": 0.012534015007833759, + "grad_norm": 5.0980706214904785, + "learning_rate": 4.149491618576532e-08, + "loss": 1.229, + "mean_token_accuracy": 0.6651886105537415, + "num_tokens": 4610344.0, + "step": 152 + }, + { + "epoch": 0.012616475632885297, + "grad_norm": 5.472636699676514, + "learning_rate": 4.176971695520747e-08, + "loss": 1.2925, + "mean_token_accuracy": 0.6441056728363037, + "num_tokens": 4642259.0, + "step": 153 + }, + { + "epoch": 0.012698936257936834, + "grad_norm": 5.020190715789795, + "learning_rate": 4.204451772464963e-08, + "loss": 1.2575, + "mean_token_accuracy": 0.6581822633743286, + "num_tokens": 4678793.0, + "step": 154 + }, + { + "epoch": 0.012781396882988374, + "grad_norm": 5.385252952575684, + "learning_rate": 4.231931849409178e-08, + "loss": 1.2962, + "mean_token_accuracy": 0.6531580686569214, + "num_tokens": 4710757.0, + "step": 155 + }, + { + "epoch": 0.012863857508039912, + "grad_norm": 5.073395252227783, + "learning_rate": 4.259411926353394e-08, + "loss": 1.1578, + "mean_token_accuracy": 0.683093786239624, + "num_tokens": 4744775.0, + "step": 156 + }, + { + "epoch": 0.01294631813309145, + "grad_norm": 5.111137390136719, + "learning_rate": 4.286892003297609e-08, + "loss": 1.1515, + "mean_token_accuracy": 0.6884509921073914, + "num_tokens": 4777319.0, + "step": 157 + }, + { + "epoch": 0.013028778758142987, + "grad_norm": 5.476365089416504, + "learning_rate": 4.314372080241824e-08, + "loss": 1.2198, + "mean_token_accuracy": 0.6749355792999268, + "num_tokens": 4806590.0, + "step": 158 + }, + { + "epoch": 0.013111239383194525, + "grad_norm": 5.713489532470703, + "learning_rate": 4.34185215718604e-08, + "loss": 1.2901, + "mean_token_accuracy": 0.6450214982032776, + "num_tokens": 4836549.0, + "step": 159 + }, + { + "epoch": 0.013193700008246063, + "grad_norm": 6.3802266120910645, + "learning_rate": 4.369332234130255e-08, + "loss": 1.2999, + "mean_token_accuracy": 0.6519960165023804, + "num_tokens": 4863155.0, + "step": 160 + }, + { + "epoch": 0.0132761606332976, + "grad_norm": 5.729377746582031, + "learning_rate": 4.396812311074471e-08, + "loss": 1.3079, + "mean_token_accuracy": 0.6619119644165039, + "num_tokens": 4893069.0, + "step": 161 + }, + { + "epoch": 0.013358621258349138, + "grad_norm": 5.992189884185791, + "learning_rate": 4.424292388018686e-08, + "loss": 1.2425, + "mean_token_accuracy": 0.6650122404098511, + "num_tokens": 4919654.0, + "step": 162 + }, + { + "epoch": 0.013441081883400676, + "grad_norm": 5.799715995788574, + "learning_rate": 4.4517724649629014e-08, + "loss": 1.3378, + "mean_token_accuracy": 0.6385664939880371, + "num_tokens": 4950089.0, + "step": 163 + }, + { + "epoch": 0.013523542508452214, + "grad_norm": 5.253135681152344, + "learning_rate": 4.479252541907117e-08, + "loss": 1.0639, + "mean_token_accuracy": 0.7029356956481934, + "num_tokens": 4981010.0, + "step": 164 + }, + { + "epoch": 0.013606003133503752, + "grad_norm": 6.125741481781006, + "learning_rate": 4.5067326188513324e-08, + "loss": 1.3144, + "mean_token_accuracy": 0.6424979567527771, + "num_tokens": 5010048.0, + "step": 165 + }, + { + "epoch": 0.01368846375855529, + "grad_norm": 5.747437477111816, + "learning_rate": 4.534212695795548e-08, + "loss": 1.192, + "mean_token_accuracy": 0.6656852960586548, + "num_tokens": 5037672.0, + "step": 166 + }, + { + "epoch": 0.013770924383606827, + "grad_norm": 5.433465003967285, + "learning_rate": 4.5616927727397634e-08, + "loss": 1.1917, + "mean_token_accuracy": 0.672165036201477, + "num_tokens": 5069298.0, + "step": 167 + }, + { + "epoch": 0.013853385008658365, + "grad_norm": 5.981449127197266, + "learning_rate": 4.589172849683979e-08, + "loss": 1.1756, + "mean_token_accuracy": 0.683349609375, + "num_tokens": 5096372.0, + "step": 168 + }, + { + "epoch": 0.013935845633709903, + "grad_norm": 4.893197536468506, + "learning_rate": 4.6166529266281944e-08, + "loss": 1.1322, + "mean_token_accuracy": 0.6861275434494019, + "num_tokens": 5130351.0, + "step": 169 + }, + { + "epoch": 0.01401830625876144, + "grad_norm": 5.12287712097168, + "learning_rate": 4.6441330035724096e-08, + "loss": 1.156, + "mean_token_accuracy": 0.6763931512832642, + "num_tokens": 5162056.0, + "step": 170 + }, + { + "epoch": 0.01410076688381298, + "grad_norm": 5.1534647941589355, + "learning_rate": 4.6716130805166254e-08, + "loss": 0.9942, + "mean_token_accuracy": 0.7017965316772461, + "num_tokens": 5190046.0, + "step": 171 + }, + { + "epoch": 0.014183227508864518, + "grad_norm": 5.526709079742432, + "learning_rate": 4.6990931574608406e-08, + "loss": 1.1481, + "mean_token_accuracy": 0.6813735961914062, + "num_tokens": 5216856.0, + "step": 172 + }, + { + "epoch": 0.014265688133916056, + "grad_norm": 4.96182107925415, + "learning_rate": 4.7265732344050564e-08, + "loss": 1.0732, + "mean_token_accuracy": 0.7047591209411621, + "num_tokens": 5245968.0, + "step": 173 + }, + { + "epoch": 0.014348148758967593, + "grad_norm": 4.964123725891113, + "learning_rate": 4.7540533113492716e-08, + "loss": 1.2202, + "mean_token_accuracy": 0.6655603051185608, + "num_tokens": 5280190.0, + "step": 174 + }, + { + "epoch": 0.014430609384019131, + "grad_norm": 5.721307754516602, + "learning_rate": 4.781533388293487e-08, + "loss": 1.5337, + "mean_token_accuracy": 0.5912930965423584, + "num_tokens": 5310952.0, + "step": 175 + }, + { + "epoch": 0.014513070009070669, + "grad_norm": 5.598477840423584, + "learning_rate": 4.8090134652377026e-08, + "loss": 1.4727, + "mean_token_accuracy": 0.6041402816772461, + "num_tokens": 5343708.0, + "step": 176 + }, + { + "epoch": 0.014595530634122207, + "grad_norm": 4.935248374938965, + "learning_rate": 4.836493542181918e-08, + "loss": 1.2588, + "mean_token_accuracy": 0.6575685739517212, + "num_tokens": 5373930.0, + "step": 177 + }, + { + "epoch": 0.014677991259173745, + "grad_norm": 6.347020626068115, + "learning_rate": 4.8639736191261336e-08, + "loss": 1.1755, + "mean_token_accuracy": 0.670225977897644, + "num_tokens": 5394307.0, + "step": 178 + }, + { + "epoch": 0.014760451884225282, + "grad_norm": 6.24328088760376, + "learning_rate": 4.891453696070349e-08, + "loss": 1.305, + "mean_token_accuracy": 0.6417940855026245, + "num_tokens": 5416910.0, + "step": 179 + }, + { + "epoch": 0.01484291250927682, + "grad_norm": 4.341771602630615, + "learning_rate": 4.9189337730145646e-08, + "loss": 1.2918, + "mean_token_accuracy": 0.6588448882102966, + "num_tokens": 5454869.0, + "step": 180 + }, + { + "epoch": 0.014925373134328358, + "grad_norm": 4.263097286224365, + "learning_rate": 4.94641384995878e-08, + "loss": 1.2195, + "mean_token_accuracy": 0.669669508934021, + "num_tokens": 5490769.0, + "step": 181 + }, + { + "epoch": 0.015007833759379896, + "grad_norm": 4.392491817474365, + "learning_rate": 4.973893926902995e-08, + "loss": 1.0304, + "mean_token_accuracy": 0.7177757620811462, + "num_tokens": 5522268.0, + "step": 182 + }, + { + "epoch": 0.015090294384431433, + "grad_norm": 4.348973751068115, + "learning_rate": 5.001374003847211e-08, + "loss": 1.0412, + "mean_token_accuracy": 0.7003046870231628, + "num_tokens": 5550632.0, + "step": 183 + }, + { + "epoch": 0.015172755009482971, + "grad_norm": 4.202242374420166, + "learning_rate": 5.028854080791426e-08, + "loss": 1.0101, + "mean_token_accuracy": 0.7154175043106079, + "num_tokens": 5581758.0, + "step": 184 + }, + { + "epoch": 0.015255215634534509, + "grad_norm": 4.724694728851318, + "learning_rate": 5.056334157735642e-08, + "loss": 1.2449, + "mean_token_accuracy": 0.6657639145851135, + "num_tokens": 5610827.0, + "step": 185 + }, + { + "epoch": 0.015337676259586047, + "grad_norm": 4.679598331451416, + "learning_rate": 5.083814234679857e-08, + "loss": 1.2588, + "mean_token_accuracy": 0.6576530933380127, + "num_tokens": 5641599.0, + "step": 186 + }, + { + "epoch": 0.015420136884637586, + "grad_norm": 5.345077991485596, + "learning_rate": 5.111294311624072e-08, + "loss": 1.2864, + "mean_token_accuracy": 0.6474964022636414, + "num_tokens": 5665161.0, + "step": 187 + }, + { + "epoch": 0.015502597509689124, + "grad_norm": 3.946035385131836, + "learning_rate": 5.138774388568288e-08, + "loss": 1.0231, + "mean_token_accuracy": 0.7042633891105652, + "num_tokens": 5699452.0, + "step": 188 + }, + { + "epoch": 0.015585058134740662, + "grad_norm": 4.401646137237549, + "learning_rate": 5.166254465512503e-08, + "loss": 1.3118, + "mean_token_accuracy": 0.6436024308204651, + "num_tokens": 5735270.0, + "step": 189 + }, + { + "epoch": 0.0156675187597922, + "grad_norm": 3.975121259689331, + "learning_rate": 5.193734542456719e-08, + "loss": 1.1385, + "mean_token_accuracy": 0.6822680830955505, + "num_tokens": 5769330.0, + "step": 190 + }, + { + "epoch": 0.015749979384843737, + "grad_norm": 4.236950874328613, + "learning_rate": 5.221214619400934e-08, + "loss": 1.2375, + "mean_token_accuracy": 0.6556273698806763, + "num_tokens": 5802261.0, + "step": 191 + }, + { + "epoch": 0.015832440009895275, + "grad_norm": 5.078976631164551, + "learning_rate": 5.24869469634515e-08, + "loss": 1.1932, + "mean_token_accuracy": 0.6758015155792236, + "num_tokens": 5826535.0, + "step": 192 + }, + { + "epoch": 0.015914900634946813, + "grad_norm": 3.7230005264282227, + "learning_rate": 5.276174773289365e-08, + "loss": 1.1333, + "mean_token_accuracy": 0.6801905632019043, + "num_tokens": 5867218.0, + "step": 193 + }, + { + "epoch": 0.01599736125999835, + "grad_norm": 5.337279319763184, + "learning_rate": 5.30365485023358e-08, + "loss": 1.1759, + "mean_token_accuracy": 0.6812642812728882, + "num_tokens": 5889808.0, + "step": 194 + }, + { + "epoch": 0.01607982188504989, + "grad_norm": 4.124512672424316, + "learning_rate": 5.331134927177796e-08, + "loss": 1.1061, + "mean_token_accuracy": 0.6920101046562195, + "num_tokens": 5921521.0, + "step": 195 + }, + { + "epoch": 0.016162282510101426, + "grad_norm": 4.59129524230957, + "learning_rate": 5.358615004122011e-08, + "loss": 1.2624, + "mean_token_accuracy": 0.6555290818214417, + "num_tokens": 5949507.0, + "step": 196 + }, + { + "epoch": 0.016244743135152964, + "grad_norm": 3.885606288909912, + "learning_rate": 5.386095081066227e-08, + "loss": 1.1255, + "mean_token_accuracy": 0.7001742720603943, + "num_tokens": 5986069.0, + "step": 197 + }, + { + "epoch": 0.016327203760204502, + "grad_norm": 4.467042446136475, + "learning_rate": 5.413575158010442e-08, + "loss": 1.3992, + "mean_token_accuracy": 0.6268547773361206, + "num_tokens": 6018909.0, + "step": 198 + }, + { + "epoch": 0.01640966438525604, + "grad_norm": 4.295664310455322, + "learning_rate": 5.4410552349546575e-08, + "loss": 1.1961, + "mean_token_accuracy": 0.669953465461731, + "num_tokens": 6048697.0, + "step": 199 + }, + { + "epoch": 0.016492125010307578, + "grad_norm": 4.025097370147705, + "learning_rate": 5.468535311898873e-08, + "loss": 1.181, + "mean_token_accuracy": 0.6677866578102112, + "num_tokens": 6081338.0, + "step": 200 + }, + { + "epoch": 0.016574585635359115, + "grad_norm": 4.1666741371154785, + "learning_rate": 5.4960153888430885e-08, + "loss": 1.0697, + "mean_token_accuracy": 0.6797927618026733, + "num_tokens": 6110714.0, + "step": 201 + }, + { + "epoch": 0.016657046260410653, + "grad_norm": 4.209021091461182, + "learning_rate": 5.523495465787304e-08, + "loss": 1.3329, + "mean_token_accuracy": 0.6326000094413757, + "num_tokens": 6144887.0, + "step": 202 + }, + { + "epoch": 0.01673950688546219, + "grad_norm": 4.050515651702881, + "learning_rate": 5.5509755427315195e-08, + "loss": 1.1572, + "mean_token_accuracy": 0.676637589931488, + "num_tokens": 6177829.0, + "step": 203 + }, + { + "epoch": 0.01682196751051373, + "grad_norm": 4.419314384460449, + "learning_rate": 5.578455619675735e-08, + "loss": 1.2924, + "mean_token_accuracy": 0.6473572850227356, + "num_tokens": 6208372.0, + "step": 204 + }, + { + "epoch": 0.016904428135565266, + "grad_norm": 3.8562264442443848, + "learning_rate": 5.6059356966199505e-08, + "loss": 1.1897, + "mean_token_accuracy": 0.6613662242889404, + "num_tokens": 6243332.0, + "step": 205 + }, + { + "epoch": 0.016986888760616804, + "grad_norm": 4.011408805847168, + "learning_rate": 5.6334157735641657e-08, + "loss": 1.1242, + "mean_token_accuracy": 0.6879323720932007, + "num_tokens": 6275980.0, + "step": 206 + }, + { + "epoch": 0.017069349385668342, + "grad_norm": 4.130151748657227, + "learning_rate": 5.6608958505083815e-08, + "loss": 0.997, + "mean_token_accuracy": 0.7196018099784851, + "num_tokens": 6305726.0, + "step": 207 + }, + { + "epoch": 0.01715181001071988, + "grad_norm": 4.411670684814453, + "learning_rate": 5.6883759274525967e-08, + "loss": 1.1465, + "mean_token_accuracy": 0.6854462027549744, + "num_tokens": 6332364.0, + "step": 208 + }, + { + "epoch": 0.017234270635771418, + "grad_norm": 3.6695992946624756, + "learning_rate": 5.7158560043968125e-08, + "loss": 1.1724, + "mean_token_accuracy": 0.6834310293197632, + "num_tokens": 6368055.0, + "step": 209 + }, + { + "epoch": 0.017316731260822955, + "grad_norm": 3.5931901931762695, + "learning_rate": 5.7433360813410277e-08, + "loss": 1.0882, + "mean_token_accuracy": 0.6997263431549072, + "num_tokens": 6403591.0, + "step": 210 + }, + { + "epoch": 0.017399191885874497, + "grad_norm": 3.5179240703582764, + "learning_rate": 5.770816158285243e-08, + "loss": 0.9336, + "mean_token_accuracy": 0.7399542331695557, + "num_tokens": 6439804.0, + "step": 211 + }, + { + "epoch": 0.017481652510926034, + "grad_norm": 4.110238075256348, + "learning_rate": 5.7982962352294587e-08, + "loss": 1.0867, + "mean_token_accuracy": 0.6995066404342651, + "num_tokens": 6471667.0, + "step": 212 + }, + { + "epoch": 0.017564113135977572, + "grad_norm": 4.1698431968688965, + "learning_rate": 5.825776312173674e-08, + "loss": 1.0508, + "mean_token_accuracy": 0.702564001083374, + "num_tokens": 6500425.0, + "step": 213 + }, + { + "epoch": 0.01764657376102911, + "grad_norm": 4.50399112701416, + "learning_rate": 5.85325638911789e-08, + "loss": 1.2021, + "mean_token_accuracy": 0.6695556640625, + "num_tokens": 6528024.0, + "step": 214 + }, + { + "epoch": 0.017729034386080648, + "grad_norm": 4.6144890785217285, + "learning_rate": 5.880736466062105e-08, + "loss": 1.2405, + "mean_token_accuracy": 0.6599380970001221, + "num_tokens": 6557748.0, + "step": 215 + }, + { + "epoch": 0.017811495011132186, + "grad_norm": 4.120750427246094, + "learning_rate": 5.908216543006321e-08, + "loss": 1.1117, + "mean_token_accuracy": 0.690737783908844, + "num_tokens": 6586844.0, + "step": 216 + }, + { + "epoch": 0.017893955636183723, + "grad_norm": 3.9217355251312256, + "learning_rate": 5.935696619950536e-08, + "loss": 1.018, + "mean_token_accuracy": 0.7071444988250732, + "num_tokens": 6614681.0, + "step": 217 + }, + { + "epoch": 0.01797641626123526, + "grad_norm": 4.281632900238037, + "learning_rate": 5.963176696894751e-08, + "loss": 1.1468, + "mean_token_accuracy": 0.6758751273155212, + "num_tokens": 6642814.0, + "step": 218 + }, + { + "epoch": 0.0180588768862868, + "grad_norm": 4.271881103515625, + "learning_rate": 5.990656773838967e-08, + "loss": 1.1245, + "mean_token_accuracy": 0.6787959337234497, + "num_tokens": 6667167.0, + "step": 219 + }, + { + "epoch": 0.018141337511338337, + "grad_norm": 3.9239273071289062, + "learning_rate": 6.018136850783183e-08, + "loss": 1.1781, + "mean_token_accuracy": 0.6683339476585388, + "num_tokens": 6696431.0, + "step": 220 + }, + { + "epoch": 0.018223798136389874, + "grad_norm": 3.62180233001709, + "learning_rate": 6.045616927727397e-08, + "loss": 1.0471, + "mean_token_accuracy": 0.7036471366882324, + "num_tokens": 6725957.0, + "step": 221 + }, + { + "epoch": 0.018306258761441412, + "grad_norm": 3.959604501724243, + "learning_rate": 6.073097004671613e-08, + "loss": 1.1065, + "mean_token_accuracy": 0.6970102787017822, + "num_tokens": 6758334.0, + "step": 222 + }, + { + "epoch": 0.01838871938649295, + "grad_norm": 4.5127692222595215, + "learning_rate": 6.100577081615829e-08, + "loss": 1.3037, + "mean_token_accuracy": 0.6406650543212891, + "num_tokens": 6784941.0, + "step": 223 + }, + { + "epoch": 0.018471180011544488, + "grad_norm": 4.1722798347473145, + "learning_rate": 6.128057158560043e-08, + "loss": 1.2167, + "mean_token_accuracy": 0.6602485775947571, + "num_tokens": 6815298.0, + "step": 224 + }, + { + "epoch": 0.018553640636596026, + "grad_norm": 3.7816543579101562, + "learning_rate": 6.155537235504259e-08, + "loss": 1.1837, + "mean_token_accuracy": 0.6679960489273071, + "num_tokens": 6850413.0, + "step": 225 + }, + { + "epoch": 0.018636101261647563, + "grad_norm": 4.21476411819458, + "learning_rate": 6.183017312448475e-08, + "loss": 1.2751, + "mean_token_accuracy": 0.6564576625823975, + "num_tokens": 6877398.0, + "step": 226 + }, + { + "epoch": 0.0187185618866991, + "grad_norm": 3.8235509395599365, + "learning_rate": 6.210497389392691e-08, + "loss": 1.0487, + "mean_token_accuracy": 0.7046782970428467, + "num_tokens": 6911787.0, + "step": 227 + }, + { + "epoch": 0.01880102251175064, + "grad_norm": 4.1496806144714355, + "learning_rate": 6.237977466336905e-08, + "loss": 1.28, + "mean_token_accuracy": 0.6408675909042358, + "num_tokens": 6940099.0, + "step": 228 + }, + { + "epoch": 0.018883483136802177, + "grad_norm": 4.23558235168457, + "learning_rate": 6.26545754328112e-08, + "loss": 1.2799, + "mean_token_accuracy": 0.6512067914009094, + "num_tokens": 6968056.0, + "step": 229 + }, + { + "epoch": 0.018965943761853715, + "grad_norm": 4.255162715911865, + "learning_rate": 6.292937620225336e-08, + "loss": 1.287, + "mean_token_accuracy": 0.652564287185669, + "num_tokens": 6995933.0, + "step": 230 + }, + { + "epoch": 0.019048404386905252, + "grad_norm": 3.8184962272644043, + "learning_rate": 6.320417697169552e-08, + "loss": 0.9737, + "mean_token_accuracy": 0.7292981147766113, + "num_tokens": 7027539.0, + "step": 231 + }, + { + "epoch": 0.01913086501195679, + "grad_norm": 4.1403679847717285, + "learning_rate": 6.347897774113767e-08, + "loss": 1.2235, + "mean_token_accuracy": 0.6609923839569092, + "num_tokens": 7056986.0, + "step": 232 + }, + { + "epoch": 0.019213325637008328, + "grad_norm": 3.7161502838134766, + "learning_rate": 6.375377851057982e-08, + "loss": 1.212, + "mean_token_accuracy": 0.6596711874008179, + "num_tokens": 7091532.0, + "step": 233 + }, + { + "epoch": 0.019295786262059866, + "grad_norm": 4.0466485023498535, + "learning_rate": 6.402857928002198e-08, + "loss": 1.1004, + "mean_token_accuracy": 0.6961380839347839, + "num_tokens": 7122218.0, + "step": 234 + }, + { + "epoch": 0.019378246887111403, + "grad_norm": 3.863215446472168, + "learning_rate": 6.430338004946414e-08, + "loss": 1.1128, + "mean_token_accuracy": 0.6841765642166138, + "num_tokens": 7149629.0, + "step": 235 + }, + { + "epoch": 0.01946070751216294, + "grad_norm": 3.993668556213379, + "learning_rate": 6.457818081890628e-08, + "loss": 1.1324, + "mean_token_accuracy": 0.6793721914291382, + "num_tokens": 7177340.0, + "step": 236 + }, + { + "epoch": 0.01954316813721448, + "grad_norm": 3.708120822906494, + "learning_rate": 6.485298158834844e-08, + "loss": 0.9866, + "mean_token_accuracy": 0.718280553817749, + "num_tokens": 7205610.0, + "step": 237 + }, + { + "epoch": 0.019625628762266017, + "grad_norm": 3.3811028003692627, + "learning_rate": 6.51277823577906e-08, + "loss": 0.9741, + "mean_token_accuracy": 0.7132773399353027, + "num_tokens": 7241448.0, + "step": 238 + }, + { + "epoch": 0.019708089387317555, + "grad_norm": 4.047371864318848, + "learning_rate": 6.540258312723276e-08, + "loss": 1.1753, + "mean_token_accuracy": 0.6717459559440613, + "num_tokens": 7269775.0, + "step": 239 + }, + { + "epoch": 0.019790550012369092, + "grad_norm": 3.560676097869873, + "learning_rate": 6.56773838966749e-08, + "loss": 1.1645, + "mean_token_accuracy": 0.6807835102081299, + "num_tokens": 7301140.0, + "step": 240 + }, + { + "epoch": 0.01987301063742063, + "grad_norm": 3.5871541500091553, + "learning_rate": 6.595218466611706e-08, + "loss": 1.3134, + "mean_token_accuracy": 0.6470252275466919, + "num_tokens": 7338656.0, + "step": 241 + }, + { + "epoch": 0.019955471262472168, + "grad_norm": 3.8842148780822754, + "learning_rate": 6.622698543555922e-08, + "loss": 1.1081, + "mean_token_accuracy": 0.6825306415557861, + "num_tokens": 7368329.0, + "step": 242 + }, + { + "epoch": 0.020037931887523706, + "grad_norm": 4.492695331573486, + "learning_rate": 6.650178620500136e-08, + "loss": 1.1319, + "mean_token_accuracy": 0.6779546141624451, + "num_tokens": 7397557.0, + "step": 243 + }, + { + "epoch": 0.020120392512575247, + "grad_norm": 5.318395614624023, + "learning_rate": 6.677658697444352e-08, + "loss": 1.2829, + "mean_token_accuracy": 0.6420520544052124, + "num_tokens": 7421549.0, + "step": 244 + }, + { + "epoch": 0.020202853137626785, + "grad_norm": 4.064792156219482, + "learning_rate": 6.705138774388568e-08, + "loss": 1.2484, + "mean_token_accuracy": 0.6509985327720642, + "num_tokens": 7450731.0, + "step": 245 + }, + { + "epoch": 0.020285313762678323, + "grad_norm": 4.088617324829102, + "learning_rate": 6.732618851332782e-08, + "loss": 1.1808, + "mean_token_accuracy": 0.6592028141021729, + "num_tokens": 7478251.0, + "step": 246 + }, + { + "epoch": 0.02036777438772986, + "grad_norm": 3.297945737838745, + "learning_rate": 6.760098928276998e-08, + "loss": 1.2087, + "mean_token_accuracy": 0.6640618443489075, + "num_tokens": 7517918.0, + "step": 247 + }, + { + "epoch": 0.020450235012781398, + "grad_norm": 3.973628520965576, + "learning_rate": 6.787579005221214e-08, + "loss": 0.9152, + "mean_token_accuracy": 0.7287558317184448, + "num_tokens": 7543937.0, + "step": 248 + }, + { + "epoch": 0.020532695637832936, + "grad_norm": 4.104936122894287, + "learning_rate": 6.81505908216543e-08, + "loss": 1.2932, + "mean_token_accuracy": 0.6487190127372742, + "num_tokens": 7577192.0, + "step": 249 + }, + { + "epoch": 0.020615156262884474, + "grad_norm": 4.4902215003967285, + "learning_rate": 6.842539159109644e-08, + "loss": 1.2794, + "mean_token_accuracy": 0.6390405297279358, + "num_tokens": 7605481.0, + "step": 250 + }, + { + "epoch": 0.02069761688793601, + "grad_norm": 4.016014099121094, + "learning_rate": 6.87001923605386e-08, + "loss": 1.1738, + "mean_token_accuracy": 0.6798156499862671, + "num_tokens": 7637067.0, + "step": 251 + }, + { + "epoch": 0.02078007751298755, + "grad_norm": 3.7815186977386475, + "learning_rate": 6.897499312998076e-08, + "loss": 1.1855, + "mean_token_accuracy": 0.6709626913070679, + "num_tokens": 7665738.0, + "step": 252 + }, + { + "epoch": 0.020862538138039087, + "grad_norm": 4.041879177093506, + "learning_rate": 6.92497938994229e-08, + "loss": 1.2111, + "mean_token_accuracy": 0.6537415385246277, + "num_tokens": 7700497.0, + "step": 253 + }, + { + "epoch": 0.020944998763090625, + "grad_norm": 3.7669339179992676, + "learning_rate": 6.952459466886506e-08, + "loss": 0.9706, + "mean_token_accuracy": 0.7144832015037537, + "num_tokens": 7735578.0, + "step": 254 + }, + { + "epoch": 0.021027459388142163, + "grad_norm": 3.98296856880188, + "learning_rate": 6.979939543830722e-08, + "loss": 1.0701, + "mean_token_accuracy": 0.6922503113746643, + "num_tokens": 7762550.0, + "step": 255 + }, + { + "epoch": 0.0211099200131937, + "grad_norm": 4.12508487701416, + "learning_rate": 7.007419620774938e-08, + "loss": 1.2727, + "mean_token_accuracy": 0.6470746397972107, + "num_tokens": 7793375.0, + "step": 256 + }, + { + "epoch": 0.021192380638245238, + "grad_norm": 3.642504930496216, + "learning_rate": 7.034899697719153e-08, + "loss": 1.2293, + "mean_token_accuracy": 0.6596113443374634, + "num_tokens": 7825662.0, + "step": 257 + }, + { + "epoch": 0.021274841263296776, + "grad_norm": 3.5116801261901855, + "learning_rate": 7.062379774663368e-08, + "loss": 1.1338, + "mean_token_accuracy": 0.6879931688308716, + "num_tokens": 7857923.0, + "step": 258 + }, + { + "epoch": 0.021357301888348314, + "grad_norm": 3.6259665489196777, + "learning_rate": 7.089859851607584e-08, + "loss": 0.9866, + "mean_token_accuracy": 0.7208870649337769, + "num_tokens": 7890112.0, + "step": 259 + }, + { + "epoch": 0.02143976251339985, + "grad_norm": 5.941770076751709, + "learning_rate": 7.117339928551799e-08, + "loss": 1.2692, + "mean_token_accuracy": 0.6460452079772949, + "num_tokens": 7909963.0, + "step": 260 + }, + { + "epoch": 0.02152222313845139, + "grad_norm": 3.881340265274048, + "learning_rate": 7.144820005496015e-08, + "loss": 1.1176, + "mean_token_accuracy": 0.67945396900177, + "num_tokens": 7946141.0, + "step": 261 + }, + { + "epoch": 0.021604683763502927, + "grad_norm": 3.931180000305176, + "learning_rate": 7.17230008244023e-08, + "loss": 1.0735, + "mean_token_accuracy": 0.7053285837173462, + "num_tokens": 7975824.0, + "step": 262 + }, + { + "epoch": 0.021687144388554465, + "grad_norm": 3.8988964557647705, + "learning_rate": 7.199780159384446e-08, + "loss": 1.095, + "mean_token_accuracy": 0.6827343106269836, + "num_tokens": 8004456.0, + "step": 263 + }, + { + "epoch": 0.021769605013606003, + "grad_norm": 3.765608310699463, + "learning_rate": 7.227260236328661e-08, + "loss": 1.2393, + "mean_token_accuracy": 0.6515863537788391, + "num_tokens": 8044137.0, + "step": 264 + }, + { + "epoch": 0.02185206563865754, + "grad_norm": 4.150010108947754, + "learning_rate": 7.254740313272877e-08, + "loss": 1.1476, + "mean_token_accuracy": 0.6733556389808655, + "num_tokens": 8076729.0, + "step": 265 + }, + { + "epoch": 0.021934526263709078, + "grad_norm": 4.913898944854736, + "learning_rate": 7.282220390217092e-08, + "loss": 1.0799, + "mean_token_accuracy": 0.6890175938606262, + "num_tokens": 8100968.0, + "step": 266 + }, + { + "epoch": 0.022016986888760616, + "grad_norm": 4.217159748077393, + "learning_rate": 7.309700467161307e-08, + "loss": 1.0169, + "mean_token_accuracy": 0.7090623378753662, + "num_tokens": 8136139.0, + "step": 267 + }, + { + "epoch": 0.022099447513812154, + "grad_norm": 3.50313138961792, + "learning_rate": 7.337180544105523e-08, + "loss": 1.1492, + "mean_token_accuracy": 0.6772158741950989, + "num_tokens": 8167522.0, + "step": 268 + }, + { + "epoch": 0.02218190813886369, + "grad_norm": 4.687294960021973, + "learning_rate": 7.364660621049739e-08, + "loss": 1.3033, + "mean_token_accuracy": 0.6420485973358154, + "num_tokens": 8197270.0, + "step": 269 + }, + { + "epoch": 0.02226436876391523, + "grad_norm": 5.3743977546691895, + "learning_rate": 7.392140697993953e-08, + "loss": 1.2101, + "mean_token_accuracy": 0.655724287033081, + "num_tokens": 8236653.0, + "step": 270 + }, + { + "epoch": 0.022346829388966767, + "grad_norm": 4.280730247497559, + "learning_rate": 7.419620774938169e-08, + "loss": 1.0295, + "mean_token_accuracy": 0.7063848972320557, + "num_tokens": 8264357.0, + "step": 271 + }, + { + "epoch": 0.022429290014018305, + "grad_norm": 3.6797773838043213, + "learning_rate": 7.447100851882385e-08, + "loss": 0.9673, + "mean_token_accuracy": 0.7187896966934204, + "num_tokens": 8295216.0, + "step": 272 + }, + { + "epoch": 0.022511750639069843, + "grad_norm": 3.8935415744781494, + "learning_rate": 7.4745809288266e-08, + "loss": 1.0687, + "mean_token_accuracy": 0.6938145160675049, + "num_tokens": 8333470.0, + "step": 273 + }, + { + "epoch": 0.02259421126412138, + "grad_norm": 4.086860179901123, + "learning_rate": 7.502061005770815e-08, + "loss": 1.1933, + "mean_token_accuracy": 0.6657329797744751, + "num_tokens": 8366179.0, + "step": 274 + }, + { + "epoch": 0.022676671889172918, + "grad_norm": 4.194552898406982, + "learning_rate": 7.529541082715031e-08, + "loss": 1.0769, + "mean_token_accuracy": 0.6924054622650146, + "num_tokens": 8399123.0, + "step": 275 + }, + { + "epoch": 0.02275913251422446, + "grad_norm": 7.151051998138428, + "learning_rate": 7.557021159659247e-08, + "loss": 1.1932, + "mean_token_accuracy": 0.6612610816955566, + "num_tokens": 8421890.0, + "step": 276 + }, + { + "epoch": 0.022841593139275997, + "grad_norm": 3.834172010421753, + "learning_rate": 7.584501236603461e-08, + "loss": 1.0815, + "mean_token_accuracy": 0.6847587823867798, + "num_tokens": 8457938.0, + "step": 277 + }, + { + "epoch": 0.022924053764327535, + "grad_norm": 5.8365797996521, + "learning_rate": 7.611981313547677e-08, + "loss": 1.0742, + "mean_token_accuracy": 0.688838005065918, + "num_tokens": 8484334.0, + "step": 278 + }, + { + "epoch": 0.023006514389379073, + "grad_norm": 4.641779899597168, + "learning_rate": 7.639461390491893e-08, + "loss": 1.0737, + "mean_token_accuracy": 0.6937015056610107, + "num_tokens": 8518427.0, + "step": 279 + }, + { + "epoch": 0.02308897501443061, + "grad_norm": 4.478398323059082, + "learning_rate": 7.666941467436109e-08, + "loss": 1.075, + "mean_token_accuracy": 0.6806389093399048, + "num_tokens": 8543704.0, + "step": 280 + }, + { + "epoch": 0.02317143563948215, + "grad_norm": 3.7042934894561768, + "learning_rate": 7.694421544380323e-08, + "loss": 1.0013, + "mean_token_accuracy": 0.7119972109794617, + "num_tokens": 8580104.0, + "step": 281 + }, + { + "epoch": 0.023253896264533686, + "grad_norm": 5.431618690490723, + "learning_rate": 7.721901621324539e-08, + "loss": 1.0202, + "mean_token_accuracy": 0.6943958401679993, + "num_tokens": 8609838.0, + "step": 282 + }, + { + "epoch": 0.023336356889585224, + "grad_norm": 4.600614070892334, + "learning_rate": 7.749381698268755e-08, + "loss": 1.0319, + "mean_token_accuracy": 0.7078181505203247, + "num_tokens": 8640535.0, + "step": 283 + }, + { + "epoch": 0.023418817514636762, + "grad_norm": 4.452020645141602, + "learning_rate": 7.77686177521297e-08, + "loss": 1.0123, + "mean_token_accuracy": 0.7114675045013428, + "num_tokens": 8668699.0, + "step": 284 + }, + { + "epoch": 0.0235012781396883, + "grad_norm": 4.531774044036865, + "learning_rate": 7.804341852157185e-08, + "loss": 1.0628, + "mean_token_accuracy": 0.6953094601631165, + "num_tokens": 8705005.0, + "step": 285 + }, + { + "epoch": 0.023583738764739837, + "grad_norm": 4.915223598480225, + "learning_rate": 7.831821929101401e-08, + "loss": 1.0528, + "mean_token_accuracy": 0.7006457448005676, + "num_tokens": 8737145.0, + "step": 286 + }, + { + "epoch": 0.023666199389791375, + "grad_norm": 4.498976230621338, + "learning_rate": 7.859302006045617e-08, + "loss": 1.1071, + "mean_token_accuracy": 0.6904487609863281, + "num_tokens": 8771502.0, + "step": 287 + }, + { + "epoch": 0.023748660014842913, + "grad_norm": 4.278553485870361, + "learning_rate": 7.886782082989832e-08, + "loss": 0.9737, + "mean_token_accuracy": 0.7171822190284729, + "num_tokens": 8799758.0, + "step": 288 + }, + { + "epoch": 0.02383112063989445, + "grad_norm": 4.47084903717041, + "learning_rate": 7.914262159934047e-08, + "loss": 1.0314, + "mean_token_accuracy": 0.697587251663208, + "num_tokens": 8829735.0, + "step": 289 + }, + { + "epoch": 0.02391358126494599, + "grad_norm": 5.907037258148193, + "learning_rate": 7.941742236878263e-08, + "loss": 1.1424, + "mean_token_accuracy": 0.6715341806411743, + "num_tokens": 8855424.0, + "step": 290 + }, + { + "epoch": 0.023996041889997526, + "grad_norm": 4.871565818786621, + "learning_rate": 7.969222313822478e-08, + "loss": 1.0805, + "mean_token_accuracy": 0.6841116547584534, + "num_tokens": 8885257.0, + "step": 291 + }, + { + "epoch": 0.024078502515049064, + "grad_norm": 4.526889324188232, + "learning_rate": 7.996702390766694e-08, + "loss": 1.056, + "mean_token_accuracy": 0.6999532580375671, + "num_tokens": 8913077.0, + "step": 292 + }, + { + "epoch": 0.024160963140100602, + "grad_norm": 5.950772285461426, + "learning_rate": 8.02418246771091e-08, + "loss": 1.3972, + "mean_token_accuracy": 0.6148780584335327, + "num_tokens": 8945031.0, + "step": 293 + }, + { + "epoch": 0.02424342376515214, + "grad_norm": 5.771305084228516, + "learning_rate": 8.051662544655124e-08, + "loss": 0.9638, + "mean_token_accuracy": 0.7248409986495972, + "num_tokens": 8978158.0, + "step": 294 + }, + { + "epoch": 0.024325884390203677, + "grad_norm": 5.671107292175293, + "learning_rate": 8.07914262159934e-08, + "loss": 1.2364, + "mean_token_accuracy": 0.6550912857055664, + "num_tokens": 9006122.0, + "step": 295 + }, + { + "epoch": 0.024408345015255215, + "grad_norm": 6.016450881958008, + "learning_rate": 8.106622698543556e-08, + "loss": 1.0792, + "mean_token_accuracy": 0.6883476972579956, + "num_tokens": 9036212.0, + "step": 296 + }, + { + "epoch": 0.024490805640306753, + "grad_norm": 4.349691867828369, + "learning_rate": 8.134102775487771e-08, + "loss": 0.9658, + "mean_token_accuracy": 0.7233712673187256, + "num_tokens": 9070827.0, + "step": 297 + }, + { + "epoch": 0.02457326626535829, + "grad_norm": 4.442138671875, + "learning_rate": 8.161582852431986e-08, + "loss": 1.0676, + "mean_token_accuracy": 0.6962677240371704, + "num_tokens": 9103584.0, + "step": 298 + }, + { + "epoch": 0.02465572689040983, + "grad_norm": 5.826080799102783, + "learning_rate": 8.189062929376202e-08, + "loss": 1.0233, + "mean_token_accuracy": 0.699803352355957, + "num_tokens": 9133498.0, + "step": 299 + }, + { + "epoch": 0.024738187515461366, + "grad_norm": 5.3808465003967285, + "learning_rate": 8.216543006320418e-08, + "loss": 1.0315, + "mean_token_accuracy": 0.7066638469696045, + "num_tokens": 9164611.0, + "step": 300 + }, + { + "epoch": 0.024820648140512904, + "grad_norm": 5.166574478149414, + "learning_rate": 8.244023083264632e-08, + "loss": 1.168, + "mean_token_accuracy": 0.6666162610054016, + "num_tokens": 9191501.0, + "step": 301 + }, + { + "epoch": 0.024903108765564442, + "grad_norm": 4.921683311462402, + "learning_rate": 8.271503160208848e-08, + "loss": 1.0402, + "mean_token_accuracy": 0.7018457651138306, + "num_tokens": 9222012.0, + "step": 302 + }, + { + "epoch": 0.02498556939061598, + "grad_norm": 4.799124717712402, + "learning_rate": 8.298983237153064e-08, + "loss": 1.0203, + "mean_token_accuracy": 0.6947497129440308, + "num_tokens": 9248777.0, + "step": 303 + }, + { + "epoch": 0.025068030015667517, + "grad_norm": 4.664659023284912, + "learning_rate": 8.32646331409728e-08, + "loss": 1.201, + "mean_token_accuracy": 0.6631577014923096, + "num_tokens": 9279245.0, + "step": 304 + }, + { + "epoch": 0.025150490640719055, + "grad_norm": 4.945487022399902, + "learning_rate": 8.353943391041494e-08, + "loss": 1.0697, + "mean_token_accuracy": 0.6789401769638062, + "num_tokens": 9310475.0, + "step": 305 + }, + { + "epoch": 0.025232951265770593, + "grad_norm": 5.661356449127197, + "learning_rate": 8.38142346798571e-08, + "loss": 1.1972, + "mean_token_accuracy": 0.6579902172088623, + "num_tokens": 9337329.0, + "step": 306 + }, + { + "epoch": 0.02531541189082213, + "grad_norm": 4.785332679748535, + "learning_rate": 8.408903544929926e-08, + "loss": 1.1466, + "mean_token_accuracy": 0.6775115728378296, + "num_tokens": 9370566.0, + "step": 307 + }, + { + "epoch": 0.02539787251587367, + "grad_norm": 5.989325523376465, + "learning_rate": 8.43638362187414e-08, + "loss": 1.148, + "mean_token_accuracy": 0.6827669143676758, + "num_tokens": 9395831.0, + "step": 308 + }, + { + "epoch": 0.02548033314092521, + "grad_norm": 5.122068405151367, + "learning_rate": 8.463863698818356e-08, + "loss": 1.1799, + "mean_token_accuracy": 0.6652275919914246, + "num_tokens": 9421821.0, + "step": 309 + }, + { + "epoch": 0.025562793765976748, + "grad_norm": 4.78012752532959, + "learning_rate": 8.491343775762572e-08, + "loss": 1.1236, + "mean_token_accuracy": 0.6867064237594604, + "num_tokens": 9451411.0, + "step": 310 + }, + { + "epoch": 0.025645254391028285, + "grad_norm": 3.9747302532196045, + "learning_rate": 8.518823852706788e-08, + "loss": 1.0635, + "mean_token_accuracy": 0.6942881345748901, + "num_tokens": 9479939.0, + "step": 311 + }, + { + "epoch": 0.025727715016079823, + "grad_norm": 3.945019006729126, + "learning_rate": 8.546303929651002e-08, + "loss": 1.027, + "mean_token_accuracy": 0.7010728120803833, + "num_tokens": 9511386.0, + "step": 312 + }, + { + "epoch": 0.02581017564113136, + "grad_norm": 4.833399772644043, + "learning_rate": 8.573784006595218e-08, + "loss": 1.0053, + "mean_token_accuracy": 0.7106674909591675, + "num_tokens": 9537026.0, + "step": 313 + }, + { + "epoch": 0.0258926362661829, + "grad_norm": 4.586074352264404, + "learning_rate": 8.601264083539434e-08, + "loss": 1.0542, + "mean_token_accuracy": 0.6997315883636475, + "num_tokens": 9564964.0, + "step": 314 + }, + { + "epoch": 0.025975096891234437, + "grad_norm": 6.025402069091797, + "learning_rate": 8.628744160483648e-08, + "loss": 1.0526, + "mean_token_accuracy": 0.6977773308753967, + "num_tokens": 9592291.0, + "step": 315 + }, + { + "epoch": 0.026057557516285974, + "grad_norm": 7.686981678009033, + "learning_rate": 8.656224237427864e-08, + "loss": 0.9573, + "mean_token_accuracy": 0.7238781452178955, + "num_tokens": 9627330.0, + "step": 316 + }, + { + "epoch": 0.026140018141337512, + "grad_norm": 7.028507709503174, + "learning_rate": 8.68370431437208e-08, + "loss": 1.1198, + "mean_token_accuracy": 0.678891658782959, + "num_tokens": 9659543.0, + "step": 317 + }, + { + "epoch": 0.02622247876638905, + "grad_norm": 4.401634693145752, + "learning_rate": 8.711184391316295e-08, + "loss": 1.003, + "mean_token_accuracy": 0.706773042678833, + "num_tokens": 9698967.0, + "step": 318 + }, + { + "epoch": 0.026304939391440588, + "grad_norm": 6.839646816253662, + "learning_rate": 8.73866446826051e-08, + "loss": 1.0989, + "mean_token_accuracy": 0.6967049837112427, + "num_tokens": 9728011.0, + "step": 319 + }, + { + "epoch": 0.026387400016492125, + "grad_norm": 5.201809406280518, + "learning_rate": 8.766144545204726e-08, + "loss": 1.1852, + "mean_token_accuracy": 0.6653717756271362, + "num_tokens": 9762449.0, + "step": 320 + }, + { + "epoch": 0.026469860641543663, + "grad_norm": 4.96702241897583, + "learning_rate": 8.793624622148942e-08, + "loss": 1.0892, + "mean_token_accuracy": 0.6888754367828369, + "num_tokens": 9795176.0, + "step": 321 + }, + { + "epoch": 0.0265523212665952, + "grad_norm": 4.76415491104126, + "learning_rate": 8.821104699093157e-08, + "loss": 0.9903, + "mean_token_accuracy": 0.7168942093849182, + "num_tokens": 9831435.0, + "step": 322 + }, + { + "epoch": 0.02663478189164674, + "grad_norm": 4.1059489250183105, + "learning_rate": 8.848584776037372e-08, + "loss": 1.0217, + "mean_token_accuracy": 0.7031457424163818, + "num_tokens": 9862527.0, + "step": 323 + }, + { + "epoch": 0.026717242516698277, + "grad_norm": 6.0018534660339355, + "learning_rate": 8.876064852981588e-08, + "loss": 1.0819, + "mean_token_accuracy": 0.6846128702163696, + "num_tokens": 9896686.0, + "step": 324 + }, + { + "epoch": 0.026799703141749814, + "grad_norm": 6.013288974761963, + "learning_rate": 8.903544929925803e-08, + "loss": 0.992, + "mean_token_accuracy": 0.7157648801803589, + "num_tokens": 9931400.0, + "step": 325 + }, + { + "epoch": 0.026882163766801352, + "grad_norm": 6.299633979797363, + "learning_rate": 8.931025006870019e-08, + "loss": 1.1299, + "mean_token_accuracy": 0.673222541809082, + "num_tokens": 9959209.0, + "step": 326 + }, + { + "epoch": 0.02696462439185289, + "grad_norm": 5.093557834625244, + "learning_rate": 8.958505083814234e-08, + "loss": 1.176, + "mean_token_accuracy": 0.6647504568099976, + "num_tokens": 9987090.0, + "step": 327 + }, + { + "epoch": 0.027047085016904428, + "grad_norm": 4.241351127624512, + "learning_rate": 8.98598516075845e-08, + "loss": 0.9717, + "mean_token_accuracy": 0.7234897613525391, + "num_tokens": 10021425.0, + "step": 328 + }, + { + "epoch": 0.027129545641955966, + "grad_norm": 5.905039310455322, + "learning_rate": 9.013465237702665e-08, + "loss": 1.0876, + "mean_token_accuracy": 0.681319534778595, + "num_tokens": 10049677.0, + "step": 329 + }, + { + "epoch": 0.027212006267007503, + "grad_norm": 5.502197265625, + "learning_rate": 9.04094531464688e-08, + "loss": 1.1406, + "mean_token_accuracy": 0.6700173020362854, + "num_tokens": 10072862.0, + "step": 330 + }, + { + "epoch": 0.02729446689205904, + "grad_norm": 5.349047660827637, + "learning_rate": 9.068425391591096e-08, + "loss": 1.0517, + "mean_token_accuracy": 0.6985857486724854, + "num_tokens": 10107483.0, + "step": 331 + }, + { + "epoch": 0.02737692751711058, + "grad_norm": 4.411892414093018, + "learning_rate": 9.095905468535311e-08, + "loss": 0.8856, + "mean_token_accuracy": 0.7386624217033386, + "num_tokens": 10145221.0, + "step": 332 + }, + { + "epoch": 0.027459388142162117, + "grad_norm": 5.408709526062012, + "learning_rate": 9.123385545479527e-08, + "loss": 1.084, + "mean_token_accuracy": 0.6932909488677979, + "num_tokens": 10175912.0, + "step": 333 + }, + { + "epoch": 0.027541848767213654, + "grad_norm": 4.228156566619873, + "learning_rate": 9.150865622423743e-08, + "loss": 1.258, + "mean_token_accuracy": 0.6472073793411255, + "num_tokens": 10213434.0, + "step": 334 + }, + { + "epoch": 0.027624309392265192, + "grad_norm": 7.474115371704102, + "learning_rate": 9.178345699367958e-08, + "loss": 1.0523, + "mean_token_accuracy": 0.6971520781517029, + "num_tokens": 10236594.0, + "step": 335 + }, + { + "epoch": 0.02770677001731673, + "grad_norm": 5.027649402618408, + "learning_rate": 9.205825776312173e-08, + "loss": 1.1668, + "mean_token_accuracy": 0.6781877279281616, + "num_tokens": 10269375.0, + "step": 336 + }, + { + "epoch": 0.027789230642368268, + "grad_norm": 8.208491325378418, + "learning_rate": 9.233305853256389e-08, + "loss": 1.1897, + "mean_token_accuracy": 0.6577918529510498, + "num_tokens": 10295844.0, + "step": 337 + }, + { + "epoch": 0.027871691267419806, + "grad_norm": 7.441732883453369, + "learning_rate": 9.260785930200605e-08, + "loss": 1.0792, + "mean_token_accuracy": 0.6890542507171631, + "num_tokens": 10328280.0, + "step": 338 + }, + { + "epoch": 0.027954151892471343, + "grad_norm": 5.444562911987305, + "learning_rate": 9.288266007144819e-08, + "loss": 1.0385, + "mean_token_accuracy": 0.710965633392334, + "num_tokens": 10353043.0, + "step": 339 + }, + { + "epoch": 0.02803661251752288, + "grad_norm": 4.6165666580200195, + "learning_rate": 9.315746084089035e-08, + "loss": 1.1514, + "mean_token_accuracy": 0.6721846461296082, + "num_tokens": 10388341.0, + "step": 340 + }, + { + "epoch": 0.028119073142574422, + "grad_norm": 5.903135299682617, + "learning_rate": 9.343226161033251e-08, + "loss": 0.9873, + "mean_token_accuracy": 0.7139780521392822, + "num_tokens": 10417850.0, + "step": 341 + }, + { + "epoch": 0.02820153376762596, + "grad_norm": 6.401676177978516, + "learning_rate": 9.370706237977465e-08, + "loss": 1.0388, + "mean_token_accuracy": 0.7054314017295837, + "num_tokens": 10444496.0, + "step": 342 + }, + { + "epoch": 0.028283994392677498, + "grad_norm": 4.5270514488220215, + "learning_rate": 9.398186314921681e-08, + "loss": 1.177, + "mean_token_accuracy": 0.6644185781478882, + "num_tokens": 10472516.0, + "step": 343 + }, + { + "epoch": 0.028366455017729036, + "grad_norm": 4.453205108642578, + "learning_rate": 9.425666391865897e-08, + "loss": 1.0502, + "mean_token_accuracy": 0.6931577324867249, + "num_tokens": 10502625.0, + "step": 344 + }, + { + "epoch": 0.028448915642780574, + "grad_norm": 5.299770355224609, + "learning_rate": 9.453146468810113e-08, + "loss": 1.0291, + "mean_token_accuracy": 0.7040730714797974, + "num_tokens": 10527699.0, + "step": 345 + }, + { + "epoch": 0.02853137626783211, + "grad_norm": 5.026134967803955, + "learning_rate": 9.480626545754327e-08, + "loss": 1.0462, + "mean_token_accuracy": 0.6961838006973267, + "num_tokens": 10562502.0, + "step": 346 + }, + { + "epoch": 0.02861383689288365, + "grad_norm": 6.637182712554932, + "learning_rate": 9.508106622698543e-08, + "loss": 1.0039, + "mean_token_accuracy": 0.7014429569244385, + "num_tokens": 10591809.0, + "step": 347 + }, + { + "epoch": 0.028696297517935187, + "grad_norm": 5.502130031585693, + "learning_rate": 9.535586699642759e-08, + "loss": 1.2237, + "mean_token_accuracy": 0.6589092016220093, + "num_tokens": 10624206.0, + "step": 348 + }, + { + "epoch": 0.028778758142986725, + "grad_norm": 4.931995868682861, + "learning_rate": 9.563066776586973e-08, + "loss": 1.0579, + "mean_token_accuracy": 0.7035638093948364, + "num_tokens": 10658909.0, + "step": 349 + }, + { + "epoch": 0.028861218768038262, + "grad_norm": 5.980095386505127, + "learning_rate": 9.590546853531189e-08, + "loss": 1.0919, + "mean_token_accuracy": 0.6861798763275146, + "num_tokens": 10683937.0, + "step": 350 + }, + { + "epoch": 0.0289436793930898, + "grad_norm": 5.412120819091797, + "learning_rate": 9.618026930475405e-08, + "loss": 0.8835, + "mean_token_accuracy": 0.7369436621665955, + "num_tokens": 10711302.0, + "step": 351 + }, + { + "epoch": 0.029026140018141338, + "grad_norm": 5.521242618560791, + "learning_rate": 9.645507007419621e-08, + "loss": 1.0413, + "mean_token_accuracy": 0.6962478756904602, + "num_tokens": 10742786.0, + "step": 352 + }, + { + "epoch": 0.029108600643192876, + "grad_norm": 4.957467079162598, + "learning_rate": 9.672987084363836e-08, + "loss": 1.1178, + "mean_token_accuracy": 0.6875641942024231, + "num_tokens": 10769170.0, + "step": 353 + }, + { + "epoch": 0.029191061268244414, + "grad_norm": 4.9935526847839355, + "learning_rate": 9.700467161308051e-08, + "loss": 1.0333, + "mean_token_accuracy": 0.699486255645752, + "num_tokens": 10795895.0, + "step": 354 + }, + { + "epoch": 0.02927352189329595, + "grad_norm": 5.258847713470459, + "learning_rate": 9.727947238252267e-08, + "loss": 1.0629, + "mean_token_accuracy": 0.6961636543273926, + "num_tokens": 10830256.0, + "step": 355 + }, + { + "epoch": 0.02935598251834749, + "grad_norm": 6.459052085876465, + "learning_rate": 9.755427315196482e-08, + "loss": 1.1026, + "mean_token_accuracy": 0.6758142113685608, + "num_tokens": 10855089.0, + "step": 356 + }, + { + "epoch": 0.029438443143399027, + "grad_norm": 6.398245334625244, + "learning_rate": 9.782907392140698e-08, + "loss": 1.2549, + "mean_token_accuracy": 0.6545199751853943, + "num_tokens": 10878802.0, + "step": 357 + }, + { + "epoch": 0.029520903768450565, + "grad_norm": 4.608811855316162, + "learning_rate": 9.810387469084913e-08, + "loss": 1.1195, + "mean_token_accuracy": 0.6763865947723389, + "num_tokens": 10911942.0, + "step": 358 + }, + { + "epoch": 0.029603364393502102, + "grad_norm": 4.465142726898193, + "learning_rate": 9.837867546029129e-08, + "loss": 1.1197, + "mean_token_accuracy": 0.6868385076522827, + "num_tokens": 10942283.0, + "step": 359 + }, + { + "epoch": 0.02968582501855364, + "grad_norm": 5.038747787475586, + "learning_rate": 9.865347622973344e-08, + "loss": 1.0246, + "mean_token_accuracy": 0.7055107355117798, + "num_tokens": 10973810.0, + "step": 360 + }, + { + "epoch": 0.029768285643605178, + "grad_norm": 6.255395412445068, + "learning_rate": 9.89282769991756e-08, + "loss": 1.0788, + "mean_token_accuracy": 0.6903204917907715, + "num_tokens": 11004717.0, + "step": 361 + }, + { + "epoch": 0.029850746268656716, + "grad_norm": 4.422749042510986, + "learning_rate": 9.920307776861775e-08, + "loss": 1.1572, + "mean_token_accuracy": 0.6771333813667297, + "num_tokens": 11036852.0, + "step": 362 + }, + { + "epoch": 0.029933206893708254, + "grad_norm": 4.686171531677246, + "learning_rate": 9.94778785380599e-08, + "loss": 0.82, + "mean_token_accuracy": 0.7574095726013184, + "num_tokens": 11066809.0, + "step": 363 + }, + { + "epoch": 0.03001566751875979, + "grad_norm": 5.210658073425293, + "learning_rate": 9.975267930750206e-08, + "loss": 1.0476, + "mean_token_accuracy": 0.6837996244430542, + "num_tokens": 11092618.0, + "step": 364 + }, + { + "epoch": 0.03009812814381133, + "grad_norm": 5.414539337158203, + "learning_rate": 1.0002748007694422e-07, + "loss": 1.1297, + "mean_token_accuracy": 0.6720084547996521, + "num_tokens": 11124604.0, + "step": 365 + }, + { + "epoch": 0.030180588768862867, + "grad_norm": 6.225357532501221, + "learning_rate": 1.0030228084638636e-07, + "loss": 1.0572, + "mean_token_accuracy": 0.6952575445175171, + "num_tokens": 11153954.0, + "step": 366 + }, + { + "epoch": 0.030263049393914405, + "grad_norm": 6.1720099449157715, + "learning_rate": 1.0057708161582852e-07, + "loss": 1.2024, + "mean_token_accuracy": 0.6674478054046631, + "num_tokens": 11185443.0, + "step": 367 + }, + { + "epoch": 0.030345510018965943, + "grad_norm": 6.781430244445801, + "learning_rate": 1.0085188238527068e-07, + "loss": 0.983, + "mean_token_accuracy": 0.7155237197875977, + "num_tokens": 11217168.0, + "step": 368 + }, + { + "epoch": 0.03042797064401748, + "grad_norm": 5.128620624542236, + "learning_rate": 1.0112668315471284e-07, + "loss": 1.0836, + "mean_token_accuracy": 0.6846739649772644, + "num_tokens": 11249742.0, + "step": 369 + }, + { + "epoch": 0.030510431269069018, + "grad_norm": 5.110124111175537, + "learning_rate": 1.0140148392415498e-07, + "loss": 1.0352, + "mean_token_accuracy": 0.6994126439094543, + "num_tokens": 11279161.0, + "step": 370 + }, + { + "epoch": 0.030592891894120556, + "grad_norm": 3.937368631362915, + "learning_rate": 1.0167628469359714e-07, + "loss": 1.1112, + "mean_token_accuracy": 0.6833343505859375, + "num_tokens": 11314989.0, + "step": 371 + }, + { + "epoch": 0.030675352519172094, + "grad_norm": 4.980045795440674, + "learning_rate": 1.019510854630393e-07, + "loss": 1.0187, + "mean_token_accuracy": 0.7037530541419983, + "num_tokens": 11342893.0, + "step": 372 + }, + { + "epoch": 0.030757813144223635, + "grad_norm": 5.885363578796387, + "learning_rate": 1.0222588623248144e-07, + "loss": 0.8547, + "mean_token_accuracy": 0.7432265877723694, + "num_tokens": 11375575.0, + "step": 373 + }, + { + "epoch": 0.030840273769275173, + "grad_norm": 5.56925106048584, + "learning_rate": 1.025006870019236e-07, + "loss": 0.8891, + "mean_token_accuracy": 0.7302370071411133, + "num_tokens": 11404191.0, + "step": 374 + }, + { + "epoch": 0.03092273439432671, + "grad_norm": 4.241403102874756, + "learning_rate": 1.0277548777136576e-07, + "loss": 1.1568, + "mean_token_accuracy": 0.6690861582756042, + "num_tokens": 11434421.0, + "step": 375 + }, + { + "epoch": 0.03100519501937825, + "grad_norm": 5.780751705169678, + "learning_rate": 1.0305028854080792e-07, + "loss": 1.0457, + "mean_token_accuracy": 0.6923712491989136, + "num_tokens": 11463228.0, + "step": 376 + }, + { + "epoch": 0.031087655644429786, + "grad_norm": 5.7516584396362305, + "learning_rate": 1.0332508931025006e-07, + "loss": 1.0551, + "mean_token_accuracy": 0.7027170658111572, + "num_tokens": 11493159.0, + "step": 377 + }, + { + "epoch": 0.031170116269481324, + "grad_norm": 6.439063549041748, + "learning_rate": 1.0359989007969222e-07, + "loss": 1.1603, + "mean_token_accuracy": 0.6609545350074768, + "num_tokens": 11522396.0, + "step": 378 + }, + { + "epoch": 0.03125257689453286, + "grad_norm": 6.1448445320129395, + "learning_rate": 1.0387469084913438e-07, + "loss": 1.1364, + "mean_token_accuracy": 0.6791951656341553, + "num_tokens": 11556209.0, + "step": 379 + }, + { + "epoch": 0.0313350375195844, + "grad_norm": 8.58962345123291, + "learning_rate": 1.0414949161857652e-07, + "loss": 1.0779, + "mean_token_accuracy": 0.6919496059417725, + "num_tokens": 11584008.0, + "step": 380 + }, + { + "epoch": 0.03141749814463594, + "grad_norm": 7.8014936447143555, + "learning_rate": 1.0442429238801868e-07, + "loss": 1.0938, + "mean_token_accuracy": 0.6876906156539917, + "num_tokens": 11607856.0, + "step": 381 + }, + { + "epoch": 0.031499958769687475, + "grad_norm": 4.910446643829346, + "learning_rate": 1.0469909315746084e-07, + "loss": 1.0383, + "mean_token_accuracy": 0.7073549032211304, + "num_tokens": 11642620.0, + "step": 382 + }, + { + "epoch": 0.03158241939473901, + "grad_norm": 5.677976131439209, + "learning_rate": 1.04973893926903e-07, + "loss": 1.0815, + "mean_token_accuracy": 0.6911696195602417, + "num_tokens": 11669324.0, + "step": 383 + }, + { + "epoch": 0.03166488001979055, + "grad_norm": 5.270987033843994, + "learning_rate": 1.0524869469634514e-07, + "loss": 1.0553, + "mean_token_accuracy": 0.6953511834144592, + "num_tokens": 11699271.0, + "step": 384 + }, + { + "epoch": 0.03174734064484209, + "grad_norm": 6.259089469909668, + "learning_rate": 1.055234954657873e-07, + "loss": 0.9568, + "mean_token_accuracy": 0.7004121541976929, + "num_tokens": 11723689.0, + "step": 385 + }, + { + "epoch": 0.031829801269893626, + "grad_norm": 4.991820812225342, + "learning_rate": 1.0579829623522946e-07, + "loss": 1.1928, + "mean_token_accuracy": 0.6653648614883423, + "num_tokens": 11755015.0, + "step": 386 + }, + { + "epoch": 0.031912261894945164, + "grad_norm": 5.89603328704834, + "learning_rate": 1.060730970046716e-07, + "loss": 0.971, + "mean_token_accuracy": 0.7138271331787109, + "num_tokens": 11783411.0, + "step": 387 + }, + { + "epoch": 0.0319947225199967, + "grad_norm": 4.083029747009277, + "learning_rate": 1.0634789777411376e-07, + "loss": 1.0097, + "mean_token_accuracy": 0.694365918636322, + "num_tokens": 11810224.0, + "step": 388 + }, + { + "epoch": 0.03207718314504824, + "grad_norm": 5.770605087280273, + "learning_rate": 1.0662269854355592e-07, + "loss": 1.0329, + "mean_token_accuracy": 0.7050026655197144, + "num_tokens": 11832786.0, + "step": 389 + }, + { + "epoch": 0.03215964377009978, + "grad_norm": 5.988123416900635, + "learning_rate": 1.0689749931299807e-07, + "loss": 1.0832, + "mean_token_accuracy": 0.6902873516082764, + "num_tokens": 11862187.0, + "step": 390 + }, + { + "epoch": 0.032242104395151315, + "grad_norm": 4.543367385864258, + "learning_rate": 1.0717230008244023e-07, + "loss": 1.0278, + "mean_token_accuracy": 0.6864785552024841, + "num_tokens": 11890509.0, + "step": 391 + }, + { + "epoch": 0.03232456502020285, + "grad_norm": 6.190674781799316, + "learning_rate": 1.0744710085188238e-07, + "loss": 1.1367, + "mean_token_accuracy": 0.6690788865089417, + "num_tokens": 11913618.0, + "step": 392 + }, + { + "epoch": 0.03240702564525439, + "grad_norm": 4.533934593200684, + "learning_rate": 1.0772190162132454e-07, + "loss": 1.2254, + "mean_token_accuracy": 0.6568992137908936, + "num_tokens": 11943339.0, + "step": 393 + }, + { + "epoch": 0.03248948627030593, + "grad_norm": 6.157672882080078, + "learning_rate": 1.0799670239076669e-07, + "loss": 0.9685, + "mean_token_accuracy": 0.7239092588424683, + "num_tokens": 11975149.0, + "step": 394 + }, + { + "epoch": 0.032571946895357466, + "grad_norm": 6.369676113128662, + "learning_rate": 1.0827150316020885e-07, + "loss": 1.1431, + "mean_token_accuracy": 0.6678471565246582, + "num_tokens": 11999460.0, + "step": 395 + }, + { + "epoch": 0.032654407520409004, + "grad_norm": 6.511314392089844, + "learning_rate": 1.08546303929651e-07, + "loss": 1.1185, + "mean_token_accuracy": 0.6771736145019531, + "num_tokens": 12020856.0, + "step": 396 + }, + { + "epoch": 0.03273686814546054, + "grad_norm": 4.756557464599609, + "learning_rate": 1.0882110469909315e-07, + "loss": 1.1189, + "mean_token_accuracy": 0.6778240203857422, + "num_tokens": 12055253.0, + "step": 397 + }, + { + "epoch": 0.03281932877051208, + "grad_norm": 6.355703353881836, + "learning_rate": 1.0909590546853531e-07, + "loss": 1.0904, + "mean_token_accuracy": 0.6938263177871704, + "num_tokens": 12081078.0, + "step": 398 + }, + { + "epoch": 0.03290178939556362, + "grad_norm": 5.261900424957275, + "learning_rate": 1.0937070623797747e-07, + "loss": 1.0809, + "mean_token_accuracy": 0.6889784336090088, + "num_tokens": 12111124.0, + "step": 399 + }, + { + "epoch": 0.032984250020615155, + "grad_norm": 5.879858493804932, + "learning_rate": 1.0964550700741962e-07, + "loss": 1.0982, + "mean_token_accuracy": 0.6899465322494507, + "num_tokens": 12137243.0, + "step": 400 + }, + { + "epoch": 0.03306671064566669, + "grad_norm": 7.348457336425781, + "learning_rate": 1.0992030777686177e-07, + "loss": 1.1172, + "mean_token_accuracy": 0.6860415935516357, + "num_tokens": 12162750.0, + "step": 401 + }, + { + "epoch": 0.03314917127071823, + "grad_norm": 7.543182373046875, + "learning_rate": 1.1019510854630393e-07, + "loss": 1.0456, + "mean_token_accuracy": 0.7055906057357788, + "num_tokens": 12192138.0, + "step": 402 + }, + { + "epoch": 0.03323163189576977, + "grad_norm": 5.320096969604492, + "learning_rate": 1.1046990931574609e-07, + "loss": 1.1297, + "mean_token_accuracy": 0.6804822087287903, + "num_tokens": 12226029.0, + "step": 403 + }, + { + "epoch": 0.033314092520821306, + "grad_norm": 4.586093902587891, + "learning_rate": 1.1074471008518823e-07, + "loss": 0.9941, + "mean_token_accuracy": 0.7095060348510742, + "num_tokens": 12264153.0, + "step": 404 + }, + { + "epoch": 0.033396553145872844, + "grad_norm": 4.9447712898254395, + "learning_rate": 1.1101951085463039e-07, + "loss": 0.9669, + "mean_token_accuracy": 0.7179563045501709, + "num_tokens": 12298089.0, + "step": 405 + }, + { + "epoch": 0.03347901377092438, + "grad_norm": 5.697714328765869, + "learning_rate": 1.1129431162407255e-07, + "loss": 0.8596, + "mean_token_accuracy": 0.745978593826294, + "num_tokens": 12325966.0, + "step": 406 + }, + { + "epoch": 0.03356147439597592, + "grad_norm": 5.637890815734863, + "learning_rate": 1.115691123935147e-07, + "loss": 1.0679, + "mean_token_accuracy": 0.6930242776870728, + "num_tokens": 12361425.0, + "step": 407 + }, + { + "epoch": 0.03364393502102746, + "grad_norm": 5.704819679260254, + "learning_rate": 1.1184391316295685e-07, + "loss": 0.92, + "mean_token_accuracy": 0.7222591638565063, + "num_tokens": 12384789.0, + "step": 408 + }, + { + "epoch": 0.033726395646078995, + "grad_norm": 4.904279708862305, + "learning_rate": 1.1211871393239901e-07, + "loss": 1.1493, + "mean_token_accuracy": 0.6770716905593872, + "num_tokens": 12416698.0, + "step": 409 + }, + { + "epoch": 0.03380885627113053, + "grad_norm": 5.696394920349121, + "learning_rate": 1.1239351470184117e-07, + "loss": 1.0699, + "mean_token_accuracy": 0.6961718797683716, + "num_tokens": 12449891.0, + "step": 410 + }, + { + "epoch": 0.03389131689618207, + "grad_norm": 4.488653182983398, + "learning_rate": 1.1266831547128331e-07, + "loss": 0.9705, + "mean_token_accuracy": 0.7134507894515991, + "num_tokens": 12480164.0, + "step": 411 + }, + { + "epoch": 0.03397377752123361, + "grad_norm": 4.840902328491211, + "learning_rate": 1.1294311624072547e-07, + "loss": 0.9707, + "mean_token_accuracy": 0.7162784934043884, + "num_tokens": 12516940.0, + "step": 412 + }, + { + "epoch": 0.034056238146285146, + "grad_norm": 4.637014865875244, + "learning_rate": 1.1321791701016763e-07, + "loss": 1.0542, + "mean_token_accuracy": 0.69876629114151, + "num_tokens": 12546576.0, + "step": 413 + }, + { + "epoch": 0.034138698771336684, + "grad_norm": 4.509403705596924, + "learning_rate": 1.1349271777960977e-07, + "loss": 0.9981, + "mean_token_accuracy": 0.7084206342697144, + "num_tokens": 12575417.0, + "step": 414 + }, + { + "epoch": 0.03422115939638822, + "grad_norm": 3.643341064453125, + "learning_rate": 1.1376751854905193e-07, + "loss": 0.9617, + "mean_token_accuracy": 0.7166204452514648, + "num_tokens": 12605555.0, + "step": 415 + }, + { + "epoch": 0.03430362002143976, + "grad_norm": 6.333124160766602, + "learning_rate": 1.1404231931849409e-07, + "loss": 1.1152, + "mean_token_accuracy": 0.682636559009552, + "num_tokens": 12632568.0, + "step": 416 + }, + { + "epoch": 0.0343860806464913, + "grad_norm": 4.648547649383545, + "learning_rate": 1.1431712008793625e-07, + "loss": 1.1116, + "mean_token_accuracy": 0.6858898401260376, + "num_tokens": 12665395.0, + "step": 417 + }, + { + "epoch": 0.034468541271542835, + "grad_norm": 5.851778507232666, + "learning_rate": 1.145919208573784e-07, + "loss": 0.9779, + "mean_token_accuracy": 0.7134439945220947, + "num_tokens": 12690751.0, + "step": 418 + }, + { + "epoch": 0.03455100189659437, + "grad_norm": 4.837076663970947, + "learning_rate": 1.1486672162682055e-07, + "loss": 1.0571, + "mean_token_accuracy": 0.6917853951454163, + "num_tokens": 12719073.0, + "step": 419 + }, + { + "epoch": 0.03463346252164591, + "grad_norm": 4.226434707641602, + "learning_rate": 1.1514152239626271e-07, + "loss": 0.9608, + "mean_token_accuracy": 0.72480309009552, + "num_tokens": 12750745.0, + "step": 420 + }, + { + "epoch": 0.03471592314669745, + "grad_norm": 5.294960975646973, + "learning_rate": 1.1541632316570486e-07, + "loss": 1.0247, + "mean_token_accuracy": 0.701513409614563, + "num_tokens": 12775343.0, + "step": 421 + }, + { + "epoch": 0.03479838377174899, + "grad_norm": 4.61802339553833, + "learning_rate": 1.1569112393514701e-07, + "loss": 1.0216, + "mean_token_accuracy": 0.7079809904098511, + "num_tokens": 12805528.0, + "step": 422 + }, + { + "epoch": 0.03488084439680053, + "grad_norm": 5.179559707641602, + "learning_rate": 1.1596592470458917e-07, + "loss": 0.9212, + "mean_token_accuracy": 0.7315980195999146, + "num_tokens": 12840336.0, + "step": 423 + }, + { + "epoch": 0.03496330502185207, + "grad_norm": 4.531215190887451, + "learning_rate": 1.1624072547403133e-07, + "loss": 0.9056, + "mean_token_accuracy": 0.7258950471878052, + "num_tokens": 12873949.0, + "step": 424 + }, + { + "epoch": 0.03504576564690361, + "grad_norm": 3.8001151084899902, + "learning_rate": 1.1651552624347348e-07, + "loss": 0.9376, + "mean_token_accuracy": 0.7183583974838257, + "num_tokens": 12914819.0, + "step": 425 + }, + { + "epoch": 0.035128226271955144, + "grad_norm": 4.7982940673828125, + "learning_rate": 1.1679032701291564e-07, + "loss": 0.7978, + "mean_token_accuracy": 0.7574738264083862, + "num_tokens": 12952144.0, + "step": 426 + }, + { + "epoch": 0.03521068689700668, + "grad_norm": 4.859533786773682, + "learning_rate": 1.170651277823578e-07, + "loss": 1.2178, + "mean_token_accuracy": 0.6667141914367676, + "num_tokens": 12985220.0, + "step": 427 + }, + { + "epoch": 0.03529314752205822, + "grad_norm": 5.5518388748168945, + "learning_rate": 1.1733992855179994e-07, + "loss": 1.0181, + "mean_token_accuracy": 0.704883337020874, + "num_tokens": 13011635.0, + "step": 428 + }, + { + "epoch": 0.03537560814710976, + "grad_norm": 5.492663383483887, + "learning_rate": 1.176147293212421e-07, + "loss": 1.0604, + "mean_token_accuracy": 0.6924668550491333, + "num_tokens": 13047061.0, + "step": 429 + }, + { + "epoch": 0.035458068772161296, + "grad_norm": 6.133967876434326, + "learning_rate": 1.1788953009068426e-07, + "loss": 1.1074, + "mean_token_accuracy": 0.6780788898468018, + "num_tokens": 13077516.0, + "step": 430 + }, + { + "epoch": 0.03554052939721283, + "grad_norm": 5.20629358291626, + "learning_rate": 1.1816433086012641e-07, + "loss": 1.1996, + "mean_token_accuracy": 0.662056565284729, + "num_tokens": 13108255.0, + "step": 431 + }, + { + "epoch": 0.03562299002226437, + "grad_norm": 4.156383514404297, + "learning_rate": 1.1843913162956856e-07, + "loss": 1.0656, + "mean_token_accuracy": 0.6928490400314331, + "num_tokens": 13144075.0, + "step": 432 + }, + { + "epoch": 0.03570545064731591, + "grad_norm": 4.210156440734863, + "learning_rate": 1.1871393239901072e-07, + "loss": 1.263, + "mean_token_accuracy": 0.6493101119995117, + "num_tokens": 13177598.0, + "step": 433 + }, + { + "epoch": 0.03578791127236745, + "grad_norm": 4.6002116203308105, + "learning_rate": 1.1898873316845288e-07, + "loss": 0.8778, + "mean_token_accuracy": 0.7354974746704102, + "num_tokens": 13208088.0, + "step": 434 + }, + { + "epoch": 0.035870371897418984, + "grad_norm": 3.800363779067993, + "learning_rate": 1.1926353393789502e-07, + "loss": 1.0255, + "mean_token_accuracy": 0.6930979490280151, + "num_tokens": 13238793.0, + "step": 435 + }, + { + "epoch": 0.03595283252247052, + "grad_norm": 5.32861852645874, + "learning_rate": 1.195383347073372e-07, + "loss": 1.1992, + "mean_token_accuracy": 0.6610348224639893, + "num_tokens": 13267741.0, + "step": 436 + }, + { + "epoch": 0.03603529314752206, + "grad_norm": 5.157675266265869, + "learning_rate": 1.1981313547677934e-07, + "loss": 1.0679, + "mean_token_accuracy": 0.6913200616836548, + "num_tokens": 13291616.0, + "step": 437 + }, + { + "epoch": 0.0361177537725736, + "grad_norm": 4.242971897125244, + "learning_rate": 1.2008793624622148e-07, + "loss": 0.9605, + "mean_token_accuracy": 0.7202513217926025, + "num_tokens": 13330435.0, + "step": 438 + }, + { + "epoch": 0.036200214397625136, + "grad_norm": 5.596362590789795, + "learning_rate": 1.2036273701566365e-07, + "loss": 1.1407, + "mean_token_accuracy": 0.6771824955940247, + "num_tokens": 13355301.0, + "step": 439 + }, + { + "epoch": 0.03628267502267667, + "grad_norm": 4.856388092041016, + "learning_rate": 1.206375377851058e-07, + "loss": 1.0167, + "mean_token_accuracy": 0.7057791948318481, + "num_tokens": 13387895.0, + "step": 440 + }, + { + "epoch": 0.03636513564772821, + "grad_norm": 4.356229782104492, + "learning_rate": 1.2091233855454794e-07, + "loss": 1.145, + "mean_token_accuracy": 0.669914722442627, + "num_tokens": 13413588.0, + "step": 441 + }, + { + "epoch": 0.03644759627277975, + "grad_norm": 5.353043556213379, + "learning_rate": 1.2118713932399012e-07, + "loss": 1.0141, + "mean_token_accuracy": 0.7098038196563721, + "num_tokens": 13439343.0, + "step": 442 + }, + { + "epoch": 0.03653005689783129, + "grad_norm": 4.449638843536377, + "learning_rate": 1.2146194009343226e-07, + "loss": 1.1591, + "mean_token_accuracy": 0.6803992390632629, + "num_tokens": 13474681.0, + "step": 443 + }, + { + "epoch": 0.036612517522882825, + "grad_norm": 4.834943771362305, + "learning_rate": 1.217367408628744e-07, + "loss": 1.0922, + "mean_token_accuracy": 0.6881239414215088, + "num_tokens": 13502989.0, + "step": 444 + }, + { + "epoch": 0.03669497814793436, + "grad_norm": 4.862467288970947, + "learning_rate": 1.2201154163231658e-07, + "loss": 1.0486, + "mean_token_accuracy": 0.6897081136703491, + "num_tokens": 13526448.0, + "step": 445 + }, + { + "epoch": 0.0367774387729859, + "grad_norm": 4.928224563598633, + "learning_rate": 1.2228634240175872e-07, + "loss": 1.1527, + "mean_token_accuracy": 0.6731460094451904, + "num_tokens": 13556199.0, + "step": 446 + }, + { + "epoch": 0.03685989939803744, + "grad_norm": 4.654381275177002, + "learning_rate": 1.2256114317120087e-07, + "loss": 1.0434, + "mean_token_accuracy": 0.6999683380126953, + "num_tokens": 13586188.0, + "step": 447 + }, + { + "epoch": 0.036942360023088976, + "grad_norm": 4.631624221801758, + "learning_rate": 1.2283594394064304e-07, + "loss": 1.1783, + "mean_token_accuracy": 0.6635637283325195, + "num_tokens": 13626061.0, + "step": 448 + }, + { + "epoch": 0.03702482064814051, + "grad_norm": 4.516923427581787, + "learning_rate": 1.2311074471008518e-07, + "loss": 1.0036, + "mean_token_accuracy": 0.7095522880554199, + "num_tokens": 13658733.0, + "step": 449 + }, + { + "epoch": 0.03710728127319205, + "grad_norm": 5.77588415145874, + "learning_rate": 1.2338554547952733e-07, + "loss": 1.2252, + "mean_token_accuracy": 0.6497097015380859, + "num_tokens": 13686157.0, + "step": 450 + }, + { + "epoch": 0.03718974189824359, + "grad_norm": 4.688918113708496, + "learning_rate": 1.236603462489695e-07, + "loss": 1.0449, + "mean_token_accuracy": 0.6997228860855103, + "num_tokens": 13717579.0, + "step": 451 + }, + { + "epoch": 0.03727220252329513, + "grad_norm": 4.715508937835693, + "learning_rate": 1.2393514701841165e-07, + "loss": 1.0653, + "mean_token_accuracy": 0.6998425126075745, + "num_tokens": 13748866.0, + "step": 452 + }, + { + "epoch": 0.037354663148346665, + "grad_norm": 5.133382320404053, + "learning_rate": 1.2420994778785382e-07, + "loss": 1.0586, + "mean_token_accuracy": 0.690837025642395, + "num_tokens": 13779288.0, + "step": 453 + }, + { + "epoch": 0.0374371237733982, + "grad_norm": 4.148893356323242, + "learning_rate": 1.2448474855729596e-07, + "loss": 1.0111, + "mean_token_accuracy": 0.7011115550994873, + "num_tokens": 13809541.0, + "step": 454 + }, + { + "epoch": 0.03751958439844974, + "grad_norm": 5.813617706298828, + "learning_rate": 1.247595493267381e-07, + "loss": 1.0466, + "mean_token_accuracy": 0.696711540222168, + "num_tokens": 13833066.0, + "step": 455 + }, + { + "epoch": 0.03760204502350128, + "grad_norm": 4.305210113525391, + "learning_rate": 1.2503435009618028e-07, + "loss": 1.0362, + "mean_token_accuracy": 0.6978673934936523, + "num_tokens": 13863437.0, + "step": 456 + }, + { + "epoch": 0.037684505648552816, + "grad_norm": 4.899416446685791, + "learning_rate": 1.253091508656224e-07, + "loss": 0.9414, + "mean_token_accuracy": 0.7226890325546265, + "num_tokens": 13901699.0, + "step": 457 + }, + { + "epoch": 0.037766966273604353, + "grad_norm": 4.550666332244873, + "learning_rate": 1.2558395163506457e-07, + "loss": 1.0391, + "mean_token_accuracy": 0.6968322992324829, + "num_tokens": 13937493.0, + "step": 458 + }, + { + "epoch": 0.03784942689865589, + "grad_norm": 5.6942033767700195, + "learning_rate": 1.2585875240450671e-07, + "loss": 0.9541, + "mean_token_accuracy": 0.7203657627105713, + "num_tokens": 13961540.0, + "step": 459 + }, + { + "epoch": 0.03793188752370743, + "grad_norm": 4.0701003074646, + "learning_rate": 1.2613355317394889e-07, + "loss": 0.8966, + "mean_token_accuracy": 0.7337546944618225, + "num_tokens": 13993762.0, + "step": 460 + }, + { + "epoch": 0.03801434814875897, + "grad_norm": 4.688321113586426, + "learning_rate": 1.2640835394339103e-07, + "loss": 0.9606, + "mean_token_accuracy": 0.7305984497070312, + "num_tokens": 14033676.0, + "step": 461 + }, + { + "epoch": 0.038096808773810505, + "grad_norm": 4.828301429748535, + "learning_rate": 1.266831547128332e-07, + "loss": 0.9928, + "mean_token_accuracy": 0.710822343826294, + "num_tokens": 14065330.0, + "step": 462 + }, + { + "epoch": 0.03817926939886204, + "grad_norm": 4.200056076049805, + "learning_rate": 1.2695795548227535e-07, + "loss": 0.8519, + "mean_token_accuracy": 0.742777407169342, + "num_tokens": 14095858.0, + "step": 463 + }, + { + "epoch": 0.03826173002391358, + "grad_norm": 4.770076274871826, + "learning_rate": 1.272327562517175e-07, + "loss": 1.1879, + "mean_token_accuracy": 0.6612261533737183, + "num_tokens": 14124091.0, + "step": 464 + }, + { + "epoch": 0.03834419064896512, + "grad_norm": 4.518112659454346, + "learning_rate": 1.2750755702115964e-07, + "loss": 1.1802, + "mean_token_accuracy": 0.664354681968689, + "num_tokens": 14156424.0, + "step": 465 + }, + { + "epoch": 0.038426651274016656, + "grad_norm": 4.108942985534668, + "learning_rate": 1.277823577906018e-07, + "loss": 0.9658, + "mean_token_accuracy": 0.7140756845474243, + "num_tokens": 14184292.0, + "step": 466 + }, + { + "epoch": 0.038509111899068194, + "grad_norm": 5.21198034286499, + "learning_rate": 1.2805715856004395e-07, + "loss": 1.0351, + "mean_token_accuracy": 0.6960062980651855, + "num_tokens": 14211116.0, + "step": 467 + }, + { + "epoch": 0.03859157252411973, + "grad_norm": 4.262142181396484, + "learning_rate": 1.2833195932948613e-07, + "loss": 0.8755, + "mean_token_accuracy": 0.7444535493850708, + "num_tokens": 14243424.0, + "step": 468 + }, + { + "epoch": 0.03867403314917127, + "grad_norm": 4.139091968536377, + "learning_rate": 1.2860676009892827e-07, + "loss": 0.8602, + "mean_token_accuracy": 0.7327604293823242, + "num_tokens": 14277946.0, + "step": 469 + }, + { + "epoch": 0.03875649377422281, + "grad_norm": 4.523189544677734, + "learning_rate": 1.2888156086837044e-07, + "loss": 1.0395, + "mean_token_accuracy": 0.6965677738189697, + "num_tokens": 14312760.0, + "step": 470 + }, + { + "epoch": 0.038838954399274345, + "grad_norm": 5.154651641845703, + "learning_rate": 1.2915636163781256e-07, + "loss": 0.9726, + "mean_token_accuracy": 0.7161046862602234, + "num_tokens": 14338121.0, + "step": 471 + }, + { + "epoch": 0.03892141502432588, + "grad_norm": 3.7244975566864014, + "learning_rate": 1.2943116240725473e-07, + "loss": 0.982, + "mean_token_accuracy": 0.7106634378433228, + "num_tokens": 14373633.0, + "step": 472 + }, + { + "epoch": 0.03900387564937742, + "grad_norm": 4.847822189331055, + "learning_rate": 1.2970596317669688e-07, + "loss": 1.1586, + "mean_token_accuracy": 0.6645189523696899, + "num_tokens": 14403688.0, + "step": 473 + }, + { + "epoch": 0.03908633627442896, + "grad_norm": 4.234280109405518, + "learning_rate": 1.2998076394613905e-07, + "loss": 0.9067, + "mean_token_accuracy": 0.7324267029762268, + "num_tokens": 14434601.0, + "step": 474 + }, + { + "epoch": 0.039168796899480496, + "grad_norm": 6.269776344299316, + "learning_rate": 1.302555647155812e-07, + "loss": 1.1556, + "mean_token_accuracy": 0.6689049005508423, + "num_tokens": 14456310.0, + "step": 475 + }, + { + "epoch": 0.039251257524532034, + "grad_norm": 4.660717964172363, + "learning_rate": 1.3053036548502337e-07, + "loss": 0.9329, + "mean_token_accuracy": 0.7231298685073853, + "num_tokens": 14485914.0, + "step": 476 + }, + { + "epoch": 0.03933371814958357, + "grad_norm": 4.890573501586914, + "learning_rate": 1.308051662544655e-07, + "loss": 0.9498, + "mean_token_accuracy": 0.7188484072685242, + "num_tokens": 14516507.0, + "step": 477 + }, + { + "epoch": 0.03941617877463511, + "grad_norm": 5.525663375854492, + "learning_rate": 1.3107996702390766e-07, + "loss": 1.1141, + "mean_token_accuracy": 0.6778415441513062, + "num_tokens": 14539000.0, + "step": 478 + }, + { + "epoch": 0.03949863939968665, + "grad_norm": 4.904911518096924, + "learning_rate": 1.313547677933498e-07, + "loss": 1.0383, + "mean_token_accuracy": 0.7049028277397156, + "num_tokens": 14563280.0, + "step": 479 + }, + { + "epoch": 0.039581100024738185, + "grad_norm": 4.151944160461426, + "learning_rate": 1.3162956856279197e-07, + "loss": 0.974, + "mean_token_accuracy": 0.708322286605835, + "num_tokens": 14591988.0, + "step": 480 + }, + { + "epoch": 0.03966356064978972, + "grad_norm": 3.8121399879455566, + "learning_rate": 1.3190436933223412e-07, + "loss": 1.0071, + "mean_token_accuracy": 0.7109956741333008, + "num_tokens": 14625755.0, + "step": 481 + }, + { + "epoch": 0.03974602127484126, + "grad_norm": 5.527960300445557, + "learning_rate": 1.321791701016763e-07, + "loss": 0.952, + "mean_token_accuracy": 0.7169830799102783, + "num_tokens": 14653126.0, + "step": 482 + }, + { + "epoch": 0.0398284818998928, + "grad_norm": 6.509961128234863, + "learning_rate": 1.3245397087111843e-07, + "loss": 0.9816, + "mean_token_accuracy": 0.7099634408950806, + "num_tokens": 14678708.0, + "step": 483 + }, + { + "epoch": 0.039910942524944336, + "grad_norm": 4.397010803222656, + "learning_rate": 1.327287716405606e-07, + "loss": 1.1197, + "mean_token_accuracy": 0.679036021232605, + "num_tokens": 14717407.0, + "step": 484 + }, + { + "epoch": 0.039993403149995874, + "grad_norm": 4.562992095947266, + "learning_rate": 1.3300357241000272e-07, + "loss": 1.1937, + "mean_token_accuracy": 0.673396110534668, + "num_tokens": 14743414.0, + "step": 485 + }, + { + "epoch": 0.04007586377504741, + "grad_norm": 3.611464023590088, + "learning_rate": 1.332783731794449e-07, + "loss": 1.0393, + "mean_token_accuracy": 0.7001224160194397, + "num_tokens": 14780765.0, + "step": 486 + }, + { + "epoch": 0.040158324400098956, + "grad_norm": 4.070435047149658, + "learning_rate": 1.3355317394888704e-07, + "loss": 0.9958, + "mean_token_accuracy": 0.6979522705078125, + "num_tokens": 14808771.0, + "step": 487 + }, + { + "epoch": 0.040240785025150494, + "grad_norm": 4.028872013092041, + "learning_rate": 1.338279747183292e-07, + "loss": 1.1291, + "mean_token_accuracy": 0.6815671920776367, + "num_tokens": 14843308.0, + "step": 488 + }, + { + "epoch": 0.04032324565020203, + "grad_norm": 5.346286296844482, + "learning_rate": 1.3410277548777136e-07, + "loss": 1.1566, + "mean_token_accuracy": 0.6728834509849548, + "num_tokens": 14870303.0, + "step": 489 + }, + { + "epoch": 0.04040570627525357, + "grad_norm": 4.400002479553223, + "learning_rate": 1.3437757625721353e-07, + "loss": 1.1345, + "mean_token_accuracy": 0.6730496287345886, + "num_tokens": 14900192.0, + "step": 490 + }, + { + "epoch": 0.04048816690030511, + "grad_norm": 4.526518821716309, + "learning_rate": 1.3465237702665565e-07, + "loss": 1.0403, + "mean_token_accuracy": 0.6882377862930298, + "num_tokens": 14931470.0, + "step": 491 + }, + { + "epoch": 0.040570627525356645, + "grad_norm": 5.517192363739014, + "learning_rate": 1.3492717779609782e-07, + "loss": 1.0293, + "mean_token_accuracy": 0.7019895315170288, + "num_tokens": 14955121.0, + "step": 492 + }, + { + "epoch": 0.04065308815040818, + "grad_norm": 4.681016445159912, + "learning_rate": 1.3520197856553997e-07, + "loss": 0.9194, + "mean_token_accuracy": 0.726089358329773, + "num_tokens": 14983901.0, + "step": 493 + }, + { + "epoch": 0.04073554877545972, + "grad_norm": 4.07907772064209, + "learning_rate": 1.3547677933498214e-07, + "loss": 1.0725, + "mean_token_accuracy": 0.6859234571456909, + "num_tokens": 15018016.0, + "step": 494 + }, + { + "epoch": 0.04081800940051126, + "grad_norm": 5.739263534545898, + "learning_rate": 1.3575158010442428e-07, + "loss": 0.9773, + "mean_token_accuracy": 0.7116990089416504, + "num_tokens": 15039416.0, + "step": 495 + }, + { + "epoch": 0.040900470025562796, + "grad_norm": 3.7092857360839844, + "learning_rate": 1.3602638087386645e-07, + "loss": 1.1123, + "mean_token_accuracy": 0.6817543506622314, + "num_tokens": 15079190.0, + "step": 496 + }, + { + "epoch": 0.040982930650614334, + "grad_norm": 3.564054012298584, + "learning_rate": 1.363011816433086e-07, + "loss": 1.031, + "mean_token_accuracy": 0.6977669596672058, + "num_tokens": 15110081.0, + "step": 497 + }, + { + "epoch": 0.04106539127566587, + "grad_norm": 4.342982769012451, + "learning_rate": 1.3657598241275074e-07, + "loss": 1.0372, + "mean_token_accuracy": 0.6888636946678162, + "num_tokens": 15135845.0, + "step": 498 + }, + { + "epoch": 0.04114785190071741, + "grad_norm": 4.2198076248168945, + "learning_rate": 1.368507831821929e-07, + "loss": 0.9215, + "mean_token_accuracy": 0.7241662740707397, + "num_tokens": 15161234.0, + "step": 499 + }, + { + "epoch": 0.04123031252576895, + "grad_norm": 5.376754283905029, + "learning_rate": 1.3712558395163506e-07, + "loss": 1.0416, + "mean_token_accuracy": 0.6985645294189453, + "num_tokens": 15182986.0, + "step": 500 + }, + { + "epoch": 0.041312773150820485, + "grad_norm": 5.401961803436279, + "learning_rate": 1.374003847210772e-07, + "loss": 1.0846, + "mean_token_accuracy": 0.6816807985305786, + "num_tokens": 15209065.0, + "step": 501 + }, + { + "epoch": 0.04139523377587202, + "grad_norm": 4.635659217834473, + "learning_rate": 1.3767518549051938e-07, + "loss": 1.1205, + "mean_token_accuracy": 0.6791418790817261, + "num_tokens": 15240846.0, + "step": 502 + }, + { + "epoch": 0.04147769440092356, + "grad_norm": 3.775330066680908, + "learning_rate": 1.3794998625996152e-07, + "loss": 1.06, + "mean_token_accuracy": 0.6832715272903442, + "num_tokens": 15270053.0, + "step": 503 + }, + { + "epoch": 0.0415601550259751, + "grad_norm": 4.971876621246338, + "learning_rate": 1.382247870294037e-07, + "loss": 1.1533, + "mean_token_accuracy": 0.6668470501899719, + "num_tokens": 15294435.0, + "step": 504 + }, + { + "epoch": 0.041642615651026636, + "grad_norm": 4.49626350402832, + "learning_rate": 1.384995877988458e-07, + "loss": 0.9548, + "mean_token_accuracy": 0.7084646224975586, + "num_tokens": 15319737.0, + "step": 505 + }, + { + "epoch": 0.041725076276078174, + "grad_norm": 5.448184013366699, + "learning_rate": 1.3877438856828798e-07, + "loss": 0.996, + "mean_token_accuracy": 0.7030441761016846, + "num_tokens": 15346082.0, + "step": 506 + }, + { + "epoch": 0.04180753690112971, + "grad_norm": 4.485584735870361, + "learning_rate": 1.3904918933773013e-07, + "loss": 1.0637, + "mean_token_accuracy": 0.694570779800415, + "num_tokens": 15377385.0, + "step": 507 + }, + { + "epoch": 0.04188999752618125, + "grad_norm": 5.281003952026367, + "learning_rate": 1.393239901071723e-07, + "loss": 1.1261, + "mean_token_accuracy": 0.6797637343406677, + "num_tokens": 15403939.0, + "step": 508 + }, + { + "epoch": 0.04197245815123279, + "grad_norm": 4.830862522125244, + "learning_rate": 1.3959879087661445e-07, + "loss": 1.115, + "mean_token_accuracy": 0.6700702905654907, + "num_tokens": 15431187.0, + "step": 509 + }, + { + "epoch": 0.042054918776284325, + "grad_norm": 4.216586112976074, + "learning_rate": 1.3987359164605662e-07, + "loss": 1.1471, + "mean_token_accuracy": 0.6648067831993103, + "num_tokens": 15461138.0, + "step": 510 + }, + { + "epoch": 0.04213737940133586, + "grad_norm": 4.883526802062988, + "learning_rate": 1.4014839241549876e-07, + "loss": 0.9189, + "mean_token_accuracy": 0.7209670543670654, + "num_tokens": 15483398.0, + "step": 511 + }, + { + "epoch": 0.0422198400263874, + "grad_norm": 4.295525550842285, + "learning_rate": 1.404231931849409e-07, + "loss": 1.156, + "mean_token_accuracy": 0.6730816960334778, + "num_tokens": 15509536.0, + "step": 512 + }, + { + "epoch": 0.04230230065143894, + "grad_norm": 4.136619567871094, + "learning_rate": 1.4069799395438305e-07, + "loss": 0.9654, + "mean_token_accuracy": 0.7193925380706787, + "num_tokens": 15540424.0, + "step": 513 + }, + { + "epoch": 0.042384761276490476, + "grad_norm": 4.983861446380615, + "learning_rate": 1.4097279472382522e-07, + "loss": 1.0731, + "mean_token_accuracy": 0.6962389349937439, + "num_tokens": 15568562.0, + "step": 514 + }, + { + "epoch": 0.042467221901542014, + "grad_norm": 5.167044162750244, + "learning_rate": 1.4124759549326737e-07, + "loss": 0.9676, + "mean_token_accuracy": 0.7152812480926514, + "num_tokens": 15595971.0, + "step": 515 + }, + { + "epoch": 0.04254968252659355, + "grad_norm": 4.434226036071777, + "learning_rate": 1.4152239626270954e-07, + "loss": 1.0405, + "mean_token_accuracy": 0.6995260715484619, + "num_tokens": 15627136.0, + "step": 516 + }, + { + "epoch": 0.04263214315164509, + "grad_norm": 5.215696334838867, + "learning_rate": 1.4179719703215169e-07, + "loss": 1.1437, + "mean_token_accuracy": 0.6671396493911743, + "num_tokens": 15651115.0, + "step": 517 + }, + { + "epoch": 0.04271460377669663, + "grad_norm": 4.053630828857422, + "learning_rate": 1.4207199780159386e-07, + "loss": 0.8936, + "mean_token_accuracy": 0.7327508926391602, + "num_tokens": 15680099.0, + "step": 518 + }, + { + "epoch": 0.042797064401748165, + "grad_norm": 4.632723808288574, + "learning_rate": 1.4234679857103598e-07, + "loss": 0.9031, + "mean_token_accuracy": 0.7135870456695557, + "num_tokens": 15704631.0, + "step": 519 + }, + { + "epoch": 0.0428795250267997, + "grad_norm": 4.183765888214111, + "learning_rate": 1.4262159934047815e-07, + "loss": 0.7922, + "mean_token_accuracy": 0.7584472894668579, + "num_tokens": 15737901.0, + "step": 520 + }, + { + "epoch": 0.04296198565185124, + "grad_norm": 4.551049709320068, + "learning_rate": 1.428964001099203e-07, + "loss": 0.9568, + "mean_token_accuracy": 0.7165443897247314, + "num_tokens": 15765735.0, + "step": 521 + }, + { + "epoch": 0.04304444627690278, + "grad_norm": 4.164209842681885, + "learning_rate": 1.4317120087936246e-07, + "loss": 0.9799, + "mean_token_accuracy": 0.7209702134132385, + "num_tokens": 15794509.0, + "step": 522 + }, + { + "epoch": 0.043126906901954316, + "grad_norm": 3.9399752616882324, + "learning_rate": 1.434460016488046e-07, + "loss": 1.1016, + "mean_token_accuracy": 0.6785260438919067, + "num_tokens": 15823519.0, + "step": 523 + }, + { + "epoch": 0.043209367527005854, + "grad_norm": 4.494845390319824, + "learning_rate": 1.4372080241824678e-07, + "loss": 1.0928, + "mean_token_accuracy": 0.6858407855033875, + "num_tokens": 15852102.0, + "step": 524 + }, + { + "epoch": 0.04329182815205739, + "grad_norm": 4.064519882202148, + "learning_rate": 1.4399560318768893e-07, + "loss": 1.2104, + "mean_token_accuracy": 0.6615128517150879, + "num_tokens": 15884505.0, + "step": 525 + }, + { + "epoch": 0.04337428877710893, + "grad_norm": 3.806948661804199, + "learning_rate": 1.4427040395713107e-07, + "loss": 0.9945, + "mean_token_accuracy": 0.6941375136375427, + "num_tokens": 15912693.0, + "step": 526 + }, + { + "epoch": 0.04345674940216047, + "grad_norm": 3.6080093383789062, + "learning_rate": 1.4454520472657322e-07, + "loss": 0.9881, + "mean_token_accuracy": 0.7194955945014954, + "num_tokens": 15955300.0, + "step": 527 + }, + { + "epoch": 0.043539210027212005, + "grad_norm": 3.7805423736572266, + "learning_rate": 1.448200054960154e-07, + "loss": 1.0174, + "mean_token_accuracy": 0.7021155953407288, + "num_tokens": 15995605.0, + "step": 528 + }, + { + "epoch": 0.04362167065226354, + "grad_norm": 3.020883560180664, + "learning_rate": 1.4509480626545753e-07, + "loss": 1.021, + "mean_token_accuracy": 0.7036924958229065, + "num_tokens": 16035259.0, + "step": 529 + }, + { + "epoch": 0.04370413127731508, + "grad_norm": 4.665146827697754, + "learning_rate": 1.453696070348997e-07, + "loss": 1.0167, + "mean_token_accuracy": 0.7088425755500793, + "num_tokens": 16063363.0, + "step": 530 + }, + { + "epoch": 0.04378659190236662, + "grad_norm": 4.151666641235352, + "learning_rate": 1.4564440780434185e-07, + "loss": 1.0836, + "mean_token_accuracy": 0.691832959651947, + "num_tokens": 16095178.0, + "step": 531 + }, + { + "epoch": 0.043869052527418156, + "grad_norm": 4.715571880340576, + "learning_rate": 1.4591920857378402e-07, + "loss": 1.126, + "mean_token_accuracy": 0.675773024559021, + "num_tokens": 16119216.0, + "step": 532 + }, + { + "epoch": 0.043951513152469694, + "grad_norm": 3.751617670059204, + "learning_rate": 1.4619400934322614e-07, + "loss": 0.9899, + "mean_token_accuracy": 0.7085219621658325, + "num_tokens": 16150894.0, + "step": 533 + }, + { + "epoch": 0.04403397377752123, + "grad_norm": 3.650418758392334, + "learning_rate": 1.464688101126683e-07, + "loss": 1.0802, + "mean_token_accuracy": 0.690751314163208, + "num_tokens": 16185057.0, + "step": 534 + }, + { + "epoch": 0.04411643440257277, + "grad_norm": 5.088624000549316, + "learning_rate": 1.4674361088211046e-07, + "loss": 1.1451, + "mean_token_accuracy": 0.6690176725387573, + "num_tokens": 16210673.0, + "step": 535 + }, + { + "epoch": 0.04419889502762431, + "grad_norm": 4.926764965057373, + "learning_rate": 1.4701841165155263e-07, + "loss": 1.0878, + "mean_token_accuracy": 0.6884994506835938, + "num_tokens": 16242529.0, + "step": 536 + }, + { + "epoch": 0.044281355652675845, + "grad_norm": 3.9357776641845703, + "learning_rate": 1.4729321242099477e-07, + "loss": 1.031, + "mean_token_accuracy": 0.7094292044639587, + "num_tokens": 16276398.0, + "step": 537 + }, + { + "epoch": 0.04436381627772738, + "grad_norm": 3.4708011150360107, + "learning_rate": 1.4756801319043694e-07, + "loss": 0.948, + "mean_token_accuracy": 0.7232272624969482, + "num_tokens": 16310659.0, + "step": 538 + }, + { + "epoch": 0.04444627690277892, + "grad_norm": 4.614837646484375, + "learning_rate": 1.4784281395987906e-07, + "loss": 0.9935, + "mean_token_accuracy": 0.7015576362609863, + "num_tokens": 16335018.0, + "step": 539 + }, + { + "epoch": 0.04452873752783046, + "grad_norm": 3.4738569259643555, + "learning_rate": 1.4811761472932123e-07, + "loss": 1.1677, + "mean_token_accuracy": 0.6652349233627319, + "num_tokens": 16367479.0, + "step": 540 + }, + { + "epoch": 0.044611198152881996, + "grad_norm": 4.428458213806152, + "learning_rate": 1.4839241549876338e-07, + "loss": 1.1618, + "mean_token_accuracy": 0.6659497022628784, + "num_tokens": 16392198.0, + "step": 541 + }, + { + "epoch": 0.044693658777933534, + "grad_norm": 3.926597833633423, + "learning_rate": 1.4866721626820555e-07, + "loss": 1.0977, + "mean_token_accuracy": 0.686291515827179, + "num_tokens": 16423709.0, + "step": 542 + }, + { + "epoch": 0.04477611940298507, + "grad_norm": 6.159482479095459, + "learning_rate": 1.489420170376477e-07, + "loss": 0.9198, + "mean_token_accuracy": 0.721812903881073, + "num_tokens": 16447638.0, + "step": 543 + }, + { + "epoch": 0.04485858002803661, + "grad_norm": 4.980267524719238, + "learning_rate": 1.4921681780708987e-07, + "loss": 1.1478, + "mean_token_accuracy": 0.6635361909866333, + "num_tokens": 16479711.0, + "step": 544 + }, + { + "epoch": 0.04494104065308815, + "grad_norm": 4.666571617126465, + "learning_rate": 1.49491618576532e-07, + "loss": 1.045, + "mean_token_accuracy": 0.6887800693511963, + "num_tokens": 16503279.0, + "step": 545 + }, + { + "epoch": 0.045023501278139685, + "grad_norm": 3.9530065059661865, + "learning_rate": 1.4976641934597416e-07, + "loss": 1.0459, + "mean_token_accuracy": 0.6999658346176147, + "num_tokens": 16537196.0, + "step": 546 + }, + { + "epoch": 0.04510596190319122, + "grad_norm": 5.542712688446045, + "learning_rate": 1.500412201154163e-07, + "loss": 0.9917, + "mean_token_accuracy": 0.7011712193489075, + "num_tokens": 16561411.0, + "step": 547 + }, + { + "epoch": 0.04518842252824276, + "grad_norm": 3.9079017639160156, + "learning_rate": 1.5031602088485847e-07, + "loss": 1.0367, + "mean_token_accuracy": 0.6993895173072815, + "num_tokens": 16590382.0, + "step": 548 + }, + { + "epoch": 0.0452708831532943, + "grad_norm": 4.668263912200928, + "learning_rate": 1.5059082165430062e-07, + "loss": 1.0759, + "mean_token_accuracy": 0.6896135807037354, + "num_tokens": 16617321.0, + "step": 549 + }, + { + "epoch": 0.045353343778345837, + "grad_norm": 3.9555857181549072, + "learning_rate": 1.508656224237428e-07, + "loss": 1.0701, + "mean_token_accuracy": 0.68265700340271, + "num_tokens": 16646293.0, + "step": 550 + }, + { + "epoch": 0.045435804403397374, + "grad_norm": 4.079385757446289, + "learning_rate": 1.5114042319318494e-07, + "loss": 0.9943, + "mean_token_accuracy": 0.7083858847618103, + "num_tokens": 16676794.0, + "step": 551 + }, + { + "epoch": 0.04551826502844892, + "grad_norm": 4.305849075317383, + "learning_rate": 1.514152239626271e-07, + "loss": 0.9927, + "mean_token_accuracy": 0.7043666243553162, + "num_tokens": 16700907.0, + "step": 552 + }, + { + "epoch": 0.04560072565350046, + "grad_norm": 3.4771642684936523, + "learning_rate": 1.5169002473206923e-07, + "loss": 0.8916, + "mean_token_accuracy": 0.7351064085960388, + "num_tokens": 16736725.0, + "step": 553 + }, + { + "epoch": 0.045683186278551995, + "grad_norm": 4.276869297027588, + "learning_rate": 1.519648255015114e-07, + "loss": 1.0132, + "mean_token_accuracy": 0.7064852118492126, + "num_tokens": 16763459.0, + "step": 554 + }, + { + "epoch": 0.04576564690360353, + "grad_norm": 3.772475242614746, + "learning_rate": 1.5223962627095354e-07, + "loss": 1.0568, + "mean_token_accuracy": 0.6977787017822266, + "num_tokens": 16798383.0, + "step": 555 + }, + { + "epoch": 0.04584810752865507, + "grad_norm": 4.073163032531738, + "learning_rate": 1.5251442704039571e-07, + "loss": 1.1286, + "mean_token_accuracy": 0.6771054267883301, + "num_tokens": 16829576.0, + "step": 556 + }, + { + "epoch": 0.04593056815370661, + "grad_norm": 3.1840150356292725, + "learning_rate": 1.5278922780983786e-07, + "loss": 0.9518, + "mean_token_accuracy": 0.7190777063369751, + "num_tokens": 16864479.0, + "step": 557 + }, + { + "epoch": 0.046013028778758146, + "grad_norm": 3.620654821395874, + "learning_rate": 1.5306402857928003e-07, + "loss": 0.888, + "mean_token_accuracy": 0.7333621382713318, + "num_tokens": 16901519.0, + "step": 558 + }, + { + "epoch": 0.046095489403809684, + "grad_norm": 4.9404616355896, + "learning_rate": 1.5333882934872218e-07, + "loss": 1.0118, + "mean_token_accuracy": 0.7019839286804199, + "num_tokens": 16925970.0, + "step": 559 + }, + { + "epoch": 0.04617795002886122, + "grad_norm": 3.7930755615234375, + "learning_rate": 1.5361363011816432e-07, + "loss": 0.7846, + "mean_token_accuracy": 0.7611290812492371, + "num_tokens": 16954956.0, + "step": 560 + }, + { + "epoch": 0.04626041065391276, + "grad_norm": 3.4686391353607178, + "learning_rate": 1.5388843088760647e-07, + "loss": 0.989, + "mean_token_accuracy": 0.7100855112075806, + "num_tokens": 16983545.0, + "step": 561 + }, + { + "epoch": 0.0463428712789643, + "grad_norm": 4.48547887802124, + "learning_rate": 1.5416323165704864e-07, + "loss": 1.0597, + "mean_token_accuracy": 0.696077287197113, + "num_tokens": 17009749.0, + "step": 562 + }, + { + "epoch": 0.046425331904015835, + "grad_norm": 3.9617702960968018, + "learning_rate": 1.5443803242649078e-07, + "loss": 1.2054, + "mean_token_accuracy": 0.6606231927871704, + "num_tokens": 17035297.0, + "step": 563 + }, + { + "epoch": 0.04650779252906737, + "grad_norm": 3.831883192062378, + "learning_rate": 1.5471283319593295e-07, + "loss": 0.8945, + "mean_token_accuracy": 0.7412411570549011, + "num_tokens": 17067517.0, + "step": 564 + }, + { + "epoch": 0.04659025315411891, + "grad_norm": 5.207947731018066, + "learning_rate": 1.549876339653751e-07, + "loss": 0.9734, + "mean_token_accuracy": 0.7161318063735962, + "num_tokens": 17094651.0, + "step": 565 + }, + { + "epoch": 0.04667271377917045, + "grad_norm": 4.419576644897461, + "learning_rate": 1.5526243473481727e-07, + "loss": 1.1031, + "mean_token_accuracy": 0.682977020740509, + "num_tokens": 17117173.0, + "step": 566 + }, + { + "epoch": 0.046755174404221986, + "grad_norm": 4.161253929138184, + "learning_rate": 1.555372355042594e-07, + "loss": 1.0221, + "mean_token_accuracy": 0.7017285823822021, + "num_tokens": 17147485.0, + "step": 567 + }, + { + "epoch": 0.046837635029273524, + "grad_norm": 3.1472160816192627, + "learning_rate": 1.5581203627370156e-07, + "loss": 1.0292, + "mean_token_accuracy": 0.7028848528862, + "num_tokens": 17177716.0, + "step": 568 + }, + { + "epoch": 0.04692009565432506, + "grad_norm": 3.509080648422241, + "learning_rate": 1.560868370431437e-07, + "loss": 1.0039, + "mean_token_accuracy": 0.7031606435775757, + "num_tokens": 17210291.0, + "step": 569 + }, + { + "epoch": 0.0470025562793766, + "grad_norm": 3.764631509780884, + "learning_rate": 1.5636163781258588e-07, + "loss": 0.9453, + "mean_token_accuracy": 0.7194751501083374, + "num_tokens": 17238841.0, + "step": 570 + }, + { + "epoch": 0.04708501690442814, + "grad_norm": 3.5326013565063477, + "learning_rate": 1.5663643858202802e-07, + "loss": 1.0987, + "mean_token_accuracy": 0.6862012147903442, + "num_tokens": 17271352.0, + "step": 571 + }, + { + "epoch": 0.047167477529479675, + "grad_norm": 3.5689520835876465, + "learning_rate": 1.569112393514702e-07, + "loss": 1.085, + "mean_token_accuracy": 0.6812868118286133, + "num_tokens": 17302868.0, + "step": 572 + }, + { + "epoch": 0.04724993815453121, + "grad_norm": 3.7645485401153564, + "learning_rate": 1.5718604012091234e-07, + "loss": 1.0352, + "mean_token_accuracy": 0.7029575109481812, + "num_tokens": 17330871.0, + "step": 573 + }, + { + "epoch": 0.04733239877958275, + "grad_norm": 3.90879225730896, + "learning_rate": 1.5746084089035449e-07, + "loss": 0.9547, + "mean_token_accuracy": 0.7123581171035767, + "num_tokens": 17361883.0, + "step": 574 + }, + { + "epoch": 0.04741485940463429, + "grad_norm": 3.4213809967041016, + "learning_rate": 1.5773564165979663e-07, + "loss": 1.1028, + "mean_token_accuracy": 0.6776801347732544, + "num_tokens": 17392014.0, + "step": 575 + }, + { + "epoch": 0.047497320029685826, + "grad_norm": 3.774348020553589, + "learning_rate": 1.580104424292388e-07, + "loss": 0.9675, + "mean_token_accuracy": 0.7029868364334106, + "num_tokens": 17415739.0, + "step": 576 + }, + { + "epoch": 0.047579780654737364, + "grad_norm": 3.1582376956939697, + "learning_rate": 1.5828524319868095e-07, + "loss": 0.8367, + "mean_token_accuracy": 0.750035285949707, + "num_tokens": 17454775.0, + "step": 577 + }, + { + "epoch": 0.0476622412797889, + "grad_norm": 3.5375373363494873, + "learning_rate": 1.5856004396812312e-07, + "loss": 0.8963, + "mean_token_accuracy": 0.7407832741737366, + "num_tokens": 17484948.0, + "step": 578 + }, + { + "epoch": 0.04774470190484044, + "grad_norm": 3.6256909370422363, + "learning_rate": 1.5883484473756526e-07, + "loss": 0.9752, + "mean_token_accuracy": 0.712343156337738, + "num_tokens": 17521083.0, + "step": 579 + }, + { + "epoch": 0.04782716252989198, + "grad_norm": 3.5074384212493896, + "learning_rate": 1.5910964550700744e-07, + "loss": 0.9148, + "mean_token_accuracy": 0.7260607481002808, + "num_tokens": 17557127.0, + "step": 580 + }, + { + "epoch": 0.047909623154943515, + "grad_norm": 3.0813939571380615, + "learning_rate": 1.5938444627644955e-07, + "loss": 0.9223, + "mean_token_accuracy": 0.7284286618232727, + "num_tokens": 17589796.0, + "step": 581 + }, + { + "epoch": 0.04799208377999505, + "grad_norm": 3.719925880432129, + "learning_rate": 1.5965924704589173e-07, + "loss": 0.9166, + "mean_token_accuracy": 0.7301337718963623, + "num_tokens": 17621763.0, + "step": 582 + }, + { + "epoch": 0.04807454440504659, + "grad_norm": 2.993288993835449, + "learning_rate": 1.5993404781533387e-07, + "loss": 1.0218, + "mean_token_accuracy": 0.700447678565979, + "num_tokens": 17657314.0, + "step": 583 + }, + { + "epoch": 0.04815700503009813, + "grad_norm": 3.601595163345337, + "learning_rate": 1.6020884858477604e-07, + "loss": 1.0085, + "mean_token_accuracy": 0.7135157585144043, + "num_tokens": 17686834.0, + "step": 584 + }, + { + "epoch": 0.048239465655149666, + "grad_norm": 3.4932072162628174, + "learning_rate": 1.604836493542182e-07, + "loss": 0.8648, + "mean_token_accuracy": 0.7288352251052856, + "num_tokens": 17714352.0, + "step": 585 + }, + { + "epoch": 0.048321926280201204, + "grad_norm": 3.8073418140411377, + "learning_rate": 1.6075845012366036e-07, + "loss": 1.0691, + "mean_token_accuracy": 0.6898394823074341, + "num_tokens": 17745388.0, + "step": 586 + }, + { + "epoch": 0.04840438690525274, + "grad_norm": 4.072897434234619, + "learning_rate": 1.6103325089310248e-07, + "loss": 1.1273, + "mean_token_accuracy": 0.6628600358963013, + "num_tokens": 17767239.0, + "step": 587 + }, + { + "epoch": 0.04848684753030428, + "grad_norm": 3.0891079902648926, + "learning_rate": 1.6130805166254465e-07, + "loss": 0.8133, + "mean_token_accuracy": 0.7423436641693115, + "num_tokens": 17797116.0, + "step": 588 + }, + { + "epoch": 0.04856930815535582, + "grad_norm": 3.2506508827209473, + "learning_rate": 1.615828524319868e-07, + "loss": 0.9268, + "mean_token_accuracy": 0.7205811738967896, + "num_tokens": 17827786.0, + "step": 589 + }, + { + "epoch": 0.048651768780407355, + "grad_norm": 3.079639434814453, + "learning_rate": 1.6185765320142897e-07, + "loss": 0.9753, + "mean_token_accuracy": 0.7133547067642212, + "num_tokens": 17861556.0, + "step": 590 + }, + { + "epoch": 0.04873422940545889, + "grad_norm": 3.1487221717834473, + "learning_rate": 1.621324539708711e-07, + "loss": 1.0394, + "mean_token_accuracy": 0.6958233714103699, + "num_tokens": 17899251.0, + "step": 591 + }, + { + "epoch": 0.04881669003051043, + "grad_norm": 3.201345682144165, + "learning_rate": 1.6240725474031328e-07, + "loss": 0.8967, + "mean_token_accuracy": 0.7338463664054871, + "num_tokens": 17933331.0, + "step": 592 + }, + { + "epoch": 0.04889915065556197, + "grad_norm": 3.8071560859680176, + "learning_rate": 1.6268205550975543e-07, + "loss": 1.0346, + "mean_token_accuracy": 0.6958250999450684, + "num_tokens": 17962639.0, + "step": 593 + }, + { + "epoch": 0.048981611280613506, + "grad_norm": 4.163147926330566, + "learning_rate": 1.6295685627919757e-07, + "loss": 0.9379, + "mean_token_accuracy": 0.7258360385894775, + "num_tokens": 17994967.0, + "step": 594 + }, + { + "epoch": 0.049064071905665044, + "grad_norm": 4.295119285583496, + "learning_rate": 1.6323165704863972e-07, + "loss": 1.1382, + "mean_token_accuracy": 0.6673653721809387, + "num_tokens": 18019580.0, + "step": 595 + }, + { + "epoch": 0.04914653253071658, + "grad_norm": 4.2669243812561035, + "learning_rate": 1.635064578180819e-07, + "loss": 1.0016, + "mean_token_accuracy": 0.7031918168067932, + "num_tokens": 18044796.0, + "step": 596 + }, + { + "epoch": 0.04922899315576812, + "grad_norm": 4.065331935882568, + "learning_rate": 1.6378125858752403e-07, + "loss": 0.9561, + "mean_token_accuracy": 0.7100266814231873, + "num_tokens": 18071934.0, + "step": 597 + }, + { + "epoch": 0.04931145378081966, + "grad_norm": 3.317343235015869, + "learning_rate": 1.640560593569662e-07, + "loss": 0.9417, + "mean_token_accuracy": 0.7243518829345703, + "num_tokens": 18100679.0, + "step": 598 + }, + { + "epoch": 0.049393914405871195, + "grad_norm": 3.37341046333313, + "learning_rate": 1.6433086012640835e-07, + "loss": 1.0116, + "mean_token_accuracy": 0.7033110857009888, + "num_tokens": 18130321.0, + "step": 599 + }, + { + "epoch": 0.04947637503092273, + "grad_norm": 3.6265721321105957, + "learning_rate": 1.6460566089585052e-07, + "loss": 1.0371, + "mean_token_accuracy": 0.7040106058120728, + "num_tokens": 18158518.0, + "step": 600 + }, + { + "epoch": 0.04955883565597427, + "grad_norm": 3.160498857498169, + "learning_rate": 1.6488046166529264e-07, + "loss": 1.0347, + "mean_token_accuracy": 0.700425922870636, + "num_tokens": 18193720.0, + "step": 601 + }, + { + "epoch": 0.04964129628102581, + "grad_norm": 3.1256375312805176, + "learning_rate": 1.651552624347348e-07, + "loss": 0.9665, + "mean_token_accuracy": 0.7018600702285767, + "num_tokens": 18230108.0, + "step": 602 + }, + { + "epoch": 0.049723756906077346, + "grad_norm": 3.1498641967773438, + "learning_rate": 1.6543006320417696e-07, + "loss": 1.1531, + "mean_token_accuracy": 0.674281656742096, + "num_tokens": 18265074.0, + "step": 603 + }, + { + "epoch": 0.049806217531128884, + "grad_norm": 3.7772538661956787, + "learning_rate": 1.6570486397361913e-07, + "loss": 0.9368, + "mean_token_accuracy": 0.7218013405799866, + "num_tokens": 18294659.0, + "step": 604 + }, + { + "epoch": 0.04988867815618042, + "grad_norm": 3.236173629760742, + "learning_rate": 1.6597966474306127e-07, + "loss": 0.9954, + "mean_token_accuracy": 0.6991817951202393, + "num_tokens": 18332534.0, + "step": 605 + }, + { + "epoch": 0.04997113878123196, + "grad_norm": 3.245250701904297, + "learning_rate": 1.6625446551250345e-07, + "loss": 1.0813, + "mean_token_accuracy": 0.687475323677063, + "num_tokens": 18363019.0, + "step": 606 + }, + { + "epoch": 0.0500535994062835, + "grad_norm": 3.1937448978424072, + "learning_rate": 1.665292662819456e-07, + "loss": 1.0004, + "mean_token_accuracy": 0.7092249393463135, + "num_tokens": 18391309.0, + "step": 607 + }, + { + "epoch": 0.050136060031335035, + "grad_norm": 3.001758337020874, + "learning_rate": 1.6680406705138774e-07, + "loss": 0.8401, + "mean_token_accuracy": 0.739920973777771, + "num_tokens": 18423327.0, + "step": 608 + }, + { + "epoch": 0.05021852065638657, + "grad_norm": 2.7099173069000244, + "learning_rate": 1.6707886782082988e-07, + "loss": 0.9372, + "mean_token_accuracy": 0.7255048751831055, + "num_tokens": 18457926.0, + "step": 609 + }, + { + "epoch": 0.05030098128143811, + "grad_norm": 3.177976369857788, + "learning_rate": 1.6735366859027205e-07, + "loss": 0.9558, + "mean_token_accuracy": 0.7280767560005188, + "num_tokens": 18488840.0, + "step": 610 + }, + { + "epoch": 0.05038344190648965, + "grad_norm": 3.1263911724090576, + "learning_rate": 1.676284693597142e-07, + "loss": 1.0134, + "mean_token_accuracy": 0.7016510963439941, + "num_tokens": 18522872.0, + "step": 611 + }, + { + "epoch": 0.050465902531541186, + "grad_norm": 3.773547410964966, + "learning_rate": 1.6790327012915637e-07, + "loss": 1.0818, + "mean_token_accuracy": 0.6760439872741699, + "num_tokens": 18550275.0, + "step": 612 + }, + { + "epoch": 0.050548363156592724, + "grad_norm": 3.2908623218536377, + "learning_rate": 1.6817807089859851e-07, + "loss": 0.9625, + "mean_token_accuracy": 0.7126534581184387, + "num_tokens": 18579192.0, + "step": 613 + }, + { + "epoch": 0.05063082378164426, + "grad_norm": 3.0089783668518066, + "learning_rate": 1.6845287166804069e-07, + "loss": 1.12, + "mean_token_accuracy": 0.6788918375968933, + "num_tokens": 18611846.0, + "step": 614 + }, + { + "epoch": 0.0507132844066958, + "grad_norm": 3.2103664875030518, + "learning_rate": 1.687276724374828e-07, + "loss": 0.9836, + "mean_token_accuracy": 0.7057093381881714, + "num_tokens": 18647264.0, + "step": 615 + }, + { + "epoch": 0.05079574503174734, + "grad_norm": 3.4371323585510254, + "learning_rate": 1.6900247320692498e-07, + "loss": 0.7517, + "mean_token_accuracy": 0.7705564498901367, + "num_tokens": 18678028.0, + "step": 616 + }, + { + "epoch": 0.05087820565679888, + "grad_norm": 3.8942458629608154, + "learning_rate": 1.6927727397636712e-07, + "loss": 1.0186, + "mean_token_accuracy": 0.6988271474838257, + "num_tokens": 18704305.0, + "step": 617 + }, + { + "epoch": 0.05096066628185042, + "grad_norm": 2.750528573989868, + "learning_rate": 1.695520747458093e-07, + "loss": 0.999, + "mean_token_accuracy": 0.7101117968559265, + "num_tokens": 18737344.0, + "step": 618 + }, + { + "epoch": 0.05104312690690196, + "grad_norm": 3.246530771255493, + "learning_rate": 1.6982687551525144e-07, + "loss": 0.9543, + "mean_token_accuracy": 0.7268282175064087, + "num_tokens": 18770067.0, + "step": 619 + }, + { + "epoch": 0.051125587531953495, + "grad_norm": 3.111365556716919, + "learning_rate": 1.701016762846936e-07, + "loss": 0.8739, + "mean_token_accuracy": 0.7229824662208557, + "num_tokens": 18800970.0, + "step": 620 + }, + { + "epoch": 0.05120804815700503, + "grad_norm": 3.305438280105591, + "learning_rate": 1.7037647705413575e-07, + "loss": 0.9328, + "mean_token_accuracy": 0.7192466259002686, + "num_tokens": 18835862.0, + "step": 621 + }, + { + "epoch": 0.05129050878205657, + "grad_norm": 3.0836708545684814, + "learning_rate": 1.706512778235779e-07, + "loss": 0.9492, + "mean_token_accuracy": 0.7160561084747314, + "num_tokens": 18871077.0, + "step": 622 + }, + { + "epoch": 0.05137296940710811, + "grad_norm": 3.030169725418091, + "learning_rate": 1.7092607859302004e-07, + "loss": 0.8178, + "mean_token_accuracy": 0.7503569722175598, + "num_tokens": 18906023.0, + "step": 623 + }, + { + "epoch": 0.051455430032159646, + "grad_norm": 3.0134081840515137, + "learning_rate": 1.7120087936246222e-07, + "loss": 0.8831, + "mean_token_accuracy": 0.7437264919281006, + "num_tokens": 18941421.0, + "step": 624 + }, + { + "epoch": 0.051537890657211184, + "grad_norm": 3.3927581310272217, + "learning_rate": 1.7147568013190436e-07, + "loss": 0.85, + "mean_token_accuracy": 0.7441840171813965, + "num_tokens": 18967898.0, + "step": 625 + }, + { + "epoch": 0.05162035128226272, + "grad_norm": 3.0239412784576416, + "learning_rate": 1.7175048090134653e-07, + "loss": 1.0193, + "mean_token_accuracy": 0.6916870474815369, + "num_tokens": 18997616.0, + "step": 626 + }, + { + "epoch": 0.05170281190731426, + "grad_norm": 3.0212342739105225, + "learning_rate": 1.7202528167078868e-07, + "loss": 1.0746, + "mean_token_accuracy": 0.6767152547836304, + "num_tokens": 19026655.0, + "step": 627 + }, + { + "epoch": 0.0517852725323658, + "grad_norm": 3.6917922496795654, + "learning_rate": 1.7230008244023085e-07, + "loss": 1.0848, + "mean_token_accuracy": 0.685280978679657, + "num_tokens": 19049039.0, + "step": 628 + }, + { + "epoch": 0.051867733157417335, + "grad_norm": 3.5011510848999023, + "learning_rate": 1.7257488320967297e-07, + "loss": 0.9967, + "mean_token_accuracy": 0.7009508609771729, + "num_tokens": 19081085.0, + "step": 629 + }, + { + "epoch": 0.05195019378246887, + "grad_norm": 3.4952194690704346, + "learning_rate": 1.7284968397911514e-07, + "loss": 1.0412, + "mean_token_accuracy": 0.6915426254272461, + "num_tokens": 19112331.0, + "step": 630 + }, + { + "epoch": 0.05203265440752041, + "grad_norm": 3.866880178451538, + "learning_rate": 1.7312448474855728e-07, + "loss": 0.943, + "mean_token_accuracy": 0.7221431732177734, + "num_tokens": 19142125.0, + "step": 631 + }, + { + "epoch": 0.05211511503257195, + "grad_norm": 3.291843891143799, + "learning_rate": 1.7339928551799946e-07, + "loss": 0.9904, + "mean_token_accuracy": 0.7061485052108765, + "num_tokens": 19178629.0, + "step": 632 + }, + { + "epoch": 0.052197575657623486, + "grad_norm": 3.3020291328430176, + "learning_rate": 1.736740862874416e-07, + "loss": 1.0554, + "mean_token_accuracy": 0.6977693438529968, + "num_tokens": 19213233.0, + "step": 633 + }, + { + "epoch": 0.052280036282675024, + "grad_norm": 3.4709742069244385, + "learning_rate": 1.7394888705688377e-07, + "loss": 0.7794, + "mean_token_accuracy": 0.7635511755943298, + "num_tokens": 19243135.0, + "step": 634 + }, + { + "epoch": 0.05236249690772656, + "grad_norm": 3.6537587642669678, + "learning_rate": 1.742236878263259e-07, + "loss": 1.0047, + "mean_token_accuracy": 0.7068238258361816, + "num_tokens": 19274310.0, + "step": 635 + }, + { + "epoch": 0.0524449575327781, + "grad_norm": 2.97763729095459, + "learning_rate": 1.7449848859576806e-07, + "loss": 1.0807, + "mean_token_accuracy": 0.6853899359703064, + "num_tokens": 19305529.0, + "step": 636 + }, + { + "epoch": 0.05252741815782964, + "grad_norm": 3.7354047298431396, + "learning_rate": 1.747732893652102e-07, + "loss": 1.0374, + "mean_token_accuracy": 0.7007061243057251, + "num_tokens": 19328000.0, + "step": 637 + }, + { + "epoch": 0.052609878782881175, + "grad_norm": 3.6869049072265625, + "learning_rate": 1.7504809013465238e-07, + "loss": 1.1057, + "mean_token_accuracy": 0.6832923293113708, + "num_tokens": 19352402.0, + "step": 638 + }, + { + "epoch": 0.05269233940793271, + "grad_norm": 3.3632657527923584, + "learning_rate": 1.7532289090409453e-07, + "loss": 0.7776, + "mean_token_accuracy": 0.752758264541626, + "num_tokens": 19387439.0, + "step": 639 + }, + { + "epoch": 0.05277480003298425, + "grad_norm": 3.782585382461548, + "learning_rate": 1.755976916735367e-07, + "loss": 0.9186, + "mean_token_accuracy": 0.7238110899925232, + "num_tokens": 19409279.0, + "step": 640 + }, + { + "epoch": 0.05285726065803579, + "grad_norm": 3.6905102729797363, + "learning_rate": 1.7587249244297884e-07, + "loss": 0.9453, + "mean_token_accuracy": 0.7178840041160583, + "num_tokens": 19436329.0, + "step": 641 + }, + { + "epoch": 0.052939721283087326, + "grad_norm": 3.2102465629577637, + "learning_rate": 1.76147293212421e-07, + "loss": 1.0033, + "mean_token_accuracy": 0.7063635587692261, + "num_tokens": 19469766.0, + "step": 642 + }, + { + "epoch": 0.053022181908138864, + "grad_norm": 2.972078323364258, + "learning_rate": 1.7642209398186313e-07, + "loss": 1.0812, + "mean_token_accuracy": 0.6921777129173279, + "num_tokens": 19500473.0, + "step": 643 + }, + { + "epoch": 0.0531046425331904, + "grad_norm": 3.064680814743042, + "learning_rate": 1.766968947513053e-07, + "loss": 0.8181, + "mean_token_accuracy": 0.7589741349220276, + "num_tokens": 19538074.0, + "step": 644 + }, + { + "epoch": 0.05318710315824194, + "grad_norm": 3.1914944648742676, + "learning_rate": 1.7697169552074745e-07, + "loss": 1.0061, + "mean_token_accuracy": 0.7065145969390869, + "num_tokens": 19565181.0, + "step": 645 + }, + { + "epoch": 0.05326956378329348, + "grad_norm": 3.2083659172058105, + "learning_rate": 1.7724649629018962e-07, + "loss": 0.9138, + "mean_token_accuracy": 0.7283425331115723, + "num_tokens": 19595243.0, + "step": 646 + }, + { + "epoch": 0.053352024408345015, + "grad_norm": 2.632481813430786, + "learning_rate": 1.7752129705963177e-07, + "loss": 0.8414, + "mean_token_accuracy": 0.7467983961105347, + "num_tokens": 19631816.0, + "step": 647 + }, + { + "epoch": 0.05343448503339655, + "grad_norm": 3.66471791267395, + "learning_rate": 1.7779609782907394e-07, + "loss": 1.1412, + "mean_token_accuracy": 0.6723348498344421, + "num_tokens": 19658621.0, + "step": 648 + }, + { + "epoch": 0.05351694565844809, + "grad_norm": 3.652658224105835, + "learning_rate": 1.7807089859851606e-07, + "loss": 1.1305, + "mean_token_accuracy": 0.6710769534111023, + "num_tokens": 19684811.0, + "step": 649 + }, + { + "epoch": 0.05359940628349963, + "grad_norm": 3.020285129547119, + "learning_rate": 1.7834569936795823e-07, + "loss": 0.7581, + "mean_token_accuracy": 0.7465521693229675, + "num_tokens": 19712737.0, + "step": 650 + }, + { + "epoch": 0.05368186690855117, + "grad_norm": 2.6368050575256348, + "learning_rate": 1.7862050013740037e-07, + "loss": 1.0174, + "mean_token_accuracy": 0.7058005332946777, + "num_tokens": 19757101.0, + "step": 651 + }, + { + "epoch": 0.053764327533602704, + "grad_norm": 3.6210811138153076, + "learning_rate": 1.7889530090684254e-07, + "loss": 1.0031, + "mean_token_accuracy": 0.7078064680099487, + "num_tokens": 19779978.0, + "step": 652 + }, + { + "epoch": 0.05384678815865424, + "grad_norm": 3.3941125869750977, + "learning_rate": 1.791701016762847e-07, + "loss": 0.9362, + "mean_token_accuracy": 0.7195713520050049, + "num_tokens": 19803898.0, + "step": 653 + }, + { + "epoch": 0.05392924878370578, + "grad_norm": 2.733492136001587, + "learning_rate": 1.7944490244572686e-07, + "loss": 0.9035, + "mean_token_accuracy": 0.7240769267082214, + "num_tokens": 19830542.0, + "step": 654 + }, + { + "epoch": 0.05401170940875732, + "grad_norm": 3.392083168029785, + "learning_rate": 1.79719703215169e-07, + "loss": 1.0043, + "mean_token_accuracy": 0.7077898383140564, + "num_tokens": 19856900.0, + "step": 655 + }, + { + "epoch": 0.054094170033808855, + "grad_norm": 3.999022960662842, + "learning_rate": 1.7999450398461115e-07, + "loss": 1.0026, + "mean_token_accuracy": 0.7060786485671997, + "num_tokens": 19881796.0, + "step": 656 + }, + { + "epoch": 0.05417663065886039, + "grad_norm": 3.341810703277588, + "learning_rate": 1.802693047540533e-07, + "loss": 0.91, + "mean_token_accuracy": 0.7254839539527893, + "num_tokens": 19907844.0, + "step": 657 + }, + { + "epoch": 0.05425909128391193, + "grad_norm": 3.16719913482666, + "learning_rate": 1.8054410552349547e-07, + "loss": 0.8778, + "mean_token_accuracy": 0.7281817197799683, + "num_tokens": 19935419.0, + "step": 658 + }, + { + "epoch": 0.05434155190896347, + "grad_norm": 3.155726671218872, + "learning_rate": 1.808189062929376e-07, + "loss": 0.9044, + "mean_token_accuracy": 0.7310791015625, + "num_tokens": 19961902.0, + "step": 659 + }, + { + "epoch": 0.05442401253401501, + "grad_norm": 3.165825366973877, + "learning_rate": 1.8109370706237978e-07, + "loss": 1.0275, + "mean_token_accuracy": 0.7031319737434387, + "num_tokens": 19989165.0, + "step": 660 + }, + { + "epoch": 0.054506473159066544, + "grad_norm": 3.309558868408203, + "learning_rate": 1.8136850783182193e-07, + "loss": 0.9207, + "mean_token_accuracy": 0.7346657514572144, + "num_tokens": 20013402.0, + "step": 661 + }, + { + "epoch": 0.05458893378411808, + "grad_norm": 2.908111572265625, + "learning_rate": 1.816433086012641e-07, + "loss": 0.9045, + "mean_token_accuracy": 0.7292393445968628, + "num_tokens": 20042188.0, + "step": 662 + }, + { + "epoch": 0.05467139440916962, + "grad_norm": 3.2005629539489746, + "learning_rate": 1.8191810937070622e-07, + "loss": 0.9222, + "mean_token_accuracy": 0.7286347150802612, + "num_tokens": 20068803.0, + "step": 663 + }, + { + "epoch": 0.05475385503422116, + "grad_norm": 3.7413389682769775, + "learning_rate": 1.8219291014014836e-07, + "loss": 1.0052, + "mean_token_accuracy": 0.697097659111023, + "num_tokens": 20091619.0, + "step": 664 + }, + { + "epoch": 0.054836315659272696, + "grad_norm": 3.249478578567505, + "learning_rate": 1.8246771090959054e-07, + "loss": 1.1952, + "mean_token_accuracy": 0.6557796001434326, + "num_tokens": 20125643.0, + "step": 665 + }, + { + "epoch": 0.05491877628432423, + "grad_norm": 2.9884884357452393, + "learning_rate": 1.8274251167903268e-07, + "loss": 0.9736, + "mean_token_accuracy": 0.7112988233566284, + "num_tokens": 20153346.0, + "step": 666 + }, + { + "epoch": 0.05500123690937577, + "grad_norm": 2.999285936355591, + "learning_rate": 1.8301731244847485e-07, + "loss": 0.9527, + "mean_token_accuracy": 0.7126274108886719, + "num_tokens": 20185319.0, + "step": 667 + }, + { + "epoch": 0.05508369753442731, + "grad_norm": 3.3095970153808594, + "learning_rate": 1.83292113217917e-07, + "loss": 1.1297, + "mean_token_accuracy": 0.671471893787384, + "num_tokens": 20209437.0, + "step": 668 + }, + { + "epoch": 0.05516615815947885, + "grad_norm": 3.080510139465332, + "learning_rate": 1.8356691398735917e-07, + "loss": 0.9972, + "mean_token_accuracy": 0.710478663444519, + "num_tokens": 20243129.0, + "step": 669 + }, + { + "epoch": 0.055248618784530384, + "grad_norm": 2.6987686157226562, + "learning_rate": 1.838417147568013e-07, + "loss": 0.8924, + "mean_token_accuracy": 0.7318994998931885, + "num_tokens": 20276435.0, + "step": 670 + }, + { + "epoch": 0.05533107940958192, + "grad_norm": 2.5070955753326416, + "learning_rate": 1.8411651552624346e-07, + "loss": 1.0025, + "mean_token_accuracy": 0.7011441588401794, + "num_tokens": 20308698.0, + "step": 671 + }, + { + "epoch": 0.05541354003463346, + "grad_norm": 2.941579818725586, + "learning_rate": 1.843913162956856e-07, + "loss": 1.0013, + "mean_token_accuracy": 0.7075217366218567, + "num_tokens": 20336710.0, + "step": 672 + }, + { + "epoch": 0.055496000659685, + "grad_norm": 3.232816457748413, + "learning_rate": 1.8466611706512778e-07, + "loss": 0.9717, + "mean_token_accuracy": 0.7080988883972168, + "num_tokens": 20363928.0, + "step": 673 + }, + { + "epoch": 0.055578461284736536, + "grad_norm": 3.009439706802368, + "learning_rate": 1.8494091783456992e-07, + "loss": 1.0133, + "mean_token_accuracy": 0.6946759819984436, + "num_tokens": 20399280.0, + "step": 674 + }, + { + "epoch": 0.05566092190978807, + "grad_norm": 2.5818779468536377, + "learning_rate": 1.852157186040121e-07, + "loss": 0.9722, + "mean_token_accuracy": 0.7170858383178711, + "num_tokens": 20434105.0, + "step": 675 + }, + { + "epoch": 0.05574338253483961, + "grad_norm": 3.6069657802581787, + "learning_rate": 1.854905193734542e-07, + "loss": 0.858, + "mean_token_accuracy": 0.7435169219970703, + "num_tokens": 20460345.0, + "step": 676 + }, + { + "epoch": 0.05582584315989115, + "grad_norm": 3.1753928661346436, + "learning_rate": 1.8576532014289638e-07, + "loss": 1.0273, + "mean_token_accuracy": 0.6854223012924194, + "num_tokens": 20486036.0, + "step": 677 + }, + { + "epoch": 0.05590830378494269, + "grad_norm": 2.6780714988708496, + "learning_rate": 1.8604012091233853e-07, + "loss": 1.0144, + "mean_token_accuracy": 0.6991140246391296, + "num_tokens": 20524086.0, + "step": 678 + }, + { + "epoch": 0.055990764409994224, + "grad_norm": 3.0523645877838135, + "learning_rate": 1.863149216817807e-07, + "loss": 0.9927, + "mean_token_accuracy": 0.7116067409515381, + "num_tokens": 20558500.0, + "step": 679 + }, + { + "epoch": 0.05607322503504576, + "grad_norm": 3.5343337059020996, + "learning_rate": 1.8658972245122284e-07, + "loss": 0.9908, + "mean_token_accuracy": 0.7201929092407227, + "num_tokens": 20588328.0, + "step": 680 + }, + { + "epoch": 0.05615568566009731, + "grad_norm": 2.8070483207702637, + "learning_rate": 1.8686452322066502e-07, + "loss": 0.885, + "mean_token_accuracy": 0.7358851432800293, + "num_tokens": 20622834.0, + "step": 681 + }, + { + "epoch": 0.056238146285148845, + "grad_norm": 2.905647039413452, + "learning_rate": 1.8713932399010716e-07, + "loss": 1.0251, + "mean_token_accuracy": 0.6968527436256409, + "num_tokens": 20651323.0, + "step": 682 + }, + { + "epoch": 0.05632060691020038, + "grad_norm": 2.770406723022461, + "learning_rate": 1.874141247595493e-07, + "loss": 1.0742, + "mean_token_accuracy": 0.6852627396583557, + "num_tokens": 20682843.0, + "step": 683 + }, + { + "epoch": 0.05640306753525192, + "grad_norm": 3.1661548614501953, + "learning_rate": 1.8768892552899145e-07, + "loss": 0.8147, + "mean_token_accuracy": 0.7517116069793701, + "num_tokens": 20713790.0, + "step": 684 + }, + { + "epoch": 0.05648552816030346, + "grad_norm": 2.477414131164551, + "learning_rate": 1.8796372629843362e-07, + "loss": 1.0049, + "mean_token_accuracy": 0.7084061503410339, + "num_tokens": 20752628.0, + "step": 685 + }, + { + "epoch": 0.056567988785354996, + "grad_norm": 2.7619388103485107, + "learning_rate": 1.8823852706787577e-07, + "loss": 0.9101, + "mean_token_accuracy": 0.7296514511108398, + "num_tokens": 20782121.0, + "step": 686 + }, + { + "epoch": 0.056650449410406534, + "grad_norm": 3.1182053089141846, + "learning_rate": 1.8851332783731794e-07, + "loss": 1.0376, + "mean_token_accuracy": 0.6976374387741089, + "num_tokens": 20809229.0, + "step": 687 + }, + { + "epoch": 0.05673291003545807, + "grad_norm": 2.8594396114349365, + "learning_rate": 1.8878812860676008e-07, + "loss": 1.0314, + "mean_token_accuracy": 0.6963785886764526, + "num_tokens": 20841299.0, + "step": 688 + }, + { + "epoch": 0.05681537066050961, + "grad_norm": 3.9006433486938477, + "learning_rate": 1.8906292937620226e-07, + "loss": 1.0246, + "mean_token_accuracy": 0.6994427442550659, + "num_tokens": 20861965.0, + "step": 689 + }, + { + "epoch": 0.05689783128556115, + "grad_norm": 3.3839335441589355, + "learning_rate": 1.8933773014564437e-07, + "loss": 0.9439, + "mean_token_accuracy": 0.7181525230407715, + "num_tokens": 20887609.0, + "step": 690 + }, + { + "epoch": 0.056980291910612685, + "grad_norm": 2.8786509037017822, + "learning_rate": 1.8961253091508655e-07, + "loss": 0.9178, + "mean_token_accuracy": 0.7363471388816833, + "num_tokens": 20916066.0, + "step": 691 + }, + { + "epoch": 0.05706275253566422, + "grad_norm": 3.105114698410034, + "learning_rate": 1.898873316845287e-07, + "loss": 1.0674, + "mean_token_accuracy": 0.6828792691230774, + "num_tokens": 20942132.0, + "step": 692 + }, + { + "epoch": 0.05714521316071576, + "grad_norm": 3.0110301971435547, + "learning_rate": 1.9016213245397086e-07, + "loss": 0.856, + "mean_token_accuracy": 0.7410976886749268, + "num_tokens": 20971286.0, + "step": 693 + }, + { + "epoch": 0.0572276737857673, + "grad_norm": 3.031477928161621, + "learning_rate": 1.90436933223413e-07, + "loss": 0.8654, + "mean_token_accuracy": 0.7389216423034668, + "num_tokens": 20998235.0, + "step": 694 + }, + { + "epoch": 0.057310134410818836, + "grad_norm": 2.839439630508423, + "learning_rate": 1.9071173399285518e-07, + "loss": 0.9086, + "mean_token_accuracy": 0.7242906093597412, + "num_tokens": 21027098.0, + "step": 695 + }, + { + "epoch": 0.057392595035870374, + "grad_norm": 2.7623491287231445, + "learning_rate": 1.9098653476229732e-07, + "loss": 1.0363, + "mean_token_accuracy": 0.6879069805145264, + "num_tokens": 21058899.0, + "step": 696 + }, + { + "epoch": 0.05747505566092191, + "grad_norm": 3.139589548110962, + "learning_rate": 1.9126133553173947e-07, + "loss": 0.9637, + "mean_token_accuracy": 0.7093906998634338, + "num_tokens": 21081642.0, + "step": 697 + }, + { + "epoch": 0.05755751628597345, + "grad_norm": 2.977581739425659, + "learning_rate": 1.9153613630118162e-07, + "loss": 1.0759, + "mean_token_accuracy": 0.6865653991699219, + "num_tokens": 21109081.0, + "step": 698 + }, + { + "epoch": 0.05763997691102499, + "grad_norm": 3.292327880859375, + "learning_rate": 1.9181093707062379e-07, + "loss": 0.9998, + "mean_token_accuracy": 0.7104979157447815, + "num_tokens": 21135653.0, + "step": 699 + }, + { + "epoch": 0.057722437536076525, + "grad_norm": 2.63094162940979, + "learning_rate": 1.9208573784006593e-07, + "loss": 0.9258, + "mean_token_accuracy": 0.7225830554962158, + "num_tokens": 21164281.0, + "step": 700 + }, + { + "epoch": 0.05780489816112806, + "grad_norm": 2.33884596824646, + "learning_rate": 1.923605386095081e-07, + "loss": 1.0549, + "mean_token_accuracy": 0.6998014450073242, + "num_tokens": 21203454.0, + "step": 701 + }, + { + "epoch": 0.0578873587861796, + "grad_norm": 2.4289498329162598, + "learning_rate": 1.9263533937895025e-07, + "loss": 1.0945, + "mean_token_accuracy": 0.6825656890869141, + "num_tokens": 21243376.0, + "step": 702 + }, + { + "epoch": 0.05796981941123114, + "grad_norm": 2.847536563873291, + "learning_rate": 1.9291014014839242e-07, + "loss": 0.8804, + "mean_token_accuracy": 0.7324576377868652, + "num_tokens": 21271722.0, + "step": 703 + }, + { + "epoch": 0.058052280036282676, + "grad_norm": 3.2178449630737305, + "learning_rate": 1.9318494091783454e-07, + "loss": 0.9508, + "mean_token_accuracy": 0.72232985496521, + "num_tokens": 21295661.0, + "step": 704 + }, + { + "epoch": 0.058134740661334214, + "grad_norm": 2.7960333824157715, + "learning_rate": 1.934597416872767e-07, + "loss": 0.9952, + "mean_token_accuracy": 0.7040393948554993, + "num_tokens": 21324003.0, + "step": 705 + }, + { + "epoch": 0.05821720128638575, + "grad_norm": 2.570627212524414, + "learning_rate": 1.9373454245671886e-07, + "loss": 1.0052, + "mean_token_accuracy": 0.7036068439483643, + "num_tokens": 21358071.0, + "step": 706 + }, + { + "epoch": 0.05829966191143729, + "grad_norm": 2.9628026485443115, + "learning_rate": 1.9400934322616103e-07, + "loss": 0.9027, + "mean_token_accuracy": 0.7169076204299927, + "num_tokens": 21386722.0, + "step": 707 + }, + { + "epoch": 0.05838212253648883, + "grad_norm": 2.75140118598938, + "learning_rate": 1.9428414399560317e-07, + "loss": 0.9834, + "mean_token_accuracy": 0.7078765630722046, + "num_tokens": 21414453.0, + "step": 708 + }, + { + "epoch": 0.058464583161540365, + "grad_norm": 2.6935372352600098, + "learning_rate": 1.9455894476504534e-07, + "loss": 0.997, + "mean_token_accuracy": 0.7128314971923828, + "num_tokens": 21446858.0, + "step": 709 + }, + { + "epoch": 0.0585470437865919, + "grad_norm": 2.634154796600342, + "learning_rate": 1.948337455344875e-07, + "loss": 0.9686, + "mean_token_accuracy": 0.7083452343940735, + "num_tokens": 21476748.0, + "step": 710 + }, + { + "epoch": 0.05862950441164344, + "grad_norm": 2.3967225551605225, + "learning_rate": 1.9510854630392963e-07, + "loss": 1.0006, + "mean_token_accuracy": 0.6992347240447998, + "num_tokens": 21515106.0, + "step": 711 + }, + { + "epoch": 0.05871196503669498, + "grad_norm": 2.6706762313842773, + "learning_rate": 1.9538334707337178e-07, + "loss": 1.0553, + "mean_token_accuracy": 0.6997218132019043, + "num_tokens": 21547561.0, + "step": 712 + }, + { + "epoch": 0.058794425661746516, + "grad_norm": 2.5602996349334717, + "learning_rate": 1.9565814784281395e-07, + "loss": 1.126, + "mean_token_accuracy": 0.6755434274673462, + "num_tokens": 21577602.0, + "step": 713 + }, + { + "epoch": 0.058876886286798054, + "grad_norm": 2.604144334793091, + "learning_rate": 1.959329486122561e-07, + "loss": 1.1803, + "mean_token_accuracy": 0.6656613945960999, + "num_tokens": 21611406.0, + "step": 714 + }, + { + "epoch": 0.05895934691184959, + "grad_norm": 3.191283702850342, + "learning_rate": 1.9620774938169827e-07, + "loss": 0.9986, + "mean_token_accuracy": 0.7009061574935913, + "num_tokens": 21634692.0, + "step": 715 + }, + { + "epoch": 0.05904180753690113, + "grad_norm": 3.1561639308929443, + "learning_rate": 1.964825501511404e-07, + "loss": 0.927, + "mean_token_accuracy": 0.7200424671173096, + "num_tokens": 21664424.0, + "step": 716 + }, + { + "epoch": 0.05912426816195267, + "grad_norm": 2.7290146350860596, + "learning_rate": 1.9675735092058258e-07, + "loss": 0.7892, + "mean_token_accuracy": 0.7553091049194336, + "num_tokens": 21693827.0, + "step": 717 + }, + { + "epoch": 0.059206728787004205, + "grad_norm": 2.482792615890503, + "learning_rate": 1.970321516900247e-07, + "loss": 0.8444, + "mean_token_accuracy": 0.7481518983840942, + "num_tokens": 21725394.0, + "step": 718 + }, + { + "epoch": 0.05928918941205574, + "grad_norm": 2.5337164402008057, + "learning_rate": 1.9730695245946687e-07, + "loss": 0.8105, + "mean_token_accuracy": 0.7444206476211548, + "num_tokens": 21763397.0, + "step": 719 + }, + { + "epoch": 0.05937165003710728, + "grad_norm": 2.475869655609131, + "learning_rate": 1.9758175322890902e-07, + "loss": 0.9767, + "mean_token_accuracy": 0.7085509300231934, + "num_tokens": 21794910.0, + "step": 720 + }, + { + "epoch": 0.05945411066215882, + "grad_norm": 2.6831984519958496, + "learning_rate": 1.978565539983512e-07, + "loss": 0.8826, + "mean_token_accuracy": 0.7394924163818359, + "num_tokens": 21824333.0, + "step": 721 + }, + { + "epoch": 0.059536571287210356, + "grad_norm": 3.0281498432159424, + "learning_rate": 1.9813135476779334e-07, + "loss": 0.9533, + "mean_token_accuracy": 0.7140673398971558, + "num_tokens": 21850530.0, + "step": 722 + }, + { + "epoch": 0.059619031912261894, + "grad_norm": 3.0260071754455566, + "learning_rate": 1.984061555372355e-07, + "loss": 1.1097, + "mean_token_accuracy": 0.6777607202529907, + "num_tokens": 21876185.0, + "step": 723 + }, + { + "epoch": 0.05970149253731343, + "grad_norm": 3.2115328311920166, + "learning_rate": 1.9868095630667763e-07, + "loss": 0.9803, + "mean_token_accuracy": 0.6972770690917969, + "num_tokens": 21897754.0, + "step": 724 + }, + { + "epoch": 0.05978395316236497, + "grad_norm": 2.8682940006256104, + "learning_rate": 1.989557570761198e-07, + "loss": 1.0591, + "mean_token_accuracy": 0.6838381290435791, + "num_tokens": 21924458.0, + "step": 725 + }, + { + "epoch": 0.05986641378741651, + "grad_norm": 3.002315044403076, + "learning_rate": 1.9923055784556194e-07, + "loss": 1.0626, + "mean_token_accuracy": 0.6856480836868286, + "num_tokens": 21950682.0, + "step": 726 + }, + { + "epoch": 0.059948874412468045, + "grad_norm": 2.9071385860443115, + "learning_rate": 1.9950535861500411e-07, + "loss": 0.8429, + "mean_token_accuracy": 0.7393008470535278, + "num_tokens": 21977789.0, + "step": 727 + }, + { + "epoch": 0.06003133503751958, + "grad_norm": 2.5601327419281006, + "learning_rate": 1.9978015938444626e-07, + "loss": 0.9407, + "mean_token_accuracy": 0.7287883758544922, + "num_tokens": 22013109.0, + "step": 728 + }, + { + "epoch": 0.06011379566257112, + "grad_norm": 2.428555727005005, + "learning_rate": 2.0005496015388843e-07, + "loss": 0.8857, + "mean_token_accuracy": 0.7381727695465088, + "num_tokens": 22044108.0, + "step": 729 + }, + { + "epoch": 0.06019625628762266, + "grad_norm": 2.7490813732147217, + "learning_rate": 2.0032976092333058e-07, + "loss": 0.9336, + "mean_token_accuracy": 0.7257232069969177, + "num_tokens": 22078229.0, + "step": 730 + }, + { + "epoch": 0.060278716912674196, + "grad_norm": 2.7297425270080566, + "learning_rate": 2.0060456169277272e-07, + "loss": 0.952, + "mean_token_accuracy": 0.7263073921203613, + "num_tokens": 22105884.0, + "step": 731 + }, + { + "epoch": 0.060361177537725734, + "grad_norm": 2.4667952060699463, + "learning_rate": 2.0087936246221487e-07, + "loss": 0.8064, + "mean_token_accuracy": 0.7521403431892395, + "num_tokens": 22142357.0, + "step": 732 + }, + { + "epoch": 0.06044363816277727, + "grad_norm": 2.424875497817993, + "learning_rate": 2.0115416323165704e-07, + "loss": 1.1208, + "mean_token_accuracy": 0.6731448173522949, + "num_tokens": 22178562.0, + "step": 733 + }, + { + "epoch": 0.06052609878782881, + "grad_norm": 2.7157137393951416, + "learning_rate": 2.0142896400109918e-07, + "loss": 0.9495, + "mean_token_accuracy": 0.703540027141571, + "num_tokens": 22208264.0, + "step": 734 + }, + { + "epoch": 0.06060855941288035, + "grad_norm": 2.3838205337524414, + "learning_rate": 2.0170376477054135e-07, + "loss": 0.9301, + "mean_token_accuracy": 0.721613883972168, + "num_tokens": 22243612.0, + "step": 735 + }, + { + "epoch": 0.060691020037931885, + "grad_norm": 2.418247699737549, + "learning_rate": 2.019785655399835e-07, + "loss": 0.9398, + "mean_token_accuracy": 0.7334630489349365, + "num_tokens": 22280366.0, + "step": 736 + }, + { + "epoch": 0.06077348066298342, + "grad_norm": 2.727174997329712, + "learning_rate": 2.0225336630942567e-07, + "loss": 1.0185, + "mean_token_accuracy": 0.7011879682540894, + "num_tokens": 22310202.0, + "step": 737 + }, + { + "epoch": 0.06085594128803496, + "grad_norm": 2.668088674545288, + "learning_rate": 2.025281670788678e-07, + "loss": 0.9239, + "mean_token_accuracy": 0.7248497605323792, + "num_tokens": 22339672.0, + "step": 738 + }, + { + "epoch": 0.0609384019130865, + "grad_norm": 3.1732985973358154, + "learning_rate": 2.0280296784830996e-07, + "loss": 1.1414, + "mean_token_accuracy": 0.6659507155418396, + "num_tokens": 22363996.0, + "step": 739 + }, + { + "epoch": 0.061020862538138036, + "grad_norm": 2.3413195610046387, + "learning_rate": 2.030777686177521e-07, + "loss": 0.9437, + "mean_token_accuracy": 0.7116901278495789, + "num_tokens": 22397269.0, + "step": 740 + }, + { + "epoch": 0.061103323163189574, + "grad_norm": 2.801759719848633, + "learning_rate": 2.0335256938719428e-07, + "loss": 0.8479, + "mean_token_accuracy": 0.7427554130554199, + "num_tokens": 22422733.0, + "step": 741 + }, + { + "epoch": 0.06118578378824111, + "grad_norm": 3.4697248935699463, + "learning_rate": 2.0362737015663642e-07, + "loss": 1.0796, + "mean_token_accuracy": 0.6809610724449158, + "num_tokens": 22447407.0, + "step": 742 + }, + { + "epoch": 0.06126824441329265, + "grad_norm": 3.826723098754883, + "learning_rate": 2.039021709260786e-07, + "loss": 1.0824, + "mean_token_accuracy": 0.685306191444397, + "num_tokens": 22468940.0, + "step": 743 + }, + { + "epoch": 0.06135070503834419, + "grad_norm": 2.4207828044891357, + "learning_rate": 2.0417697169552074e-07, + "loss": 0.8677, + "mean_token_accuracy": 0.7338758707046509, + "num_tokens": 22501639.0, + "step": 744 + }, + { + "epoch": 0.061433165663395725, + "grad_norm": 2.6534922122955322, + "learning_rate": 2.0445177246496288e-07, + "loss": 0.9684, + "mean_token_accuracy": 0.7150545120239258, + "num_tokens": 22530977.0, + "step": 745 + }, + { + "epoch": 0.06151562628844727, + "grad_norm": 2.5508968830108643, + "learning_rate": 2.0472657323440503e-07, + "loss": 0.9786, + "mean_token_accuracy": 0.7108112573623657, + "num_tokens": 22562952.0, + "step": 746 + }, + { + "epoch": 0.06159808691349881, + "grad_norm": 3.0700771808624268, + "learning_rate": 2.050013740038472e-07, + "loss": 0.9304, + "mean_token_accuracy": 0.7171639204025269, + "num_tokens": 22587505.0, + "step": 747 + }, + { + "epoch": 0.061680547538550345, + "grad_norm": 2.544412612915039, + "learning_rate": 2.0527617477328935e-07, + "loss": 1.1823, + "mean_token_accuracy": 0.6586083769798279, + "num_tokens": 22621013.0, + "step": 748 + }, + { + "epoch": 0.06176300816360188, + "grad_norm": 2.9449703693389893, + "learning_rate": 2.0555097554273152e-07, + "loss": 1.0002, + "mean_token_accuracy": 0.7096898555755615, + "num_tokens": 22646227.0, + "step": 749 + }, + { + "epoch": 0.06184546878865342, + "grad_norm": 2.7871897220611572, + "learning_rate": 2.0582577631217366e-07, + "loss": 1.0669, + "mean_token_accuracy": 0.6847690343856812, + "num_tokens": 22676878.0, + "step": 750 + }, + { + "epoch": 0.06192792941370496, + "grad_norm": 2.749694347381592, + "learning_rate": 2.0610057708161583e-07, + "loss": 1.0041, + "mean_token_accuracy": 0.7022327184677124, + "num_tokens": 22704873.0, + "step": 751 + }, + { + "epoch": 0.0620103900387565, + "grad_norm": 2.781615734100342, + "learning_rate": 2.0637537785105795e-07, + "loss": 1.0608, + "mean_token_accuracy": 0.6949086785316467, + "num_tokens": 22732690.0, + "step": 752 + }, + { + "epoch": 0.062092850663808034, + "grad_norm": 2.7087512016296387, + "learning_rate": 2.0665017862050012e-07, + "loss": 1.0061, + "mean_token_accuracy": 0.7008106708526611, + "num_tokens": 22761010.0, + "step": 753 + }, + { + "epoch": 0.06217531128885957, + "grad_norm": 2.323737382888794, + "learning_rate": 2.0692497938994227e-07, + "loss": 0.8835, + "mean_token_accuracy": 0.729609489440918, + "num_tokens": 22797011.0, + "step": 754 + }, + { + "epoch": 0.06225777191391111, + "grad_norm": 2.669870376586914, + "learning_rate": 2.0719978015938444e-07, + "loss": 0.8018, + "mean_token_accuracy": 0.7569133043289185, + "num_tokens": 22836152.0, + "step": 755 + }, + { + "epoch": 0.06234023253896265, + "grad_norm": 2.7029218673706055, + "learning_rate": 2.0747458092882659e-07, + "loss": 1.0525, + "mean_token_accuracy": 0.691249668598175, + "num_tokens": 22864514.0, + "step": 756 + }, + { + "epoch": 0.062422693164014185, + "grad_norm": 2.7310545444488525, + "learning_rate": 2.0774938169826876e-07, + "loss": 1.1725, + "mean_token_accuracy": 0.6674242615699768, + "num_tokens": 22891290.0, + "step": 757 + }, + { + "epoch": 0.06250515378906572, + "grad_norm": 2.341264009475708, + "learning_rate": 2.080241824677109e-07, + "loss": 0.7805, + "mean_token_accuracy": 0.7639495730400085, + "num_tokens": 22921344.0, + "step": 758 + }, + { + "epoch": 0.06258761441411725, + "grad_norm": 2.565513849258423, + "learning_rate": 2.0829898323715305e-07, + "loss": 1.1295, + "mean_token_accuracy": 0.6732931733131409, + "num_tokens": 22951326.0, + "step": 759 + }, + { + "epoch": 0.0626700750391688, + "grad_norm": 2.7447752952575684, + "learning_rate": 2.085737840065952e-07, + "loss": 1.1359, + "mean_token_accuracy": 0.6720055937767029, + "num_tokens": 22979569.0, + "step": 760 + }, + { + "epoch": 0.06275253566422033, + "grad_norm": 2.737974166870117, + "learning_rate": 2.0884858477603736e-07, + "loss": 0.9654, + "mean_token_accuracy": 0.7131686210632324, + "num_tokens": 23006401.0, + "step": 761 + }, + { + "epoch": 0.06283499628927187, + "grad_norm": 2.1480178833007812, + "learning_rate": 2.091233855454795e-07, + "loss": 0.8716, + "mean_token_accuracy": 0.7364578247070312, + "num_tokens": 23043959.0, + "step": 762 + }, + { + "epoch": 0.0629174569143234, + "grad_norm": 2.5878102779388428, + "learning_rate": 2.0939818631492168e-07, + "loss": 0.9459, + "mean_token_accuracy": 0.7213497161865234, + "num_tokens": 23073063.0, + "step": 763 + }, + { + "epoch": 0.06299991753937495, + "grad_norm": 2.3021931648254395, + "learning_rate": 2.0967298708436383e-07, + "loss": 0.8707, + "mean_token_accuracy": 0.7385721206665039, + "num_tokens": 23107464.0, + "step": 764 + }, + { + "epoch": 0.06308237816442648, + "grad_norm": 2.572007179260254, + "learning_rate": 2.09947787853806e-07, + "loss": 0.9628, + "mean_token_accuracy": 0.7194913029670715, + "num_tokens": 23137183.0, + "step": 765 + }, + { + "epoch": 0.06316483878947803, + "grad_norm": 2.910667896270752, + "learning_rate": 2.1022258862324812e-07, + "loss": 1.0271, + "mean_token_accuracy": 0.6927753686904907, + "num_tokens": 23161617.0, + "step": 766 + }, + { + "epoch": 0.06324729941452956, + "grad_norm": 2.4953320026397705, + "learning_rate": 2.104973893926903e-07, + "loss": 0.9887, + "mean_token_accuracy": 0.7087677717208862, + "num_tokens": 23193824.0, + "step": 767 + }, + { + "epoch": 0.0633297600395811, + "grad_norm": 2.5179495811462402, + "learning_rate": 2.1077219016213243e-07, + "loss": 1.0898, + "mean_token_accuracy": 0.6908871531486511, + "num_tokens": 23224133.0, + "step": 768 + }, + { + "epoch": 0.06341222066463263, + "grad_norm": 2.2466607093811035, + "learning_rate": 2.110469909315746e-07, + "loss": 0.8927, + "mean_token_accuracy": 0.7277114987373352, + "num_tokens": 23260634.0, + "step": 769 + }, + { + "epoch": 0.06349468128968418, + "grad_norm": 2.3769004344940186, + "learning_rate": 2.1132179170101675e-07, + "loss": 0.8919, + "mean_token_accuracy": 0.7277786135673523, + "num_tokens": 23292320.0, + "step": 770 + }, + { + "epoch": 0.06357714191473571, + "grad_norm": 2.3198082447052, + "learning_rate": 2.1159659247045892e-07, + "loss": 1.0699, + "mean_token_accuracy": 0.6946768760681152, + "num_tokens": 23329827.0, + "step": 771 + }, + { + "epoch": 0.06365960253978725, + "grad_norm": 2.279608964920044, + "learning_rate": 2.1187139323990104e-07, + "loss": 0.8897, + "mean_token_accuracy": 0.7306249141693115, + "num_tokens": 23363924.0, + "step": 772 + }, + { + "epoch": 0.06374206316483878, + "grad_norm": 2.278999090194702, + "learning_rate": 2.121461940093432e-07, + "loss": 0.7856, + "mean_token_accuracy": 0.7629410028457642, + "num_tokens": 23400243.0, + "step": 773 + }, + { + "epoch": 0.06382452378989033, + "grad_norm": 2.1711227893829346, + "learning_rate": 2.1242099477878536e-07, + "loss": 0.927, + "mean_token_accuracy": 0.7233618497848511, + "num_tokens": 23435818.0, + "step": 774 + }, + { + "epoch": 0.06390698441494186, + "grad_norm": 2.616701364517212, + "learning_rate": 2.1269579554822753e-07, + "loss": 0.7798, + "mean_token_accuracy": 0.757564902305603, + "num_tokens": 23469784.0, + "step": 775 + }, + { + "epoch": 0.0639894450399934, + "grad_norm": 2.8759913444519043, + "learning_rate": 2.1297059631766967e-07, + "loss": 1.018, + "mean_token_accuracy": 0.7026112079620361, + "num_tokens": 23499517.0, + "step": 776 + }, + { + "epoch": 0.06407190566504493, + "grad_norm": 2.660353899002075, + "learning_rate": 2.1324539708711185e-07, + "loss": 1.0612, + "mean_token_accuracy": 0.6850818395614624, + "num_tokens": 23530233.0, + "step": 777 + }, + { + "epoch": 0.06415436629009648, + "grad_norm": 2.5558841228485107, + "learning_rate": 2.13520197856554e-07, + "loss": 0.9972, + "mean_token_accuracy": 0.7065430879592896, + "num_tokens": 23559811.0, + "step": 778 + }, + { + "epoch": 0.06423682691514802, + "grad_norm": 2.4326059818267822, + "learning_rate": 2.1379499862599614e-07, + "loss": 0.9671, + "mean_token_accuracy": 0.7098103761672974, + "num_tokens": 23594528.0, + "step": 779 + }, + { + "epoch": 0.06431928754019955, + "grad_norm": 2.23271107673645, + "learning_rate": 2.1406979939543828e-07, + "loss": 0.856, + "mean_token_accuracy": 0.7482224702835083, + "num_tokens": 23627406.0, + "step": 780 + }, + { + "epoch": 0.0644017481652511, + "grad_norm": 2.7080535888671875, + "learning_rate": 2.1434460016488045e-07, + "loss": 1.0194, + "mean_token_accuracy": 0.7048466205596924, + "num_tokens": 23654838.0, + "step": 781 + }, + { + "epoch": 0.06448420879030263, + "grad_norm": 2.560786724090576, + "learning_rate": 2.146194009343226e-07, + "loss": 1.0648, + "mean_token_accuracy": 0.6924061179161072, + "num_tokens": 23686823.0, + "step": 782 + }, + { + "epoch": 0.06456666941535417, + "grad_norm": 2.6201326847076416, + "learning_rate": 2.1489420170376477e-07, + "loss": 0.8284, + "mean_token_accuracy": 0.7497735023498535, + "num_tokens": 23712363.0, + "step": 783 + }, + { + "epoch": 0.0646491300404057, + "grad_norm": 2.345917224884033, + "learning_rate": 2.1516900247320691e-07, + "loss": 0.9927, + "mean_token_accuracy": 0.7078331112861633, + "num_tokens": 23747388.0, + "step": 784 + }, + { + "epoch": 0.06473159066545725, + "grad_norm": 2.47448468208313, + "learning_rate": 2.1544380324264909e-07, + "loss": 0.8586, + "mean_token_accuracy": 0.7278915643692017, + "num_tokens": 23774628.0, + "step": 785 + }, + { + "epoch": 0.06481405129050878, + "grad_norm": 2.560152530670166, + "learning_rate": 2.157186040120912e-07, + "loss": 0.9033, + "mean_token_accuracy": 0.7281396389007568, + "num_tokens": 23801750.0, + "step": 786 + }, + { + "epoch": 0.06489651191556033, + "grad_norm": 2.1750707626342773, + "learning_rate": 2.1599340478153338e-07, + "loss": 1.139, + "mean_token_accuracy": 0.6719571352005005, + "num_tokens": 23839226.0, + "step": 787 + }, + { + "epoch": 0.06497897254061186, + "grad_norm": 2.4111766815185547, + "learning_rate": 2.1626820555097552e-07, + "loss": 0.9615, + "mean_token_accuracy": 0.7115905284881592, + "num_tokens": 23870187.0, + "step": 788 + }, + { + "epoch": 0.0650614331656634, + "grad_norm": 2.419278621673584, + "learning_rate": 2.165430063204177e-07, + "loss": 0.9577, + "mean_token_accuracy": 0.707419216632843, + "num_tokens": 23901887.0, + "step": 789 + }, + { + "epoch": 0.06514389379071493, + "grad_norm": 3.000333547592163, + "learning_rate": 2.1681780708985984e-07, + "loss": 1.0549, + "mean_token_accuracy": 0.691678524017334, + "num_tokens": 23923284.0, + "step": 790 + }, + { + "epoch": 0.06522635441576648, + "grad_norm": 2.2831480503082275, + "learning_rate": 2.17092607859302e-07, + "loss": 1.0185, + "mean_token_accuracy": 0.6989438533782959, + "num_tokens": 23958694.0, + "step": 791 + }, + { + "epoch": 0.06530881504081801, + "grad_norm": 2.3745908737182617, + "learning_rate": 2.1736740862874415e-07, + "loss": 0.9595, + "mean_token_accuracy": 0.7147995829582214, + "num_tokens": 23989260.0, + "step": 792 + }, + { + "epoch": 0.06539127566586955, + "grad_norm": 2.7144670486450195, + "learning_rate": 2.176422093981863e-07, + "loss": 0.9632, + "mean_token_accuracy": 0.7160773277282715, + "num_tokens": 24016320.0, + "step": 793 + }, + { + "epoch": 0.06547373629092108, + "grad_norm": 2.697114944458008, + "learning_rate": 2.1791701016762844e-07, + "loss": 0.9736, + "mean_token_accuracy": 0.7049291133880615, + "num_tokens": 24040491.0, + "step": 794 + }, + { + "epoch": 0.06555619691597263, + "grad_norm": 2.8116228580474854, + "learning_rate": 2.1819181093707062e-07, + "loss": 1.0201, + "mean_token_accuracy": 0.7032392024993896, + "num_tokens": 24065662.0, + "step": 795 + }, + { + "epoch": 0.06563865754102416, + "grad_norm": 2.588658571243286, + "learning_rate": 2.1846661170651276e-07, + "loss": 0.894, + "mean_token_accuracy": 0.7308117151260376, + "num_tokens": 24090741.0, + "step": 796 + }, + { + "epoch": 0.0657211181660757, + "grad_norm": 2.6994950771331787, + "learning_rate": 2.1874141247595493e-07, + "loss": 1.0563, + "mean_token_accuracy": 0.6909680366516113, + "num_tokens": 24117269.0, + "step": 797 + }, + { + "epoch": 0.06580357879112723, + "grad_norm": 3.105729818344116, + "learning_rate": 2.1901621324539708e-07, + "loss": 0.9547, + "mean_token_accuracy": 0.7090911269187927, + "num_tokens": 24136755.0, + "step": 798 + }, + { + "epoch": 0.06588603941617878, + "grad_norm": 2.1143746376037598, + "learning_rate": 2.1929101401483925e-07, + "loss": 0.8631, + "mean_token_accuracy": 0.7337855696678162, + "num_tokens": 24172858.0, + "step": 799 + }, + { + "epoch": 0.06596850004123031, + "grad_norm": 2.2891125679016113, + "learning_rate": 2.1956581478428137e-07, + "loss": 0.93, + "mean_token_accuracy": 0.7324391007423401, + "num_tokens": 24208817.0, + "step": 800 + }, + { + "epoch": 0.06605096066628185, + "grad_norm": 2.6571152210235596, + "learning_rate": 2.1984061555372354e-07, + "loss": 0.8237, + "mean_token_accuracy": 0.7520148158073425, + "num_tokens": 24234417.0, + "step": 801 + }, + { + "epoch": 0.06613342129133339, + "grad_norm": 2.480447769165039, + "learning_rate": 2.2011541632316568e-07, + "loss": 1.0052, + "mean_token_accuracy": 0.70771324634552, + "num_tokens": 24267098.0, + "step": 802 + }, + { + "epoch": 0.06621588191638493, + "grad_norm": 2.822566270828247, + "learning_rate": 2.2039021709260786e-07, + "loss": 1.054, + "mean_token_accuracy": 0.6961683034896851, + "num_tokens": 24296637.0, + "step": 803 + }, + { + "epoch": 0.06629834254143646, + "grad_norm": 2.1249687671661377, + "learning_rate": 2.2066501786205e-07, + "loss": 0.8676, + "mean_token_accuracy": 0.7500913143157959, + "num_tokens": 24334280.0, + "step": 804 + }, + { + "epoch": 0.066380803166488, + "grad_norm": 2.6733007431030273, + "learning_rate": 2.2093981863149217e-07, + "loss": 0.9929, + "mean_token_accuracy": 0.7068202495574951, + "num_tokens": 24362168.0, + "step": 805 + }, + { + "epoch": 0.06646326379153954, + "grad_norm": 2.8574893474578857, + "learning_rate": 2.2121461940093432e-07, + "loss": 1.1754, + "mean_token_accuracy": 0.6702730655670166, + "num_tokens": 24388592.0, + "step": 806 + }, + { + "epoch": 0.06654572441659108, + "grad_norm": 2.8796651363372803, + "learning_rate": 2.2148942017037646e-07, + "loss": 1.0036, + "mean_token_accuracy": 0.703001081943512, + "num_tokens": 24413007.0, + "step": 807 + }, + { + "epoch": 0.06662818504164261, + "grad_norm": 2.5913898944854736, + "learning_rate": 2.217642209398186e-07, + "loss": 0.899, + "mean_token_accuracy": 0.7281142473220825, + "num_tokens": 24442784.0, + "step": 808 + }, + { + "epoch": 0.06671064566669416, + "grad_norm": 2.8602254390716553, + "learning_rate": 2.2203902170926078e-07, + "loss": 0.9267, + "mean_token_accuracy": 0.7115074396133423, + "num_tokens": 24466019.0, + "step": 809 + }, + { + "epoch": 0.06679310629174569, + "grad_norm": 2.4972476959228516, + "learning_rate": 2.2231382247870292e-07, + "loss": 0.8525, + "mean_token_accuracy": 0.7444584965705872, + "num_tokens": 24495943.0, + "step": 810 + }, + { + "epoch": 0.06687556691679723, + "grad_norm": 2.5217032432556152, + "learning_rate": 2.225886232481451e-07, + "loss": 0.9103, + "mean_token_accuracy": 0.7283675670623779, + "num_tokens": 24522317.0, + "step": 811 + }, + { + "epoch": 0.06695802754184876, + "grad_norm": 2.257963180541992, + "learning_rate": 2.2286342401758724e-07, + "loss": 0.8974, + "mean_token_accuracy": 0.724678099155426, + "num_tokens": 24553598.0, + "step": 812 + }, + { + "epoch": 0.06704048816690031, + "grad_norm": 2.447312116622925, + "learning_rate": 2.231382247870294e-07, + "loss": 0.8708, + "mean_token_accuracy": 0.7367665767669678, + "num_tokens": 24585050.0, + "step": 813 + }, + { + "epoch": 0.06712294879195184, + "grad_norm": 2.4575746059417725, + "learning_rate": 2.2341302555647153e-07, + "loss": 0.8576, + "mean_token_accuracy": 0.737172544002533, + "num_tokens": 24615716.0, + "step": 814 + }, + { + "epoch": 0.06720540941700338, + "grad_norm": 2.4026241302490234, + "learning_rate": 2.236878263259137e-07, + "loss": 0.8315, + "mean_token_accuracy": 0.7420417070388794, + "num_tokens": 24644570.0, + "step": 815 + }, + { + "epoch": 0.06728787004205491, + "grad_norm": 2.469097375869751, + "learning_rate": 2.2396262709535585e-07, + "loss": 1.0166, + "mean_token_accuracy": 0.6907176971435547, + "num_tokens": 24672215.0, + "step": 816 + }, + { + "epoch": 0.06737033066710646, + "grad_norm": 2.553290605545044, + "learning_rate": 2.2423742786479802e-07, + "loss": 1.0329, + "mean_token_accuracy": 0.6928033828735352, + "num_tokens": 24697596.0, + "step": 817 + }, + { + "epoch": 0.06745279129215799, + "grad_norm": 2.709857940673828, + "learning_rate": 2.2451222863424016e-07, + "loss": 0.9843, + "mean_token_accuracy": 0.7124001383781433, + "num_tokens": 24722751.0, + "step": 818 + }, + { + "epoch": 0.06753525191720954, + "grad_norm": 2.4306085109710693, + "learning_rate": 2.2478702940368234e-07, + "loss": 0.7959, + "mean_token_accuracy": 0.7616783976554871, + "num_tokens": 24754179.0, + "step": 819 + }, + { + "epoch": 0.06761771254226107, + "grad_norm": 2.0355019569396973, + "learning_rate": 2.2506183017312445e-07, + "loss": 0.8525, + "mean_token_accuracy": 0.7420663237571716, + "num_tokens": 24793192.0, + "step": 820 + }, + { + "epoch": 0.06770017316731261, + "grad_norm": 2.334256887435913, + "learning_rate": 2.2533663094256663e-07, + "loss": 0.854, + "mean_token_accuracy": 0.7437018752098083, + "num_tokens": 24823389.0, + "step": 821 + }, + { + "epoch": 0.06778263379236414, + "grad_norm": 2.3246257305145264, + "learning_rate": 2.2561143171200877e-07, + "loss": 0.8562, + "mean_token_accuracy": 0.727616012096405, + "num_tokens": 24853669.0, + "step": 822 + }, + { + "epoch": 0.06786509441741569, + "grad_norm": 2.249109983444214, + "learning_rate": 2.2588623248145094e-07, + "loss": 0.7888, + "mean_token_accuracy": 0.7608472108840942, + "num_tokens": 24887314.0, + "step": 823 + }, + { + "epoch": 0.06794755504246722, + "grad_norm": 2.8213860988616943, + "learning_rate": 2.261610332508931e-07, + "loss": 1.0235, + "mean_token_accuracy": 0.6985840201377869, + "num_tokens": 24910771.0, + "step": 824 + }, + { + "epoch": 0.06803001566751876, + "grad_norm": 2.7699389457702637, + "learning_rate": 2.2643583402033526e-07, + "loss": 0.8905, + "mean_token_accuracy": 0.7293692827224731, + "num_tokens": 24933587.0, + "step": 825 + }, + { + "epoch": 0.06811247629257029, + "grad_norm": 2.474755048751831, + "learning_rate": 2.267106347897774e-07, + "loss": 1.0123, + "mean_token_accuracy": 0.6982953548431396, + "num_tokens": 24966780.0, + "step": 826 + }, + { + "epoch": 0.06819493691762184, + "grad_norm": 2.945997953414917, + "learning_rate": 2.2698543555921955e-07, + "loss": 0.9883, + "mean_token_accuracy": 0.6956835985183716, + "num_tokens": 25001699.0, + "step": 827 + }, + { + "epoch": 0.06827739754267337, + "grad_norm": 2.496234893798828, + "learning_rate": 2.272602363286617e-07, + "loss": 1.0888, + "mean_token_accuracy": 0.6810837984085083, + "num_tokens": 25031991.0, + "step": 828 + }, + { + "epoch": 0.06835985816772491, + "grad_norm": 2.045482635498047, + "learning_rate": 2.2753503709810387e-07, + "loss": 0.9889, + "mean_token_accuracy": 0.7041285634040833, + "num_tokens": 25073068.0, + "step": 829 + }, + { + "epoch": 0.06844231879277644, + "grad_norm": 2.2126941680908203, + "learning_rate": 2.27809837867546e-07, + "loss": 0.9802, + "mean_token_accuracy": 0.7110644578933716, + "num_tokens": 25108288.0, + "step": 830 + }, + { + "epoch": 0.06852477941782799, + "grad_norm": 2.2696738243103027, + "learning_rate": 2.2808463863698818e-07, + "loss": 0.9183, + "mean_token_accuracy": 0.7207636833190918, + "num_tokens": 25138164.0, + "step": 831 + }, + { + "epoch": 0.06860724004287952, + "grad_norm": 2.6655285358428955, + "learning_rate": 2.2835943940643033e-07, + "loss": 1.0094, + "mean_token_accuracy": 0.6922081112861633, + "num_tokens": 25163373.0, + "step": 832 + }, + { + "epoch": 0.06868970066793106, + "grad_norm": 2.1321356296539307, + "learning_rate": 2.286342401758725e-07, + "loss": 0.8993, + "mean_token_accuracy": 0.7269656658172607, + "num_tokens": 25196816.0, + "step": 833 + }, + { + "epoch": 0.0687721612929826, + "grad_norm": 2.2061963081359863, + "learning_rate": 2.2890904094531462e-07, + "loss": 1.0267, + "mean_token_accuracy": 0.691566526889801, + "num_tokens": 25232493.0, + "step": 834 + }, + { + "epoch": 0.06885462191803414, + "grad_norm": 2.348926544189453, + "learning_rate": 2.291838417147568e-07, + "loss": 0.9974, + "mean_token_accuracy": 0.7083061933517456, + "num_tokens": 25263130.0, + "step": 835 + }, + { + "epoch": 0.06893708254308567, + "grad_norm": 2.4431703090667725, + "learning_rate": 2.2945864248419893e-07, + "loss": 0.9609, + "mean_token_accuracy": 0.7109242081642151, + "num_tokens": 25292195.0, + "step": 836 + }, + { + "epoch": 0.06901954316813722, + "grad_norm": 2.77128267288208, + "learning_rate": 2.297334432536411e-07, + "loss": 1.0037, + "mean_token_accuracy": 0.7008329629898071, + "num_tokens": 25314565.0, + "step": 837 + }, + { + "epoch": 0.06910200379318875, + "grad_norm": 2.70113205909729, + "learning_rate": 2.3000824402308325e-07, + "loss": 1.0275, + "mean_token_accuracy": 0.7010009288787842, + "num_tokens": 25342884.0, + "step": 838 + }, + { + "epoch": 0.06918446441824029, + "grad_norm": 2.340453863143921, + "learning_rate": 2.3028304479252542e-07, + "loss": 1.0089, + "mean_token_accuracy": 0.6949679255485535, + "num_tokens": 25374825.0, + "step": 839 + }, + { + "epoch": 0.06926692504329182, + "grad_norm": 2.441068172454834, + "learning_rate": 2.3055784556196757e-07, + "loss": 0.9649, + "mean_token_accuracy": 0.7046759128570557, + "num_tokens": 25404515.0, + "step": 840 + }, + { + "epoch": 0.06934938566834337, + "grad_norm": 2.4452717304229736, + "learning_rate": 2.3083264633140971e-07, + "loss": 0.8906, + "mean_token_accuracy": 0.7356621623039246, + "num_tokens": 25443050.0, + "step": 841 + }, + { + "epoch": 0.0694318462933949, + "grad_norm": 2.2528269290924072, + "learning_rate": 2.3110744710085186e-07, + "loss": 0.8652, + "mean_token_accuracy": 0.7331151962280273, + "num_tokens": 25472931.0, + "step": 842 + }, + { + "epoch": 0.06951430691844644, + "grad_norm": 2.320394277572632, + "learning_rate": 2.3138224787029403e-07, + "loss": 0.9178, + "mean_token_accuracy": 0.7330987453460693, + "num_tokens": 25503141.0, + "step": 843 + }, + { + "epoch": 0.06959676754349799, + "grad_norm": 2.563469171524048, + "learning_rate": 2.3165704863973618e-07, + "loss": 0.9226, + "mean_token_accuracy": 0.7209997177124023, + "num_tokens": 25528970.0, + "step": 844 + }, + { + "epoch": 0.06967922816854952, + "grad_norm": 2.383526086807251, + "learning_rate": 2.3193184940917835e-07, + "loss": 0.9172, + "mean_token_accuracy": 0.7237555980682373, + "num_tokens": 25559255.0, + "step": 845 + }, + { + "epoch": 0.06976168879360106, + "grad_norm": 2.0026583671569824, + "learning_rate": 2.322066501786205e-07, + "loss": 0.98, + "mean_token_accuracy": 0.7075604200363159, + "num_tokens": 25595695.0, + "step": 846 + }, + { + "epoch": 0.06984414941865259, + "grad_norm": 2.1402840614318848, + "learning_rate": 2.3248145094806266e-07, + "loss": 0.797, + "mean_token_accuracy": 0.7526004314422607, + "num_tokens": 25628691.0, + "step": 847 + }, + { + "epoch": 0.06992661004370414, + "grad_norm": 2.445591926574707, + "learning_rate": 2.3275625171750478e-07, + "loss": 0.7749, + "mean_token_accuracy": 0.7525128722190857, + "num_tokens": 25654290.0, + "step": 848 + }, + { + "epoch": 0.07000907066875567, + "grad_norm": 2.4901909828186035, + "learning_rate": 2.3303105248694695e-07, + "loss": 0.8896, + "mean_token_accuracy": 0.7277095317840576, + "num_tokens": 25681795.0, + "step": 849 + }, + { + "epoch": 0.07009153129380721, + "grad_norm": 2.8539233207702637, + "learning_rate": 2.333058532563891e-07, + "loss": 1.007, + "mean_token_accuracy": 0.6993942260742188, + "num_tokens": 25705189.0, + "step": 850 + }, + { + "epoch": 0.07017399191885874, + "grad_norm": 2.8905506134033203, + "learning_rate": 2.3358065402583127e-07, + "loss": 1.0868, + "mean_token_accuracy": 0.680228590965271, + "num_tokens": 25727236.0, + "step": 851 + }, + { + "epoch": 0.07025645254391029, + "grad_norm": 2.378915548324585, + "learning_rate": 2.3385545479527342e-07, + "loss": 0.9041, + "mean_token_accuracy": 0.7222111225128174, + "num_tokens": 25758400.0, + "step": 852 + }, + { + "epoch": 0.07033891316896182, + "grad_norm": 2.5110924243927, + "learning_rate": 2.341302555647156e-07, + "loss": 1.0841, + "mean_token_accuracy": 0.6834540367126465, + "num_tokens": 25786901.0, + "step": 853 + }, + { + "epoch": 0.07042137379401336, + "grad_norm": 1.9979469776153564, + "learning_rate": 2.3440505633415773e-07, + "loss": 0.9193, + "mean_token_accuracy": 0.7254970073699951, + "num_tokens": 25826027.0, + "step": 854 + }, + { + "epoch": 0.0705038344190649, + "grad_norm": 2.437314987182617, + "learning_rate": 2.3467985710359988e-07, + "loss": 0.9958, + "mean_token_accuracy": 0.7034879922866821, + "num_tokens": 25854838.0, + "step": 855 + }, + { + "epoch": 0.07058629504411644, + "grad_norm": 2.537827253341675, + "learning_rate": 2.3495465787304202e-07, + "loss": 0.931, + "mean_token_accuracy": 0.7197819352149963, + "num_tokens": 25880315.0, + "step": 856 + }, + { + "epoch": 0.07066875566916797, + "grad_norm": 2.4534213542938232, + "learning_rate": 2.352294586424842e-07, + "loss": 1.0568, + "mean_token_accuracy": 0.6862059831619263, + "num_tokens": 25906894.0, + "step": 857 + }, + { + "epoch": 0.07075121629421952, + "grad_norm": 2.6783435344696045, + "learning_rate": 2.3550425941192634e-07, + "loss": 1.0467, + "mean_token_accuracy": 0.6864767074584961, + "num_tokens": 25931214.0, + "step": 858 + }, + { + "epoch": 0.07083367691927105, + "grad_norm": 1.9330806732177734, + "learning_rate": 2.357790601813685e-07, + "loss": 0.735, + "mean_token_accuracy": 0.7758082747459412, + "num_tokens": 25971151.0, + "step": 859 + }, + { + "epoch": 0.07091613754432259, + "grad_norm": 2.2615888118743896, + "learning_rate": 2.3605386095081066e-07, + "loss": 1.0535, + "mean_token_accuracy": 0.6811195611953735, + "num_tokens": 26002397.0, + "step": 860 + }, + { + "epoch": 0.07099859816937412, + "grad_norm": 2.2306954860687256, + "learning_rate": 2.3632866172025283e-07, + "loss": 0.8736, + "mean_token_accuracy": 0.7263501882553101, + "num_tokens": 26032353.0, + "step": 861 + }, + { + "epoch": 0.07108105879442567, + "grad_norm": 2.4045403003692627, + "learning_rate": 2.3660346248969495e-07, + "loss": 0.9626, + "mean_token_accuracy": 0.715686559677124, + "num_tokens": 26059907.0, + "step": 862 + }, + { + "epoch": 0.0711635194194772, + "grad_norm": 2.355947256088257, + "learning_rate": 2.3687826325913712e-07, + "loss": 0.9714, + "mean_token_accuracy": 0.7099372744560242, + "num_tokens": 26089384.0, + "step": 863 + }, + { + "epoch": 0.07124598004452874, + "grad_norm": 2.498657703399658, + "learning_rate": 2.3715306402857926e-07, + "loss": 0.975, + "mean_token_accuracy": 0.7192074060440063, + "num_tokens": 26119483.0, + "step": 864 + }, + { + "epoch": 0.07132844066958027, + "grad_norm": 2.5661070346832275, + "learning_rate": 2.3742786479802143e-07, + "loss": 0.9671, + "mean_token_accuracy": 0.7033607959747314, + "num_tokens": 26147670.0, + "step": 865 + }, + { + "epoch": 0.07141090129463182, + "grad_norm": 3.0381522178649902, + "learning_rate": 2.3770266556746358e-07, + "loss": 1.0262, + "mean_token_accuracy": 0.6968949437141418, + "num_tokens": 26169806.0, + "step": 866 + }, + { + "epoch": 0.07149336191968335, + "grad_norm": 2.6132636070251465, + "learning_rate": 2.3797746633690575e-07, + "loss": 0.8521, + "mean_token_accuracy": 0.7310311198234558, + "num_tokens": 26196935.0, + "step": 867 + }, + { + "epoch": 0.0715758225447349, + "grad_norm": 2.1324362754821777, + "learning_rate": 2.3825226710634787e-07, + "loss": 0.7885, + "mean_token_accuracy": 0.7510719895362854, + "num_tokens": 26229813.0, + "step": 868 + }, + { + "epoch": 0.07165828316978642, + "grad_norm": 2.133310079574585, + "learning_rate": 2.3852706787579004e-07, + "loss": 0.8017, + "mean_token_accuracy": 0.7470506429672241, + "num_tokens": 26259276.0, + "step": 869 + }, + { + "epoch": 0.07174074379483797, + "grad_norm": 2.7512948513031006, + "learning_rate": 2.388018686452322e-07, + "loss": 0.9287, + "mean_token_accuracy": 0.7261373996734619, + "num_tokens": 26281903.0, + "step": 870 + }, + { + "epoch": 0.0718232044198895, + "grad_norm": 2.5225439071655273, + "learning_rate": 2.390766694146744e-07, + "loss": 1.1148, + "mean_token_accuracy": 0.673260509967804, + "num_tokens": 26311349.0, + "step": 871 + }, + { + "epoch": 0.07190566504494104, + "grad_norm": 2.571380138397217, + "learning_rate": 2.393514701841165e-07, + "loss": 1.0149, + "mean_token_accuracy": 0.6966145038604736, + "num_tokens": 26334884.0, + "step": 872 + }, + { + "epoch": 0.07198812566999258, + "grad_norm": 2.446113348007202, + "learning_rate": 2.396262709535587e-07, + "loss": 0.8628, + "mean_token_accuracy": 0.7352155447006226, + "num_tokens": 26360510.0, + "step": 873 + }, + { + "epoch": 0.07207058629504412, + "grad_norm": 2.5074539184570312, + "learning_rate": 2.399010717230008e-07, + "loss": 1.036, + "mean_token_accuracy": 0.7014179229736328, + "num_tokens": 26386571.0, + "step": 874 + }, + { + "epoch": 0.07215304692009565, + "grad_norm": 2.39729642868042, + "learning_rate": 2.4017587249244296e-07, + "loss": 0.7593, + "mean_token_accuracy": 0.7698537111282349, + "num_tokens": 26412214.0, + "step": 875 + }, + { + "epoch": 0.0722355075451472, + "grad_norm": 2.3212616443634033, + "learning_rate": 2.4045067326188514e-07, + "loss": 0.948, + "mean_token_accuracy": 0.7214453816413879, + "num_tokens": 26441730.0, + "step": 876 + }, + { + "epoch": 0.07231796817019873, + "grad_norm": 2.0599348545074463, + "learning_rate": 2.407254740313273e-07, + "loss": 0.7961, + "mean_token_accuracy": 0.7554175853729248, + "num_tokens": 26475268.0, + "step": 877 + }, + { + "epoch": 0.07240042879525027, + "grad_norm": 2.137415885925293, + "learning_rate": 2.410002748007694e-07, + "loss": 0.9334, + "mean_token_accuracy": 0.7269635796546936, + "num_tokens": 26510654.0, + "step": 878 + }, + { + "epoch": 0.0724828894203018, + "grad_norm": 1.9959591627120972, + "learning_rate": 2.412750755702116e-07, + "loss": 0.8553, + "mean_token_accuracy": 0.7471638917922974, + "num_tokens": 26550649.0, + "step": 879 + }, + { + "epoch": 0.07256535004535335, + "grad_norm": 2.4241716861724854, + "learning_rate": 2.415498763396537e-07, + "loss": 0.7823, + "mean_token_accuracy": 0.7596645951271057, + "num_tokens": 26578435.0, + "step": 880 + }, + { + "epoch": 0.07264781067040488, + "grad_norm": 2.713653326034546, + "learning_rate": 2.418246771090959e-07, + "loss": 0.918, + "mean_token_accuracy": 0.7207231521606445, + "num_tokens": 26601380.0, + "step": 881 + }, + { + "epoch": 0.07273027129545642, + "grad_norm": 2.6578407287597656, + "learning_rate": 2.4209947787853806e-07, + "loss": 0.9356, + "mean_token_accuracy": 0.7250710129737854, + "num_tokens": 26624494.0, + "step": 882 + }, + { + "epoch": 0.07281273192050795, + "grad_norm": 2.476212739944458, + "learning_rate": 2.4237427864798023e-07, + "loss": 1.0122, + "mean_token_accuracy": 0.6971966028213501, + "num_tokens": 26652048.0, + "step": 883 + }, + { + "epoch": 0.0728951925455595, + "grad_norm": 2.059431314468384, + "learning_rate": 2.4264907941742235e-07, + "loss": 0.7666, + "mean_token_accuracy": 0.7696902751922607, + "num_tokens": 26684078.0, + "step": 884 + }, + { + "epoch": 0.07297765317061103, + "grad_norm": 2.319502353668213, + "learning_rate": 2.429238801868645e-07, + "loss": 0.8838, + "mean_token_accuracy": 0.7420690655708313, + "num_tokens": 26712609.0, + "step": 885 + }, + { + "epoch": 0.07306011379566257, + "grad_norm": 2.266831398010254, + "learning_rate": 2.4319868095630664e-07, + "loss": 0.7282, + "mean_token_accuracy": 0.7783255577087402, + "num_tokens": 26744860.0, + "step": 886 + }, + { + "epoch": 0.0731425744207141, + "grad_norm": 2.4541985988616943, + "learning_rate": 2.434734817257488e-07, + "loss": 0.9397, + "mean_token_accuracy": 0.7148136496543884, + "num_tokens": 26774046.0, + "step": 887 + }, + { + "epoch": 0.07322503504576565, + "grad_norm": 2.4067251682281494, + "learning_rate": 2.43748282495191e-07, + "loss": 0.9923, + "mean_token_accuracy": 0.7065757513046265, + "num_tokens": 26800886.0, + "step": 888 + }, + { + "epoch": 0.07330749567081718, + "grad_norm": 2.0139920711517334, + "learning_rate": 2.4402308326463315e-07, + "loss": 0.8516, + "mean_token_accuracy": 0.7441893815994263, + "num_tokens": 26835951.0, + "step": 889 + }, + { + "epoch": 0.07338995629586872, + "grad_norm": 2.1262571811676025, + "learning_rate": 2.4429788403407527e-07, + "loss": 0.891, + "mean_token_accuracy": 0.7270762920379639, + "num_tokens": 26870477.0, + "step": 890 + }, + { + "epoch": 0.07347241692092026, + "grad_norm": 2.1462559700012207, + "learning_rate": 2.4457268480351744e-07, + "loss": 0.9825, + "mean_token_accuracy": 0.7068077921867371, + "num_tokens": 26906192.0, + "step": 891 + }, + { + "epoch": 0.0735548775459718, + "grad_norm": 2.2153658866882324, + "learning_rate": 2.4484748557295956e-07, + "loss": 0.8952, + "mean_token_accuracy": 0.7256723642349243, + "num_tokens": 26938158.0, + "step": 892 + }, + { + "epoch": 0.07363733817102333, + "grad_norm": 2.165210247039795, + "learning_rate": 2.4512228634240173e-07, + "loss": 0.898, + "mean_token_accuracy": 0.7421982288360596, + "num_tokens": 26970054.0, + "step": 893 + }, + { + "epoch": 0.07371979879607488, + "grad_norm": 2.839128017425537, + "learning_rate": 2.453970871118439e-07, + "loss": 0.9789, + "mean_token_accuracy": 0.7055510878562927, + "num_tokens": 26994625.0, + "step": 894 + }, + { + "epoch": 0.0738022594211264, + "grad_norm": 2.2801332473754883, + "learning_rate": 2.456718878812861e-07, + "loss": 0.7857, + "mean_token_accuracy": 0.7593759298324585, + "num_tokens": 27021892.0, + "step": 895 + }, + { + "epoch": 0.07388472004617795, + "grad_norm": 2.0890233516693115, + "learning_rate": 2.459466886507282e-07, + "loss": 0.9027, + "mean_token_accuracy": 0.7301205396652222, + "num_tokens": 27057624.0, + "step": 896 + }, + { + "epoch": 0.07396718067122948, + "grad_norm": 2.12429141998291, + "learning_rate": 2.4622148942017037e-07, + "loss": 0.9634, + "mean_token_accuracy": 0.7111847400665283, + "num_tokens": 27090324.0, + "step": 897 + }, + { + "epoch": 0.07404964129628103, + "grad_norm": 2.5103254318237305, + "learning_rate": 2.4649629018961254e-07, + "loss": 1.0486, + "mean_token_accuracy": 0.6847168803215027, + "num_tokens": 27116694.0, + "step": 898 + }, + { + "epoch": 0.07413210192133256, + "grad_norm": 2.3642406463623047, + "learning_rate": 2.4677109095905466e-07, + "loss": 0.906, + "mean_token_accuracy": 0.7258450984954834, + "num_tokens": 27144907.0, + "step": 899 + }, + { + "epoch": 0.0742145625463841, + "grad_norm": 2.465859889984131, + "learning_rate": 2.4704589172849683e-07, + "loss": 0.9759, + "mean_token_accuracy": 0.7109360098838806, + "num_tokens": 27169482.0, + "step": 900 + }, + { + "epoch": 0.07429702317143563, + "grad_norm": 2.2170233726501465, + "learning_rate": 2.47320692497939e-07, + "loss": 1.0346, + "mean_token_accuracy": 0.6904748678207397, + "num_tokens": 27201501.0, + "step": 901 + }, + { + "epoch": 0.07437948379648718, + "grad_norm": 2.145578145980835, + "learning_rate": 2.475954932673811e-07, + "loss": 0.9237, + "mean_token_accuracy": 0.7207221388816833, + "num_tokens": 27234162.0, + "step": 902 + }, + { + "epoch": 0.07446194442153871, + "grad_norm": 2.4689137935638428, + "learning_rate": 2.478702940368233e-07, + "loss": 0.773, + "mean_token_accuracy": 0.7649416327476501, + "num_tokens": 27262953.0, + "step": 903 + }, + { + "epoch": 0.07454440504659025, + "grad_norm": 2.4967446327209473, + "learning_rate": 2.4814509480626546e-07, + "loss": 1.0104, + "mean_token_accuracy": 0.7062439918518066, + "num_tokens": 27290223.0, + "step": 904 + }, + { + "epoch": 0.07462686567164178, + "grad_norm": 2.8091437816619873, + "learning_rate": 2.4841989557570763e-07, + "loss": 0.8225, + "mean_token_accuracy": 0.7460731267929077, + "num_tokens": 27312292.0, + "step": 905 + }, + { + "epoch": 0.07470932629669333, + "grad_norm": 2.4351541996002197, + "learning_rate": 2.4869469634514975e-07, + "loss": 0.9991, + "mean_token_accuracy": 0.6987947225570679, + "num_tokens": 27339957.0, + "step": 906 + }, + { + "epoch": 0.07479178692174486, + "grad_norm": 2.113868474960327, + "learning_rate": 2.489694971145919e-07, + "loss": 0.8317, + "mean_token_accuracy": 0.7400506734848022, + "num_tokens": 27368908.0, + "step": 907 + }, + { + "epoch": 0.0748742475467964, + "grad_norm": 2.2413651943206787, + "learning_rate": 2.4924429788403404e-07, + "loss": 0.9648, + "mean_token_accuracy": 0.7039275169372559, + "num_tokens": 27401244.0, + "step": 908 + }, + { + "epoch": 0.07495670817184795, + "grad_norm": 2.012364387512207, + "learning_rate": 2.495190986534762e-07, + "loss": 0.9719, + "mean_token_accuracy": 0.7053535580635071, + "num_tokens": 27436524.0, + "step": 909 + }, + { + "epoch": 0.07503916879689948, + "grad_norm": 1.9733855724334717, + "learning_rate": 2.497938994229184e-07, + "loss": 0.771, + "mean_token_accuracy": 0.7677605152130127, + "num_tokens": 27470159.0, + "step": 910 + }, + { + "epoch": 0.07512162942195102, + "grad_norm": 2.378288984298706, + "learning_rate": 2.5006870019236056e-07, + "loss": 0.9795, + "mean_token_accuracy": 0.7056481242179871, + "num_tokens": 27502444.0, + "step": 911 + }, + { + "epoch": 0.07520409004700256, + "grad_norm": 2.4342288970947266, + "learning_rate": 2.503435009618027e-07, + "loss": 0.8419, + "mean_token_accuracy": 0.7342323064804077, + "num_tokens": 27529752.0, + "step": 912 + }, + { + "epoch": 0.0752865506720541, + "grad_norm": 2.296811819076538, + "learning_rate": 2.506183017312448e-07, + "loss": 0.7405, + "mean_token_accuracy": 0.7678407430648804, + "num_tokens": 27560011.0, + "step": 913 + }, + { + "epoch": 0.07536901129710563, + "grad_norm": 2.5793802738189697, + "learning_rate": 2.50893102500687e-07, + "loss": 1.1303, + "mean_token_accuracy": 0.6706790924072266, + "num_tokens": 27587833.0, + "step": 914 + }, + { + "epoch": 0.07545147192215718, + "grad_norm": 2.185004711151123, + "learning_rate": 2.5116790327012914e-07, + "loss": 0.9894, + "mean_token_accuracy": 0.7016828060150146, + "num_tokens": 27620000.0, + "step": 915 + }, + { + "epoch": 0.07553393254720871, + "grad_norm": 2.2830607891082764, + "learning_rate": 2.514427040395713e-07, + "loss": 0.9092, + "mean_token_accuracy": 0.7186462879180908, + "num_tokens": 27647963.0, + "step": 916 + }, + { + "epoch": 0.07561639317226025, + "grad_norm": 2.167327404022217, + "learning_rate": 2.5171750480901343e-07, + "loss": 0.9601, + "mean_token_accuracy": 0.7160711288452148, + "num_tokens": 27683963.0, + "step": 917 + }, + { + "epoch": 0.07569885379731178, + "grad_norm": 2.632089138031006, + "learning_rate": 2.5199230557845565e-07, + "loss": 0.8864, + "mean_token_accuracy": 0.7248727083206177, + "num_tokens": 27708264.0, + "step": 918 + }, + { + "epoch": 0.07578131442236333, + "grad_norm": 2.1277730464935303, + "learning_rate": 2.5226710634789777e-07, + "loss": 0.9652, + "mean_token_accuracy": 0.7100851535797119, + "num_tokens": 27743122.0, + "step": 919 + }, + { + "epoch": 0.07586377504741486, + "grad_norm": 2.052899122238159, + "learning_rate": 2.525419071173399e-07, + "loss": 0.7984, + "mean_token_accuracy": 0.7607134580612183, + "num_tokens": 27778939.0, + "step": 920 + }, + { + "epoch": 0.0759462356724664, + "grad_norm": 2.15161395072937, + "learning_rate": 2.5281670788678206e-07, + "loss": 0.9284, + "mean_token_accuracy": 0.7186484336853027, + "num_tokens": 27812651.0, + "step": 921 + }, + { + "epoch": 0.07602869629751793, + "grad_norm": 2.1009066104888916, + "learning_rate": 2.5309150865622423e-07, + "loss": 1.0366, + "mean_token_accuracy": 0.6954798698425293, + "num_tokens": 27845750.0, + "step": 922 + }, + { + "epoch": 0.07611115692256948, + "grad_norm": 2.2515339851379395, + "learning_rate": 2.533663094256664e-07, + "loss": 1.0398, + "mean_token_accuracy": 0.7005565166473389, + "num_tokens": 27879916.0, + "step": 923 + }, + { + "epoch": 0.07619361754762101, + "grad_norm": 2.79296612739563, + "learning_rate": 2.536411101951085e-07, + "loss": 0.9255, + "mean_token_accuracy": 0.7160238027572632, + "num_tokens": 27904268.0, + "step": 924 + }, + { + "epoch": 0.07627607817267255, + "grad_norm": 2.4599099159240723, + "learning_rate": 2.539159109645507e-07, + "loss": 1.0217, + "mean_token_accuracy": 0.6989666223526001, + "num_tokens": 27932008.0, + "step": 925 + }, + { + "epoch": 0.07635853879772408, + "grad_norm": 2.4110162258148193, + "learning_rate": 2.5419071173399287e-07, + "loss": 1.0055, + "mean_token_accuracy": 0.7026255130767822, + "num_tokens": 27957897.0, + "step": 926 + }, + { + "epoch": 0.07644099942277563, + "grad_norm": 2.4548227787017822, + "learning_rate": 2.54465512503435e-07, + "loss": 0.9552, + "mean_token_accuracy": 0.7149970531463623, + "num_tokens": 27985243.0, + "step": 927 + }, + { + "epoch": 0.07652346004782716, + "grad_norm": 2.3720030784606934, + "learning_rate": 2.5474031327287716e-07, + "loss": 0.8975, + "mean_token_accuracy": 0.7220115661621094, + "num_tokens": 28010127.0, + "step": 928 + }, + { + "epoch": 0.0766059206728787, + "grad_norm": 2.0246083736419678, + "learning_rate": 2.550151140423193e-07, + "loss": 0.7804, + "mean_token_accuracy": 0.7577241659164429, + "num_tokens": 28045858.0, + "step": 929 + }, + { + "epoch": 0.07668838129793024, + "grad_norm": 2.2812113761901855, + "learning_rate": 2.552899148117615e-07, + "loss": 1.0373, + "mean_token_accuracy": 0.700753927230835, + "num_tokens": 28075492.0, + "step": 930 + }, + { + "epoch": 0.07677084192298178, + "grad_norm": 2.4929709434509277, + "learning_rate": 2.555647155812036e-07, + "loss": 0.998, + "mean_token_accuracy": 0.7006492614746094, + "num_tokens": 28102039.0, + "step": 931 + }, + { + "epoch": 0.07685330254803331, + "grad_norm": 2.443540573120117, + "learning_rate": 2.558395163506458e-07, + "loss": 0.9306, + "mean_token_accuracy": 0.7128456830978394, + "num_tokens": 28127222.0, + "step": 932 + }, + { + "epoch": 0.07693576317308486, + "grad_norm": 2.676053524017334, + "learning_rate": 2.561143171200879e-07, + "loss": 1.0599, + "mean_token_accuracy": 0.6924147009849548, + "num_tokens": 28152257.0, + "step": 933 + }, + { + "epoch": 0.07701822379813639, + "grad_norm": 2.209604024887085, + "learning_rate": 2.563891178895301e-07, + "loss": 0.9229, + "mean_token_accuracy": 0.7139480113983154, + "num_tokens": 28182399.0, + "step": 934 + }, + { + "epoch": 0.07710068442318793, + "grad_norm": 2.5439538955688477, + "learning_rate": 2.5666391865897225e-07, + "loss": 1.0185, + "mean_token_accuracy": 0.6996201872825623, + "num_tokens": 28208301.0, + "step": 935 + }, + { + "epoch": 0.07718314504823946, + "grad_norm": 2.6925249099731445, + "learning_rate": 2.5693871942841437e-07, + "loss": 1.0366, + "mean_token_accuracy": 0.6934782266616821, + "num_tokens": 28232481.0, + "step": 936 + }, + { + "epoch": 0.07726560567329101, + "grad_norm": 2.3472747802734375, + "learning_rate": 2.5721352019785654e-07, + "loss": 0.9673, + "mean_token_accuracy": 0.7065139412879944, + "num_tokens": 28260037.0, + "step": 937 + }, + { + "epoch": 0.07734806629834254, + "grad_norm": 2.448902130126953, + "learning_rate": 2.574883209672987e-07, + "loss": 0.789, + "mean_token_accuracy": 0.7616350054740906, + "num_tokens": 28284960.0, + "step": 938 + }, + { + "epoch": 0.07743052692339408, + "grad_norm": 2.122835636138916, + "learning_rate": 2.577631217367409e-07, + "loss": 0.9398, + "mean_token_accuracy": 0.7230260372161865, + "num_tokens": 28321518.0, + "step": 939 + }, + { + "epoch": 0.07751298754844561, + "grad_norm": 2.241265058517456, + "learning_rate": 2.58037922506183e-07, + "loss": 0.873, + "mean_token_accuracy": 0.7342028617858887, + "num_tokens": 28350171.0, + "step": 940 + }, + { + "epoch": 0.07759544817349716, + "grad_norm": 2.233555793762207, + "learning_rate": 2.583127232756251e-07, + "loss": 0.7893, + "mean_token_accuracy": 0.757283091545105, + "num_tokens": 28379072.0, + "step": 941 + }, + { + "epoch": 0.07767790879854869, + "grad_norm": 2.1079204082489014, + "learning_rate": 2.5858752404506735e-07, + "loss": 0.853, + "mean_token_accuracy": 0.7291576862335205, + "num_tokens": 28412488.0, + "step": 942 + }, + { + "epoch": 0.07776036942360023, + "grad_norm": 2.103734016418457, + "learning_rate": 2.5886232481450947e-07, + "loss": 0.8484, + "mean_token_accuracy": 0.7456963062286377, + "num_tokens": 28445794.0, + "step": 943 + }, + { + "epoch": 0.07784283004865176, + "grad_norm": 2.385098695755005, + "learning_rate": 2.5913712558395164e-07, + "loss": 0.9887, + "mean_token_accuracy": 0.7032197713851929, + "num_tokens": 28475059.0, + "step": 944 + }, + { + "epoch": 0.07792529067370331, + "grad_norm": 2.2440242767333984, + "learning_rate": 2.5941192635339376e-07, + "loss": 0.9766, + "mean_token_accuracy": 0.7122355103492737, + "num_tokens": 28508807.0, + "step": 945 + }, + { + "epoch": 0.07800775129875484, + "grad_norm": 2.846609592437744, + "learning_rate": 2.59686727122836e-07, + "loss": 1.0525, + "mean_token_accuracy": 0.687404990196228, + "num_tokens": 28531429.0, + "step": 946 + }, + { + "epoch": 0.07809021192380639, + "grad_norm": 2.4037013053894043, + "learning_rate": 2.599615278922781e-07, + "loss": 0.9954, + "mean_token_accuracy": 0.7008355855941772, + "num_tokens": 28559194.0, + "step": 947 + }, + { + "epoch": 0.07817267254885792, + "grad_norm": 2.0532217025756836, + "learning_rate": 2.602363286617202e-07, + "loss": 0.9249, + "mean_token_accuracy": 0.7222945690155029, + "num_tokens": 28591755.0, + "step": 948 + }, + { + "epoch": 0.07825513317390946, + "grad_norm": 2.4935178756713867, + "learning_rate": 2.605111294311624e-07, + "loss": 0.9734, + "mean_token_accuracy": 0.7091808915138245, + "num_tokens": 28618393.0, + "step": 949 + }, + { + "epoch": 0.07833759379896099, + "grad_norm": 2.3973536491394043, + "learning_rate": 2.6078593020060456e-07, + "loss": 0.8212, + "mean_token_accuracy": 0.750989556312561, + "num_tokens": 28646870.0, + "step": 950 + }, + { + "epoch": 0.07842005442401254, + "grad_norm": 2.0342469215393066, + "learning_rate": 2.6106073097004673e-07, + "loss": 0.8571, + "mean_token_accuracy": 0.7413069009780884, + "num_tokens": 28680760.0, + "step": 951 + }, + { + "epoch": 0.07850251504906407, + "grad_norm": 2.8125455379486084, + "learning_rate": 2.6133553173948885e-07, + "loss": 0.9972, + "mean_token_accuracy": 0.7109227776527405, + "num_tokens": 28700918.0, + "step": 952 + }, + { + "epoch": 0.07858497567411561, + "grad_norm": 2.354962110519409, + "learning_rate": 2.61610332508931e-07, + "loss": 0.8709, + "mean_token_accuracy": 0.731782078742981, + "num_tokens": 28726467.0, + "step": 953 + }, + { + "epoch": 0.07866743629916714, + "grad_norm": 2.1405153274536133, + "learning_rate": 2.618851332783732e-07, + "loss": 0.9039, + "mean_token_accuracy": 0.7336090207099915, + "num_tokens": 28760425.0, + "step": 954 + }, + { + "epoch": 0.07874989692421869, + "grad_norm": 2.487196922302246, + "learning_rate": 2.621599340478153e-07, + "loss": 1.0005, + "mean_token_accuracy": 0.6914699077606201, + "num_tokens": 28785620.0, + "step": 955 + }, + { + "epoch": 0.07883235754927022, + "grad_norm": 2.8400871753692627, + "learning_rate": 2.624347348172575e-07, + "loss": 0.9406, + "mean_token_accuracy": 0.7164096832275391, + "num_tokens": 28808487.0, + "step": 956 + }, + { + "epoch": 0.07891481817432176, + "grad_norm": 2.572951078414917, + "learning_rate": 2.627095355866996e-07, + "loss": 0.7927, + "mean_token_accuracy": 0.754783034324646, + "num_tokens": 28833882.0, + "step": 957 + }, + { + "epoch": 0.0789972787993733, + "grad_norm": 2.5798280239105225, + "learning_rate": 2.6298433635614183e-07, + "loss": 0.9681, + "mean_token_accuracy": 0.7104064226150513, + "num_tokens": 28856953.0, + "step": 958 + }, + { + "epoch": 0.07907973942442484, + "grad_norm": 2.113161563873291, + "learning_rate": 2.6325913712558395e-07, + "loss": 0.8418, + "mean_token_accuracy": 0.7474382519721985, + "num_tokens": 28890158.0, + "step": 959 + }, + { + "epoch": 0.07916220004947637, + "grad_norm": 2.4609286785125732, + "learning_rate": 2.635339378950261e-07, + "loss": 0.9544, + "mean_token_accuracy": 0.7147408723831177, + "num_tokens": 28916037.0, + "step": 960 + }, + { + "epoch": 0.07924466067452791, + "grad_norm": 2.451228380203247, + "learning_rate": 2.6380873866446824e-07, + "loss": 1.066, + "mean_token_accuracy": 0.6875990629196167, + "num_tokens": 28940994.0, + "step": 961 + }, + { + "epoch": 0.07932712129957945, + "grad_norm": 1.9867490530014038, + "learning_rate": 2.640835394339104e-07, + "loss": 0.835, + "mean_token_accuracy": 0.7429400682449341, + "num_tokens": 28973587.0, + "step": 962 + }, + { + "epoch": 0.07940958192463099, + "grad_norm": 2.2005414962768555, + "learning_rate": 2.643583402033526e-07, + "loss": 0.9666, + "mean_token_accuracy": 0.717656135559082, + "num_tokens": 29002198.0, + "step": 963 + }, + { + "epoch": 0.07949204254968252, + "grad_norm": 2.5134594440460205, + "learning_rate": 2.646331409727947e-07, + "loss": 0.9435, + "mean_token_accuracy": 0.7261828780174255, + "num_tokens": 29027008.0, + "step": 964 + }, + { + "epoch": 0.07957450317473407, + "grad_norm": 2.210324764251709, + "learning_rate": 2.6490794174223687e-07, + "loss": 0.8369, + "mean_token_accuracy": 0.7432659864425659, + "num_tokens": 29059199.0, + "step": 965 + }, + { + "epoch": 0.0796569637997856, + "grad_norm": 2.036329507827759, + "learning_rate": 2.6518274251167904e-07, + "loss": 0.7814, + "mean_token_accuracy": 0.7638797163963318, + "num_tokens": 29092044.0, + "step": 966 + }, + { + "epoch": 0.07973942442483714, + "grad_norm": 2.162806272506714, + "learning_rate": 2.654575432811212e-07, + "loss": 0.9597, + "mean_token_accuracy": 0.7195420265197754, + "num_tokens": 29123738.0, + "step": 967 + }, + { + "epoch": 0.07982188504988867, + "grad_norm": 2.589096784591675, + "learning_rate": 2.6573234405056333e-07, + "loss": 0.9962, + "mean_token_accuracy": 0.7042597532272339, + "num_tokens": 29148514.0, + "step": 968 + }, + { + "epoch": 0.07990434567494022, + "grad_norm": 2.2631568908691406, + "learning_rate": 2.6600714482000545e-07, + "loss": 0.8574, + "mean_token_accuracy": 0.7381428480148315, + "num_tokens": 29177214.0, + "step": 969 + }, + { + "epoch": 0.07998680629999175, + "grad_norm": 1.9190053939819336, + "learning_rate": 2.662819455894477e-07, + "loss": 0.8134, + "mean_token_accuracy": 0.7495971918106079, + "num_tokens": 29212535.0, + "step": 970 + }, + { + "epoch": 0.08006926692504329, + "grad_norm": 2.2808077335357666, + "learning_rate": 2.665567463588898e-07, + "loss": 0.941, + "mean_token_accuracy": 0.7121448516845703, + "num_tokens": 29243146.0, + "step": 971 + }, + { + "epoch": 0.08015172755009482, + "grad_norm": 2.4431962966918945, + "learning_rate": 2.6683154712833196e-07, + "loss": 1.0344, + "mean_token_accuracy": 0.683850109577179, + "num_tokens": 29268972.0, + "step": 972 + }, + { + "epoch": 0.08023418817514637, + "grad_norm": 2.2326793670654297, + "learning_rate": 2.671063478977741e-07, + "loss": 0.971, + "mean_token_accuracy": 0.7021933794021606, + "num_tokens": 29302381.0, + "step": 973 + }, + { + "epoch": 0.08031664880019791, + "grad_norm": 2.1856794357299805, + "learning_rate": 2.673811486672163e-07, + "loss": 1.0923, + "mean_token_accuracy": 0.6800248622894287, + "num_tokens": 29335837.0, + "step": 974 + }, + { + "epoch": 0.08039910942524944, + "grad_norm": 2.338404893875122, + "learning_rate": 2.676559494366584e-07, + "loss": 1.0733, + "mean_token_accuracy": 0.6845941543579102, + "num_tokens": 29364734.0, + "step": 975 + }, + { + "epoch": 0.08048157005030099, + "grad_norm": 2.310774803161621, + "learning_rate": 2.6793075020610054e-07, + "loss": 0.9698, + "mean_token_accuracy": 0.7204848527908325, + "num_tokens": 29393553.0, + "step": 976 + }, + { + "epoch": 0.08056403067535252, + "grad_norm": 2.0039234161376953, + "learning_rate": 2.682055509755427e-07, + "loss": 0.8885, + "mean_token_accuracy": 0.7282233238220215, + "num_tokens": 29426066.0, + "step": 977 + }, + { + "epoch": 0.08064649130040406, + "grad_norm": 1.946075439453125, + "learning_rate": 2.684803517449849e-07, + "loss": 0.8209, + "mean_token_accuracy": 0.7528659105300903, + "num_tokens": 29460940.0, + "step": 978 + }, + { + "epoch": 0.0807289519254556, + "grad_norm": 1.9165372848510742, + "learning_rate": 2.6875515251442706e-07, + "loss": 0.908, + "mean_token_accuracy": 0.7189962863922119, + "num_tokens": 29499641.0, + "step": 979 + }, + { + "epoch": 0.08081141255050714, + "grad_norm": 2.358314037322998, + "learning_rate": 2.690299532838692e-07, + "loss": 1.0587, + "mean_token_accuracy": 0.6901912689208984, + "num_tokens": 29527829.0, + "step": 980 + }, + { + "epoch": 0.08089387317555867, + "grad_norm": 2.2082080841064453, + "learning_rate": 2.693047540533113e-07, + "loss": 1.0233, + "mean_token_accuracy": 0.7075475454330444, + "num_tokens": 29557319.0, + "step": 981 + }, + { + "epoch": 0.08097633380061021, + "grad_norm": 2.174431324005127, + "learning_rate": 2.695795548227535e-07, + "loss": 0.9023, + "mean_token_accuracy": 0.7212624549865723, + "num_tokens": 29587019.0, + "step": 982 + }, + { + "epoch": 0.08105879442566175, + "grad_norm": 2.1515777111053467, + "learning_rate": 2.6985435559219564e-07, + "loss": 1.0552, + "mean_token_accuracy": 0.6855634450912476, + "num_tokens": 29619768.0, + "step": 983 + }, + { + "epoch": 0.08114125505071329, + "grad_norm": 2.268310785293579, + "learning_rate": 2.701291563616378e-07, + "loss": 1.0282, + "mean_token_accuracy": 0.6954492330551147, + "num_tokens": 29651334.0, + "step": 984 + }, + { + "epoch": 0.08122371567576482, + "grad_norm": 2.2121591567993164, + "learning_rate": 2.7040395713107993e-07, + "loss": 0.9673, + "mean_token_accuracy": 0.7186022400856018, + "num_tokens": 29681229.0, + "step": 985 + }, + { + "epoch": 0.08130617630081637, + "grad_norm": 2.610719919204712, + "learning_rate": 2.7067875790052215e-07, + "loss": 0.9609, + "mean_token_accuracy": 0.7100074887275696, + "num_tokens": 29703938.0, + "step": 986 + }, + { + "epoch": 0.0813886369258679, + "grad_norm": 2.6025516986846924, + "learning_rate": 2.7095355866996427e-07, + "loss": 0.8825, + "mean_token_accuracy": 0.7241466045379639, + "num_tokens": 29726508.0, + "step": 987 + }, + { + "epoch": 0.08147109755091944, + "grad_norm": 2.196441173553467, + "learning_rate": 2.712283594394064e-07, + "loss": 0.9467, + "mean_token_accuracy": 0.7162476778030396, + "num_tokens": 29755341.0, + "step": 988 + }, + { + "epoch": 0.08155355817597097, + "grad_norm": 2.148797035217285, + "learning_rate": 2.7150316020884856e-07, + "loss": 0.8712, + "mean_token_accuracy": 0.7303158044815063, + "num_tokens": 29784818.0, + "step": 989 + }, + { + "epoch": 0.08163601880102252, + "grad_norm": 2.522580146789551, + "learning_rate": 2.7177796097829074e-07, + "loss": 1.05, + "mean_token_accuracy": 0.6884550452232361, + "num_tokens": 29808807.0, + "step": 990 + }, + { + "epoch": 0.08171847942607405, + "grad_norm": 2.3500888347625732, + "learning_rate": 2.720527617477329e-07, + "loss": 0.9422, + "mean_token_accuracy": 0.7154930830001831, + "num_tokens": 29835908.0, + "step": 991 + }, + { + "epoch": 0.08180094005112559, + "grad_norm": 2.2261340618133545, + "learning_rate": 2.72327562517175e-07, + "loss": 0.9568, + "mean_token_accuracy": 0.7162906527519226, + "num_tokens": 29863621.0, + "step": 992 + }, + { + "epoch": 0.08188340067617712, + "grad_norm": 2.491814136505127, + "learning_rate": 2.726023632866172e-07, + "loss": 0.9795, + "mean_token_accuracy": 0.7090897560119629, + "num_tokens": 29888835.0, + "step": 993 + }, + { + "epoch": 0.08196586130122867, + "grad_norm": 2.1144356727600098, + "learning_rate": 2.728771640560593e-07, + "loss": 0.8451, + "mean_token_accuracy": 0.7360666990280151, + "num_tokens": 29920472.0, + "step": 994 + }, + { + "epoch": 0.0820483219262802, + "grad_norm": 2.206563949584961, + "learning_rate": 2.731519648255015e-07, + "loss": 0.9453, + "mean_token_accuracy": 0.7157011032104492, + "num_tokens": 29954311.0, + "step": 995 + }, + { + "epoch": 0.08213078255133174, + "grad_norm": 2.15276837348938, + "learning_rate": 2.7342676559494366e-07, + "loss": 1.0112, + "mean_token_accuracy": 0.7019349336624146, + "num_tokens": 29985556.0, + "step": 996 + }, + { + "epoch": 0.08221324317638327, + "grad_norm": 2.2860069274902344, + "learning_rate": 2.737015663643858e-07, + "loss": 0.9774, + "mean_token_accuracy": 0.712268590927124, + "num_tokens": 30017175.0, + "step": 997 + }, + { + "epoch": 0.08229570380143482, + "grad_norm": 2.2362220287323, + "learning_rate": 2.7397636713382795e-07, + "loss": 0.886, + "mean_token_accuracy": 0.7318234443664551, + "num_tokens": 30046402.0, + "step": 998 + }, + { + "epoch": 0.08237816442648635, + "grad_norm": 2.128653049468994, + "learning_rate": 2.742511679032701e-07, + "loss": 0.812, + "mean_token_accuracy": 0.745940625667572, + "num_tokens": 30074859.0, + "step": 999 + }, + { + "epoch": 0.0824606250515379, + "grad_norm": 2.289865732192993, + "learning_rate": 2.745259686727123e-07, + "loss": 0.974, + "mean_token_accuracy": 0.7122156023979187, + "num_tokens": 30102608.0, + "step": 1000 + }, + { + "epoch": 0.08254308567658943, + "grad_norm": 2.871864080429077, + "learning_rate": 2.748007694421544e-07, + "loss": 0.9302, + "mean_token_accuracy": 0.7200660705566406, + "num_tokens": 30122147.0, + "step": 1001 + }, + { + "epoch": 0.08262554630164097, + "grad_norm": 2.1749660968780518, + "learning_rate": 2.7507557021159653e-07, + "loss": 0.9496, + "mean_token_accuracy": 0.7177358269691467, + "num_tokens": 30151322.0, + "step": 1002 + }, + { + "epoch": 0.0827080069266925, + "grad_norm": 2.2160775661468506, + "learning_rate": 2.7535037098103875e-07, + "loss": 1.0096, + "mean_token_accuracy": 0.7124843597412109, + "num_tokens": 30184712.0, + "step": 1003 + }, + { + "epoch": 0.08279046755174405, + "grad_norm": 1.9701815843582153, + "learning_rate": 2.7562517175048087e-07, + "loss": 0.9365, + "mean_token_accuracy": 0.7147359848022461, + "num_tokens": 30221286.0, + "step": 1004 + }, + { + "epoch": 0.08287292817679558, + "grad_norm": 2.477426528930664, + "learning_rate": 2.7589997251992304e-07, + "loss": 0.8745, + "mean_token_accuracy": 0.7318183183670044, + "num_tokens": 30244086.0, + "step": 1005 + }, + { + "epoch": 0.08295538880184712, + "grad_norm": 2.1278700828552246, + "learning_rate": 2.7617477328936516e-07, + "loss": 0.9857, + "mean_token_accuracy": 0.705303430557251, + "num_tokens": 30274740.0, + "step": 1006 + }, + { + "epoch": 0.08303784942689865, + "grad_norm": 2.237201690673828, + "learning_rate": 2.764495740588074e-07, + "loss": 0.8311, + "mean_token_accuracy": 0.7381347417831421, + "num_tokens": 30303034.0, + "step": 1007 + }, + { + "epoch": 0.0831203100519502, + "grad_norm": 2.2826507091522217, + "learning_rate": 2.767243748282495e-07, + "loss": 0.8836, + "mean_token_accuracy": 0.7260825634002686, + "num_tokens": 30329728.0, + "step": 1008 + }, + { + "epoch": 0.08320277067700173, + "grad_norm": 2.1809213161468506, + "learning_rate": 2.769991755976916e-07, + "loss": 0.9225, + "mean_token_accuracy": 0.718695878982544, + "num_tokens": 30359948.0, + "step": 1009 + }, + { + "epoch": 0.08328523130205327, + "grad_norm": 2.5709753036499023, + "learning_rate": 2.772739763671338e-07, + "loss": 0.941, + "mean_token_accuracy": 0.7121010422706604, + "num_tokens": 30383098.0, + "step": 1010 + }, + { + "epoch": 0.0833676919271048, + "grad_norm": 2.241408348083496, + "learning_rate": 2.7754877713657597e-07, + "loss": 0.9623, + "mean_token_accuracy": 0.7128527164459229, + "num_tokens": 30413036.0, + "step": 1011 + }, + { + "epoch": 0.08345015255215635, + "grad_norm": 2.4293150901794434, + "learning_rate": 2.7782357790601814e-07, + "loss": 0.9765, + "mean_token_accuracy": 0.70417720079422, + "num_tokens": 30439778.0, + "step": 1012 + }, + { + "epoch": 0.08353261317720788, + "grad_norm": 2.262840747833252, + "learning_rate": 2.7809837867546026e-07, + "loss": 1.0113, + "mean_token_accuracy": 0.7018774747848511, + "num_tokens": 30469411.0, + "step": 1013 + }, + { + "epoch": 0.08361507380225942, + "grad_norm": 2.359069347381592, + "learning_rate": 2.7837317944490243e-07, + "loss": 1.0004, + "mean_token_accuracy": 0.7036352157592773, + "num_tokens": 30495503.0, + "step": 1014 + }, + { + "epoch": 0.08369753442731095, + "grad_norm": 2.4904444217681885, + "learning_rate": 2.786479802143446e-07, + "loss": 0.9642, + "mean_token_accuracy": 0.7043683528900146, + "num_tokens": 30520439.0, + "step": 1015 + }, + { + "epoch": 0.0837799950523625, + "grad_norm": 2.389646291732788, + "learning_rate": 2.789227809837867e-07, + "loss": 0.8946, + "mean_token_accuracy": 0.7361365556716919, + "num_tokens": 30546150.0, + "step": 1016 + }, + { + "epoch": 0.08386245567741403, + "grad_norm": 2.131420135498047, + "learning_rate": 2.791975817532289e-07, + "loss": 1.0874, + "mean_token_accuracy": 0.677773118019104, + "num_tokens": 30578309.0, + "step": 1017 + }, + { + "epoch": 0.08394491630246557, + "grad_norm": 2.176765203475952, + "learning_rate": 2.79472382522671e-07, + "loss": 0.8512, + "mean_token_accuracy": 0.7299399375915527, + "num_tokens": 30609382.0, + "step": 1018 + }, + { + "epoch": 0.0840273769275171, + "grad_norm": 2.3677029609680176, + "learning_rate": 2.7974718329211323e-07, + "loss": 1.0104, + "mean_token_accuracy": 0.702422022819519, + "num_tokens": 30638687.0, + "step": 1019 + }, + { + "epoch": 0.08410983755256865, + "grad_norm": 2.227888584136963, + "learning_rate": 2.8002198406155535e-07, + "loss": 1.0003, + "mean_token_accuracy": 0.7062816023826599, + "num_tokens": 30668042.0, + "step": 1020 + }, + { + "epoch": 0.08419229817762018, + "grad_norm": 2.4394586086273193, + "learning_rate": 2.802967848309975e-07, + "loss": 1.0256, + "mean_token_accuracy": 0.6974937915802002, + "num_tokens": 30692923.0, + "step": 1021 + }, + { + "epoch": 0.08427475880267173, + "grad_norm": 2.015636920928955, + "learning_rate": 2.8057158560043964e-07, + "loss": 0.9249, + "mean_token_accuracy": 0.7052711248397827, + "num_tokens": 30725925.0, + "step": 1022 + }, + { + "epoch": 0.08435721942772326, + "grad_norm": 2.3264429569244385, + "learning_rate": 2.808463863698818e-07, + "loss": 0.9033, + "mean_token_accuracy": 0.7207982540130615, + "num_tokens": 30755256.0, + "step": 1023 + }, + { + "epoch": 0.0844396800527748, + "grad_norm": 2.245541572570801, + "learning_rate": 2.81121187139324e-07, + "loss": 0.9126, + "mean_token_accuracy": 0.7182707786560059, + "num_tokens": 30784561.0, + "step": 1024 + }, + { + "epoch": 0.08452214067782633, + "grad_norm": 2.6026577949523926, + "learning_rate": 2.813959879087661e-07, + "loss": 0.7978, + "mean_token_accuracy": 0.7572023868560791, + "num_tokens": 30806656.0, + "step": 1025 + }, + { + "epoch": 0.08460460130287788, + "grad_norm": 1.9957308769226074, + "learning_rate": 2.816707886782083e-07, + "loss": 0.8478, + "mean_token_accuracy": 0.7430036067962646, + "num_tokens": 30841540.0, + "step": 1026 + }, + { + "epoch": 0.08468706192792941, + "grad_norm": 2.0186078548431396, + "learning_rate": 2.8194558944765045e-07, + "loss": 0.8874, + "mean_token_accuracy": 0.7293986082077026, + "num_tokens": 30874400.0, + "step": 1027 + }, + { + "epoch": 0.08476952255298095, + "grad_norm": 1.8522944450378418, + "learning_rate": 2.822203902170926e-07, + "loss": 0.9856, + "mean_token_accuracy": 0.7089660167694092, + "num_tokens": 30913017.0, + "step": 1028 + }, + { + "epoch": 0.08485198317803248, + "grad_norm": 2.336148500442505, + "learning_rate": 2.8249519098653474e-07, + "loss": 0.9631, + "mean_token_accuracy": 0.7086480855941772, + "num_tokens": 30940941.0, + "step": 1029 + }, + { + "epoch": 0.08493444380308403, + "grad_norm": 1.8945633172988892, + "learning_rate": 2.8276999175597686e-07, + "loss": 0.7969, + "mean_token_accuracy": 0.7565870881080627, + "num_tokens": 30978494.0, + "step": 1030 + }, + { + "epoch": 0.08501690442813556, + "grad_norm": 2.2932066917419434, + "learning_rate": 2.830447925254191e-07, + "loss": 1.0165, + "mean_token_accuracy": 0.6995201706886292, + "num_tokens": 31008721.0, + "step": 1031 + }, + { + "epoch": 0.0850993650531871, + "grad_norm": 2.2747838497161865, + "learning_rate": 2.833195932948612e-07, + "loss": 0.9455, + "mean_token_accuracy": 0.7063438892364502, + "num_tokens": 31037713.0, + "step": 1032 + }, + { + "epoch": 0.08518182567823863, + "grad_norm": 2.527121067047119, + "learning_rate": 2.8359439406430337e-07, + "loss": 0.959, + "mean_token_accuracy": 0.7072583436965942, + "num_tokens": 31061815.0, + "step": 1033 + }, + { + "epoch": 0.08526428630329018, + "grad_norm": 2.3565213680267334, + "learning_rate": 2.838691948337455e-07, + "loss": 0.9235, + "mean_token_accuracy": 0.7237040400505066, + "num_tokens": 31088605.0, + "step": 1034 + }, + { + "epoch": 0.08534674692834171, + "grad_norm": 2.045905828475952, + "learning_rate": 2.841439956031877e-07, + "loss": 0.7845, + "mean_token_accuracy": 0.7591684460639954, + "num_tokens": 31123895.0, + "step": 1035 + }, + { + "epoch": 0.08542920755339325, + "grad_norm": 2.1624696254730225, + "learning_rate": 2.8441879637262983e-07, + "loss": 0.9238, + "mean_token_accuracy": 0.7124428749084473, + "num_tokens": 31158026.0, + "step": 1036 + }, + { + "epoch": 0.08551166817844479, + "grad_norm": 2.2841830253601074, + "learning_rate": 2.8469359714207195e-07, + "loss": 1.0763, + "mean_token_accuracy": 0.6761177182197571, + "num_tokens": 31189335.0, + "step": 1037 + }, + { + "epoch": 0.08559412880349633, + "grad_norm": 2.1770315170288086, + "learning_rate": 2.849683979115141e-07, + "loss": 0.7644, + "mean_token_accuracy": 0.7674139738082886, + "num_tokens": 31218585.0, + "step": 1038 + }, + { + "epoch": 0.08567658942854788, + "grad_norm": 2.683372974395752, + "learning_rate": 2.852431986809563e-07, + "loss": 0.8324, + "mean_token_accuracy": 0.7385224103927612, + "num_tokens": 31240790.0, + "step": 1039 + }, + { + "epoch": 0.0857590500535994, + "grad_norm": 2.337599039077759, + "learning_rate": 2.8551799945039847e-07, + "loss": 1.1269, + "mean_token_accuracy": 0.6685028672218323, + "num_tokens": 31269252.0, + "step": 1040 + }, + { + "epoch": 0.08584151067865095, + "grad_norm": 1.9621117115020752, + "learning_rate": 2.857928002198406e-07, + "loss": 0.9304, + "mean_token_accuracy": 0.7243366241455078, + "num_tokens": 31304475.0, + "step": 1041 + }, + { + "epoch": 0.08592397130370248, + "grad_norm": 1.8748501539230347, + "learning_rate": 2.8606760098928276e-07, + "loss": 1.0076, + "mean_token_accuracy": 0.7113293409347534, + "num_tokens": 31343708.0, + "step": 1042 + }, + { + "epoch": 0.08600643192875403, + "grad_norm": 2.035475730895996, + "learning_rate": 2.8634240175872493e-07, + "loss": 1.0439, + "mean_token_accuracy": 0.694629430770874, + "num_tokens": 31378599.0, + "step": 1043 + }, + { + "epoch": 0.08608889255380556, + "grad_norm": 1.9863157272338867, + "learning_rate": 2.8661720252816705e-07, + "loss": 1.0285, + "mean_token_accuracy": 0.6946170330047607, + "num_tokens": 31414216.0, + "step": 1044 + }, + { + "epoch": 0.0861713531788571, + "grad_norm": 2.286346912384033, + "learning_rate": 2.868920032976092e-07, + "loss": 0.8896, + "mean_token_accuracy": 0.7367074489593506, + "num_tokens": 31445173.0, + "step": 1045 + }, + { + "epoch": 0.08625381380390863, + "grad_norm": 2.5195605754852295, + "learning_rate": 2.8716680406705134e-07, + "loss": 0.9864, + "mean_token_accuracy": 0.7011181116104126, + "num_tokens": 31470368.0, + "step": 1046 + }, + { + "epoch": 0.08633627442896018, + "grad_norm": 2.7756335735321045, + "learning_rate": 2.8744160483649356e-07, + "loss": 0.9705, + "mean_token_accuracy": 0.7072232961654663, + "num_tokens": 31495552.0, + "step": 1047 + }, + { + "epoch": 0.08641873505401171, + "grad_norm": 1.9694656133651733, + "learning_rate": 2.877164056059357e-07, + "loss": 1.0008, + "mean_token_accuracy": 0.7105634212493896, + "num_tokens": 31531014.0, + "step": 1048 + }, + { + "epoch": 0.08650119567906325, + "grad_norm": 2.034327983856201, + "learning_rate": 2.8799120637537785e-07, + "loss": 0.8637, + "mean_token_accuracy": 0.737167239189148, + "num_tokens": 31564879.0, + "step": 1049 + }, + { + "epoch": 0.08658365630411478, + "grad_norm": 2.1663827896118164, + "learning_rate": 2.8826600714481997e-07, + "loss": 0.9517, + "mean_token_accuracy": 0.7125064134597778, + "num_tokens": 31595038.0, + "step": 1050 + }, + { + "epoch": 0.08666611692916633, + "grad_norm": 2.1557536125183105, + "learning_rate": 2.8854080791426214e-07, + "loss": 0.8591, + "mean_token_accuracy": 0.7299388647079468, + "num_tokens": 31622987.0, + "step": 1051 + }, + { + "epoch": 0.08674857755421786, + "grad_norm": 2.0847394466400146, + "learning_rate": 2.888156086837043e-07, + "loss": 0.8987, + "mean_token_accuracy": 0.7218307256698608, + "num_tokens": 31654669.0, + "step": 1052 + }, + { + "epoch": 0.0868310381792694, + "grad_norm": 2.1658177375793457, + "learning_rate": 2.8909040945314643e-07, + "loss": 0.9599, + "mean_token_accuracy": 0.7136215567588806, + "num_tokens": 31684220.0, + "step": 1053 + }, + { + "epoch": 0.08691349880432093, + "grad_norm": 2.0210671424865723, + "learning_rate": 2.893652102225886e-07, + "loss": 0.8744, + "mean_token_accuracy": 0.7369560599327087, + "num_tokens": 31715019.0, + "step": 1054 + }, + { + "epoch": 0.08699595942937248, + "grad_norm": 2.1056876182556152, + "learning_rate": 2.896400109920308e-07, + "loss": 0.9291, + "mean_token_accuracy": 0.7180016040802002, + "num_tokens": 31746689.0, + "step": 1055 + }, + { + "epoch": 0.08707842005442401, + "grad_norm": 2.1139252185821533, + "learning_rate": 2.8991481176147295e-07, + "loss": 0.7591, + "mean_token_accuracy": 0.7635997533798218, + "num_tokens": 31775187.0, + "step": 1056 + }, + { + "epoch": 0.08716088067947556, + "grad_norm": 2.0928092002868652, + "learning_rate": 2.9018961253091507e-07, + "loss": 0.7719, + "mean_token_accuracy": 0.7538392543792725, + "num_tokens": 31805281.0, + "step": 1057 + }, + { + "epoch": 0.08724334130452709, + "grad_norm": 2.17379093170166, + "learning_rate": 2.904644133003572e-07, + "loss": 0.9764, + "mean_token_accuracy": 0.7180372476577759, + "num_tokens": 31834992.0, + "step": 1058 + }, + { + "epoch": 0.08732580192957863, + "grad_norm": 2.0468878746032715, + "learning_rate": 2.907392140697994e-07, + "loss": 0.9392, + "mean_token_accuracy": 0.7283011674880981, + "num_tokens": 31868389.0, + "step": 1059 + }, + { + "epoch": 0.08740826255463016, + "grad_norm": 2.5709760189056396, + "learning_rate": 2.9101401483924153e-07, + "loss": 1.0359, + "mean_token_accuracy": 0.6881766319274902, + "num_tokens": 31892223.0, + "step": 1060 + }, + { + "epoch": 0.0874907231796817, + "grad_norm": 2.2392632961273193, + "learning_rate": 2.912888156086837e-07, + "loss": 0.9274, + "mean_token_accuracy": 0.712801456451416, + "num_tokens": 31919576.0, + "step": 1061 + }, + { + "epoch": 0.08757318380473324, + "grad_norm": 2.028412342071533, + "learning_rate": 2.915636163781258e-07, + "loss": 0.9814, + "mean_token_accuracy": 0.7065452337265015, + "num_tokens": 31956418.0, + "step": 1062 + }, + { + "epoch": 0.08765564442978478, + "grad_norm": 2.080284595489502, + "learning_rate": 2.9183841714756804e-07, + "loss": 0.8609, + "mean_token_accuracy": 0.737427830696106, + "num_tokens": 31986538.0, + "step": 1063 + }, + { + "epoch": 0.08773810505483631, + "grad_norm": 2.638974905014038, + "learning_rate": 2.9211321791701016e-07, + "loss": 1.0468, + "mean_token_accuracy": 0.6923074722290039, + "num_tokens": 32010084.0, + "step": 1064 + }, + { + "epoch": 0.08782056567988786, + "grad_norm": 2.1586413383483887, + "learning_rate": 2.923880186864523e-07, + "loss": 0.9089, + "mean_token_accuracy": 0.7262108325958252, + "num_tokens": 32042237.0, + "step": 1065 + }, + { + "epoch": 0.08790302630493939, + "grad_norm": 2.1587793827056885, + "learning_rate": 2.9266281945589445e-07, + "loss": 0.9939, + "mean_token_accuracy": 0.7052040696144104, + "num_tokens": 32074121.0, + "step": 1066 + }, + { + "epoch": 0.08798548692999093, + "grad_norm": 2.2095019817352295, + "learning_rate": 2.929376202253366e-07, + "loss": 1.0626, + "mean_token_accuracy": 0.687225341796875, + "num_tokens": 32104997.0, + "step": 1067 + }, + { + "epoch": 0.08806794755504246, + "grad_norm": 2.2054972648620605, + "learning_rate": 2.932124209947788e-07, + "loss": 0.8344, + "mean_token_accuracy": 0.7430638670921326, + "num_tokens": 32134186.0, + "step": 1068 + }, + { + "epoch": 0.08815040818009401, + "grad_norm": 2.1531362533569336, + "learning_rate": 2.934872217642209e-07, + "loss": 0.907, + "mean_token_accuracy": 0.722321629524231, + "num_tokens": 32164045.0, + "step": 1069 + }, + { + "epoch": 0.08823286880514554, + "grad_norm": 2.2492542266845703, + "learning_rate": 2.9376202253366303e-07, + "loss": 0.8852, + "mean_token_accuracy": 0.7300453186035156, + "num_tokens": 32189967.0, + "step": 1070 + }, + { + "epoch": 0.08831532943019708, + "grad_norm": 2.9545676708221436, + "learning_rate": 2.9403682330310526e-07, + "loss": 0.7954, + "mean_token_accuracy": 0.7537407875061035, + "num_tokens": 32223965.0, + "step": 1071 + }, + { + "epoch": 0.08839779005524862, + "grad_norm": 2.537429094314575, + "learning_rate": 2.943116240725474e-07, + "loss": 1.0632, + "mean_token_accuracy": 0.6798567771911621, + "num_tokens": 32247395.0, + "step": 1072 + }, + { + "epoch": 0.08848025068030016, + "grad_norm": 1.9742138385772705, + "learning_rate": 2.9458642484198955e-07, + "loss": 0.9011, + "mean_token_accuracy": 0.7322624921798706, + "num_tokens": 32282100.0, + "step": 1073 + }, + { + "epoch": 0.08856271130535169, + "grad_norm": 2.1841111183166504, + "learning_rate": 2.9486122561143166e-07, + "loss": 0.9518, + "mean_token_accuracy": 0.7195562720298767, + "num_tokens": 32313849.0, + "step": 1074 + }, + { + "epoch": 0.08864517193040324, + "grad_norm": 1.9711133241653442, + "learning_rate": 2.951360263808739e-07, + "loss": 0.7594, + "mean_token_accuracy": 0.76231449842453, + "num_tokens": 32346565.0, + "step": 1075 + }, + { + "epoch": 0.08872763255545477, + "grad_norm": 2.192531108856201, + "learning_rate": 2.95410827150316e-07, + "loss": 0.9867, + "mean_token_accuracy": 0.710732102394104, + "num_tokens": 32376717.0, + "step": 1076 + }, + { + "epoch": 0.08881009318050631, + "grad_norm": 2.1627368927001953, + "learning_rate": 2.956856279197581e-07, + "loss": 0.9858, + "mean_token_accuracy": 0.7013511657714844, + "num_tokens": 32408618.0, + "step": 1077 + }, + { + "epoch": 0.08889255380555784, + "grad_norm": 2.079029083251953, + "learning_rate": 2.959604286892003e-07, + "loss": 0.9319, + "mean_token_accuracy": 0.7250794172286987, + "num_tokens": 32440177.0, + "step": 1078 + }, + { + "epoch": 0.08897501443060939, + "grad_norm": 2.0550215244293213, + "learning_rate": 2.9623522945864247e-07, + "loss": 0.8371, + "mean_token_accuracy": 0.7512809038162231, + "num_tokens": 32470565.0, + "step": 1079 + }, + { + "epoch": 0.08905747505566092, + "grad_norm": 2.176332473754883, + "learning_rate": 2.9651003022808464e-07, + "loss": 0.9069, + "mean_token_accuracy": 0.7184207439422607, + "num_tokens": 32498304.0, + "step": 1080 + }, + { + "epoch": 0.08913993568071246, + "grad_norm": 1.9705471992492676, + "learning_rate": 2.9678483099752676e-07, + "loss": 0.9117, + "mean_token_accuracy": 0.7297847867012024, + "num_tokens": 32531101.0, + "step": 1081 + }, + { + "epoch": 0.08922239630576399, + "grad_norm": 1.8337948322296143, + "learning_rate": 2.9705963176696893e-07, + "loss": 0.881, + "mean_token_accuracy": 0.741973876953125, + "num_tokens": 32570402.0, + "step": 1082 + }, + { + "epoch": 0.08930485693081554, + "grad_norm": 2.7663960456848145, + "learning_rate": 2.973344325364111e-07, + "loss": 1.0577, + "mean_token_accuracy": 0.6940232515335083, + "num_tokens": 32593269.0, + "step": 1083 + }, + { + "epoch": 0.08938731755586707, + "grad_norm": 2.223400115966797, + "learning_rate": 2.976092333058532e-07, + "loss": 1.0283, + "mean_token_accuracy": 0.6934139132499695, + "num_tokens": 32622022.0, + "step": 1084 + }, + { + "epoch": 0.08946977818091861, + "grad_norm": 2.4147989749908447, + "learning_rate": 2.978840340752954e-07, + "loss": 0.9677, + "mean_token_accuracy": 0.7028720378875732, + "num_tokens": 32649209.0, + "step": 1085 + }, + { + "epoch": 0.08955223880597014, + "grad_norm": 2.093050956726074, + "learning_rate": 2.981588348447375e-07, + "loss": 1.0397, + "mean_token_accuracy": 0.6852442026138306, + "num_tokens": 32683682.0, + "step": 1086 + }, + { + "epoch": 0.08963469943102169, + "grad_norm": 1.993531584739685, + "learning_rate": 2.9843363561417974e-07, + "loss": 1.0078, + "mean_token_accuracy": 0.7041193842887878, + "num_tokens": 32719681.0, + "step": 1087 + }, + { + "epoch": 0.08971716005607322, + "grad_norm": 2.1069695949554443, + "learning_rate": 2.9870843638362185e-07, + "loss": 0.9522, + "mean_token_accuracy": 0.7070248126983643, + "num_tokens": 32752335.0, + "step": 1088 + }, + { + "epoch": 0.08979962068112476, + "grad_norm": 2.2138004302978516, + "learning_rate": 2.98983237153064e-07, + "loss": 0.9299, + "mean_token_accuracy": 0.7179133892059326, + "num_tokens": 32783849.0, + "step": 1089 + }, + { + "epoch": 0.0898820813061763, + "grad_norm": 2.4244894981384277, + "learning_rate": 2.9925803792250614e-07, + "loss": 1.0095, + "mean_token_accuracy": 0.6935750246047974, + "num_tokens": 32808394.0, + "step": 1090 + }, + { + "epoch": 0.08996454193122784, + "grad_norm": 2.006889581680298, + "learning_rate": 2.995328386919483e-07, + "loss": 0.9551, + "mean_token_accuracy": 0.715869665145874, + "num_tokens": 32844089.0, + "step": 1091 + }, + { + "epoch": 0.09004700255627937, + "grad_norm": 1.941454291343689, + "learning_rate": 2.998076394613905e-07, + "loss": 0.9376, + "mean_token_accuracy": 0.7123960256576538, + "num_tokens": 32878767.0, + "step": 1092 + }, + { + "epoch": 0.09012946318133092, + "grad_norm": 2.029859781265259, + "learning_rate": 3.000824402308326e-07, + "loss": 0.6708, + "mean_token_accuracy": 0.778471827507019, + "num_tokens": 32908648.0, + "step": 1093 + }, + { + "epoch": 0.09021192380638245, + "grad_norm": 2.2889864444732666, + "learning_rate": 3.003572410002748e-07, + "loss": 0.9624, + "mean_token_accuracy": 0.7139971256256104, + "num_tokens": 32933917.0, + "step": 1094 + }, + { + "epoch": 0.09029438443143399, + "grad_norm": 2.011016607284546, + "learning_rate": 3.0063204176971695e-07, + "loss": 0.906, + "mean_token_accuracy": 0.7270622849464417, + "num_tokens": 32966693.0, + "step": 1095 + }, + { + "epoch": 0.09037684505648552, + "grad_norm": 2.1921722888946533, + "learning_rate": 3.009068425391591e-07, + "loss": 0.8885, + "mean_token_accuracy": 0.7386051416397095, + "num_tokens": 32993985.0, + "step": 1096 + }, + { + "epoch": 0.09045930568153707, + "grad_norm": 1.8977792263031006, + "learning_rate": 3.0118164330860124e-07, + "loss": 0.7747, + "mean_token_accuracy": 0.7614308595657349, + "num_tokens": 33029542.0, + "step": 1097 + }, + { + "epoch": 0.0905417663065886, + "grad_norm": 2.3306639194488525, + "learning_rate": 3.0145644407804336e-07, + "loss": 0.8667, + "mean_token_accuracy": 0.7304731607437134, + "num_tokens": 33053216.0, + "step": 1098 + }, + { + "epoch": 0.09062422693164014, + "grad_norm": 2.267002820968628, + "learning_rate": 3.017312448474856e-07, + "loss": 0.8328, + "mean_token_accuracy": 0.7374740839004517, + "num_tokens": 33081303.0, + "step": 1099 + }, + { + "epoch": 0.09070668755669167, + "grad_norm": 2.0387964248657227, + "learning_rate": 3.020060456169277e-07, + "loss": 0.8581, + "mean_token_accuracy": 0.7373381853103638, + "num_tokens": 33114932.0, + "step": 1100 + }, + { + "epoch": 0.09078914818174322, + "grad_norm": 2.0769221782684326, + "learning_rate": 3.0228084638636987e-07, + "loss": 0.8689, + "mean_token_accuracy": 0.7335523366928101, + "num_tokens": 33146995.0, + "step": 1101 + }, + { + "epoch": 0.09087160880679475, + "grad_norm": 2.0685975551605225, + "learning_rate": 3.02555647155812e-07, + "loss": 0.8283, + "mean_token_accuracy": 0.7405304908752441, + "num_tokens": 33177665.0, + "step": 1102 + }, + { + "epoch": 0.0909540694318463, + "grad_norm": 2.20536732673645, + "learning_rate": 3.028304479252542e-07, + "loss": 0.7845, + "mean_token_accuracy": 0.7589049339294434, + "num_tokens": 33206159.0, + "step": 1103 + }, + { + "epoch": 0.09103653005689784, + "grad_norm": 2.2754783630371094, + "learning_rate": 3.0310524869469633e-07, + "loss": 1.1348, + "mean_token_accuracy": 0.6734490394592285, + "num_tokens": 33235444.0, + "step": 1104 + }, + { + "epoch": 0.09111899068194937, + "grad_norm": 2.254277229309082, + "learning_rate": 3.0338004946413845e-07, + "loss": 0.9353, + "mean_token_accuracy": 0.713744044303894, + "num_tokens": 33265502.0, + "step": 1105 + }, + { + "epoch": 0.09120145130700091, + "grad_norm": 2.280832052230835, + "learning_rate": 3.036548502335806e-07, + "loss": 0.993, + "mean_token_accuracy": 0.713797926902771, + "num_tokens": 33292928.0, + "step": 1106 + }, + { + "epoch": 0.09128391193205244, + "grad_norm": 2.3929786682128906, + "learning_rate": 3.039296510030228e-07, + "loss": 0.9613, + "mean_token_accuracy": 0.7105299830436707, + "num_tokens": 33321478.0, + "step": 1107 + }, + { + "epoch": 0.09136637255710399, + "grad_norm": 2.264404296875, + "learning_rate": 3.0420445177246497e-07, + "loss": 0.835, + "mean_token_accuracy": 0.7483552694320679, + "num_tokens": 33348424.0, + "step": 1108 + }, + { + "epoch": 0.09144883318215552, + "grad_norm": 2.1214237213134766, + "learning_rate": 3.044792525419071e-07, + "loss": 0.9861, + "mean_token_accuracy": 0.7058748602867126, + "num_tokens": 33378893.0, + "step": 1109 + }, + { + "epoch": 0.09153129380720706, + "grad_norm": 2.373533010482788, + "learning_rate": 3.0475405331134926e-07, + "loss": 0.9526, + "mean_token_accuracy": 0.7098721265792847, + "num_tokens": 33405336.0, + "step": 1110 + }, + { + "epoch": 0.0916137544322586, + "grad_norm": 2.1972739696502686, + "learning_rate": 3.0502885408079143e-07, + "loss": 0.8549, + "mean_token_accuracy": 0.7319930791854858, + "num_tokens": 33434565.0, + "step": 1111 + }, + { + "epoch": 0.09169621505731014, + "grad_norm": 1.986839771270752, + "learning_rate": 3.0530365485023355e-07, + "loss": 0.8336, + "mean_token_accuracy": 0.7468766570091248, + "num_tokens": 33469226.0, + "step": 1112 + }, + { + "epoch": 0.09177867568236167, + "grad_norm": 2.1861250400543213, + "learning_rate": 3.055784556196757e-07, + "loss": 0.958, + "mean_token_accuracy": 0.7153369784355164, + "num_tokens": 33498363.0, + "step": 1113 + }, + { + "epoch": 0.09186113630741322, + "grad_norm": 2.0817577838897705, + "learning_rate": 3.0585325638911784e-07, + "loss": 0.923, + "mean_token_accuracy": 0.7179930806159973, + "num_tokens": 33528440.0, + "step": 1114 + }, + { + "epoch": 0.09194359693246475, + "grad_norm": 1.9437776803970337, + "learning_rate": 3.0612805715856006e-07, + "loss": 0.8976, + "mean_token_accuracy": 0.7279585599899292, + "num_tokens": 33568013.0, + "step": 1115 + }, + { + "epoch": 0.09202605755751629, + "grad_norm": 2.4607045650482178, + "learning_rate": 3.064028579280022e-07, + "loss": 0.9232, + "mean_token_accuracy": 0.7235750555992126, + "num_tokens": 33593234.0, + "step": 1116 + }, + { + "epoch": 0.09210851818256782, + "grad_norm": 2.275899648666382, + "learning_rate": 3.0667765869744435e-07, + "loss": 1.0464, + "mean_token_accuracy": 0.6926898956298828, + "num_tokens": 33623936.0, + "step": 1117 + }, + { + "epoch": 0.09219097880761937, + "grad_norm": 1.987847089767456, + "learning_rate": 3.0695245946688647e-07, + "loss": 0.9721, + "mean_token_accuracy": 0.7037615776062012, + "num_tokens": 33660433.0, + "step": 1118 + }, + { + "epoch": 0.0922734394326709, + "grad_norm": 2.019848108291626, + "learning_rate": 3.0722726023632864e-07, + "loss": 0.9626, + "mean_token_accuracy": 0.7142197489738464, + "num_tokens": 33694600.0, + "step": 1119 + }, + { + "epoch": 0.09235590005772244, + "grad_norm": 1.903613805770874, + "learning_rate": 3.075020610057708e-07, + "loss": 0.7571, + "mean_token_accuracy": 0.7618793845176697, + "num_tokens": 33729947.0, + "step": 1120 + }, + { + "epoch": 0.09243836068277397, + "grad_norm": 2.074462652206421, + "learning_rate": 3.0777686177521293e-07, + "loss": 0.8107, + "mean_token_accuracy": 0.7498548030853271, + "num_tokens": 33759839.0, + "step": 1121 + }, + { + "epoch": 0.09252082130782552, + "grad_norm": 1.8288863897323608, + "learning_rate": 3.080516625446551e-07, + "loss": 0.986, + "mean_token_accuracy": 0.7140556573867798, + "num_tokens": 33799182.0, + "step": 1122 + }, + { + "epoch": 0.09260328193287705, + "grad_norm": 1.8974699974060059, + "learning_rate": 3.083264633140973e-07, + "loss": 0.7973, + "mean_token_accuracy": 0.7571432590484619, + "num_tokens": 33833412.0, + "step": 1123 + }, + { + "epoch": 0.0926857425579286, + "grad_norm": 2.563840389251709, + "learning_rate": 3.0860126408353945e-07, + "loss": 0.9981, + "mean_token_accuracy": 0.7108879089355469, + "num_tokens": 33855286.0, + "step": 1124 + }, + { + "epoch": 0.09276820318298012, + "grad_norm": 2.3469760417938232, + "learning_rate": 3.0887606485298157e-07, + "loss": 0.7777, + "mean_token_accuracy": 0.7579464912414551, + "num_tokens": 33881664.0, + "step": 1125 + }, + { + "epoch": 0.09285066380803167, + "grad_norm": 2.8281350135803223, + "learning_rate": 3.091508656224237e-07, + "loss": 1.044, + "mean_token_accuracy": 0.6858003735542297, + "num_tokens": 33902199.0, + "step": 1126 + }, + { + "epoch": 0.0929331244330832, + "grad_norm": 1.9302431344985962, + "learning_rate": 3.094256663918659e-07, + "loss": 0.958, + "mean_token_accuracy": 0.7185426950454712, + "num_tokens": 33937490.0, + "step": 1127 + }, + { + "epoch": 0.09301558505813474, + "grad_norm": 2.261676549911499, + "learning_rate": 3.0970046716130803e-07, + "loss": 0.9681, + "mean_token_accuracy": 0.713731050491333, + "num_tokens": 33968510.0, + "step": 1128 + }, + { + "epoch": 0.09309804568318628, + "grad_norm": 2.1471030712127686, + "learning_rate": 3.099752679307502e-07, + "loss": 0.7752, + "mean_token_accuracy": 0.7663387060165405, + "num_tokens": 33997941.0, + "step": 1129 + }, + { + "epoch": 0.09318050630823782, + "grad_norm": 2.0691583156585693, + "learning_rate": 3.102500687001923e-07, + "loss": 0.9417, + "mean_token_accuracy": 0.710413932800293, + "num_tokens": 34030805.0, + "step": 1130 + }, + { + "epoch": 0.09326296693328935, + "grad_norm": 2.4229576587677, + "learning_rate": 3.1052486946963454e-07, + "loss": 0.9767, + "mean_token_accuracy": 0.7035362124443054, + "num_tokens": 34061237.0, + "step": 1131 + }, + { + "epoch": 0.0933454275583409, + "grad_norm": 2.0687148571014404, + "learning_rate": 3.1079967023907666e-07, + "loss": 0.8703, + "mean_token_accuracy": 0.7367270588874817, + "num_tokens": 34093546.0, + "step": 1132 + }, + { + "epoch": 0.09342788818339243, + "grad_norm": 1.739057183265686, + "learning_rate": 3.110744710085188e-07, + "loss": 0.9055, + "mean_token_accuracy": 0.7305870056152344, + "num_tokens": 34134611.0, + "step": 1133 + }, + { + "epoch": 0.09351034880844397, + "grad_norm": 1.9687836170196533, + "learning_rate": 3.1134927177796095e-07, + "loss": 0.9547, + "mean_token_accuracy": 0.7172346711158752, + "num_tokens": 34169032.0, + "step": 1134 + }, + { + "epoch": 0.0935928094334955, + "grad_norm": 2.0562760829925537, + "learning_rate": 3.116240725474031e-07, + "loss": 0.7304, + "mean_token_accuracy": 0.7713539600372314, + "num_tokens": 34197859.0, + "step": 1135 + }, + { + "epoch": 0.09367527005854705, + "grad_norm": 2.3685061931610107, + "learning_rate": 3.118988733168453e-07, + "loss": 1.0212, + "mean_token_accuracy": 0.6881908774375916, + "num_tokens": 34225558.0, + "step": 1136 + }, + { + "epoch": 0.09375773068359858, + "grad_norm": 2.1472327709198, + "learning_rate": 3.121736740862874e-07, + "loss": 0.9131, + "mean_token_accuracy": 0.7225595116615295, + "num_tokens": 34254941.0, + "step": 1137 + }, + { + "epoch": 0.09384019130865012, + "grad_norm": 2.364595413208008, + "learning_rate": 3.124484748557296e-07, + "loss": 1.0972, + "mean_token_accuracy": 0.6737436652183533, + "num_tokens": 34282044.0, + "step": 1138 + }, + { + "epoch": 0.09392265193370165, + "grad_norm": 1.9754749536514282, + "learning_rate": 3.1272327562517176e-07, + "loss": 1.071, + "mean_token_accuracy": 0.6747012734413147, + "num_tokens": 34317384.0, + "step": 1139 + }, + { + "epoch": 0.0940051125587532, + "grad_norm": 2.137821674346924, + "learning_rate": 3.129980763946139e-07, + "loss": 0.9258, + "mean_token_accuracy": 0.7290209531784058, + "num_tokens": 34349748.0, + "step": 1140 + }, + { + "epoch": 0.09408757318380473, + "grad_norm": 2.0486295223236084, + "learning_rate": 3.1327287716405605e-07, + "loss": 0.985, + "mean_token_accuracy": 0.7163176536560059, + "num_tokens": 34380808.0, + "step": 1141 + }, + { + "epoch": 0.09417003380885627, + "grad_norm": 2.05979323387146, + "learning_rate": 3.1354767793349817e-07, + "loss": 0.957, + "mean_token_accuracy": 0.717144250869751, + "num_tokens": 34415354.0, + "step": 1142 + }, + { + "epoch": 0.0942524944339078, + "grad_norm": 1.9569131135940552, + "learning_rate": 3.138224787029404e-07, + "loss": 0.8299, + "mean_token_accuracy": 0.7534568309783936, + "num_tokens": 34449547.0, + "step": 1143 + }, + { + "epoch": 0.09433495505895935, + "grad_norm": 2.257256507873535, + "learning_rate": 3.140972794723825e-07, + "loss": 0.9041, + "mean_token_accuracy": 0.7247443199157715, + "num_tokens": 34477163.0, + "step": 1144 + }, + { + "epoch": 0.09441741568401088, + "grad_norm": 1.8036105632781982, + "learning_rate": 3.143720802418247e-07, + "loss": 0.6849, + "mean_token_accuracy": 0.7833101749420166, + "num_tokens": 34509464.0, + "step": 1145 + }, + { + "epoch": 0.09449987630906242, + "grad_norm": 2.079829216003418, + "learning_rate": 3.146468810112668e-07, + "loss": 0.8185, + "mean_token_accuracy": 0.7422221899032593, + "num_tokens": 34542334.0, + "step": 1146 + }, + { + "epoch": 0.09458233693411396, + "grad_norm": 1.9057929515838623, + "learning_rate": 3.1492168178070897e-07, + "loss": 0.8586, + "mean_token_accuracy": 0.7343847751617432, + "num_tokens": 34578719.0, + "step": 1147 + }, + { + "epoch": 0.0946647975591655, + "grad_norm": 2.268904685974121, + "learning_rate": 3.1519648255015114e-07, + "loss": 0.9476, + "mean_token_accuracy": 0.7121584415435791, + "num_tokens": 34604466.0, + "step": 1148 + }, + { + "epoch": 0.09474725818421703, + "grad_norm": 2.1659350395202637, + "learning_rate": 3.1547128331959326e-07, + "loss": 0.8699, + "mean_token_accuracy": 0.7318578958511353, + "num_tokens": 34632307.0, + "step": 1149 + }, + { + "epoch": 0.09482971880926858, + "grad_norm": 2.149418592453003, + "learning_rate": 3.1574608408903543e-07, + "loss": 1.0219, + "mean_token_accuracy": 0.6874595880508423, + "num_tokens": 34664447.0, + "step": 1150 + }, + { + "epoch": 0.09491217943432011, + "grad_norm": 2.337836265563965, + "learning_rate": 3.160208848584776e-07, + "loss": 0.9688, + "mean_token_accuracy": 0.7198048830032349, + "num_tokens": 34691613.0, + "step": 1151 + }, + { + "epoch": 0.09499464005937165, + "grad_norm": 1.9608265161514282, + "learning_rate": 3.162956856279198e-07, + "loss": 0.8584, + "mean_token_accuracy": 0.7390762567520142, + "num_tokens": 34727824.0, + "step": 1152 + }, + { + "epoch": 0.09507710068442318, + "grad_norm": 2.4465339183807373, + "learning_rate": 3.165704863973619e-07, + "loss": 0.9706, + "mean_token_accuracy": 0.7053803205490112, + "num_tokens": 34752100.0, + "step": 1153 + }, + { + "epoch": 0.09515956130947473, + "grad_norm": 2.471470355987549, + "learning_rate": 3.16845287166804e-07, + "loss": 0.9743, + "mean_token_accuracy": 0.7069209218025208, + "num_tokens": 34778411.0, + "step": 1154 + }, + { + "epoch": 0.09524202193452626, + "grad_norm": 1.9979056119918823, + "learning_rate": 3.1712008793624624e-07, + "loss": 1.0186, + "mean_token_accuracy": 0.6973321437835693, + "num_tokens": 34812704.0, + "step": 1155 + }, + { + "epoch": 0.0953244825595778, + "grad_norm": 2.106297016143799, + "learning_rate": 3.1739488870568836e-07, + "loss": 0.8242, + "mean_token_accuracy": 0.737174928188324, + "num_tokens": 34844700.0, + "step": 1156 + }, + { + "epoch": 0.09540694318462933, + "grad_norm": 2.081528902053833, + "learning_rate": 3.1766968947513053e-07, + "loss": 0.9496, + "mean_token_accuracy": 0.7184405326843262, + "num_tokens": 34876175.0, + "step": 1157 + }, + { + "epoch": 0.09548940380968088, + "grad_norm": 2.1761796474456787, + "learning_rate": 3.1794449024457265e-07, + "loss": 0.875, + "mean_token_accuracy": 0.73323655128479, + "num_tokens": 34906932.0, + "step": 1158 + }, + { + "epoch": 0.09557186443473241, + "grad_norm": 2.444899559020996, + "learning_rate": 3.1821929101401487e-07, + "loss": 0.98, + "mean_token_accuracy": 0.7156383991241455, + "num_tokens": 34935521.0, + "step": 1159 + }, + { + "epoch": 0.09565432505978395, + "grad_norm": 2.1642181873321533, + "learning_rate": 3.18494091783457e-07, + "loss": 0.8311, + "mean_token_accuracy": 0.7430585622787476, + "num_tokens": 34963839.0, + "step": 1160 + }, + { + "epoch": 0.09573678568483548, + "grad_norm": 2.024779796600342, + "learning_rate": 3.187688925528991e-07, + "loss": 0.8693, + "mean_token_accuracy": 0.7435390949249268, + "num_tokens": 34994274.0, + "step": 1161 + }, + { + "epoch": 0.09581924630988703, + "grad_norm": 2.0712294578552246, + "learning_rate": 3.190436933223413e-07, + "loss": 0.8042, + "mean_token_accuracy": 0.7511001229286194, + "num_tokens": 35028409.0, + "step": 1162 + }, + { + "epoch": 0.09590170693493856, + "grad_norm": 2.2486071586608887, + "learning_rate": 3.1931849409178345e-07, + "loss": 0.995, + "mean_token_accuracy": 0.7051641941070557, + "num_tokens": 35055813.0, + "step": 1163 + }, + { + "epoch": 0.0959841675599901, + "grad_norm": 2.1149213314056396, + "learning_rate": 3.195932948612256e-07, + "loss": 0.9004, + "mean_token_accuracy": 0.7230446338653564, + "num_tokens": 35084819.0, + "step": 1164 + }, + { + "epoch": 0.09606662818504164, + "grad_norm": 2.3069653511047363, + "learning_rate": 3.1986809563066774e-07, + "loss": 1.0278, + "mean_token_accuracy": 0.701704740524292, + "num_tokens": 35112421.0, + "step": 1165 + }, + { + "epoch": 0.09614908881009318, + "grad_norm": 2.0294811725616455, + "learning_rate": 3.2014289640010986e-07, + "loss": 0.7996, + "mean_token_accuracy": 0.7518236041069031, + "num_tokens": 35142970.0, + "step": 1166 + }, + { + "epoch": 0.09623154943514471, + "grad_norm": 2.0922365188598633, + "learning_rate": 3.204176971695521e-07, + "loss": 0.956, + "mean_token_accuracy": 0.71159827709198, + "num_tokens": 35176191.0, + "step": 1167 + }, + { + "epoch": 0.09631401006019626, + "grad_norm": 2.256434440612793, + "learning_rate": 3.206924979389942e-07, + "loss": 0.9538, + "mean_token_accuracy": 0.7138963937759399, + "num_tokens": 35203190.0, + "step": 1168 + }, + { + "epoch": 0.0963964706852478, + "grad_norm": 2.1842381954193115, + "learning_rate": 3.209672987084364e-07, + "loss": 0.789, + "mean_token_accuracy": 0.753588080406189, + "num_tokens": 35231796.0, + "step": 1169 + }, + { + "epoch": 0.09647893131029933, + "grad_norm": 2.351935863494873, + "learning_rate": 3.212420994778785e-07, + "loss": 0.8685, + "mean_token_accuracy": 0.7257114052772522, + "num_tokens": 35256887.0, + "step": 1170 + }, + { + "epoch": 0.09656139193535088, + "grad_norm": 2.3235630989074707, + "learning_rate": 3.215169002473207e-07, + "loss": 0.9205, + "mean_token_accuracy": 0.7181037068367004, + "num_tokens": 35283417.0, + "step": 1171 + }, + { + "epoch": 0.09664385256040241, + "grad_norm": 2.2404892444610596, + "learning_rate": 3.2179170101676284e-07, + "loss": 0.94, + "mean_token_accuracy": 0.7227078080177307, + "num_tokens": 35311070.0, + "step": 1172 + }, + { + "epoch": 0.09672631318545395, + "grad_norm": 1.8856738805770874, + "learning_rate": 3.2206650178620495e-07, + "loss": 0.851, + "mean_token_accuracy": 0.7430887818336487, + "num_tokens": 35347904.0, + "step": 1173 + }, + { + "epoch": 0.09680877381050548, + "grad_norm": 2.2733612060546875, + "learning_rate": 3.223413025556471e-07, + "loss": 0.9641, + "mean_token_accuracy": 0.7096889019012451, + "num_tokens": 35376773.0, + "step": 1174 + }, + { + "epoch": 0.09689123443555703, + "grad_norm": 1.9778097867965698, + "learning_rate": 3.226161033250893e-07, + "loss": 0.7768, + "mean_token_accuracy": 0.7513629198074341, + "num_tokens": 35409016.0, + "step": 1175 + }, + { + "epoch": 0.09697369506060856, + "grad_norm": 2.182356834411621, + "learning_rate": 3.2289090409453147e-07, + "loss": 0.9102, + "mean_token_accuracy": 0.7230268120765686, + "num_tokens": 35441809.0, + "step": 1176 + }, + { + "epoch": 0.0970561556856601, + "grad_norm": 1.828187346458435, + "learning_rate": 3.231657048639736e-07, + "loss": 0.8811, + "mean_token_accuracy": 0.7385594844818115, + "num_tokens": 35478687.0, + "step": 1177 + }, + { + "epoch": 0.09713861631071163, + "grad_norm": 1.8558286428451538, + "learning_rate": 3.2344050563341576e-07, + "loss": 0.7195, + "mean_token_accuracy": 0.7723275423049927, + "num_tokens": 35513763.0, + "step": 1178 + }, + { + "epoch": 0.09722107693576318, + "grad_norm": 1.8502055406570435, + "learning_rate": 3.2371530640285793e-07, + "loss": 0.8648, + "mean_token_accuracy": 0.737212598323822, + "num_tokens": 35550653.0, + "step": 1179 + }, + { + "epoch": 0.09730353756081471, + "grad_norm": 2.2683541774749756, + "learning_rate": 3.2399010717230005e-07, + "loss": 0.9456, + "mean_token_accuracy": 0.7132858633995056, + "num_tokens": 35576681.0, + "step": 1180 + }, + { + "epoch": 0.09738599818586625, + "grad_norm": 2.298693895339966, + "learning_rate": 3.242649079417422e-07, + "loss": 1.0252, + "mean_token_accuracy": 0.6914993524551392, + "num_tokens": 35604860.0, + "step": 1181 + }, + { + "epoch": 0.09746845881091779, + "grad_norm": 1.880712628364563, + "learning_rate": 3.2453970871118434e-07, + "loss": 0.8168, + "mean_token_accuracy": 0.7435126304626465, + "num_tokens": 35640971.0, + "step": 1182 + }, + { + "epoch": 0.09755091943596933, + "grad_norm": 2.071467638015747, + "learning_rate": 3.2481450948062656e-07, + "loss": 0.8811, + "mean_token_accuracy": 0.7333333492279053, + "num_tokens": 35672399.0, + "step": 1183 + }, + { + "epoch": 0.09763338006102086, + "grad_norm": 2.1021111011505127, + "learning_rate": 3.250893102500687e-07, + "loss": 0.9064, + "mean_token_accuracy": 0.7192788124084473, + "num_tokens": 35705046.0, + "step": 1184 + }, + { + "epoch": 0.0977158406860724, + "grad_norm": 1.966060996055603, + "learning_rate": 3.2536411101951085e-07, + "loss": 0.9372, + "mean_token_accuracy": 0.7199410200119019, + "num_tokens": 35738812.0, + "step": 1185 + }, + { + "epoch": 0.09779830131112394, + "grad_norm": 2.0097625255584717, + "learning_rate": 3.2563891178895297e-07, + "loss": 0.9502, + "mean_token_accuracy": 0.7127976417541504, + "num_tokens": 35771445.0, + "step": 1186 + }, + { + "epoch": 0.09788076193617548, + "grad_norm": 2.4167275428771973, + "learning_rate": 3.2591371255839514e-07, + "loss": 0.9314, + "mean_token_accuracy": 0.7204985618591309, + "num_tokens": 35795377.0, + "step": 1187 + }, + { + "epoch": 0.09796322256122701, + "grad_norm": 2.085597276687622, + "learning_rate": 3.261885133278373e-07, + "loss": 0.8158, + "mean_token_accuracy": 0.7605792880058289, + "num_tokens": 35824576.0, + "step": 1188 + }, + { + "epoch": 0.09804568318627856, + "grad_norm": 2.21186900138855, + "learning_rate": 3.2646331409727944e-07, + "loss": 0.9939, + "mean_token_accuracy": 0.7024415731430054, + "num_tokens": 35854196.0, + "step": 1189 + }, + { + "epoch": 0.09812814381133009, + "grad_norm": 2.3287642002105713, + "learning_rate": 3.267381148667216e-07, + "loss": 0.964, + "mean_token_accuracy": 0.7005327939987183, + "num_tokens": 35881189.0, + "step": 1190 + }, + { + "epoch": 0.09821060443638163, + "grad_norm": 1.98771071434021, + "learning_rate": 3.270129156361638e-07, + "loss": 0.9504, + "mean_token_accuracy": 0.7076086401939392, + "num_tokens": 35914951.0, + "step": 1191 + }, + { + "epoch": 0.09829306506143316, + "grad_norm": 1.9342187643051147, + "learning_rate": 3.2728771640560595e-07, + "loss": 0.752, + "mean_token_accuracy": 0.7587350606918335, + "num_tokens": 35947009.0, + "step": 1192 + }, + { + "epoch": 0.09837552568648471, + "grad_norm": 2.0190329551696777, + "learning_rate": 3.2756251717504807e-07, + "loss": 1.0597, + "mean_token_accuracy": 0.6921601891517639, + "num_tokens": 35980816.0, + "step": 1193 + }, + { + "epoch": 0.09845798631153624, + "grad_norm": 2.211451292037964, + "learning_rate": 3.278373179444902e-07, + "loss": 0.9471, + "mean_token_accuracy": 0.7198120355606079, + "num_tokens": 36009846.0, + "step": 1194 + }, + { + "epoch": 0.09854044693658778, + "grad_norm": 2.278978109359741, + "learning_rate": 3.281121187139324e-07, + "loss": 1.0492, + "mean_token_accuracy": 0.6837566494941711, + "num_tokens": 36038278.0, + "step": 1195 + }, + { + "epoch": 0.09862290756163931, + "grad_norm": 2.6022675037384033, + "learning_rate": 3.2838691948337453e-07, + "loss": 0.9824, + "mean_token_accuracy": 0.6990740299224854, + "num_tokens": 36059527.0, + "step": 1196 + }, + { + "epoch": 0.09870536818669086, + "grad_norm": 2.012819528579712, + "learning_rate": 3.286617202528167e-07, + "loss": 0.9482, + "mean_token_accuracy": 0.7130134105682373, + "num_tokens": 36093855.0, + "step": 1197 + }, + { + "epoch": 0.09878782881174239, + "grad_norm": 2.22611141204834, + "learning_rate": 3.289365210222588e-07, + "loss": 1.0024, + "mean_token_accuracy": 0.7175321578979492, + "num_tokens": 36119702.0, + "step": 1198 + }, + { + "epoch": 0.09887028943679393, + "grad_norm": 1.9601906538009644, + "learning_rate": 3.2921132179170104e-07, + "loss": 0.874, + "mean_token_accuracy": 0.7312037348747253, + "num_tokens": 36153373.0, + "step": 1199 + }, + { + "epoch": 0.09895275006184547, + "grad_norm": 2.5778019428253174, + "learning_rate": 3.2948612256114316e-07, + "loss": 0.9552, + "mean_token_accuracy": 0.7071126103401184, + "num_tokens": 36173832.0, + "step": 1200 + }, + { + "epoch": 0.09903521068689701, + "grad_norm": 2.52750563621521, + "learning_rate": 3.297609233305853e-07, + "loss": 0.824, + "mean_token_accuracy": 0.7347512245178223, + "num_tokens": 36199892.0, + "step": 1201 + }, + { + "epoch": 0.09911767131194854, + "grad_norm": 2.424367904663086, + "learning_rate": 3.3003572410002745e-07, + "loss": 1.0973, + "mean_token_accuracy": 0.6783713102340698, + "num_tokens": 36226392.0, + "step": 1202 + }, + { + "epoch": 0.09920013193700009, + "grad_norm": 2.5237672328948975, + "learning_rate": 3.303105248694696e-07, + "loss": 1.0971, + "mean_token_accuracy": 0.6853283643722534, + "num_tokens": 36251412.0, + "step": 1203 + }, + { + "epoch": 0.09928259256205162, + "grad_norm": 2.2985856533050537, + "learning_rate": 3.305853256389118e-07, + "loss": 0.8082, + "mean_token_accuracy": 0.7504374980926514, + "num_tokens": 36276415.0, + "step": 1204 + }, + { + "epoch": 0.09936505318710316, + "grad_norm": 1.7943446636199951, + "learning_rate": 3.308601264083539e-07, + "loss": 0.6855, + "mean_token_accuracy": 0.7889133095741272, + "num_tokens": 36308721.0, + "step": 1205 + }, + { + "epoch": 0.09944751381215469, + "grad_norm": 2.2062652111053467, + "learning_rate": 3.311349271777961e-07, + "loss": 0.7311, + "mean_token_accuracy": 0.7562646865844727, + "num_tokens": 36332403.0, + "step": 1206 + }, + { + "epoch": 0.09952997443720624, + "grad_norm": 2.5824146270751953, + "learning_rate": 3.3140972794723826e-07, + "loss": 0.9884, + "mean_token_accuracy": 0.7101116180419922, + "num_tokens": 36356942.0, + "step": 1207 + }, + { + "epoch": 0.09961243506225777, + "grad_norm": 2.285006046295166, + "learning_rate": 3.316845287166804e-07, + "loss": 0.8803, + "mean_token_accuracy": 0.7278713583946228, + "num_tokens": 36385498.0, + "step": 1208 + }, + { + "epoch": 0.09969489568730931, + "grad_norm": 1.965773105621338, + "learning_rate": 3.3195932948612255e-07, + "loss": 0.8334, + "mean_token_accuracy": 0.7572652101516724, + "num_tokens": 36417234.0, + "step": 1209 + }, + { + "epoch": 0.09977735631236084, + "grad_norm": 1.9953227043151855, + "learning_rate": 3.3223413025556467e-07, + "loss": 0.7743, + "mean_token_accuracy": 0.7547795176506042, + "num_tokens": 36450961.0, + "step": 1210 + }, + { + "epoch": 0.09985981693741239, + "grad_norm": 2.467064619064331, + "learning_rate": 3.325089310250069e-07, + "loss": 1.0247, + "mean_token_accuracy": 0.6915669441223145, + "num_tokens": 36474435.0, + "step": 1211 + }, + { + "epoch": 0.09994227756246392, + "grad_norm": 2.033409833908081, + "learning_rate": 3.32783731794449e-07, + "loss": 0.9703, + "mean_token_accuracy": 0.702457070350647, + "num_tokens": 36510212.0, + "step": 1212 + }, + { + "epoch": 0.10002473818751546, + "grad_norm": 2.265517234802246, + "learning_rate": 3.330585325638912e-07, + "loss": 0.8208, + "mean_token_accuracy": 0.7328994274139404, + "num_tokens": 36535052.0, + "step": 1213 + }, + { + "epoch": 0.100107198812567, + "grad_norm": 2.172217607498169, + "learning_rate": 3.333333333333333e-07, + "loss": 0.9087, + "mean_token_accuracy": 0.7246965765953064, + "num_tokens": 36562378.0, + "step": 1214 + }, + { + "epoch": 0.10018965943761854, + "grad_norm": 2.1276888847351074, + "learning_rate": 3.3360813410277547e-07, + "loss": 0.868, + "mean_token_accuracy": 0.7255774736404419, + "num_tokens": 36593141.0, + "step": 1215 + }, + { + "epoch": 0.10027212006267007, + "grad_norm": 1.9398293495178223, + "learning_rate": 3.3388293487221764e-07, + "loss": 0.7881, + "mean_token_accuracy": 0.7550675868988037, + "num_tokens": 36626355.0, + "step": 1216 + }, + { + "epoch": 0.10035458068772161, + "grad_norm": 1.9441181421279907, + "learning_rate": 3.3415773564165976e-07, + "loss": 0.7865, + "mean_token_accuracy": 0.7564496994018555, + "num_tokens": 36657614.0, + "step": 1217 + }, + { + "epoch": 0.10043704131277315, + "grad_norm": 2.1851391792297363, + "learning_rate": 3.3443253641110193e-07, + "loss": 0.9202, + "mean_token_accuracy": 0.7221061587333679, + "num_tokens": 36685984.0, + "step": 1218 + }, + { + "epoch": 0.10051950193782469, + "grad_norm": 2.202270984649658, + "learning_rate": 3.347073371805441e-07, + "loss": 0.8849, + "mean_token_accuracy": 0.7237269878387451, + "num_tokens": 36713331.0, + "step": 1219 + }, + { + "epoch": 0.10060196256287622, + "grad_norm": 2.140443801879883, + "learning_rate": 3.349821379499863e-07, + "loss": 0.9252, + "mean_token_accuracy": 0.7204279899597168, + "num_tokens": 36744666.0, + "step": 1220 + }, + { + "epoch": 0.10068442318792777, + "grad_norm": 1.9153772592544556, + "learning_rate": 3.352569387194284e-07, + "loss": 0.6796, + "mean_token_accuracy": 0.7743455767631531, + "num_tokens": 36775239.0, + "step": 1221 + }, + { + "epoch": 0.1007668838129793, + "grad_norm": 2.1153311729431152, + "learning_rate": 3.355317394888705e-07, + "loss": 0.9531, + "mean_token_accuracy": 0.7123091816902161, + "num_tokens": 36804871.0, + "step": 1222 + }, + { + "epoch": 0.10084934443803084, + "grad_norm": 2.07938551902771, + "learning_rate": 3.3580654025831274e-07, + "loss": 0.9438, + "mean_token_accuracy": 0.7081034183502197, + "num_tokens": 36835948.0, + "step": 1223 + }, + { + "epoch": 0.10093180506308237, + "grad_norm": 1.963382363319397, + "learning_rate": 3.3608134102775486e-07, + "loss": 0.6938, + "mean_token_accuracy": 0.7843372821807861, + "num_tokens": 36866538.0, + "step": 1224 + }, + { + "epoch": 0.10101426568813392, + "grad_norm": 2.6104304790496826, + "learning_rate": 3.3635614179719703e-07, + "loss": 1.0218, + "mean_token_accuracy": 0.6969665288925171, + "num_tokens": 36888782.0, + "step": 1225 + }, + { + "epoch": 0.10109672631318545, + "grad_norm": 2.242441415786743, + "learning_rate": 3.3663094256663915e-07, + "loss": 0.9582, + "mean_token_accuracy": 0.7173416018486023, + "num_tokens": 36915831.0, + "step": 1226 + }, + { + "epoch": 0.10117918693823699, + "grad_norm": 2.11285662651062, + "learning_rate": 3.3690574333608137e-07, + "loss": 0.8787, + "mean_token_accuracy": 0.7317386865615845, + "num_tokens": 36946917.0, + "step": 1227 + }, + { + "epoch": 0.10126164756328852, + "grad_norm": 2.1975717544555664, + "learning_rate": 3.371805441055235e-07, + "loss": 0.7801, + "mean_token_accuracy": 0.7552793622016907, + "num_tokens": 36972919.0, + "step": 1228 + }, + { + "epoch": 0.10134410818834007, + "grad_norm": 2.2663283348083496, + "learning_rate": 3.374553448749656e-07, + "loss": 0.8957, + "mean_token_accuracy": 0.7217451333999634, + "num_tokens": 37001680.0, + "step": 1229 + }, + { + "epoch": 0.1014265688133916, + "grad_norm": 2.0510618686676025, + "learning_rate": 3.377301456444078e-07, + "loss": 0.7868, + "mean_token_accuracy": 0.756345272064209, + "num_tokens": 37032896.0, + "step": 1230 + }, + { + "epoch": 0.10150902943844314, + "grad_norm": 2.2539052963256836, + "learning_rate": 3.3800494641384995e-07, + "loss": 0.9999, + "mean_token_accuracy": 0.699819803237915, + "num_tokens": 37062482.0, + "step": 1231 + }, + { + "epoch": 0.10159149006349467, + "grad_norm": 2.039045810699463, + "learning_rate": 3.382797471832921e-07, + "loss": 0.8757, + "mean_token_accuracy": 0.7316949963569641, + "num_tokens": 37096746.0, + "step": 1232 + }, + { + "epoch": 0.10167395068854622, + "grad_norm": 2.3536338806152344, + "learning_rate": 3.3855454795273424e-07, + "loss": 1.0061, + "mean_token_accuracy": 0.700942873954773, + "num_tokens": 37122677.0, + "step": 1233 + }, + { + "epoch": 0.10175641131359776, + "grad_norm": 2.0356099605560303, + "learning_rate": 3.388293487221764e-07, + "loss": 1.0141, + "mean_token_accuracy": 0.6947858929634094, + "num_tokens": 37155515.0, + "step": 1234 + }, + { + "epoch": 0.1018388719386493, + "grad_norm": 2.343555450439453, + "learning_rate": 3.391041494916186e-07, + "loss": 1.1045, + "mean_token_accuracy": 0.6796568632125854, + "num_tokens": 37184083.0, + "step": 1235 + }, + { + "epoch": 0.10192133256370084, + "grad_norm": 2.0468943119049072, + "learning_rate": 3.393789502610607e-07, + "loss": 1.034, + "mean_token_accuracy": 0.6988768577575684, + "num_tokens": 37215843.0, + "step": 1236 + }, + { + "epoch": 0.10200379318875237, + "grad_norm": 2.094365358352661, + "learning_rate": 3.396537510305029e-07, + "loss": 0.9732, + "mean_token_accuracy": 0.7011001110076904, + "num_tokens": 37246091.0, + "step": 1237 + }, + { + "epoch": 0.10208625381380391, + "grad_norm": 2.1325294971466064, + "learning_rate": 3.39928551799945e-07, + "loss": 0.9523, + "mean_token_accuracy": 0.7147523164749146, + "num_tokens": 37275413.0, + "step": 1238 + }, + { + "epoch": 0.10216871443885545, + "grad_norm": 2.233198881149292, + "learning_rate": 3.402033525693872e-07, + "loss": 0.9915, + "mean_token_accuracy": 0.6956987977027893, + "num_tokens": 37303258.0, + "step": 1239 + }, + { + "epoch": 0.10225117506390699, + "grad_norm": 2.2855238914489746, + "learning_rate": 3.4047815333882934e-07, + "loss": 0.7765, + "mean_token_accuracy": 0.7550464272499084, + "num_tokens": 37330132.0, + "step": 1240 + }, + { + "epoch": 0.10233363568895852, + "grad_norm": 2.134395122528076, + "learning_rate": 3.407529541082715e-07, + "loss": 0.9025, + "mean_token_accuracy": 0.7251788377761841, + "num_tokens": 37361741.0, + "step": 1241 + }, + { + "epoch": 0.10241609631401007, + "grad_norm": 1.9492844343185425, + "learning_rate": 3.4102775487771363e-07, + "loss": 0.833, + "mean_token_accuracy": 0.7408881187438965, + "num_tokens": 37395834.0, + "step": 1242 + }, + { + "epoch": 0.1024985569390616, + "grad_norm": 2.3700509071350098, + "learning_rate": 3.413025556471558e-07, + "loss": 0.9521, + "mean_token_accuracy": 0.7080053091049194, + "num_tokens": 37422122.0, + "step": 1243 + }, + { + "epoch": 0.10258101756411314, + "grad_norm": 2.0741400718688965, + "learning_rate": 3.4157735641659797e-07, + "loss": 0.9053, + "mean_token_accuracy": 0.7254718542098999, + "num_tokens": 37452536.0, + "step": 1244 + }, + { + "epoch": 0.10266347818916467, + "grad_norm": 1.9106053113937378, + "learning_rate": 3.418521571860401e-07, + "loss": 0.8454, + "mean_token_accuracy": 0.7426043152809143, + "num_tokens": 37486322.0, + "step": 1245 + }, + { + "epoch": 0.10274593881421622, + "grad_norm": 2.1095869541168213, + "learning_rate": 3.4212695795548226e-07, + "loss": 0.9223, + "mean_token_accuracy": 0.7288050651550293, + "num_tokens": 37515926.0, + "step": 1246 + }, + { + "epoch": 0.10282839943926775, + "grad_norm": 2.2890305519104004, + "learning_rate": 3.4240175872492443e-07, + "loss": 0.9383, + "mean_token_accuracy": 0.71668940782547, + "num_tokens": 37544842.0, + "step": 1247 + }, + { + "epoch": 0.10291086006431929, + "grad_norm": 2.130392074584961, + "learning_rate": 3.426765594943666e-07, + "loss": 0.7998, + "mean_token_accuracy": 0.7447239756584167, + "num_tokens": 37572802.0, + "step": 1248 + }, + { + "epoch": 0.10299332068937082, + "grad_norm": 2.2124521732330322, + "learning_rate": 3.429513602638087e-07, + "loss": 0.8736, + "mean_token_accuracy": 0.7298523187637329, + "num_tokens": 37601394.0, + "step": 1249 + }, + { + "epoch": 0.10307578131442237, + "grad_norm": 1.8711528778076172, + "learning_rate": 3.4322616103325084e-07, + "loss": 0.9981, + "mean_token_accuracy": 0.7027552723884583, + "num_tokens": 37639976.0, + "step": 1250 + }, + { + "epoch": 0.1031582419394739, + "grad_norm": 1.9512407779693604, + "learning_rate": 3.4350096180269307e-07, + "loss": 0.8468, + "mean_token_accuracy": 0.7468028664588928, + "num_tokens": 37673800.0, + "step": 1251 + }, + { + "epoch": 0.10324070256452544, + "grad_norm": 2.1829614639282227, + "learning_rate": 3.437757625721352e-07, + "loss": 0.8599, + "mean_token_accuracy": 0.7343120574951172, + "num_tokens": 37700880.0, + "step": 1252 + }, + { + "epoch": 0.10332316318957697, + "grad_norm": 2.200608015060425, + "learning_rate": 3.4405056334157736e-07, + "loss": 0.903, + "mean_token_accuracy": 0.7230491638183594, + "num_tokens": 37730482.0, + "step": 1253 + }, + { + "epoch": 0.10340562381462852, + "grad_norm": 1.8699729442596436, + "learning_rate": 3.443253641110195e-07, + "loss": 0.9258, + "mean_token_accuracy": 0.7248353362083435, + "num_tokens": 37770710.0, + "step": 1254 + }, + { + "epoch": 0.10348808443968005, + "grad_norm": 2.459036350250244, + "learning_rate": 3.446001648804617e-07, + "loss": 0.8941, + "mean_token_accuracy": 0.7287093997001648, + "num_tokens": 37794501.0, + "step": 1255 + }, + { + "epoch": 0.1035705450647316, + "grad_norm": 2.070962429046631, + "learning_rate": 3.448749656499038e-07, + "loss": 0.8757, + "mean_token_accuracy": 0.7473536729812622, + "num_tokens": 37827551.0, + "step": 1256 + }, + { + "epoch": 0.10365300568978313, + "grad_norm": 2.2544307708740234, + "learning_rate": 3.4514976641934594e-07, + "loss": 0.9546, + "mean_token_accuracy": 0.7136033773422241, + "num_tokens": 37857956.0, + "step": 1257 + }, + { + "epoch": 0.10373546631483467, + "grad_norm": 2.4855093955993652, + "learning_rate": 3.454245671887881e-07, + "loss": 1.0663, + "mean_token_accuracy": 0.6828539371490479, + "num_tokens": 37882671.0, + "step": 1258 + }, + { + "epoch": 0.1038179269398862, + "grad_norm": 2.375493049621582, + "learning_rate": 3.456993679582303e-07, + "loss": 0.8447, + "mean_token_accuracy": 0.7421249151229858, + "num_tokens": 37905516.0, + "step": 1259 + }, + { + "epoch": 0.10390038756493775, + "grad_norm": 2.0024871826171875, + "learning_rate": 3.4597416872767245e-07, + "loss": 0.8702, + "mean_token_accuracy": 0.7395105361938477, + "num_tokens": 37937608.0, + "step": 1260 + }, + { + "epoch": 0.10398284818998928, + "grad_norm": 2.1412787437438965, + "learning_rate": 3.4624896949711457e-07, + "loss": 0.9373, + "mean_token_accuracy": 0.719931960105896, + "num_tokens": 37968886.0, + "step": 1261 + }, + { + "epoch": 0.10406530881504082, + "grad_norm": 1.9441320896148682, + "learning_rate": 3.465237702665567e-07, + "loss": 0.8736, + "mean_token_accuracy": 0.7359894514083862, + "num_tokens": 38006094.0, + "step": 1262 + }, + { + "epoch": 0.10414776944009235, + "grad_norm": 2.088123083114624, + "learning_rate": 3.467985710359989e-07, + "loss": 0.7535, + "mean_token_accuracy": 0.7673155069351196, + "num_tokens": 38037946.0, + "step": 1263 + }, + { + "epoch": 0.1042302300651439, + "grad_norm": 1.7332836389541626, + "learning_rate": 3.4707337180544103e-07, + "loss": 0.9016, + "mean_token_accuracy": 0.7266995906829834, + "num_tokens": 38077137.0, + "step": 1264 + }, + { + "epoch": 0.10431269069019543, + "grad_norm": 1.9885048866271973, + "learning_rate": 3.473481725748832e-07, + "loss": 0.9522, + "mean_token_accuracy": 0.7089892029762268, + "num_tokens": 38110833.0, + "step": 1265 + }, + { + "epoch": 0.10439515131524697, + "grad_norm": 1.9753506183624268, + "learning_rate": 3.476229733443253e-07, + "loss": 0.8661, + "mean_token_accuracy": 0.7397968173027039, + "num_tokens": 38141835.0, + "step": 1266 + }, + { + "epoch": 0.1044776119402985, + "grad_norm": 2.1703758239746094, + "learning_rate": 3.4789777411376755e-07, + "loss": 0.7989, + "mean_token_accuracy": 0.7544353604316711, + "num_tokens": 38169792.0, + "step": 1267 + }, + { + "epoch": 0.10456007256535005, + "grad_norm": 1.9580775499343872, + "learning_rate": 3.4817257488320967e-07, + "loss": 0.8824, + "mean_token_accuracy": 0.7351440191268921, + "num_tokens": 38202152.0, + "step": 1268 + }, + { + "epoch": 0.10464253319040158, + "grad_norm": 2.285370111465454, + "learning_rate": 3.484473756526518e-07, + "loss": 0.9004, + "mean_token_accuracy": 0.7240142822265625, + "num_tokens": 38229538.0, + "step": 1269 + }, + { + "epoch": 0.10472499381545312, + "grad_norm": 2.64677357673645, + "learning_rate": 3.4872217642209396e-07, + "loss": 0.8734, + "mean_token_accuracy": 0.7167177200317383, + "num_tokens": 38250478.0, + "step": 1270 + }, + { + "epoch": 0.10480745444050465, + "grad_norm": 2.165632724761963, + "learning_rate": 3.4899697719153613e-07, + "loss": 0.8412, + "mean_token_accuracy": 0.7303205728530884, + "num_tokens": 38276710.0, + "step": 1271 + }, + { + "epoch": 0.1048899150655562, + "grad_norm": 2.1297545433044434, + "learning_rate": 3.492717779609783e-07, + "loss": 0.8604, + "mean_token_accuracy": 0.742713212966919, + "num_tokens": 38307851.0, + "step": 1272 + }, + { + "epoch": 0.10497237569060773, + "grad_norm": 2.580292224884033, + "learning_rate": 3.495465787304204e-07, + "loss": 0.9102, + "mean_token_accuracy": 0.7180161476135254, + "num_tokens": 38331337.0, + "step": 1273 + }, + { + "epoch": 0.10505483631565928, + "grad_norm": 2.1869678497314453, + "learning_rate": 3.498213794998626e-07, + "loss": 0.9133, + "mean_token_accuracy": 0.7221822142601013, + "num_tokens": 38361931.0, + "step": 1274 + }, + { + "epoch": 0.1051372969407108, + "grad_norm": 2.202664375305176, + "learning_rate": 3.5009618026930476e-07, + "loss": 0.9175, + "mean_token_accuracy": 0.7277350425720215, + "num_tokens": 38389616.0, + "step": 1275 + }, + { + "epoch": 0.10521975756576235, + "grad_norm": 2.243340253829956, + "learning_rate": 3.503709810387469e-07, + "loss": 0.8697, + "mean_token_accuracy": 0.7374023199081421, + "num_tokens": 38415861.0, + "step": 1276 + }, + { + "epoch": 0.10530221819081388, + "grad_norm": 2.152585744857788, + "learning_rate": 3.5064578180818905e-07, + "loss": 0.8791, + "mean_token_accuracy": 0.7235972285270691, + "num_tokens": 38447416.0, + "step": 1277 + }, + { + "epoch": 0.10538467881586543, + "grad_norm": 1.8083364963531494, + "learning_rate": 3.5092058257763117e-07, + "loss": 0.6381, + "mean_token_accuracy": 0.7866542339324951, + "num_tokens": 38482083.0, + "step": 1278 + }, + { + "epoch": 0.10546713944091696, + "grad_norm": 1.9331763982772827, + "learning_rate": 3.511953833470734e-07, + "loss": 0.9702, + "mean_token_accuracy": 0.6993638277053833, + "num_tokens": 38517788.0, + "step": 1279 + }, + { + "epoch": 0.1055496000659685, + "grad_norm": 2.277677536010742, + "learning_rate": 3.514701841165155e-07, + "loss": 1.0917, + "mean_token_accuracy": 0.6725723743438721, + "num_tokens": 38546288.0, + "step": 1280 + }, + { + "epoch": 0.10563206069102003, + "grad_norm": 1.8805800676345825, + "learning_rate": 3.517449848859577e-07, + "loss": 0.883, + "mean_token_accuracy": 0.7367761731147766, + "num_tokens": 38582106.0, + "step": 1281 + }, + { + "epoch": 0.10571452131607158, + "grad_norm": 2.8177919387817383, + "learning_rate": 3.520197856553998e-07, + "loss": 1.0277, + "mean_token_accuracy": 0.695296585559845, + "num_tokens": 38604008.0, + "step": 1282 + }, + { + "epoch": 0.10579698194112311, + "grad_norm": 2.0595152378082275, + "learning_rate": 3.52294586424842e-07, + "loss": 0.8881, + "mean_token_accuracy": 0.7384445667266846, + "num_tokens": 38635095.0, + "step": 1283 + }, + { + "epoch": 0.10587944256617465, + "grad_norm": 1.89009690284729, + "learning_rate": 3.5256938719428415e-07, + "loss": 0.7181, + "mean_token_accuracy": 0.7742410898208618, + "num_tokens": 38669386.0, + "step": 1284 + }, + { + "epoch": 0.10596190319122618, + "grad_norm": 2.210676670074463, + "learning_rate": 3.5284418796372626e-07, + "loss": 1.0258, + "mean_token_accuracy": 0.6913540363311768, + "num_tokens": 38702084.0, + "step": 1285 + }, + { + "epoch": 0.10604436381627773, + "grad_norm": 2.003641366958618, + "learning_rate": 3.5311898873316844e-07, + "loss": 1.0738, + "mean_token_accuracy": 0.6818665266036987, + "num_tokens": 38740702.0, + "step": 1286 + }, + { + "epoch": 0.10612682444132926, + "grad_norm": 2.221912145614624, + "learning_rate": 3.533937895026106e-07, + "loss": 0.9611, + "mean_token_accuracy": 0.7108796834945679, + "num_tokens": 38767278.0, + "step": 1287 + }, + { + "epoch": 0.1062092850663808, + "grad_norm": 2.20050048828125, + "learning_rate": 3.536685902720528e-07, + "loss": 0.9527, + "mean_token_accuracy": 0.7237738966941833, + "num_tokens": 38793941.0, + "step": 1288 + }, + { + "epoch": 0.10629174569143233, + "grad_norm": 1.8261640071868896, + "learning_rate": 3.539433910414949e-07, + "loss": 1.0344, + "mean_token_accuracy": 0.6946701407432556, + "num_tokens": 38833780.0, + "step": 1289 + }, + { + "epoch": 0.10637420631648388, + "grad_norm": 1.9816316366195679, + "learning_rate": 3.54218191810937e-07, + "loss": 0.8435, + "mean_token_accuracy": 0.7373237609863281, + "num_tokens": 38866804.0, + "step": 1290 + }, + { + "epoch": 0.10645666694153541, + "grad_norm": 2.084390640258789, + "learning_rate": 3.5449299258037924e-07, + "loss": 1.0286, + "mean_token_accuracy": 0.6951139569282532, + "num_tokens": 38897795.0, + "step": 1291 + }, + { + "epoch": 0.10653912756658696, + "grad_norm": 2.2916417121887207, + "learning_rate": 3.5476779334982136e-07, + "loss": 1.0782, + "mean_token_accuracy": 0.6809105277061462, + "num_tokens": 38924992.0, + "step": 1292 + }, + { + "epoch": 0.10662158819163849, + "grad_norm": 1.9625263214111328, + "learning_rate": 3.5504259411926353e-07, + "loss": 0.9241, + "mean_token_accuracy": 0.7191088795661926, + "num_tokens": 38957065.0, + "step": 1293 + }, + { + "epoch": 0.10670404881669003, + "grad_norm": 2.2005867958068848, + "learning_rate": 3.5531739488870565e-07, + "loss": 0.9254, + "mean_token_accuracy": 0.7242832183837891, + "num_tokens": 38986711.0, + "step": 1294 + }, + { + "epoch": 0.10678650944174156, + "grad_norm": 1.8019859790802002, + "learning_rate": 3.555921956581479e-07, + "loss": 0.8399, + "mean_token_accuracy": 0.7501908540725708, + "num_tokens": 39023184.0, + "step": 1295 + }, + { + "epoch": 0.1068689700667931, + "grad_norm": 1.7930514812469482, + "learning_rate": 3.5586699642759e-07, + "loss": 0.8247, + "mean_token_accuracy": 0.7472535371780396, + "num_tokens": 39059621.0, + "step": 1296 + }, + { + "epoch": 0.10695143069184464, + "grad_norm": 1.9982151985168457, + "learning_rate": 3.561417971970321e-07, + "loss": 0.797, + "mean_token_accuracy": 0.7549675107002258, + "num_tokens": 39092564.0, + "step": 1297 + }, + { + "epoch": 0.10703389131689618, + "grad_norm": 1.9701619148254395, + "learning_rate": 3.564165979664743e-07, + "loss": 0.922, + "mean_token_accuracy": 0.7305883169174194, + "num_tokens": 39124425.0, + "step": 1298 + }, + { + "epoch": 0.10711635194194773, + "grad_norm": 2.1006014347076416, + "learning_rate": 3.5669139873591645e-07, + "loss": 0.9638, + "mean_token_accuracy": 0.7122470140457153, + "num_tokens": 39156607.0, + "step": 1299 + }, + { + "epoch": 0.10719881256699926, + "grad_norm": 2.213473320007324, + "learning_rate": 3.569661995053586e-07, + "loss": 0.8934, + "mean_token_accuracy": 0.7307826280593872, + "num_tokens": 39184421.0, + "step": 1300 + }, + { + "epoch": 0.1072812731920508, + "grad_norm": 1.6038178205490112, + "learning_rate": 3.5724100027480074e-07, + "loss": 0.6998, + "mean_token_accuracy": 0.7849804162979126, + "num_tokens": 39227294.0, + "step": 1301 + }, + { + "epoch": 0.10736373381710233, + "grad_norm": 2.0479536056518555, + "learning_rate": 3.575158010442429e-07, + "loss": 0.7728, + "mean_token_accuracy": 0.7563564777374268, + "num_tokens": 39260810.0, + "step": 1302 + }, + { + "epoch": 0.10744619444215388, + "grad_norm": 2.75776743888855, + "learning_rate": 3.577906018136851e-07, + "loss": 0.9649, + "mean_token_accuracy": 0.704898476600647, + "num_tokens": 39282008.0, + "step": 1303 + }, + { + "epoch": 0.10752865506720541, + "grad_norm": 2.330765724182129, + "learning_rate": 3.580654025831272e-07, + "loss": 0.8667, + "mean_token_accuracy": 0.7421588897705078, + "num_tokens": 39308135.0, + "step": 1304 + }, + { + "epoch": 0.10761111569225695, + "grad_norm": 2.2216391563415527, + "learning_rate": 3.583402033525694e-07, + "loss": 0.9584, + "mean_token_accuracy": 0.710018515586853, + "num_tokens": 39335852.0, + "step": 1305 + }, + { + "epoch": 0.10769357631730848, + "grad_norm": 2.1852946281433105, + "learning_rate": 3.586150041220115e-07, + "loss": 0.8207, + "mean_token_accuracy": 0.7415304780006409, + "num_tokens": 39363057.0, + "step": 1306 + }, + { + "epoch": 0.10777603694236003, + "grad_norm": 2.0409176349639893, + "learning_rate": 3.588898048914537e-07, + "loss": 1.0119, + "mean_token_accuracy": 0.6953348517417908, + "num_tokens": 39398987.0, + "step": 1307 + }, + { + "epoch": 0.10785849756741156, + "grad_norm": 2.4113223552703857, + "learning_rate": 3.5916460566089584e-07, + "loss": 1.1465, + "mean_token_accuracy": 0.6659183502197266, + "num_tokens": 39426377.0, + "step": 1308 + }, + { + "epoch": 0.1079409581924631, + "grad_norm": 2.2134311199188232, + "learning_rate": 3.59439406430338e-07, + "loss": 0.8731, + "mean_token_accuracy": 0.7389196157455444, + "num_tokens": 39458055.0, + "step": 1309 + }, + { + "epoch": 0.10802341881751464, + "grad_norm": 2.0137691497802734, + "learning_rate": 3.5971420719978013e-07, + "loss": 0.8903, + "mean_token_accuracy": 0.7314599752426147, + "num_tokens": 39490473.0, + "step": 1310 + }, + { + "epoch": 0.10810587944256618, + "grad_norm": 2.1646597385406494, + "learning_rate": 3.599890079692223e-07, + "loss": 0.9069, + "mean_token_accuracy": 0.7270756363868713, + "num_tokens": 39520193.0, + "step": 1311 + }, + { + "epoch": 0.10818834006761771, + "grad_norm": 2.089526414871216, + "learning_rate": 3.6026380873866447e-07, + "loss": 0.906, + "mean_token_accuracy": 0.7232775688171387, + "num_tokens": 39553210.0, + "step": 1312 + }, + { + "epoch": 0.10827080069266926, + "grad_norm": 1.96950364112854, + "learning_rate": 3.605386095081066e-07, + "loss": 0.7952, + "mean_token_accuracy": 0.7497720718383789, + "num_tokens": 39584216.0, + "step": 1313 + }, + { + "epoch": 0.10835326131772079, + "grad_norm": 2.299290418624878, + "learning_rate": 3.6081341027754876e-07, + "loss": 0.8196, + "mean_token_accuracy": 0.7436956763267517, + "num_tokens": 39609165.0, + "step": 1314 + }, + { + "epoch": 0.10843572194277233, + "grad_norm": 1.9884145259857178, + "learning_rate": 3.6108821104699093e-07, + "loss": 1.0204, + "mean_token_accuracy": 0.6937052011489868, + "num_tokens": 39645536.0, + "step": 1315 + }, + { + "epoch": 0.10851818256782386, + "grad_norm": 2.0695252418518066, + "learning_rate": 3.613630118164331e-07, + "loss": 0.9961, + "mean_token_accuracy": 0.699321985244751, + "num_tokens": 39679686.0, + "step": 1316 + }, + { + "epoch": 0.1086006431928754, + "grad_norm": 2.2135725021362305, + "learning_rate": 3.616378125858752e-07, + "loss": 0.7109, + "mean_token_accuracy": 0.7693525552749634, + "num_tokens": 39709129.0, + "step": 1317 + }, + { + "epoch": 0.10868310381792694, + "grad_norm": 2.0861470699310303, + "learning_rate": 3.6191261335531734e-07, + "loss": 0.954, + "mean_token_accuracy": 0.7116961479187012, + "num_tokens": 39737487.0, + "step": 1318 + }, + { + "epoch": 0.10876556444297848, + "grad_norm": 1.9521607160568237, + "learning_rate": 3.6218741412475957e-07, + "loss": 0.9682, + "mean_token_accuracy": 0.7162103056907654, + "num_tokens": 39771302.0, + "step": 1319 + }, + { + "epoch": 0.10884802506803001, + "grad_norm": 2.2930715084075928, + "learning_rate": 3.624622148942017e-07, + "loss": 1.0912, + "mean_token_accuracy": 0.6994989514350891, + "num_tokens": 39800226.0, + "step": 1320 + }, + { + "epoch": 0.10893048569308156, + "grad_norm": 2.043996810913086, + "learning_rate": 3.6273701566364386e-07, + "loss": 0.9458, + "mean_token_accuracy": 0.723782479763031, + "num_tokens": 39833283.0, + "step": 1321 + }, + { + "epoch": 0.10901294631813309, + "grad_norm": 2.4929118156433105, + "learning_rate": 3.63011816433086e-07, + "loss": 0.8944, + "mean_token_accuracy": 0.7233796119689941, + "num_tokens": 39856674.0, + "step": 1322 + }, + { + "epoch": 0.10909540694318463, + "grad_norm": 2.373054265975952, + "learning_rate": 3.632866172025282e-07, + "loss": 0.7817, + "mean_token_accuracy": 0.7575793266296387, + "num_tokens": 39880311.0, + "step": 1323 + }, + { + "epoch": 0.10917786756823616, + "grad_norm": 2.2840352058410645, + "learning_rate": 3.635614179719703e-07, + "loss": 0.9119, + "mean_token_accuracy": 0.7255232334136963, + "num_tokens": 39906811.0, + "step": 1324 + }, + { + "epoch": 0.10926032819328771, + "grad_norm": 2.2011783123016357, + "learning_rate": 3.6383621874141244e-07, + "loss": 0.8817, + "mean_token_accuracy": 0.733998715877533, + "num_tokens": 39936160.0, + "step": 1325 + }, + { + "epoch": 0.10934278881833924, + "grad_norm": 1.8076951503753662, + "learning_rate": 3.641110195108546e-07, + "loss": 0.8729, + "mean_token_accuracy": 0.7381136417388916, + "num_tokens": 39974933.0, + "step": 1326 + }, + { + "epoch": 0.10942524944339078, + "grad_norm": 2.3699393272399902, + "learning_rate": 3.6438582028029673e-07, + "loss": 0.8981, + "mean_token_accuracy": 0.7326769232749939, + "num_tokens": 40000652.0, + "step": 1327 + }, + { + "epoch": 0.10950771006844232, + "grad_norm": 2.2337985038757324, + "learning_rate": 3.6466062104973895e-07, + "loss": 0.7746, + "mean_token_accuracy": 0.7463428974151611, + "num_tokens": 40026537.0, + "step": 1328 + }, + { + "epoch": 0.10959017069349386, + "grad_norm": 1.9507352113723755, + "learning_rate": 3.6493542181918107e-07, + "loss": 0.964, + "mean_token_accuracy": 0.7149921655654907, + "num_tokens": 40064534.0, + "step": 1329 + }, + { + "epoch": 0.10967263131854539, + "grad_norm": 2.1653354167938232, + "learning_rate": 3.6521022258862324e-07, + "loss": 0.8038, + "mean_token_accuracy": 0.7538349628448486, + "num_tokens": 40092929.0, + "step": 1330 + }, + { + "epoch": 0.10975509194359694, + "grad_norm": 1.984866976737976, + "learning_rate": 3.6548502335806536e-07, + "loss": 0.7688, + "mean_token_accuracy": 0.7617655992507935, + "num_tokens": 40126133.0, + "step": 1331 + }, + { + "epoch": 0.10983755256864847, + "grad_norm": 2.002340793609619, + "learning_rate": 3.6575982412750753e-07, + "loss": 0.8301, + "mean_token_accuracy": 0.7416605949401855, + "num_tokens": 40159815.0, + "step": 1332 + }, + { + "epoch": 0.10992001319370001, + "grad_norm": 2.149228096008301, + "learning_rate": 3.660346248969497e-07, + "loss": 0.8572, + "mean_token_accuracy": 0.7367404699325562, + "num_tokens": 40188399.0, + "step": 1333 + }, + { + "epoch": 0.11000247381875154, + "grad_norm": 2.3122873306274414, + "learning_rate": 3.663094256663918e-07, + "loss": 0.848, + "mean_token_accuracy": 0.7448471188545227, + "num_tokens": 40211897.0, + "step": 1334 + }, + { + "epoch": 0.11008493444380309, + "grad_norm": 2.2368483543395996, + "learning_rate": 3.66584226435834e-07, + "loss": 0.8807, + "mean_token_accuracy": 0.7346627116203308, + "num_tokens": 40238305.0, + "step": 1335 + }, + { + "epoch": 0.11016739506885462, + "grad_norm": 2.5250449180603027, + "learning_rate": 3.6685902720527617e-07, + "loss": 0.9068, + "mean_token_accuracy": 0.7210112810134888, + "num_tokens": 40261929.0, + "step": 1336 + }, + { + "epoch": 0.11024985569390616, + "grad_norm": 1.9734240770339966, + "learning_rate": 3.6713382797471834e-07, + "loss": 0.9724, + "mean_token_accuracy": 0.7067251205444336, + "num_tokens": 40300800.0, + "step": 1337 + }, + { + "epoch": 0.1103323163189577, + "grad_norm": 2.424534797668457, + "learning_rate": 3.6740862874416046e-07, + "loss": 0.8805, + "mean_token_accuracy": 0.7313261032104492, + "num_tokens": 40325528.0, + "step": 1338 + }, + { + "epoch": 0.11041477694400924, + "grad_norm": 2.061368465423584, + "learning_rate": 3.676834295136026e-07, + "loss": 0.9549, + "mean_token_accuracy": 0.7119652032852173, + "num_tokens": 40354145.0, + "step": 1339 + }, + { + "epoch": 0.11049723756906077, + "grad_norm": 2.029785633087158, + "learning_rate": 3.679582302830448e-07, + "loss": 0.9147, + "mean_token_accuracy": 0.7249409556388855, + "num_tokens": 40386230.0, + "step": 1340 + }, + { + "epoch": 0.11057969819411231, + "grad_norm": 1.6655614376068115, + "learning_rate": 3.682330310524869e-07, + "loss": 0.7034, + "mean_token_accuracy": 0.777039647102356, + "num_tokens": 40429183.0, + "step": 1341 + }, + { + "epoch": 0.11066215881916384, + "grad_norm": 1.94557785987854, + "learning_rate": 3.685078318219291e-07, + "loss": 0.8959, + "mean_token_accuracy": 0.7233223915100098, + "num_tokens": 40464845.0, + "step": 1342 + }, + { + "epoch": 0.11074461944421539, + "grad_norm": 2.006796360015869, + "learning_rate": 3.687826325913712e-07, + "loss": 0.9045, + "mean_token_accuracy": 0.7215403914451599, + "num_tokens": 40498606.0, + "step": 1343 + }, + { + "epoch": 0.11082708006926692, + "grad_norm": 1.94917631149292, + "learning_rate": 3.6905743336081343e-07, + "loss": 0.9965, + "mean_token_accuracy": 0.7007352113723755, + "num_tokens": 40532911.0, + "step": 1344 + }, + { + "epoch": 0.11090954069431846, + "grad_norm": 1.9071564674377441, + "learning_rate": 3.6933223413025555e-07, + "loss": 0.8234, + "mean_token_accuracy": 0.7455002069473267, + "num_tokens": 40564138.0, + "step": 1345 + }, + { + "epoch": 0.11099200131937, + "grad_norm": 2.005155086517334, + "learning_rate": 3.6960703489969767e-07, + "loss": 0.9631, + "mean_token_accuracy": 0.7085983753204346, + "num_tokens": 40597749.0, + "step": 1346 + }, + { + "epoch": 0.11107446194442154, + "grad_norm": 2.0416524410247803, + "learning_rate": 3.6988183566913984e-07, + "loss": 0.6465, + "mean_token_accuracy": 0.7908802628517151, + "num_tokens": 40627915.0, + "step": 1347 + }, + { + "epoch": 0.11115692256947307, + "grad_norm": 2.2551076412200928, + "learning_rate": 3.70156636438582e-07, + "loss": 0.8077, + "mean_token_accuracy": 0.755630373954773, + "num_tokens": 40653336.0, + "step": 1348 + }, + { + "epoch": 0.11123938319452462, + "grad_norm": 1.9394699335098267, + "learning_rate": 3.704314372080242e-07, + "loss": 0.7902, + "mean_token_accuracy": 0.7548516392707825, + "num_tokens": 40689164.0, + "step": 1349 + }, + { + "epoch": 0.11132184381957615, + "grad_norm": 2.058846950531006, + "learning_rate": 3.707062379774663e-07, + "loss": 0.8394, + "mean_token_accuracy": 0.7393883466720581, + "num_tokens": 40720063.0, + "step": 1350 + }, + { + "epoch": 0.11140430444462769, + "grad_norm": 2.2071645259857178, + "learning_rate": 3.709810387469084e-07, + "loss": 0.9381, + "mean_token_accuracy": 0.7136522531509399, + "num_tokens": 40747650.0, + "step": 1351 + }, + { + "epoch": 0.11148676506967922, + "grad_norm": 2.4305083751678467, + "learning_rate": 3.7125583951635065e-07, + "loss": 1.1285, + "mean_token_accuracy": 0.6702581644058228, + "num_tokens": 40773652.0, + "step": 1352 + }, + { + "epoch": 0.11156922569473077, + "grad_norm": 2.226827621459961, + "learning_rate": 3.7153064028579277e-07, + "loss": 0.9349, + "mean_token_accuracy": 0.7166039943695068, + "num_tokens": 40798806.0, + "step": 1353 + }, + { + "epoch": 0.1116516863197823, + "grad_norm": 2.051927328109741, + "learning_rate": 3.7180544105523494e-07, + "loss": 0.7792, + "mean_token_accuracy": 0.7669134736061096, + "num_tokens": 40827101.0, + "step": 1354 + }, + { + "epoch": 0.11173414694483384, + "grad_norm": 2.0559403896331787, + "learning_rate": 3.7208024182467706e-07, + "loss": 0.7722, + "mean_token_accuracy": 0.7623270750045776, + "num_tokens": 40857420.0, + "step": 1355 + }, + { + "epoch": 0.11181660756988537, + "grad_norm": 2.115185022354126, + "learning_rate": 3.723550425941193e-07, + "loss": 0.9909, + "mean_token_accuracy": 0.7091200351715088, + "num_tokens": 40890135.0, + "step": 1356 + }, + { + "epoch": 0.11189906819493692, + "grad_norm": 1.8192660808563232, + "learning_rate": 3.726298433635614e-07, + "loss": 0.7595, + "mean_token_accuracy": 0.7614802122116089, + "num_tokens": 40927478.0, + "step": 1357 + }, + { + "epoch": 0.11198152881998845, + "grad_norm": 2.0226521492004395, + "learning_rate": 3.729046441330035e-07, + "loss": 1.0064, + "mean_token_accuracy": 0.6985230445861816, + "num_tokens": 40960880.0, + "step": 1358 + }, + { + "epoch": 0.11206398944504, + "grad_norm": 2.0766894817352295, + "learning_rate": 3.731794449024457e-07, + "loss": 0.7473, + "mean_token_accuracy": 0.7676405906677246, + "num_tokens": 40989916.0, + "step": 1359 + }, + { + "epoch": 0.11214645007009152, + "grad_norm": 2.6305274963378906, + "learning_rate": 3.7345424567188786e-07, + "loss": 0.9733, + "mean_token_accuracy": 0.7087240219116211, + "num_tokens": 41011973.0, + "step": 1360 + }, + { + "epoch": 0.11222891069514307, + "grad_norm": 2.164045810699463, + "learning_rate": 3.7372904644133003e-07, + "loss": 0.8497, + "mean_token_accuracy": 0.7448890209197998, + "num_tokens": 41041431.0, + "step": 1361 + }, + { + "epoch": 0.11231137132019461, + "grad_norm": 2.09088134765625, + "learning_rate": 3.7400384721077215e-07, + "loss": 0.9617, + "mean_token_accuracy": 0.7182782292366028, + "num_tokens": 41072094.0, + "step": 1362 + }, + { + "epoch": 0.11239383194524614, + "grad_norm": 2.0136728286743164, + "learning_rate": 3.742786479802143e-07, + "loss": 1.0405, + "mean_token_accuracy": 0.6866908669471741, + "num_tokens": 41110241.0, + "step": 1363 + }, + { + "epoch": 0.11247629257029769, + "grad_norm": 2.0315799713134766, + "learning_rate": 3.745534487496565e-07, + "loss": 0.8683, + "mean_token_accuracy": 0.7344831228256226, + "num_tokens": 41141195.0, + "step": 1364 + }, + { + "epoch": 0.11255875319534922, + "grad_norm": 2.0025558471679688, + "learning_rate": 3.748282495190986e-07, + "loss": 0.8714, + "mean_token_accuracy": 0.7539353966712952, + "num_tokens": 41172188.0, + "step": 1365 + }, + { + "epoch": 0.11264121382040077, + "grad_norm": 2.180274248123169, + "learning_rate": 3.751030502885408e-07, + "loss": 0.9157, + "mean_token_accuracy": 0.7209250926971436, + "num_tokens": 41202585.0, + "step": 1366 + }, + { + "epoch": 0.1127236744454523, + "grad_norm": 2.4467270374298096, + "learning_rate": 3.753778510579829e-07, + "loss": 0.9731, + "mean_token_accuracy": 0.7005912661552429, + "num_tokens": 41225097.0, + "step": 1367 + }, + { + "epoch": 0.11280613507050384, + "grad_norm": 1.8652859926223755, + "learning_rate": 3.7565265182742513e-07, + "loss": 0.9479, + "mean_token_accuracy": 0.7277081608772278, + "num_tokens": 41265440.0, + "step": 1368 + }, + { + "epoch": 0.11288859569555537, + "grad_norm": 1.84540855884552, + "learning_rate": 3.7592745259686725e-07, + "loss": 0.9486, + "mean_token_accuracy": 0.7245583534240723, + "num_tokens": 41301786.0, + "step": 1369 + }, + { + "epoch": 0.11297105632060692, + "grad_norm": 2.0248730182647705, + "learning_rate": 3.762022533663094e-07, + "loss": 0.9442, + "mean_token_accuracy": 0.7162798643112183, + "num_tokens": 41335759.0, + "step": 1370 + }, + { + "epoch": 0.11305351694565845, + "grad_norm": 1.8614619970321655, + "learning_rate": 3.7647705413575154e-07, + "loss": 0.7706, + "mean_token_accuracy": 0.7561070322990417, + "num_tokens": 41371247.0, + "step": 1371 + }, + { + "epoch": 0.11313597757070999, + "grad_norm": 2.131453037261963, + "learning_rate": 3.767518549051937e-07, + "loss": 0.9809, + "mean_token_accuracy": 0.709900975227356, + "num_tokens": 41402641.0, + "step": 1372 + }, + { + "epoch": 0.11321843819576152, + "grad_norm": 1.856755018234253, + "learning_rate": 3.770266556746359e-07, + "loss": 0.7789, + "mean_token_accuracy": 0.7570480108261108, + "num_tokens": 41436690.0, + "step": 1373 + }, + { + "epoch": 0.11330089882081307, + "grad_norm": 1.9101159572601318, + "learning_rate": 3.77301456444078e-07, + "loss": 0.9409, + "mean_token_accuracy": 0.7165617942810059, + "num_tokens": 41472041.0, + "step": 1374 + }, + { + "epoch": 0.1133833594458646, + "grad_norm": 2.1903951168060303, + "learning_rate": 3.7757625721352017e-07, + "loss": 0.9864, + "mean_token_accuracy": 0.7059256434440613, + "num_tokens": 41501234.0, + "step": 1375 + }, + { + "epoch": 0.11346582007091614, + "grad_norm": 2.3510472774505615, + "learning_rate": 3.7785105798296234e-07, + "loss": 1.0867, + "mean_token_accuracy": 0.68006831407547, + "num_tokens": 41528073.0, + "step": 1376 + }, + { + "epoch": 0.11354828069596767, + "grad_norm": 2.231116533279419, + "learning_rate": 3.781258587524045e-07, + "loss": 0.9083, + "mean_token_accuracy": 0.7351080179214478, + "num_tokens": 41555925.0, + "step": 1377 + }, + { + "epoch": 0.11363074132101922, + "grad_norm": 2.0626022815704346, + "learning_rate": 3.7840065952184663e-07, + "loss": 0.9785, + "mean_token_accuracy": 0.7147505283355713, + "num_tokens": 41587007.0, + "step": 1378 + }, + { + "epoch": 0.11371320194607075, + "grad_norm": 2.0300397872924805, + "learning_rate": 3.7867546029128875e-07, + "loss": 0.8638, + "mean_token_accuracy": 0.7326364517211914, + "num_tokens": 41617876.0, + "step": 1379 + }, + { + "epoch": 0.1137956625711223, + "grad_norm": 1.9395917654037476, + "learning_rate": 3.78950261060731e-07, + "loss": 0.8515, + "mean_token_accuracy": 0.7404981255531311, + "num_tokens": 41653310.0, + "step": 1380 + }, + { + "epoch": 0.11387812319617382, + "grad_norm": 2.200105905532837, + "learning_rate": 3.792250618301731e-07, + "loss": 0.766, + "mean_token_accuracy": 0.766829252243042, + "num_tokens": 41682309.0, + "step": 1381 + }, + { + "epoch": 0.11396058382122537, + "grad_norm": 2.2893311977386475, + "learning_rate": 3.7949986259961526e-07, + "loss": 0.8775, + "mean_token_accuracy": 0.7355047464370728, + "num_tokens": 41707898.0, + "step": 1382 + }, + { + "epoch": 0.1140430444462769, + "grad_norm": 2.139643430709839, + "learning_rate": 3.797746633690574e-07, + "loss": 0.9852, + "mean_token_accuracy": 0.7078078985214233, + "num_tokens": 41738015.0, + "step": 1383 + }, + { + "epoch": 0.11412550507132845, + "grad_norm": 2.1497247219085693, + "learning_rate": 3.800494641384996e-07, + "loss": 0.9064, + "mean_token_accuracy": 0.7139729857444763, + "num_tokens": 41767840.0, + "step": 1384 + }, + { + "epoch": 0.11420796569637998, + "grad_norm": 2.081936836242676, + "learning_rate": 3.803242649079417e-07, + "loss": 0.9122, + "mean_token_accuracy": 0.727202296257019, + "num_tokens": 41798426.0, + "step": 1385 + }, + { + "epoch": 0.11429042632143152, + "grad_norm": 1.9698489904403687, + "learning_rate": 3.8059906567738384e-07, + "loss": 0.8938, + "mean_token_accuracy": 0.7300816178321838, + "num_tokens": 41832474.0, + "step": 1386 + }, + { + "epoch": 0.11437288694648305, + "grad_norm": 2.1718039512634277, + "learning_rate": 3.80873866446826e-07, + "loss": 0.8207, + "mean_token_accuracy": 0.7573536038398743, + "num_tokens": 41860684.0, + "step": 1387 + }, + { + "epoch": 0.1144553475715346, + "grad_norm": 1.96500825881958, + "learning_rate": 3.811486672162682e-07, + "loss": 0.7565, + "mean_token_accuracy": 0.755339503288269, + "num_tokens": 41893773.0, + "step": 1388 + }, + { + "epoch": 0.11453780819658613, + "grad_norm": 1.9925795793533325, + "learning_rate": 3.8142346798571036e-07, + "loss": 0.871, + "mean_token_accuracy": 0.7306469678878784, + "num_tokens": 41923436.0, + "step": 1389 + }, + { + "epoch": 0.11462026882163767, + "grad_norm": 2.0822129249572754, + "learning_rate": 3.816982687551525e-07, + "loss": 0.8968, + "mean_token_accuracy": 0.7290531396865845, + "num_tokens": 41955835.0, + "step": 1390 + }, + { + "epoch": 0.1147027294466892, + "grad_norm": 1.9204453229904175, + "learning_rate": 3.8197306952459465e-07, + "loss": 0.7417, + "mean_token_accuracy": 0.7706171870231628, + "num_tokens": 41987680.0, + "step": 1391 + }, + { + "epoch": 0.11478519007174075, + "grad_norm": 2.139538526535034, + "learning_rate": 3.822478702940368e-07, + "loss": 1.0841, + "mean_token_accuracy": 0.6800360083580017, + "num_tokens": 42016917.0, + "step": 1392 + }, + { + "epoch": 0.11486765069679228, + "grad_norm": 1.8174920082092285, + "learning_rate": 3.8252267106347894e-07, + "loss": 0.8365, + "mean_token_accuracy": 0.7496597766876221, + "num_tokens": 42050922.0, + "step": 1393 + }, + { + "epoch": 0.11495011132184382, + "grad_norm": 1.9569387435913086, + "learning_rate": 3.827974718329211e-07, + "loss": 0.8372, + "mean_token_accuracy": 0.7470295429229736, + "num_tokens": 42085225.0, + "step": 1394 + }, + { + "epoch": 0.11503257194689535, + "grad_norm": 2.184128999710083, + "learning_rate": 3.8307227260236323e-07, + "loss": 0.7025, + "mean_token_accuracy": 0.7693310976028442, + "num_tokens": 42112611.0, + "step": 1395 + }, + { + "epoch": 0.1151150325719469, + "grad_norm": 2.6192851066589355, + "learning_rate": 3.8334707337180545e-07, + "loss": 1.0405, + "mean_token_accuracy": 0.6988491415977478, + "num_tokens": 42134897.0, + "step": 1396 + }, + { + "epoch": 0.11519749319699843, + "grad_norm": 2.3039543628692627, + "learning_rate": 3.8362187414124757e-07, + "loss": 0.9882, + "mean_token_accuracy": 0.7030636668205261, + "num_tokens": 42161595.0, + "step": 1397 + }, + { + "epoch": 0.11527995382204997, + "grad_norm": 2.1380841732025146, + "learning_rate": 3.8389667491068974e-07, + "loss": 0.7755, + "mean_token_accuracy": 0.7539756298065186, + "num_tokens": 42189840.0, + "step": 1398 + }, + { + "epoch": 0.1153624144471015, + "grad_norm": 1.9877467155456543, + "learning_rate": 3.8417147568013186e-07, + "loss": 1.0498, + "mean_token_accuracy": 0.6876652240753174, + "num_tokens": 42224684.0, + "step": 1399 + }, + { + "epoch": 0.11544487507215305, + "grad_norm": 1.9111428260803223, + "learning_rate": 3.8444627644957403e-07, + "loss": 0.9119, + "mean_token_accuracy": 0.7197059988975525, + "num_tokens": 42260151.0, + "step": 1400 + }, + { + "epoch": 0.11552733569720458, + "grad_norm": 2.0996639728546143, + "learning_rate": 3.847210772190162e-07, + "loss": 1.0108, + "mean_token_accuracy": 0.7046809196472168, + "num_tokens": 42293044.0, + "step": 1401 + }, + { + "epoch": 0.11560979632225613, + "grad_norm": 1.8794386386871338, + "learning_rate": 3.849958779884583e-07, + "loss": 0.7955, + "mean_token_accuracy": 0.7526528239250183, + "num_tokens": 42329526.0, + "step": 1402 + }, + { + "epoch": 0.11569225694730766, + "grad_norm": 2.201061487197876, + "learning_rate": 3.852706787579005e-07, + "loss": 0.8652, + "mean_token_accuracy": 0.7294347882270813, + "num_tokens": 42361184.0, + "step": 1403 + }, + { + "epoch": 0.1157747175723592, + "grad_norm": 2.2926244735717773, + "learning_rate": 3.8554547952734267e-07, + "loss": 0.7989, + "mean_token_accuracy": 0.7300792932510376, + "num_tokens": 42387316.0, + "step": 1404 + }, + { + "epoch": 0.11585717819741073, + "grad_norm": 2.226982831954956, + "learning_rate": 3.8582028029678484e-07, + "loss": 0.885, + "mean_token_accuracy": 0.7269067168235779, + "num_tokens": 42414274.0, + "step": 1405 + }, + { + "epoch": 0.11593963882246228, + "grad_norm": 1.8662797212600708, + "learning_rate": 3.8609508106622696e-07, + "loss": 0.8516, + "mean_token_accuracy": 0.7371129989624023, + "num_tokens": 42449199.0, + "step": 1406 + }, + { + "epoch": 0.11602209944751381, + "grad_norm": 2.3124263286590576, + "learning_rate": 3.863698818356691e-07, + "loss": 0.8791, + "mean_token_accuracy": 0.727696418762207, + "num_tokens": 42473439.0, + "step": 1407 + }, + { + "epoch": 0.11610456007256535, + "grad_norm": 1.8755234479904175, + "learning_rate": 3.866446826051113e-07, + "loss": 0.7252, + "mean_token_accuracy": 0.758804202079773, + "num_tokens": 42507490.0, + "step": 1408 + }, + { + "epoch": 0.11618702069761688, + "grad_norm": 2.2256932258605957, + "learning_rate": 3.869194833745534e-07, + "loss": 0.933, + "mean_token_accuracy": 0.7178033590316772, + "num_tokens": 42533663.0, + "step": 1409 + }, + { + "epoch": 0.11626948132266843, + "grad_norm": 2.014158010482788, + "learning_rate": 3.871942841439956e-07, + "loss": 0.9201, + "mean_token_accuracy": 0.7261185646057129, + "num_tokens": 42567921.0, + "step": 1410 + }, + { + "epoch": 0.11635194194771996, + "grad_norm": 1.9397066831588745, + "learning_rate": 3.874690849134377e-07, + "loss": 0.9392, + "mean_token_accuracy": 0.7170717716217041, + "num_tokens": 42600975.0, + "step": 1411 + }, + { + "epoch": 0.1164344025727715, + "grad_norm": 2.1583337783813477, + "learning_rate": 3.8774388568287993e-07, + "loss": 0.9518, + "mean_token_accuracy": 0.7238322496414185, + "num_tokens": 42629189.0, + "step": 1412 + }, + { + "epoch": 0.11651686319782303, + "grad_norm": 2.095757246017456, + "learning_rate": 3.8801868645232205e-07, + "loss": 0.7843, + "mean_token_accuracy": 0.7537742853164673, + "num_tokens": 42660053.0, + "step": 1413 + }, + { + "epoch": 0.11659932382287458, + "grad_norm": 1.8819408416748047, + "learning_rate": 3.8829348722176417e-07, + "loss": 0.8993, + "mean_token_accuracy": 0.731022834777832, + "num_tokens": 42694384.0, + "step": 1414 + }, + { + "epoch": 0.11668178444792611, + "grad_norm": 1.8786523342132568, + "learning_rate": 3.8856828799120634e-07, + "loss": 0.7397, + "mean_token_accuracy": 0.7761344909667969, + "num_tokens": 42729131.0, + "step": 1415 + }, + { + "epoch": 0.11676424507297765, + "grad_norm": 2.1729423999786377, + "learning_rate": 3.888430887606485e-07, + "loss": 0.8276, + "mean_token_accuracy": 0.7454015612602234, + "num_tokens": 42756503.0, + "step": 1416 + }, + { + "epoch": 0.11684670569802919, + "grad_norm": 1.9263132810592651, + "learning_rate": 3.891178895300907e-07, + "loss": 0.7332, + "mean_token_accuracy": 0.7661750316619873, + "num_tokens": 42787341.0, + "step": 1417 + }, + { + "epoch": 0.11692916632308073, + "grad_norm": 2.1693942546844482, + "learning_rate": 3.893926902995328e-07, + "loss": 0.9362, + "mean_token_accuracy": 0.7090581059455872, + "num_tokens": 42817013.0, + "step": 1418 + }, + { + "epoch": 0.11701162694813226, + "grad_norm": 2.0145697593688965, + "learning_rate": 3.89667491068975e-07, + "loss": 0.7692, + "mean_token_accuracy": 0.7635902762413025, + "num_tokens": 42845980.0, + "step": 1419 + }, + { + "epoch": 0.1170940875731838, + "grad_norm": 1.8751800060272217, + "learning_rate": 3.8994229183841715e-07, + "loss": 0.8544, + "mean_token_accuracy": 0.7373672723770142, + "num_tokens": 42882777.0, + "step": 1420 + }, + { + "epoch": 0.11717654819823534, + "grad_norm": 2.186852216720581, + "learning_rate": 3.9021709260785927e-07, + "loss": 1.0747, + "mean_token_accuracy": 0.6967514753341675, + "num_tokens": 42908937.0, + "step": 1421 + }, + { + "epoch": 0.11725900882328688, + "grad_norm": 2.0997724533081055, + "learning_rate": 3.9049189337730144e-07, + "loss": 0.9967, + "mean_token_accuracy": 0.704297661781311, + "num_tokens": 42938091.0, + "step": 1422 + }, + { + "epoch": 0.11734146944833841, + "grad_norm": 2.294269323348999, + "learning_rate": 3.9076669414674356e-07, + "loss": 0.928, + "mean_token_accuracy": 0.7182149291038513, + "num_tokens": 42964391.0, + "step": 1423 + }, + { + "epoch": 0.11742393007338996, + "grad_norm": 2.2042155265808105, + "learning_rate": 3.910414949161858e-07, + "loss": 0.864, + "mean_token_accuracy": 0.7328150868415833, + "num_tokens": 42992605.0, + "step": 1424 + }, + { + "epoch": 0.11750639069844149, + "grad_norm": 1.9616961479187012, + "learning_rate": 3.913162956856279e-07, + "loss": 0.7538, + "mean_token_accuracy": 0.7606306076049805, + "num_tokens": 43023083.0, + "step": 1425 + }, + { + "epoch": 0.11758885132349303, + "grad_norm": 1.8508795499801636, + "learning_rate": 3.9159109645507007e-07, + "loss": 0.8687, + "mean_token_accuracy": 0.7296756505966187, + "num_tokens": 43059317.0, + "step": 1426 + }, + { + "epoch": 0.11767131194854458, + "grad_norm": 2.250730514526367, + "learning_rate": 3.918658972245122e-07, + "loss": 0.8588, + "mean_token_accuracy": 0.730421781539917, + "num_tokens": 43085313.0, + "step": 1427 + }, + { + "epoch": 0.11775377257359611, + "grad_norm": 2.443739414215088, + "learning_rate": 3.9214069799395436e-07, + "loss": 1.0044, + "mean_token_accuracy": 0.7046804428100586, + "num_tokens": 43110710.0, + "step": 1428 + }, + { + "epoch": 0.11783623319864765, + "grad_norm": 1.939397931098938, + "learning_rate": 3.9241549876339653e-07, + "loss": 0.9166, + "mean_token_accuracy": 0.7205094695091248, + "num_tokens": 43146965.0, + "step": 1429 + }, + { + "epoch": 0.11791869382369918, + "grad_norm": 2.0594160556793213, + "learning_rate": 3.9269029953283865e-07, + "loss": 0.979, + "mean_token_accuracy": 0.7066776752471924, + "num_tokens": 43181736.0, + "step": 1430 + }, + { + "epoch": 0.11800115444875073, + "grad_norm": 1.780120849609375, + "learning_rate": 3.929651003022808e-07, + "loss": 0.7771, + "mean_token_accuracy": 0.7488316297531128, + "num_tokens": 43221065.0, + "step": 1431 + }, + { + "epoch": 0.11808361507380226, + "grad_norm": 2.3500916957855225, + "learning_rate": 3.93239901071723e-07, + "loss": 0.9556, + "mean_token_accuracy": 0.7135501503944397, + "num_tokens": 43246856.0, + "step": 1432 + }, + { + "epoch": 0.1181660756988538, + "grad_norm": 2.1823530197143555, + "learning_rate": 3.9351470184116517e-07, + "loss": 0.8447, + "mean_token_accuracy": 0.7400959730148315, + "num_tokens": 43273105.0, + "step": 1433 + }, + { + "epoch": 0.11824853632390533, + "grad_norm": 1.9580612182617188, + "learning_rate": 3.937895026106073e-07, + "loss": 0.9532, + "mean_token_accuracy": 0.7156270146369934, + "num_tokens": 43306497.0, + "step": 1434 + }, + { + "epoch": 0.11833099694895688, + "grad_norm": 2.38952374458313, + "learning_rate": 3.940643033800494e-07, + "loss": 0.9646, + "mean_token_accuracy": 0.7112021446228027, + "num_tokens": 43332812.0, + "step": 1435 + }, + { + "epoch": 0.11841345757400841, + "grad_norm": 1.8589389324188232, + "learning_rate": 3.9433910414949163e-07, + "loss": 0.8342, + "mean_token_accuracy": 0.7460728883743286, + "num_tokens": 43367613.0, + "step": 1436 + }, + { + "epoch": 0.11849591819905995, + "grad_norm": 1.7520450353622437, + "learning_rate": 3.9461390491893375e-07, + "loss": 0.7715, + "mean_token_accuracy": 0.7664405107498169, + "num_tokens": 43406940.0, + "step": 1437 + }, + { + "epoch": 0.11857837882411149, + "grad_norm": 2.1312215328216553, + "learning_rate": 3.948887056883759e-07, + "loss": 0.8803, + "mean_token_accuracy": 0.7293530702590942, + "num_tokens": 43437163.0, + "step": 1438 + }, + { + "epoch": 0.11866083944916303, + "grad_norm": 1.9535750150680542, + "learning_rate": 3.9516350645781804e-07, + "loss": 0.9179, + "mean_token_accuracy": 0.7227062582969666, + "num_tokens": 43472414.0, + "step": 1439 + }, + { + "epoch": 0.11874330007421456, + "grad_norm": 2.0808448791503906, + "learning_rate": 3.9543830722726026e-07, + "loss": 0.8922, + "mean_token_accuracy": 0.7252038717269897, + "num_tokens": 43505320.0, + "step": 1440 + }, + { + "epoch": 0.1188257606992661, + "grad_norm": 2.055175304412842, + "learning_rate": 3.957131079967024e-07, + "loss": 0.9148, + "mean_token_accuracy": 0.724056601524353, + "num_tokens": 43537248.0, + "step": 1441 + }, + { + "epoch": 0.11890822132431764, + "grad_norm": 2.0276033878326416, + "learning_rate": 3.959879087661445e-07, + "loss": 1.0642, + "mean_token_accuracy": 0.6836510300636292, + "num_tokens": 43568763.0, + "step": 1442 + }, + { + "epoch": 0.11899068194936918, + "grad_norm": 2.067218065261841, + "learning_rate": 3.9626270953558667e-07, + "loss": 0.8717, + "mean_token_accuracy": 0.7359845638275146, + "num_tokens": 43598741.0, + "step": 1443 + }, + { + "epoch": 0.11907314257442071, + "grad_norm": 1.9570051431655884, + "learning_rate": 3.9653751030502884e-07, + "loss": 0.7962, + "mean_token_accuracy": 0.7599301338195801, + "num_tokens": 43632182.0, + "step": 1444 + }, + { + "epoch": 0.11915560319947226, + "grad_norm": 1.9883909225463867, + "learning_rate": 3.96812311074471e-07, + "loss": 0.7518, + "mean_token_accuracy": 0.7729600667953491, + "num_tokens": 43662213.0, + "step": 1445 + }, + { + "epoch": 0.11923806382452379, + "grad_norm": 2.0569136142730713, + "learning_rate": 3.9708711184391313e-07, + "loss": 0.8658, + "mean_token_accuracy": 0.731813371181488, + "num_tokens": 43694514.0, + "step": 1446 + }, + { + "epoch": 0.11932052444957533, + "grad_norm": 2.0094246864318848, + "learning_rate": 3.9736191261335525e-07, + "loss": 0.9278, + "mean_token_accuracy": 0.7220869660377502, + "num_tokens": 43726260.0, + "step": 1447 + }, + { + "epoch": 0.11940298507462686, + "grad_norm": 2.407654047012329, + "learning_rate": 3.976367133827975e-07, + "loss": 0.8502, + "mean_token_accuracy": 0.7315276861190796, + "num_tokens": 43749359.0, + "step": 1448 + }, + { + "epoch": 0.11948544569967841, + "grad_norm": 2.0477309226989746, + "learning_rate": 3.979115141522396e-07, + "loss": 0.9093, + "mean_token_accuracy": 0.7255866527557373, + "num_tokens": 43782042.0, + "step": 1449 + }, + { + "epoch": 0.11956790632472994, + "grad_norm": 1.8783700466156006, + "learning_rate": 3.9818631492168177e-07, + "loss": 0.9517, + "mean_token_accuracy": 0.7104323506355286, + "num_tokens": 43817583.0, + "step": 1450 + }, + { + "epoch": 0.11965036694978148, + "grad_norm": 2.0733256340026855, + "learning_rate": 3.984611156911239e-07, + "loss": 1.0263, + "mean_token_accuracy": 0.7004494071006775, + "num_tokens": 43848605.0, + "step": 1451 + }, + { + "epoch": 0.11973282757483301, + "grad_norm": 2.1685473918914795, + "learning_rate": 3.987359164605661e-07, + "loss": 0.9623, + "mean_token_accuracy": 0.7099615335464478, + "num_tokens": 43879992.0, + "step": 1452 + }, + { + "epoch": 0.11981528819988456, + "grad_norm": 1.8800381422042847, + "learning_rate": 3.9901071723000823e-07, + "loss": 0.8857, + "mean_token_accuracy": 0.7397819757461548, + "num_tokens": 43915478.0, + "step": 1453 + }, + { + "epoch": 0.11989774882493609, + "grad_norm": 2.0346899032592773, + "learning_rate": 3.9928551799945035e-07, + "loss": 0.9962, + "mean_token_accuracy": 0.7084252834320068, + "num_tokens": 43947213.0, + "step": 1454 + }, + { + "epoch": 0.11998020944998763, + "grad_norm": 1.893096685409546, + "learning_rate": 3.995603187688925e-07, + "loss": 0.8947, + "mean_token_accuracy": 0.7358983159065247, + "num_tokens": 43983231.0, + "step": 1455 + }, + { + "epoch": 0.12006267007503917, + "grad_norm": 2.3333606719970703, + "learning_rate": 3.998351195383347e-07, + "loss": 0.8752, + "mean_token_accuracy": 0.7337930202484131, + "num_tokens": 44008363.0, + "step": 1456 + }, + { + "epoch": 0.12014513070009071, + "grad_norm": 2.016634464263916, + "learning_rate": 4.0010992030777686e-07, + "loss": 0.9256, + "mean_token_accuracy": 0.7094459533691406, + "num_tokens": 44040553.0, + "step": 1457 + }, + { + "epoch": 0.12022759132514224, + "grad_norm": 2.201349973678589, + "learning_rate": 4.00384721077219e-07, + "loss": 0.9452, + "mean_token_accuracy": 0.7148351073265076, + "num_tokens": 44067968.0, + "step": 1458 + }, + { + "epoch": 0.12031005195019379, + "grad_norm": 2.1382951736450195, + "learning_rate": 4.0065952184666115e-07, + "loss": 0.8208, + "mean_token_accuracy": 0.7506686449050903, + "num_tokens": 44094974.0, + "step": 1459 + }, + { + "epoch": 0.12039251257524532, + "grad_norm": 1.989112138748169, + "learning_rate": 4.009343226161033e-07, + "loss": 0.9678, + "mean_token_accuracy": 0.7044667601585388, + "num_tokens": 44128464.0, + "step": 1460 + }, + { + "epoch": 0.12047497320029686, + "grad_norm": 1.9330977201461792, + "learning_rate": 4.0120912338554544e-07, + "loss": 0.7601, + "mean_token_accuracy": 0.7534011602401733, + "num_tokens": 44159910.0, + "step": 1461 + }, + { + "epoch": 0.12055743382534839, + "grad_norm": 2.1827168464660645, + "learning_rate": 4.014839241549876e-07, + "loss": 0.7733, + "mean_token_accuracy": 0.7608629465103149, + "num_tokens": 44186404.0, + "step": 1462 + }, + { + "epoch": 0.12063989445039994, + "grad_norm": 1.7912676334381104, + "learning_rate": 4.0175872492442973e-07, + "loss": 0.8482, + "mean_token_accuracy": 0.7365294098854065, + "num_tokens": 44222790.0, + "step": 1463 + }, + { + "epoch": 0.12072235507545147, + "grad_norm": 1.8780651092529297, + "learning_rate": 4.0203352569387196e-07, + "loss": 0.6841, + "mean_token_accuracy": 0.7827097177505493, + "num_tokens": 44255432.0, + "step": 1464 + }, + { + "epoch": 0.12080481570050301, + "grad_norm": 2.1249492168426514, + "learning_rate": 4.023083264633141e-07, + "loss": 1.0145, + "mean_token_accuracy": 0.7005302309989929, + "num_tokens": 44287127.0, + "step": 1465 + }, + { + "epoch": 0.12088727632555454, + "grad_norm": 2.155423879623413, + "learning_rate": 4.0258312723275625e-07, + "loss": 0.7912, + "mean_token_accuracy": 0.7519180774688721, + "num_tokens": 44314329.0, + "step": 1466 + }, + { + "epoch": 0.12096973695060609, + "grad_norm": 2.129312515258789, + "learning_rate": 4.0285792800219836e-07, + "loss": 0.9143, + "mean_token_accuracy": 0.7153382301330566, + "num_tokens": 44345789.0, + "step": 1467 + }, + { + "epoch": 0.12105219757565762, + "grad_norm": 2.439431667327881, + "learning_rate": 4.0313272877164054e-07, + "loss": 1.0396, + "mean_token_accuracy": 0.6883304715156555, + "num_tokens": 44370069.0, + "step": 1468 + }, + { + "epoch": 0.12113465820070916, + "grad_norm": 1.8737342357635498, + "learning_rate": 4.034075295410827e-07, + "loss": 0.7283, + "mean_token_accuracy": 0.7758364677429199, + "num_tokens": 44402685.0, + "step": 1469 + }, + { + "epoch": 0.1212171188257607, + "grad_norm": 2.2971036434173584, + "learning_rate": 4.0368233031052483e-07, + "loss": 0.9244, + "mean_token_accuracy": 0.7231711149215698, + "num_tokens": 44428180.0, + "step": 1470 + }, + { + "epoch": 0.12129957945081224, + "grad_norm": 2.042788028717041, + "learning_rate": 4.03957131079967e-07, + "loss": 0.9881, + "mean_token_accuracy": 0.6966698169708252, + "num_tokens": 44462096.0, + "step": 1471 + }, + { + "epoch": 0.12138204007586377, + "grad_norm": 2.321166753768921, + "learning_rate": 4.0423193184940917e-07, + "loss": 0.9792, + "mean_token_accuracy": 0.7059205174446106, + "num_tokens": 44488032.0, + "step": 1472 + }, + { + "epoch": 0.12146450070091531, + "grad_norm": 2.007042646408081, + "learning_rate": 4.0450673261885134e-07, + "loss": 0.7882, + "mean_token_accuracy": 0.7597799301147461, + "num_tokens": 44521609.0, + "step": 1473 + }, + { + "epoch": 0.12154696132596685, + "grad_norm": 2.396237373352051, + "learning_rate": 4.0478153338829346e-07, + "loss": 1.095, + "mean_token_accuracy": 0.6824365258216858, + "num_tokens": 44546783.0, + "step": 1474 + }, + { + "epoch": 0.12162942195101839, + "grad_norm": 2.1825263500213623, + "learning_rate": 4.050563341577356e-07, + "loss": 0.8696, + "mean_token_accuracy": 0.7342220544815063, + "num_tokens": 44575835.0, + "step": 1475 + }, + { + "epoch": 0.12171188257606992, + "grad_norm": 2.2596042156219482, + "learning_rate": 4.053311349271778e-07, + "loss": 0.9954, + "mean_token_accuracy": 0.7012727856636047, + "num_tokens": 44602954.0, + "step": 1476 + }, + { + "epoch": 0.12179434320112147, + "grad_norm": 1.9796736240386963, + "learning_rate": 4.056059356966199e-07, + "loss": 0.9625, + "mean_token_accuracy": 0.7150213718414307, + "num_tokens": 44634497.0, + "step": 1477 + }, + { + "epoch": 0.121876803826173, + "grad_norm": 1.9907236099243164, + "learning_rate": 4.058807364660621e-07, + "loss": 0.8368, + "mean_token_accuracy": 0.7440563440322876, + "num_tokens": 44667129.0, + "step": 1478 + }, + { + "epoch": 0.12195926445122454, + "grad_norm": 2.0532920360565186, + "learning_rate": 4.061555372355042e-07, + "loss": 0.8105, + "mean_token_accuracy": 0.7426314949989319, + "num_tokens": 44695670.0, + "step": 1479 + }, + { + "epoch": 0.12204172507627607, + "grad_norm": 2.1565170288085938, + "learning_rate": 4.0643033800494644e-07, + "loss": 0.8116, + "mean_token_accuracy": 0.7444955706596375, + "num_tokens": 44723918.0, + "step": 1480 + }, + { + "epoch": 0.12212418570132762, + "grad_norm": 2.347506046295166, + "learning_rate": 4.0670513877438856e-07, + "loss": 0.8149, + "mean_token_accuracy": 0.7468400597572327, + "num_tokens": 44749059.0, + "step": 1481 + }, + { + "epoch": 0.12220664632637915, + "grad_norm": 2.3630106449127197, + "learning_rate": 4.069799395438307e-07, + "loss": 0.9403, + "mean_token_accuracy": 0.7291049957275391, + "num_tokens": 44775194.0, + "step": 1482 + }, + { + "epoch": 0.12228910695143069, + "grad_norm": 2.053070306777954, + "learning_rate": 4.0725474031327285e-07, + "loss": 0.8876, + "mean_token_accuracy": 0.7345468401908875, + "num_tokens": 44806448.0, + "step": 1483 + }, + { + "epoch": 0.12237156757648222, + "grad_norm": 2.1783945560455322, + "learning_rate": 4.07529541082715e-07, + "loss": 0.8057, + "mean_token_accuracy": 0.7469872236251831, + "num_tokens": 44836489.0, + "step": 1484 + }, + { + "epoch": 0.12245402820153377, + "grad_norm": 2.3485965728759766, + "learning_rate": 4.078043418521572e-07, + "loss": 0.8313, + "mean_token_accuracy": 0.7351621389389038, + "num_tokens": 44863015.0, + "step": 1485 + }, + { + "epoch": 0.1225364888265853, + "grad_norm": 2.059657573699951, + "learning_rate": 4.080791426215993e-07, + "loss": 0.7825, + "mean_token_accuracy": 0.7634377479553223, + "num_tokens": 44891485.0, + "step": 1486 + }, + { + "epoch": 0.12261894945163684, + "grad_norm": 2.0112252235412598, + "learning_rate": 4.083539433910415e-07, + "loss": 0.856, + "mean_token_accuracy": 0.7274739742279053, + "num_tokens": 44921639.0, + "step": 1487 + }, + { + "epoch": 0.12270141007668837, + "grad_norm": 1.935770034790039, + "learning_rate": 4.0862874416048365e-07, + "loss": 0.9085, + "mean_token_accuracy": 0.7258907556533813, + "num_tokens": 44954890.0, + "step": 1488 + }, + { + "epoch": 0.12278387070173992, + "grad_norm": 2.271331310272217, + "learning_rate": 4.0890354492992577e-07, + "loss": 0.9947, + "mean_token_accuracy": 0.7053195834159851, + "num_tokens": 44980485.0, + "step": 1489 + }, + { + "epoch": 0.12286633132679145, + "grad_norm": 2.069514274597168, + "learning_rate": 4.0917834569936794e-07, + "loss": 0.8774, + "mean_token_accuracy": 0.7312343120574951, + "num_tokens": 45010821.0, + "step": 1490 + }, + { + "epoch": 0.122948791951843, + "grad_norm": 2.120191812515259, + "learning_rate": 4.0945314646881006e-07, + "loss": 0.8783, + "mean_token_accuracy": 0.7236340045928955, + "num_tokens": 45040581.0, + "step": 1491 + }, + { + "epoch": 0.12303125257689454, + "grad_norm": 2.0584616661071777, + "learning_rate": 4.097279472382523e-07, + "loss": 0.8722, + "mean_token_accuracy": 0.7382363080978394, + "num_tokens": 45068299.0, + "step": 1492 + }, + { + "epoch": 0.12311371320194607, + "grad_norm": 2.134258985519409, + "learning_rate": 4.100027480076944e-07, + "loss": 0.9743, + "mean_token_accuracy": 0.7175441980361938, + "num_tokens": 45097742.0, + "step": 1493 + }, + { + "epoch": 0.12319617382699762, + "grad_norm": 1.95363187789917, + "learning_rate": 4.102775487771366e-07, + "loss": 0.912, + "mean_token_accuracy": 0.7293864488601685, + "num_tokens": 45131823.0, + "step": 1494 + }, + { + "epoch": 0.12327863445204915, + "grad_norm": 2.098853588104248, + "learning_rate": 4.105523495465787e-07, + "loss": 1.0296, + "mean_token_accuracy": 0.7001122832298279, + "num_tokens": 45167919.0, + "step": 1495 + }, + { + "epoch": 0.12336109507710069, + "grad_norm": 1.9000482559204102, + "learning_rate": 4.1082715031602086e-07, + "loss": 0.7623, + "mean_token_accuracy": 0.7675463557243347, + "num_tokens": 45198491.0, + "step": 1496 + }, + { + "epoch": 0.12344355570215222, + "grad_norm": 2.0937886238098145, + "learning_rate": 4.1110195108546304e-07, + "loss": 0.8525, + "mean_token_accuracy": 0.7345645427703857, + "num_tokens": 45226967.0, + "step": 1497 + }, + { + "epoch": 0.12352601632720377, + "grad_norm": 1.8677558898925781, + "learning_rate": 4.1137675185490515e-07, + "loss": 1.0286, + "mean_token_accuracy": 0.6992236375808716, + "num_tokens": 45260744.0, + "step": 1498 + }, + { + "epoch": 0.1236084769522553, + "grad_norm": 1.9445526599884033, + "learning_rate": 4.116515526243473e-07, + "loss": 0.7857, + "mean_token_accuracy": 0.7446334362030029, + "num_tokens": 45290475.0, + "step": 1499 + }, + { + "epoch": 0.12369093757730684, + "grad_norm": 1.8286954164505005, + "learning_rate": 4.119263533937895e-07, + "loss": 0.7442, + "mean_token_accuracy": 0.7617647647857666, + "num_tokens": 45322598.0, + "step": 1500 + }, + { + "epoch": 0.12377339820235837, + "grad_norm": 2.786271095275879, + "learning_rate": 4.1220115416323167e-07, + "loss": 0.8599, + "mean_token_accuracy": 0.7363649606704712, + "num_tokens": 45348854.0, + "step": 1501 + }, + { + "epoch": 0.12385585882740992, + "grad_norm": 2.4127538204193115, + "learning_rate": 4.124759549326738e-07, + "loss": 0.9311, + "mean_token_accuracy": 0.7126359343528748, + "num_tokens": 45374233.0, + "step": 1502 + }, + { + "epoch": 0.12393831945246145, + "grad_norm": 1.800318956375122, + "learning_rate": 4.127507557021159e-07, + "loss": 0.6588, + "mean_token_accuracy": 0.7925533652305603, + "num_tokens": 45406808.0, + "step": 1503 + }, + { + "epoch": 0.124020780077513, + "grad_norm": 2.0393447875976562, + "learning_rate": 4.1302555647155813e-07, + "loss": 0.9348, + "mean_token_accuracy": 0.7165430784225464, + "num_tokens": 45437503.0, + "step": 1504 + }, + { + "epoch": 0.12410324070256452, + "grad_norm": 2.2892191410064697, + "learning_rate": 4.1330035724100025e-07, + "loss": 0.9811, + "mean_token_accuracy": 0.6995816826820374, + "num_tokens": 45465853.0, + "step": 1505 + }, + { + "epoch": 0.12418570132761607, + "grad_norm": 2.0133190155029297, + "learning_rate": 4.135751580104424e-07, + "loss": 0.8013, + "mean_token_accuracy": 0.7593821287155151, + "num_tokens": 45496098.0, + "step": 1506 + }, + { + "epoch": 0.1242681619526676, + "grad_norm": 2.048687219619751, + "learning_rate": 4.1384995877988454e-07, + "loss": 0.8781, + "mean_token_accuracy": 0.7356740236282349, + "num_tokens": 45527264.0, + "step": 1507 + }, + { + "epoch": 0.12435062257771914, + "grad_norm": 2.075474500656128, + "learning_rate": 4.1412475954932676e-07, + "loss": 1.0204, + "mean_token_accuracy": 0.693838357925415, + "num_tokens": 45560822.0, + "step": 1508 + }, + { + "epoch": 0.12443308320277068, + "grad_norm": 2.1905882358551025, + "learning_rate": 4.143995603187689e-07, + "loss": 0.9531, + "mean_token_accuracy": 0.710883378982544, + "num_tokens": 45590107.0, + "step": 1509 + }, + { + "epoch": 0.12451554382782222, + "grad_norm": 2.1169941425323486, + "learning_rate": 4.14674361088211e-07, + "loss": 0.8682, + "mean_token_accuracy": 0.7296448945999146, + "num_tokens": 45620640.0, + "step": 1510 + }, + { + "epoch": 0.12459800445287375, + "grad_norm": 1.899823546409607, + "learning_rate": 4.1494916185765317e-07, + "loss": 0.9542, + "mean_token_accuracy": 0.7195577621459961, + "num_tokens": 45658856.0, + "step": 1511 + }, + { + "epoch": 0.1246804650779253, + "grad_norm": 2.1712381839752197, + "learning_rate": 4.1522396262709534e-07, + "loss": 0.9113, + "mean_token_accuracy": 0.7294025421142578, + "num_tokens": 45688199.0, + "step": 1512 + }, + { + "epoch": 0.12476292570297683, + "grad_norm": 2.3823037147521973, + "learning_rate": 4.154987633965375e-07, + "loss": 0.9605, + "mean_token_accuracy": 0.714146614074707, + "num_tokens": 45712885.0, + "step": 1513 + }, + { + "epoch": 0.12484538632802837, + "grad_norm": 1.89915931224823, + "learning_rate": 4.1577356416597963e-07, + "loss": 0.7343, + "mean_token_accuracy": 0.7649499177932739, + "num_tokens": 45748220.0, + "step": 1514 + }, + { + "epoch": 0.1249278469530799, + "grad_norm": 2.0651421546936035, + "learning_rate": 4.160483649354218e-07, + "loss": 0.7878, + "mean_token_accuracy": 0.748073399066925, + "num_tokens": 45776114.0, + "step": 1515 + }, + { + "epoch": 0.12501030757813145, + "grad_norm": 1.7935121059417725, + "learning_rate": 4.16323165704864e-07, + "loss": 0.8632, + "mean_token_accuracy": 0.739965558052063, + "num_tokens": 45815633.0, + "step": 1516 + }, + { + "epoch": 0.125092768203183, + "grad_norm": 1.981192946434021, + "learning_rate": 4.165979664743061e-07, + "loss": 0.7845, + "mean_token_accuracy": 0.7538737654685974, + "num_tokens": 45845504.0, + "step": 1517 + }, + { + "epoch": 0.1251752288282345, + "grad_norm": 2.27514910697937, + "learning_rate": 4.1687276724374827e-07, + "loss": 0.8849, + "mean_token_accuracy": 0.7282694578170776, + "num_tokens": 45871791.0, + "step": 1518 + }, + { + "epoch": 0.12525768945328605, + "grad_norm": 2.1811013221740723, + "learning_rate": 4.171475680131904e-07, + "loss": 0.9889, + "mean_token_accuracy": 0.7003824710845947, + "num_tokens": 45902714.0, + "step": 1519 + }, + { + "epoch": 0.1253401500783376, + "grad_norm": 2.036602020263672, + "learning_rate": 4.174223687826326e-07, + "loss": 0.9406, + "mean_token_accuracy": 0.7167147397994995, + "num_tokens": 45932097.0, + "step": 1520 + }, + { + "epoch": 0.12542261070338914, + "grad_norm": 1.9426872730255127, + "learning_rate": 4.1769716955207473e-07, + "loss": 0.7887, + "mean_token_accuracy": 0.7519102096557617, + "num_tokens": 45963602.0, + "step": 1521 + }, + { + "epoch": 0.12550507132844066, + "grad_norm": 2.0275497436523438, + "learning_rate": 4.179719703215169e-07, + "loss": 0.7563, + "mean_token_accuracy": 0.759791374206543, + "num_tokens": 45995142.0, + "step": 1522 + }, + { + "epoch": 0.1255875319534922, + "grad_norm": 1.89125657081604, + "learning_rate": 4.18246771090959e-07, + "loss": 0.784, + "mean_token_accuracy": 0.7598589658737183, + "num_tokens": 46029280.0, + "step": 1523 + }, + { + "epoch": 0.12566999257854375, + "grad_norm": 1.9972224235534668, + "learning_rate": 4.185215718604012e-07, + "loss": 0.7819, + "mean_token_accuracy": 0.7441714406013489, + "num_tokens": 46060282.0, + "step": 1524 + }, + { + "epoch": 0.1257524532035953, + "grad_norm": 1.9727131128311157, + "learning_rate": 4.1879637262984336e-07, + "loss": 0.8302, + "mean_token_accuracy": 0.7433643937110901, + "num_tokens": 46096763.0, + "step": 1525 + }, + { + "epoch": 0.1258349138286468, + "grad_norm": 2.088958740234375, + "learning_rate": 4.190711733992855e-07, + "loss": 0.7341, + "mean_token_accuracy": 0.761747419834137, + "num_tokens": 46123658.0, + "step": 1526 + }, + { + "epoch": 0.12591737445369836, + "grad_norm": 2.1675026416778564, + "learning_rate": 4.1934597416872765e-07, + "loss": 0.8104, + "mean_token_accuracy": 0.7533636689186096, + "num_tokens": 46152046.0, + "step": 1527 + }, + { + "epoch": 0.1259998350787499, + "grad_norm": 2.2458255290985107, + "learning_rate": 4.196207749381698e-07, + "loss": 0.9373, + "mean_token_accuracy": 0.7214813232421875, + "num_tokens": 46180332.0, + "step": 1528 + }, + { + "epoch": 0.12608229570380144, + "grad_norm": 1.8501249551773071, + "learning_rate": 4.19895575707612e-07, + "loss": 0.8168, + "mean_token_accuracy": 0.7538939714431763, + "num_tokens": 46215916.0, + "step": 1529 + }, + { + "epoch": 0.12616475632885296, + "grad_norm": 1.993330478668213, + "learning_rate": 4.201703764770541e-07, + "loss": 0.9565, + "mean_token_accuracy": 0.7089114189147949, + "num_tokens": 46250613.0, + "step": 1530 + }, + { + "epoch": 0.1262472169539045, + "grad_norm": 2.0238754749298096, + "learning_rate": 4.2044517724649623e-07, + "loss": 0.8542, + "mean_token_accuracy": 0.7386036515235901, + "num_tokens": 46283714.0, + "step": 1531 + }, + { + "epoch": 0.12632967757895605, + "grad_norm": 2.094867706298828, + "learning_rate": 4.2071997801593846e-07, + "loss": 0.8487, + "mean_token_accuracy": 0.7422693967819214, + "num_tokens": 46313510.0, + "step": 1532 + }, + { + "epoch": 0.1264121382040076, + "grad_norm": 2.2704946994781494, + "learning_rate": 4.209947787853806e-07, + "loss": 0.9402, + "mean_token_accuracy": 0.7182186841964722, + "num_tokens": 46339050.0, + "step": 1533 + }, + { + "epoch": 0.1264945988290591, + "grad_norm": 1.9231773614883423, + "learning_rate": 4.2126957955482275e-07, + "loss": 0.7466, + "mean_token_accuracy": 0.7724407911300659, + "num_tokens": 46369875.0, + "step": 1534 + }, + { + "epoch": 0.12657705945411066, + "grad_norm": 1.9216291904449463, + "learning_rate": 4.2154438032426487e-07, + "loss": 0.8902, + "mean_token_accuracy": 0.7378148436546326, + "num_tokens": 46403624.0, + "step": 1535 + }, + { + "epoch": 0.1266595200791622, + "grad_norm": 2.2232983112335205, + "learning_rate": 4.218191810937071e-07, + "loss": 0.9372, + "mean_token_accuracy": 0.7187195420265198, + "num_tokens": 46430674.0, + "step": 1536 + }, + { + "epoch": 0.12674198070421375, + "grad_norm": 2.027724266052246, + "learning_rate": 4.220939818631492e-07, + "loss": 0.8012, + "mean_token_accuracy": 0.7566806077957153, + "num_tokens": 46461924.0, + "step": 1537 + }, + { + "epoch": 0.12682444132926526, + "grad_norm": 1.9096806049346924, + "learning_rate": 4.2236878263259133e-07, + "loss": 0.8637, + "mean_token_accuracy": 0.7396245002746582, + "num_tokens": 46495076.0, + "step": 1538 + }, + { + "epoch": 0.1269069019543168, + "grad_norm": 2.0206902027130127, + "learning_rate": 4.226435834020335e-07, + "loss": 0.8812, + "mean_token_accuracy": 0.7427130937576294, + "num_tokens": 46524193.0, + "step": 1539 + }, + { + "epoch": 0.12698936257936835, + "grad_norm": 2.144253730773926, + "learning_rate": 4.2291838417147567e-07, + "loss": 0.765, + "mean_token_accuracy": 0.7588068246841431, + "num_tokens": 46551499.0, + "step": 1540 + }, + { + "epoch": 0.1270718232044199, + "grad_norm": 2.2388710975646973, + "learning_rate": 4.2319318494091784e-07, + "loss": 1.0225, + "mean_token_accuracy": 0.7023499011993408, + "num_tokens": 46577646.0, + "step": 1541 + }, + { + "epoch": 0.12715428382947142, + "grad_norm": 2.2422070503234863, + "learning_rate": 4.2346798571035996e-07, + "loss": 0.9593, + "mean_token_accuracy": 0.7077398300170898, + "num_tokens": 46604007.0, + "step": 1542 + }, + { + "epoch": 0.12723674445452296, + "grad_norm": 2.274787187576294, + "learning_rate": 4.237427864798021e-07, + "loss": 0.9274, + "mean_token_accuracy": 0.7218794226646423, + "num_tokens": 46631827.0, + "step": 1543 + }, + { + "epoch": 0.1273192050795745, + "grad_norm": 2.0387260913848877, + "learning_rate": 4.240175872492443e-07, + "loss": 0.8856, + "mean_token_accuracy": 0.7234577536582947, + "num_tokens": 46663155.0, + "step": 1544 + }, + { + "epoch": 0.12740166570462605, + "grad_norm": 2.255204439163208, + "learning_rate": 4.242923880186864e-07, + "loss": 1.0879, + "mean_token_accuracy": 0.6712806224822998, + "num_tokens": 46690369.0, + "step": 1545 + }, + { + "epoch": 0.12748412632967757, + "grad_norm": 2.348053455352783, + "learning_rate": 4.245671887881286e-07, + "loss": 0.9326, + "mean_token_accuracy": 0.720270037651062, + "num_tokens": 46716666.0, + "step": 1546 + }, + { + "epoch": 0.1275665869547291, + "grad_norm": 2.2191264629364014, + "learning_rate": 4.248419895575707e-07, + "loss": 0.8193, + "mean_token_accuracy": 0.7406450510025024, + "num_tokens": 46748117.0, + "step": 1547 + }, + { + "epoch": 0.12764904757978066, + "grad_norm": 2.355952024459839, + "learning_rate": 4.2511679032701294e-07, + "loss": 0.9295, + "mean_token_accuracy": 0.7130861878395081, + "num_tokens": 46772067.0, + "step": 1548 + }, + { + "epoch": 0.1277315082048322, + "grad_norm": 2.0131571292877197, + "learning_rate": 4.2539159109645506e-07, + "loss": 0.9897, + "mean_token_accuracy": 0.7055280208587646, + "num_tokens": 46808277.0, + "step": 1549 + }, + { + "epoch": 0.12781396882988372, + "grad_norm": 2.076507568359375, + "learning_rate": 4.256663918658972e-07, + "loss": 0.8872, + "mean_token_accuracy": 0.7258000373840332, + "num_tokens": 46838112.0, + "step": 1550 + }, + { + "epoch": 0.12789642945493526, + "grad_norm": 1.9285833835601807, + "learning_rate": 4.2594119263533935e-07, + "loss": 0.7772, + "mean_token_accuracy": 0.7598193287849426, + "num_tokens": 46873165.0, + "step": 1551 + }, + { + "epoch": 0.1279788900799868, + "grad_norm": 2.136676788330078, + "learning_rate": 4.262159934047815e-07, + "loss": 0.9028, + "mean_token_accuracy": 0.7197006940841675, + "num_tokens": 46901276.0, + "step": 1552 + }, + { + "epoch": 0.12806135070503835, + "grad_norm": 2.4475691318511963, + "learning_rate": 4.264907941742237e-07, + "loss": 0.9982, + "mean_token_accuracy": 0.7038230895996094, + "num_tokens": 46923831.0, + "step": 1553 + }, + { + "epoch": 0.12814381133008987, + "grad_norm": 2.088773250579834, + "learning_rate": 4.267655949436658e-07, + "loss": 0.92, + "mean_token_accuracy": 0.7261974215507507, + "num_tokens": 46953614.0, + "step": 1554 + }, + { + "epoch": 0.1282262719551414, + "grad_norm": 2.152796983718872, + "learning_rate": 4.27040395713108e-07, + "loss": 0.9067, + "mean_token_accuracy": 0.7311045527458191, + "num_tokens": 46982087.0, + "step": 1555 + }, + { + "epoch": 0.12830873258019296, + "grad_norm": 2.0170187950134277, + "learning_rate": 4.2731519648255015e-07, + "loss": 0.8892, + "mean_token_accuracy": 0.7329636216163635, + "num_tokens": 47011767.0, + "step": 1556 + }, + { + "epoch": 0.1283911932052445, + "grad_norm": 1.8653969764709473, + "learning_rate": 4.2758999725199227e-07, + "loss": 0.7021, + "mean_token_accuracy": 0.7792178392410278, + "num_tokens": 47045822.0, + "step": 1557 + }, + { + "epoch": 0.12847365383029605, + "grad_norm": 2.1577441692352295, + "learning_rate": 4.2786479802143444e-07, + "loss": 1.084, + "mean_token_accuracy": 0.6790626049041748, + "num_tokens": 47076371.0, + "step": 1558 + }, + { + "epoch": 0.12855611445534756, + "grad_norm": 1.8398061990737915, + "learning_rate": 4.2813959879087656e-07, + "loss": 0.681, + "mean_token_accuracy": 0.7877324819564819, + "num_tokens": 47110296.0, + "step": 1559 + }, + { + "epoch": 0.1286385750803991, + "grad_norm": 2.0732293128967285, + "learning_rate": 4.284143995603188e-07, + "loss": 1.0413, + "mean_token_accuracy": 0.6913267374038696, + "num_tokens": 47142340.0, + "step": 1560 + }, + { + "epoch": 0.12872103570545065, + "grad_norm": 1.8655226230621338, + "learning_rate": 4.286892003297609e-07, + "loss": 0.888, + "mean_token_accuracy": 0.7296017408370972, + "num_tokens": 47179655.0, + "step": 1561 + }, + { + "epoch": 0.1288034963305022, + "grad_norm": 1.9198576211929321, + "learning_rate": 4.289640010992031e-07, + "loss": 0.827, + "mean_token_accuracy": 0.7500269412994385, + "num_tokens": 47209897.0, + "step": 1562 + }, + { + "epoch": 0.12888595695555372, + "grad_norm": 2.156834363937378, + "learning_rate": 4.292388018686452e-07, + "loss": 1.0286, + "mean_token_accuracy": 0.6966935992240906, + "num_tokens": 47241255.0, + "step": 1563 + }, + { + "epoch": 0.12896841758060526, + "grad_norm": 1.9640065431594849, + "learning_rate": 4.2951360263808737e-07, + "loss": 0.8813, + "mean_token_accuracy": 0.7269206047058105, + "num_tokens": 47273131.0, + "step": 1564 + }, + { + "epoch": 0.1290508782056568, + "grad_norm": 2.0160467624664307, + "learning_rate": 4.2978840340752954e-07, + "loss": 0.7875, + "mean_token_accuracy": 0.7499583959579468, + "num_tokens": 47303714.0, + "step": 1565 + }, + { + "epoch": 0.12913333883070835, + "grad_norm": 1.956241488456726, + "learning_rate": 4.3006320417697166e-07, + "loss": 0.8868, + "mean_token_accuracy": 0.7372139692306519, + "num_tokens": 47335771.0, + "step": 1566 + }, + { + "epoch": 0.12921579945575987, + "grad_norm": 1.984337568283081, + "learning_rate": 4.3033800494641383e-07, + "loss": 0.8235, + "mean_token_accuracy": 0.7543558478355408, + "num_tokens": 47366807.0, + "step": 1567 + }, + { + "epoch": 0.1292982600808114, + "grad_norm": 2.260270595550537, + "learning_rate": 4.30612805715856e-07, + "loss": 0.9296, + "mean_token_accuracy": 0.7189995050430298, + "num_tokens": 47394191.0, + "step": 1568 + }, + { + "epoch": 0.12938072070586296, + "grad_norm": 2.169987916946411, + "learning_rate": 4.3088760648529817e-07, + "loss": 1.066, + "mean_token_accuracy": 0.6792118549346924, + "num_tokens": 47423943.0, + "step": 1569 + }, + { + "epoch": 0.1294631813309145, + "grad_norm": 2.01930570602417, + "learning_rate": 4.311624072547403e-07, + "loss": 0.9494, + "mean_token_accuracy": 0.719319224357605, + "num_tokens": 47455669.0, + "step": 1570 + }, + { + "epoch": 0.12954564195596602, + "grad_norm": 2.015421152114868, + "learning_rate": 4.314372080241824e-07, + "loss": 0.8049, + "mean_token_accuracy": 0.7467403411865234, + "num_tokens": 47485190.0, + "step": 1571 + }, + { + "epoch": 0.12962810258101756, + "grad_norm": 2.3981692790985107, + "learning_rate": 4.3171200879362463e-07, + "loss": 0.9564, + "mean_token_accuracy": 0.7100774049758911, + "num_tokens": 47509645.0, + "step": 1572 + }, + { + "epoch": 0.1297105632060691, + "grad_norm": 1.9825496673583984, + "learning_rate": 4.3198680956306675e-07, + "loss": 0.9518, + "mean_token_accuracy": 0.7231554985046387, + "num_tokens": 47546696.0, + "step": 1573 + }, + { + "epoch": 0.12979302383112065, + "grad_norm": 1.9620606899261475, + "learning_rate": 4.322616103325089e-07, + "loss": 0.7818, + "mean_token_accuracy": 0.7551836967468262, + "num_tokens": 47577672.0, + "step": 1574 + }, + { + "epoch": 0.12987548445617217, + "grad_norm": 2.246917963027954, + "learning_rate": 4.3253641110195104e-07, + "loss": 0.9894, + "mean_token_accuracy": 0.7131004333496094, + "num_tokens": 47608952.0, + "step": 1575 + }, + { + "epoch": 0.1299579450812237, + "grad_norm": 1.8991763591766357, + "learning_rate": 4.3281121187139327e-07, + "loss": 0.85, + "mean_token_accuracy": 0.7445151209831238, + "num_tokens": 47642549.0, + "step": 1576 + }, + { + "epoch": 0.13004040570627526, + "grad_norm": 2.183638572692871, + "learning_rate": 4.330860126408354e-07, + "loss": 0.8481, + "mean_token_accuracy": 0.7266889214515686, + "num_tokens": 47671390.0, + "step": 1577 + }, + { + "epoch": 0.1301228663313268, + "grad_norm": 2.0523624420166016, + "learning_rate": 4.333608134102775e-07, + "loss": 0.7184, + "mean_token_accuracy": 0.7818061113357544, + "num_tokens": 47701409.0, + "step": 1578 + }, + { + "epoch": 0.13020532695637832, + "grad_norm": 2.173044204711914, + "learning_rate": 4.336356141797197e-07, + "loss": 0.7939, + "mean_token_accuracy": 0.7549862265586853, + "num_tokens": 47728657.0, + "step": 1579 + }, + { + "epoch": 0.13028778758142986, + "grad_norm": 2.1465351581573486, + "learning_rate": 4.3391041494916185e-07, + "loss": 0.8502, + "mean_token_accuracy": 0.7386811971664429, + "num_tokens": 47757519.0, + "step": 1580 + }, + { + "epoch": 0.1303702482064814, + "grad_norm": 2.0412933826446533, + "learning_rate": 4.34185215718604e-07, + "loss": 0.8105, + "mean_token_accuracy": 0.7481038570404053, + "num_tokens": 47787213.0, + "step": 1581 + }, + { + "epoch": 0.13045270883153295, + "grad_norm": 2.0226621627807617, + "learning_rate": 4.3446001648804614e-07, + "loss": 0.8798, + "mean_token_accuracy": 0.7345763444900513, + "num_tokens": 47819593.0, + "step": 1582 + }, + { + "epoch": 0.13053516945658447, + "grad_norm": 2.49055814743042, + "learning_rate": 4.347348172574883e-07, + "loss": 0.9107, + "mean_token_accuracy": 0.721087634563446, + "num_tokens": 47841795.0, + "step": 1583 + }, + { + "epoch": 0.13061763008163602, + "grad_norm": 2.1897852420806885, + "learning_rate": 4.350096180269305e-07, + "loss": 0.8997, + "mean_token_accuracy": 0.7286381125450134, + "num_tokens": 47869933.0, + "step": 1584 + }, + { + "epoch": 0.13070009070668756, + "grad_norm": 2.2085366249084473, + "learning_rate": 4.352844187963726e-07, + "loss": 0.9407, + "mean_token_accuracy": 0.7183750867843628, + "num_tokens": 47898532.0, + "step": 1585 + }, + { + "epoch": 0.1307825513317391, + "grad_norm": 2.0651636123657227, + "learning_rate": 4.3555921956581477e-07, + "loss": 0.938, + "mean_token_accuracy": 0.720497727394104, + "num_tokens": 47931061.0, + "step": 1586 + }, + { + "epoch": 0.13086501195679062, + "grad_norm": 2.3742620944976807, + "learning_rate": 4.358340203352569e-07, + "loss": 0.8156, + "mean_token_accuracy": 0.7439684867858887, + "num_tokens": 47955549.0, + "step": 1587 + }, + { + "epoch": 0.13094747258184217, + "grad_norm": 2.0014092922210693, + "learning_rate": 4.361088211046991e-07, + "loss": 0.9362, + "mean_token_accuracy": 0.7221314311027527, + "num_tokens": 47989065.0, + "step": 1588 + }, + { + "epoch": 0.1310299332068937, + "grad_norm": 2.4434142112731934, + "learning_rate": 4.3638362187414123e-07, + "loss": 0.9591, + "mean_token_accuracy": 0.7114866971969604, + "num_tokens": 48010509.0, + "step": 1589 + }, + { + "epoch": 0.13111239383194526, + "grad_norm": 2.257378101348877, + "learning_rate": 4.366584226435834e-07, + "loss": 0.8865, + "mean_token_accuracy": 0.7342751026153564, + "num_tokens": 48039054.0, + "step": 1590 + }, + { + "epoch": 0.13119485445699677, + "grad_norm": 2.1219327449798584, + "learning_rate": 4.369332234130255e-07, + "loss": 0.8446, + "mean_token_accuracy": 0.7403661012649536, + "num_tokens": 48071320.0, + "step": 1591 + }, + { + "epoch": 0.13127731508204832, + "grad_norm": 2.2403464317321777, + "learning_rate": 4.372080241824677e-07, + "loss": 0.8926, + "mean_token_accuracy": 0.7212316989898682, + "num_tokens": 48096818.0, + "step": 1592 + }, + { + "epoch": 0.13135977570709986, + "grad_norm": 2.5986886024475098, + "learning_rate": 4.3748282495190986e-07, + "loss": 0.8744, + "mean_token_accuracy": 0.7222633957862854, + "num_tokens": 48116840.0, + "step": 1593 + }, + { + "epoch": 0.1314422363321514, + "grad_norm": 2.0896780490875244, + "learning_rate": 4.37757625721352e-07, + "loss": 0.9451, + "mean_token_accuracy": 0.7129908800125122, + "num_tokens": 48148114.0, + "step": 1594 + }, + { + "epoch": 0.13152469695720292, + "grad_norm": 2.185074806213379, + "learning_rate": 4.3803242649079415e-07, + "loss": 0.9993, + "mean_token_accuracy": 0.7019674777984619, + "num_tokens": 48178807.0, + "step": 1595 + }, + { + "epoch": 0.13160715758225447, + "grad_norm": 1.8104478120803833, + "learning_rate": 4.383072272602363e-07, + "loss": 0.815, + "mean_token_accuracy": 0.7486482262611389, + "num_tokens": 48214358.0, + "step": 1596 + }, + { + "epoch": 0.13168961820730601, + "grad_norm": 2.2753894329071045, + "learning_rate": 4.385820280296785e-07, + "loss": 0.9228, + "mean_token_accuracy": 0.7261634469032288, + "num_tokens": 48238162.0, + "step": 1597 + }, + { + "epoch": 0.13177207883235756, + "grad_norm": 2.3725221157073975, + "learning_rate": 4.388568287991206e-07, + "loss": 0.9284, + "mean_token_accuracy": 0.7151937484741211, + "num_tokens": 48263929.0, + "step": 1598 + }, + { + "epoch": 0.13185453945740908, + "grad_norm": 1.9361261129379272, + "learning_rate": 4.3913162956856273e-07, + "loss": 0.9214, + "mean_token_accuracy": 0.7231380343437195, + "num_tokens": 48300262.0, + "step": 1599 + }, + { + "epoch": 0.13193700008246062, + "grad_norm": 2.0961174964904785, + "learning_rate": 4.3940643033800496e-07, + "loss": 0.9494, + "mean_token_accuracy": 0.7187715768814087, + "num_tokens": 48329671.0, + "step": 1600 + }, + { + "epoch": 0.13201946070751217, + "grad_norm": 2.0125038623809814, + "learning_rate": 4.396812311074471e-07, + "loss": 0.7181, + "mean_token_accuracy": 0.7745144367218018, + "num_tokens": 48357837.0, + "step": 1601 + }, + { + "epoch": 0.1321019213325637, + "grad_norm": 2.0974810123443604, + "learning_rate": 4.3995603187688925e-07, + "loss": 0.9027, + "mean_token_accuracy": 0.7330954074859619, + "num_tokens": 48386609.0, + "step": 1602 + }, + { + "epoch": 0.13218438195761523, + "grad_norm": 2.0580179691314697, + "learning_rate": 4.4023083264633137e-07, + "loss": 0.9256, + "mean_token_accuracy": 0.7164292335510254, + "num_tokens": 48415188.0, + "step": 1603 + }, + { + "epoch": 0.13226684258266677, + "grad_norm": 2.0528435707092285, + "learning_rate": 4.405056334157736e-07, + "loss": 0.7579, + "mean_token_accuracy": 0.7648659944534302, + "num_tokens": 48441005.0, + "step": 1604 + }, + { + "epoch": 0.13234930320771832, + "grad_norm": 2.1568779945373535, + "learning_rate": 4.407804341852157e-07, + "loss": 0.831, + "mean_token_accuracy": 0.733048677444458, + "num_tokens": 48470358.0, + "step": 1605 + }, + { + "epoch": 0.13243176383276986, + "grad_norm": 2.140446186065674, + "learning_rate": 4.4105523495465783e-07, + "loss": 0.8396, + "mean_token_accuracy": 0.7420947551727295, + "num_tokens": 48501356.0, + "step": 1606 + }, + { + "epoch": 0.13251422445782138, + "grad_norm": 1.9136215448379517, + "learning_rate": 4.413300357241e-07, + "loss": 0.8162, + "mean_token_accuracy": 0.7488313913345337, + "num_tokens": 48533990.0, + "step": 1607 + }, + { + "epoch": 0.13259668508287292, + "grad_norm": 2.1324379444122314, + "learning_rate": 4.4160483649354217e-07, + "loss": 0.93, + "mean_token_accuracy": 0.7145793437957764, + "num_tokens": 48561704.0, + "step": 1608 + }, + { + "epoch": 0.13267914570792447, + "grad_norm": 2.1826303005218506, + "learning_rate": 4.4187963726298434e-07, + "loss": 0.9826, + "mean_token_accuracy": 0.706373929977417, + "num_tokens": 48588867.0, + "step": 1609 + }, + { + "epoch": 0.132761606332976, + "grad_norm": 2.0558738708496094, + "learning_rate": 4.4215443803242646e-07, + "loss": 0.8543, + "mean_token_accuracy": 0.7304118871688843, + "num_tokens": 48620029.0, + "step": 1610 + }, + { + "epoch": 0.13284406695802753, + "grad_norm": 2.2329444885253906, + "learning_rate": 4.4242923880186863e-07, + "loss": 0.9081, + "mean_token_accuracy": 0.7326167821884155, + "num_tokens": 48646140.0, + "step": 1611 + }, + { + "epoch": 0.13292652758307907, + "grad_norm": 2.1419551372528076, + "learning_rate": 4.427040395713108e-07, + "loss": 0.9151, + "mean_token_accuracy": 0.729802668094635, + "num_tokens": 48678122.0, + "step": 1612 + }, + { + "epoch": 0.13300898820813062, + "grad_norm": 2.267876148223877, + "learning_rate": 4.429788403407529e-07, + "loss": 0.8978, + "mean_token_accuracy": 0.7115591168403625, + "num_tokens": 48704165.0, + "step": 1613 + }, + { + "epoch": 0.13309144883318216, + "grad_norm": 2.397486448287964, + "learning_rate": 4.432536411101951e-07, + "loss": 1.1278, + "mean_token_accuracy": 0.6688622832298279, + "num_tokens": 48732681.0, + "step": 1614 + }, + { + "epoch": 0.13317390945823368, + "grad_norm": 2.0317554473876953, + "learning_rate": 4.435284418796372e-07, + "loss": 0.987, + "mean_token_accuracy": 0.6940829753875732, + "num_tokens": 48767281.0, + "step": 1615 + }, + { + "epoch": 0.13325637008328522, + "grad_norm": 2.2497246265411377, + "learning_rate": 4.4380324264907944e-07, + "loss": 0.7861, + "mean_token_accuracy": 0.7500901222229004, + "num_tokens": 48791487.0, + "step": 1616 + }, + { + "epoch": 0.13333883070833677, + "grad_norm": 2.222471237182617, + "learning_rate": 4.4407804341852156e-07, + "loss": 1.0411, + "mean_token_accuracy": 0.6876806020736694, + "num_tokens": 48818688.0, + "step": 1617 + }, + { + "epoch": 0.13342129133338831, + "grad_norm": 2.152998447418213, + "learning_rate": 4.4435284418796373e-07, + "loss": 0.9383, + "mean_token_accuracy": 0.7131169438362122, + "num_tokens": 48847089.0, + "step": 1618 + }, + { + "epoch": 0.13350375195843983, + "grad_norm": 1.9534075260162354, + "learning_rate": 4.4462764495740585e-07, + "loss": 0.9308, + "mean_token_accuracy": 0.7110995054244995, + "num_tokens": 48882427.0, + "step": 1619 + }, + { + "epoch": 0.13358621258349138, + "grad_norm": 2.0064899921417236, + "learning_rate": 4.44902445726848e-07, + "loss": 0.8521, + "mean_token_accuracy": 0.7415265440940857, + "num_tokens": 48912726.0, + "step": 1620 + }, + { + "epoch": 0.13366867320854292, + "grad_norm": 1.959647297859192, + "learning_rate": 4.451772464962902e-07, + "loss": 0.7868, + "mean_token_accuracy": 0.7581895589828491, + "num_tokens": 48943249.0, + "step": 1621 + }, + { + "epoch": 0.13375113383359447, + "grad_norm": 2.258348226547241, + "learning_rate": 4.454520472657323e-07, + "loss": 0.9287, + "mean_token_accuracy": 0.7191278338432312, + "num_tokens": 48969243.0, + "step": 1622 + }, + { + "epoch": 0.133833594458646, + "grad_norm": 1.7916733026504517, + "learning_rate": 4.457268480351745e-07, + "loss": 0.9022, + "mean_token_accuracy": 0.7372636198997498, + "num_tokens": 49009786.0, + "step": 1623 + }, + { + "epoch": 0.13391605508369753, + "grad_norm": 1.9419676065444946, + "learning_rate": 4.4600164880461665e-07, + "loss": 0.8687, + "mean_token_accuracy": 0.7251266241073608, + "num_tokens": 49041287.0, + "step": 1624 + }, + { + "epoch": 0.13399851570874907, + "grad_norm": 2.2571940422058105, + "learning_rate": 4.462764495740588e-07, + "loss": 0.8365, + "mean_token_accuracy": 0.7329846620559692, + "num_tokens": 49066499.0, + "step": 1625 + }, + { + "epoch": 0.13408097633380062, + "grad_norm": 2.1222918033599854, + "learning_rate": 4.4655125034350094e-07, + "loss": 0.9693, + "mean_token_accuracy": 0.7114278674125671, + "num_tokens": 49096269.0, + "step": 1626 + }, + { + "epoch": 0.13416343695885216, + "grad_norm": 1.8655835390090942, + "learning_rate": 4.4682605111294306e-07, + "loss": 0.8773, + "mean_token_accuracy": 0.7294366359710693, + "num_tokens": 49130393.0, + "step": 1627 + }, + { + "epoch": 0.13424589758390368, + "grad_norm": 2.0259506702423096, + "learning_rate": 4.471008518823853e-07, + "loss": 0.7748, + "mean_token_accuracy": 0.7635329961776733, + "num_tokens": 49163449.0, + "step": 1628 + }, + { + "epoch": 0.13432835820895522, + "grad_norm": 2.147282123565674, + "learning_rate": 4.473756526518274e-07, + "loss": 0.8707, + "mean_token_accuracy": 0.724371075630188, + "num_tokens": 49190294.0, + "step": 1629 + }, + { + "epoch": 0.13441081883400677, + "grad_norm": 2.4691905975341797, + "learning_rate": 4.476504534212696e-07, + "loss": 0.9194, + "mean_token_accuracy": 0.7189136743545532, + "num_tokens": 49214609.0, + "step": 1630 + }, + { + "epoch": 0.1344932794590583, + "grad_norm": 1.9391493797302246, + "learning_rate": 4.479252541907117e-07, + "loss": 0.9652, + "mean_token_accuracy": 0.7115057706832886, + "num_tokens": 49252083.0, + "step": 1631 + }, + { + "epoch": 0.13457574008410983, + "grad_norm": 2.1624159812927246, + "learning_rate": 4.482000549601539e-07, + "loss": 0.782, + "mean_token_accuracy": 0.7519997954368591, + "num_tokens": 49281720.0, + "step": 1632 + }, + { + "epoch": 0.13465820070916137, + "grad_norm": 2.130887508392334, + "learning_rate": 4.4847485572959604e-07, + "loss": 0.8265, + "mean_token_accuracy": 0.7521308660507202, + "num_tokens": 49307287.0, + "step": 1633 + }, + { + "epoch": 0.13474066133421292, + "grad_norm": 2.0371463298797607, + "learning_rate": 4.4874965649903816e-07, + "loss": 0.9072, + "mean_token_accuracy": 0.7286190986633301, + "num_tokens": 49336988.0, + "step": 1634 + }, + { + "epoch": 0.13482312195926446, + "grad_norm": 2.1578829288482666, + "learning_rate": 4.4902445726848033e-07, + "loss": 0.874, + "mean_token_accuracy": 0.7357889413833618, + "num_tokens": 49364680.0, + "step": 1635 + }, + { + "epoch": 0.13490558258431598, + "grad_norm": 2.006329298019409, + "learning_rate": 4.492992580379225e-07, + "loss": 0.93, + "mean_token_accuracy": 0.7172128558158875, + "num_tokens": 49397654.0, + "step": 1636 + }, + { + "epoch": 0.13498804320936753, + "grad_norm": 2.1941099166870117, + "learning_rate": 4.4957405880736467e-07, + "loss": 0.7997, + "mean_token_accuracy": 0.7564677000045776, + "num_tokens": 49424973.0, + "step": 1637 + }, + { + "epoch": 0.13507050383441907, + "grad_norm": 2.0912535190582275, + "learning_rate": 4.498488595768068e-07, + "loss": 0.8584, + "mean_token_accuracy": 0.7392517924308777, + "num_tokens": 49453775.0, + "step": 1638 + }, + { + "epoch": 0.13515296445947061, + "grad_norm": 2.1877024173736572, + "learning_rate": 4.501236603462489e-07, + "loss": 0.8326, + "mean_token_accuracy": 0.7429982423782349, + "num_tokens": 49480703.0, + "step": 1639 + }, + { + "epoch": 0.13523542508452213, + "grad_norm": 1.9282951354980469, + "learning_rate": 4.5039846111569113e-07, + "loss": 0.8751, + "mean_token_accuracy": 0.725446343421936, + "num_tokens": 49516372.0, + "step": 1640 + }, + { + "epoch": 0.13531788570957368, + "grad_norm": 2.0827736854553223, + "learning_rate": 4.5067326188513325e-07, + "loss": 0.9687, + "mean_token_accuracy": 0.7080146074295044, + "num_tokens": 49545052.0, + "step": 1641 + }, + { + "epoch": 0.13540034633462522, + "grad_norm": 2.0245532989501953, + "learning_rate": 4.509480626545754e-07, + "loss": 0.7845, + "mean_token_accuracy": 0.7621080875396729, + "num_tokens": 49574379.0, + "step": 1642 + }, + { + "epoch": 0.13548280695967677, + "grad_norm": 2.196094036102295, + "learning_rate": 4.5122286342401754e-07, + "loss": 1.0572, + "mean_token_accuracy": 0.6897295713424683, + "num_tokens": 49604076.0, + "step": 1643 + }, + { + "epoch": 0.13556526758472828, + "grad_norm": 2.186030864715576, + "learning_rate": 4.5149766419345977e-07, + "loss": 1.0027, + "mean_token_accuracy": 0.6973799467086792, + "num_tokens": 49632615.0, + "step": 1644 + }, + { + "epoch": 0.13564772820977983, + "grad_norm": 1.9106826782226562, + "learning_rate": 4.517724649629019e-07, + "loss": 0.7492, + "mean_token_accuracy": 0.7586303949356079, + "num_tokens": 49663797.0, + "step": 1645 + }, + { + "epoch": 0.13573018883483137, + "grad_norm": 2.3819661140441895, + "learning_rate": 4.52047265732344e-07, + "loss": 0.9192, + "mean_token_accuracy": 0.7261579036712646, + "num_tokens": 49687961.0, + "step": 1646 + }, + { + "epoch": 0.13581264945988292, + "grad_norm": 2.17146372795105, + "learning_rate": 4.523220665017862e-07, + "loss": 1.0307, + "mean_token_accuracy": 0.6896770000457764, + "num_tokens": 49719583.0, + "step": 1647 + }, + { + "epoch": 0.13589511008493443, + "grad_norm": 1.8872759342193604, + "learning_rate": 4.5259686727122835e-07, + "loss": 0.8061, + "mean_token_accuracy": 0.7508621215820312, + "num_tokens": 49754444.0, + "step": 1648 + }, + { + "epoch": 0.13597757070998598, + "grad_norm": 2.1311001777648926, + "learning_rate": 4.528716680406705e-07, + "loss": 0.7869, + "mean_token_accuracy": 0.7551379799842834, + "num_tokens": 49784715.0, + "step": 1649 + }, + { + "epoch": 0.13606003133503752, + "grad_norm": 2.1756417751312256, + "learning_rate": 4.5314646881011264e-07, + "loss": 0.933, + "mean_token_accuracy": 0.7208852171897888, + "num_tokens": 49815950.0, + "step": 1650 + }, + { + "epoch": 0.13614249196008907, + "grad_norm": 2.077937602996826, + "learning_rate": 4.534212695795548e-07, + "loss": 1.0023, + "mean_token_accuracy": 0.7140227556228638, + "num_tokens": 49844312.0, + "step": 1651 + }, + { + "epoch": 0.13622495258514059, + "grad_norm": 2.033888816833496, + "learning_rate": 4.53696070348997e-07, + "loss": 0.7492, + "mean_token_accuracy": 0.7649842500686646, + "num_tokens": 49871094.0, + "step": 1652 + }, + { + "epoch": 0.13630741321019213, + "grad_norm": 1.9816217422485352, + "learning_rate": 4.539708711184391e-07, + "loss": 0.7024, + "mean_token_accuracy": 0.7683992385864258, + "num_tokens": 49904149.0, + "step": 1653 + }, + { + "epoch": 0.13638987383524367, + "grad_norm": 1.9816360473632812, + "learning_rate": 4.5424567188788127e-07, + "loss": 0.8403, + "mean_token_accuracy": 0.7477820515632629, + "num_tokens": 49935050.0, + "step": 1654 + }, + { + "epoch": 0.13647233446029522, + "grad_norm": 2.070646286010742, + "learning_rate": 4.545204726573234e-07, + "loss": 0.9864, + "mean_token_accuracy": 0.7069747447967529, + "num_tokens": 49965947.0, + "step": 1655 + }, + { + "epoch": 0.13655479508534674, + "grad_norm": 1.9730061292648315, + "learning_rate": 4.5479527342676556e-07, + "loss": 0.8332, + "mean_token_accuracy": 0.7449889183044434, + "num_tokens": 49998201.0, + "step": 1656 + }, + { + "epoch": 0.13663725571039828, + "grad_norm": 2.062145233154297, + "learning_rate": 4.5507007419620773e-07, + "loss": 0.8897, + "mean_token_accuracy": 0.7267539501190186, + "num_tokens": 50028186.0, + "step": 1657 + }, + { + "epoch": 0.13671971633544983, + "grad_norm": 2.1159703731536865, + "learning_rate": 4.553448749656499e-07, + "loss": 0.6583, + "mean_token_accuracy": 0.787714958190918, + "num_tokens": 50053663.0, + "step": 1658 + }, + { + "epoch": 0.13680217696050137, + "grad_norm": 2.3487205505371094, + "learning_rate": 4.55619675735092e-07, + "loss": 1.0229, + "mean_token_accuracy": 0.7064402103424072, + "num_tokens": 50079471.0, + "step": 1659 + }, + { + "epoch": 0.1368846375855529, + "grad_norm": 2.0796775817871094, + "learning_rate": 4.5589447650453414e-07, + "loss": 0.9663, + "mean_token_accuracy": 0.7168182134628296, + "num_tokens": 50111178.0, + "step": 1660 + }, + { + "epoch": 0.13696709821060443, + "grad_norm": 2.2452709674835205, + "learning_rate": 4.5616927727397637e-07, + "loss": 0.9492, + "mean_token_accuracy": 0.7098708152770996, + "num_tokens": 50137719.0, + "step": 1661 + }, + { + "epoch": 0.13704955883565598, + "grad_norm": 2.5023834705352783, + "learning_rate": 4.564440780434185e-07, + "loss": 0.9169, + "mean_token_accuracy": 0.7187298536300659, + "num_tokens": 50160524.0, + "step": 1662 + }, + { + "epoch": 0.13713201946070752, + "grad_norm": 1.9373270273208618, + "learning_rate": 4.5671887881286066e-07, + "loss": 0.8976, + "mean_token_accuracy": 0.738940417766571, + "num_tokens": 50194228.0, + "step": 1663 + }, + { + "epoch": 0.13721448008575904, + "grad_norm": 2.004133701324463, + "learning_rate": 4.569936795823028e-07, + "loss": 0.6582, + "mean_token_accuracy": 0.7924010753631592, + "num_tokens": 50221880.0, + "step": 1664 + }, + { + "epoch": 0.13729694071081058, + "grad_norm": 2.3169310092926025, + "learning_rate": 4.57268480351745e-07, + "loss": 0.9854, + "mean_token_accuracy": 0.7009056210517883, + "num_tokens": 50249067.0, + "step": 1665 + }, + { + "epoch": 0.13737940133586213, + "grad_norm": 2.4220449924468994, + "learning_rate": 4.575432811211871e-07, + "loss": 0.9796, + "mean_token_accuracy": 0.7184774279594421, + "num_tokens": 50273205.0, + "step": 1666 + }, + { + "epoch": 0.13746186196091367, + "grad_norm": 2.236206531524658, + "learning_rate": 4.5781808189062924e-07, + "loss": 0.9295, + "mean_token_accuracy": 0.7142399549484253, + "num_tokens": 50300414.0, + "step": 1667 + }, + { + "epoch": 0.1375443225859652, + "grad_norm": 1.9871718883514404, + "learning_rate": 4.580928826600714e-07, + "loss": 0.8598, + "mean_token_accuracy": 0.7356272339820862, + "num_tokens": 50331084.0, + "step": 1668 + }, + { + "epoch": 0.13762678321101673, + "grad_norm": 1.8960299491882324, + "learning_rate": 4.583676834295136e-07, + "loss": 0.9333, + "mean_token_accuracy": 0.7156428694725037, + "num_tokens": 50369030.0, + "step": 1669 + }, + { + "epoch": 0.13770924383606828, + "grad_norm": 2.405473470687866, + "learning_rate": 4.5864248419895575e-07, + "loss": 1.0281, + "mean_token_accuracy": 0.7005497813224792, + "num_tokens": 50396136.0, + "step": 1670 + }, + { + "epoch": 0.13779170446111982, + "grad_norm": 2.2954587936401367, + "learning_rate": 4.5891728496839787e-07, + "loss": 0.8166, + "mean_token_accuracy": 0.7528100609779358, + "num_tokens": 50421696.0, + "step": 1671 + }, + { + "epoch": 0.13787416508617134, + "grad_norm": 1.9811409711837769, + "learning_rate": 4.5919208573784004e-07, + "loss": 0.8509, + "mean_token_accuracy": 0.7369825839996338, + "num_tokens": 50453081.0, + "step": 1672 + }, + { + "epoch": 0.13795662571122289, + "grad_norm": 1.9876842498779297, + "learning_rate": 4.594668865072822e-07, + "loss": 0.8354, + "mean_token_accuracy": 0.7459633350372314, + "num_tokens": 50486085.0, + "step": 1673 + }, + { + "epoch": 0.13803908633627443, + "grad_norm": 1.9752154350280762, + "learning_rate": 4.5974168727672433e-07, + "loss": 0.8099, + "mean_token_accuracy": 0.7512794137001038, + "num_tokens": 50516627.0, + "step": 1674 + }, + { + "epoch": 0.13812154696132597, + "grad_norm": 2.0856525897979736, + "learning_rate": 4.600164880461665e-07, + "loss": 0.8369, + "mean_token_accuracy": 0.7427292466163635, + "num_tokens": 50544983.0, + "step": 1675 + }, + { + "epoch": 0.1382040075863775, + "grad_norm": 2.1062967777252197, + "learning_rate": 4.602912888156086e-07, + "loss": 0.7955, + "mean_token_accuracy": 0.7629863023757935, + "num_tokens": 50580807.0, + "step": 1676 + }, + { + "epoch": 0.13828646821142904, + "grad_norm": 2.3892104625701904, + "learning_rate": 4.6056608958505085e-07, + "loss": 0.9637, + "mean_token_accuracy": 0.7157956957817078, + "num_tokens": 50605037.0, + "step": 1677 + }, + { + "epoch": 0.13836892883648058, + "grad_norm": 1.9878910779953003, + "learning_rate": 4.6084089035449296e-07, + "loss": 0.69, + "mean_token_accuracy": 0.7742583155632019, + "num_tokens": 50632852.0, + "step": 1678 + }, + { + "epoch": 0.13845138946153213, + "grad_norm": 1.8206900358200073, + "learning_rate": 4.6111569112393514e-07, + "loss": 0.8784, + "mean_token_accuracy": 0.730437159538269, + "num_tokens": 50671352.0, + "step": 1679 + }, + { + "epoch": 0.13853385008658364, + "grad_norm": 1.9193482398986816, + "learning_rate": 4.6139049189337726e-07, + "loss": 0.8294, + "mean_token_accuracy": 0.7411342263221741, + "num_tokens": 50704255.0, + "step": 1680 + }, + { + "epoch": 0.1386163107116352, + "grad_norm": 2.2308919429779053, + "learning_rate": 4.6166529266281943e-07, + "loss": 0.889, + "mean_token_accuracy": 0.7314280271530151, + "num_tokens": 50731751.0, + "step": 1681 + }, + { + "epoch": 0.13869877133668673, + "grad_norm": 2.2965433597564697, + "learning_rate": 4.619400934322616e-07, + "loss": 0.7386, + "mean_token_accuracy": 0.7560648918151855, + "num_tokens": 50753468.0, + "step": 1682 + }, + { + "epoch": 0.13878123196173828, + "grad_norm": 1.972327709197998, + "learning_rate": 4.622148942017037e-07, + "loss": 0.8017, + "mean_token_accuracy": 0.7457852363586426, + "num_tokens": 50784917.0, + "step": 1683 + }, + { + "epoch": 0.1388636925867898, + "grad_norm": 1.8525594472885132, + "learning_rate": 4.624896949711459e-07, + "loss": 0.678, + "mean_token_accuracy": 0.7850761413574219, + "num_tokens": 50819049.0, + "step": 1684 + }, + { + "epoch": 0.13894615321184134, + "grad_norm": 2.05877423286438, + "learning_rate": 4.6276449574058806e-07, + "loss": 0.9063, + "mean_token_accuracy": 0.7086782455444336, + "num_tokens": 50847173.0, + "step": 1685 + }, + { + "epoch": 0.13902861383689288, + "grad_norm": 1.845391869544983, + "learning_rate": 4.6303929651003023e-07, + "loss": 0.7904, + "mean_token_accuracy": 0.7502480149269104, + "num_tokens": 50881579.0, + "step": 1686 + }, + { + "epoch": 0.13911107446194443, + "grad_norm": 2.151362657546997, + "learning_rate": 4.6331409727947235e-07, + "loss": 0.8444, + "mean_token_accuracy": 0.7471015453338623, + "num_tokens": 50908531.0, + "step": 1687 + }, + { + "epoch": 0.13919353508699597, + "grad_norm": 2.0041451454162598, + "learning_rate": 4.6358889804891447e-07, + "loss": 0.9776, + "mean_token_accuracy": 0.7048242092132568, + "num_tokens": 50942054.0, + "step": 1688 + }, + { + "epoch": 0.1392759957120475, + "grad_norm": 1.7936811447143555, + "learning_rate": 4.638636988183567e-07, + "loss": 0.7518, + "mean_token_accuracy": 0.7754173278808594, + "num_tokens": 50977014.0, + "step": 1689 + }, + { + "epoch": 0.13935845633709903, + "grad_norm": 1.8799525499343872, + "learning_rate": 4.641384995877988e-07, + "loss": 0.9013, + "mean_token_accuracy": 0.7357406616210938, + "num_tokens": 51014267.0, + "step": 1690 + }, + { + "epoch": 0.13944091696215058, + "grad_norm": 1.9140936136245728, + "learning_rate": 4.64413300357241e-07, + "loss": 0.7619, + "mean_token_accuracy": 0.7656726837158203, + "num_tokens": 51043644.0, + "step": 1691 + }, + { + "epoch": 0.13952337758720212, + "grad_norm": 2.024651288986206, + "learning_rate": 4.646881011266831e-07, + "loss": 0.8699, + "mean_token_accuracy": 0.7337296009063721, + "num_tokens": 51075159.0, + "step": 1692 + }, + { + "epoch": 0.13960583821225364, + "grad_norm": 1.8355345726013184, + "learning_rate": 4.6496290189612533e-07, + "loss": 0.7143, + "mean_token_accuracy": 0.774918794631958, + "num_tokens": 51109043.0, + "step": 1693 + }, + { + "epoch": 0.13968829883730519, + "grad_norm": 1.9900274276733398, + "learning_rate": 4.6523770266556745e-07, + "loss": 0.7663, + "mean_token_accuracy": 0.7582066059112549, + "num_tokens": 51143793.0, + "step": 1694 + }, + { + "epoch": 0.13977075946235673, + "grad_norm": 2.544347047805786, + "learning_rate": 4.6551250343500956e-07, + "loss": 0.9882, + "mean_token_accuracy": 0.6967458724975586, + "num_tokens": 51165547.0, + "step": 1695 + }, + { + "epoch": 0.13985322008740828, + "grad_norm": 2.0098817348480225, + "learning_rate": 4.6578730420445174e-07, + "loss": 0.8358, + "mean_token_accuracy": 0.7431207895278931, + "num_tokens": 51197547.0, + "step": 1696 + }, + { + "epoch": 0.1399356807124598, + "grad_norm": 2.3463451862335205, + "learning_rate": 4.660621049738939e-07, + "loss": 0.8311, + "mean_token_accuracy": 0.7437360286712646, + "num_tokens": 51220565.0, + "step": 1697 + }, + { + "epoch": 0.14001814133751134, + "grad_norm": 2.1957223415374756, + "learning_rate": 4.663369057433361e-07, + "loss": 0.9135, + "mean_token_accuracy": 0.7256888151168823, + "num_tokens": 51246949.0, + "step": 1698 + }, + { + "epoch": 0.14010060196256288, + "grad_norm": 2.202453136444092, + "learning_rate": 4.666117065127782e-07, + "loss": 1.0312, + "mean_token_accuracy": 0.6862379908561707, + "num_tokens": 51277185.0, + "step": 1699 + }, + { + "epoch": 0.14018306258761443, + "grad_norm": 2.111021041870117, + "learning_rate": 4.6688650728222037e-07, + "loss": 0.8077, + "mean_token_accuracy": 0.7450732588768005, + "num_tokens": 51306549.0, + "step": 1700 + }, + { + "epoch": 0.14026552321266594, + "grad_norm": 1.9432272911071777, + "learning_rate": 4.6716130805166254e-07, + "loss": 0.9655, + "mean_token_accuracy": 0.7171384692192078, + "num_tokens": 51341154.0, + "step": 1701 + }, + { + "epoch": 0.1403479838377175, + "grad_norm": 1.8533519506454468, + "learning_rate": 4.6743610882110466e-07, + "loss": 0.8361, + "mean_token_accuracy": 0.7457166910171509, + "num_tokens": 51378166.0, + "step": 1702 + }, + { + "epoch": 0.14043044446276903, + "grad_norm": 2.235452651977539, + "learning_rate": 4.6771090959054683e-07, + "loss": 1.0111, + "mean_token_accuracy": 0.6981621980667114, + "num_tokens": 51405059.0, + "step": 1703 + }, + { + "epoch": 0.14051290508782058, + "grad_norm": 2.3346197605133057, + "learning_rate": 4.6798571035998895e-07, + "loss": 0.8149, + "mean_token_accuracy": 0.7445928454399109, + "num_tokens": 51430566.0, + "step": 1704 + }, + { + "epoch": 0.1405953657128721, + "grad_norm": 2.0474588871002197, + "learning_rate": 4.682605111294312e-07, + "loss": 0.7652, + "mean_token_accuracy": 0.7683714628219604, + "num_tokens": 51460345.0, + "step": 1705 + }, + { + "epoch": 0.14067782633792364, + "grad_norm": 2.2726926803588867, + "learning_rate": 4.685353118988733e-07, + "loss": 0.6837, + "mean_token_accuracy": 0.7779980897903442, + "num_tokens": 51483001.0, + "step": 1706 + }, + { + "epoch": 0.14076028696297518, + "grad_norm": 1.9682821035385132, + "learning_rate": 4.6881011266831546e-07, + "loss": 0.7999, + "mean_token_accuracy": 0.7476588487625122, + "num_tokens": 51515657.0, + "step": 1707 + }, + { + "epoch": 0.14084274758802673, + "grad_norm": 1.9849177598953247, + "learning_rate": 4.690849134377576e-07, + "loss": 0.8582, + "mean_token_accuracy": 0.7398413419723511, + "num_tokens": 51547066.0, + "step": 1708 + }, + { + "epoch": 0.14092520821307825, + "grad_norm": 2.1883976459503174, + "learning_rate": 4.6935971420719975e-07, + "loss": 0.9413, + "mean_token_accuracy": 0.7120356559753418, + "num_tokens": 51576782.0, + "step": 1709 + }, + { + "epoch": 0.1410076688381298, + "grad_norm": 3.2107884883880615, + "learning_rate": 4.696345149766419e-07, + "loss": 0.7763, + "mean_token_accuracy": 0.7620803117752075, + "num_tokens": 51603019.0, + "step": 1710 + }, + { + "epoch": 0.14109012946318134, + "grad_norm": 2.0088367462158203, + "learning_rate": 4.6990931574608404e-07, + "loss": 0.9415, + "mean_token_accuracy": 0.7155007719993591, + "num_tokens": 51636344.0, + "step": 1711 + }, + { + "epoch": 0.14117259008823288, + "grad_norm": 1.7529559135437012, + "learning_rate": 4.701841165155262e-07, + "loss": 1.0072, + "mean_token_accuracy": 0.7009587287902832, + "num_tokens": 51677358.0, + "step": 1712 + }, + { + "epoch": 0.1412550507132844, + "grad_norm": 2.119814157485962, + "learning_rate": 4.704589172849684e-07, + "loss": 0.7669, + "mean_token_accuracy": 0.7604952454566956, + "num_tokens": 51705593.0, + "step": 1713 + }, + { + "epoch": 0.14133751133833594, + "grad_norm": 2.25174880027771, + "learning_rate": 4.7073371805441056e-07, + "loss": 0.8561, + "mean_token_accuracy": 0.7327386140823364, + "num_tokens": 51730477.0, + "step": 1714 + }, + { + "epoch": 0.1414199719633875, + "grad_norm": 2.0146827697753906, + "learning_rate": 4.710085188238527e-07, + "loss": 0.7163, + "mean_token_accuracy": 0.7669206857681274, + "num_tokens": 51757811.0, + "step": 1715 + }, + { + "epoch": 0.14150243258843903, + "grad_norm": 1.944381833076477, + "learning_rate": 4.712833195932948e-07, + "loss": 0.7418, + "mean_token_accuracy": 0.7620489001274109, + "num_tokens": 51788237.0, + "step": 1716 + }, + { + "epoch": 0.14158489321349055, + "grad_norm": 2.143124580383301, + "learning_rate": 4.71558120362737e-07, + "loss": 0.8257, + "mean_token_accuracy": 0.7385540008544922, + "num_tokens": 51817286.0, + "step": 1717 + }, + { + "epoch": 0.1416673538385421, + "grad_norm": 1.957619547843933, + "learning_rate": 4.7183292113217914e-07, + "loss": 0.8063, + "mean_token_accuracy": 0.7534171938896179, + "num_tokens": 51849004.0, + "step": 1718 + }, + { + "epoch": 0.14174981446359364, + "grad_norm": 2.3467156887054443, + "learning_rate": 4.721077219016213e-07, + "loss": 0.942, + "mean_token_accuracy": 0.7116962671279907, + "num_tokens": 51876847.0, + "step": 1719 + }, + { + "epoch": 0.14183227508864518, + "grad_norm": 1.9438271522521973, + "learning_rate": 4.7238252267106343e-07, + "loss": 0.7865, + "mean_token_accuracy": 0.7668686509132385, + "num_tokens": 51909017.0, + "step": 1720 + }, + { + "epoch": 0.1419147357136967, + "grad_norm": 2.1584348678588867, + "learning_rate": 4.7265732344050565e-07, + "loss": 0.9837, + "mean_token_accuracy": 0.702038586139679, + "num_tokens": 51936953.0, + "step": 1721 + }, + { + "epoch": 0.14199719633874824, + "grad_norm": 2.294525623321533, + "learning_rate": 4.7293212420994777e-07, + "loss": 0.8855, + "mean_token_accuracy": 0.7392202615737915, + "num_tokens": 51962624.0, + "step": 1722 + }, + { + "epoch": 0.1420796569637998, + "grad_norm": 1.9847036600112915, + "learning_rate": 4.732069249793899e-07, + "loss": 0.9369, + "mean_token_accuracy": 0.7230961322784424, + "num_tokens": 51995490.0, + "step": 1723 + }, + { + "epoch": 0.14216211758885133, + "grad_norm": 2.2233357429504395, + "learning_rate": 4.7348172574883206e-07, + "loss": 0.9466, + "mean_token_accuracy": 0.7209491729736328, + "num_tokens": 52021661.0, + "step": 1724 + }, + { + "epoch": 0.14224457821390285, + "grad_norm": 2.0466699600219727, + "learning_rate": 4.7375652651827423e-07, + "loss": 0.8319, + "mean_token_accuracy": 0.7480276226997375, + "num_tokens": 52052328.0, + "step": 1725 + }, + { + "epoch": 0.1423270388389544, + "grad_norm": 2.0111083984375, + "learning_rate": 4.740313272877164e-07, + "loss": 0.9732, + "mean_token_accuracy": 0.7088425755500793, + "num_tokens": 52084629.0, + "step": 1726 + }, + { + "epoch": 0.14240949946400594, + "grad_norm": 2.0065462589263916, + "learning_rate": 4.743061280571585e-07, + "loss": 0.966, + "mean_token_accuracy": 0.7084306478500366, + "num_tokens": 52117772.0, + "step": 1727 + }, + { + "epoch": 0.14249196008905748, + "grad_norm": 2.066622734069824, + "learning_rate": 4.7458092882660064e-07, + "loss": 0.8033, + "mean_token_accuracy": 0.7489135265350342, + "num_tokens": 52145624.0, + "step": 1728 + }, + { + "epoch": 0.142574420714109, + "grad_norm": 2.0997769832611084, + "learning_rate": 4.7485572959604287e-07, + "loss": 0.8558, + "mean_token_accuracy": 0.7411609888076782, + "num_tokens": 52178271.0, + "step": 1729 + }, + { + "epoch": 0.14265688133916055, + "grad_norm": 1.8715949058532715, + "learning_rate": 4.75130530365485e-07, + "loss": 0.9651, + "mean_token_accuracy": 0.7118431329727173, + "num_tokens": 52213457.0, + "step": 1730 + }, + { + "epoch": 0.1427393419642121, + "grad_norm": 2.1048054695129395, + "learning_rate": 4.7540533113492716e-07, + "loss": 0.8415, + "mean_token_accuracy": 0.7442131638526917, + "num_tokens": 52245835.0, + "step": 1731 + }, + { + "epoch": 0.14282180258926364, + "grad_norm": 2.2973408699035645, + "learning_rate": 4.756801319043693e-07, + "loss": 0.7506, + "mean_token_accuracy": 0.7618380784988403, + "num_tokens": 52269353.0, + "step": 1732 + }, + { + "epoch": 0.14290426321431515, + "grad_norm": 2.104304313659668, + "learning_rate": 4.759549326738115e-07, + "loss": 0.9458, + "mean_token_accuracy": 0.7146650552749634, + "num_tokens": 52298883.0, + "step": 1733 + }, + { + "epoch": 0.1429867238393667, + "grad_norm": 2.157771587371826, + "learning_rate": 4.762297334432536e-07, + "loss": 0.7586, + "mean_token_accuracy": 0.7568196654319763, + "num_tokens": 52323502.0, + "step": 1734 + }, + { + "epoch": 0.14306918446441824, + "grad_norm": 2.2524454593658447, + "learning_rate": 4.7650453421269574e-07, + "loss": 0.9392, + "mean_token_accuracy": 0.7186516523361206, + "num_tokens": 52350423.0, + "step": 1735 + }, + { + "epoch": 0.1431516450894698, + "grad_norm": 2.3754303455352783, + "learning_rate": 4.767793349821379e-07, + "loss": 0.7741, + "mean_token_accuracy": 0.7530540227890015, + "num_tokens": 52373686.0, + "step": 1736 + }, + { + "epoch": 0.1432341057145213, + "grad_norm": 2.120004653930664, + "learning_rate": 4.770541357515801e-07, + "loss": 0.9494, + "mean_token_accuracy": 0.7125406265258789, + "num_tokens": 52402072.0, + "step": 1737 + }, + { + "epoch": 0.14331656633957285, + "grad_norm": 1.972173810005188, + "learning_rate": 4.773289365210223e-07, + "loss": 0.8383, + "mean_token_accuracy": 0.7422926425933838, + "num_tokens": 52433896.0, + "step": 1738 + }, + { + "epoch": 0.1433990269646244, + "grad_norm": 1.943491816520691, + "learning_rate": 4.776037372904644e-07, + "loss": 0.8496, + "mean_token_accuracy": 0.7357045412063599, + "num_tokens": 52467932.0, + "step": 1739 + }, + { + "epoch": 0.14348148758967594, + "grad_norm": 1.9774001836776733, + "learning_rate": 4.778785380599065e-07, + "loss": 0.9578, + "mean_token_accuracy": 0.7129608988761902, + "num_tokens": 52501603.0, + "step": 1740 + }, + { + "epoch": 0.14356394821472745, + "grad_norm": 2.1179134845733643, + "learning_rate": 4.781533388293488e-07, + "loss": 0.8974, + "mean_token_accuracy": 0.72100430727005, + "num_tokens": 52530141.0, + "step": 1741 + }, + { + "epoch": 0.143646408839779, + "grad_norm": 2.4072823524475098, + "learning_rate": 4.784281395987908e-07, + "loss": 1.0894, + "mean_token_accuracy": 0.6798986196517944, + "num_tokens": 52555490.0, + "step": 1742 + }, + { + "epoch": 0.14372886946483054, + "grad_norm": 1.989848017692566, + "learning_rate": 4.78702940368233e-07, + "loss": 0.783, + "mean_token_accuracy": 0.7519559860229492, + "num_tokens": 52586275.0, + "step": 1743 + }, + { + "epoch": 0.1438113300898821, + "grad_norm": 1.9308134317398071, + "learning_rate": 4.789777411376752e-07, + "loss": 0.9521, + "mean_token_accuracy": 0.7147394418716431, + "num_tokens": 52620224.0, + "step": 1744 + }, + { + "epoch": 0.1438937907149336, + "grad_norm": 1.9106812477111816, + "learning_rate": 4.792525419071173e-07, + "loss": 0.8324, + "mean_token_accuracy": 0.7441853880882263, + "num_tokens": 52653721.0, + "step": 1745 + }, + { + "epoch": 0.14397625133998515, + "grad_norm": 2.0013551712036133, + "learning_rate": 4.795273426765595e-07, + "loss": 0.7941, + "mean_token_accuracy": 0.7504277229309082, + "num_tokens": 52681608.0, + "step": 1746 + }, + { + "epoch": 0.1440587119650367, + "grad_norm": 2.072406053543091, + "learning_rate": 4.798021434460016e-07, + "loss": 0.929, + "mean_token_accuracy": 0.7225048542022705, + "num_tokens": 52711568.0, + "step": 1747 + }, + { + "epoch": 0.14414117259008824, + "grad_norm": 1.9118096828460693, + "learning_rate": 4.800769442154438e-07, + "loss": 0.9233, + "mean_token_accuracy": 0.7240248918533325, + "num_tokens": 52745142.0, + "step": 1748 + }, + { + "epoch": 0.14422363321513976, + "grad_norm": 1.9938162565231323, + "learning_rate": 4.803517449848859e-07, + "loss": 0.8999, + "mean_token_accuracy": 0.7280124425888062, + "num_tokens": 52775553.0, + "step": 1749 + }, + { + "epoch": 0.1443060938401913, + "grad_norm": 1.9553031921386719, + "learning_rate": 4.806265457543281e-07, + "loss": 0.8739, + "mean_token_accuracy": 0.7308230400085449, + "num_tokens": 52807110.0, + "step": 1750 + }, + { + "epoch": 0.14438855446524285, + "grad_norm": 2.1959776878356934, + "learning_rate": 4.809013465237703e-07, + "loss": 1.0471, + "mean_token_accuracy": 0.6995769739151001, + "num_tokens": 52837116.0, + "step": 1751 + }, + { + "epoch": 0.1444710150902944, + "grad_norm": 2.0767831802368164, + "learning_rate": 4.811761472932123e-07, + "loss": 0.6767, + "mean_token_accuracy": 0.7774621844291687, + "num_tokens": 52865349.0, + "step": 1752 + }, + { + "epoch": 0.14455347571534594, + "grad_norm": 2.1969857215881348, + "learning_rate": 4.814509480626546e-07, + "loss": 0.9631, + "mean_token_accuracy": 0.6996597051620483, + "num_tokens": 52894042.0, + "step": 1753 + }, + { + "epoch": 0.14463593634039745, + "grad_norm": 2.2758333683013916, + "learning_rate": 4.817257488320967e-07, + "loss": 0.9537, + "mean_token_accuracy": 0.717389702796936, + "num_tokens": 52920669.0, + "step": 1754 + }, + { + "epoch": 0.144718396965449, + "grad_norm": 1.9260668754577637, + "learning_rate": 4.820005496015389e-07, + "loss": 0.9199, + "mean_token_accuracy": 0.720063328742981, + "num_tokens": 52955371.0, + "step": 1755 + }, + { + "epoch": 0.14480085759050054, + "grad_norm": 2.304940938949585, + "learning_rate": 4.82275350370981e-07, + "loss": 0.9877, + "mean_token_accuracy": 0.707699716091156, + "num_tokens": 52980219.0, + "step": 1756 + }, + { + "epoch": 0.1448833182155521, + "grad_norm": 2.0793826580047607, + "learning_rate": 4.825501511404232e-07, + "loss": 0.8857, + "mean_token_accuracy": 0.7203917503356934, + "num_tokens": 53008892.0, + "step": 1757 + }, + { + "epoch": 0.1449657788406036, + "grad_norm": 1.753701090812683, + "learning_rate": 4.828249519098654e-07, + "loss": 0.7758, + "mean_token_accuracy": 0.7575061321258545, + "num_tokens": 53048418.0, + "step": 1758 + }, + { + "epoch": 0.14504823946565515, + "grad_norm": 1.984602451324463, + "learning_rate": 4.830997526793074e-07, + "loss": 0.9711, + "mean_token_accuracy": 0.7076729536056519, + "num_tokens": 53080535.0, + "step": 1759 + }, + { + "epoch": 0.1451307000907067, + "grad_norm": 1.8825918436050415, + "learning_rate": 4.833745534487496e-07, + "loss": 0.8073, + "mean_token_accuracy": 0.7564477920532227, + "num_tokens": 53114258.0, + "step": 1760 + }, + { + "epoch": 0.14521316071575824, + "grad_norm": 2.1789257526397705, + "learning_rate": 4.836493542181918e-07, + "loss": 0.755, + "mean_token_accuracy": 0.7633332014083862, + "num_tokens": 53143169.0, + "step": 1761 + }, + { + "epoch": 0.14529562134080976, + "grad_norm": 2.2266016006469727, + "learning_rate": 4.839241549876339e-07, + "loss": 0.7741, + "mean_token_accuracy": 0.7583499550819397, + "num_tokens": 53169904.0, + "step": 1762 + }, + { + "epoch": 0.1453780819658613, + "grad_norm": 1.7977296113967896, + "learning_rate": 4.841989557570761e-07, + "loss": 0.8627, + "mean_token_accuracy": 0.739262044429779, + "num_tokens": 53208328.0, + "step": 1763 + }, + { + "epoch": 0.14546054259091284, + "grad_norm": 1.9617938995361328, + "learning_rate": 4.844737565265182e-07, + "loss": 0.9816, + "mean_token_accuracy": 0.7028167843818665, + "num_tokens": 53242659.0, + "step": 1764 + }, + { + "epoch": 0.1455430032159644, + "grad_norm": 1.9263410568237305, + "learning_rate": 4.847485572959605e-07, + "loss": 0.8316, + "mean_token_accuracy": 0.7437317967414856, + "num_tokens": 53276165.0, + "step": 1765 + }, + { + "epoch": 0.1456254638410159, + "grad_norm": 2.15061092376709, + "learning_rate": 4.850233580654025e-07, + "loss": 0.9961, + "mean_token_accuracy": 0.7021395564079285, + "num_tokens": 53306783.0, + "step": 1766 + }, + { + "epoch": 0.14570792446606745, + "grad_norm": 2.0116641521453857, + "learning_rate": 4.852981588348447e-07, + "loss": 0.796, + "mean_token_accuracy": 0.7471225261688232, + "num_tokens": 53338404.0, + "step": 1767 + }, + { + "epoch": 0.145790385091119, + "grad_norm": 2.0668511390686035, + "learning_rate": 4.855729596042869e-07, + "loss": 0.7447, + "mean_token_accuracy": 0.7774937152862549, + "num_tokens": 53366093.0, + "step": 1768 + }, + { + "epoch": 0.14587284571617054, + "grad_norm": 1.872800350189209, + "learning_rate": 4.85847760373729e-07, + "loss": 1.0323, + "mean_token_accuracy": 0.6852450370788574, + "num_tokens": 53401204.0, + "step": 1769 + }, + { + "epoch": 0.14595530634122206, + "grad_norm": 2.20345401763916, + "learning_rate": 4.861225611431712e-07, + "loss": 0.8657, + "mean_token_accuracy": 0.7323510646820068, + "num_tokens": 53429731.0, + "step": 1770 + }, + { + "epoch": 0.1460377669662736, + "grad_norm": 1.8910011053085327, + "learning_rate": 4.863973619126133e-07, + "loss": 0.6543, + "mean_token_accuracy": 0.7830115556716919, + "num_tokens": 53459708.0, + "step": 1771 + }, + { + "epoch": 0.14612022759132515, + "grad_norm": 1.8578280210494995, + "learning_rate": 4.866721626820555e-07, + "loss": 0.7698, + "mean_token_accuracy": 0.7693190574645996, + "num_tokens": 53494277.0, + "step": 1772 + }, + { + "epoch": 0.1462026882163767, + "grad_norm": 2.2554972171783447, + "learning_rate": 4.869469634514976e-07, + "loss": 0.8652, + "mean_token_accuracy": 0.7420366406440735, + "num_tokens": 53520402.0, + "step": 1773 + }, + { + "epoch": 0.1462851488414282, + "grad_norm": 2.007028818130493, + "learning_rate": 4.872217642209398e-07, + "loss": 0.8169, + "mean_token_accuracy": 0.738685131072998, + "num_tokens": 53551315.0, + "step": 1774 + }, + { + "epoch": 0.14636760946647975, + "grad_norm": 2.0575709342956543, + "learning_rate": 4.87496564990382e-07, + "loss": 0.7602, + "mean_token_accuracy": 0.7555232644081116, + "num_tokens": 53579029.0, + "step": 1775 + }, + { + "epoch": 0.1464500700915313, + "grad_norm": 2.169990062713623, + "learning_rate": 4.87771365759824e-07, + "loss": 0.8485, + "mean_token_accuracy": 0.7374513745307922, + "num_tokens": 53605697.0, + "step": 1776 + }, + { + "epoch": 0.14653253071658284, + "grad_norm": 2.167548894882202, + "learning_rate": 4.880461665292663e-07, + "loss": 0.804, + "mean_token_accuracy": 0.7510885000228882, + "num_tokens": 53632079.0, + "step": 1777 + }, + { + "epoch": 0.14661499134163436, + "grad_norm": 2.417888879776001, + "learning_rate": 4.883209672987084e-07, + "loss": 0.9173, + "mean_token_accuracy": 0.7170870304107666, + "num_tokens": 53656414.0, + "step": 1778 + }, + { + "epoch": 0.1466974519666859, + "grad_norm": 2.095160722732544, + "learning_rate": 4.885957680681505e-07, + "loss": 0.8608, + "mean_token_accuracy": 0.7413516044616699, + "num_tokens": 53688603.0, + "step": 1779 + }, + { + "epoch": 0.14677991259173745, + "grad_norm": 1.9518201351165771, + "learning_rate": 4.888705688375927e-07, + "loss": 0.8634, + "mean_token_accuracy": 0.7388441562652588, + "num_tokens": 53723613.0, + "step": 1780 + }, + { + "epoch": 0.146862373216789, + "grad_norm": 1.9953880310058594, + "learning_rate": 4.891453696070349e-07, + "loss": 0.7871, + "mean_token_accuracy": 0.7524806261062622, + "num_tokens": 53751991.0, + "step": 1781 + }, + { + "epoch": 0.1469448338418405, + "grad_norm": 1.790615200996399, + "learning_rate": 4.894201703764771e-07, + "loss": 1.033, + "mean_token_accuracy": 0.701306939125061, + "num_tokens": 53792339.0, + "step": 1782 + }, + { + "epoch": 0.14702729446689206, + "grad_norm": 2.5902628898620605, + "learning_rate": 4.896949711459191e-07, + "loss": 1.0985, + "mean_token_accuracy": 0.6670498847961426, + "num_tokens": 53815519.0, + "step": 1783 + }, + { + "epoch": 0.1471097550919436, + "grad_norm": 2.1230006217956543, + "learning_rate": 4.899697719153613e-07, + "loss": 0.8866, + "mean_token_accuracy": 0.7241868376731873, + "num_tokens": 53842123.0, + "step": 1784 + }, + { + "epoch": 0.14719221571699515, + "grad_norm": 2.2320075035095215, + "learning_rate": 4.902445726848035e-07, + "loss": 0.9109, + "mean_token_accuracy": 0.7217571139335632, + "num_tokens": 53869605.0, + "step": 1785 + }, + { + "epoch": 0.14727467634204666, + "grad_norm": 2.3092517852783203, + "learning_rate": 4.905193734542456e-07, + "loss": 0.8627, + "mean_token_accuracy": 0.7414803504943848, + "num_tokens": 53897276.0, + "step": 1786 + }, + { + "epoch": 0.1473571369670982, + "grad_norm": 1.999781608581543, + "learning_rate": 4.907941742236878e-07, + "loss": 0.9236, + "mean_token_accuracy": 0.729097306728363, + "num_tokens": 53930012.0, + "step": 1787 + }, + { + "epoch": 0.14743959759214975, + "grad_norm": 2.029168128967285, + "learning_rate": 4.9106897499313e-07, + "loss": 0.9045, + "mean_token_accuracy": 0.7259600758552551, + "num_tokens": 53961797.0, + "step": 1788 + }, + { + "epoch": 0.1475220582172013, + "grad_norm": 1.9572128057479858, + "learning_rate": 4.913437757625722e-07, + "loss": 0.9163, + "mean_token_accuracy": 0.7307217121124268, + "num_tokens": 53994349.0, + "step": 1789 + }, + { + "epoch": 0.1476045188422528, + "grad_norm": 2.216562509536743, + "learning_rate": 4.916185765320142e-07, + "loss": 0.9389, + "mean_token_accuracy": 0.7147018909454346, + "num_tokens": 54020473.0, + "step": 1790 + }, + { + "epoch": 0.14768697946730436, + "grad_norm": 1.9390121698379517, + "learning_rate": 4.918933773014564e-07, + "loss": 0.8799, + "mean_token_accuracy": 0.7318960428237915, + "num_tokens": 54055835.0, + "step": 1791 + }, + { + "epoch": 0.1477694400923559, + "grad_norm": 2.3216137886047363, + "learning_rate": 4.921681780708986e-07, + "loss": 0.7768, + "mean_token_accuracy": 0.7557121515274048, + "num_tokens": 54082884.0, + "step": 1792 + }, + { + "epoch": 0.14785190071740745, + "grad_norm": 2.28859543800354, + "learning_rate": 4.924429788403407e-07, + "loss": 0.8153, + "mean_token_accuracy": 0.7464511394500732, + "num_tokens": 54109739.0, + "step": 1793 + }, + { + "epoch": 0.14793436134245896, + "grad_norm": 2.0158917903900146, + "learning_rate": 4.927177796097829e-07, + "loss": 0.7728, + "mean_token_accuracy": 0.7610437273979187, + "num_tokens": 54139574.0, + "step": 1794 + }, + { + "epoch": 0.1480168219675105, + "grad_norm": 1.925517201423645, + "learning_rate": 4.929925803792251e-07, + "loss": 0.7885, + "mean_token_accuracy": 0.7594107389450073, + "num_tokens": 54172854.0, + "step": 1795 + }, + { + "epoch": 0.14809928259256205, + "grad_norm": 2.3011271953582764, + "learning_rate": 4.932673811486671e-07, + "loss": 1.0865, + "mean_token_accuracy": 0.684407114982605, + "num_tokens": 54199513.0, + "step": 1796 + }, + { + "epoch": 0.1481817432176136, + "grad_norm": 1.9007765054702759, + "learning_rate": 4.935421819181093e-07, + "loss": 0.888, + "mean_token_accuracy": 0.7341702580451965, + "num_tokens": 54237278.0, + "step": 1797 + }, + { + "epoch": 0.14826420384266512, + "grad_norm": 2.3002567291259766, + "learning_rate": 4.938169826875515e-07, + "loss": 0.8598, + "mean_token_accuracy": 0.740615963935852, + "num_tokens": 54262051.0, + "step": 1798 + }, + { + "epoch": 0.14834666446771666, + "grad_norm": 2.162287473678589, + "learning_rate": 4.940917834569937e-07, + "loss": 0.9427, + "mean_token_accuracy": 0.7143123149871826, + "num_tokens": 54290489.0, + "step": 1799 + }, + { + "epoch": 0.1484291250927682, + "grad_norm": 1.827702522277832, + "learning_rate": 4.943665842264358e-07, + "loss": 0.8758, + "mean_token_accuracy": 0.7356058359146118, + "num_tokens": 54328869.0, + "step": 1800 + }, + { + "epoch": 0.14851158571781975, + "grad_norm": 2.1300406455993652, + "learning_rate": 4.94641384995878e-07, + "loss": 1.0014, + "mean_token_accuracy": 0.7057164907455444, + "num_tokens": 54358503.0, + "step": 1801 + }, + { + "epoch": 0.14859404634287127, + "grad_norm": 1.9458543062210083, + "learning_rate": 4.949161857653202e-07, + "loss": 0.6995, + "mean_token_accuracy": 0.7839722633361816, + "num_tokens": 54390434.0, + "step": 1802 + }, + { + "epoch": 0.1486765069679228, + "grad_norm": 1.9540562629699707, + "learning_rate": 4.951909865347622e-07, + "loss": 0.8152, + "mean_token_accuracy": 0.7478034496307373, + "num_tokens": 54424798.0, + "step": 1803 + }, + { + "epoch": 0.14875896759297436, + "grad_norm": 1.9320064783096313, + "learning_rate": 4.954657873042044e-07, + "loss": 0.9027, + "mean_token_accuracy": 0.7322335243225098, + "num_tokens": 54460000.0, + "step": 1804 + }, + { + "epoch": 0.1488414282180259, + "grad_norm": 2.1902217864990234, + "learning_rate": 4.957405880736466e-07, + "loss": 0.9077, + "mean_token_accuracy": 0.7235999703407288, + "num_tokens": 54485469.0, + "step": 1805 + }, + { + "epoch": 0.14892388884307742, + "grad_norm": 2.325775623321533, + "learning_rate": 4.960153888430888e-07, + "loss": 0.9385, + "mean_token_accuracy": 0.7115559577941895, + "num_tokens": 54509420.0, + "step": 1806 + }, + { + "epoch": 0.14900634946812896, + "grad_norm": 2.2769532203674316, + "learning_rate": 4.962901896125309e-07, + "loss": 0.8845, + "mean_token_accuracy": 0.7311135530471802, + "num_tokens": 54537324.0, + "step": 1807 + }, + { + "epoch": 0.1490888100931805, + "grad_norm": 1.9967052936553955, + "learning_rate": 4.96564990381973e-07, + "loss": 0.7929, + "mean_token_accuracy": 0.7560660243034363, + "num_tokens": 54568641.0, + "step": 1808 + }, + { + "epoch": 0.14917127071823205, + "grad_norm": 2.055105686187744, + "learning_rate": 4.968397911514153e-07, + "loss": 0.8612, + "mean_token_accuracy": 0.7411644458770752, + "num_tokens": 54598268.0, + "step": 1809 + }, + { + "epoch": 0.14925373134328357, + "grad_norm": 1.8060295581817627, + "learning_rate": 4.971145919208573e-07, + "loss": 0.7658, + "mean_token_accuracy": 0.761658787727356, + "num_tokens": 54632707.0, + "step": 1810 + }, + { + "epoch": 0.1493361919683351, + "grad_norm": 2.175990581512451, + "learning_rate": 4.973893926902995e-07, + "loss": 0.8999, + "mean_token_accuracy": 0.7192807793617249, + "num_tokens": 54660884.0, + "step": 1811 + }, + { + "epoch": 0.14941865259338666, + "grad_norm": 2.104097604751587, + "learning_rate": 4.976641934597417e-07, + "loss": 0.8135, + "mean_token_accuracy": 0.7479180693626404, + "num_tokens": 54688729.0, + "step": 1812 + }, + { + "epoch": 0.1495011132184382, + "grad_norm": 1.755334496498108, + "learning_rate": 4.979389942291838e-07, + "loss": 0.7588, + "mean_token_accuracy": 0.7722519636154175, + "num_tokens": 54726448.0, + "step": 1813 + }, + { + "epoch": 0.14958357384348972, + "grad_norm": 2.1793015003204346, + "learning_rate": 4.98213794998626e-07, + "loss": 0.9774, + "mean_token_accuracy": 0.7046716809272766, + "num_tokens": 54755350.0, + "step": 1814 + }, + { + "epoch": 0.14966603446854126, + "grad_norm": 2.1134450435638428, + "learning_rate": 4.984885957680681e-07, + "loss": 0.9091, + "mean_token_accuracy": 0.7214349508285522, + "num_tokens": 54783830.0, + "step": 1815 + }, + { + "epoch": 0.1497484950935928, + "grad_norm": 2.1846230030059814, + "learning_rate": 4.987633965375103e-07, + "loss": 0.8689, + "mean_token_accuracy": 0.7365109920501709, + "num_tokens": 54812429.0, + "step": 1816 + }, + { + "epoch": 0.14983095571864435, + "grad_norm": 2.6861732006073, + "learning_rate": 4.990381973069524e-07, + "loss": 0.942, + "mean_token_accuracy": 0.7073571085929871, + "num_tokens": 54843395.0, + "step": 1817 + }, + { + "epoch": 0.1499134163436959, + "grad_norm": 1.7961152791976929, + "learning_rate": 4.993129980763946e-07, + "loss": 0.8087, + "mean_token_accuracy": 0.753034234046936, + "num_tokens": 54879374.0, + "step": 1818 + }, + { + "epoch": 0.14999587696874742, + "grad_norm": 1.9997234344482422, + "learning_rate": 4.995877988458368e-07, + "loss": 0.86, + "mean_token_accuracy": 0.7375285625457764, + "num_tokens": 54909819.0, + "step": 1819 + }, + { + "epoch": 0.15007833759379896, + "grad_norm": 2.1220178604125977, + "learning_rate": 4.998625996152788e-07, + "loss": 1.0811, + "mean_token_accuracy": 0.6891776323318481, + "num_tokens": 54939300.0, + "step": 1820 + }, + { + "epoch": 0.1501607982188505, + "grad_norm": 2.1112027168273926, + "learning_rate": 5.001374003847211e-07, + "loss": 0.7765, + "mean_token_accuracy": 0.7601979970932007, + "num_tokens": 54969298.0, + "step": 1821 + }, + { + "epoch": 0.15024325884390205, + "grad_norm": 1.837648630142212, + "learning_rate": 5.004122011541632e-07, + "loss": 0.8884, + "mean_token_accuracy": 0.7330594658851624, + "num_tokens": 55008628.0, + "step": 1822 + }, + { + "epoch": 0.15032571946895357, + "grad_norm": 2.0293099880218506, + "learning_rate": 5.006870019236054e-07, + "loss": 0.977, + "mean_token_accuracy": 0.7146437168121338, + "num_tokens": 55043221.0, + "step": 1823 + }, + { + "epoch": 0.1504081800940051, + "grad_norm": 1.8843249082565308, + "learning_rate": 5.009618026930475e-07, + "loss": 0.8003, + "mean_token_accuracy": 0.7577842473983765, + "num_tokens": 55075954.0, + "step": 1824 + }, + { + "epoch": 0.15049064071905666, + "grad_norm": 2.1333749294281006, + "learning_rate": 5.012366034624896e-07, + "loss": 0.905, + "mean_token_accuracy": 0.7239571213722229, + "num_tokens": 55102801.0, + "step": 1825 + }, + { + "epoch": 0.1505731013441082, + "grad_norm": 2.022869348526001, + "learning_rate": 5.015114042319319e-07, + "loss": 0.9625, + "mean_token_accuracy": 0.7083941698074341, + "num_tokens": 55136103.0, + "step": 1826 + }, + { + "epoch": 0.15065556196915972, + "grad_norm": 2.0214033126831055, + "learning_rate": 5.01786205001374e-07, + "loss": 0.9691, + "mean_token_accuracy": 0.7024414539337158, + "num_tokens": 55166639.0, + "step": 1827 + }, + { + "epoch": 0.15073802259421126, + "grad_norm": 1.850500226020813, + "learning_rate": 5.020610057708161e-07, + "loss": 0.8208, + "mean_token_accuracy": 0.756506621837616, + "num_tokens": 55203245.0, + "step": 1828 + }, + { + "epoch": 0.1508204832192628, + "grad_norm": 2.238138198852539, + "learning_rate": 5.023358065402583e-07, + "loss": 0.8068, + "mean_token_accuracy": 0.7500932812690735, + "num_tokens": 55229258.0, + "step": 1829 + }, + { + "epoch": 0.15090294384431435, + "grad_norm": 2.132136821746826, + "learning_rate": 5.026106073097004e-07, + "loss": 0.8025, + "mean_token_accuracy": 0.7455651760101318, + "num_tokens": 55259221.0, + "step": 1830 + }, + { + "epoch": 0.15098540446936587, + "grad_norm": 1.743604063987732, + "learning_rate": 5.028854080791426e-07, + "loss": 0.8056, + "mean_token_accuracy": 0.7522326707839966, + "num_tokens": 55297516.0, + "step": 1831 + }, + { + "epoch": 0.15106786509441741, + "grad_norm": 2.029782295227051, + "learning_rate": 5.031602088485848e-07, + "loss": 0.9558, + "mean_token_accuracy": 0.7172785401344299, + "num_tokens": 55328528.0, + "step": 1832 + }, + { + "epoch": 0.15115032571946896, + "grad_norm": 2.098684787750244, + "learning_rate": 5.034350096180269e-07, + "loss": 0.9116, + "mean_token_accuracy": 0.7367525100708008, + "num_tokens": 55360139.0, + "step": 1833 + }, + { + "epoch": 0.1512327863445205, + "grad_norm": 1.9165658950805664, + "learning_rate": 5.03709810387469e-07, + "loss": 0.8523, + "mean_token_accuracy": 0.7446428537368774, + "num_tokens": 55394313.0, + "step": 1834 + }, + { + "epoch": 0.15131524696957202, + "grad_norm": 2.022843599319458, + "learning_rate": 5.039846111569113e-07, + "loss": 0.8934, + "mean_token_accuracy": 0.7252094745635986, + "num_tokens": 55427687.0, + "step": 1835 + }, + { + "epoch": 0.15139770759462357, + "grad_norm": 1.857958197593689, + "learning_rate": 5.042594119263534e-07, + "loss": 0.8171, + "mean_token_accuracy": 0.7482229471206665, + "num_tokens": 55463687.0, + "step": 1836 + }, + { + "epoch": 0.1514801682196751, + "grad_norm": 2.3366196155548096, + "learning_rate": 5.045342126957955e-07, + "loss": 0.9311, + "mean_token_accuracy": 0.718806803226471, + "num_tokens": 55488784.0, + "step": 1837 + }, + { + "epoch": 0.15156262884472665, + "grad_norm": 2.1123769283294678, + "learning_rate": 5.048090134652376e-07, + "loss": 0.9745, + "mean_token_accuracy": 0.7175321578979492, + "num_tokens": 55521779.0, + "step": 1838 + }, + { + "epoch": 0.15164508946977817, + "grad_norm": 1.887294888496399, + "learning_rate": 5.050838142346798e-07, + "loss": 0.764, + "mean_token_accuracy": 0.7593780755996704, + "num_tokens": 55553918.0, + "step": 1839 + }, + { + "epoch": 0.15172755009482972, + "grad_norm": 2.411526918411255, + "learning_rate": 5.053586150041221e-07, + "loss": 0.9092, + "mean_token_accuracy": 0.7264455556869507, + "num_tokens": 55576708.0, + "step": 1840 + }, + { + "epoch": 0.15181001071988126, + "grad_norm": 1.8163325786590576, + "learning_rate": 5.056334157735641e-07, + "loss": 0.8178, + "mean_token_accuracy": 0.7430169582366943, + "num_tokens": 55612753.0, + "step": 1841 + }, + { + "epoch": 0.1518924713449328, + "grad_norm": 2.068171501159668, + "learning_rate": 5.059082165430063e-07, + "loss": 0.9106, + "mean_token_accuracy": 0.7252832055091858, + "num_tokens": 55641858.0, + "step": 1842 + }, + { + "epoch": 0.15197493196998432, + "grad_norm": 1.9425196647644043, + "learning_rate": 5.061830173124485e-07, + "loss": 0.8141, + "mean_token_accuracy": 0.7466685771942139, + "num_tokens": 55672422.0, + "step": 1843 + }, + { + "epoch": 0.15205739259503587, + "grad_norm": 1.961548089981079, + "learning_rate": 5.064578180818906e-07, + "loss": 0.9984, + "mean_token_accuracy": 0.707135796546936, + "num_tokens": 55704858.0, + "step": 1844 + }, + { + "epoch": 0.1521398532200874, + "grad_norm": 2.1071529388427734, + "learning_rate": 5.067326188513328e-07, + "loss": 0.7833, + "mean_token_accuracy": 0.7623915076255798, + "num_tokens": 55732140.0, + "step": 1845 + }, + { + "epoch": 0.15222231384513896, + "grad_norm": 2.069014310836792, + "learning_rate": 5.070074196207749e-07, + "loss": 0.7288, + "mean_token_accuracy": 0.7742234468460083, + "num_tokens": 55762080.0, + "step": 1846 + }, + { + "epoch": 0.15230477447019047, + "grad_norm": 1.8690862655639648, + "learning_rate": 5.07282220390217e-07, + "loss": 0.9663, + "mean_token_accuracy": 0.7016614675521851, + "num_tokens": 55796951.0, + "step": 1847 + }, + { + "epoch": 0.15238723509524202, + "grad_norm": 2.1025211811065674, + "learning_rate": 5.075570211596592e-07, + "loss": 0.89, + "mean_token_accuracy": 0.7330448031425476, + "num_tokens": 55824550.0, + "step": 1848 + }, + { + "epoch": 0.15246969572029356, + "grad_norm": 2.0572245121002197, + "learning_rate": 5.078318219291014e-07, + "loss": 0.758, + "mean_token_accuracy": 0.7643921375274658, + "num_tokens": 55852823.0, + "step": 1849 + }, + { + "epoch": 0.1525521563453451, + "grad_norm": 1.9485044479370117, + "learning_rate": 5.081066226985436e-07, + "loss": 0.9268, + "mean_token_accuracy": 0.7140077948570251, + "num_tokens": 55886387.0, + "step": 1850 + }, + { + "epoch": 0.15263461697039662, + "grad_norm": 2.3220088481903076, + "learning_rate": 5.083814234679857e-07, + "loss": 1.0667, + "mean_token_accuracy": 0.6842535138130188, + "num_tokens": 55913718.0, + "step": 1851 + }, + { + "epoch": 0.15271707759544817, + "grad_norm": 1.8145289421081543, + "learning_rate": 5.086562242374278e-07, + "loss": 0.9348, + "mean_token_accuracy": 0.7166938781738281, + "num_tokens": 55949889.0, + "step": 1852 + }, + { + "epoch": 0.15279953822049971, + "grad_norm": 2.331336259841919, + "learning_rate": 5.0893102500687e-07, + "loss": 0.9233, + "mean_token_accuracy": 0.7230398654937744, + "num_tokens": 55973066.0, + "step": 1853 + }, + { + "epoch": 0.15288199884555126, + "grad_norm": 1.9276326894760132, + "learning_rate": 5.092058257763121e-07, + "loss": 0.6846, + "mean_token_accuracy": 0.786616325378418, + "num_tokens": 56004380.0, + "step": 1854 + }, + { + "epoch": 0.15296445947060278, + "grad_norm": 1.8556193113327026, + "learning_rate": 5.094806265457543e-07, + "loss": 0.8268, + "mean_token_accuracy": 0.754289984703064, + "num_tokens": 56036624.0, + "step": 1855 + }, + { + "epoch": 0.15304692009565432, + "grad_norm": 1.9102586507797241, + "learning_rate": 5.097554273151965e-07, + "loss": 0.866, + "mean_token_accuracy": 0.7342514991760254, + "num_tokens": 56069303.0, + "step": 1856 + }, + { + "epoch": 0.15312938072070587, + "grad_norm": 2.035576105117798, + "learning_rate": 5.100302280846386e-07, + "loss": 0.9104, + "mean_token_accuracy": 0.7190113067626953, + "num_tokens": 56099790.0, + "step": 1857 + }, + { + "epoch": 0.1532118413457574, + "grad_norm": 2.0167574882507324, + "learning_rate": 5.103050288540808e-07, + "loss": 0.816, + "mean_token_accuracy": 0.751401424407959, + "num_tokens": 56129824.0, + "step": 1858 + }, + { + "epoch": 0.15329430197080893, + "grad_norm": 2.2296507358551025, + "learning_rate": 5.10579829623523e-07, + "loss": 0.7929, + "mean_token_accuracy": 0.7555414438247681, + "num_tokens": 56155194.0, + "step": 1859 + }, + { + "epoch": 0.15337676259586047, + "grad_norm": 1.9441633224487305, + "learning_rate": 5.108546303929651e-07, + "loss": 0.8416, + "mean_token_accuracy": 0.7389636039733887, + "num_tokens": 56185650.0, + "step": 1860 + }, + { + "epoch": 0.15345922322091202, + "grad_norm": 2.0110831260681152, + "learning_rate": 5.111294311624072e-07, + "loss": 0.7281, + "mean_token_accuracy": 0.754863977432251, + "num_tokens": 56214730.0, + "step": 1861 + }, + { + "epoch": 0.15354168384596356, + "grad_norm": 2.2372007369995117, + "learning_rate": 5.114042319318493e-07, + "loss": 0.9858, + "mean_token_accuracy": 0.7161402702331543, + "num_tokens": 56242159.0, + "step": 1862 + }, + { + "epoch": 0.15362414447101508, + "grad_norm": 2.124419927597046, + "learning_rate": 5.116790327012916e-07, + "loss": 0.8936, + "mean_token_accuracy": 0.7273122072219849, + "num_tokens": 56275328.0, + "step": 1863 + }, + { + "epoch": 0.15370660509606662, + "grad_norm": 1.933325171470642, + "learning_rate": 5.119538334707338e-07, + "loss": 0.8905, + "mean_token_accuracy": 0.7259213924407959, + "num_tokens": 56308351.0, + "step": 1864 + }, + { + "epoch": 0.15378906572111817, + "grad_norm": 2.0444140434265137, + "learning_rate": 5.122286342401758e-07, + "loss": 0.893, + "mean_token_accuracy": 0.7287408113479614, + "num_tokens": 56338568.0, + "step": 1865 + }, + { + "epoch": 0.1538715263461697, + "grad_norm": 1.9405524730682373, + "learning_rate": 5.12503435009618e-07, + "loss": 0.874, + "mean_token_accuracy": 0.7426609992980957, + "num_tokens": 56370707.0, + "step": 1866 + }, + { + "epoch": 0.15395398697122123, + "grad_norm": 2.3508448600769043, + "learning_rate": 5.127782357790602e-07, + "loss": 0.889, + "mean_token_accuracy": 0.7189972400665283, + "num_tokens": 56395886.0, + "step": 1867 + }, + { + "epoch": 0.15403644759627277, + "grad_norm": 2.2890572547912598, + "learning_rate": 5.130530365485023e-07, + "loss": 0.9403, + "mean_token_accuracy": 0.7133245468139648, + "num_tokens": 56423394.0, + "step": 1868 + }, + { + "epoch": 0.15411890822132432, + "grad_norm": 2.239947557449341, + "learning_rate": 5.133278373179445e-07, + "loss": 0.9466, + "mean_token_accuracy": 0.7110059261322021, + "num_tokens": 56450320.0, + "step": 1869 + }, + { + "epoch": 0.15420136884637586, + "grad_norm": 1.7809350490570068, + "learning_rate": 5.136026380873866e-07, + "loss": 0.6942, + "mean_token_accuracy": 0.7790764570236206, + "num_tokens": 56487133.0, + "step": 1870 + }, + { + "epoch": 0.15428382947142738, + "grad_norm": 2.021724224090576, + "learning_rate": 5.138774388568287e-07, + "loss": 0.8485, + "mean_token_accuracy": 0.7410566806793213, + "num_tokens": 56519300.0, + "step": 1871 + }, + { + "epoch": 0.15436629009647893, + "grad_norm": 2.2232918739318848, + "learning_rate": 5.14152239626271e-07, + "loss": 1.0191, + "mean_token_accuracy": 0.7025390863418579, + "num_tokens": 56547920.0, + "step": 1872 + }, + { + "epoch": 0.15444875072153047, + "grad_norm": 1.9414582252502441, + "learning_rate": 5.144270403957131e-07, + "loss": 0.7203, + "mean_token_accuracy": 0.762307345867157, + "num_tokens": 56577287.0, + "step": 1873 + }, + { + "epoch": 0.15453121134658201, + "grad_norm": 2.114431858062744, + "learning_rate": 5.147018411651553e-07, + "loss": 0.8305, + "mean_token_accuracy": 0.7505825757980347, + "num_tokens": 56604594.0, + "step": 1874 + }, + { + "epoch": 0.15461367197163353, + "grad_norm": 2.142359972000122, + "learning_rate": 5.149766419345974e-07, + "loss": 0.9471, + "mean_token_accuracy": 0.710605263710022, + "num_tokens": 56635279.0, + "step": 1875 + }, + { + "epoch": 0.15469613259668508, + "grad_norm": 2.1010146141052246, + "learning_rate": 5.152514427040395e-07, + "loss": 0.9168, + "mean_token_accuracy": 0.729682445526123, + "num_tokens": 56666777.0, + "step": 1876 + }, + { + "epoch": 0.15477859322173662, + "grad_norm": 2.330125093460083, + "learning_rate": 5.155262434734818e-07, + "loss": 0.8776, + "mean_token_accuracy": 0.7417199611663818, + "num_tokens": 56691230.0, + "step": 1877 + }, + { + "epoch": 0.15486105384678817, + "grad_norm": 2.0293233394622803, + "learning_rate": 5.158010442429238e-07, + "loss": 0.8001, + "mean_token_accuracy": 0.7524474859237671, + "num_tokens": 56718901.0, + "step": 1878 + }, + { + "epoch": 0.15494351447183968, + "grad_norm": 2.198007583618164, + "learning_rate": 5.16075845012366e-07, + "loss": 0.7986, + "mean_token_accuracy": 0.7504761815071106, + "num_tokens": 56743237.0, + "step": 1879 + }, + { + "epoch": 0.15502597509689123, + "grad_norm": 2.0274651050567627, + "learning_rate": 5.163506457818082e-07, + "loss": 0.7871, + "mean_token_accuracy": 0.7553460001945496, + "num_tokens": 56772842.0, + "step": 1880 + }, + { + "epoch": 0.15510843572194277, + "grad_norm": 2.0712571144104004, + "learning_rate": 5.166254465512502e-07, + "loss": 0.8982, + "mean_token_accuracy": 0.7194044589996338, + "num_tokens": 56803311.0, + "step": 1881 + }, + { + "epoch": 0.15519089634699432, + "grad_norm": 2.0742499828338623, + "learning_rate": 5.169002473206925e-07, + "loss": 1.0329, + "mean_token_accuracy": 0.6895038485527039, + "num_tokens": 56836375.0, + "step": 1882 + }, + { + "epoch": 0.15527335697204586, + "grad_norm": 1.8372310400009155, + "learning_rate": 5.171750480901347e-07, + "loss": 0.7447, + "mean_token_accuracy": 0.7671487927436829, + "num_tokens": 56872199.0, + "step": 1883 + }, + { + "epoch": 0.15535581759709738, + "grad_norm": 2.411149024963379, + "learning_rate": 5.174498488595768e-07, + "loss": 1.0445, + "mean_token_accuracy": 0.6886780261993408, + "num_tokens": 56896496.0, + "step": 1884 + }, + { + "epoch": 0.15543827822214892, + "grad_norm": 2.0040483474731445, + "learning_rate": 5.177246496290189e-07, + "loss": 0.997, + "mean_token_accuracy": 0.7076826095581055, + "num_tokens": 56927689.0, + "step": 1885 + }, + { + "epoch": 0.15552073884720047, + "grad_norm": 1.8695622682571411, + "learning_rate": 5.17999450398461e-07, + "loss": 0.7825, + "mean_token_accuracy": 0.751777172088623, + "num_tokens": 56962418.0, + "step": 1886 + }, + { + "epoch": 0.155603199472252, + "grad_norm": 1.9737390279769897, + "learning_rate": 5.182742511679033e-07, + "loss": 0.7858, + "mean_token_accuracy": 0.7574151754379272, + "num_tokens": 56991169.0, + "step": 1887 + }, + { + "epoch": 0.15568566009730353, + "grad_norm": 2.169602632522583, + "learning_rate": 5.185490519373454e-07, + "loss": 0.8634, + "mean_token_accuracy": 0.7328963279724121, + "num_tokens": 57019742.0, + "step": 1888 + }, + { + "epoch": 0.15576812072235507, + "grad_norm": 2.121305465698242, + "learning_rate": 5.188238527067875e-07, + "loss": 0.7943, + "mean_token_accuracy": 0.7490081787109375, + "num_tokens": 57047638.0, + "step": 1889 + }, + { + "epoch": 0.15585058134740662, + "grad_norm": 2.359823703765869, + "learning_rate": 5.190986534762297e-07, + "loss": 0.9054, + "mean_token_accuracy": 0.7213444709777832, + "num_tokens": 57071452.0, + "step": 1890 + }, + { + "epoch": 0.15593304197245816, + "grad_norm": 2.1187124252319336, + "learning_rate": 5.19373454245672e-07, + "loss": 0.8297, + "mean_token_accuracy": 0.7467620372772217, + "num_tokens": 57100453.0, + "step": 1891 + }, + { + "epoch": 0.15601550259750968, + "grad_norm": 1.965823769569397, + "learning_rate": 5.19648255015114e-07, + "loss": 1.0729, + "mean_token_accuracy": 0.6935751438140869, + "num_tokens": 57138807.0, + "step": 1892 + }, + { + "epoch": 0.15609796322256123, + "grad_norm": 1.8629857301712036, + "learning_rate": 5.199230557845562e-07, + "loss": 0.9224, + "mean_token_accuracy": 0.723006546497345, + "num_tokens": 57176682.0, + "step": 1893 + }, + { + "epoch": 0.15618042384761277, + "grad_norm": 2.032832622528076, + "learning_rate": 5.201978565539983e-07, + "loss": 0.9693, + "mean_token_accuracy": 0.7101889848709106, + "num_tokens": 57206199.0, + "step": 1894 + }, + { + "epoch": 0.15626288447266432, + "grad_norm": 2.3545644283294678, + "learning_rate": 5.204726573234404e-07, + "loss": 0.9, + "mean_token_accuracy": 0.7303619384765625, + "num_tokens": 57231297.0, + "step": 1895 + }, + { + "epoch": 0.15634534509771583, + "grad_norm": 2.1709868907928467, + "learning_rate": 5.207474580928827e-07, + "loss": 0.9324, + "mean_token_accuracy": 0.7083973288536072, + "num_tokens": 57258295.0, + "step": 1896 + }, + { + "epoch": 0.15642780572276738, + "grad_norm": 1.8707513809204102, + "learning_rate": 5.210222588623248e-07, + "loss": 0.7004, + "mean_token_accuracy": 0.7830420732498169, + "num_tokens": 57288842.0, + "step": 1897 + }, + { + "epoch": 0.15651026634781892, + "grad_norm": 2.2337567806243896, + "learning_rate": 5.21297059631767e-07, + "loss": 0.9649, + "mean_token_accuracy": 0.7072205543518066, + "num_tokens": 57313528.0, + "step": 1898 + }, + { + "epoch": 0.15659272697287047, + "grad_norm": 2.060922145843506, + "learning_rate": 5.215718604012091e-07, + "loss": 0.9613, + "mean_token_accuracy": 0.7168091535568237, + "num_tokens": 57344477.0, + "step": 1899 + }, + { + "epoch": 0.15667518759792198, + "grad_norm": 2.0957131385803223, + "learning_rate": 5.218466611706512e-07, + "loss": 0.8591, + "mean_token_accuracy": 0.7373028993606567, + "num_tokens": 57374091.0, + "step": 1900 + }, + { + "epoch": 0.15675764822297353, + "grad_norm": 2.2340645790100098, + "learning_rate": 5.221214619400935e-07, + "loss": 0.9472, + "mean_token_accuracy": 0.7089908719062805, + "num_tokens": 57399928.0, + "step": 1901 + }, + { + "epoch": 0.15684010884802507, + "grad_norm": 2.0062217712402344, + "learning_rate": 5.223962627095355e-07, + "loss": 0.7519, + "mean_token_accuracy": 0.7673367857933044, + "num_tokens": 57431235.0, + "step": 1902 + }, + { + "epoch": 0.15692256947307662, + "grad_norm": 2.1579947471618652, + "learning_rate": 5.226710634789777e-07, + "loss": 0.7758, + "mean_token_accuracy": 0.756822943687439, + "num_tokens": 57460178.0, + "step": 1903 + }, + { + "epoch": 0.15700503009812813, + "grad_norm": 1.9283965826034546, + "learning_rate": 5.229458642484199e-07, + "loss": 0.9116, + "mean_token_accuracy": 0.7103390693664551, + "num_tokens": 57492260.0, + "step": 1904 + }, + { + "epoch": 0.15708749072317968, + "grad_norm": 2.0176053047180176, + "learning_rate": 5.23220665017862e-07, + "loss": 0.8705, + "mean_token_accuracy": 0.7320363521575928, + "num_tokens": 57523070.0, + "step": 1905 + }, + { + "epoch": 0.15716995134823122, + "grad_norm": 1.910685658454895, + "learning_rate": 5.234954657873042e-07, + "loss": 0.8491, + "mean_token_accuracy": 0.7358726263046265, + "num_tokens": 57556893.0, + "step": 1906 + }, + { + "epoch": 0.15725241197328277, + "grad_norm": 2.48663592338562, + "learning_rate": 5.237702665567464e-07, + "loss": 0.9931, + "mean_token_accuracy": 0.6972435712814331, + "num_tokens": 57579705.0, + "step": 1907 + }, + { + "epoch": 0.15733487259833429, + "grad_norm": 2.0898258686065674, + "learning_rate": 5.240450673261885e-07, + "loss": 0.9011, + "mean_token_accuracy": 0.7320390939712524, + "num_tokens": 57606755.0, + "step": 1908 + }, + { + "epoch": 0.15741733322338583, + "grad_norm": 1.9907459020614624, + "learning_rate": 5.243198680956306e-07, + "loss": 0.9217, + "mean_token_accuracy": 0.7287893295288086, + "num_tokens": 57637667.0, + "step": 1909 + }, + { + "epoch": 0.15749979384843737, + "grad_norm": 2.2820281982421875, + "learning_rate": 5.245946688650728e-07, + "loss": 0.9601, + "mean_token_accuracy": 0.7085834741592407, + "num_tokens": 57662344.0, + "step": 1910 + }, + { + "epoch": 0.15758225447348892, + "grad_norm": 2.0257909297943115, + "learning_rate": 5.24869469634515e-07, + "loss": 1.0106, + "mean_token_accuracy": 0.6936131715774536, + "num_tokens": 57691291.0, + "step": 1911 + }, + { + "epoch": 0.15766471509854044, + "grad_norm": 2.0972635746002197, + "learning_rate": 5.251442704039571e-07, + "loss": 0.9237, + "mean_token_accuracy": 0.7270301580429077, + "num_tokens": 57721942.0, + "step": 1912 + }, + { + "epoch": 0.15774717572359198, + "grad_norm": 2.1124720573425293, + "learning_rate": 5.254190711733992e-07, + "loss": 0.875, + "mean_token_accuracy": 0.7359429001808167, + "num_tokens": 57749713.0, + "step": 1913 + }, + { + "epoch": 0.15782963634864353, + "grad_norm": 2.162151336669922, + "learning_rate": 5.256938719428414e-07, + "loss": 0.8821, + "mean_token_accuracy": 0.7430218458175659, + "num_tokens": 57777647.0, + "step": 1914 + }, + { + "epoch": 0.15791209697369507, + "grad_norm": 1.9515252113342285, + "learning_rate": 5.259686727122837e-07, + "loss": 0.7746, + "mean_token_accuracy": 0.7392846345901489, + "num_tokens": 57811002.0, + "step": 1915 + }, + { + "epoch": 0.1579945575987466, + "grad_norm": 1.8487331867218018, + "learning_rate": 5.262434734817257e-07, + "loss": 0.8228, + "mean_token_accuracy": 0.741500973701477, + "num_tokens": 57845079.0, + "step": 1916 + }, + { + "epoch": 0.15807701822379813, + "grad_norm": 2.0747952461242676, + "learning_rate": 5.265182742511679e-07, + "loss": 0.9682, + "mean_token_accuracy": 0.7097042202949524, + "num_tokens": 57875419.0, + "step": 1917 + }, + { + "epoch": 0.15815947884884968, + "grad_norm": 2.0599312782287598, + "learning_rate": 5.2679307502061e-07, + "loss": 0.8146, + "mean_token_accuracy": 0.7428438663482666, + "num_tokens": 57904223.0, + "step": 1918 + }, + { + "epoch": 0.15824193947390122, + "grad_norm": 2.120936155319214, + "learning_rate": 5.270678757900522e-07, + "loss": 0.8111, + "mean_token_accuracy": 0.7432650327682495, + "num_tokens": 57932255.0, + "step": 1919 + }, + { + "epoch": 0.15832440009895274, + "grad_norm": 1.855492115020752, + "learning_rate": 5.273426765594944e-07, + "loss": 0.9115, + "mean_token_accuracy": 0.7297208309173584, + "num_tokens": 57968649.0, + "step": 1920 + }, + { + "epoch": 0.15840686072400428, + "grad_norm": 2.001593828201294, + "learning_rate": 5.276174773289365e-07, + "loss": 0.7849, + "mean_token_accuracy": 0.758399486541748, + "num_tokens": 57997965.0, + "step": 1921 + }, + { + "epoch": 0.15848932134905583, + "grad_norm": 2.0043067932128906, + "learning_rate": 5.278922780983786e-07, + "loss": 0.9653, + "mean_token_accuracy": 0.7096554040908813, + "num_tokens": 58030924.0, + "step": 1922 + }, + { + "epoch": 0.15857178197410737, + "grad_norm": 1.9571641683578491, + "learning_rate": 5.281670788678208e-07, + "loss": 0.8765, + "mean_token_accuracy": 0.7267829179763794, + "num_tokens": 58060248.0, + "step": 1923 + }, + { + "epoch": 0.1586542425991589, + "grad_norm": 1.9210771322250366, + "learning_rate": 5.28441879637263e-07, + "loss": 0.7762, + "mean_token_accuracy": 0.7519308924674988, + "num_tokens": 58089309.0, + "step": 1924 + }, + { + "epoch": 0.15873670322421043, + "grad_norm": 2.18916392326355, + "learning_rate": 5.287166804067052e-07, + "loss": 0.8448, + "mean_token_accuracy": 0.736831545829773, + "num_tokens": 58114722.0, + "step": 1925 + }, + { + "epoch": 0.15881916384926198, + "grad_norm": 1.8919309377670288, + "learning_rate": 5.289914811761472e-07, + "loss": 1.002, + "mean_token_accuracy": 0.6923055648803711, + "num_tokens": 58151318.0, + "step": 1926 + }, + { + "epoch": 0.15890162447431352, + "grad_norm": 1.9263653755187988, + "learning_rate": 5.292662819455894e-07, + "loss": 0.739, + "mean_token_accuracy": 0.7749878764152527, + "num_tokens": 58181158.0, + "step": 1927 + }, + { + "epoch": 0.15898408509936504, + "grad_norm": 2.07185697555542, + "learning_rate": 5.295410827150316e-07, + "loss": 0.9154, + "mean_token_accuracy": 0.7234501838684082, + "num_tokens": 58213845.0, + "step": 1928 + }, + { + "epoch": 0.15906654572441659, + "grad_norm": 1.9195197820663452, + "learning_rate": 5.298158834844737e-07, + "loss": 0.7303, + "mean_token_accuracy": 0.7722153067588806, + "num_tokens": 58244468.0, + "step": 1929 + }, + { + "epoch": 0.15914900634946813, + "grad_norm": 1.8426156044006348, + "learning_rate": 5.300906842539159e-07, + "loss": 0.8915, + "mean_token_accuracy": 0.7272101044654846, + "num_tokens": 58280796.0, + "step": 1930 + }, + { + "epoch": 0.15923146697451968, + "grad_norm": 2.0302817821502686, + "learning_rate": 5.303654850233581e-07, + "loss": 0.9015, + "mean_token_accuracy": 0.7275641560554504, + "num_tokens": 58309093.0, + "step": 1931 + }, + { + "epoch": 0.1593139275995712, + "grad_norm": 2.057421922683716, + "learning_rate": 5.306402857928001e-07, + "loss": 0.9317, + "mean_token_accuracy": 0.7206166982650757, + "num_tokens": 58340079.0, + "step": 1932 + }, + { + "epoch": 0.15939638822462274, + "grad_norm": 1.8274613618850708, + "learning_rate": 5.309150865622424e-07, + "loss": 0.8986, + "mean_token_accuracy": 0.7323462963104248, + "num_tokens": 58376687.0, + "step": 1933 + }, + { + "epoch": 0.15947884884967428, + "grad_norm": 2.225940704345703, + "learning_rate": 5.311898873316845e-07, + "loss": 0.8885, + "mean_token_accuracy": 0.7249643802642822, + "num_tokens": 58403966.0, + "step": 1934 + }, + { + "epoch": 0.15956130947472583, + "grad_norm": 2.0699362754821777, + "learning_rate": 5.314646881011267e-07, + "loss": 0.9364, + "mean_token_accuracy": 0.7033332586288452, + "num_tokens": 58433129.0, + "step": 1935 + }, + { + "epoch": 0.15964377009977734, + "grad_norm": 1.8783340454101562, + "learning_rate": 5.317394888705688e-07, + "loss": 0.8043, + "mean_token_accuracy": 0.7575483322143555, + "num_tokens": 58472149.0, + "step": 1936 + }, + { + "epoch": 0.1597262307248289, + "grad_norm": 2.0032167434692383, + "learning_rate": 5.320142896400109e-07, + "loss": 0.864, + "mean_token_accuracy": 0.7426046133041382, + "num_tokens": 58503171.0, + "step": 1937 + }, + { + "epoch": 0.15980869134988043, + "grad_norm": 1.960252046585083, + "learning_rate": 5.322890904094532e-07, + "loss": 0.95, + "mean_token_accuracy": 0.7103514671325684, + "num_tokens": 58534357.0, + "step": 1938 + }, + { + "epoch": 0.15989115197493198, + "grad_norm": 2.1383235454559326, + "learning_rate": 5.325638911788953e-07, + "loss": 1.0905, + "mean_token_accuracy": 0.6848423480987549, + "num_tokens": 58564082.0, + "step": 1939 + }, + { + "epoch": 0.1599736125999835, + "grad_norm": 2.2121078968048096, + "learning_rate": 5.328386919483374e-07, + "loss": 0.7656, + "mean_token_accuracy": 0.7574381232261658, + "num_tokens": 58592379.0, + "step": 1940 + }, + { + "epoch": 0.16005607322503504, + "grad_norm": 1.9712953567504883, + "learning_rate": 5.331134927177796e-07, + "loss": 0.8955, + "mean_token_accuracy": 0.7268815636634827, + "num_tokens": 58626324.0, + "step": 1941 + }, + { + "epoch": 0.16013853385008658, + "grad_norm": 2.1655895709991455, + "learning_rate": 5.333882934872217e-07, + "loss": 0.8325, + "mean_token_accuracy": 0.7391678690910339, + "num_tokens": 58654834.0, + "step": 1942 + }, + { + "epoch": 0.16022099447513813, + "grad_norm": 2.007699489593506, + "learning_rate": 5.336630942566639e-07, + "loss": 0.7791, + "mean_token_accuracy": 0.765920877456665, + "num_tokens": 58685461.0, + "step": 1943 + }, + { + "epoch": 0.16030345510018965, + "grad_norm": 1.7511433362960815, + "learning_rate": 5.339378950261061e-07, + "loss": 0.7774, + "mean_token_accuracy": 0.7723842859268188, + "num_tokens": 58720592.0, + "step": 1944 + }, + { + "epoch": 0.1603859157252412, + "grad_norm": 2.202033758163452, + "learning_rate": 5.342126957955482e-07, + "loss": 0.7089, + "mean_token_accuracy": 0.7675821781158447, + "num_tokens": 58745806.0, + "step": 1945 + }, + { + "epoch": 0.16046837635029274, + "grad_norm": 2.0631003379821777, + "learning_rate": 5.344874965649903e-07, + "loss": 0.8508, + "mean_token_accuracy": 0.7353035807609558, + "num_tokens": 58775101.0, + "step": 1946 + }, + { + "epoch": 0.16055083697534428, + "grad_norm": 2.2722153663635254, + "learning_rate": 5.347622973344326e-07, + "loss": 1.1173, + "mean_token_accuracy": 0.6772485971450806, + "num_tokens": 58804230.0, + "step": 1947 + }, + { + "epoch": 0.16063329760039582, + "grad_norm": 1.824138879776001, + "learning_rate": 5.350370981038747e-07, + "loss": 0.691, + "mean_token_accuracy": 0.7798889875411987, + "num_tokens": 58837864.0, + "step": 1948 + }, + { + "epoch": 0.16071575822544734, + "grad_norm": 2.12318754196167, + "learning_rate": 5.353118988733169e-07, + "loss": 0.8079, + "mean_token_accuracy": 0.7391672134399414, + "num_tokens": 58869892.0, + "step": 1949 + }, + { + "epoch": 0.1607982188504989, + "grad_norm": 1.9208505153656006, + "learning_rate": 5.355866996427589e-07, + "loss": 0.7907, + "mean_token_accuracy": 0.7559675574302673, + "num_tokens": 58901982.0, + "step": 1950 + }, + { + "epoch": 0.16088067947555043, + "grad_norm": 2.0349488258361816, + "learning_rate": 5.358615004122011e-07, + "loss": 0.8803, + "mean_token_accuracy": 0.7204020023345947, + "num_tokens": 58932061.0, + "step": 1951 + }, + { + "epoch": 0.16096314010060198, + "grad_norm": 2.4448680877685547, + "learning_rate": 5.361363011816434e-07, + "loss": 0.9146, + "mean_token_accuracy": 0.7237802147865295, + "num_tokens": 58955439.0, + "step": 1952 + }, + { + "epoch": 0.1610456007256535, + "grad_norm": 2.0451104640960693, + "learning_rate": 5.364111019510854e-07, + "loss": 0.8919, + "mean_token_accuracy": 0.7314690947532654, + "num_tokens": 58986417.0, + "step": 1953 + }, + { + "epoch": 0.16112806135070504, + "grad_norm": 2.04472017288208, + "learning_rate": 5.366859027205276e-07, + "loss": 0.9563, + "mean_token_accuracy": 0.7192728519439697, + "num_tokens": 59015866.0, + "step": 1954 + }, + { + "epoch": 0.16121052197575658, + "grad_norm": 2.379957914352417, + "learning_rate": 5.369607034899698e-07, + "loss": 1.0102, + "mean_token_accuracy": 0.6907830238342285, + "num_tokens": 59039399.0, + "step": 1955 + }, + { + "epoch": 0.16129298260080813, + "grad_norm": 2.029301404953003, + "learning_rate": 5.372355042594118e-07, + "loss": 0.975, + "mean_token_accuracy": 0.713354229927063, + "num_tokens": 59073623.0, + "step": 1956 + }, + { + "epoch": 0.16137544322585964, + "grad_norm": 2.0076956748962402, + "learning_rate": 5.375103050288541e-07, + "loss": 0.8001, + "mean_token_accuracy": 0.7527633905410767, + "num_tokens": 59102656.0, + "step": 1957 + }, + { + "epoch": 0.1614579038509112, + "grad_norm": 2.0983622074127197, + "learning_rate": 5.377851057982962e-07, + "loss": 0.9047, + "mean_token_accuracy": 0.7274239659309387, + "num_tokens": 59131133.0, + "step": 1958 + }, + { + "epoch": 0.16154036447596273, + "grad_norm": 2.010514497756958, + "learning_rate": 5.380599065677384e-07, + "loss": 0.7274, + "mean_token_accuracy": 0.7689942717552185, + "num_tokens": 59161360.0, + "step": 1959 + }, + { + "epoch": 0.16162282510101428, + "grad_norm": 1.961027979850769, + "learning_rate": 5.383347073371805e-07, + "loss": 0.8787, + "mean_token_accuracy": 0.7299865484237671, + "num_tokens": 59196351.0, + "step": 1960 + }, + { + "epoch": 0.1617052857260658, + "grad_norm": 1.7049263715744019, + "learning_rate": 5.386095081066226e-07, + "loss": 0.723, + "mean_token_accuracy": 0.7646509408950806, + "num_tokens": 59233178.0, + "step": 1961 + }, + { + "epoch": 0.16178774635111734, + "grad_norm": 1.80672025680542, + "learning_rate": 5.388843088760649e-07, + "loss": 0.8295, + "mean_token_accuracy": 0.7456047534942627, + "num_tokens": 59269333.0, + "step": 1962 + }, + { + "epoch": 0.16187020697616888, + "grad_norm": 2.3800177574157715, + "learning_rate": 5.39159109645507e-07, + "loss": 0.958, + "mean_token_accuracy": 0.708531379699707, + "num_tokens": 59294854.0, + "step": 1963 + }, + { + "epoch": 0.16195266760122043, + "grad_norm": 1.8908379077911377, + "learning_rate": 5.394339104149491e-07, + "loss": 0.9318, + "mean_token_accuracy": 0.7228379249572754, + "num_tokens": 59328819.0, + "step": 1964 + }, + { + "epoch": 0.16203512822627195, + "grad_norm": 1.8992840051651, + "learning_rate": 5.397087111843913e-07, + "loss": 0.6676, + "mean_token_accuracy": 0.7890290021896362, + "num_tokens": 59359907.0, + "step": 1965 + }, + { + "epoch": 0.1621175888513235, + "grad_norm": 2.0477347373962402, + "learning_rate": 5.399835119538335e-07, + "loss": 0.8078, + "mean_token_accuracy": 0.7439822554588318, + "num_tokens": 59386494.0, + "step": 1966 + }, + { + "epoch": 0.16220004947637504, + "grad_norm": 2.3125786781311035, + "learning_rate": 5.402583127232756e-07, + "loss": 0.8935, + "mean_token_accuracy": 0.727706253528595, + "num_tokens": 59410210.0, + "step": 1967 + }, + { + "epoch": 0.16228251010142658, + "grad_norm": 1.8434059619903564, + "learning_rate": 5.405331134927178e-07, + "loss": 0.9014, + "mean_token_accuracy": 0.73048996925354, + "num_tokens": 59446995.0, + "step": 1968 + }, + { + "epoch": 0.1623649707264781, + "grad_norm": 2.0409255027770996, + "learning_rate": 5.408079142621599e-07, + "loss": 0.7184, + "mean_token_accuracy": 0.7652630805969238, + "num_tokens": 59476371.0, + "step": 1969 + }, + { + "epoch": 0.16244743135152964, + "grad_norm": 2.1856000423431396, + "learning_rate": 5.41082715031602e-07, + "loss": 1.0282, + "mean_token_accuracy": 0.6883177161216736, + "num_tokens": 59506699.0, + "step": 1970 + }, + { + "epoch": 0.1625298919765812, + "grad_norm": 1.9696986675262451, + "learning_rate": 5.413575158010443e-07, + "loss": 0.8877, + "mean_token_accuracy": 0.7258217334747314, + "num_tokens": 59541649.0, + "step": 1971 + }, + { + "epoch": 0.16261235260163273, + "grad_norm": 1.9656505584716797, + "learning_rate": 5.416323165704864e-07, + "loss": 0.7882, + "mean_token_accuracy": 0.7550392150878906, + "num_tokens": 59571800.0, + "step": 1972 + }, + { + "epoch": 0.16269481322668425, + "grad_norm": 2.0946640968322754, + "learning_rate": 5.419071173399285e-07, + "loss": 0.7712, + "mean_token_accuracy": 0.7573607563972473, + "num_tokens": 59601773.0, + "step": 1973 + }, + { + "epoch": 0.1627772738517358, + "grad_norm": 1.8225822448730469, + "learning_rate": 5.421819181093706e-07, + "loss": 0.7257, + "mean_token_accuracy": 0.7708569765090942, + "num_tokens": 59635994.0, + "step": 1974 + }, + { + "epoch": 0.16285973447678734, + "grad_norm": 2.2230632305145264, + "learning_rate": 5.424567188788128e-07, + "loss": 0.8076, + "mean_token_accuracy": 0.7427929639816284, + "num_tokens": 59660343.0, + "step": 1975 + }, + { + "epoch": 0.16294219510183888, + "grad_norm": 2.2332403659820557, + "learning_rate": 5.427315196482551e-07, + "loss": 1.1474, + "mean_token_accuracy": 0.6621050238609314, + "num_tokens": 59689602.0, + "step": 1976 + }, + { + "epoch": 0.1630246557268904, + "grad_norm": 2.1379239559173584, + "learning_rate": 5.430063204176971e-07, + "loss": 1.0259, + "mean_token_accuracy": 0.6945847272872925, + "num_tokens": 59721492.0, + "step": 1977 + }, + { + "epoch": 0.16310711635194194, + "grad_norm": 2.0960299968719482, + "learning_rate": 5.432811211871393e-07, + "loss": 0.7946, + "mean_token_accuracy": 0.7530333995819092, + "num_tokens": 59748751.0, + "step": 1978 + }, + { + "epoch": 0.1631895769769935, + "grad_norm": 2.1915106773376465, + "learning_rate": 5.435559219565815e-07, + "loss": 0.9223, + "mean_token_accuracy": 0.7196632623672485, + "num_tokens": 59778581.0, + "step": 1979 + }, + { + "epoch": 0.16327203760204503, + "grad_norm": 1.9625204801559448, + "learning_rate": 5.438307227260236e-07, + "loss": 0.9434, + "mean_token_accuracy": 0.7195153832435608, + "num_tokens": 59814489.0, + "step": 1980 + }, + { + "epoch": 0.16335449822709655, + "grad_norm": 1.8598952293395996, + "learning_rate": 5.441055234954658e-07, + "loss": 0.83, + "mean_token_accuracy": 0.749034583568573, + "num_tokens": 59848205.0, + "step": 1981 + }, + { + "epoch": 0.1634369588521481, + "grad_norm": 2.3018271923065186, + "learning_rate": 5.443803242649079e-07, + "loss": 0.9788, + "mean_token_accuracy": 0.7188276052474976, + "num_tokens": 59875555.0, + "step": 1982 + }, + { + "epoch": 0.16351941947719964, + "grad_norm": 2.1396119594573975, + "learning_rate": 5.4465512503435e-07, + "loss": 0.8435, + "mean_token_accuracy": 0.7353528738021851, + "num_tokens": 59903349.0, + "step": 1983 + }, + { + "epoch": 0.16360188010225118, + "grad_norm": 1.772236704826355, + "learning_rate": 5.449299258037922e-07, + "loss": 0.786, + "mean_token_accuracy": 0.7583116888999939, + "num_tokens": 59943294.0, + "step": 1984 + }, + { + "epoch": 0.1636843407273027, + "grad_norm": 1.9327266216278076, + "learning_rate": 5.452047265732344e-07, + "loss": 0.8423, + "mean_token_accuracy": 0.7397441864013672, + "num_tokens": 59974851.0, + "step": 1985 + }, + { + "epoch": 0.16376680135235425, + "grad_norm": 2.082601308822632, + "learning_rate": 5.454795273426766e-07, + "loss": 0.9649, + "mean_token_accuracy": 0.7024078369140625, + "num_tokens": 60005509.0, + "step": 1986 + }, + { + "epoch": 0.1638492619774058, + "grad_norm": 2.2576098442077637, + "learning_rate": 5.457543281121186e-07, + "loss": 0.9553, + "mean_token_accuracy": 0.7119102478027344, + "num_tokens": 60030850.0, + "step": 1987 + }, + { + "epoch": 0.16393172260245734, + "grad_norm": 2.0398693084716797, + "learning_rate": 5.460291288815608e-07, + "loss": 0.8146, + "mean_token_accuracy": 0.7529458403587341, + "num_tokens": 60060332.0, + "step": 1988 + }, + { + "epoch": 0.16401418322750885, + "grad_norm": 1.9109936952590942, + "learning_rate": 5.46303929651003e-07, + "loss": 0.7238, + "mean_token_accuracy": 0.7576925754547119, + "num_tokens": 60090697.0, + "step": 1989 + }, + { + "epoch": 0.1640966438525604, + "grad_norm": 1.8415652513504028, + "learning_rate": 5.465787304204451e-07, + "loss": 0.7533, + "mean_token_accuracy": 0.7639830112457275, + "num_tokens": 60124166.0, + "step": 1990 + }, + { + "epoch": 0.16417910447761194, + "grad_norm": 1.991509199142456, + "learning_rate": 5.468535311898873e-07, + "loss": 0.8969, + "mean_token_accuracy": 0.7288036346435547, + "num_tokens": 60154275.0, + "step": 1991 + }, + { + "epoch": 0.1642615651026635, + "grad_norm": 2.0430233478546143, + "learning_rate": 5.471283319593295e-07, + "loss": 0.8682, + "mean_token_accuracy": 0.7280598282814026, + "num_tokens": 60185180.0, + "step": 1992 + }, + { + "epoch": 0.164344025727715, + "grad_norm": 1.981476068496704, + "learning_rate": 5.474031327287716e-07, + "loss": 0.9522, + "mean_token_accuracy": 0.7153571844100952, + "num_tokens": 60218864.0, + "step": 1993 + }, + { + "epoch": 0.16442648635276655, + "grad_norm": 2.2509353160858154, + "learning_rate": 5.476779334982138e-07, + "loss": 1.0733, + "mean_token_accuracy": 0.6814702749252319, + "num_tokens": 60247734.0, + "step": 1994 + }, + { + "epoch": 0.1645089469778181, + "grad_norm": 1.853586196899414, + "learning_rate": 5.479527342676559e-07, + "loss": 0.7473, + "mean_token_accuracy": 0.7683985233306885, + "num_tokens": 60280913.0, + "step": 1995 + }, + { + "epoch": 0.16459140760286964, + "grad_norm": 1.8198186159133911, + "learning_rate": 5.482275350370981e-07, + "loss": 0.8222, + "mean_token_accuracy": 0.7454094290733337, + "num_tokens": 60317263.0, + "step": 1996 + }, + { + "epoch": 0.16467386822792116, + "grad_norm": 2.42832088470459, + "learning_rate": 5.485023358065402e-07, + "loss": 1.0047, + "mean_token_accuracy": 0.6982740163803101, + "num_tokens": 60343295.0, + "step": 1997 + }, + { + "epoch": 0.1647563288529727, + "grad_norm": 2.2388651371002197, + "learning_rate": 5.487771365759823e-07, + "loss": 0.9561, + "mean_token_accuracy": 0.7095003724098206, + "num_tokens": 60373072.0, + "step": 1998 + }, + { + "epoch": 0.16483878947802424, + "grad_norm": 2.0901122093200684, + "learning_rate": 5.490519373454246e-07, + "loss": 0.8005, + "mean_token_accuracy": 0.7455300092697144, + "num_tokens": 60401004.0, + "step": 1999 + }, + { + "epoch": 0.1649212501030758, + "grad_norm": 2.170046329498291, + "learning_rate": 5.493267381148668e-07, + "loss": 0.9107, + "mean_token_accuracy": 0.7151706218719482, + "num_tokens": 60429174.0, + "step": 2000 + }, + { + "epoch": 0.1650037107281273, + "grad_norm": 2.167649745941162, + "learning_rate": 5.496015388843088e-07, + "loss": 0.8985, + "mean_token_accuracy": 0.7186275720596313, + "num_tokens": 60456135.0, + "step": 2001 + }, + { + "epoch": 0.16508617135317885, + "grad_norm": 1.8324894905090332, + "learning_rate": 5.49876339653751e-07, + "loss": 0.8787, + "mean_token_accuracy": 0.7288339138031006, + "num_tokens": 60492599.0, + "step": 2002 + }, + { + "epoch": 0.1651686319782304, + "grad_norm": 2.084960699081421, + "learning_rate": 5.501511404231931e-07, + "loss": 0.8746, + "mean_token_accuracy": 0.7316613793373108, + "num_tokens": 60523951.0, + "step": 2003 + }, + { + "epoch": 0.16525109260328194, + "grad_norm": 2.275813341140747, + "learning_rate": 5.504259411926353e-07, + "loss": 0.8696, + "mean_token_accuracy": 0.7358349561691284, + "num_tokens": 60550223.0, + "step": 2004 + }, + { + "epoch": 0.16533355322833346, + "grad_norm": 1.7807888984680176, + "learning_rate": 5.507007419620775e-07, + "loss": 0.777, + "mean_token_accuracy": 0.7450129985809326, + "num_tokens": 60586221.0, + "step": 2005 + }, + { + "epoch": 0.165416013853385, + "grad_norm": 2.071499824523926, + "learning_rate": 5.509755427315196e-07, + "loss": 0.9507, + "mean_token_accuracy": 0.7041199803352356, + "num_tokens": 60617612.0, + "step": 2006 + }, + { + "epoch": 0.16549847447843655, + "grad_norm": 2.0799973011016846, + "learning_rate": 5.512503435009617e-07, + "loss": 0.798, + "mean_token_accuracy": 0.7417788505554199, + "num_tokens": 60644923.0, + "step": 2007 + }, + { + "epoch": 0.1655809351034881, + "grad_norm": 2.3604207038879395, + "learning_rate": 5.51525144270404e-07, + "loss": 0.887, + "mean_token_accuracy": 0.7430332899093628, + "num_tokens": 60671750.0, + "step": 2008 + }, + { + "epoch": 0.1656633957285396, + "grad_norm": 2.221208095550537, + "learning_rate": 5.517999450398461e-07, + "loss": 0.8009, + "mean_token_accuracy": 0.7364823222160339, + "num_tokens": 60696375.0, + "step": 2009 + }, + { + "epoch": 0.16574585635359115, + "grad_norm": 1.914542317390442, + "learning_rate": 5.520747458092883e-07, + "loss": 0.8548, + "mean_token_accuracy": 0.7364325523376465, + "num_tokens": 60731683.0, + "step": 2010 + }, + { + "epoch": 0.1658283169786427, + "grad_norm": 1.9694479703903198, + "learning_rate": 5.523495465787303e-07, + "loss": 0.8875, + "mean_token_accuracy": 0.7278320789337158, + "num_tokens": 60767738.0, + "step": 2011 + }, + { + "epoch": 0.16591077760369424, + "grad_norm": 1.9993512630462646, + "learning_rate": 5.526243473481725e-07, + "loss": 0.8843, + "mean_token_accuracy": 0.7344401478767395, + "num_tokens": 60799657.0, + "step": 2012 + }, + { + "epoch": 0.1659932382287458, + "grad_norm": 2.0464189052581787, + "learning_rate": 5.528991481176148e-07, + "loss": 0.9231, + "mean_token_accuracy": 0.7167290449142456, + "num_tokens": 60829620.0, + "step": 2013 + }, + { + "epoch": 0.1660756988537973, + "grad_norm": 2.015251398086548, + "learning_rate": 5.531739488870568e-07, + "loss": 0.8145, + "mean_token_accuracy": 0.7496448755264282, + "num_tokens": 60857764.0, + "step": 2014 + }, + { + "epoch": 0.16615815947884885, + "grad_norm": 2.1843066215515137, + "learning_rate": 5.53448749656499e-07, + "loss": 1.0097, + "mean_token_accuracy": 0.7050005793571472, + "num_tokens": 60890089.0, + "step": 2015 + }, + { + "epoch": 0.1662406201039004, + "grad_norm": 2.2176175117492676, + "learning_rate": 5.537235504259412e-07, + "loss": 0.998, + "mean_token_accuracy": 0.6991809606552124, + "num_tokens": 60919010.0, + "step": 2016 + }, + { + "epoch": 0.16632308072895194, + "grad_norm": 1.8947110176086426, + "learning_rate": 5.539983511953832e-07, + "loss": 0.8241, + "mean_token_accuracy": 0.7492181062698364, + "num_tokens": 60952369.0, + "step": 2017 + }, + { + "epoch": 0.16640554135400346, + "grad_norm": 1.9447247982025146, + "learning_rate": 5.542731519648255e-07, + "loss": 0.7706, + "mean_token_accuracy": 0.749796986579895, + "num_tokens": 60983427.0, + "step": 2018 + }, + { + "epoch": 0.166488001979055, + "grad_norm": 1.9669808149337769, + "learning_rate": 5.545479527342676e-07, + "loss": 0.793, + "mean_token_accuracy": 0.7551445960998535, + "num_tokens": 61014419.0, + "step": 2019 + }, + { + "epoch": 0.16657046260410655, + "grad_norm": 2.3485875129699707, + "learning_rate": 5.548227535037098e-07, + "loss": 0.9986, + "mean_token_accuracy": 0.7038121223449707, + "num_tokens": 61043560.0, + "step": 2020 + }, + { + "epoch": 0.1666529232291581, + "grad_norm": 1.8904876708984375, + "learning_rate": 5.550975542731519e-07, + "loss": 0.7855, + "mean_token_accuracy": 0.7604044079780579, + "num_tokens": 61075137.0, + "step": 2021 + }, + { + "epoch": 0.1667353838542096, + "grad_norm": 2.0439493656158447, + "learning_rate": 5.553723550425941e-07, + "loss": 0.7437, + "mean_token_accuracy": 0.7669874429702759, + "num_tokens": 61105554.0, + "step": 2022 + }, + { + "epoch": 0.16681784447926115, + "grad_norm": 2.3512203693389893, + "learning_rate": 5.556471558120363e-07, + "loss": 0.9859, + "mean_token_accuracy": 0.7027463316917419, + "num_tokens": 61132589.0, + "step": 2023 + }, + { + "epoch": 0.1669003051043127, + "grad_norm": 1.9928765296936035, + "learning_rate": 5.559219565814784e-07, + "loss": 0.8508, + "mean_token_accuracy": 0.7391364574432373, + "num_tokens": 61164984.0, + "step": 2024 + }, + { + "epoch": 0.16698276572936424, + "grad_norm": 2.042522668838501, + "learning_rate": 5.561967573509205e-07, + "loss": 0.6217, + "mean_token_accuracy": 0.7867984771728516, + "num_tokens": 61190505.0, + "step": 2025 + }, + { + "epoch": 0.16706522635441576, + "grad_norm": 1.944833517074585, + "learning_rate": 5.564715581203627e-07, + "loss": 0.889, + "mean_token_accuracy": 0.7253875136375427, + "num_tokens": 61220766.0, + "step": 2026 + }, + { + "epoch": 0.1671476869794673, + "grad_norm": 1.9553474187850952, + "learning_rate": 5.567463588898049e-07, + "loss": 0.7785, + "mean_token_accuracy": 0.7594171762466431, + "num_tokens": 61254594.0, + "step": 2027 + }, + { + "epoch": 0.16723014760451885, + "grad_norm": 2.2160277366638184, + "learning_rate": 5.57021159659247e-07, + "loss": 0.8742, + "mean_token_accuracy": 0.7243433594703674, + "num_tokens": 61279436.0, + "step": 2028 + }, + { + "epoch": 0.1673126082295704, + "grad_norm": 2.2746262550354004, + "learning_rate": 5.572959604286892e-07, + "loss": 1.0205, + "mean_token_accuracy": 0.6950541734695435, + "num_tokens": 61305169.0, + "step": 2029 + }, + { + "epoch": 0.1673950688546219, + "grad_norm": 2.1715450286865234, + "learning_rate": 5.575707611981313e-07, + "loss": 0.7175, + "mean_token_accuracy": 0.7731161713600159, + "num_tokens": 61334623.0, + "step": 2030 + }, + { + "epoch": 0.16747752947967345, + "grad_norm": 2.1257927417755127, + "learning_rate": 5.578455619675734e-07, + "loss": 0.9704, + "mean_token_accuracy": 0.7062625885009766, + "num_tokens": 61366251.0, + "step": 2031 + }, + { + "epoch": 0.167559990104725, + "grad_norm": 2.18967604637146, + "learning_rate": 5.581203627370157e-07, + "loss": 0.8382, + "mean_token_accuracy": 0.7361307740211487, + "num_tokens": 61392583.0, + "step": 2032 + }, + { + "epoch": 0.16764245072977654, + "grad_norm": 2.2945051193237305, + "learning_rate": 5.583951635064578e-07, + "loss": 0.8835, + "mean_token_accuracy": 0.7285972833633423, + "num_tokens": 61418633.0, + "step": 2033 + }, + { + "epoch": 0.16772491135482806, + "grad_norm": 1.8047685623168945, + "learning_rate": 5.586699642759e-07, + "loss": 0.7449, + "mean_token_accuracy": 0.7628445029258728, + "num_tokens": 61453878.0, + "step": 2034 + }, + { + "epoch": 0.1678073719798796, + "grad_norm": 1.9071847200393677, + "learning_rate": 5.58944765045342e-07, + "loss": 0.7315, + "mean_token_accuracy": 0.7676417827606201, + "num_tokens": 61483953.0, + "step": 2035 + }, + { + "epoch": 0.16788983260493115, + "grad_norm": 1.8893340826034546, + "learning_rate": 5.592195658147843e-07, + "loss": 0.8249, + "mean_token_accuracy": 0.7365633845329285, + "num_tokens": 61515115.0, + "step": 2036 + }, + { + "epoch": 0.1679722932299827, + "grad_norm": 1.7123967409133911, + "learning_rate": 5.594943665842265e-07, + "loss": 0.7148, + "mean_token_accuracy": 0.7765827178955078, + "num_tokens": 61553745.0, + "step": 2037 + }, + { + "epoch": 0.1680547538550342, + "grad_norm": 1.722883939743042, + "learning_rate": 5.597691673536685e-07, + "loss": 0.8154, + "mean_token_accuracy": 0.7511028051376343, + "num_tokens": 61592889.0, + "step": 2038 + }, + { + "epoch": 0.16813721448008576, + "grad_norm": 2.1659512519836426, + "learning_rate": 5.600439681231107e-07, + "loss": 0.993, + "mean_token_accuracy": 0.6986595988273621, + "num_tokens": 61621767.0, + "step": 2039 + }, + { + "epoch": 0.1682196751051373, + "grad_norm": 1.7638123035430908, + "learning_rate": 5.603187688925529e-07, + "loss": 0.952, + "mean_token_accuracy": 0.7154674530029297, + "num_tokens": 61659366.0, + "step": 2040 + }, + { + "epoch": 0.16830213573018885, + "grad_norm": 2.0302276611328125, + "learning_rate": 5.60593569661995e-07, + "loss": 0.8234, + "mean_token_accuracy": 0.7437499165534973, + "num_tokens": 61689538.0, + "step": 2041 + }, + { + "epoch": 0.16838459635524036, + "grad_norm": 2.182473659515381, + "learning_rate": 5.608683704314372e-07, + "loss": 0.8223, + "mean_token_accuracy": 0.7463316917419434, + "num_tokens": 61715605.0, + "step": 2042 + }, + { + "epoch": 0.1684670569802919, + "grad_norm": 2.2756898403167725, + "learning_rate": 5.611431712008793e-07, + "loss": 1.1026, + "mean_token_accuracy": 0.6778181791305542, + "num_tokens": 61744578.0, + "step": 2043 + }, + { + "epoch": 0.16854951760534345, + "grad_norm": 2.2024729251861572, + "learning_rate": 5.614179719703215e-07, + "loss": 0.8427, + "mean_token_accuracy": 0.7471648454666138, + "num_tokens": 61775433.0, + "step": 2044 + }, + { + "epoch": 0.168631978230395, + "grad_norm": 2.2736058235168457, + "learning_rate": 5.616927727397636e-07, + "loss": 0.7609, + "mean_token_accuracy": 0.7593288421630859, + "num_tokens": 61799783.0, + "step": 2045 + }, + { + "epoch": 0.1687144388554465, + "grad_norm": 2.545117139816284, + "learning_rate": 5.619675735092058e-07, + "loss": 0.8693, + "mean_token_accuracy": 0.72488933801651, + "num_tokens": 61819981.0, + "step": 2046 + }, + { + "epoch": 0.16879689948049806, + "grad_norm": 2.10841703414917, + "learning_rate": 5.62242374278648e-07, + "loss": 0.8547, + "mean_token_accuracy": 0.7343822121620178, + "num_tokens": 61848474.0, + "step": 2047 + }, + { + "epoch": 0.1688793601055496, + "grad_norm": 2.01162052154541, + "learning_rate": 5.625171750480901e-07, + "loss": 0.8588, + "mean_token_accuracy": 0.7301830053329468, + "num_tokens": 61881021.0, + "step": 2048 + }, + { + "epoch": 0.16896182073060115, + "grad_norm": 1.9424742460250854, + "learning_rate": 5.627919758175322e-07, + "loss": 0.9491, + "mean_token_accuracy": 0.7108215093612671, + "num_tokens": 61915342.0, + "step": 2049 + }, + { + "epoch": 0.16904428135565266, + "grad_norm": 2.2295022010803223, + "learning_rate": 5.630667765869745e-07, + "loss": 0.885, + "mean_token_accuracy": 0.7258465886116028, + "num_tokens": 61943381.0, + "step": 2050 + }, + { + "epoch": 0.1691267419807042, + "grad_norm": 2.346977710723877, + "learning_rate": 5.633415773564166e-07, + "loss": 0.8888, + "mean_token_accuracy": 0.7253333926200867, + "num_tokens": 61967452.0, + "step": 2051 + }, + { + "epoch": 0.16920920260575575, + "grad_norm": 2.153974771499634, + "learning_rate": 5.636163781258587e-07, + "loss": 0.9395, + "mean_token_accuracy": 0.7121273279190063, + "num_tokens": 62001299.0, + "step": 2052 + }, + { + "epoch": 0.1692916632308073, + "grad_norm": 2.3675341606140137, + "learning_rate": 5.638911788953009e-07, + "loss": 0.9761, + "mean_token_accuracy": 0.7145035266876221, + "num_tokens": 62025419.0, + "step": 2053 + }, + { + "epoch": 0.16937412385585882, + "grad_norm": 2.0733911991119385, + "learning_rate": 5.64165979664743e-07, + "loss": 0.7795, + "mean_token_accuracy": 0.7551143169403076, + "num_tokens": 62053380.0, + "step": 2054 + }, + { + "epoch": 0.16945658448091036, + "grad_norm": 2.0816144943237305, + "learning_rate": 5.644407804341852e-07, + "loss": 0.8057, + "mean_token_accuracy": 0.744344174861908, + "num_tokens": 62080470.0, + "step": 2055 + }, + { + "epoch": 0.1695390451059619, + "grad_norm": 2.325977087020874, + "learning_rate": 5.647155812036274e-07, + "loss": 0.8935, + "mean_token_accuracy": 0.7239512801170349, + "num_tokens": 62106278.0, + "step": 2056 + }, + { + "epoch": 0.16962150573101345, + "grad_norm": 1.8749223947525024, + "learning_rate": 5.649903819730695e-07, + "loss": 0.8217, + "mean_token_accuracy": 0.7476544380187988, + "num_tokens": 62140510.0, + "step": 2057 + }, + { + "epoch": 0.16970396635606497, + "grad_norm": 1.9842736721038818, + "learning_rate": 5.652651827425116e-07, + "loss": 0.9708, + "mean_token_accuracy": 0.7001789808273315, + "num_tokens": 62171239.0, + "step": 2058 + }, + { + "epoch": 0.1697864269811165, + "grad_norm": 1.985754132270813, + "learning_rate": 5.655399835119537e-07, + "loss": 0.8025, + "mean_token_accuracy": 0.7416729927062988, + "num_tokens": 62203251.0, + "step": 2059 + }, + { + "epoch": 0.16986888760616806, + "grad_norm": 1.9928865432739258, + "learning_rate": 5.65814784281396e-07, + "loss": 0.8714, + "mean_token_accuracy": 0.7296357154846191, + "num_tokens": 62236840.0, + "step": 2060 + }, + { + "epoch": 0.1699513482312196, + "grad_norm": 2.2564096450805664, + "learning_rate": 5.660895850508382e-07, + "loss": 0.8817, + "mean_token_accuracy": 0.729287326335907, + "num_tokens": 62264557.0, + "step": 2061 + }, + { + "epoch": 0.17003380885627112, + "grad_norm": 2.201629638671875, + "learning_rate": 5.663643858202802e-07, + "loss": 1.0622, + "mean_token_accuracy": 0.684874951839447, + "num_tokens": 62293453.0, + "step": 2062 + }, + { + "epoch": 0.17011626948132266, + "grad_norm": 2.0825464725494385, + "learning_rate": 5.666391865897224e-07, + "loss": 0.9165, + "mean_token_accuracy": 0.7205172181129456, + "num_tokens": 62320830.0, + "step": 2063 + }, + { + "epoch": 0.1701987301063742, + "grad_norm": 1.7849658727645874, + "learning_rate": 5.669139873591647e-07, + "loss": 0.7365, + "mean_token_accuracy": 0.7635064125061035, + "num_tokens": 62353637.0, + "step": 2064 + }, + { + "epoch": 0.17028119073142575, + "grad_norm": 2.1132423877716064, + "learning_rate": 5.671887881286067e-07, + "loss": 0.9008, + "mean_token_accuracy": 0.7276737689971924, + "num_tokens": 62383779.0, + "step": 2065 + }, + { + "epoch": 0.17036365135647727, + "grad_norm": 2.1376616954803467, + "learning_rate": 5.674635888980489e-07, + "loss": 0.9074, + "mean_token_accuracy": 0.7309685945510864, + "num_tokens": 62413142.0, + "step": 2066 + }, + { + "epoch": 0.17044611198152881, + "grad_norm": 2.355787515640259, + "learning_rate": 5.67738389667491e-07, + "loss": 0.9798, + "mean_token_accuracy": 0.701736569404602, + "num_tokens": 62438108.0, + "step": 2067 + }, + { + "epoch": 0.17052857260658036, + "grad_norm": 2.347902536392212, + "learning_rate": 5.680131904369332e-07, + "loss": 0.9009, + "mean_token_accuracy": 0.729655385017395, + "num_tokens": 62464537.0, + "step": 2068 + }, + { + "epoch": 0.1706110332316319, + "grad_norm": 1.9382588863372803, + "learning_rate": 5.682879912063754e-07, + "loss": 0.9978, + "mean_token_accuracy": 0.7149825096130371, + "num_tokens": 62499342.0, + "step": 2069 + }, + { + "epoch": 0.17069349385668342, + "grad_norm": 1.9039299488067627, + "learning_rate": 5.685627919758175e-07, + "loss": 0.8547, + "mean_token_accuracy": 0.7408872246742249, + "num_tokens": 62534063.0, + "step": 2070 + }, + { + "epoch": 0.17077595448173497, + "grad_norm": 2.1460928916931152, + "learning_rate": 5.688375927452597e-07, + "loss": 0.9949, + "mean_token_accuracy": 0.694998025894165, + "num_tokens": 62563436.0, + "step": 2071 + }, + { + "epoch": 0.1708584151067865, + "grad_norm": 2.0668859481811523, + "learning_rate": 5.691123935147018e-07, + "loss": 0.8806, + "mean_token_accuracy": 0.7314630746841431, + "num_tokens": 62597094.0, + "step": 2072 + }, + { + "epoch": 0.17094087573183805, + "grad_norm": 2.081535577774048, + "learning_rate": 5.693871942841439e-07, + "loss": 0.873, + "mean_token_accuracy": 0.7322946190834045, + "num_tokens": 62626134.0, + "step": 2073 + }, + { + "epoch": 0.17102333635688957, + "grad_norm": 1.8536404371261597, + "learning_rate": 5.696619950535862e-07, + "loss": 0.8361, + "mean_token_accuracy": 0.7356479167938232, + "num_tokens": 62662044.0, + "step": 2074 + }, + { + "epoch": 0.17110579698194112, + "grad_norm": 2.2132062911987305, + "learning_rate": 5.699367958230282e-07, + "loss": 0.9413, + "mean_token_accuracy": 0.7137727737426758, + "num_tokens": 62687663.0, + "step": 2075 + }, + { + "epoch": 0.17118825760699266, + "grad_norm": 2.466282844543457, + "learning_rate": 5.702115965924704e-07, + "loss": 1.0067, + "mean_token_accuracy": 0.7002744078636169, + "num_tokens": 62715684.0, + "step": 2076 + }, + { + "epoch": 0.1712707182320442, + "grad_norm": 2.6269803047180176, + "learning_rate": 5.704863973619126e-07, + "loss": 0.9602, + "mean_token_accuracy": 0.7102690935134888, + "num_tokens": 62737447.0, + "step": 2077 + }, + { + "epoch": 0.17135317885709575, + "grad_norm": 1.8833192586898804, + "learning_rate": 5.707611981313547e-07, + "loss": 0.8814, + "mean_token_accuracy": 0.7416027784347534, + "num_tokens": 62775545.0, + "step": 2078 + }, + { + "epoch": 0.17143563948214727, + "grad_norm": 2.2651829719543457, + "learning_rate": 5.710359989007969e-07, + "loss": 0.7654, + "mean_token_accuracy": 0.7623432874679565, + "num_tokens": 62801458.0, + "step": 2079 + }, + { + "epoch": 0.1715181001071988, + "grad_norm": 2.239570140838623, + "learning_rate": 5.713107996702391e-07, + "loss": 0.964, + "mean_token_accuracy": 0.7103276252746582, + "num_tokens": 62828229.0, + "step": 2080 + }, + { + "epoch": 0.17160056073225036, + "grad_norm": 2.0489699840545654, + "learning_rate": 5.715856004396812e-07, + "loss": 0.7939, + "mean_token_accuracy": 0.7686282396316528, + "num_tokens": 62856540.0, + "step": 2081 + }, + { + "epoch": 0.1716830213573019, + "grad_norm": 1.8878602981567383, + "learning_rate": 5.718604012091233e-07, + "loss": 0.7724, + "mean_token_accuracy": 0.7592417001724243, + "num_tokens": 62887869.0, + "step": 2082 + }, + { + "epoch": 0.17176548198235342, + "grad_norm": 1.9222311973571777, + "learning_rate": 5.721352019785655e-07, + "loss": 0.7381, + "mean_token_accuracy": 0.7569898366928101, + "num_tokens": 62917509.0, + "step": 2083 + }, + { + "epoch": 0.17184794260740496, + "grad_norm": 1.8991345167160034, + "learning_rate": 5.724100027480077e-07, + "loss": 0.7922, + "mean_token_accuracy": 0.7511452436447144, + "num_tokens": 62947520.0, + "step": 2084 + }, + { + "epoch": 0.1719304032324565, + "grad_norm": 2.0687098503112793, + "learning_rate": 5.726848035174499e-07, + "loss": 1.0057, + "mean_token_accuracy": 0.7023791074752808, + "num_tokens": 62983575.0, + "step": 2085 + }, + { + "epoch": 0.17201286385750805, + "grad_norm": 2.269911766052246, + "learning_rate": 5.729596042868919e-07, + "loss": 1.0036, + "mean_token_accuracy": 0.7043541073799133, + "num_tokens": 63010363.0, + "step": 2086 + }, + { + "epoch": 0.17209532448255957, + "grad_norm": 2.2662250995635986, + "learning_rate": 5.732344050563341e-07, + "loss": 0.9172, + "mean_token_accuracy": 0.7181608080863953, + "num_tokens": 63036237.0, + "step": 2087 + }, + { + "epoch": 0.17217778510761111, + "grad_norm": 1.8024505376815796, + "learning_rate": 5.735092058257764e-07, + "loss": 0.7527, + "mean_token_accuracy": 0.7614847421646118, + "num_tokens": 63069530.0, + "step": 2088 + }, + { + "epoch": 0.17226024573266266, + "grad_norm": 1.9476431608200073, + "learning_rate": 5.737840065952184e-07, + "loss": 0.9446, + "mean_token_accuracy": 0.7142715454101562, + "num_tokens": 63103573.0, + "step": 2089 + }, + { + "epoch": 0.1723427063577142, + "grad_norm": 2.196537494659424, + "learning_rate": 5.740588073646606e-07, + "loss": 0.9028, + "mean_token_accuracy": 0.7245672941207886, + "num_tokens": 63130225.0, + "step": 2090 + }, + { + "epoch": 0.17242516698276572, + "grad_norm": 2.0390822887420654, + "learning_rate": 5.743336081341027e-07, + "loss": 0.9054, + "mean_token_accuracy": 0.7154731154441833, + "num_tokens": 63162123.0, + "step": 2091 + }, + { + "epoch": 0.17250762760781727, + "grad_norm": 1.931343674659729, + "learning_rate": 5.746084089035448e-07, + "loss": 0.9539, + "mean_token_accuracy": 0.7124407887458801, + "num_tokens": 63198425.0, + "step": 2092 + }, + { + "epoch": 0.1725900882328688, + "grad_norm": 2.139134407043457, + "learning_rate": 5.748832096729871e-07, + "loss": 0.9377, + "mean_token_accuracy": 0.7177853584289551, + "num_tokens": 63225612.0, + "step": 2093 + }, + { + "epoch": 0.17267254885792035, + "grad_norm": 2.161769151687622, + "learning_rate": 5.751580104424292e-07, + "loss": 0.8431, + "mean_token_accuracy": 0.7379804849624634, + "num_tokens": 63252067.0, + "step": 2094 + }, + { + "epoch": 0.17275500948297187, + "grad_norm": 2.6013288497924805, + "learning_rate": 5.754328112118714e-07, + "loss": 0.9603, + "mean_token_accuracy": 0.707880437374115, + "num_tokens": 63273209.0, + "step": 2095 + }, + { + "epoch": 0.17283747010802342, + "grad_norm": 1.9065011739730835, + "learning_rate": 5.757076119813135e-07, + "loss": 0.7371, + "mean_token_accuracy": 0.7753681540489197, + "num_tokens": 63303188.0, + "step": 2096 + }, + { + "epoch": 0.17291993073307496, + "grad_norm": 1.78939950466156, + "learning_rate": 5.759824127507557e-07, + "loss": 0.8642, + "mean_token_accuracy": 0.7327036261558533, + "num_tokens": 63341136.0, + "step": 2097 + }, + { + "epoch": 0.1730023913581265, + "grad_norm": 2.1242637634277344, + "learning_rate": 5.762572135201979e-07, + "loss": 0.8558, + "mean_token_accuracy": 0.7380306720733643, + "num_tokens": 63369204.0, + "step": 2098 + }, + { + "epoch": 0.17308485198317802, + "grad_norm": 2.53189754486084, + "learning_rate": 5.765320142896399e-07, + "loss": 1.1578, + "mean_token_accuracy": 0.6616733074188232, + "num_tokens": 63396409.0, + "step": 2099 + }, + { + "epoch": 0.17316731260822957, + "grad_norm": 2.218925714492798, + "learning_rate": 5.768068150590821e-07, + "loss": 0.8419, + "mean_token_accuracy": 0.7245692014694214, + "num_tokens": 63422766.0, + "step": 2100 + }, + { + "epoch": 0.1732497732332811, + "grad_norm": 2.0786736011505127, + "learning_rate": 5.770816158285243e-07, + "loss": 0.7381, + "mean_token_accuracy": 0.7654886841773987, + "num_tokens": 63453104.0, + "step": 2101 + }, + { + "epoch": 0.17333223385833266, + "grad_norm": 2.0552146434783936, + "learning_rate": 5.773564165979665e-07, + "loss": 0.832, + "mean_token_accuracy": 0.7401512861251831, + "num_tokens": 63483805.0, + "step": 2102 + }, + { + "epoch": 0.17341469448338417, + "grad_norm": 2.1660187244415283, + "learning_rate": 5.776312173674086e-07, + "loss": 0.9202, + "mean_token_accuracy": 0.7223033905029297, + "num_tokens": 63511909.0, + "step": 2103 + }, + { + "epoch": 0.17349715510843572, + "grad_norm": 1.9367878437042236, + "learning_rate": 5.779060181368508e-07, + "loss": 0.899, + "mean_token_accuracy": 0.7278795838356018, + "num_tokens": 63546927.0, + "step": 2104 + }, + { + "epoch": 0.17357961573348726, + "grad_norm": 2.054607391357422, + "learning_rate": 5.781808189062929e-07, + "loss": 0.7512, + "mean_token_accuracy": 0.764183759689331, + "num_tokens": 63576765.0, + "step": 2105 + }, + { + "epoch": 0.1736620763585388, + "grad_norm": 2.707493305206299, + "learning_rate": 5.78455619675735e-07, + "loss": 0.9543, + "mean_token_accuracy": 0.7090968489646912, + "num_tokens": 63600371.0, + "step": 2106 + }, + { + "epoch": 0.17374453698359033, + "grad_norm": 1.9019334316253662, + "learning_rate": 5.787304204451772e-07, + "loss": 0.8431, + "mean_token_accuracy": 0.737576961517334, + "num_tokens": 63634615.0, + "step": 2107 + }, + { + "epoch": 0.17382699760864187, + "grad_norm": 2.074748992919922, + "learning_rate": 5.790052212146194e-07, + "loss": 0.9398, + "mean_token_accuracy": 0.7226858139038086, + "num_tokens": 63662630.0, + "step": 2108 + }, + { + "epoch": 0.17390945823369341, + "grad_norm": 2.2980594635009766, + "learning_rate": 5.792800219840615e-07, + "loss": 0.9037, + "mean_token_accuracy": 0.7208499312400818, + "num_tokens": 63690191.0, + "step": 2109 + }, + { + "epoch": 0.17399191885874496, + "grad_norm": 2.0972182750701904, + "learning_rate": 5.795548227535036e-07, + "loss": 0.7601, + "mean_token_accuracy": 0.7583439946174622, + "num_tokens": 63717642.0, + "step": 2110 + }, + { + "epoch": 0.17407437948379648, + "grad_norm": 1.9392437934875488, + "learning_rate": 5.798296235229459e-07, + "loss": 0.9681, + "mean_token_accuracy": 0.7016993761062622, + "num_tokens": 63753955.0, + "step": 2111 + }, + { + "epoch": 0.17415684010884802, + "grad_norm": 1.9618514776229858, + "learning_rate": 5.801044242923881e-07, + "loss": 0.9853, + "mean_token_accuracy": 0.6959215402603149, + "num_tokens": 63789263.0, + "step": 2112 + }, + { + "epoch": 0.17423930073389957, + "grad_norm": 1.7641563415527344, + "learning_rate": 5.803792250618301e-07, + "loss": 0.7725, + "mean_token_accuracy": 0.7603769302368164, + "num_tokens": 63825835.0, + "step": 2113 + }, + { + "epoch": 0.1743217613589511, + "grad_norm": 2.2060177326202393, + "learning_rate": 5.806540258312723e-07, + "loss": 0.6926, + "mean_token_accuracy": 0.7782909274101257, + "num_tokens": 63850707.0, + "step": 2114 + }, + { + "epoch": 0.17440422198400263, + "grad_norm": 1.9163131713867188, + "learning_rate": 5.809288266007144e-07, + "loss": 0.9505, + "mean_token_accuracy": 0.7104604244232178, + "num_tokens": 63885777.0, + "step": 2115 + }, + { + "epoch": 0.17448668260905417, + "grad_norm": 1.9071578979492188, + "learning_rate": 5.812036273701566e-07, + "loss": 0.773, + "mean_token_accuracy": 0.7605434060096741, + "num_tokens": 63917687.0, + "step": 2116 + }, + { + "epoch": 0.17456914323410572, + "grad_norm": 2.246582269668579, + "learning_rate": 5.814784281395988e-07, + "loss": 0.9109, + "mean_token_accuracy": 0.7253592014312744, + "num_tokens": 63942865.0, + "step": 2117 + }, + { + "epoch": 0.17465160385915726, + "grad_norm": 2.120361328125, + "learning_rate": 5.817532289090409e-07, + "loss": 0.9664, + "mean_token_accuracy": 0.7129337787628174, + "num_tokens": 63973146.0, + "step": 2118 + }, + { + "epoch": 0.17473406448420878, + "grad_norm": 2.0296313762664795, + "learning_rate": 5.820280296784831e-07, + "loss": 0.7162, + "mean_token_accuracy": 0.7657577395439148, + "num_tokens": 63999856.0, + "step": 2119 + }, + { + "epoch": 0.17481652510926032, + "grad_norm": 1.786117672920227, + "learning_rate": 5.823028304479252e-07, + "loss": 0.5761, + "mean_token_accuracy": 0.8086272478103638, + "num_tokens": 64031537.0, + "step": 2120 + }, + { + "epoch": 0.17489898573431187, + "grad_norm": 1.9584600925445557, + "learning_rate": 5.825776312173674e-07, + "loss": 0.8012, + "mean_token_accuracy": 0.7468841075897217, + "num_tokens": 64062868.0, + "step": 2121 + }, + { + "epoch": 0.1749814463593634, + "grad_norm": 2.3815276622772217, + "learning_rate": 5.828524319868096e-07, + "loss": 0.9733, + "mean_token_accuracy": 0.699030339717865, + "num_tokens": 64090324.0, + "step": 2122 + }, + { + "epoch": 0.17506390698441493, + "grad_norm": 1.906366229057312, + "learning_rate": 5.831272327562516e-07, + "loss": 0.8555, + "mean_token_accuracy": 0.7408373951911926, + "num_tokens": 64127253.0, + "step": 2123 + }, + { + "epoch": 0.17514636760946647, + "grad_norm": 2.273905038833618, + "learning_rate": 5.834020335256938e-07, + "loss": 0.9568, + "mean_token_accuracy": 0.7128313779830933, + "num_tokens": 64153948.0, + "step": 2124 + }, + { + "epoch": 0.17522882823451802, + "grad_norm": 2.0618810653686523, + "learning_rate": 5.836768342951361e-07, + "loss": 0.8804, + "mean_token_accuracy": 0.7319632172584534, + "num_tokens": 64182047.0, + "step": 2125 + }, + { + "epoch": 0.17531128885956956, + "grad_norm": 2.0873146057128906, + "learning_rate": 5.839516350645781e-07, + "loss": 0.9727, + "mean_token_accuracy": 0.7096214294433594, + "num_tokens": 64209502.0, + "step": 2126 + }, + { + "epoch": 0.17539374948462108, + "grad_norm": 2.061264991760254, + "learning_rate": 5.842264358340203e-07, + "loss": 0.854, + "mean_token_accuracy": 0.7422325015068054, + "num_tokens": 64238826.0, + "step": 2127 + }, + { + "epoch": 0.17547621010967263, + "grad_norm": 2.2294809818267822, + "learning_rate": 5.845012366034625e-07, + "loss": 1.0014, + "mean_token_accuracy": 0.6985730528831482, + "num_tokens": 64267661.0, + "step": 2128 + }, + { + "epoch": 0.17555867073472417, + "grad_norm": 2.278655529022217, + "learning_rate": 5.847760373729046e-07, + "loss": 0.8936, + "mean_token_accuracy": 0.7327805757522583, + "num_tokens": 64292232.0, + "step": 2129 + }, + { + "epoch": 0.17564113135977572, + "grad_norm": 1.995100498199463, + "learning_rate": 5.850508381423468e-07, + "loss": 1.0184, + "mean_token_accuracy": 0.6975722908973694, + "num_tokens": 64325197.0, + "step": 2130 + }, + { + "epoch": 0.17572359198482723, + "grad_norm": 2.006822109222412, + "learning_rate": 5.853256389117889e-07, + "loss": 0.9501, + "mean_token_accuracy": 0.7170014381408691, + "num_tokens": 64358034.0, + "step": 2131 + }, + { + "epoch": 0.17580605260987878, + "grad_norm": 2.1318795680999756, + "learning_rate": 5.856004396812311e-07, + "loss": 0.9618, + "mean_token_accuracy": 0.7099152207374573, + "num_tokens": 64385409.0, + "step": 2132 + }, + { + "epoch": 0.17588851323493032, + "grad_norm": 2.071352481842041, + "learning_rate": 5.858752404506732e-07, + "loss": 0.7486, + "mean_token_accuracy": 0.7554016709327698, + "num_tokens": 64415386.0, + "step": 2133 + }, + { + "epoch": 0.17597097385998187, + "grad_norm": 2.0298116207122803, + "learning_rate": 5.861500412201153e-07, + "loss": 0.9813, + "mean_token_accuracy": 0.7041655778884888, + "num_tokens": 64446214.0, + "step": 2134 + }, + { + "epoch": 0.17605343448503338, + "grad_norm": 2.2863216400146484, + "learning_rate": 5.864248419895576e-07, + "loss": 0.9129, + "mean_token_accuracy": 0.720211386680603, + "num_tokens": 64469974.0, + "step": 2135 + }, + { + "epoch": 0.17613589511008493, + "grad_norm": 2.090453863143921, + "learning_rate": 5.866996427589998e-07, + "loss": 0.9015, + "mean_token_accuracy": 0.7244338989257812, + "num_tokens": 64498293.0, + "step": 2136 + }, + { + "epoch": 0.17621835573513647, + "grad_norm": 1.9494556188583374, + "learning_rate": 5.869744435284418e-07, + "loss": 0.8404, + "mean_token_accuracy": 0.7306016683578491, + "num_tokens": 64528372.0, + "step": 2137 + }, + { + "epoch": 0.17630081636018802, + "grad_norm": 2.0307137966156006, + "learning_rate": 5.87249244297884e-07, + "loss": 0.8467, + "mean_token_accuracy": 0.7439531087875366, + "num_tokens": 64559189.0, + "step": 2138 + }, + { + "epoch": 0.17638327698523953, + "grad_norm": 2.301593065261841, + "learning_rate": 5.875240450673261e-07, + "loss": 1.0002, + "mean_token_accuracy": 0.6997897624969482, + "num_tokens": 64583461.0, + "step": 2139 + }, + { + "epoch": 0.17646573761029108, + "grad_norm": 1.8192975521087646, + "learning_rate": 5.877988458367683e-07, + "loss": 0.7368, + "mean_token_accuracy": 0.7729976177215576, + "num_tokens": 64619583.0, + "step": 2140 + }, + { + "epoch": 0.17654819823534262, + "grad_norm": 2.1495211124420166, + "learning_rate": 5.880736466062105e-07, + "loss": 0.871, + "mean_token_accuracy": 0.7302600741386414, + "num_tokens": 64646519.0, + "step": 2141 + }, + { + "epoch": 0.17663065886039417, + "grad_norm": 1.9756709337234497, + "learning_rate": 5.883484473756526e-07, + "loss": 0.7391, + "mean_token_accuracy": 0.7674711346626282, + "num_tokens": 64677057.0, + "step": 2142 + }, + { + "epoch": 0.1767131194854457, + "grad_norm": 2.0251963138580322, + "learning_rate": 5.886232481450947e-07, + "loss": 0.7137, + "mean_token_accuracy": 0.7682346105575562, + "num_tokens": 64706068.0, + "step": 2143 + }, + { + "epoch": 0.17679558011049723, + "grad_norm": 2.3561315536499023, + "learning_rate": 5.88898048914537e-07, + "loss": 0.8978, + "mean_token_accuracy": 0.7208757400512695, + "num_tokens": 64731684.0, + "step": 2144 + }, + { + "epoch": 0.17687804073554877, + "grad_norm": 1.9247236251831055, + "learning_rate": 5.891728496839791e-07, + "loss": 0.8026, + "mean_token_accuracy": 0.747455358505249, + "num_tokens": 64763707.0, + "step": 2145 + }, + { + "epoch": 0.17696050136060032, + "grad_norm": 1.817876935005188, + "learning_rate": 5.894476504534213e-07, + "loss": 0.8683, + "mean_token_accuracy": 0.7308446168899536, + "num_tokens": 64801812.0, + "step": 2146 + }, + { + "epoch": 0.17704296198565186, + "grad_norm": 2.204054117202759, + "learning_rate": 5.897224512228633e-07, + "loss": 0.7423, + "mean_token_accuracy": 0.7580150961875916, + "num_tokens": 64825945.0, + "step": 2147 + }, + { + "epoch": 0.17712542261070338, + "grad_norm": 1.960205316543579, + "learning_rate": 5.899972519923055e-07, + "loss": 0.8421, + "mean_token_accuracy": 0.7392404079437256, + "num_tokens": 64856664.0, + "step": 2148 + }, + { + "epoch": 0.17720788323575493, + "grad_norm": 2.006572723388672, + "learning_rate": 5.902720527617478e-07, + "loss": 0.9217, + "mean_token_accuracy": 0.7248421311378479, + "num_tokens": 64887899.0, + "step": 2149 + }, + { + "epoch": 0.17729034386080647, + "grad_norm": 2.2219796180725098, + "learning_rate": 5.905468535311898e-07, + "loss": 0.9491, + "mean_token_accuracy": 0.7119932174682617, + "num_tokens": 64915409.0, + "step": 2150 + }, + { + "epoch": 0.17737280448585802, + "grad_norm": 1.7765388488769531, + "learning_rate": 5.90821654300632e-07, + "loss": 0.7275, + "mean_token_accuracy": 0.7778245210647583, + "num_tokens": 64949178.0, + "step": 2151 + }, + { + "epoch": 0.17745526511090953, + "grad_norm": 1.9422332048416138, + "learning_rate": 5.910964550700742e-07, + "loss": 0.818, + "mean_token_accuracy": 0.7412781715393066, + "num_tokens": 64980249.0, + "step": 2152 + }, + { + "epoch": 0.17753772573596108, + "grad_norm": 1.817924976348877, + "learning_rate": 5.913712558395163e-07, + "loss": 0.7876, + "mean_token_accuracy": 0.758813738822937, + "num_tokens": 65014248.0, + "step": 2153 + }, + { + "epoch": 0.17762018636101262, + "grad_norm": 2.14282488822937, + "learning_rate": 5.916460566089585e-07, + "loss": 0.8471, + "mean_token_accuracy": 0.7428890466690063, + "num_tokens": 65039939.0, + "step": 2154 + }, + { + "epoch": 0.17770264698606417, + "grad_norm": 2.153980016708374, + "learning_rate": 5.919208573784006e-07, + "loss": 0.8777, + "mean_token_accuracy": 0.7374365925788879, + "num_tokens": 65066766.0, + "step": 2155 + }, + { + "epoch": 0.17778510761111568, + "grad_norm": 1.869339108467102, + "learning_rate": 5.921956581478428e-07, + "loss": 0.9559, + "mean_token_accuracy": 0.7058311104774475, + "num_tokens": 65103417.0, + "step": 2156 + }, + { + "epoch": 0.17786756823616723, + "grad_norm": 2.0499460697174072, + "learning_rate": 5.924704589172849e-07, + "loss": 0.8517, + "mean_token_accuracy": 0.7402349710464478, + "num_tokens": 65133439.0, + "step": 2157 + }, + { + "epoch": 0.17795002886121877, + "grad_norm": 1.9459160566329956, + "learning_rate": 5.927452596867271e-07, + "loss": 0.7729, + "mean_token_accuracy": 0.7546190023422241, + "num_tokens": 65166514.0, + "step": 2158 + }, + { + "epoch": 0.17803248948627032, + "grad_norm": 1.9959990978240967, + "learning_rate": 5.930200604561693e-07, + "loss": 0.8824, + "mean_token_accuracy": 0.7266654372215271, + "num_tokens": 65199786.0, + "step": 2159 + }, + { + "epoch": 0.17811495011132183, + "grad_norm": 2.040956735610962, + "learning_rate": 5.932948612256115e-07, + "loss": 0.8436, + "mean_token_accuracy": 0.7389944791793823, + "num_tokens": 65226896.0, + "step": 2160 + }, + { + "epoch": 0.17819741073637338, + "grad_norm": 2.0637292861938477, + "learning_rate": 5.935696619950535e-07, + "loss": 0.8235, + "mean_token_accuracy": 0.7423043251037598, + "num_tokens": 65258446.0, + "step": 2161 + }, + { + "epoch": 0.17827987136142492, + "grad_norm": 2.0374414920806885, + "learning_rate": 5.938444627644957e-07, + "loss": 0.8665, + "mean_token_accuracy": 0.734740138053894, + "num_tokens": 65287509.0, + "step": 2162 + }, + { + "epoch": 0.17836233198647647, + "grad_norm": 1.9951162338256836, + "learning_rate": 5.941192635339379e-07, + "loss": 0.8572, + "mean_token_accuracy": 0.7244640588760376, + "num_tokens": 65318388.0, + "step": 2163 + }, + { + "epoch": 0.17844479261152799, + "grad_norm": 1.7300729751586914, + "learning_rate": 5.9439406430338e-07, + "loss": 0.9822, + "mean_token_accuracy": 0.7114285826683044, + "num_tokens": 65357746.0, + "step": 2164 + }, + { + "epoch": 0.17852725323657953, + "grad_norm": 2.1554083824157715, + "learning_rate": 5.946688650728222e-07, + "loss": 0.8242, + "mean_token_accuracy": 0.7403252124786377, + "num_tokens": 65385255.0, + "step": 2165 + }, + { + "epoch": 0.17860971386163108, + "grad_norm": 2.1415328979492188, + "learning_rate": 5.949436658422643e-07, + "loss": 0.8985, + "mean_token_accuracy": 0.7299051284790039, + "num_tokens": 65414223.0, + "step": 2166 + }, + { + "epoch": 0.17869217448668262, + "grad_norm": 2.208075523376465, + "learning_rate": 5.952184666117064e-07, + "loss": 0.9483, + "mean_token_accuracy": 0.7122540473937988, + "num_tokens": 65439811.0, + "step": 2167 + }, + { + "epoch": 0.17877463511173414, + "grad_norm": 2.322143316268921, + "learning_rate": 5.954932673811487e-07, + "loss": 0.8291, + "mean_token_accuracy": 0.7365913391113281, + "num_tokens": 65466179.0, + "step": 2168 + }, + { + "epoch": 0.17885709573678568, + "grad_norm": 2.329533100128174, + "learning_rate": 5.957680681505908e-07, + "loss": 0.9067, + "mean_token_accuracy": 0.720384418964386, + "num_tokens": 65492440.0, + "step": 2169 + }, + { + "epoch": 0.17893955636183723, + "grad_norm": 2.1498632431030273, + "learning_rate": 5.96042868920033e-07, + "loss": 0.8729, + "mean_token_accuracy": 0.7293941974639893, + "num_tokens": 65520798.0, + "step": 2170 + }, + { + "epoch": 0.17902201698688877, + "grad_norm": 2.0155999660491943, + "learning_rate": 5.96317669689475e-07, + "loss": 0.7879, + "mean_token_accuracy": 0.7586917877197266, + "num_tokens": 65552576.0, + "step": 2171 + }, + { + "epoch": 0.1791044776119403, + "grad_norm": 1.873840093612671, + "learning_rate": 5.965924704589173e-07, + "loss": 0.9055, + "mean_token_accuracy": 0.7239565849304199, + "num_tokens": 65588008.0, + "step": 2172 + }, + { + "epoch": 0.17918693823699183, + "grad_norm": 1.97608482837677, + "learning_rate": 5.968672712283595e-07, + "loss": 0.8749, + "mean_token_accuracy": 0.7233425378799438, + "num_tokens": 65620870.0, + "step": 2173 + }, + { + "epoch": 0.17926939886204338, + "grad_norm": 2.529712200164795, + "learning_rate": 5.971420719978015e-07, + "loss": 1.0734, + "mean_token_accuracy": 0.6728518009185791, + "num_tokens": 65643960.0, + "step": 2174 + }, + { + "epoch": 0.17935185948709492, + "grad_norm": 2.0638539791107178, + "learning_rate": 5.974168727672437e-07, + "loss": 0.8419, + "mean_token_accuracy": 0.7409147024154663, + "num_tokens": 65671373.0, + "step": 2175 + }, + { + "epoch": 0.17943432011214644, + "grad_norm": 2.3853936195373535, + "learning_rate": 5.976916735366859e-07, + "loss": 0.9252, + "mean_token_accuracy": 0.7258416414260864, + "num_tokens": 65694368.0, + "step": 2176 + }, + { + "epoch": 0.17951678073719798, + "grad_norm": 1.900363802909851, + "learning_rate": 5.97966474306128e-07, + "loss": 0.9283, + "mean_token_accuracy": 0.7139133214950562, + "num_tokens": 65727549.0, + "step": 2177 + }, + { + "epoch": 0.17959924136224953, + "grad_norm": 1.9541949033737183, + "learning_rate": 5.982412750755702e-07, + "loss": 0.8805, + "mean_token_accuracy": 0.7392981648445129, + "num_tokens": 65761549.0, + "step": 2178 + }, + { + "epoch": 0.17968170198730107, + "grad_norm": 2.281447172164917, + "learning_rate": 5.985160758450123e-07, + "loss": 0.8902, + "mean_token_accuracy": 0.7201948761940002, + "num_tokens": 65787251.0, + "step": 2179 + }, + { + "epoch": 0.1797641626123526, + "grad_norm": 1.8681275844573975, + "learning_rate": 5.987908766144545e-07, + "loss": 0.7955, + "mean_token_accuracy": 0.7467873692512512, + "num_tokens": 65820135.0, + "step": 2180 + }, + { + "epoch": 0.17984662323740414, + "grad_norm": 2.30523681640625, + "learning_rate": 5.990656773838966e-07, + "loss": 0.8305, + "mean_token_accuracy": 0.7380280494689941, + "num_tokens": 65845298.0, + "step": 2181 + }, + { + "epoch": 0.17992908386245568, + "grad_norm": 2.3046765327453613, + "learning_rate": 5.993404781533388e-07, + "loss": 0.8142, + "mean_token_accuracy": 0.7460610866546631, + "num_tokens": 65869052.0, + "step": 2182 + }, + { + "epoch": 0.18001154448750722, + "grad_norm": 2.018613576889038, + "learning_rate": 5.99615278922781e-07, + "loss": 1.0214, + "mean_token_accuracy": 0.6910146474838257, + "num_tokens": 65899392.0, + "step": 2183 + }, + { + "epoch": 0.18009400511255874, + "grad_norm": 1.9188401699066162, + "learning_rate": 5.998900796922231e-07, + "loss": 0.884, + "mean_token_accuracy": 0.7388159036636353, + "num_tokens": 65932037.0, + "step": 2184 + }, + { + "epoch": 0.1801764657376103, + "grad_norm": 2.3108177185058594, + "learning_rate": 6.001648804616652e-07, + "loss": 1.0509, + "mean_token_accuracy": 0.6891099214553833, + "num_tokens": 65957648.0, + "step": 2185 + }, + { + "epoch": 0.18025892636266183, + "grad_norm": 2.05928111076355, + "learning_rate": 6.004396812311075e-07, + "loss": 0.924, + "mean_token_accuracy": 0.7243841886520386, + "num_tokens": 65988631.0, + "step": 2186 + }, + { + "epoch": 0.18034138698771338, + "grad_norm": 1.9199491739273071, + "learning_rate": 6.007144820005496e-07, + "loss": 0.7452, + "mean_token_accuracy": 0.7648890018463135, + "num_tokens": 66021134.0, + "step": 2187 + }, + { + "epoch": 0.1804238476127649, + "grad_norm": 2.1749660968780518, + "learning_rate": 6.009892827699917e-07, + "loss": 0.7987, + "mean_token_accuracy": 0.749224066734314, + "num_tokens": 66049706.0, + "step": 2188 + }, + { + "epoch": 0.18050630823781644, + "grad_norm": 2.3147873878479004, + "learning_rate": 6.012640835394339e-07, + "loss": 0.8335, + "mean_token_accuracy": 0.7527575492858887, + "num_tokens": 66077091.0, + "step": 2189 + }, + { + "epoch": 0.18058876886286798, + "grad_norm": 2.094989538192749, + "learning_rate": 6.01538884308876e-07, + "loss": 0.8192, + "mean_token_accuracy": 0.747717559337616, + "num_tokens": 66109543.0, + "step": 2190 + }, + { + "epoch": 0.18067122948791953, + "grad_norm": 2.1164438724517822, + "learning_rate": 6.018136850783182e-07, + "loss": 0.8843, + "mean_token_accuracy": 0.7247068285942078, + "num_tokens": 66139620.0, + "step": 2191 + }, + { + "epoch": 0.18075369011297104, + "grad_norm": 2.281830072402954, + "learning_rate": 6.020884858477604e-07, + "loss": 0.9323, + "mean_token_accuracy": 0.7155448198318481, + "num_tokens": 66165402.0, + "step": 2192 + }, + { + "epoch": 0.1808361507380226, + "grad_norm": 2.1136858463287354, + "learning_rate": 6.023632866172025e-07, + "loss": 0.9376, + "mean_token_accuracy": 0.7182445526123047, + "num_tokens": 66193612.0, + "step": 2193 + }, + { + "epoch": 0.18091861136307413, + "grad_norm": 2.0049967765808105, + "learning_rate": 6.026380873866447e-07, + "loss": 0.8803, + "mean_token_accuracy": 0.7267665863037109, + "num_tokens": 66224352.0, + "step": 2194 + }, + { + "epoch": 0.18100107198812568, + "grad_norm": 1.9588477611541748, + "learning_rate": 6.029128881560867e-07, + "loss": 0.9637, + "mean_token_accuracy": 0.7095160484313965, + "num_tokens": 66256551.0, + "step": 2195 + }, + { + "epoch": 0.1810835326131772, + "grad_norm": 2.2719643115997314, + "learning_rate": 6.03187688925529e-07, + "loss": 0.8622, + "mean_token_accuracy": 0.7306114435195923, + "num_tokens": 66281988.0, + "step": 2196 + }, + { + "epoch": 0.18116599323822874, + "grad_norm": 2.1334314346313477, + "learning_rate": 6.034624896949712e-07, + "loss": 0.8662, + "mean_token_accuracy": 0.7337970733642578, + "num_tokens": 66309938.0, + "step": 2197 + }, + { + "epoch": 0.18124845386328028, + "grad_norm": 2.0379977226257324, + "learning_rate": 6.037372904644132e-07, + "loss": 0.8328, + "mean_token_accuracy": 0.7375301122665405, + "num_tokens": 66338187.0, + "step": 2198 + }, + { + "epoch": 0.18133091448833183, + "grad_norm": 1.9882636070251465, + "learning_rate": 6.040120912338554e-07, + "loss": 0.8331, + "mean_token_accuracy": 0.739071249961853, + "num_tokens": 66370043.0, + "step": 2199 + }, + { + "epoch": 0.18141337511338335, + "grad_norm": 2.4853882789611816, + "learning_rate": 6.042868920032977e-07, + "loss": 0.9306, + "mean_token_accuracy": 0.7078879475593567, + "num_tokens": 66394875.0, + "step": 2200 + }, + { + "epoch": 0.1814958357384349, + "grad_norm": 2.123617172241211, + "learning_rate": 6.045616927727397e-07, + "loss": 0.7476, + "mean_token_accuracy": 0.7566802501678467, + "num_tokens": 66421337.0, + "step": 2201 + }, + { + "epoch": 0.18157829636348644, + "grad_norm": 2.122694253921509, + "learning_rate": 6.048364935421819e-07, + "loss": 0.8481, + "mean_token_accuracy": 0.7352448105812073, + "num_tokens": 66449445.0, + "step": 2202 + }, + { + "epoch": 0.18166075698853798, + "grad_norm": 1.9774246215820312, + "learning_rate": 6.05111294311624e-07, + "loss": 0.832, + "mean_token_accuracy": 0.7416164875030518, + "num_tokens": 66483435.0, + "step": 2203 + }, + { + "epoch": 0.1817432176135895, + "grad_norm": 2.0742604732513428, + "learning_rate": 6.053860950810662e-07, + "loss": 0.832, + "mean_token_accuracy": 0.7441304326057434, + "num_tokens": 66511093.0, + "step": 2204 + }, + { + "epoch": 0.18182567823864104, + "grad_norm": 2.127173662185669, + "learning_rate": 6.056608958505084e-07, + "loss": 0.8497, + "mean_token_accuracy": 0.733798086643219, + "num_tokens": 66541755.0, + "step": 2205 + }, + { + "epoch": 0.1819081388636926, + "grad_norm": 1.9262343645095825, + "learning_rate": 6.059356966199505e-07, + "loss": 0.8322, + "mean_token_accuracy": 0.7424299716949463, + "num_tokens": 66572019.0, + "step": 2206 + }, + { + "epoch": 0.18199059948874413, + "grad_norm": 2.146034002304077, + "learning_rate": 6.062104973893927e-07, + "loss": 0.8123, + "mean_token_accuracy": 0.7549070715904236, + "num_tokens": 66602242.0, + "step": 2207 + }, + { + "epoch": 0.18207306011379568, + "grad_norm": 1.937963843345642, + "learning_rate": 6.064852981588348e-07, + "loss": 0.8747, + "mean_token_accuracy": 0.7331517338752747, + "num_tokens": 66637487.0, + "step": 2208 + }, + { + "epoch": 0.1821555207388472, + "grad_norm": 1.9095394611358643, + "learning_rate": 6.067600989282769e-07, + "loss": 0.8804, + "mean_token_accuracy": 0.7382134199142456, + "num_tokens": 66672898.0, + "step": 2209 + }, + { + "epoch": 0.18223798136389874, + "grad_norm": 1.8053627014160156, + "learning_rate": 6.070348996977192e-07, + "loss": 0.8348, + "mean_token_accuracy": 0.7475044131278992, + "num_tokens": 66708901.0, + "step": 2210 + }, + { + "epoch": 0.18232044198895028, + "grad_norm": 1.900917887687683, + "learning_rate": 6.073097004671612e-07, + "loss": 0.8195, + "mean_token_accuracy": 0.7521653175354004, + "num_tokens": 66741159.0, + "step": 2211 + }, + { + "epoch": 0.18240290261400183, + "grad_norm": 1.9445830583572388, + "learning_rate": 6.075845012366034e-07, + "loss": 0.7302, + "mean_token_accuracy": 0.7689406871795654, + "num_tokens": 66772375.0, + "step": 2212 + }, + { + "epoch": 0.18248536323905334, + "grad_norm": 2.1321444511413574, + "learning_rate": 6.078593020060456e-07, + "loss": 0.8408, + "mean_token_accuracy": 0.7372334599494934, + "num_tokens": 66799783.0, + "step": 2213 + }, + { + "epoch": 0.1825678238641049, + "grad_norm": 1.8838995695114136, + "learning_rate": 6.081341027754878e-07, + "loss": 0.804, + "mean_token_accuracy": 0.7535148859024048, + "num_tokens": 66834927.0, + "step": 2214 + }, + { + "epoch": 0.18265028448915643, + "grad_norm": 2.477288007736206, + "learning_rate": 6.084089035449299e-07, + "loss": 0.9639, + "mean_token_accuracy": 0.7049069404602051, + "num_tokens": 66859708.0, + "step": 2215 + }, + { + "epoch": 0.18273274511420798, + "grad_norm": 1.934116244316101, + "learning_rate": 6.086837043143721e-07, + "loss": 0.7985, + "mean_token_accuracy": 0.7512194514274597, + "num_tokens": 66892975.0, + "step": 2216 + }, + { + "epoch": 0.1828152057392595, + "grad_norm": 1.743238925933838, + "learning_rate": 6.089585050838142e-07, + "loss": 0.6062, + "mean_token_accuracy": 0.8035449385643005, + "num_tokens": 66930383.0, + "step": 2217 + }, + { + "epoch": 0.18289766636431104, + "grad_norm": 2.110553741455078, + "learning_rate": 6.092333058532563e-07, + "loss": 0.8788, + "mean_token_accuracy": 0.7292197346687317, + "num_tokens": 66958215.0, + "step": 2218 + }, + { + "epoch": 0.18298012698936258, + "grad_norm": 1.8731517791748047, + "learning_rate": 6.095081066226985e-07, + "loss": 0.8435, + "mean_token_accuracy": 0.75047767162323, + "num_tokens": 66994494.0, + "step": 2219 + }, + { + "epoch": 0.18306258761441413, + "grad_norm": 2.0684075355529785, + "learning_rate": 6.097829073921407e-07, + "loss": 0.7906, + "mean_token_accuracy": 0.7516710758209229, + "num_tokens": 67021174.0, + "step": 2220 + }, + { + "epoch": 0.18314504823946565, + "grad_norm": 1.911033034324646, + "learning_rate": 6.100577081615829e-07, + "loss": 0.7719, + "mean_token_accuracy": 0.7513386607170105, + "num_tokens": 67053622.0, + "step": 2221 + }, + { + "epoch": 0.1832275088645172, + "grad_norm": 2.240455389022827, + "learning_rate": 6.103325089310249e-07, + "loss": 0.7798, + "mean_token_accuracy": 0.7531282901763916, + "num_tokens": 67079134.0, + "step": 2222 + }, + { + "epoch": 0.18330996948956874, + "grad_norm": 1.9255367517471313, + "learning_rate": 6.106073097004671e-07, + "loss": 1.0335, + "mean_token_accuracy": 0.693274974822998, + "num_tokens": 67116063.0, + "step": 2223 + }, + { + "epoch": 0.18339243011462028, + "grad_norm": 1.9766885042190552, + "learning_rate": 6.108821104699094e-07, + "loss": 0.8541, + "mean_token_accuracy": 0.736835777759552, + "num_tokens": 67150566.0, + "step": 2224 + }, + { + "epoch": 0.1834748907396718, + "grad_norm": 2.250858783721924, + "learning_rate": 6.111569112393514e-07, + "loss": 0.9653, + "mean_token_accuracy": 0.7096788883209229, + "num_tokens": 67176255.0, + "step": 2225 + }, + { + "epoch": 0.18355735136472334, + "grad_norm": 1.9219492673873901, + "learning_rate": 6.114317120087936e-07, + "loss": 0.76, + "mean_token_accuracy": 0.7708268761634827, + "num_tokens": 67207850.0, + "step": 2226 + }, + { + "epoch": 0.1836398119897749, + "grad_norm": 1.9978772401809692, + "learning_rate": 6.117065127782357e-07, + "loss": 0.727, + "mean_token_accuracy": 0.7722371816635132, + "num_tokens": 67234682.0, + "step": 2227 + }, + { + "epoch": 0.18372227261482643, + "grad_norm": 2.106340169906616, + "learning_rate": 6.11981313547678e-07, + "loss": 0.8663, + "mean_token_accuracy": 0.7344421148300171, + "num_tokens": 67262876.0, + "step": 2228 + }, + { + "epoch": 0.18380473323987795, + "grad_norm": 1.9993350505828857, + "learning_rate": 6.122561143171201e-07, + "loss": 0.8286, + "mean_token_accuracy": 0.7370607852935791, + "num_tokens": 67294716.0, + "step": 2229 + }, + { + "epoch": 0.1838871938649295, + "grad_norm": 2.100485324859619, + "learning_rate": 6.125309150865622e-07, + "loss": 0.9329, + "mean_token_accuracy": 0.7228190898895264, + "num_tokens": 67325306.0, + "step": 2230 + }, + { + "epoch": 0.18396965448998104, + "grad_norm": 2.277048110961914, + "learning_rate": 6.128057158560044e-07, + "loss": 0.9051, + "mean_token_accuracy": 0.7204943895339966, + "num_tokens": 67348326.0, + "step": 2231 + }, + { + "epoch": 0.18405211511503258, + "grad_norm": 1.8374193906784058, + "learning_rate": 6.130805166254465e-07, + "loss": 0.7269, + "mean_token_accuracy": 0.7688210010528564, + "num_tokens": 67382727.0, + "step": 2232 + }, + { + "epoch": 0.1841345757400841, + "grad_norm": 2.3001489639282227, + "learning_rate": 6.133553173948887e-07, + "loss": 0.9864, + "mean_token_accuracy": 0.7033143043518066, + "num_tokens": 67410777.0, + "step": 2233 + }, + { + "epoch": 0.18421703636513564, + "grad_norm": 2.072336435317993, + "learning_rate": 6.136301181643309e-07, + "loss": 0.7924, + "mean_token_accuracy": 0.74737548828125, + "num_tokens": 67439110.0, + "step": 2234 + }, + { + "epoch": 0.1842994969901872, + "grad_norm": 2.1110920906066895, + "learning_rate": 6.139049189337729e-07, + "loss": 0.9809, + "mean_token_accuracy": 0.7112285494804382, + "num_tokens": 67466995.0, + "step": 2235 + }, + { + "epoch": 0.18438195761523873, + "grad_norm": 1.8791134357452393, + "learning_rate": 6.141797197032151e-07, + "loss": 0.7576, + "mean_token_accuracy": 0.7605605125427246, + "num_tokens": 67500205.0, + "step": 2236 + }, + { + "epoch": 0.18446441824029025, + "grad_norm": 1.9675737619400024, + "learning_rate": 6.144545204726573e-07, + "loss": 0.7193, + "mean_token_accuracy": 0.7706539034843445, + "num_tokens": 67530343.0, + "step": 2237 + }, + { + "epoch": 0.1845468788653418, + "grad_norm": 2.15010929107666, + "learning_rate": 6.147293212420995e-07, + "loss": 0.7466, + "mean_token_accuracy": 0.7629244923591614, + "num_tokens": 67559757.0, + "step": 2238 + }, + { + "epoch": 0.18462933949039334, + "grad_norm": 2.215325355529785, + "learning_rate": 6.150041220115416e-07, + "loss": 0.8003, + "mean_token_accuracy": 0.7464509010314941, + "num_tokens": 67582947.0, + "step": 2239 + }, + { + "epoch": 0.18471180011544489, + "grad_norm": 2.1401631832122803, + "learning_rate": 6.152789227809838e-07, + "loss": 0.8442, + "mean_token_accuracy": 0.7325496673583984, + "num_tokens": 67609596.0, + "step": 2240 + }, + { + "epoch": 0.1847942607404964, + "grad_norm": 2.0014495849609375, + "learning_rate": 6.155537235504259e-07, + "loss": 0.6961, + "mean_token_accuracy": 0.7771811485290527, + "num_tokens": 67640012.0, + "step": 2241 + }, + { + "epoch": 0.18487672136554795, + "grad_norm": 2.161133289337158, + "learning_rate": 6.158285243198681e-07, + "loss": 0.8951, + "mean_token_accuracy": 0.7260287404060364, + "num_tokens": 67668257.0, + "step": 2242 + }, + { + "epoch": 0.1849591819905995, + "grad_norm": 2.144883632659912, + "learning_rate": 6.161033250893102e-07, + "loss": 0.951, + "mean_token_accuracy": 0.7082509994506836, + "num_tokens": 67694497.0, + "step": 2243 + }, + { + "epoch": 0.18504164261565104, + "grad_norm": 1.7791873216629028, + "learning_rate": 6.163781258587524e-07, + "loss": 0.7627, + "mean_token_accuracy": 0.7670036554336548, + "num_tokens": 67732678.0, + "step": 2244 + }, + { + "epoch": 0.18512410324070255, + "grad_norm": 1.8266280889511108, + "learning_rate": 6.166529266281946e-07, + "loss": 0.8349, + "mean_token_accuracy": 0.7480456829071045, + "num_tokens": 67768008.0, + "step": 2245 + }, + { + "epoch": 0.1852065638657541, + "grad_norm": 2.1453585624694824, + "learning_rate": 6.169277273976366e-07, + "loss": 0.7645, + "mean_token_accuracy": 0.7637383937835693, + "num_tokens": 67794678.0, + "step": 2246 + }, + { + "epoch": 0.18528902449080564, + "grad_norm": 1.9427618980407715, + "learning_rate": 6.172025281670789e-07, + "loss": 0.9645, + "mean_token_accuracy": 0.7092040777206421, + "num_tokens": 67828161.0, + "step": 2247 + }, + { + "epoch": 0.1853714851158572, + "grad_norm": 2.219489336013794, + "learning_rate": 6.174773289365211e-07, + "loss": 0.9679, + "mean_token_accuracy": 0.7036007642745972, + "num_tokens": 67859946.0, + "step": 2248 + }, + { + "epoch": 0.1854539457409087, + "grad_norm": 2.0825865268707275, + "learning_rate": 6.177521297059631e-07, + "loss": 0.8127, + "mean_token_accuracy": 0.7518148422241211, + "num_tokens": 67885591.0, + "step": 2249 + }, + { + "epoch": 0.18553640636596025, + "grad_norm": 2.4353866577148438, + "learning_rate": 6.180269304754053e-07, + "loss": 0.9594, + "mean_token_accuracy": 0.7057174444198608, + "num_tokens": 67909379.0, + "step": 2250 + }, + { + "epoch": 0.1856188669910118, + "grad_norm": 2.137195587158203, + "learning_rate": 6.183017312448474e-07, + "loss": 0.9501, + "mean_token_accuracy": 0.7120466232299805, + "num_tokens": 67940113.0, + "step": 2251 + }, + { + "epoch": 0.18570132761606334, + "grad_norm": 1.9965344667434692, + "learning_rate": 6.185765320142896e-07, + "loss": 0.8435, + "mean_token_accuracy": 0.7484246492385864, + "num_tokens": 67973308.0, + "step": 2252 + }, + { + "epoch": 0.18578378824111486, + "grad_norm": 2.0130324363708496, + "learning_rate": 6.188513327837318e-07, + "loss": 0.8218, + "mean_token_accuracy": 0.7434560060501099, + "num_tokens": 68004937.0, + "step": 2253 + }, + { + "epoch": 0.1858662488661664, + "grad_norm": 1.9353164434432983, + "learning_rate": 6.191261335531739e-07, + "loss": 0.8826, + "mean_token_accuracy": 0.7255194187164307, + "num_tokens": 68039410.0, + "step": 2254 + }, + { + "epoch": 0.18594870949121795, + "grad_norm": 2.570120334625244, + "learning_rate": 6.194009343226161e-07, + "loss": 0.866, + "mean_token_accuracy": 0.7274097204208374, + "num_tokens": 68063554.0, + "step": 2255 + }, + { + "epoch": 0.1860311701162695, + "grad_norm": 2.300010919570923, + "learning_rate": 6.196757350920583e-07, + "loss": 0.8228, + "mean_token_accuracy": 0.7438790798187256, + "num_tokens": 68091199.0, + "step": 2256 + }, + { + "epoch": 0.186113630741321, + "grad_norm": 2.080531358718872, + "learning_rate": 6.199505358615004e-07, + "loss": 0.8743, + "mean_token_accuracy": 0.736781120300293, + "num_tokens": 68120062.0, + "step": 2257 + }, + { + "epoch": 0.18619609136637255, + "grad_norm": 1.9678919315338135, + "learning_rate": 6.202253366309426e-07, + "loss": 0.9415, + "mean_token_accuracy": 0.7095876932144165, + "num_tokens": 68154765.0, + "step": 2258 + }, + { + "epoch": 0.1862785519914241, + "grad_norm": 2.0102739334106445, + "learning_rate": 6.205001374003846e-07, + "loss": 0.9296, + "mean_token_accuracy": 0.7167536020278931, + "num_tokens": 68185899.0, + "step": 2259 + }, + { + "epoch": 0.18636101261647564, + "grad_norm": 2.269232749938965, + "learning_rate": 6.207749381698268e-07, + "loss": 0.8897, + "mean_token_accuracy": 0.724913477897644, + "num_tokens": 68210828.0, + "step": 2260 + }, + { + "epoch": 0.18644347324152716, + "grad_norm": 1.9658719301223755, + "learning_rate": 6.210497389392691e-07, + "loss": 0.926, + "mean_token_accuracy": 0.7174134850502014, + "num_tokens": 68240221.0, + "step": 2261 + }, + { + "epoch": 0.1865259338665787, + "grad_norm": 2.2763352394104004, + "learning_rate": 6.213245397087112e-07, + "loss": 0.7693, + "mean_token_accuracy": 0.7538816928863525, + "num_tokens": 68265633.0, + "step": 2262 + }, + { + "epoch": 0.18660839449163025, + "grad_norm": 1.9568133354187012, + "learning_rate": 6.215993404781533e-07, + "loss": 0.7432, + "mean_token_accuracy": 0.7650355100631714, + "num_tokens": 68297609.0, + "step": 2263 + }, + { + "epoch": 0.1866908551166818, + "grad_norm": 2.1565654277801514, + "learning_rate": 6.218741412475955e-07, + "loss": 0.9428, + "mean_token_accuracy": 0.7106567621231079, + "num_tokens": 68325710.0, + "step": 2264 + }, + { + "epoch": 0.1867733157417333, + "grad_norm": 1.9640710353851318, + "learning_rate": 6.221489420170376e-07, + "loss": 0.8088, + "mean_token_accuracy": 0.7582015991210938, + "num_tokens": 68357195.0, + "step": 2265 + }, + { + "epoch": 0.18685577636678485, + "grad_norm": 1.979195237159729, + "learning_rate": 6.224237427864798e-07, + "loss": 0.7511, + "mean_token_accuracy": 0.7764274477958679, + "num_tokens": 68389947.0, + "step": 2266 + }, + { + "epoch": 0.1869382369918364, + "grad_norm": 1.9211764335632324, + "learning_rate": 6.226985435559219e-07, + "loss": 0.938, + "mean_token_accuracy": 0.7181965112686157, + "num_tokens": 68422194.0, + "step": 2267 + }, + { + "epoch": 0.18702069761688794, + "grad_norm": 2.2571306228637695, + "learning_rate": 6.229733443253641e-07, + "loss": 1.0474, + "mean_token_accuracy": 0.6809492111206055, + "num_tokens": 68449349.0, + "step": 2268 + }, + { + "epoch": 0.18710315824193946, + "grad_norm": 1.7230461835861206, + "learning_rate": 6.232481450948062e-07, + "loss": 0.6094, + "mean_token_accuracy": 0.8050206899642944, + "num_tokens": 68487076.0, + "step": 2269 + }, + { + "epoch": 0.187185618866991, + "grad_norm": 2.154165029525757, + "learning_rate": 6.235229458642483e-07, + "loss": 0.8037, + "mean_token_accuracy": 0.7474712133407593, + "num_tokens": 68512432.0, + "step": 2270 + }, + { + "epoch": 0.18726807949204255, + "grad_norm": 2.363881826400757, + "learning_rate": 6.237977466336906e-07, + "loss": 0.9362, + "mean_token_accuracy": 0.7117995023727417, + "num_tokens": 68540548.0, + "step": 2271 + }, + { + "epoch": 0.1873505401170941, + "grad_norm": 2.2325150966644287, + "learning_rate": 6.240725474031328e-07, + "loss": 0.9592, + "mean_token_accuracy": 0.7155036330223083, + "num_tokens": 68565271.0, + "step": 2272 + }, + { + "epoch": 0.18743300074214564, + "grad_norm": 2.0259149074554443, + "learning_rate": 6.243473481725748e-07, + "loss": 0.8458, + "mean_token_accuracy": 0.7446849346160889, + "num_tokens": 68594518.0, + "step": 2273 + }, + { + "epoch": 0.18751546136719716, + "grad_norm": 1.928207278251648, + "learning_rate": 6.24622148942017e-07, + "loss": 0.7413, + "mean_token_accuracy": 0.7624290585517883, + "num_tokens": 68625460.0, + "step": 2274 + }, + { + "epoch": 0.1875979219922487, + "grad_norm": 2.206652879714966, + "learning_rate": 6.248969497114592e-07, + "loss": 0.8474, + "mean_token_accuracy": 0.7450675964355469, + "num_tokens": 68652320.0, + "step": 2275 + }, + { + "epoch": 0.18768038261730025, + "grad_norm": 2.1509506702423096, + "learning_rate": 6.251717504809013e-07, + "loss": 0.9226, + "mean_token_accuracy": 0.7121586799621582, + "num_tokens": 68679177.0, + "step": 2276 + }, + { + "epoch": 0.1877628432423518, + "grad_norm": 2.2386200428009033, + "learning_rate": 6.254465512503435e-07, + "loss": 0.878, + "mean_token_accuracy": 0.7289509773254395, + "num_tokens": 68708173.0, + "step": 2277 + }, + { + "epoch": 0.1878453038674033, + "grad_norm": 2.0531396865844727, + "learning_rate": 6.257213520197856e-07, + "loss": 0.9947, + "mean_token_accuracy": 0.7004314064979553, + "num_tokens": 68740031.0, + "step": 2278 + }, + { + "epoch": 0.18792776449245485, + "grad_norm": 1.9658708572387695, + "learning_rate": 6.259961527892278e-07, + "loss": 0.8663, + "mean_token_accuracy": 0.7367220520973206, + "num_tokens": 68771347.0, + "step": 2279 + }, + { + "epoch": 0.1880102251175064, + "grad_norm": 1.9863479137420654, + "learning_rate": 6.2627095355867e-07, + "loss": 0.9375, + "mean_token_accuracy": 0.7363364696502686, + "num_tokens": 68804790.0, + "step": 2280 + }, + { + "epoch": 0.18809268574255794, + "grad_norm": 1.9808320999145508, + "learning_rate": 6.265457543281121e-07, + "loss": 0.8978, + "mean_token_accuracy": 0.7275524735450745, + "num_tokens": 68837328.0, + "step": 2281 + }, + { + "epoch": 0.18817514636760946, + "grad_norm": 2.065345287322998, + "learning_rate": 6.268205550975543e-07, + "loss": 0.8495, + "mean_token_accuracy": 0.7350069284439087, + "num_tokens": 68867788.0, + "step": 2282 + }, + { + "epoch": 0.188257606992661, + "grad_norm": 2.2628021240234375, + "learning_rate": 6.270953558669963e-07, + "loss": 0.7772, + "mean_token_accuracy": 0.7562698125839233, + "num_tokens": 68891815.0, + "step": 2283 + }, + { + "epoch": 0.18834006761771255, + "grad_norm": 2.1607296466827393, + "learning_rate": 6.273701566364385e-07, + "loss": 0.9738, + "mean_token_accuracy": 0.7168681621551514, + "num_tokens": 68920399.0, + "step": 2284 + }, + { + "epoch": 0.1884225282427641, + "grad_norm": 2.1135013103485107, + "learning_rate": 6.276449574058808e-07, + "loss": 0.8909, + "mean_token_accuracy": 0.7366048693656921, + "num_tokens": 68948731.0, + "step": 2285 + }, + { + "epoch": 0.1885049888678156, + "grad_norm": 2.0554380416870117, + "learning_rate": 6.279197581753228e-07, + "loss": 0.8225, + "mean_token_accuracy": 0.7564795613288879, + "num_tokens": 68978274.0, + "step": 2286 + }, + { + "epoch": 0.18858744949286715, + "grad_norm": 1.8302855491638184, + "learning_rate": 6.28194558944765e-07, + "loss": 0.7646, + "mean_token_accuracy": 0.7700861096382141, + "num_tokens": 69014829.0, + "step": 2287 + }, + { + "epoch": 0.1886699101179187, + "grad_norm": 1.9339016675949097, + "learning_rate": 6.284693597142072e-07, + "loss": 0.6529, + "mean_token_accuracy": 0.7841840982437134, + "num_tokens": 69046616.0, + "step": 2288 + }, + { + "epoch": 0.18875237074297024, + "grad_norm": 1.9993616342544556, + "learning_rate": 6.287441604836494e-07, + "loss": 0.9066, + "mean_token_accuracy": 0.7139160633087158, + "num_tokens": 69079543.0, + "step": 2289 + }, + { + "epoch": 0.18883483136802176, + "grad_norm": 2.1365692615509033, + "learning_rate": 6.290189612530915e-07, + "loss": 0.9524, + "mean_token_accuracy": 0.7005301713943481, + "num_tokens": 69108754.0, + "step": 2290 + }, + { + "epoch": 0.1889172919930733, + "grad_norm": 2.0422017574310303, + "learning_rate": 6.292937620225336e-07, + "loss": 0.8319, + "mean_token_accuracy": 0.739001989364624, + "num_tokens": 69136951.0, + "step": 2291 + }, + { + "epoch": 0.18899975261812485, + "grad_norm": 2.1564109325408936, + "learning_rate": 6.295685627919758e-07, + "loss": 0.859, + "mean_token_accuracy": 0.7407863140106201, + "num_tokens": 69163451.0, + "step": 2292 + }, + { + "epoch": 0.1890822132431764, + "grad_norm": 2.0408260822296143, + "learning_rate": 6.298433635614179e-07, + "loss": 0.8793, + "mean_token_accuracy": 0.7280500531196594, + "num_tokens": 69193134.0, + "step": 2293 + }, + { + "epoch": 0.1891646738682279, + "grad_norm": 2.025625228881836, + "learning_rate": 6.301181643308601e-07, + "loss": 0.8443, + "mean_token_accuracy": 0.7400133013725281, + "num_tokens": 69222428.0, + "step": 2294 + }, + { + "epoch": 0.18924713449327946, + "grad_norm": 2.020122766494751, + "learning_rate": 6.303929651003023e-07, + "loss": 0.9046, + "mean_token_accuracy": 0.7236413955688477, + "num_tokens": 69254415.0, + "step": 2295 + }, + { + "epoch": 0.189329595118331, + "grad_norm": 2.5331947803497314, + "learning_rate": 6.306677658697445e-07, + "loss": 0.8355, + "mean_token_accuracy": 0.7379772663116455, + "num_tokens": 69275345.0, + "step": 2296 + }, + { + "epoch": 0.18941205574338255, + "grad_norm": 1.7360036373138428, + "learning_rate": 6.309425666391865e-07, + "loss": 0.8632, + "mean_token_accuracy": 0.7274636030197144, + "num_tokens": 69316534.0, + "step": 2297 + }, + { + "epoch": 0.18949451636843406, + "grad_norm": 2.3232972621917725, + "learning_rate": 6.312173674086287e-07, + "loss": 0.7409, + "mean_token_accuracy": 0.7629525065422058, + "num_tokens": 69339029.0, + "step": 2298 + }, + { + "epoch": 0.1895769769934856, + "grad_norm": 1.9691253900527954, + "learning_rate": 6.314921681780709e-07, + "loss": 0.7538, + "mean_token_accuracy": 0.7564710378646851, + "num_tokens": 69370138.0, + "step": 2299 + }, + { + "epoch": 0.18965943761853715, + "grad_norm": 1.9355685710906982, + "learning_rate": 6.31766968947513e-07, + "loss": 0.9633, + "mean_token_accuracy": 0.7080135345458984, + "num_tokens": 69402802.0, + "step": 2300 + }, + { + "epoch": 0.1897418982435887, + "grad_norm": 2.2245023250579834, + "learning_rate": 6.320417697169552e-07, + "loss": 0.9986, + "mean_token_accuracy": 0.6993838548660278, + "num_tokens": 69430080.0, + "step": 2301 + }, + { + "epoch": 0.18982435886864021, + "grad_norm": 1.852309226989746, + "learning_rate": 6.323165704863973e-07, + "loss": 0.8164, + "mean_token_accuracy": 0.7456220388412476, + "num_tokens": 69461973.0, + "step": 2302 + }, + { + "epoch": 0.18990681949369176, + "grad_norm": 1.8542194366455078, + "learning_rate": 6.325913712558396e-07, + "loss": 0.7832, + "mean_token_accuracy": 0.7558555603027344, + "num_tokens": 69496956.0, + "step": 2303 + }, + { + "epoch": 0.1899892801187433, + "grad_norm": 1.925903081893921, + "learning_rate": 6.328661720252817e-07, + "loss": 0.8253, + "mean_token_accuracy": 0.741621196269989, + "num_tokens": 69532514.0, + "step": 2304 + }, + { + "epoch": 0.19007174074379485, + "grad_norm": 1.9371318817138672, + "learning_rate": 6.331409727947238e-07, + "loss": 0.9278, + "mean_token_accuracy": 0.7233750224113464, + "num_tokens": 69568026.0, + "step": 2305 + }, + { + "epoch": 0.19015420136884637, + "grad_norm": 2.1780834197998047, + "learning_rate": 6.33415773564166e-07, + "loss": 0.8801, + "mean_token_accuracy": 0.728009819984436, + "num_tokens": 69597551.0, + "step": 2306 + }, + { + "epoch": 0.1902366619938979, + "grad_norm": 2.015808343887329, + "learning_rate": 6.33690574333608e-07, + "loss": 0.7901, + "mean_token_accuracy": 0.7550503611564636, + "num_tokens": 69630362.0, + "step": 2307 + }, + { + "epoch": 0.19031912261894945, + "grad_norm": 2.0654478073120117, + "learning_rate": 6.339653751030503e-07, + "loss": 0.903, + "mean_token_accuracy": 0.7307332754135132, + "num_tokens": 69662522.0, + "step": 2308 + }, + { + "epoch": 0.190401583244001, + "grad_norm": 1.9701584577560425, + "learning_rate": 6.342401758724925e-07, + "loss": 0.8271, + "mean_token_accuracy": 0.7386654615402222, + "num_tokens": 69692473.0, + "step": 2309 + }, + { + "epoch": 0.19048404386905252, + "grad_norm": 2.054100275039673, + "learning_rate": 6.345149766419345e-07, + "loss": 0.9312, + "mean_token_accuracy": 0.7110805511474609, + "num_tokens": 69722233.0, + "step": 2310 + }, + { + "epoch": 0.19056650449410406, + "grad_norm": 1.9884591102600098, + "learning_rate": 6.347897774113767e-07, + "loss": 0.9618, + "mean_token_accuracy": 0.7036853432655334, + "num_tokens": 69757812.0, + "step": 2311 + }, + { + "epoch": 0.1906489651191556, + "grad_norm": 2.174501419067383, + "learning_rate": 6.350645781808189e-07, + "loss": 0.8468, + "mean_token_accuracy": 0.729017436504364, + "num_tokens": 69785322.0, + "step": 2312 + }, + { + "epoch": 0.19073142574420715, + "grad_norm": 1.9825127124786377, + "learning_rate": 6.353393789502611e-07, + "loss": 0.8783, + "mean_token_accuracy": 0.7370282411575317, + "num_tokens": 69817267.0, + "step": 2313 + }, + { + "epoch": 0.19081388636925867, + "grad_norm": 2.2389776706695557, + "learning_rate": 6.356141797197032e-07, + "loss": 0.7068, + "mean_token_accuracy": 0.7749263644218445, + "num_tokens": 69840317.0, + "step": 2314 + }, + { + "epoch": 0.1908963469943102, + "grad_norm": 1.7658370733261108, + "learning_rate": 6.358889804891453e-07, + "loss": 0.6911, + "mean_token_accuracy": 0.7894213795661926, + "num_tokens": 69875180.0, + "step": 2315 + }, + { + "epoch": 0.19097880761936176, + "grad_norm": 2.11575984954834, + "learning_rate": 6.361637812585875e-07, + "loss": 0.7757, + "mean_token_accuracy": 0.7581590414047241, + "num_tokens": 69905304.0, + "step": 2316 + }, + { + "epoch": 0.1910612682444133, + "grad_norm": 1.8323907852172852, + "learning_rate": 6.364385820280297e-07, + "loss": 0.8519, + "mean_token_accuracy": 0.7388482689857483, + "num_tokens": 69942463.0, + "step": 2317 + }, + { + "epoch": 0.19114372886946482, + "grad_norm": 2.220064163208008, + "learning_rate": 6.367133827974718e-07, + "loss": 0.7051, + "mean_token_accuracy": 0.7625762224197388, + "num_tokens": 69965626.0, + "step": 2318 + }, + { + "epoch": 0.19122618949451636, + "grad_norm": 2.0359067916870117, + "learning_rate": 6.36988183566914e-07, + "loss": 0.8507, + "mean_token_accuracy": 0.7327566742897034, + "num_tokens": 69996108.0, + "step": 2319 + }, + { + "epoch": 0.1913086501195679, + "grad_norm": 2.1089746952056885, + "learning_rate": 6.37262984336356e-07, + "loss": 0.8885, + "mean_token_accuracy": 0.7344229221343994, + "num_tokens": 70026276.0, + "step": 2320 + }, + { + "epoch": 0.19139111074461945, + "grad_norm": 2.1432852745056152, + "learning_rate": 6.375377851057982e-07, + "loss": 0.8018, + "mean_token_accuracy": 0.7495038509368896, + "num_tokens": 70053721.0, + "step": 2321 + }, + { + "epoch": 0.19147357136967097, + "grad_norm": 2.5730316638946533, + "learning_rate": 6.378125858752405e-07, + "loss": 0.8811, + "mean_token_accuracy": 0.7294206619262695, + "num_tokens": 70075254.0, + "step": 2322 + }, + { + "epoch": 0.19155603199472251, + "grad_norm": 1.8647902011871338, + "learning_rate": 6.380873866446826e-07, + "loss": 0.8101, + "mean_token_accuracy": 0.7510364055633545, + "num_tokens": 70110780.0, + "step": 2323 + }, + { + "epoch": 0.19163849261977406, + "grad_norm": 2.0429089069366455, + "learning_rate": 6.383621874141247e-07, + "loss": 0.9594, + "mean_token_accuracy": 0.7147541642189026, + "num_tokens": 70143644.0, + "step": 2324 + }, + { + "epoch": 0.1917209532448256, + "grad_norm": 1.852028727531433, + "learning_rate": 6.386369881835669e-07, + "loss": 0.8872, + "mean_token_accuracy": 0.7317536473274231, + "num_tokens": 70177390.0, + "step": 2325 + }, + { + "epoch": 0.19180341386987712, + "grad_norm": 2.273106813430786, + "learning_rate": 6.38911788953009e-07, + "loss": 0.8906, + "mean_token_accuracy": 0.7324609160423279, + "num_tokens": 70203135.0, + "step": 2326 + }, + { + "epoch": 0.19188587449492867, + "grad_norm": 2.2173376083374023, + "learning_rate": 6.391865897224512e-07, + "loss": 0.7732, + "mean_token_accuracy": 0.7472094297409058, + "num_tokens": 70229025.0, + "step": 2327 + }, + { + "epoch": 0.1919683351199802, + "grad_norm": 1.9684308767318726, + "learning_rate": 6.394613904918933e-07, + "loss": 0.7965, + "mean_token_accuracy": 0.7559791803359985, + "num_tokens": 70259274.0, + "step": 2328 + }, + { + "epoch": 0.19205079574503175, + "grad_norm": 2.3105998039245605, + "learning_rate": 6.397361912613355e-07, + "loss": 0.9821, + "mean_token_accuracy": 0.7106884717941284, + "num_tokens": 70286935.0, + "step": 2329 + }, + { + "epoch": 0.19213325637008327, + "grad_norm": 2.136603832244873, + "learning_rate": 6.400109920307777e-07, + "loss": 0.9526, + "mean_token_accuracy": 0.7189773321151733, + "num_tokens": 70316531.0, + "step": 2330 + }, + { + "epoch": 0.19221571699513482, + "grad_norm": 2.2951297760009766, + "learning_rate": 6.402857928002197e-07, + "loss": 0.9844, + "mean_token_accuracy": 0.7072654366493225, + "num_tokens": 70343381.0, + "step": 2331 + }, + { + "epoch": 0.19229817762018636, + "grad_norm": 2.041747808456421, + "learning_rate": 6.40560593569662e-07, + "loss": 0.898, + "mean_token_accuracy": 0.7317155599594116, + "num_tokens": 70376370.0, + "step": 2332 + }, + { + "epoch": 0.1923806382452379, + "grad_norm": 2.0153257846832275, + "learning_rate": 6.408353943391042e-07, + "loss": 0.8354, + "mean_token_accuracy": 0.7387596964836121, + "num_tokens": 70408816.0, + "step": 2333 + }, + { + "epoch": 0.19246309887028942, + "grad_norm": 1.8907736539840698, + "learning_rate": 6.411101951085462e-07, + "loss": 0.7873, + "mean_token_accuracy": 0.7628771066665649, + "num_tokens": 70443624.0, + "step": 2334 + }, + { + "epoch": 0.19254555949534097, + "grad_norm": 2.0817692279815674, + "learning_rate": 6.413849958779884e-07, + "loss": 0.9915, + "mean_token_accuracy": 0.6987605690956116, + "num_tokens": 70475069.0, + "step": 2335 + }, + { + "epoch": 0.1926280201203925, + "grad_norm": 1.864863634109497, + "learning_rate": 6.416597966474306e-07, + "loss": 0.8549, + "mean_token_accuracy": 0.733317494392395, + "num_tokens": 70509887.0, + "step": 2336 + }, + { + "epoch": 0.19271048074544406, + "grad_norm": 1.7668209075927734, + "learning_rate": 6.419345974168727e-07, + "loss": 0.7488, + "mean_token_accuracy": 0.7600290775299072, + "num_tokens": 70543971.0, + "step": 2337 + }, + { + "epoch": 0.1927929413704956, + "grad_norm": 1.9903050661087036, + "learning_rate": 6.422093981863149e-07, + "loss": 0.8926, + "mean_token_accuracy": 0.7284696102142334, + "num_tokens": 70575409.0, + "step": 2338 + }, + { + "epoch": 0.19287540199554712, + "grad_norm": 1.962903618812561, + "learning_rate": 6.42484198955757e-07, + "loss": 0.8139, + "mean_token_accuracy": 0.7507119178771973, + "num_tokens": 70607697.0, + "step": 2339 + }, + { + "epoch": 0.19295786262059866, + "grad_norm": 2.106872797012329, + "learning_rate": 6.427589997251992e-07, + "loss": 0.9593, + "mean_token_accuracy": 0.7108416557312012, + "num_tokens": 70636086.0, + "step": 2340 + }, + { + "epoch": 0.1930403232456502, + "grad_norm": 2.347687005996704, + "learning_rate": 6.430338004946414e-07, + "loss": 0.9351, + "mean_token_accuracy": 0.708877682685852, + "num_tokens": 70659718.0, + "step": 2341 + }, + { + "epoch": 0.19312278387070175, + "grad_norm": 1.8253731727600098, + "learning_rate": 6.433086012640835e-07, + "loss": 0.8018, + "mean_token_accuracy": 0.7598893642425537, + "num_tokens": 70693141.0, + "step": 2342 + }, + { + "epoch": 0.19320524449575327, + "grad_norm": 2.176539659500122, + "learning_rate": 6.435834020335257e-07, + "loss": 0.8135, + "mean_token_accuracy": 0.7475861310958862, + "num_tokens": 70719754.0, + "step": 2343 + }, + { + "epoch": 0.19328770512080481, + "grad_norm": 2.626775026321411, + "learning_rate": 6.438582028029677e-07, + "loss": 0.8761, + "mean_token_accuracy": 0.7197669744491577, + "num_tokens": 70739402.0, + "step": 2344 + }, + { + "epoch": 0.19337016574585636, + "grad_norm": 2.2266080379486084, + "learning_rate": 6.441330035724099e-07, + "loss": 0.8693, + "mean_token_accuracy": 0.7411713600158691, + "num_tokens": 70765953.0, + "step": 2345 + }, + { + "epoch": 0.1934526263709079, + "grad_norm": 2.0076117515563965, + "learning_rate": 6.444078043418522e-07, + "loss": 0.8963, + "mean_token_accuracy": 0.7322173118591309, + "num_tokens": 70796656.0, + "step": 2346 + }, + { + "epoch": 0.19353508699595942, + "grad_norm": 2.1979501247406006, + "learning_rate": 6.446826051112943e-07, + "loss": 0.8154, + "mean_token_accuracy": 0.7536759376525879, + "num_tokens": 70821313.0, + "step": 2347 + }, + { + "epoch": 0.19361754762101097, + "grad_norm": 1.8757808208465576, + "learning_rate": 6.449574058807364e-07, + "loss": 0.9516, + "mean_token_accuracy": 0.7125358581542969, + "num_tokens": 70855585.0, + "step": 2348 + }, + { + "epoch": 0.1937000082460625, + "grad_norm": 2.117851972579956, + "learning_rate": 6.452322066501786e-07, + "loss": 0.9402, + "mean_token_accuracy": 0.7132500410079956, + "num_tokens": 70886059.0, + "step": 2349 + }, + { + "epoch": 0.19378246887111406, + "grad_norm": 2.1039459705352783, + "learning_rate": 6.455070074196208e-07, + "loss": 0.7516, + "mean_token_accuracy": 0.7613468170166016, + "num_tokens": 70913802.0, + "step": 2350 + }, + { + "epoch": 0.19386492949616557, + "grad_norm": 2.375861167907715, + "learning_rate": 6.457818081890629e-07, + "loss": 0.93, + "mean_token_accuracy": 0.7202264070510864, + "num_tokens": 70940141.0, + "step": 2351 + }, + { + "epoch": 0.19394739012121712, + "grad_norm": 1.9024757146835327, + "learning_rate": 6.46056608958505e-07, + "loss": 0.6854, + "mean_token_accuracy": 0.7939953207969666, + "num_tokens": 70968402.0, + "step": 2352 + }, + { + "epoch": 0.19402985074626866, + "grad_norm": 2.4619548320770264, + "learning_rate": 6.463314097279472e-07, + "loss": 0.9002, + "mean_token_accuracy": 0.7270994782447815, + "num_tokens": 70990617.0, + "step": 2353 + }, + { + "epoch": 0.1941123113713202, + "grad_norm": 1.8758723735809326, + "learning_rate": 6.466062104973893e-07, + "loss": 0.7493, + "mean_token_accuracy": 0.7705410122871399, + "num_tokens": 71024730.0, + "step": 2354 + }, + { + "epoch": 0.19419477199637172, + "grad_norm": 2.112971544265747, + "learning_rate": 6.468810112668315e-07, + "loss": 0.8595, + "mean_token_accuracy": 0.7351844310760498, + "num_tokens": 71053449.0, + "step": 2355 + }, + { + "epoch": 0.19427723262142327, + "grad_norm": 2.2323150634765625, + "learning_rate": 6.471558120362737e-07, + "loss": 1.0522, + "mean_token_accuracy": 0.6816068887710571, + "num_tokens": 71077832.0, + "step": 2356 + }, + { + "epoch": 0.1943596932464748, + "grad_norm": 2.134854316711426, + "learning_rate": 6.474306128057159e-07, + "loss": 0.8055, + "mean_token_accuracy": 0.7386572360992432, + "num_tokens": 71106364.0, + "step": 2357 + }, + { + "epoch": 0.19444215387152636, + "grad_norm": 2.198471784591675, + "learning_rate": 6.477054135751579e-07, + "loss": 0.7742, + "mean_token_accuracy": 0.7532641291618347, + "num_tokens": 71132927.0, + "step": 2358 + }, + { + "epoch": 0.19452461449657787, + "grad_norm": 2.125227451324463, + "learning_rate": 6.479802143446001e-07, + "loss": 0.879, + "mean_token_accuracy": 0.7212669849395752, + "num_tokens": 71165863.0, + "step": 2359 + }, + { + "epoch": 0.19460707512162942, + "grad_norm": 2.1238434314727783, + "learning_rate": 6.482550151140423e-07, + "loss": 0.9132, + "mean_token_accuracy": 0.722495436668396, + "num_tokens": 71193165.0, + "step": 2360 + }, + { + "epoch": 0.19468953574668096, + "grad_norm": 1.9577075242996216, + "learning_rate": 6.485298158834844e-07, + "loss": 0.8369, + "mean_token_accuracy": 0.7413164973258972, + "num_tokens": 71227118.0, + "step": 2361 + }, + { + "epoch": 0.1947719963717325, + "grad_norm": 1.9570724964141846, + "learning_rate": 6.488046166529266e-07, + "loss": 0.7303, + "mean_token_accuracy": 0.771539032459259, + "num_tokens": 71258721.0, + "step": 2362 + }, + { + "epoch": 0.19485445699678403, + "grad_norm": 2.2046427726745605, + "learning_rate": 6.490794174223687e-07, + "loss": 0.9015, + "mean_token_accuracy": 0.721750795841217, + "num_tokens": 71283754.0, + "step": 2363 + }, + { + "epoch": 0.19493691762183557, + "grad_norm": 1.9870134592056274, + "learning_rate": 6.49354218191811e-07, + "loss": 0.9637, + "mean_token_accuracy": 0.7153565287590027, + "num_tokens": 71318572.0, + "step": 2364 + }, + { + "epoch": 0.19501937824688712, + "grad_norm": 1.8176887035369873, + "learning_rate": 6.496290189612531e-07, + "loss": 0.8371, + "mean_token_accuracy": 0.7400875687599182, + "num_tokens": 71356408.0, + "step": 2365 + }, + { + "epoch": 0.19510183887193866, + "grad_norm": 2.10078501701355, + "learning_rate": 6.499038197306952e-07, + "loss": 0.7738, + "mean_token_accuracy": 0.7668285369873047, + "num_tokens": 71385782.0, + "step": 2366 + }, + { + "epoch": 0.19518429949699018, + "grad_norm": 2.126376152038574, + "learning_rate": 6.501786205001374e-07, + "loss": 0.9676, + "mean_token_accuracy": 0.7085942029953003, + "num_tokens": 71416936.0, + "step": 2367 + }, + { + "epoch": 0.19526676012204172, + "grad_norm": 1.8326689004898071, + "learning_rate": 6.504534212695794e-07, + "loss": 0.899, + "mean_token_accuracy": 0.7341902256011963, + "num_tokens": 71454674.0, + "step": 2368 + }, + { + "epoch": 0.19534922074709327, + "grad_norm": 1.8917174339294434, + "learning_rate": 6.507282220390217e-07, + "loss": 0.7943, + "mean_token_accuracy": 0.755429744720459, + "num_tokens": 71488390.0, + "step": 2369 + }, + { + "epoch": 0.1954316813721448, + "grad_norm": 2.323601007461548, + "learning_rate": 6.510030228084639e-07, + "loss": 0.9975, + "mean_token_accuracy": 0.7062281370162964, + "num_tokens": 71513620.0, + "step": 2370 + }, + { + "epoch": 0.19551414199719633, + "grad_norm": 1.7965991497039795, + "learning_rate": 6.512778235779059e-07, + "loss": 0.9344, + "mean_token_accuracy": 0.7181904315948486, + "num_tokens": 71552263.0, + "step": 2371 + }, + { + "epoch": 0.19559660262224787, + "grad_norm": 1.8030720949172974, + "learning_rate": 6.515526243473481e-07, + "loss": 0.7493, + "mean_token_accuracy": 0.7657244205474854, + "num_tokens": 71587065.0, + "step": 2372 + }, + { + "epoch": 0.19567906324729942, + "grad_norm": 2.390268325805664, + "learning_rate": 6.518274251167903e-07, + "loss": 0.9637, + "mean_token_accuracy": 0.703181266784668, + "num_tokens": 71614708.0, + "step": 2373 + }, + { + "epoch": 0.19576152387235096, + "grad_norm": 1.9625954627990723, + "learning_rate": 6.521022258862325e-07, + "loss": 0.8632, + "mean_token_accuracy": 0.7367439866065979, + "num_tokens": 71647849.0, + "step": 2374 + }, + { + "epoch": 0.19584398449740248, + "grad_norm": 2.210479736328125, + "learning_rate": 6.523770266556746e-07, + "loss": 0.9255, + "mean_token_accuracy": 0.7125680446624756, + "num_tokens": 71677979.0, + "step": 2375 + }, + { + "epoch": 0.19592644512245402, + "grad_norm": 2.1628129482269287, + "learning_rate": 6.526518274251167e-07, + "loss": 0.9437, + "mean_token_accuracy": 0.7180332541465759, + "num_tokens": 71707823.0, + "step": 2376 + }, + { + "epoch": 0.19600890574750557, + "grad_norm": 2.000633478164673, + "learning_rate": 6.529266281945589e-07, + "loss": 0.7184, + "mean_token_accuracy": 0.7631494998931885, + "num_tokens": 71735859.0, + "step": 2377 + }, + { + "epoch": 0.1960913663725571, + "grad_norm": 2.0183470249176025, + "learning_rate": 6.532014289640011e-07, + "loss": 0.8919, + "mean_token_accuracy": 0.723739504814148, + "num_tokens": 71766672.0, + "step": 2378 + }, + { + "epoch": 0.19617382699760863, + "grad_norm": 2.249382734298706, + "learning_rate": 6.534762297334432e-07, + "loss": 0.8288, + "mean_token_accuracy": 0.7397507429122925, + "num_tokens": 71791424.0, + "step": 2379 + }, + { + "epoch": 0.19625628762266017, + "grad_norm": 2.3616816997528076, + "learning_rate": 6.537510305028854e-07, + "loss": 0.9241, + "mean_token_accuracy": 0.721355676651001, + "num_tokens": 71817123.0, + "step": 2380 + }, + { + "epoch": 0.19633874824771172, + "grad_norm": 2.038825750350952, + "learning_rate": 6.540258312723276e-07, + "loss": 0.9761, + "mean_token_accuracy": 0.7055118083953857, + "num_tokens": 71850617.0, + "step": 2381 + }, + { + "epoch": 0.19642120887276326, + "grad_norm": 2.3532052040100098, + "learning_rate": 6.543006320417696e-07, + "loss": 0.821, + "mean_token_accuracy": 0.748945415019989, + "num_tokens": 71875436.0, + "step": 2382 + }, + { + "epoch": 0.19650366949781478, + "grad_norm": 1.9239853620529175, + "learning_rate": 6.545754328112119e-07, + "loss": 0.8505, + "mean_token_accuracy": 0.7444579601287842, + "num_tokens": 71908050.0, + "step": 2383 + }, + { + "epoch": 0.19658613012286633, + "grad_norm": 1.9882210493087769, + "learning_rate": 6.54850233580654e-07, + "loss": 0.9508, + "mean_token_accuracy": 0.712222695350647, + "num_tokens": 71938568.0, + "step": 2384 + }, + { + "epoch": 0.19666859074791787, + "grad_norm": 2.07625675201416, + "learning_rate": 6.551250343500961e-07, + "loss": 0.9009, + "mean_token_accuracy": 0.7235968112945557, + "num_tokens": 71971460.0, + "step": 2385 + }, + { + "epoch": 0.19675105137296942, + "grad_norm": 2.183706521987915, + "learning_rate": 6.553998351195383e-07, + "loss": 1.0356, + "mean_token_accuracy": 0.6960832476615906, + "num_tokens": 71997561.0, + "step": 2386 + }, + { + "epoch": 0.19683351199802093, + "grad_norm": 1.8976958990097046, + "learning_rate": 6.556746358889804e-07, + "loss": 0.8563, + "mean_token_accuracy": 0.7324737906455994, + "num_tokens": 72030537.0, + "step": 2387 + }, + { + "epoch": 0.19691597262307248, + "grad_norm": 2.2488112449645996, + "learning_rate": 6.559494366584227e-07, + "loss": 0.8752, + "mean_token_accuracy": 0.7276034355163574, + "num_tokens": 72059322.0, + "step": 2388 + }, + { + "epoch": 0.19699843324812402, + "grad_norm": 2.1861069202423096, + "learning_rate": 6.562242374278648e-07, + "loss": 0.8634, + "mean_token_accuracy": 0.7334606647491455, + "num_tokens": 72085820.0, + "step": 2389 + }, + { + "epoch": 0.19708089387317557, + "grad_norm": 2.1475582122802734, + "learning_rate": 6.564990381973069e-07, + "loss": 0.9958, + "mean_token_accuracy": 0.6914352178573608, + "num_tokens": 72115630.0, + "step": 2390 + }, + { + "epoch": 0.19716335449822708, + "grad_norm": 2.0292766094207764, + "learning_rate": 6.567738389667491e-07, + "loss": 0.7966, + "mean_token_accuracy": 0.7463798522949219, + "num_tokens": 72144191.0, + "step": 2391 + }, + { + "epoch": 0.19724581512327863, + "grad_norm": 2.2461612224578857, + "learning_rate": 6.570486397361912e-07, + "loss": 0.9916, + "mean_token_accuracy": 0.7121171355247498, + "num_tokens": 72171113.0, + "step": 2392 + }, + { + "epoch": 0.19732827574833017, + "grad_norm": 2.1682310104370117, + "learning_rate": 6.573234405056334e-07, + "loss": 0.8417, + "mean_token_accuracy": 0.7367330193519592, + "num_tokens": 72197172.0, + "step": 2393 + }, + { + "epoch": 0.19741073637338172, + "grad_norm": 2.192509651184082, + "learning_rate": 6.575982412750756e-07, + "loss": 0.7862, + "mean_token_accuracy": 0.7606461048126221, + "num_tokens": 72222971.0, + "step": 2394 + }, + { + "epoch": 0.19749319699843323, + "grad_norm": 2.085054874420166, + "learning_rate": 6.578730420445176e-07, + "loss": 0.8442, + "mean_token_accuracy": 0.7265884876251221, + "num_tokens": 72250997.0, + "step": 2395 + }, + { + "epoch": 0.19757565762348478, + "grad_norm": 2.056385040283203, + "learning_rate": 6.581478428139598e-07, + "loss": 1.0139, + "mean_token_accuracy": 0.6973865628242493, + "num_tokens": 72281769.0, + "step": 2396 + }, + { + "epoch": 0.19765811824853632, + "grad_norm": 1.9799553155899048, + "learning_rate": 6.584226435834021e-07, + "loss": 0.8727, + "mean_token_accuracy": 0.7379887104034424, + "num_tokens": 72316473.0, + "step": 2397 + }, + { + "epoch": 0.19774057887358787, + "grad_norm": 1.8332931995391846, + "learning_rate": 6.586974443528442e-07, + "loss": 0.8008, + "mean_token_accuracy": 0.7532175779342651, + "num_tokens": 72356361.0, + "step": 2398 + }, + { + "epoch": 0.19782303949863939, + "grad_norm": 2.416405200958252, + "learning_rate": 6.589722451222863e-07, + "loss": 0.9728, + "mean_token_accuracy": 0.7060790657997131, + "num_tokens": 72380447.0, + "step": 2399 + }, + { + "epoch": 0.19790550012369093, + "grad_norm": 2.3181252479553223, + "learning_rate": 6.592470458917284e-07, + "loss": 0.9923, + "mean_token_accuracy": 0.7027307748794556, + "num_tokens": 72405086.0, + "step": 2400 + }, + { + "epoch": 0.19798796074874248, + "grad_norm": 2.068854808807373, + "learning_rate": 6.595218466611706e-07, + "loss": 1.0068, + "mean_token_accuracy": 0.7012544870376587, + "num_tokens": 72437612.0, + "step": 2401 + }, + { + "epoch": 0.19807042137379402, + "grad_norm": 1.810164451599121, + "learning_rate": 6.597966474306128e-07, + "loss": 0.761, + "mean_token_accuracy": 0.7641777992248535, + "num_tokens": 72469608.0, + "step": 2402 + }, + { + "epoch": 0.19815288199884556, + "grad_norm": 1.7893266677856445, + "learning_rate": 6.600714482000549e-07, + "loss": 0.7914, + "mean_token_accuracy": 0.7409276962280273, + "num_tokens": 72506281.0, + "step": 2403 + }, + { + "epoch": 0.19823534262389708, + "grad_norm": 2.2085700035095215, + "learning_rate": 6.603462489694971e-07, + "loss": 0.8137, + "mean_token_accuracy": 0.7380874156951904, + "num_tokens": 72534892.0, + "step": 2404 + }, + { + "epoch": 0.19831780324894863, + "grad_norm": 1.9805140495300293, + "learning_rate": 6.606210497389393e-07, + "loss": 0.9067, + "mean_token_accuracy": 0.7211213111877441, + "num_tokens": 72567003.0, + "step": 2405 + }, + { + "epoch": 0.19840026387400017, + "grad_norm": 2.0368363857269287, + "learning_rate": 6.608958505083814e-07, + "loss": 1.0655, + "mean_token_accuracy": 0.6769062280654907, + "num_tokens": 72599548.0, + "step": 2406 + }, + { + "epoch": 0.19848272449905172, + "grad_norm": 2.3755667209625244, + "learning_rate": 6.611706512778236e-07, + "loss": 0.9732, + "mean_token_accuracy": 0.7133071422576904, + "num_tokens": 72623538.0, + "step": 2407 + }, + { + "epoch": 0.19856518512410323, + "grad_norm": 1.86786949634552, + "learning_rate": 6.614454520472657e-07, + "loss": 0.8146, + "mean_token_accuracy": 0.7495149374008179, + "num_tokens": 72658364.0, + "step": 2408 + }, + { + "epoch": 0.19864764574915478, + "grad_norm": 2.351311206817627, + "learning_rate": 6.617202528167078e-07, + "loss": 0.8925, + "mean_token_accuracy": 0.7246725559234619, + "num_tokens": 72681303.0, + "step": 2409 + }, + { + "epoch": 0.19873010637420632, + "grad_norm": 2.0057120323181152, + "learning_rate": 6.6199505358615e-07, + "loss": 0.9229, + "mean_token_accuracy": 0.7218419313430786, + "num_tokens": 72713521.0, + "step": 2410 + }, + { + "epoch": 0.19881256699925787, + "grad_norm": 1.964141845703125, + "learning_rate": 6.622698543555922e-07, + "loss": 0.8641, + "mean_token_accuracy": 0.7329114675521851, + "num_tokens": 72745260.0, + "step": 2411 + }, + { + "epoch": 0.19889502762430938, + "grad_norm": 1.97533118724823, + "learning_rate": 6.625446551250343e-07, + "loss": 1.0023, + "mean_token_accuracy": 0.6982477903366089, + "num_tokens": 72778149.0, + "step": 2412 + }, + { + "epoch": 0.19897748824936093, + "grad_norm": 1.9920731782913208, + "learning_rate": 6.628194558944765e-07, + "loss": 0.8244, + "mean_token_accuracy": 0.7392204999923706, + "num_tokens": 72806671.0, + "step": 2413 + }, + { + "epoch": 0.19905994887441247, + "grad_norm": 1.8816252946853638, + "learning_rate": 6.630942566639186e-07, + "loss": 0.9118, + "mean_token_accuracy": 0.7214726209640503, + "num_tokens": 72843961.0, + "step": 2414 + }, + { + "epoch": 0.19914240949946402, + "grad_norm": 2.0269885063171387, + "learning_rate": 6.633690574333608e-07, + "loss": 0.8244, + "mean_token_accuracy": 0.7447600364685059, + "num_tokens": 72872994.0, + "step": 2415 + }, + { + "epoch": 0.19922487012451554, + "grad_norm": 2.0611352920532227, + "learning_rate": 6.636438582028029e-07, + "loss": 0.9252, + "mean_token_accuracy": 0.7170296907424927, + "num_tokens": 72904543.0, + "step": 2416 + }, + { + "epoch": 0.19930733074956708, + "grad_norm": 2.093501567840576, + "learning_rate": 6.639186589722451e-07, + "loss": 0.9116, + "mean_token_accuracy": 0.7189697623252869, + "num_tokens": 72934702.0, + "step": 2417 + }, + { + "epoch": 0.19938979137461862, + "grad_norm": 1.7319364547729492, + "learning_rate": 6.641934597416873e-07, + "loss": 0.7347, + "mean_token_accuracy": 0.7563570737838745, + "num_tokens": 72972638.0, + "step": 2418 + }, + { + "epoch": 0.19947225199967017, + "grad_norm": 2.1664276123046875, + "learning_rate": 6.644682605111293e-07, + "loss": 0.8985, + "mean_token_accuracy": 0.7193621397018433, + "num_tokens": 72999597.0, + "step": 2419 + }, + { + "epoch": 0.1995547126247217, + "grad_norm": 2.2002010345458984, + "learning_rate": 6.647430612805716e-07, + "loss": 0.8606, + "mean_token_accuracy": 0.742111325263977, + "num_tokens": 73025511.0, + "step": 2420 + }, + { + "epoch": 0.19963717324977323, + "grad_norm": 2.3881430625915527, + "learning_rate": 6.650178620500138e-07, + "loss": 0.768, + "mean_token_accuracy": 0.7587279081344604, + "num_tokens": 73049908.0, + "step": 2421 + }, + { + "epoch": 0.19971963387482478, + "grad_norm": 1.9800606966018677, + "learning_rate": 6.652926628194558e-07, + "loss": 0.8769, + "mean_token_accuracy": 0.725591778755188, + "num_tokens": 73081594.0, + "step": 2422 + }, + { + "epoch": 0.19980209449987632, + "grad_norm": 2.630749464035034, + "learning_rate": 6.65567463588898e-07, + "loss": 0.9378, + "mean_token_accuracy": 0.7102943658828735, + "num_tokens": 73103017.0, + "step": 2423 + }, + { + "epoch": 0.19988455512492784, + "grad_norm": 2.205204725265503, + "learning_rate": 6.658422643583401e-07, + "loss": 0.7283, + "mean_token_accuracy": 0.7729459404945374, + "num_tokens": 73128355.0, + "step": 2424 + }, + { + "epoch": 0.19996701574997938, + "grad_norm": 1.7583088874816895, + "learning_rate": 6.661170651277824e-07, + "loss": 0.7796, + "mean_token_accuracy": 0.7636103630065918, + "num_tokens": 73165632.0, + "step": 2425 + }, + { + "epoch": 0.20004947637503093, + "grad_norm": 2.006618022918701, + "learning_rate": 6.663918658972245e-07, + "loss": 0.9287, + "mean_token_accuracy": 0.7203341722488403, + "num_tokens": 73198982.0, + "step": 2426 + }, + { + "epoch": 0.20013193700008247, + "grad_norm": 1.7986513376235962, + "learning_rate": 6.666666666666666e-07, + "loss": 0.7677, + "mean_token_accuracy": 0.756123423576355, + "num_tokens": 73236945.0, + "step": 2427 + }, + { + "epoch": 0.200214397625134, + "grad_norm": 1.9987562894821167, + "learning_rate": 6.669414674361088e-07, + "loss": 0.8438, + "mean_token_accuracy": 0.7427761554718018, + "num_tokens": 73267193.0, + "step": 2428 + }, + { + "epoch": 0.20029685825018553, + "grad_norm": 2.1512410640716553, + "learning_rate": 6.672162682055509e-07, + "loss": 0.9014, + "mean_token_accuracy": 0.7269668579101562, + "num_tokens": 73298042.0, + "step": 2429 + }, + { + "epoch": 0.20037931887523708, + "grad_norm": 2.021164655685425, + "learning_rate": 6.674910689749931e-07, + "loss": 0.7314, + "mean_token_accuracy": 0.7699324488639832, + "num_tokens": 73327801.0, + "step": 2430 + }, + { + "epoch": 0.20046177950028862, + "grad_norm": 2.0221474170684814, + "learning_rate": 6.677658697444353e-07, + "loss": 0.936, + "mean_token_accuracy": 0.7110230922698975, + "num_tokens": 73360116.0, + "step": 2431 + }, + { + "epoch": 0.20054424012534014, + "grad_norm": 1.9844141006469727, + "learning_rate": 6.680406705138774e-07, + "loss": 0.6902, + "mean_token_accuracy": 0.778213620185852, + "num_tokens": 73387712.0, + "step": 2432 + }, + { + "epoch": 0.20062670075039168, + "grad_norm": 1.990349531173706, + "learning_rate": 6.683154712833195e-07, + "loss": 0.7498, + "mean_token_accuracy": 0.7529619336128235, + "num_tokens": 73415191.0, + "step": 2433 + }, + { + "epoch": 0.20070916137544323, + "grad_norm": 2.109370708465576, + "learning_rate": 6.685902720527618e-07, + "loss": 0.9032, + "mean_token_accuracy": 0.7280092239379883, + "num_tokens": 73442452.0, + "step": 2434 + }, + { + "epoch": 0.20079162200049477, + "grad_norm": 1.8694645166397095, + "learning_rate": 6.688650728222039e-07, + "loss": 0.7145, + "mean_token_accuracy": 0.7799092531204224, + "num_tokens": 73474218.0, + "step": 2435 + }, + { + "epoch": 0.2008740826255463, + "grad_norm": 2.1529879570007324, + "learning_rate": 6.69139873591646e-07, + "loss": 0.9531, + "mean_token_accuracy": 0.7212119698524475, + "num_tokens": 73502952.0, + "step": 2436 + }, + { + "epoch": 0.20095654325059784, + "grad_norm": 2.1662380695343018, + "learning_rate": 6.694146743610882e-07, + "loss": 0.7943, + "mean_token_accuracy": 0.7564496994018555, + "num_tokens": 73530248.0, + "step": 2437 + }, + { + "epoch": 0.20103900387564938, + "grad_norm": 2.4079599380493164, + "learning_rate": 6.696894751305303e-07, + "loss": 0.9905, + "mean_token_accuracy": 0.7005162239074707, + "num_tokens": 73552696.0, + "step": 2438 + }, + { + "epoch": 0.20112146450070092, + "grad_norm": 1.8853167295455933, + "learning_rate": 6.699642758999726e-07, + "loss": 0.7353, + "mean_token_accuracy": 0.7632014751434326, + "num_tokens": 73585451.0, + "step": 2439 + }, + { + "epoch": 0.20120392512575244, + "grad_norm": 1.8458620309829712, + "learning_rate": 6.702390766694146e-07, + "loss": 0.855, + "mean_token_accuracy": 0.7386734485626221, + "num_tokens": 73621131.0, + "step": 2440 + }, + { + "epoch": 0.201286385750804, + "grad_norm": 1.876204490661621, + "learning_rate": 6.705138774388568e-07, + "loss": 0.7001, + "mean_token_accuracy": 0.768340528011322, + "num_tokens": 73654075.0, + "step": 2441 + }, + { + "epoch": 0.20136884637585553, + "grad_norm": 2.143777370452881, + "learning_rate": 6.70788678208299e-07, + "loss": 0.8387, + "mean_token_accuracy": 0.7418729662895203, + "num_tokens": 73681963.0, + "step": 2442 + }, + { + "epoch": 0.20145130700090708, + "grad_norm": 1.8147509098052979, + "learning_rate": 6.71063478977741e-07, + "loss": 0.985, + "mean_token_accuracy": 0.7043489217758179, + "num_tokens": 73720437.0, + "step": 2443 + }, + { + "epoch": 0.2015337676259586, + "grad_norm": 1.9105892181396484, + "learning_rate": 6.713382797471833e-07, + "loss": 0.7835, + "mean_token_accuracy": 0.7520047426223755, + "num_tokens": 73752086.0, + "step": 2444 + }, + { + "epoch": 0.20161622825101014, + "grad_norm": 2.0225830078125, + "learning_rate": 6.716130805166255e-07, + "loss": 0.9312, + "mean_token_accuracy": 0.7180671691894531, + "num_tokens": 73782276.0, + "step": 2445 + }, + { + "epoch": 0.20169868887606168, + "grad_norm": 2.0409739017486572, + "learning_rate": 6.718878812860675e-07, + "loss": 0.8806, + "mean_token_accuracy": 0.7356061935424805, + "num_tokens": 73811819.0, + "step": 2446 + }, + { + "epoch": 0.20178114950111323, + "grad_norm": 2.4201009273529053, + "learning_rate": 6.721626820555097e-07, + "loss": 0.8446, + "mean_token_accuracy": 0.7364508509635925, + "num_tokens": 73833180.0, + "step": 2447 + }, + { + "epoch": 0.20186361012616474, + "grad_norm": 1.9668419361114502, + "learning_rate": 6.724374828249518e-07, + "loss": 0.8639, + "mean_token_accuracy": 0.7193729877471924, + "num_tokens": 73864700.0, + "step": 2448 + }, + { + "epoch": 0.2019460707512163, + "grad_norm": 2.1473324298858643, + "learning_rate": 6.727122835943941e-07, + "loss": 0.7652, + "mean_token_accuracy": 0.7500659227371216, + "num_tokens": 73889747.0, + "step": 2449 + }, + { + "epoch": 0.20202853137626783, + "grad_norm": 1.9761073589324951, + "learning_rate": 6.729870843638362e-07, + "loss": 0.7402, + "mean_token_accuracy": 0.7648200392723083, + "num_tokens": 73922780.0, + "step": 2450 + }, + { + "epoch": 0.20211099200131938, + "grad_norm": 2.1031696796417236, + "learning_rate": 6.732618851332783e-07, + "loss": 0.8667, + "mean_token_accuracy": 0.736113965511322, + "num_tokens": 73954204.0, + "step": 2451 + }, + { + "epoch": 0.2021934526263709, + "grad_norm": 2.3908183574676514, + "learning_rate": 6.735366859027205e-07, + "loss": 0.844, + "mean_token_accuracy": 0.7505725622177124, + "num_tokens": 73978907.0, + "step": 2452 + }, + { + "epoch": 0.20227591325142244, + "grad_norm": 1.8147755861282349, + "learning_rate": 6.738114866721627e-07, + "loss": 0.7123, + "mean_token_accuracy": 0.7774757742881775, + "num_tokens": 74011205.0, + "step": 2453 + }, + { + "epoch": 0.20235837387647398, + "grad_norm": 2.0763728618621826, + "learning_rate": 6.740862874416048e-07, + "loss": 0.9983, + "mean_token_accuracy": 0.7205536365509033, + "num_tokens": 74045165.0, + "step": 2454 + }, + { + "epoch": 0.20244083450152553, + "grad_norm": 2.1339268684387207, + "learning_rate": 6.74361088211047e-07, + "loss": 0.9299, + "mean_token_accuracy": 0.7130947709083557, + "num_tokens": 74073059.0, + "step": 2455 + }, + { + "epoch": 0.20252329512657705, + "grad_norm": 2.122657060623169, + "learning_rate": 6.74635888980489e-07, + "loss": 0.8573, + "mean_token_accuracy": 0.7322395443916321, + "num_tokens": 74101144.0, + "step": 2456 + }, + { + "epoch": 0.2026057557516286, + "grad_norm": 2.1596930027008057, + "learning_rate": 6.749106897499312e-07, + "loss": 0.7494, + "mean_token_accuracy": 0.7580983638763428, + "num_tokens": 74128359.0, + "step": 2457 + }, + { + "epoch": 0.20268821637668014, + "grad_norm": 2.0190186500549316, + "learning_rate": 6.751854905193735e-07, + "loss": 0.7386, + "mean_token_accuracy": 0.7585405707359314, + "num_tokens": 74158869.0, + "step": 2458 + }, + { + "epoch": 0.20277067700173168, + "grad_norm": 2.0464346408843994, + "learning_rate": 6.754602912888156e-07, + "loss": 0.8863, + "mean_token_accuracy": 0.7243049144744873, + "num_tokens": 74189081.0, + "step": 2459 + }, + { + "epoch": 0.2028531376267832, + "grad_norm": 2.147271156311035, + "learning_rate": 6.757350920582577e-07, + "loss": 0.9167, + "mean_token_accuracy": 0.7205410003662109, + "num_tokens": 74217538.0, + "step": 2460 + }, + { + "epoch": 0.20293559825183474, + "grad_norm": 2.069610118865967, + "learning_rate": 6.760098928276999e-07, + "loss": 0.829, + "mean_token_accuracy": 0.7468355894088745, + "num_tokens": 74244066.0, + "step": 2461 + }, + { + "epoch": 0.2030180588768863, + "grad_norm": 1.8735698461532593, + "learning_rate": 6.76284693597142e-07, + "loss": 0.8207, + "mean_token_accuracy": 0.7518970966339111, + "num_tokens": 74277735.0, + "step": 2462 + }, + { + "epoch": 0.20310051950193783, + "grad_norm": 1.976391315460205, + "learning_rate": 6.765594943665842e-07, + "loss": 0.7349, + "mean_token_accuracy": 0.764487624168396, + "num_tokens": 74310824.0, + "step": 2463 + }, + { + "epoch": 0.20318298012698935, + "grad_norm": 1.9785875082015991, + "learning_rate": 6.768342951360263e-07, + "loss": 0.7316, + "mean_token_accuracy": 0.7581707835197449, + "num_tokens": 74339884.0, + "step": 2464 + }, + { + "epoch": 0.2032654407520409, + "grad_norm": 2.0194079875946045, + "learning_rate": 6.771090959054685e-07, + "loss": 0.7241, + "mean_token_accuracy": 0.780013382434845, + "num_tokens": 74367782.0, + "step": 2465 + }, + { + "epoch": 0.20334790137709244, + "grad_norm": 1.9946694374084473, + "learning_rate": 6.773838966749107e-07, + "loss": 0.7757, + "mean_token_accuracy": 0.7673142552375793, + "num_tokens": 74395981.0, + "step": 2466 + }, + { + "epoch": 0.20343036200214398, + "grad_norm": 1.9899840354919434, + "learning_rate": 6.776586974443528e-07, + "loss": 0.8875, + "mean_token_accuracy": 0.7143080234527588, + "num_tokens": 74428408.0, + "step": 2467 + }, + { + "epoch": 0.20351282262719553, + "grad_norm": 1.8975634574890137, + "learning_rate": 6.77933498213795e-07, + "loss": 0.7698, + "mean_token_accuracy": 0.7615916728973389, + "num_tokens": 74459985.0, + "step": 2468 + }, + { + "epoch": 0.20359528325224704, + "grad_norm": 2.043147563934326, + "learning_rate": 6.782082989832372e-07, + "loss": 0.7736, + "mean_token_accuracy": 0.7499510645866394, + "num_tokens": 74492039.0, + "step": 2469 + }, + { + "epoch": 0.2036777438772986, + "grad_norm": 2.141563653945923, + "learning_rate": 6.784830997526792e-07, + "loss": 0.8801, + "mean_token_accuracy": 0.7348748445510864, + "num_tokens": 74520036.0, + "step": 2470 + }, + { + "epoch": 0.20376020450235013, + "grad_norm": 2.198479652404785, + "learning_rate": 6.787579005221214e-07, + "loss": 0.799, + "mean_token_accuracy": 0.7483334541320801, + "num_tokens": 74547128.0, + "step": 2471 + }, + { + "epoch": 0.20384266512740168, + "grad_norm": 2.198726177215576, + "learning_rate": 6.790327012915636e-07, + "loss": 0.8311, + "mean_token_accuracy": 0.7449815273284912, + "num_tokens": 74575748.0, + "step": 2472 + }, + { + "epoch": 0.2039251257524532, + "grad_norm": 1.9760295152664185, + "learning_rate": 6.793075020610058e-07, + "loss": 0.8297, + "mean_token_accuracy": 0.7451308965682983, + "num_tokens": 74608342.0, + "step": 2473 + }, + { + "epoch": 0.20400758637750474, + "grad_norm": 2.0895094871520996, + "learning_rate": 6.795823028304479e-07, + "loss": 0.979, + "mean_token_accuracy": 0.7085798978805542, + "num_tokens": 74639222.0, + "step": 2474 + }, + { + "epoch": 0.20409004700255629, + "grad_norm": 1.9322258234024048, + "learning_rate": 6.7985710359989e-07, + "loss": 0.7792, + "mean_token_accuracy": 0.7567487955093384, + "num_tokens": 74671059.0, + "step": 2475 + }, + { + "epoch": 0.20417250762760783, + "grad_norm": 2.1617491245269775, + "learning_rate": 6.801319043693322e-07, + "loss": 0.8892, + "mean_token_accuracy": 0.7208133935928345, + "num_tokens": 74698207.0, + "step": 2476 + }, + { + "epoch": 0.20425496825265935, + "grad_norm": 1.935829520225525, + "learning_rate": 6.804067051387744e-07, + "loss": 0.7099, + "mean_token_accuracy": 0.768328070640564, + "num_tokens": 74727469.0, + "step": 2477 + }, + { + "epoch": 0.2043374288777109, + "grad_norm": 2.0180952548980713, + "learning_rate": 6.806815059082165e-07, + "loss": 0.8026, + "mean_token_accuracy": 0.754700243473053, + "num_tokens": 74757246.0, + "step": 2478 + }, + { + "epoch": 0.20441988950276244, + "grad_norm": 1.9830302000045776, + "learning_rate": 6.809563066776587e-07, + "loss": 0.9976, + "mean_token_accuracy": 0.70119309425354, + "num_tokens": 74792138.0, + "step": 2479 + }, + { + "epoch": 0.20450235012781398, + "grad_norm": 1.9106488227844238, + "learning_rate": 6.812311074471007e-07, + "loss": 0.8236, + "mean_token_accuracy": 0.7413995265960693, + "num_tokens": 74822606.0, + "step": 2480 + }, + { + "epoch": 0.2045848107528655, + "grad_norm": 2.045018434524536, + "learning_rate": 6.81505908216543e-07, + "loss": 0.9387, + "mean_token_accuracy": 0.7161116003990173, + "num_tokens": 74856180.0, + "step": 2481 + }, + { + "epoch": 0.20466727137791704, + "grad_norm": 2.200775384902954, + "learning_rate": 6.817807089859852e-07, + "loss": 0.9475, + "mean_token_accuracy": 0.7111808061599731, + "num_tokens": 74884190.0, + "step": 2482 + }, + { + "epoch": 0.2047497320029686, + "grad_norm": 2.3386332988739014, + "learning_rate": 6.820555097554273e-07, + "loss": 0.9202, + "mean_token_accuracy": 0.7130080461502075, + "num_tokens": 74907618.0, + "step": 2483 + }, + { + "epoch": 0.20483219262802013, + "grad_norm": 2.2719566822052, + "learning_rate": 6.823303105248694e-07, + "loss": 0.7747, + "mean_token_accuracy": 0.741007387638092, + "num_tokens": 74936699.0, + "step": 2484 + }, + { + "epoch": 0.20491465325307165, + "grad_norm": 2.1441879272460938, + "learning_rate": 6.826051112943116e-07, + "loss": 0.9737, + "mean_token_accuracy": 0.709674596786499, + "num_tokens": 74966658.0, + "step": 2485 + }, + { + "epoch": 0.2049971138781232, + "grad_norm": 1.8996307849884033, + "learning_rate": 6.828799120637538e-07, + "loss": 0.8561, + "mean_token_accuracy": 0.7331873774528503, + "num_tokens": 74998215.0, + "step": 2486 + }, + { + "epoch": 0.20507957450317474, + "grad_norm": 1.915948748588562, + "learning_rate": 6.831547128331959e-07, + "loss": 0.9069, + "mean_token_accuracy": 0.7218433618545532, + "num_tokens": 75030182.0, + "step": 2487 + }, + { + "epoch": 0.20516203512822628, + "grad_norm": 1.9889494180679321, + "learning_rate": 6.83429513602638e-07, + "loss": 0.8724, + "mean_token_accuracy": 0.7316683530807495, + "num_tokens": 75061487.0, + "step": 2488 + }, + { + "epoch": 0.2052444957532778, + "grad_norm": 2.121086597442627, + "learning_rate": 6.837043143720802e-07, + "loss": 0.9808, + "mean_token_accuracy": 0.7020283341407776, + "num_tokens": 75089995.0, + "step": 2489 + }, + { + "epoch": 0.20532695637832934, + "grad_norm": 2.2707808017730713, + "learning_rate": 6.839791151415224e-07, + "loss": 0.9843, + "mean_token_accuracy": 0.703437864780426, + "num_tokens": 75116650.0, + "step": 2490 + }, + { + "epoch": 0.2054094170033809, + "grad_norm": 1.9602117538452148, + "learning_rate": 6.842539159109645e-07, + "loss": 0.8457, + "mean_token_accuracy": 0.7354024648666382, + "num_tokens": 75150140.0, + "step": 2491 + }, + { + "epoch": 0.20549187762843243, + "grad_norm": 2.43001389503479, + "learning_rate": 6.845287166804067e-07, + "loss": 0.9842, + "mean_token_accuracy": 0.6910221576690674, + "num_tokens": 75176460.0, + "step": 2492 + }, + { + "epoch": 0.20557433825348395, + "grad_norm": 1.7622265815734863, + "learning_rate": 6.848035174498489e-07, + "loss": 0.7781, + "mean_token_accuracy": 0.7654834985733032, + "num_tokens": 75213565.0, + "step": 2493 + }, + { + "epoch": 0.2056567988785355, + "grad_norm": 2.0166468620300293, + "learning_rate": 6.850783182192909e-07, + "loss": 0.7842, + "mean_token_accuracy": 0.7513344287872314, + "num_tokens": 75245203.0, + "step": 2494 + }, + { + "epoch": 0.20573925950358704, + "grad_norm": 1.7736374139785767, + "learning_rate": 6.853531189887332e-07, + "loss": 0.679, + "mean_token_accuracy": 0.7731654644012451, + "num_tokens": 75279140.0, + "step": 2495 + }, + { + "epoch": 0.20582172012863859, + "grad_norm": 1.9873363971710205, + "learning_rate": 6.856279197581753e-07, + "loss": 0.8799, + "mean_token_accuracy": 0.7349574565887451, + "num_tokens": 75310919.0, + "step": 2496 + }, + { + "epoch": 0.2059041807536901, + "grad_norm": 2.1203083992004395, + "learning_rate": 6.859027205276174e-07, + "loss": 0.8071, + "mean_token_accuracy": 0.7526293992996216, + "num_tokens": 75339821.0, + "step": 2497 + }, + { + "epoch": 0.20598664137874165, + "grad_norm": 1.9845811128616333, + "learning_rate": 6.861775212970596e-07, + "loss": 0.8303, + "mean_token_accuracy": 0.7281421422958374, + "num_tokens": 75371599.0, + "step": 2498 + }, + { + "epoch": 0.2060691020037932, + "grad_norm": 2.3194541931152344, + "learning_rate": 6.864523220665017e-07, + "loss": 0.7923, + "mean_token_accuracy": 0.7542937994003296, + "num_tokens": 75395638.0, + "step": 2499 + }, + { + "epoch": 0.20615156262884474, + "grad_norm": 1.7749850749969482, + "learning_rate": 6.86727122835944e-07, + "loss": 0.8152, + "mean_token_accuracy": 0.7445012331008911, + "num_tokens": 75433086.0, + "step": 2500 + }, + { + "epoch": 0.20623402325389625, + "grad_norm": 1.8247419595718384, + "learning_rate": 6.870019236053861e-07, + "loss": 0.8898, + "mean_token_accuracy": 0.7294207811355591, + "num_tokens": 75472134.0, + "step": 2501 + }, + { + "epoch": 0.2063164838789478, + "grad_norm": 2.004133701324463, + "learning_rate": 6.872767243748282e-07, + "loss": 0.908, + "mean_token_accuracy": 0.7185105085372925, + "num_tokens": 75502668.0, + "step": 2502 + }, + { + "epoch": 0.20639894450399934, + "grad_norm": 2.068565607070923, + "learning_rate": 6.875515251442704e-07, + "loss": 0.7618, + "mean_token_accuracy": 0.7656917572021484, + "num_tokens": 75534545.0, + "step": 2503 + }, + { + "epoch": 0.2064814051290509, + "grad_norm": 2.121993064880371, + "learning_rate": 6.878263259137124e-07, + "loss": 0.8695, + "mean_token_accuracy": 0.7341827154159546, + "num_tokens": 75560984.0, + "step": 2504 + }, + { + "epoch": 0.2065638657541024, + "grad_norm": 2.0719447135925293, + "learning_rate": 6.881011266831547e-07, + "loss": 0.8376, + "mean_token_accuracy": 0.7417665719985962, + "num_tokens": 75586502.0, + "step": 2505 + }, + { + "epoch": 0.20664632637915395, + "grad_norm": 2.332587957382202, + "learning_rate": 6.883759274525969e-07, + "loss": 0.8059, + "mean_token_accuracy": 0.7488102912902832, + "num_tokens": 75611105.0, + "step": 2506 + }, + { + "epoch": 0.2067287870042055, + "grad_norm": 2.005875825881958, + "learning_rate": 6.88650728222039e-07, + "loss": 0.8113, + "mean_token_accuracy": 0.7474141716957092, + "num_tokens": 75642231.0, + "step": 2507 + }, + { + "epoch": 0.20681124762925704, + "grad_norm": 2.509908437728882, + "learning_rate": 6.889255289914811e-07, + "loss": 0.9732, + "mean_token_accuracy": 0.7095094919204712, + "num_tokens": 75663710.0, + "step": 2508 + }, + { + "epoch": 0.20689370825430856, + "grad_norm": 2.087153673171997, + "learning_rate": 6.892003297609234e-07, + "loss": 0.8198, + "mean_token_accuracy": 0.7337878942489624, + "num_tokens": 75691610.0, + "step": 2509 + }, + { + "epoch": 0.2069761688793601, + "grad_norm": 2.088820695877075, + "learning_rate": 6.894751305303655e-07, + "loss": 0.8933, + "mean_token_accuracy": 0.7266222238540649, + "num_tokens": 75719475.0, + "step": 2510 + }, + { + "epoch": 0.20705862950441165, + "grad_norm": 1.823957085609436, + "learning_rate": 6.897499312998076e-07, + "loss": 0.7458, + "mean_token_accuracy": 0.7692508697509766, + "num_tokens": 75752360.0, + "step": 2511 + }, + { + "epoch": 0.2071410901294632, + "grad_norm": 2.0279221534729004, + "learning_rate": 6.900247320692497e-07, + "loss": 0.7617, + "mean_token_accuracy": 0.7581486701965332, + "num_tokens": 75782281.0, + "step": 2512 + }, + { + "epoch": 0.2072235507545147, + "grad_norm": 1.9279999732971191, + "learning_rate": 6.902995328386919e-07, + "loss": 0.814, + "mean_token_accuracy": 0.7543815970420837, + "num_tokens": 75814989.0, + "step": 2513 + }, + { + "epoch": 0.20730601137956625, + "grad_norm": 1.868293046951294, + "learning_rate": 6.905743336081342e-07, + "loss": 0.8514, + "mean_token_accuracy": 0.7274972200393677, + "num_tokens": 75852338.0, + "step": 2514 + }, + { + "epoch": 0.2073884720046178, + "grad_norm": 1.8475098609924316, + "learning_rate": 6.908491343775762e-07, + "loss": 0.671, + "mean_token_accuracy": 0.7882729768753052, + "num_tokens": 75885111.0, + "step": 2515 + }, + { + "epoch": 0.20747093262966934, + "grad_norm": 1.9838234186172485, + "learning_rate": 6.911239351470184e-07, + "loss": 0.8265, + "mean_token_accuracy": 0.7356150150299072, + "num_tokens": 75915911.0, + "step": 2516 + }, + { + "epoch": 0.20755339325472086, + "grad_norm": 1.7800005674362183, + "learning_rate": 6.913987359164606e-07, + "loss": 0.8101, + "mean_token_accuracy": 0.7561479210853577, + "num_tokens": 75950650.0, + "step": 2517 + }, + { + "epoch": 0.2076358538797724, + "grad_norm": 1.8716826438903809, + "learning_rate": 6.916735366859026e-07, + "loss": 0.7957, + "mean_token_accuracy": 0.7529168128967285, + "num_tokens": 75983396.0, + "step": 2518 + }, + { + "epoch": 0.20771831450482395, + "grad_norm": 2.3315091133117676, + "learning_rate": 6.919483374553449e-07, + "loss": 0.9697, + "mean_token_accuracy": 0.7012147307395935, + "num_tokens": 76010268.0, + "step": 2519 + }, + { + "epoch": 0.2078007751298755, + "grad_norm": 1.9190037250518799, + "learning_rate": 6.92223138224787e-07, + "loss": 1.0661, + "mean_token_accuracy": 0.6769826412200928, + "num_tokens": 76043134.0, + "step": 2520 + }, + { + "epoch": 0.207883235754927, + "grad_norm": 2.330970048904419, + "learning_rate": 6.924979389942291e-07, + "loss": 0.8723, + "mean_token_accuracy": 0.7268625497817993, + "num_tokens": 76068017.0, + "step": 2521 + }, + { + "epoch": 0.20796569637997855, + "grad_norm": 2.238008737564087, + "learning_rate": 6.927727397636713e-07, + "loss": 0.843, + "mean_token_accuracy": 0.7415707111358643, + "num_tokens": 76094034.0, + "step": 2522 + }, + { + "epoch": 0.2080481570050301, + "grad_norm": 2.144752025604248, + "learning_rate": 6.930475405331134e-07, + "loss": 0.8466, + "mean_token_accuracy": 0.7358875274658203, + "num_tokens": 76121203.0, + "step": 2523 + }, + { + "epoch": 0.20813061763008164, + "grad_norm": 2.088745355606079, + "learning_rate": 6.933223413025557e-07, + "loss": 0.7657, + "mean_token_accuracy": 0.7622612714767456, + "num_tokens": 76149960.0, + "step": 2524 + }, + { + "epoch": 0.20821307825513316, + "grad_norm": 1.9207991361618042, + "learning_rate": 6.935971420719978e-07, + "loss": 0.7495, + "mean_token_accuracy": 0.7643743753433228, + "num_tokens": 76181154.0, + "step": 2525 + }, + { + "epoch": 0.2082955388801847, + "grad_norm": 2.2294070720672607, + "learning_rate": 6.938719428414399e-07, + "loss": 0.8759, + "mean_token_accuracy": 0.7251396179199219, + "num_tokens": 76208086.0, + "step": 2526 + }, + { + "epoch": 0.20837799950523625, + "grad_norm": 1.9505566358566284, + "learning_rate": 6.941467436108821e-07, + "loss": 0.8563, + "mean_token_accuracy": 0.7322226762771606, + "num_tokens": 76241952.0, + "step": 2527 + }, + { + "epoch": 0.2084604601302878, + "grad_norm": 1.981136679649353, + "learning_rate": 6.944215443803242e-07, + "loss": 0.731, + "mean_token_accuracy": 0.777805507183075, + "num_tokens": 76272004.0, + "step": 2528 + }, + { + "epoch": 0.2085429207553393, + "grad_norm": 2.0996689796447754, + "learning_rate": 6.946963451497664e-07, + "loss": 0.7699, + "mean_token_accuracy": 0.7553051114082336, + "num_tokens": 76299095.0, + "step": 2529 + }, + { + "epoch": 0.20862538138039086, + "grad_norm": 1.9206740856170654, + "learning_rate": 6.949711459192086e-07, + "loss": 0.83, + "mean_token_accuracy": 0.7282752394676208, + "num_tokens": 76333200.0, + "step": 2530 + }, + { + "epoch": 0.2087078420054424, + "grad_norm": 2.066840648651123, + "learning_rate": 6.952459466886506e-07, + "loss": 0.9644, + "mean_token_accuracy": 0.7179713249206543, + "num_tokens": 76365639.0, + "step": 2531 + }, + { + "epoch": 0.20879030263049395, + "grad_norm": 1.84452223777771, + "learning_rate": 6.955207474580928e-07, + "loss": 0.7613, + "mean_token_accuracy": 0.7623421549797058, + "num_tokens": 76400485.0, + "step": 2532 + }, + { + "epoch": 0.2088727632555455, + "grad_norm": 1.941788911819458, + "learning_rate": 6.957955482275351e-07, + "loss": 0.8925, + "mean_token_accuracy": 0.7246928215026855, + "num_tokens": 76434932.0, + "step": 2533 + }, + { + "epoch": 0.208955223880597, + "grad_norm": 1.9189140796661377, + "learning_rate": 6.960703489969772e-07, + "loss": 1.0516, + "mean_token_accuracy": 0.6904180645942688, + "num_tokens": 76471619.0, + "step": 2534 + }, + { + "epoch": 0.20903768450564855, + "grad_norm": 1.9019864797592163, + "learning_rate": 6.963451497664193e-07, + "loss": 0.9016, + "mean_token_accuracy": 0.7248802185058594, + "num_tokens": 76506888.0, + "step": 2535 + }, + { + "epoch": 0.2091201451307001, + "grad_norm": 2.038156270980835, + "learning_rate": 6.966199505358614e-07, + "loss": 0.7081, + "mean_token_accuracy": 0.774146556854248, + "num_tokens": 76537102.0, + "step": 2536 + }, + { + "epoch": 0.20920260575575164, + "grad_norm": 1.7679612636566162, + "learning_rate": 6.968947513053036e-07, + "loss": 0.6466, + "mean_token_accuracy": 0.7925276160240173, + "num_tokens": 76573312.0, + "step": 2537 + }, + { + "epoch": 0.20928506638080316, + "grad_norm": 1.9558175802230835, + "learning_rate": 6.971695520747458e-07, + "loss": 0.8757, + "mean_token_accuracy": 0.7153781056404114, + "num_tokens": 76606126.0, + "step": 2538 + }, + { + "epoch": 0.2093675270058547, + "grad_norm": 2.285050868988037, + "learning_rate": 6.974443528441879e-07, + "loss": 1.0189, + "mean_token_accuracy": 0.6905347108840942, + "num_tokens": 76632561.0, + "step": 2539 + }, + { + "epoch": 0.20944998763090625, + "grad_norm": 1.9046037197113037, + "learning_rate": 6.977191536136301e-07, + "loss": 0.9583, + "mean_token_accuracy": 0.7064018845558167, + "num_tokens": 76665583.0, + "step": 2540 + }, + { + "epoch": 0.2095324482559578, + "grad_norm": 2.0538668632507324, + "learning_rate": 6.979939543830723e-07, + "loss": 0.8812, + "mean_token_accuracy": 0.7300233244895935, + "num_tokens": 76697142.0, + "step": 2541 + }, + { + "epoch": 0.2096149088810093, + "grad_norm": 2.005403757095337, + "learning_rate": 6.982687551525144e-07, + "loss": 0.7183, + "mean_token_accuracy": 0.7768757343292236, + "num_tokens": 76727905.0, + "step": 2542 + }, + { + "epoch": 0.20969736950606085, + "grad_norm": 1.816319227218628, + "learning_rate": 6.985435559219566e-07, + "loss": 0.9027, + "mean_token_accuracy": 0.7246758937835693, + "num_tokens": 76766047.0, + "step": 2543 + }, + { + "epoch": 0.2097798301311124, + "grad_norm": 2.1842923164367676, + "learning_rate": 6.988183566913987e-07, + "loss": 0.8601, + "mean_token_accuracy": 0.7293670177459717, + "num_tokens": 76799017.0, + "step": 2544 + }, + { + "epoch": 0.20986229075616394, + "grad_norm": 1.8667008876800537, + "learning_rate": 6.990931574608408e-07, + "loss": 0.7532, + "mean_token_accuracy": 0.7587798833847046, + "num_tokens": 76832724.0, + "step": 2545 + }, + { + "epoch": 0.20994475138121546, + "grad_norm": 1.9337178468704224, + "learning_rate": 6.99367958230283e-07, + "loss": 0.9613, + "mean_token_accuracy": 0.7138046622276306, + "num_tokens": 76867434.0, + "step": 2546 + }, + { + "epoch": 0.210027212006267, + "grad_norm": 2.1187047958374023, + "learning_rate": 6.996427589997252e-07, + "loss": 0.9082, + "mean_token_accuracy": 0.7213776111602783, + "num_tokens": 76895228.0, + "step": 2547 + }, + { + "epoch": 0.21010967263131855, + "grad_norm": 1.864626407623291, + "learning_rate": 6.999175597691673e-07, + "loss": 0.6566, + "mean_token_accuracy": 0.7813625931739807, + "num_tokens": 76923962.0, + "step": 2548 + }, + { + "epoch": 0.2101921332563701, + "grad_norm": 2.058652400970459, + "learning_rate": 7.001923605386095e-07, + "loss": 0.9611, + "mean_token_accuracy": 0.7068091034889221, + "num_tokens": 76955358.0, + "step": 2549 + }, + { + "epoch": 0.2102745938814216, + "grad_norm": 2.098440647125244, + "learning_rate": 7.004671613080516e-07, + "loss": 0.9369, + "mean_token_accuracy": 0.7152495980262756, + "num_tokens": 76988452.0, + "step": 2550 + }, + { + "epoch": 0.21035705450647316, + "grad_norm": 1.8303858041763306, + "learning_rate": 7.007419620774938e-07, + "loss": 0.7667, + "mean_token_accuracy": 0.7633402347564697, + "num_tokens": 77024191.0, + "step": 2551 + }, + { + "epoch": 0.2104395151315247, + "grad_norm": 2.0097382068634033, + "learning_rate": 7.010167628469359e-07, + "loss": 0.7955, + "mean_token_accuracy": 0.7371186017990112, + "num_tokens": 77055681.0, + "step": 2552 + }, + { + "epoch": 0.21052197575657625, + "grad_norm": 2.472935914993286, + "learning_rate": 7.012915636163781e-07, + "loss": 0.7879, + "mean_token_accuracy": 0.7474480867385864, + "num_tokens": 77078719.0, + "step": 2553 + }, + { + "epoch": 0.21060443638162776, + "grad_norm": 1.900888442993164, + "learning_rate": 7.015663643858203e-07, + "loss": 0.8542, + "mean_token_accuracy": 0.7384078502655029, + "num_tokens": 77110563.0, + "step": 2554 + }, + { + "epoch": 0.2106868970066793, + "grad_norm": 1.9481366872787476, + "learning_rate": 7.018411651552623e-07, + "loss": 0.8074, + "mean_token_accuracy": 0.7636147737503052, + "num_tokens": 77144189.0, + "step": 2555 + }, + { + "epoch": 0.21076935763173085, + "grad_norm": 2.0242562294006348, + "learning_rate": 7.021159659247046e-07, + "loss": 0.7437, + "mean_token_accuracy": 0.7681643962860107, + "num_tokens": 77171116.0, + "step": 2556 + }, + { + "epoch": 0.2108518182567824, + "grad_norm": 2.229240655899048, + "learning_rate": 7.023907666941468e-07, + "loss": 0.7612, + "mean_token_accuracy": 0.7626796960830688, + "num_tokens": 77197420.0, + "step": 2557 + }, + { + "epoch": 0.21093427888183391, + "grad_norm": 1.9900537729263306, + "learning_rate": 7.026655674635889e-07, + "loss": 0.8186, + "mean_token_accuracy": 0.7508015036582947, + "num_tokens": 77226729.0, + "step": 2558 + }, + { + "epoch": 0.21101673950688546, + "grad_norm": 2.0226597785949707, + "learning_rate": 7.02940368233031e-07, + "loss": 0.8531, + "mean_token_accuracy": 0.7368243336677551, + "num_tokens": 77258113.0, + "step": 2559 + }, + { + "epoch": 0.211099200131937, + "grad_norm": 2.0845298767089844, + "learning_rate": 7.032151690024731e-07, + "loss": 0.8651, + "mean_token_accuracy": 0.7365744113922119, + "num_tokens": 77290477.0, + "step": 2560 + }, + { + "epoch": 0.21118166075698855, + "grad_norm": 1.8578684329986572, + "learning_rate": 7.034899697719154e-07, + "loss": 0.7815, + "mean_token_accuracy": 0.7479082942008972, + "num_tokens": 77323622.0, + "step": 2561 + }, + { + "epoch": 0.21126412138204007, + "grad_norm": 2.1778135299682617, + "learning_rate": 7.037647705413575e-07, + "loss": 0.956, + "mean_token_accuracy": 0.7068285942077637, + "num_tokens": 77352115.0, + "step": 2562 + }, + { + "epoch": 0.2113465820070916, + "grad_norm": 1.9371334314346313, + "learning_rate": 7.040395713107996e-07, + "loss": 0.7161, + "mean_token_accuracy": 0.7671825885772705, + "num_tokens": 77384791.0, + "step": 2563 + }, + { + "epoch": 0.21142904263214315, + "grad_norm": 2.3543453216552734, + "learning_rate": 7.043143720802418e-07, + "loss": 0.8451, + "mean_token_accuracy": 0.7248322367668152, + "num_tokens": 77410693.0, + "step": 2564 + }, + { + "epoch": 0.2115115032571947, + "grad_norm": 1.8696959018707275, + "learning_rate": 7.04589172849684e-07, + "loss": 0.9025, + "mean_token_accuracy": 0.7246320843696594, + "num_tokens": 77444472.0, + "step": 2565 + }, + { + "epoch": 0.21159396388224622, + "grad_norm": 1.775354266166687, + "learning_rate": 7.048639736191261e-07, + "loss": 0.7172, + "mean_token_accuracy": 0.7585111260414124, + "num_tokens": 77479113.0, + "step": 2566 + }, + { + "epoch": 0.21167642450729776, + "grad_norm": 1.8938623666763306, + "learning_rate": 7.051387743885683e-07, + "loss": 0.8831, + "mean_token_accuracy": 0.7327471971511841, + "num_tokens": 77516532.0, + "step": 2567 + }, + { + "epoch": 0.2117588851323493, + "grad_norm": 1.9291421175003052, + "learning_rate": 7.054135751580104e-07, + "loss": 0.8348, + "mean_token_accuracy": 0.7428665161132812, + "num_tokens": 77548921.0, + "step": 2568 + }, + { + "epoch": 0.21184134575740085, + "grad_norm": 1.8494747877120972, + "learning_rate": 7.056883759274525e-07, + "loss": 0.7947, + "mean_token_accuracy": 0.7395891547203064, + "num_tokens": 77582564.0, + "step": 2569 + }, + { + "epoch": 0.21192380638245237, + "grad_norm": 1.9342761039733887, + "learning_rate": 7.059631766968948e-07, + "loss": 0.9294, + "mean_token_accuracy": 0.7194843888282776, + "num_tokens": 77619200.0, + "step": 2570 + }, + { + "epoch": 0.2120062670075039, + "grad_norm": 2.0069072246551514, + "learning_rate": 7.062379774663369e-07, + "loss": 0.8368, + "mean_token_accuracy": 0.74684739112854, + "num_tokens": 77652170.0, + "step": 2571 + }, + { + "epoch": 0.21208872763255546, + "grad_norm": 2.00477933883667, + "learning_rate": 7.06512778235779e-07, + "loss": 0.8592, + "mean_token_accuracy": 0.7405598163604736, + "num_tokens": 77683557.0, + "step": 2572 + }, + { + "epoch": 0.212171188257607, + "grad_norm": 2.0825307369232178, + "learning_rate": 7.067875790052212e-07, + "loss": 1.0015, + "mean_token_accuracy": 0.7055160403251648, + "num_tokens": 77712338.0, + "step": 2573 + }, + { + "epoch": 0.21225364888265852, + "grad_norm": 1.9988301992416382, + "learning_rate": 7.070623797746633e-07, + "loss": 0.7434, + "mean_token_accuracy": 0.7616744041442871, + "num_tokens": 77738735.0, + "step": 2574 + }, + { + "epoch": 0.21233610950771006, + "grad_norm": 2.019341230392456, + "learning_rate": 7.073371805441056e-07, + "loss": 0.7849, + "mean_token_accuracy": 0.748740017414093, + "num_tokens": 77770184.0, + "step": 2575 + }, + { + "epoch": 0.2124185701327616, + "grad_norm": 2.0448813438415527, + "learning_rate": 7.076119813135476e-07, + "loss": 0.8178, + "mean_token_accuracy": 0.7425835132598877, + "num_tokens": 77802373.0, + "step": 2576 + }, + { + "epoch": 0.21250103075781315, + "grad_norm": 2.1389284133911133, + "learning_rate": 7.078867820829898e-07, + "loss": 0.7131, + "mean_token_accuracy": 0.7689486145973206, + "num_tokens": 77830705.0, + "step": 2577 + }, + { + "epoch": 0.21258349138286467, + "grad_norm": 2.0414834022521973, + "learning_rate": 7.08161582852432e-07, + "loss": 0.6691, + "mean_token_accuracy": 0.7915334701538086, + "num_tokens": 77856747.0, + "step": 2578 + }, + { + "epoch": 0.21266595200791621, + "grad_norm": 1.914366364479065, + "learning_rate": 7.08436383621874e-07, + "loss": 0.8674, + "mean_token_accuracy": 0.739303708076477, + "num_tokens": 77890146.0, + "step": 2579 + }, + { + "epoch": 0.21274841263296776, + "grad_norm": 1.9332764148712158, + "learning_rate": 7.087111843913163e-07, + "loss": 0.8769, + "mean_token_accuracy": 0.7255654335021973, + "num_tokens": 77924328.0, + "step": 2580 + }, + { + "epoch": 0.2128308732580193, + "grad_norm": 2.1519219875335693, + "learning_rate": 7.089859851607585e-07, + "loss": 0.8295, + "mean_token_accuracy": 0.7356494665145874, + "num_tokens": 77954159.0, + "step": 2581 + }, + { + "epoch": 0.21291333388307082, + "grad_norm": 2.41947340965271, + "learning_rate": 7.092607859302005e-07, + "loss": 1.0256, + "mean_token_accuracy": 0.6896710991859436, + "num_tokens": 77981614.0, + "step": 2582 + }, + { + "epoch": 0.21299579450812237, + "grad_norm": 2.107351064682007, + "learning_rate": 7.095355866996427e-07, + "loss": 0.7488, + "mean_token_accuracy": 0.750404953956604, + "num_tokens": 78008563.0, + "step": 2583 + }, + { + "epoch": 0.2130782551331739, + "grad_norm": 2.073578119277954, + "learning_rate": 7.098103874690849e-07, + "loss": 1.0221, + "mean_token_accuracy": 0.7014495134353638, + "num_tokens": 78038319.0, + "step": 2584 + }, + { + "epoch": 0.21316071575822546, + "grad_norm": 2.0182485580444336, + "learning_rate": 7.100851882385271e-07, + "loss": 0.7779, + "mean_token_accuracy": 0.7540038824081421, + "num_tokens": 78067569.0, + "step": 2585 + }, + { + "epoch": 0.21324317638327697, + "grad_norm": 2.2926185131073, + "learning_rate": 7.103599890079692e-07, + "loss": 0.881, + "mean_token_accuracy": 0.728908360004425, + "num_tokens": 78090578.0, + "step": 2586 + }, + { + "epoch": 0.21332563700832852, + "grad_norm": 2.2848868370056152, + "learning_rate": 7.106347897774113e-07, + "loss": 1.0813, + "mean_token_accuracy": 0.6872992515563965, + "num_tokens": 78116095.0, + "step": 2587 + }, + { + "epoch": 0.21340809763338006, + "grad_norm": 1.718827724456787, + "learning_rate": 7.109095905468535e-07, + "loss": 0.7446, + "mean_token_accuracy": 0.7592388391494751, + "num_tokens": 78153648.0, + "step": 2588 + }, + { + "epoch": 0.2134905582584316, + "grad_norm": 1.7884694337844849, + "learning_rate": 7.111843913162957e-07, + "loss": 0.907, + "mean_token_accuracy": 0.7248749136924744, + "num_tokens": 78191368.0, + "step": 2589 + }, + { + "epoch": 0.21357301888348312, + "grad_norm": 2.1849851608276367, + "learning_rate": 7.114591920857378e-07, + "loss": 0.9714, + "mean_token_accuracy": 0.7137632369995117, + "num_tokens": 78218959.0, + "step": 2590 + }, + { + "epoch": 0.21365547950853467, + "grad_norm": 1.8605908155441284, + "learning_rate": 7.1173399285518e-07, + "loss": 0.7722, + "mean_token_accuracy": 0.7618352174758911, + "num_tokens": 78254071.0, + "step": 2591 + }, + { + "epoch": 0.2137379401335862, + "grad_norm": 2.211393117904663, + "learning_rate": 7.12008793624622e-07, + "loss": 0.8434, + "mean_token_accuracy": 0.7368955016136169, + "num_tokens": 78280066.0, + "step": 2592 + }, + { + "epoch": 0.21382040075863776, + "grad_norm": 2.0645883083343506, + "learning_rate": 7.122835943940642e-07, + "loss": 0.8247, + "mean_token_accuracy": 0.7433204650878906, + "num_tokens": 78311469.0, + "step": 2593 + }, + { + "epoch": 0.21390286138368927, + "grad_norm": 2.079779624938965, + "learning_rate": 7.125583951635065e-07, + "loss": 0.9088, + "mean_token_accuracy": 0.7217963933944702, + "num_tokens": 78338662.0, + "step": 2594 + }, + { + "epoch": 0.21398532200874082, + "grad_norm": 2.5880329608917236, + "learning_rate": 7.128331959329486e-07, + "loss": 0.88, + "mean_token_accuracy": 0.7291198372840881, + "num_tokens": 78371957.0, + "step": 2595 + }, + { + "epoch": 0.21406778263379236, + "grad_norm": 1.8570034503936768, + "learning_rate": 7.131079967023907e-07, + "loss": 0.8277, + "mean_token_accuracy": 0.7351164221763611, + "num_tokens": 78409231.0, + "step": 2596 + }, + { + "epoch": 0.2141502432588439, + "grad_norm": 2.0520904064178467, + "learning_rate": 7.133827974718329e-07, + "loss": 0.8329, + "mean_token_accuracy": 0.7395590543746948, + "num_tokens": 78437165.0, + "step": 2597 + }, + { + "epoch": 0.21423270388389545, + "grad_norm": 2.324014663696289, + "learning_rate": 7.136575982412751e-07, + "loss": 0.9093, + "mean_token_accuracy": 0.7182493805885315, + "num_tokens": 78461701.0, + "step": 2598 + }, + { + "epoch": 0.21431516450894697, + "grad_norm": 2.240835428237915, + "learning_rate": 7.139323990107173e-07, + "loss": 0.7683, + "mean_token_accuracy": 0.7554762363433838, + "num_tokens": 78489126.0, + "step": 2599 + }, + { + "epoch": 0.21439762513399852, + "grad_norm": 1.6322811841964722, + "learning_rate": 7.142071997801593e-07, + "loss": 0.7802, + "mean_token_accuracy": 0.7560683488845825, + "num_tokens": 78527492.0, + "step": 2600 + }, + { + "epoch": 0.21448008575905006, + "grad_norm": 1.9771836996078491, + "learning_rate": 7.144820005496015e-07, + "loss": 0.83, + "mean_token_accuracy": 0.7528209686279297, + "num_tokens": 78559696.0, + "step": 2601 + }, + { + "epoch": 0.2145625463841016, + "grad_norm": 2.1714627742767334, + "learning_rate": 7.147568013190437e-07, + "loss": 0.8121, + "mean_token_accuracy": 0.7432830929756165, + "num_tokens": 78586664.0, + "step": 2602 + }, + { + "epoch": 0.21464500700915312, + "grad_norm": 1.916172981262207, + "learning_rate": 7.150316020884858e-07, + "loss": 0.8606, + "mean_token_accuracy": 0.7357479333877563, + "num_tokens": 78620158.0, + "step": 2603 + }, + { + "epoch": 0.21472746763420467, + "grad_norm": 2.114642858505249, + "learning_rate": 7.15306402857928e-07, + "loss": 0.8569, + "mean_token_accuracy": 0.7377569079399109, + "num_tokens": 78647420.0, + "step": 2604 + }, + { + "epoch": 0.2148099282592562, + "grad_norm": 2.0424909591674805, + "learning_rate": 7.155812036273702e-07, + "loss": 0.9657, + "mean_token_accuracy": 0.7087818384170532, + "num_tokens": 78678421.0, + "step": 2605 + }, + { + "epoch": 0.21489238888430776, + "grad_norm": 2.38671875, + "learning_rate": 7.158560043968122e-07, + "loss": 0.8964, + "mean_token_accuracy": 0.723603367805481, + "num_tokens": 78702779.0, + "step": 2606 + }, + { + "epoch": 0.21497484950935927, + "grad_norm": 2.0334224700927734, + "learning_rate": 7.161308051662544e-07, + "loss": 0.7404, + "mean_token_accuracy": 0.7694445848464966, + "num_tokens": 78731567.0, + "step": 2607 + }, + { + "epoch": 0.21505731013441082, + "grad_norm": 2.188377857208252, + "learning_rate": 7.164056059356966e-07, + "loss": 1.0426, + "mean_token_accuracy": 0.6838760375976562, + "num_tokens": 78759493.0, + "step": 2608 + }, + { + "epoch": 0.21513977075946236, + "grad_norm": 1.908316969871521, + "learning_rate": 7.166804067051388e-07, + "loss": 0.8031, + "mean_token_accuracy": 0.7419538497924805, + "num_tokens": 78793445.0, + "step": 2609 + }, + { + "epoch": 0.2152222313845139, + "grad_norm": 2.051342010498047, + "learning_rate": 7.169552074745809e-07, + "loss": 1.0032, + "mean_token_accuracy": 0.7001985311508179, + "num_tokens": 78823963.0, + "step": 2610 + }, + { + "epoch": 0.21530469200956542, + "grad_norm": 2.2515769004821777, + "learning_rate": 7.17230008244023e-07, + "loss": 0.9783, + "mean_token_accuracy": 0.7009145021438599, + "num_tokens": 78847844.0, + "step": 2611 + }, + { + "epoch": 0.21538715263461697, + "grad_norm": 1.806711196899414, + "learning_rate": 7.175048090134653e-07, + "loss": 0.8485, + "mean_token_accuracy": 0.7423166632652283, + "num_tokens": 78885895.0, + "step": 2612 + }, + { + "epoch": 0.2154696132596685, + "grad_norm": 2.075956106185913, + "learning_rate": 7.177796097829074e-07, + "loss": 0.937, + "mean_token_accuracy": 0.7119300961494446, + "num_tokens": 78915358.0, + "step": 2613 + }, + { + "epoch": 0.21555207388472006, + "grad_norm": 1.8791463375091553, + "learning_rate": 7.180544105523495e-07, + "loss": 0.9054, + "mean_token_accuracy": 0.721845269203186, + "num_tokens": 78948054.0, + "step": 2614 + }, + { + "epoch": 0.21563453450977157, + "grad_norm": 2.08916974067688, + "learning_rate": 7.183292113217917e-07, + "loss": 0.9701, + "mean_token_accuracy": 0.7196394801139832, + "num_tokens": 78978442.0, + "step": 2615 + }, + { + "epoch": 0.21571699513482312, + "grad_norm": 1.8976163864135742, + "learning_rate": 7.186040120912337e-07, + "loss": 0.9162, + "mean_token_accuracy": 0.7289448380470276, + "num_tokens": 79009854.0, + "step": 2616 + }, + { + "epoch": 0.21579945575987466, + "grad_norm": 2.089588165283203, + "learning_rate": 7.18878812860676e-07, + "loss": 0.9436, + "mean_token_accuracy": 0.7143666744232178, + "num_tokens": 79039608.0, + "step": 2617 + }, + { + "epoch": 0.2158819163849262, + "grad_norm": 2.07148814201355, + "learning_rate": 7.191536136301182e-07, + "loss": 0.9224, + "mean_token_accuracy": 0.7231871485710144, + "num_tokens": 79070350.0, + "step": 2618 + }, + { + "epoch": 0.21596437700997773, + "grad_norm": 2.214442014694214, + "learning_rate": 7.194284143995603e-07, + "loss": 0.9366, + "mean_token_accuracy": 0.7138582468032837, + "num_tokens": 79097099.0, + "step": 2619 + }, + { + "epoch": 0.21604683763502927, + "grad_norm": 2.1279523372650146, + "learning_rate": 7.197032151690024e-07, + "loss": 0.9249, + "mean_token_accuracy": 0.717401385307312, + "num_tokens": 79126805.0, + "step": 2620 + }, + { + "epoch": 0.21612929826008082, + "grad_norm": 1.7335309982299805, + "learning_rate": 7.199780159384446e-07, + "loss": 0.7318, + "mean_token_accuracy": 0.7683326005935669, + "num_tokens": 79162246.0, + "step": 2621 + }, + { + "epoch": 0.21621175888513236, + "grad_norm": 1.860431432723999, + "learning_rate": 7.202528167078868e-07, + "loss": 0.6741, + "mean_token_accuracy": 0.771660327911377, + "num_tokens": 79194590.0, + "step": 2622 + }, + { + "epoch": 0.21629421951018388, + "grad_norm": 2.354156494140625, + "learning_rate": 7.205276174773289e-07, + "loss": 0.9977, + "mean_token_accuracy": 0.7051688432693481, + "num_tokens": 79220592.0, + "step": 2623 + }, + { + "epoch": 0.21637668013523542, + "grad_norm": 1.939192771911621, + "learning_rate": 7.20802418246771e-07, + "loss": 0.7921, + "mean_token_accuracy": 0.7545566558837891, + "num_tokens": 79252072.0, + "step": 2624 + }, + { + "epoch": 0.21645914076028697, + "grad_norm": 1.8877557516098022, + "learning_rate": 7.210772190162132e-07, + "loss": 0.9198, + "mean_token_accuracy": 0.719751238822937, + "num_tokens": 79286796.0, + "step": 2625 + }, + { + "epoch": 0.2165416013853385, + "grad_norm": 2.100605010986328, + "learning_rate": 7.213520197856555e-07, + "loss": 0.8563, + "mean_token_accuracy": 0.7311835885047913, + "num_tokens": 79316292.0, + "step": 2626 + }, + { + "epoch": 0.21662406201039003, + "grad_norm": 2.0194225311279297, + "learning_rate": 7.216268205550975e-07, + "loss": 0.8762, + "mean_token_accuracy": 0.72409987449646, + "num_tokens": 79348086.0, + "step": 2627 + }, + { + "epoch": 0.21670652263544157, + "grad_norm": 2.0408284664154053, + "learning_rate": 7.219016213245397e-07, + "loss": 0.9314, + "mean_token_accuracy": 0.7155554294586182, + "num_tokens": 79381633.0, + "step": 2628 + }, + { + "epoch": 0.21678898326049312, + "grad_norm": 2.060713529586792, + "learning_rate": 7.221764220939819e-07, + "loss": 0.89, + "mean_token_accuracy": 0.7337980270385742, + "num_tokens": 79412563.0, + "step": 2629 + }, + { + "epoch": 0.21687144388554466, + "grad_norm": 2.168191909790039, + "learning_rate": 7.224512228634239e-07, + "loss": 0.7673, + "mean_token_accuracy": 0.7646677494049072, + "num_tokens": 79439481.0, + "step": 2630 + }, + { + "epoch": 0.21695390451059618, + "grad_norm": 1.880114197731018, + "learning_rate": 7.227260236328662e-07, + "loss": 0.8756, + "mean_token_accuracy": 0.7386662364006042, + "num_tokens": 79473499.0, + "step": 2631 + }, + { + "epoch": 0.21703636513564772, + "grad_norm": 2.4373834133148193, + "learning_rate": 7.230008244023083e-07, + "loss": 0.8075, + "mean_token_accuracy": 0.7465602159500122, + "num_tokens": 79495655.0, + "step": 2632 + }, + { + "epoch": 0.21711882576069927, + "grad_norm": 2.171949863433838, + "learning_rate": 7.232756251717504e-07, + "loss": 0.9331, + "mean_token_accuracy": 0.7133431434631348, + "num_tokens": 79524773.0, + "step": 2633 + }, + { + "epoch": 0.2172012863857508, + "grad_norm": 2.3131394386291504, + "learning_rate": 7.235504259411926e-07, + "loss": 0.7788, + "mean_token_accuracy": 0.756553053855896, + "num_tokens": 79547464.0, + "step": 2634 + }, + { + "epoch": 0.21728374701080233, + "grad_norm": 2.080268144607544, + "learning_rate": 7.238252267106347e-07, + "loss": 0.8083, + "mean_token_accuracy": 0.7469620704650879, + "num_tokens": 79577090.0, + "step": 2635 + }, + { + "epoch": 0.21736620763585388, + "grad_norm": 1.9900994300842285, + "learning_rate": 7.24100027480077e-07, + "loss": 0.7646, + "mean_token_accuracy": 0.7551175355911255, + "num_tokens": 79612048.0, + "step": 2636 + }, + { + "epoch": 0.21744866826090542, + "grad_norm": 2.3780765533447266, + "learning_rate": 7.243748282495191e-07, + "loss": 0.7446, + "mean_token_accuracy": 0.7608609199523926, + "num_tokens": 79634534.0, + "step": 2637 + }, + { + "epoch": 0.21753112888595696, + "grad_norm": 2.2949235439300537, + "learning_rate": 7.246496290189612e-07, + "loss": 0.8775, + "mean_token_accuracy": 0.732521653175354, + "num_tokens": 79658746.0, + "step": 2638 + }, + { + "epoch": 0.21761358951100848, + "grad_norm": 1.924580454826355, + "learning_rate": 7.249244297884034e-07, + "loss": 0.8038, + "mean_token_accuracy": 0.7548014521598816, + "num_tokens": 79692645.0, + "step": 2639 + }, + { + "epoch": 0.21769605013606003, + "grad_norm": 1.7388584613800049, + "learning_rate": 7.251992305578454e-07, + "loss": 0.761, + "mean_token_accuracy": 0.7707446217536926, + "num_tokens": 79727472.0, + "step": 2640 + }, + { + "epoch": 0.21777851076111157, + "grad_norm": 2.2540159225463867, + "learning_rate": 7.254740313272877e-07, + "loss": 0.8185, + "mean_token_accuracy": 0.749592661857605, + "num_tokens": 79752624.0, + "step": 2641 + }, + { + "epoch": 0.21786097138616312, + "grad_norm": 1.9241427183151245, + "learning_rate": 7.257488320967299e-07, + "loss": 0.6961, + "mean_token_accuracy": 0.7822953462600708, + "num_tokens": 79782874.0, + "step": 2642 + }, + { + "epoch": 0.21794343201121463, + "grad_norm": 1.7766348123550415, + "learning_rate": 7.26023632866172e-07, + "loss": 0.8067, + "mean_token_accuracy": 0.7513980865478516, + "num_tokens": 79820499.0, + "step": 2643 + }, + { + "epoch": 0.21802589263626618, + "grad_norm": 1.9355740547180176, + "learning_rate": 7.262984336356141e-07, + "loss": 0.8134, + "mean_token_accuracy": 0.7477370500564575, + "num_tokens": 79855457.0, + "step": 2644 + }, + { + "epoch": 0.21810835326131772, + "grad_norm": 2.061707019805908, + "learning_rate": 7.265732344050564e-07, + "loss": 0.8, + "mean_token_accuracy": 0.7501004338264465, + "num_tokens": 79883809.0, + "step": 2645 + }, + { + "epoch": 0.21819081388636927, + "grad_norm": 1.9511520862579346, + "learning_rate": 7.268480351744985e-07, + "loss": 0.9353, + "mean_token_accuracy": 0.7140922546386719, + "num_tokens": 79917972.0, + "step": 2646 + }, + { + "epoch": 0.21827327451142078, + "grad_norm": 2.047419309616089, + "learning_rate": 7.271228359439406e-07, + "loss": 0.7096, + "mean_token_accuracy": 0.7807750701904297, + "num_tokens": 79942896.0, + "step": 2647 + }, + { + "epoch": 0.21835573513647233, + "grad_norm": 1.841664433479309, + "learning_rate": 7.273976367133827e-07, + "loss": 0.9177, + "mean_token_accuracy": 0.721673846244812, + "num_tokens": 79978023.0, + "step": 2648 + }, + { + "epoch": 0.21843819576152387, + "grad_norm": 2.137723445892334, + "learning_rate": 7.276724374828249e-07, + "loss": 0.7942, + "mean_token_accuracy": 0.7490983605384827, + "num_tokens": 80003736.0, + "step": 2649 + }, + { + "epoch": 0.21852065638657542, + "grad_norm": 2.0686659812927246, + "learning_rate": 7.279472382522672e-07, + "loss": 0.9046, + "mean_token_accuracy": 0.7290642261505127, + "num_tokens": 80032420.0, + "step": 2650 + }, + { + "epoch": 0.21860311701162694, + "grad_norm": 2.0076894760131836, + "learning_rate": 7.282220390217092e-07, + "loss": 0.7896, + "mean_token_accuracy": 0.750324010848999, + "num_tokens": 80060385.0, + "step": 2651 + }, + { + "epoch": 0.21868557763667848, + "grad_norm": 2.029729127883911, + "learning_rate": 7.284968397911514e-07, + "loss": 0.7404, + "mean_token_accuracy": 0.7583057880401611, + "num_tokens": 80090958.0, + "step": 2652 + }, + { + "epoch": 0.21876803826173002, + "grad_norm": 2.0680272579193115, + "learning_rate": 7.287716405605935e-07, + "loss": 0.774, + "mean_token_accuracy": 0.7575745582580566, + "num_tokens": 80119767.0, + "step": 2653 + }, + { + "epoch": 0.21885049888678157, + "grad_norm": 2.227766275405884, + "learning_rate": 7.290464413300356e-07, + "loss": 0.9761, + "mean_token_accuracy": 0.6955506801605225, + "num_tokens": 80145176.0, + "step": 2654 + }, + { + "epoch": 0.2189329595118331, + "grad_norm": 1.8615151643753052, + "learning_rate": 7.293212420994779e-07, + "loss": 0.8858, + "mean_token_accuracy": 0.7332279086112976, + "num_tokens": 80181781.0, + "step": 2655 + }, + { + "epoch": 0.21901542013688463, + "grad_norm": 2.2116308212280273, + "learning_rate": 7.2959604286892e-07, + "loss": 0.8601, + "mean_token_accuracy": 0.7277252674102783, + "num_tokens": 80207606.0, + "step": 2656 + }, + { + "epoch": 0.21909788076193618, + "grad_norm": 1.9630944728851318, + "learning_rate": 7.298708436383621e-07, + "loss": 0.9403, + "mean_token_accuracy": 0.7174110412597656, + "num_tokens": 80241725.0, + "step": 2657 + }, + { + "epoch": 0.21918034138698772, + "grad_norm": 2.1692612171173096, + "learning_rate": 7.301456444078043e-07, + "loss": 0.9849, + "mean_token_accuracy": 0.6983025670051575, + "num_tokens": 80271254.0, + "step": 2658 + }, + { + "epoch": 0.21926280201203927, + "grad_norm": 2.0169284343719482, + "learning_rate": 7.304204451772465e-07, + "loss": 0.745, + "mean_token_accuracy": 0.7463425397872925, + "num_tokens": 80301557.0, + "step": 2659 + }, + { + "epoch": 0.21934526263709078, + "grad_norm": 2.002976655960083, + "learning_rate": 7.306952459466887e-07, + "loss": 0.9083, + "mean_token_accuracy": 0.7234466075897217, + "num_tokens": 80335745.0, + "step": 2660 + }, + { + "epoch": 0.21942772326214233, + "grad_norm": 2.0699961185455322, + "learning_rate": 7.309700467161307e-07, + "loss": 0.8149, + "mean_token_accuracy": 0.7428590655326843, + "num_tokens": 80363846.0, + "step": 2661 + }, + { + "epoch": 0.21951018388719387, + "grad_norm": 1.941754937171936, + "learning_rate": 7.312448474855729e-07, + "loss": 0.6985, + "mean_token_accuracy": 0.778656005859375, + "num_tokens": 80394819.0, + "step": 2662 + }, + { + "epoch": 0.21959264451224542, + "grad_norm": 2.2754716873168945, + "learning_rate": 7.315196482550151e-07, + "loss": 0.7758, + "mean_token_accuracy": 0.7485525608062744, + "num_tokens": 80419957.0, + "step": 2663 + }, + { + "epoch": 0.21967510513729693, + "grad_norm": 1.9189636707305908, + "learning_rate": 7.317944490244572e-07, + "loss": 0.6371, + "mean_token_accuracy": 0.7928805947303772, + "num_tokens": 80450779.0, + "step": 2664 + }, + { + "epoch": 0.21975756576234848, + "grad_norm": 1.9087531566619873, + "learning_rate": 7.320692497938994e-07, + "loss": 0.8209, + "mean_token_accuracy": 0.7434282302856445, + "num_tokens": 80483976.0, + "step": 2665 + }, + { + "epoch": 0.21984002638740002, + "grad_norm": 2.1500232219696045, + "learning_rate": 7.323440505633416e-07, + "loss": 0.8606, + "mean_token_accuracy": 0.7347559928894043, + "num_tokens": 80513808.0, + "step": 2666 + }, + { + "epoch": 0.21992248701245157, + "grad_norm": 2.083956480026245, + "learning_rate": 7.326188513327836e-07, + "loss": 0.8958, + "mean_token_accuracy": 0.7280943393707275, + "num_tokens": 80544331.0, + "step": 2667 + }, + { + "epoch": 0.22000494763750308, + "grad_norm": 2.1305809020996094, + "learning_rate": 7.328936521022258e-07, + "loss": 0.9035, + "mean_token_accuracy": 0.715561032295227, + "num_tokens": 80575213.0, + "step": 2668 + }, + { + "epoch": 0.22008740826255463, + "grad_norm": 2.310800075531006, + "learning_rate": 7.33168452871668e-07, + "loss": 0.8345, + "mean_token_accuracy": 0.7385168075561523, + "num_tokens": 80600973.0, + "step": 2669 + }, + { + "epoch": 0.22016986888760617, + "grad_norm": 2.3248167037963867, + "learning_rate": 7.334432536411102e-07, + "loss": 0.8333, + "mean_token_accuracy": 0.7386378645896912, + "num_tokens": 80625051.0, + "step": 2670 + }, + { + "epoch": 0.22025232951265772, + "grad_norm": 2.081845283508301, + "learning_rate": 7.337180544105523e-07, + "loss": 0.9629, + "mean_token_accuracy": 0.714725136756897, + "num_tokens": 80654850.0, + "step": 2671 + }, + { + "epoch": 0.22033479013770924, + "grad_norm": 2.070319890975952, + "learning_rate": 7.339928551799944e-07, + "loss": 0.7194, + "mean_token_accuracy": 0.7694286108016968, + "num_tokens": 80682386.0, + "step": 2672 + }, + { + "epoch": 0.22041725076276078, + "grad_norm": 1.9324480295181274, + "learning_rate": 7.342676559494367e-07, + "loss": 0.9256, + "mean_token_accuracy": 0.7102110385894775, + "num_tokens": 80714180.0, + "step": 2673 + }, + { + "epoch": 0.22049971138781232, + "grad_norm": 1.8199090957641602, + "learning_rate": 7.345424567188788e-07, + "loss": 0.8235, + "mean_token_accuracy": 0.7432385087013245, + "num_tokens": 80749092.0, + "step": 2674 + }, + { + "epoch": 0.22058217201286387, + "grad_norm": 2.577787399291992, + "learning_rate": 7.348172574883209e-07, + "loss": 0.9883, + "mean_token_accuracy": 0.706494927406311, + "num_tokens": 80770075.0, + "step": 2675 + }, + { + "epoch": 0.2206646326379154, + "grad_norm": 2.0282204151153564, + "learning_rate": 7.350920582577631e-07, + "loss": 0.9713, + "mean_token_accuracy": 0.7049773335456848, + "num_tokens": 80798985.0, + "step": 2676 + }, + { + "epoch": 0.22074709326296693, + "grad_norm": 2.137812852859497, + "learning_rate": 7.353668590272052e-07, + "loss": 0.8651, + "mean_token_accuracy": 0.7378124594688416, + "num_tokens": 80828229.0, + "step": 2677 + }, + { + "epoch": 0.22082955388801848, + "grad_norm": 2.0602471828460693, + "learning_rate": 7.356416597966474e-07, + "loss": 0.7901, + "mean_token_accuracy": 0.7556182742118835, + "num_tokens": 80854301.0, + "step": 2678 + }, + { + "epoch": 0.22091201451307002, + "grad_norm": 2.260612726211548, + "learning_rate": 7.359164605660896e-07, + "loss": 0.8816, + "mean_token_accuracy": 0.7410905361175537, + "num_tokens": 80880086.0, + "step": 2679 + }, + { + "epoch": 0.22099447513812154, + "grad_norm": 1.943347692489624, + "learning_rate": 7.361912613355317e-07, + "loss": 0.7391, + "mean_token_accuracy": 0.7764577269554138, + "num_tokens": 80910055.0, + "step": 2680 + }, + { + "epoch": 0.22107693576317308, + "grad_norm": 1.8877506256103516, + "learning_rate": 7.364660621049738e-07, + "loss": 0.8341, + "mean_token_accuracy": 0.7453892230987549, + "num_tokens": 80941967.0, + "step": 2681 + }, + { + "epoch": 0.22115939638822463, + "grad_norm": 1.9949097633361816, + "learning_rate": 7.36740862874416e-07, + "loss": 0.8137, + "mean_token_accuracy": 0.7432973384857178, + "num_tokens": 80971659.0, + "step": 2682 + }, + { + "epoch": 0.22124185701327617, + "grad_norm": 1.9620221853256226, + "learning_rate": 7.370156636438582e-07, + "loss": 0.8762, + "mean_token_accuracy": 0.7345893383026123, + "num_tokens": 81008436.0, + "step": 2683 + }, + { + "epoch": 0.2213243176383277, + "grad_norm": 2.140756607055664, + "learning_rate": 7.372904644133004e-07, + "loss": 1.0238, + "mean_token_accuracy": 0.6934177875518799, + "num_tokens": 81038716.0, + "step": 2684 + }, + { + "epoch": 0.22140677826337923, + "grad_norm": 1.845634937286377, + "learning_rate": 7.375652651827424e-07, + "loss": 0.7816, + "mean_token_accuracy": 0.7533012628555298, + "num_tokens": 81071893.0, + "step": 2685 + }, + { + "epoch": 0.22148923888843078, + "grad_norm": 1.9511548280715942, + "learning_rate": 7.378400659521846e-07, + "loss": 0.8592, + "mean_token_accuracy": 0.7406413555145264, + "num_tokens": 81106159.0, + "step": 2686 + }, + { + "epoch": 0.22157169951348232, + "grad_norm": 2.0543770790100098, + "learning_rate": 7.381148667216269e-07, + "loss": 0.8657, + "mean_token_accuracy": 0.7356932163238525, + "num_tokens": 81137687.0, + "step": 2687 + }, + { + "epoch": 0.22165416013853384, + "grad_norm": 2.016601085662842, + "learning_rate": 7.383896674910689e-07, + "loss": 1.0356, + "mean_token_accuracy": 0.6899042129516602, + "num_tokens": 81171085.0, + "step": 2688 + }, + { + "epoch": 0.22173662076358538, + "grad_norm": 2.1917824745178223, + "learning_rate": 7.386644682605111e-07, + "loss": 0.7016, + "mean_token_accuracy": 0.7654902338981628, + "num_tokens": 81197285.0, + "step": 2689 + }, + { + "epoch": 0.22181908138863693, + "grad_norm": 2.083423137664795, + "learning_rate": 7.389392690299533e-07, + "loss": 0.9934, + "mean_token_accuracy": 0.6950976252555847, + "num_tokens": 81227998.0, + "step": 2690 + }, + { + "epoch": 0.22190154201368847, + "grad_norm": 1.9582936763763428, + "learning_rate": 7.392140697993953e-07, + "loss": 0.6891, + "mean_token_accuracy": 0.7851238250732422, + "num_tokens": 81262841.0, + "step": 2691 + }, + { + "epoch": 0.22198400263874, + "grad_norm": 1.8565192222595215, + "learning_rate": 7.394888705688376e-07, + "loss": 0.8761, + "mean_token_accuracy": 0.7262952923774719, + "num_tokens": 81294915.0, + "step": 2692 + }, + { + "epoch": 0.22206646326379154, + "grad_norm": 1.8163657188415527, + "learning_rate": 7.397636713382797e-07, + "loss": 0.7144, + "mean_token_accuracy": 0.7770524024963379, + "num_tokens": 81329045.0, + "step": 2693 + }, + { + "epoch": 0.22214892388884308, + "grad_norm": 2.2384214401245117, + "learning_rate": 7.400384721077219e-07, + "loss": 0.8132, + "mean_token_accuracy": 0.7422913312911987, + "num_tokens": 81356897.0, + "step": 2694 + }, + { + "epoch": 0.22223138451389463, + "grad_norm": 1.9667843580245972, + "learning_rate": 7.40313272877164e-07, + "loss": 0.8821, + "mean_token_accuracy": 0.7298004627227783, + "num_tokens": 81389293.0, + "step": 2695 + }, + { + "epoch": 0.22231384513894614, + "grad_norm": 2.236814498901367, + "learning_rate": 7.405880736466061e-07, + "loss": 0.8408, + "mean_token_accuracy": 0.7422981262207031, + "num_tokens": 81415853.0, + "step": 2696 + }, + { + "epoch": 0.2223963057639977, + "grad_norm": 1.8606914281845093, + "learning_rate": 7.408628744160484e-07, + "loss": 0.8785, + "mean_token_accuracy": 0.7367094159126282, + "num_tokens": 81455844.0, + "step": 2697 + }, + { + "epoch": 0.22247876638904923, + "grad_norm": 2.033363103866577, + "learning_rate": 7.411376751854905e-07, + "loss": 0.9368, + "mean_token_accuracy": 0.7161157131195068, + "num_tokens": 81486187.0, + "step": 2698 + }, + { + "epoch": 0.22256122701410078, + "grad_norm": 1.754690170288086, + "learning_rate": 7.414124759549326e-07, + "loss": 0.8806, + "mean_token_accuracy": 0.7241355776786804, + "num_tokens": 81524363.0, + "step": 2699 + }, + { + "epoch": 0.2226436876391523, + "grad_norm": 2.0122618675231934, + "learning_rate": 7.416872767243748e-07, + "loss": 0.7441, + "mean_token_accuracy": 0.7639057636260986, + "num_tokens": 81551078.0, + "step": 2700 + }, + { + "epoch": 0.22272614826420384, + "grad_norm": 1.993978500366211, + "learning_rate": 7.419620774938168e-07, + "loss": 0.7713, + "mean_token_accuracy": 0.7616648077964783, + "num_tokens": 81580914.0, + "step": 2701 + }, + { + "epoch": 0.22280860888925538, + "grad_norm": 1.9176548719406128, + "learning_rate": 7.422368782632591e-07, + "loss": 0.7065, + "mean_token_accuracy": 0.7752344012260437, + "num_tokens": 81612666.0, + "step": 2702 + }, + { + "epoch": 0.22289106951430693, + "grad_norm": 2.2904014587402344, + "learning_rate": 7.425116790327013e-07, + "loss": 0.9897, + "mean_token_accuracy": 0.6992826461791992, + "num_tokens": 81641730.0, + "step": 2703 + }, + { + "epoch": 0.22297353013935844, + "grad_norm": 2.2665092945098877, + "learning_rate": 7.427864798021434e-07, + "loss": 0.8086, + "mean_token_accuracy": 0.7467564940452576, + "num_tokens": 81667158.0, + "step": 2704 + }, + { + "epoch": 0.22305599076441, + "grad_norm": 2.5696895122528076, + "learning_rate": 7.430612805715855e-07, + "loss": 0.8339, + "mean_token_accuracy": 0.738223671913147, + "num_tokens": 81686470.0, + "step": 2705 + }, + { + "epoch": 0.22313845138946153, + "grad_norm": 2.0267248153686523, + "learning_rate": 7.433360813410278e-07, + "loss": 0.7391, + "mean_token_accuracy": 0.7700996398925781, + "num_tokens": 81716681.0, + "step": 2706 + }, + { + "epoch": 0.22322091201451308, + "grad_norm": 1.8988227844238281, + "learning_rate": 7.436108821104699e-07, + "loss": 0.7463, + "mean_token_accuracy": 0.7663331627845764, + "num_tokens": 81747917.0, + "step": 2707 + }, + { + "epoch": 0.2233033726395646, + "grad_norm": 1.9812475442886353, + "learning_rate": 7.43885682879912e-07, + "loss": 0.8486, + "mean_token_accuracy": 0.7394485473632812, + "num_tokens": 81780523.0, + "step": 2708 + }, + { + "epoch": 0.22338583326461614, + "grad_norm": 2.0381786823272705, + "learning_rate": 7.441604836493541e-07, + "loss": 0.7864, + "mean_token_accuracy": 0.7559022903442383, + "num_tokens": 81810806.0, + "step": 2709 + }, + { + "epoch": 0.22346829388966769, + "grad_norm": 2.0944156646728516, + "learning_rate": 7.444352844187963e-07, + "loss": 0.8637, + "mean_token_accuracy": 0.7288573980331421, + "num_tokens": 81841336.0, + "step": 2710 + }, + { + "epoch": 0.22355075451471923, + "grad_norm": 2.2053921222686768, + "learning_rate": 7.447100851882386e-07, + "loss": 0.9156, + "mean_token_accuracy": 0.7279903888702393, + "num_tokens": 81866296.0, + "step": 2711 + }, + { + "epoch": 0.22363321513977075, + "grad_norm": 1.9981094598770142, + "learning_rate": 7.449848859576806e-07, + "loss": 0.8425, + "mean_token_accuracy": 0.7445144653320312, + "num_tokens": 81896390.0, + "step": 2712 + }, + { + "epoch": 0.2237156757648223, + "grad_norm": 1.983343243598938, + "learning_rate": 7.452596867271228e-07, + "loss": 0.7643, + "mean_token_accuracy": 0.7683299779891968, + "num_tokens": 81926354.0, + "step": 2713 + }, + { + "epoch": 0.22379813638987384, + "grad_norm": 2.31760835647583, + "learning_rate": 7.45534487496565e-07, + "loss": 0.8889, + "mean_token_accuracy": 0.7217978239059448, + "num_tokens": 81951573.0, + "step": 2714 + }, + { + "epoch": 0.22388059701492538, + "grad_norm": 1.9906755685806274, + "learning_rate": 7.45809288266007e-07, + "loss": 0.9838, + "mean_token_accuracy": 0.7052680253982544, + "num_tokens": 81983746.0, + "step": 2715 + }, + { + "epoch": 0.2239630576399769, + "grad_norm": 2.002131938934326, + "learning_rate": 7.460840890354493e-07, + "loss": 0.8862, + "mean_token_accuracy": 0.7326591610908508, + "num_tokens": 82012842.0, + "step": 2716 + }, + { + "epoch": 0.22404551826502844, + "grad_norm": 1.9373810291290283, + "learning_rate": 7.463588898048914e-07, + "loss": 0.9352, + "mean_token_accuracy": 0.713107705116272, + "num_tokens": 82044507.0, + "step": 2717 + }, + { + "epoch": 0.22412797889008, + "grad_norm": 2.0683650970458984, + "learning_rate": 7.466336905743335e-07, + "loss": 0.8827, + "mean_token_accuracy": 0.7363553643226624, + "num_tokens": 82073714.0, + "step": 2718 + }, + { + "epoch": 0.22421043951513153, + "grad_norm": 2.036487579345703, + "learning_rate": 7.469084913437757e-07, + "loss": 0.7679, + "mean_token_accuracy": 0.7516089081764221, + "num_tokens": 82099995.0, + "step": 2719 + }, + { + "epoch": 0.22429290014018305, + "grad_norm": 2.056147336959839, + "learning_rate": 7.471832921132179e-07, + "loss": 0.9172, + "mean_token_accuracy": 0.7166299819946289, + "num_tokens": 82129122.0, + "step": 2720 + }, + { + "epoch": 0.2243753607652346, + "grad_norm": 1.9119911193847656, + "learning_rate": 7.474580928826601e-07, + "loss": 0.8561, + "mean_token_accuracy": 0.7265573740005493, + "num_tokens": 82164480.0, + "step": 2721 + }, + { + "epoch": 0.22445782139028614, + "grad_norm": 1.989681363105774, + "learning_rate": 7.477328936521022e-07, + "loss": 0.7639, + "mean_token_accuracy": 0.7619332075119019, + "num_tokens": 82194668.0, + "step": 2722 + }, + { + "epoch": 0.22454028201533768, + "grad_norm": 1.9485268592834473, + "learning_rate": 7.480076944215443e-07, + "loss": 0.9006, + "mean_token_accuracy": 0.7300729155540466, + "num_tokens": 82227536.0, + "step": 2723 + }, + { + "epoch": 0.22462274264038923, + "grad_norm": 2.053983211517334, + "learning_rate": 7.482824951909865e-07, + "loss": 0.9109, + "mean_token_accuracy": 0.7255200147628784, + "num_tokens": 82258952.0, + "step": 2724 + }, + { + "epoch": 0.22470520326544074, + "grad_norm": 1.8732943534851074, + "learning_rate": 7.485572959604286e-07, + "loss": 0.8547, + "mean_token_accuracy": 0.7379970550537109, + "num_tokens": 82293225.0, + "step": 2725 + }, + { + "epoch": 0.2247876638904923, + "grad_norm": 1.7994011640548706, + "learning_rate": 7.488320967298708e-07, + "loss": 0.6895, + "mean_token_accuracy": 0.7814545631408691, + "num_tokens": 82326982.0, + "step": 2726 + }, + { + "epoch": 0.22487012451554383, + "grad_norm": 1.9175763130187988, + "learning_rate": 7.49106897499313e-07, + "loss": 0.8678, + "mean_token_accuracy": 0.7438633441925049, + "num_tokens": 82360924.0, + "step": 2727 + }, + { + "epoch": 0.22495258514059538, + "grad_norm": 1.9093258380889893, + "learning_rate": 7.493816982687551e-07, + "loss": 0.8566, + "mean_token_accuracy": 0.737669050693512, + "num_tokens": 82392548.0, + "step": 2728 + }, + { + "epoch": 0.2250350457656469, + "grad_norm": 1.8853206634521484, + "learning_rate": 7.496564990381972e-07, + "loss": 0.7317, + "mean_token_accuracy": 0.7739404439926147, + "num_tokens": 82427648.0, + "step": 2729 + }, + { + "epoch": 0.22511750639069844, + "grad_norm": 2.0017471313476562, + "learning_rate": 7.499312998076395e-07, + "loss": 0.8146, + "mean_token_accuracy": 0.741174578666687, + "num_tokens": 82456236.0, + "step": 2730 + }, + { + "epoch": 0.22519996701574999, + "grad_norm": 1.9144294261932373, + "learning_rate": 7.502061005770816e-07, + "loss": 0.7365, + "mean_token_accuracy": 0.7654203772544861, + "num_tokens": 82485545.0, + "step": 2731 + }, + { + "epoch": 0.22528242764080153, + "grad_norm": 2.0884602069854736, + "learning_rate": 7.504809013465237e-07, + "loss": 0.9211, + "mean_token_accuracy": 0.7230295538902283, + "num_tokens": 82519482.0, + "step": 2732 + }, + { + "epoch": 0.22536488826585305, + "grad_norm": 1.9272549152374268, + "learning_rate": 7.507557021159658e-07, + "loss": 0.6324, + "mean_token_accuracy": 0.7973251342773438, + "num_tokens": 82552268.0, + "step": 2733 + }, + { + "epoch": 0.2254473488909046, + "grad_norm": 2.034020185470581, + "learning_rate": 7.510305028854081e-07, + "loss": 0.879, + "mean_token_accuracy": 0.7251454591751099, + "num_tokens": 82583081.0, + "step": 2734 + }, + { + "epoch": 0.22552980951595614, + "grad_norm": 1.8792716264724731, + "learning_rate": 7.513053036548503e-07, + "loss": 0.7896, + "mean_token_accuracy": 0.757631242275238, + "num_tokens": 82619997.0, + "step": 2735 + }, + { + "epoch": 0.22561227014100768, + "grad_norm": 1.773738980293274, + "learning_rate": 7.515801044242923e-07, + "loss": 0.6843, + "mean_token_accuracy": 0.7800796031951904, + "num_tokens": 82655769.0, + "step": 2736 + }, + { + "epoch": 0.2256947307660592, + "grad_norm": 2.0277464389801025, + "learning_rate": 7.518549051937345e-07, + "loss": 0.7943, + "mean_token_accuracy": 0.7597637176513672, + "num_tokens": 82685149.0, + "step": 2737 + }, + { + "epoch": 0.22577719139111074, + "grad_norm": 2.093170404434204, + "learning_rate": 7.521297059631767e-07, + "loss": 0.8715, + "mean_token_accuracy": 0.729781448841095, + "num_tokens": 82713583.0, + "step": 2738 + }, + { + "epoch": 0.2258596520161623, + "grad_norm": 2.0475213527679443, + "learning_rate": 7.524045067326188e-07, + "loss": 0.7243, + "mean_token_accuracy": 0.768130362033844, + "num_tokens": 82740975.0, + "step": 2739 + }, + { + "epoch": 0.22594211264121383, + "grad_norm": 1.9681035280227661, + "learning_rate": 7.52679307502061e-07, + "loss": 0.8344, + "mean_token_accuracy": 0.7423874139785767, + "num_tokens": 82772009.0, + "step": 2740 + }, + { + "epoch": 0.22602457326626535, + "grad_norm": 1.9357913732528687, + "learning_rate": 7.529541082715031e-07, + "loss": 0.8379, + "mean_token_accuracy": 0.7354817390441895, + "num_tokens": 82803620.0, + "step": 2741 + }, + { + "epoch": 0.2261070338913169, + "grad_norm": 1.945307970046997, + "learning_rate": 7.532289090409452e-07, + "loss": 0.8015, + "mean_token_accuracy": 0.742707371711731, + "num_tokens": 82833145.0, + "step": 2742 + }, + { + "epoch": 0.22618949451636844, + "grad_norm": 2.041048049926758, + "learning_rate": 7.535037098103874e-07, + "loss": 0.8218, + "mean_token_accuracy": 0.7430242300033569, + "num_tokens": 82862899.0, + "step": 2743 + }, + { + "epoch": 0.22627195514141998, + "grad_norm": 1.9321088790893555, + "learning_rate": 7.537785105798296e-07, + "loss": 0.7385, + "mean_token_accuracy": 0.765122652053833, + "num_tokens": 82895470.0, + "step": 2744 + }, + { + "epoch": 0.2263544157664715, + "grad_norm": 2.25854229927063, + "learning_rate": 7.540533113492718e-07, + "loss": 0.8163, + "mean_token_accuracy": 0.7504265308380127, + "num_tokens": 82926947.0, + "step": 2745 + }, + { + "epoch": 0.22643687639152305, + "grad_norm": 2.0461721420288086, + "learning_rate": 7.543281121187139e-07, + "loss": 0.8567, + "mean_token_accuracy": 0.7293409705162048, + "num_tokens": 82955959.0, + "step": 2746 + }, + { + "epoch": 0.2265193370165746, + "grad_norm": 2.0112550258636475, + "learning_rate": 7.54602912888156e-07, + "loss": 0.8902, + "mean_token_accuracy": 0.7330959439277649, + "num_tokens": 82987975.0, + "step": 2747 + }, + { + "epoch": 0.22660179764162613, + "grad_norm": 2.025219202041626, + "learning_rate": 7.548777136575983e-07, + "loss": 0.7151, + "mean_token_accuracy": 0.7688711881637573, + "num_tokens": 83017263.0, + "step": 2748 + }, + { + "epoch": 0.22668425826667765, + "grad_norm": 2.1602489948272705, + "learning_rate": 7.551525144270403e-07, + "loss": 0.9616, + "mean_token_accuracy": 0.7092478275299072, + "num_tokens": 83045339.0, + "step": 2749 + }, + { + "epoch": 0.2267667188917292, + "grad_norm": 2.4207115173339844, + "learning_rate": 7.554273151964825e-07, + "loss": 0.9099, + "mean_token_accuracy": 0.725718080997467, + "num_tokens": 83069022.0, + "step": 2750 + }, + { + "epoch": 0.22684917951678074, + "grad_norm": 1.8709416389465332, + "learning_rate": 7.557021159659247e-07, + "loss": 0.9085, + "mean_token_accuracy": 0.7208677530288696, + "num_tokens": 83106241.0, + "step": 2751 + }, + { + "epoch": 0.22693164014183229, + "grad_norm": 1.9164544343948364, + "learning_rate": 7.559769167353667e-07, + "loss": 0.8449, + "mean_token_accuracy": 0.730628490447998, + "num_tokens": 83139996.0, + "step": 2752 + }, + { + "epoch": 0.2270141007668838, + "grad_norm": 2.028219223022461, + "learning_rate": 7.56251717504809e-07, + "loss": 0.8451, + "mean_token_accuracy": 0.7414940595626831, + "num_tokens": 83171404.0, + "step": 2753 + }, + { + "epoch": 0.22709656139193535, + "grad_norm": 2.0516176223754883, + "learning_rate": 7.565265182742512e-07, + "loss": 0.8812, + "mean_token_accuracy": 0.7327568531036377, + "num_tokens": 83204101.0, + "step": 2754 + }, + { + "epoch": 0.2271790220169869, + "grad_norm": 1.9659785032272339, + "learning_rate": 7.568013190436933e-07, + "loss": 0.8043, + "mean_token_accuracy": 0.7436848878860474, + "num_tokens": 83234498.0, + "step": 2755 + }, + { + "epoch": 0.22726148264203844, + "grad_norm": 2.4082181453704834, + "learning_rate": 7.570761198131354e-07, + "loss": 0.7698, + "mean_token_accuracy": 0.7593070268630981, + "num_tokens": 83256164.0, + "step": 2756 + }, + { + "epoch": 0.22734394326708995, + "grad_norm": 2.0043561458587646, + "learning_rate": 7.573509205825775e-07, + "loss": 0.9216, + "mean_token_accuracy": 0.7154951095581055, + "num_tokens": 83285942.0, + "step": 2757 + }, + { + "epoch": 0.2274264038921415, + "grad_norm": 2.049314022064209, + "learning_rate": 7.576257213520198e-07, + "loss": 0.8903, + "mean_token_accuracy": 0.732976496219635, + "num_tokens": 83316856.0, + "step": 2758 + }, + { + "epoch": 0.22750886451719304, + "grad_norm": 2.4728806018829346, + "learning_rate": 7.57900522121462e-07, + "loss": 0.9002, + "mean_token_accuracy": 0.7358952760696411, + "num_tokens": 83338707.0, + "step": 2759 + }, + { + "epoch": 0.2275913251422446, + "grad_norm": 2.0857841968536377, + "learning_rate": 7.58175322890904e-07, + "loss": 0.9111, + "mean_token_accuracy": 0.7132269144058228, + "num_tokens": 83371158.0, + "step": 2760 + }, + { + "epoch": 0.2276737857672961, + "grad_norm": 2.1351962089538574, + "learning_rate": 7.584501236603462e-07, + "loss": 0.8344, + "mean_token_accuracy": 0.7346871495246887, + "num_tokens": 83398415.0, + "step": 2761 + }, + { + "epoch": 0.22775624639234765, + "grad_norm": 2.2958686351776123, + "learning_rate": 7.587249244297885e-07, + "loss": 0.8093, + "mean_token_accuracy": 0.7456135749816895, + "num_tokens": 83422050.0, + "step": 2762 + }, + { + "epoch": 0.2278387070173992, + "grad_norm": 2.132596969604492, + "learning_rate": 7.589997251992305e-07, + "loss": 0.8359, + "mean_token_accuracy": 0.7495065927505493, + "num_tokens": 83448293.0, + "step": 2763 + }, + { + "epoch": 0.22792116764245074, + "grad_norm": 2.2850074768066406, + "learning_rate": 7.592745259686727e-07, + "loss": 1.0884, + "mean_token_accuracy": 0.6784820556640625, + "num_tokens": 83476269.0, + "step": 2764 + }, + { + "epoch": 0.22800362826750226, + "grad_norm": 2.019930362701416, + "learning_rate": 7.595493267381148e-07, + "loss": 0.8091, + "mean_token_accuracy": 0.7419009208679199, + "num_tokens": 83504306.0, + "step": 2765 + }, + { + "epoch": 0.2280860888925538, + "grad_norm": 1.9411935806274414, + "learning_rate": 7.598241275075569e-07, + "loss": 0.7654, + "mean_token_accuracy": 0.7529876232147217, + "num_tokens": 83536914.0, + "step": 2766 + }, + { + "epoch": 0.22816854951760535, + "grad_norm": 1.8320293426513672, + "learning_rate": 7.600989282769992e-07, + "loss": 0.7981, + "mean_token_accuracy": 0.7368391752243042, + "num_tokens": 83572248.0, + "step": 2767 + }, + { + "epoch": 0.2282510101426569, + "grad_norm": 2.1304712295532227, + "learning_rate": 7.603737290464413e-07, + "loss": 0.8258, + "mean_token_accuracy": 0.7428147196769714, + "num_tokens": 83597529.0, + "step": 2768 + }, + { + "epoch": 0.2283334707677084, + "grad_norm": 2.1193795204162598, + "learning_rate": 7.606485298158835e-07, + "loss": 0.7826, + "mean_token_accuracy": 0.7494561672210693, + "num_tokens": 83624032.0, + "step": 2769 + }, + { + "epoch": 0.22841593139275995, + "grad_norm": 1.944008231163025, + "learning_rate": 7.609233305853256e-07, + "loss": 0.781, + "mean_token_accuracy": 0.7598984241485596, + "num_tokens": 83653163.0, + "step": 2770 + }, + { + "epoch": 0.2284983920178115, + "grad_norm": 2.1207051277160645, + "learning_rate": 7.611981313547677e-07, + "loss": 0.8951, + "mean_token_accuracy": 0.7300752401351929, + "num_tokens": 83681704.0, + "step": 2771 + }, + { + "epoch": 0.22858085264286304, + "grad_norm": 2.261524200439453, + "learning_rate": 7.6147293212421e-07, + "loss": 0.7348, + "mean_token_accuracy": 0.7720218300819397, + "num_tokens": 83703969.0, + "step": 2772 + }, + { + "epoch": 0.22866331326791456, + "grad_norm": 2.06026554107666, + "learning_rate": 7.61747732893652e-07, + "loss": 0.8082, + "mean_token_accuracy": 0.7440429329872131, + "num_tokens": 83735557.0, + "step": 2773 + }, + { + "epoch": 0.2287457738929661, + "grad_norm": 2.127159833908081, + "learning_rate": 7.620225336630942e-07, + "loss": 0.9035, + "mean_token_accuracy": 0.7303279042243958, + "num_tokens": 83762646.0, + "step": 2774 + }, + { + "epoch": 0.22882823451801765, + "grad_norm": 2.047489643096924, + "learning_rate": 7.622973344325364e-07, + "loss": 0.7962, + "mean_token_accuracy": 0.7609827518463135, + "num_tokens": 83794480.0, + "step": 2775 + }, + { + "epoch": 0.2289106951430692, + "grad_norm": 2.0584232807159424, + "learning_rate": 7.625721352019785e-07, + "loss": 0.8363, + "mean_token_accuracy": 0.7318737506866455, + "num_tokens": 83823029.0, + "step": 2776 + }, + { + "epoch": 0.2289931557681207, + "grad_norm": 2.3708865642547607, + "learning_rate": 7.628469359714207e-07, + "loss": 0.9248, + "mean_token_accuracy": 0.7154517769813538, + "num_tokens": 83849302.0, + "step": 2777 + }, + { + "epoch": 0.22907561639317225, + "grad_norm": 2.0775604248046875, + "learning_rate": 7.631217367408629e-07, + "loss": 0.9349, + "mean_token_accuracy": 0.725847065448761, + "num_tokens": 83880143.0, + "step": 2778 + }, + { + "epoch": 0.2291580770182238, + "grad_norm": 2.0828380584716797, + "learning_rate": 7.63396537510305e-07, + "loss": 0.9332, + "mean_token_accuracy": 0.7166590094566345, + "num_tokens": 83913481.0, + "step": 2779 + }, + { + "epoch": 0.22924053764327534, + "grad_norm": 2.0978269577026367, + "learning_rate": 7.636713382797471e-07, + "loss": 0.8897, + "mean_token_accuracy": 0.7244192361831665, + "num_tokens": 83942032.0, + "step": 2780 + }, + { + "epoch": 0.22932299826832686, + "grad_norm": 1.835831880569458, + "learning_rate": 7.639461390491893e-07, + "loss": 0.7098, + "mean_token_accuracy": 0.7685962319374084, + "num_tokens": 83973260.0, + "step": 2781 + }, + { + "epoch": 0.2294054588933784, + "grad_norm": 2.024416923522949, + "learning_rate": 7.642209398186315e-07, + "loss": 0.8104, + "mean_token_accuracy": 0.7418243288993835, + "num_tokens": 84002581.0, + "step": 2782 + }, + { + "epoch": 0.22948791951842995, + "grad_norm": 1.9949839115142822, + "learning_rate": 7.644957405880736e-07, + "loss": 0.7233, + "mean_token_accuracy": 0.7681878209114075, + "num_tokens": 84031028.0, + "step": 2783 + }, + { + "epoch": 0.2295703801434815, + "grad_norm": 2.185600519180298, + "learning_rate": 7.647705413575157e-07, + "loss": 0.8462, + "mean_token_accuracy": 0.7563387155532837, + "num_tokens": 84059067.0, + "step": 2784 + }, + { + "epoch": 0.229652840768533, + "grad_norm": 2.103360414505005, + "learning_rate": 7.650453421269579e-07, + "loss": 0.8223, + "mean_token_accuracy": 0.7404800653457642, + "num_tokens": 84087295.0, + "step": 2785 + }, + { + "epoch": 0.22973530139358456, + "grad_norm": 2.3461496829986572, + "learning_rate": 7.653201428964002e-07, + "loss": 0.858, + "mean_token_accuracy": 0.7213935852050781, + "num_tokens": 84110885.0, + "step": 2786 + }, + { + "epoch": 0.2298177620186361, + "grad_norm": 2.0056755542755127, + "learning_rate": 7.655949436658422e-07, + "loss": 0.6957, + "mean_token_accuracy": 0.7650933265686035, + "num_tokens": 84141038.0, + "step": 2787 + }, + { + "epoch": 0.22990022264368765, + "grad_norm": 2.230100631713867, + "learning_rate": 7.658697444352844e-07, + "loss": 0.7233, + "mean_token_accuracy": 0.7675287127494812, + "num_tokens": 84165579.0, + "step": 2788 + }, + { + "epoch": 0.2299826832687392, + "grad_norm": 1.8540067672729492, + "learning_rate": 7.661445452047265e-07, + "loss": 0.7407, + "mean_token_accuracy": 0.7703126072883606, + "num_tokens": 84197267.0, + "step": 2789 + }, + { + "epoch": 0.2300651438937907, + "grad_norm": 2.224440336227417, + "learning_rate": 7.664193459741687e-07, + "loss": 0.8609, + "mean_token_accuracy": 0.7363563776016235, + "num_tokens": 84225704.0, + "step": 2790 + }, + { + "epoch": 0.23014760451884225, + "grad_norm": 2.3149590492248535, + "learning_rate": 7.666941467436109e-07, + "loss": 0.8085, + "mean_token_accuracy": 0.7409564256668091, + "num_tokens": 84249476.0, + "step": 2791 + }, + { + "epoch": 0.2302300651438938, + "grad_norm": 1.8049203157424927, + "learning_rate": 7.66968947513053e-07, + "loss": 0.7564, + "mean_token_accuracy": 0.7615553140640259, + "num_tokens": 84285206.0, + "step": 2792 + }, + { + "epoch": 0.23031252576894534, + "grad_norm": 2.2383718490600586, + "learning_rate": 7.672437482824951e-07, + "loss": 0.8323, + "mean_token_accuracy": 0.7447332739830017, + "num_tokens": 84311809.0, + "step": 2793 + }, + { + "epoch": 0.23039498639399686, + "grad_norm": 1.8661000728607178, + "learning_rate": 7.675185490519373e-07, + "loss": 0.8012, + "mean_token_accuracy": 0.7518207430839539, + "num_tokens": 84344547.0, + "step": 2794 + }, + { + "epoch": 0.2304774470190484, + "grad_norm": 2.418771505355835, + "learning_rate": 7.677933498213795e-07, + "loss": 0.8871, + "mean_token_accuracy": 0.7310587167739868, + "num_tokens": 84370128.0, + "step": 2795 + }, + { + "epoch": 0.23055990764409995, + "grad_norm": 1.9800655841827393, + "learning_rate": 7.680681505908217e-07, + "loss": 0.641, + "mean_token_accuracy": 0.789546012878418, + "num_tokens": 84402894.0, + "step": 2796 + }, + { + "epoch": 0.2306423682691515, + "grad_norm": 1.9105064868927002, + "learning_rate": 7.683429513602637e-07, + "loss": 0.7168, + "mean_token_accuracy": 0.7700726985931396, + "num_tokens": 84434271.0, + "step": 2797 + }, + { + "epoch": 0.230724828894203, + "grad_norm": 2.25258731842041, + "learning_rate": 7.686177521297059e-07, + "loss": 0.9194, + "mean_token_accuracy": 0.7218350172042847, + "num_tokens": 84462500.0, + "step": 2798 + }, + { + "epoch": 0.23080728951925455, + "grad_norm": 2.014096260070801, + "learning_rate": 7.688925528991481e-07, + "loss": 0.7746, + "mean_token_accuracy": 0.7631586194038391, + "num_tokens": 84490501.0, + "step": 2799 + }, + { + "epoch": 0.2308897501443061, + "grad_norm": 1.8319534063339233, + "learning_rate": 7.691673536685902e-07, + "loss": 0.7326, + "mean_token_accuracy": 0.7715283632278442, + "num_tokens": 84523772.0, + "step": 2800 + }, + { + "epoch": 0.23097221076935764, + "grad_norm": 2.2714295387268066, + "learning_rate": 7.694421544380324e-07, + "loss": 0.9916, + "mean_token_accuracy": 0.7207728028297424, + "num_tokens": 84548547.0, + "step": 2801 + }, + { + "epoch": 0.23105467139440916, + "grad_norm": 1.7925143241882324, + "learning_rate": 7.697169552074746e-07, + "loss": 0.7708, + "mean_token_accuracy": 0.752578854560852, + "num_tokens": 84583192.0, + "step": 2802 + }, + { + "epoch": 0.2311371320194607, + "grad_norm": 2.124598979949951, + "learning_rate": 7.699917559769167e-07, + "loss": 0.8382, + "mean_token_accuracy": 0.7367593050003052, + "num_tokens": 84612827.0, + "step": 2803 + }, + { + "epoch": 0.23121959264451225, + "grad_norm": 1.977428674697876, + "learning_rate": 7.702665567463589e-07, + "loss": 0.7921, + "mean_token_accuracy": 0.7486159205436707, + "num_tokens": 84644024.0, + "step": 2804 + }, + { + "epoch": 0.2313020532695638, + "grad_norm": 1.8844574689865112, + "learning_rate": 7.70541357515801e-07, + "loss": 1.0225, + "mean_token_accuracy": 0.6945847272872925, + "num_tokens": 84682982.0, + "step": 2805 + }, + { + "epoch": 0.2313845138946153, + "grad_norm": 1.9836063385009766, + "learning_rate": 7.708161582852432e-07, + "loss": 0.848, + "mean_token_accuracy": 0.7368612885475159, + "num_tokens": 84715241.0, + "step": 2806 + }, + { + "epoch": 0.23146697451966686, + "grad_norm": 2.0298452377319336, + "learning_rate": 7.710909590546853e-07, + "loss": 0.9127, + "mean_token_accuracy": 0.7338269948959351, + "num_tokens": 84744294.0, + "step": 2807 + }, + { + "epoch": 0.2315494351447184, + "grad_norm": 1.9439297914505005, + "learning_rate": 7.713657598241274e-07, + "loss": 0.7664, + "mean_token_accuracy": 0.7491698265075684, + "num_tokens": 84775091.0, + "step": 2808 + }, + { + "epoch": 0.23163189576976995, + "grad_norm": 1.9916986227035522, + "learning_rate": 7.716405605935697e-07, + "loss": 0.6816, + "mean_token_accuracy": 0.7797372341156006, + "num_tokens": 84802622.0, + "step": 2809 + }, + { + "epoch": 0.23171435639482146, + "grad_norm": 2.0878350734710693, + "learning_rate": 7.719153613630119e-07, + "loss": 0.8989, + "mean_token_accuracy": 0.7287666201591492, + "num_tokens": 84830632.0, + "step": 2810 + }, + { + "epoch": 0.231796817019873, + "grad_norm": 1.960439920425415, + "learning_rate": 7.721901621324539e-07, + "loss": 0.9318, + "mean_token_accuracy": 0.714330792427063, + "num_tokens": 84861869.0, + "step": 2811 + }, + { + "epoch": 0.23187927764492455, + "grad_norm": 1.9030792713165283, + "learning_rate": 7.724649629018961e-07, + "loss": 0.7702, + "mean_token_accuracy": 0.7587711215019226, + "num_tokens": 84896049.0, + "step": 2812 + }, + { + "epoch": 0.2319617382699761, + "grad_norm": 2.403294324874878, + "learning_rate": 7.727397636713382e-07, + "loss": 0.8458, + "mean_token_accuracy": 0.7329800128936768, + "num_tokens": 84918552.0, + "step": 2813 + }, + { + "epoch": 0.23204419889502761, + "grad_norm": 1.966475486755371, + "learning_rate": 7.730145644407804e-07, + "loss": 0.8709, + "mean_token_accuracy": 0.7311322689056396, + "num_tokens": 84951897.0, + "step": 2814 + }, + { + "epoch": 0.23212665952007916, + "grad_norm": 1.8583760261535645, + "learning_rate": 7.732893652102226e-07, + "loss": 0.663, + "mean_token_accuracy": 0.7834858894348145, + "num_tokens": 84983277.0, + "step": 2815 + }, + { + "epoch": 0.2322091201451307, + "grad_norm": 1.8818261623382568, + "learning_rate": 7.735641659796647e-07, + "loss": 0.8981, + "mean_token_accuracy": 0.7194645404815674, + "num_tokens": 85021027.0, + "step": 2816 + }, + { + "epoch": 0.23229158077018225, + "grad_norm": 1.9651070833206177, + "learning_rate": 7.738389667491068e-07, + "loss": 0.9337, + "mean_token_accuracy": 0.720249593257904, + "num_tokens": 85053755.0, + "step": 2817 + }, + { + "epoch": 0.23237404139523377, + "grad_norm": 2.0669267177581787, + "learning_rate": 7.741137675185491e-07, + "loss": 0.8308, + "mean_token_accuracy": 0.7449716329574585, + "num_tokens": 85081688.0, + "step": 2818 + }, + { + "epoch": 0.2324565020202853, + "grad_norm": 1.7584863901138306, + "learning_rate": 7.743885682879912e-07, + "loss": 0.7238, + "mean_token_accuracy": 0.7598780393600464, + "num_tokens": 85116690.0, + "step": 2819 + }, + { + "epoch": 0.23253896264533686, + "grad_norm": 2.0254409313201904, + "learning_rate": 7.746633690574334e-07, + "loss": 0.9815, + "mean_token_accuracy": 0.7113269567489624, + "num_tokens": 85146874.0, + "step": 2820 + }, + { + "epoch": 0.2326214232703884, + "grad_norm": 2.390366315841675, + "learning_rate": 7.749381698268754e-07, + "loss": 1.0173, + "mean_token_accuracy": 0.6965104341506958, + "num_tokens": 85168603.0, + "step": 2821 + }, + { + "epoch": 0.23270388389543992, + "grad_norm": 2.0028462409973145, + "learning_rate": 7.752129705963176e-07, + "loss": 0.8242, + "mean_token_accuracy": 0.7498526573181152, + "num_tokens": 85199282.0, + "step": 2822 + }, + { + "epoch": 0.23278634452049146, + "grad_norm": 2.123701572418213, + "learning_rate": 7.754877713657599e-07, + "loss": 0.7881, + "mean_token_accuracy": 0.7546727061271667, + "num_tokens": 85227208.0, + "step": 2823 + }, + { + "epoch": 0.232868805145543, + "grad_norm": 2.011383295059204, + "learning_rate": 7.757625721352019e-07, + "loss": 0.8103, + "mean_token_accuracy": 0.7424169778823853, + "num_tokens": 85259396.0, + "step": 2824 + }, + { + "epoch": 0.23295126577059455, + "grad_norm": 1.9927479028701782, + "learning_rate": 7.760373729046441e-07, + "loss": 0.9186, + "mean_token_accuracy": 0.7196203470230103, + "num_tokens": 85292036.0, + "step": 2825 + }, + { + "epoch": 0.23303372639564607, + "grad_norm": 1.8787287473678589, + "learning_rate": 7.763121736740863e-07, + "loss": 0.9362, + "mean_token_accuracy": 0.7123875021934509, + "num_tokens": 85325344.0, + "step": 2826 + }, + { + "epoch": 0.2331161870206976, + "grad_norm": 2.083540439605713, + "learning_rate": 7.765869744435283e-07, + "loss": 0.8396, + "mean_token_accuracy": 0.7437264919281006, + "num_tokens": 85352333.0, + "step": 2827 + }, + { + "epoch": 0.23319864764574916, + "grad_norm": 2.0327584743499756, + "learning_rate": 7.768617752129706e-07, + "loss": 0.9186, + "mean_token_accuracy": 0.7213307023048401, + "num_tokens": 85385601.0, + "step": 2828 + }, + { + "epoch": 0.2332811082708007, + "grad_norm": 2.076646327972412, + "learning_rate": 7.771365759824127e-07, + "loss": 0.8527, + "mean_token_accuracy": 0.7284291386604309, + "num_tokens": 85415849.0, + "step": 2829 + }, + { + "epoch": 0.23336356889585222, + "grad_norm": 2.095735549926758, + "learning_rate": 7.774113767518549e-07, + "loss": 0.9415, + "mean_token_accuracy": 0.7148407697677612, + "num_tokens": 85444657.0, + "step": 2830 + }, + { + "epoch": 0.23344602952090376, + "grad_norm": 2.0683345794677734, + "learning_rate": 7.77686177521297e-07, + "loss": 0.844, + "mean_token_accuracy": 0.7412149310112, + "num_tokens": 85475370.0, + "step": 2831 + }, + { + "epoch": 0.2335284901459553, + "grad_norm": 1.8362222909927368, + "learning_rate": 7.779609782907391e-07, + "loss": 0.7671, + "mean_token_accuracy": 0.760413646697998, + "num_tokens": 85507842.0, + "step": 2832 + }, + { + "epoch": 0.23361095077100685, + "grad_norm": 2.2240259647369385, + "learning_rate": 7.782357790601814e-07, + "loss": 0.9637, + "mean_token_accuracy": 0.7049970626831055, + "num_tokens": 85533401.0, + "step": 2833 + }, + { + "epoch": 0.23369341139605837, + "grad_norm": 2.1414737701416016, + "learning_rate": 7.785105798296235e-07, + "loss": 0.8392, + "mean_token_accuracy": 0.7419110536575317, + "num_tokens": 85559957.0, + "step": 2834 + }, + { + "epoch": 0.23377587202110992, + "grad_norm": 2.070678472518921, + "learning_rate": 7.787853805990656e-07, + "loss": 0.7861, + "mean_token_accuracy": 0.7454505562782288, + "num_tokens": 85589498.0, + "step": 2835 + }, + { + "epoch": 0.23385833264616146, + "grad_norm": 1.7908408641815186, + "learning_rate": 7.790601813685078e-07, + "loss": 0.753, + "mean_token_accuracy": 0.7586437463760376, + "num_tokens": 85627605.0, + "step": 2836 + }, + { + "epoch": 0.233940793271213, + "grad_norm": 1.8179974555969238, + "learning_rate": 7.7933498213795e-07, + "loss": 0.6852, + "mean_token_accuracy": 0.7790992856025696, + "num_tokens": 85662478.0, + "step": 2837 + }, + { + "epoch": 0.23402325389626452, + "grad_norm": 2.2022101879119873, + "learning_rate": 7.796097829073921e-07, + "loss": 0.8228, + "mean_token_accuracy": 0.7481242418289185, + "num_tokens": 85689924.0, + "step": 2838 + }, + { + "epoch": 0.23410571452131607, + "grad_norm": 2.6659998893737793, + "learning_rate": 7.798845836768343e-07, + "loss": 0.8168, + "mean_token_accuracy": 0.739068329334259, + "num_tokens": 85709579.0, + "step": 2839 + }, + { + "epoch": 0.2341881751463676, + "grad_norm": 2.0610709190368652, + "learning_rate": 7.801593844462764e-07, + "loss": 0.8909, + "mean_token_accuracy": 0.7237709760665894, + "num_tokens": 85740358.0, + "step": 2840 + }, + { + "epoch": 0.23427063577141916, + "grad_norm": 2.12074875831604, + "learning_rate": 7.804341852157185e-07, + "loss": 0.8519, + "mean_token_accuracy": 0.7326991558074951, + "num_tokens": 85767298.0, + "step": 2841 + }, + { + "epoch": 0.23435309639647067, + "grad_norm": 2.355600595474243, + "learning_rate": 7.807089859851608e-07, + "loss": 0.9284, + "mean_token_accuracy": 0.724760115146637, + "num_tokens": 85790989.0, + "step": 2842 + }, + { + "epoch": 0.23443555702152222, + "grad_norm": 2.0073392391204834, + "learning_rate": 7.809837867546029e-07, + "loss": 0.8585, + "mean_token_accuracy": 0.7361856698989868, + "num_tokens": 85820219.0, + "step": 2843 + }, + { + "epoch": 0.23451801764657376, + "grad_norm": 2.1661088466644287, + "learning_rate": 7.81258587524045e-07, + "loss": 0.8323, + "mean_token_accuracy": 0.7336546182632446, + "num_tokens": 85844626.0, + "step": 2844 + }, + { + "epoch": 0.2346004782716253, + "grad_norm": 2.0297234058380127, + "learning_rate": 7.815333882934871e-07, + "loss": 0.9151, + "mean_token_accuracy": 0.7214933037757874, + "num_tokens": 85875375.0, + "step": 2845 + }, + { + "epoch": 0.23468293889667682, + "grad_norm": 2.014331817626953, + "learning_rate": 7.818081890629293e-07, + "loss": 0.7699, + "mean_token_accuracy": 0.7575321793556213, + "num_tokens": 85903328.0, + "step": 2846 + }, + { + "epoch": 0.23476539952172837, + "grad_norm": 2.3139595985412598, + "learning_rate": 7.820829898323716e-07, + "loss": 0.8704, + "mean_token_accuracy": 0.7249996662139893, + "num_tokens": 85926854.0, + "step": 2847 + }, + { + "epoch": 0.2348478601467799, + "grad_norm": 2.331535577774048, + "learning_rate": 7.823577906018136e-07, + "loss": 0.9011, + "mean_token_accuracy": 0.7177160978317261, + "num_tokens": 85952279.0, + "step": 2848 + }, + { + "epoch": 0.23493032077183146, + "grad_norm": 2.283867359161377, + "learning_rate": 7.826325913712558e-07, + "loss": 1.084, + "mean_token_accuracy": 0.6769775152206421, + "num_tokens": 85980663.0, + "step": 2849 + }, + { + "epoch": 0.23501278139688297, + "grad_norm": 2.281782627105713, + "learning_rate": 7.82907392140698e-07, + "loss": 0.8548, + "mean_token_accuracy": 0.7290191650390625, + "num_tokens": 86006690.0, + "step": 2850 + }, + { + "epoch": 0.23509524202193452, + "grad_norm": 1.8577609062194824, + "learning_rate": 7.831821929101401e-07, + "loss": 0.7135, + "mean_token_accuracy": 0.7731411457061768, + "num_tokens": 86037902.0, + "step": 2851 + }, + { + "epoch": 0.23517770264698606, + "grad_norm": 2.077704668045044, + "learning_rate": 7.834569936795823e-07, + "loss": 0.9499, + "mean_token_accuracy": 0.7105525732040405, + "num_tokens": 86067446.0, + "step": 2852 + }, + { + "epoch": 0.2352601632720376, + "grad_norm": 2.0144736766815186, + "learning_rate": 7.837317944490244e-07, + "loss": 0.7783, + "mean_token_accuracy": 0.7582852244377136, + "num_tokens": 86099390.0, + "step": 2853 + }, + { + "epoch": 0.23534262389708915, + "grad_norm": 2.175619125366211, + "learning_rate": 7.840065952184666e-07, + "loss": 0.8129, + "mean_token_accuracy": 0.7513447403907776, + "num_tokens": 86127958.0, + "step": 2854 + }, + { + "epoch": 0.23542508452214067, + "grad_norm": 2.1720011234283447, + "learning_rate": 7.842813959879087e-07, + "loss": 0.7615, + "mean_token_accuracy": 0.7652835249900818, + "num_tokens": 86153476.0, + "step": 2855 + }, + { + "epoch": 0.23550754514719222, + "grad_norm": 1.9593769311904907, + "learning_rate": 7.845561967573509e-07, + "loss": 0.7825, + "mean_token_accuracy": 0.759804368019104, + "num_tokens": 86185553.0, + "step": 2856 + }, + { + "epoch": 0.23559000577224376, + "grad_norm": 2.2250494956970215, + "learning_rate": 7.848309975267931e-07, + "loss": 0.8841, + "mean_token_accuracy": 0.7235831022262573, + "num_tokens": 86210754.0, + "step": 2857 + }, + { + "epoch": 0.2356724663972953, + "grad_norm": 2.1810545921325684, + "learning_rate": 7.851057982962352e-07, + "loss": 0.9216, + "mean_token_accuracy": 0.7350524663925171, + "num_tokens": 86238109.0, + "step": 2858 + }, + { + "epoch": 0.23575492702234682, + "grad_norm": 1.8164085149765015, + "learning_rate": 7.853805990656773e-07, + "loss": 0.9197, + "mean_token_accuracy": 0.7236336469650269, + "num_tokens": 86275166.0, + "step": 2859 + }, + { + "epoch": 0.23583738764739837, + "grad_norm": 1.951925277709961, + "learning_rate": 7.856553998351195e-07, + "loss": 0.7608, + "mean_token_accuracy": 0.7623002529144287, + "num_tokens": 86305009.0, + "step": 2860 + }, + { + "epoch": 0.2359198482724499, + "grad_norm": 1.9908180236816406, + "learning_rate": 7.859302006045616e-07, + "loss": 0.7725, + "mean_token_accuracy": 0.7578967809677124, + "num_tokens": 86333972.0, + "step": 2861 + }, + { + "epoch": 0.23600230889750146, + "grad_norm": 1.8575539588928223, + "learning_rate": 7.862050013740038e-07, + "loss": 0.8643, + "mean_token_accuracy": 0.7447640299797058, + "num_tokens": 86368387.0, + "step": 2862 + }, + { + "epoch": 0.23608476952255297, + "grad_norm": 2.151063919067383, + "learning_rate": 7.86479802143446e-07, + "loss": 1.0232, + "mean_token_accuracy": 0.6906960010528564, + "num_tokens": 86397314.0, + "step": 2863 + }, + { + "epoch": 0.23616723014760452, + "grad_norm": 1.7360483407974243, + "learning_rate": 7.867546029128881e-07, + "loss": 0.6912, + "mean_token_accuracy": 0.772977352142334, + "num_tokens": 86432450.0, + "step": 2864 + }, + { + "epoch": 0.23624969077265606, + "grad_norm": 2.145202875137329, + "learning_rate": 7.870294036823303e-07, + "loss": 0.8464, + "mean_token_accuracy": 0.7374927401542664, + "num_tokens": 86459692.0, + "step": 2865 + }, + { + "epoch": 0.2363321513977076, + "grad_norm": 2.510244846343994, + "learning_rate": 7.873042044517725e-07, + "loss": 0.971, + "mean_token_accuracy": 0.7029048204421997, + "num_tokens": 86484912.0, + "step": 2866 + }, + { + "epoch": 0.23641461202275912, + "grad_norm": 1.8885077238082886, + "learning_rate": 7.875790052212146e-07, + "loss": 0.9222, + "mean_token_accuracy": 0.717637300491333, + "num_tokens": 86522681.0, + "step": 2867 + }, + { + "epoch": 0.23649707264781067, + "grad_norm": 2.4095630645751953, + "learning_rate": 7.878538059906567e-07, + "loss": 0.8375, + "mean_token_accuracy": 0.7384510636329651, + "num_tokens": 86543827.0, + "step": 2868 + }, + { + "epoch": 0.2365795332728622, + "grad_norm": 1.8086315393447876, + "learning_rate": 7.881286067600988e-07, + "loss": 0.7961, + "mean_token_accuracy": 0.7610061168670654, + "num_tokens": 86579954.0, + "step": 2869 + }, + { + "epoch": 0.23666199389791376, + "grad_norm": 2.203402280807495, + "learning_rate": 7.884034075295411e-07, + "loss": 0.8668, + "mean_token_accuracy": 0.7349355220794678, + "num_tokens": 86609182.0, + "step": 2870 + }, + { + "epoch": 0.23674445452296528, + "grad_norm": 1.9911863803863525, + "learning_rate": 7.886782082989833e-07, + "loss": 0.8125, + "mean_token_accuracy": 0.7392383813858032, + "num_tokens": 86637072.0, + "step": 2871 + }, + { + "epoch": 0.23682691514801682, + "grad_norm": 2.2217230796813965, + "learning_rate": 7.889530090684253e-07, + "loss": 0.8874, + "mean_token_accuracy": 0.722663402557373, + "num_tokens": 86663932.0, + "step": 2872 + }, + { + "epoch": 0.23690937577306836, + "grad_norm": 2.0323076248168945, + "learning_rate": 7.892278098378675e-07, + "loss": 0.9206, + "mean_token_accuracy": 0.7188233137130737, + "num_tokens": 86694127.0, + "step": 2873 + }, + { + "epoch": 0.2369918363981199, + "grad_norm": 1.9328358173370361, + "learning_rate": 7.895026106073097e-07, + "loss": 0.7108, + "mean_token_accuracy": 0.7811681032180786, + "num_tokens": 86724303.0, + "step": 2874 + }, + { + "epoch": 0.23707429702317143, + "grad_norm": 1.8346339464187622, + "learning_rate": 7.897774113767518e-07, + "loss": 0.9564, + "mean_token_accuracy": 0.7053040266036987, + "num_tokens": 86762185.0, + "step": 2875 + }, + { + "epoch": 0.23715675764822297, + "grad_norm": 2.002307891845703, + "learning_rate": 7.90052212146194e-07, + "loss": 1.0007, + "mean_token_accuracy": 0.6978682279586792, + "num_tokens": 86794742.0, + "step": 2876 + }, + { + "epoch": 0.23723921827327452, + "grad_norm": 1.9544907808303833, + "learning_rate": 7.903270129156361e-07, + "loss": 0.7684, + "mean_token_accuracy": 0.756192147731781, + "num_tokens": 86825146.0, + "step": 2877 + }, + { + "epoch": 0.23732167889832606, + "grad_norm": 1.6774243116378784, + "learning_rate": 7.906018136850782e-07, + "loss": 0.7171, + "mean_token_accuracy": 0.7761411666870117, + "num_tokens": 86863534.0, + "step": 2878 + }, + { + "epoch": 0.23740413952337758, + "grad_norm": 2.055065393447876, + "learning_rate": 7.908766144545205e-07, + "loss": 0.943, + "mean_token_accuracy": 0.7203270196914673, + "num_tokens": 86894389.0, + "step": 2879 + }, + { + "epoch": 0.23748660014842912, + "grad_norm": 1.9718248844146729, + "learning_rate": 7.911514152239626e-07, + "loss": 0.781, + "mean_token_accuracy": 0.758500337600708, + "num_tokens": 86928237.0, + "step": 2880 + }, + { + "epoch": 0.23756906077348067, + "grad_norm": 2.219834089279175, + "learning_rate": 7.914262159934048e-07, + "loss": 0.799, + "mean_token_accuracy": 0.7488945722579956, + "num_tokens": 86953572.0, + "step": 2881 + }, + { + "epoch": 0.2376515213985322, + "grad_norm": 2.005408763885498, + "learning_rate": 7.917010167628469e-07, + "loss": 0.8466, + "mean_token_accuracy": 0.7372280955314636, + "num_tokens": 86985220.0, + "step": 2882 + }, + { + "epoch": 0.23773398202358373, + "grad_norm": 2.1813712120056152, + "learning_rate": 7.91975817532289e-07, + "loss": 0.7664, + "mean_token_accuracy": 0.7523099780082703, + "num_tokens": 87010421.0, + "step": 2883 + }, + { + "epoch": 0.23781644264863527, + "grad_norm": 1.968675971031189, + "learning_rate": 7.922506183017313e-07, + "loss": 0.946, + "mean_token_accuracy": 0.716749906539917, + "num_tokens": 87040457.0, + "step": 2884 + }, + { + "epoch": 0.23789890327368682, + "grad_norm": 1.9127780199050903, + "learning_rate": 7.925254190711733e-07, + "loss": 0.8692, + "mean_token_accuracy": 0.7342392802238464, + "num_tokens": 87072927.0, + "step": 2885 + }, + { + "epoch": 0.23798136389873836, + "grad_norm": 1.8649117946624756, + "learning_rate": 7.928002198406155e-07, + "loss": 0.7744, + "mean_token_accuracy": 0.7669743299484253, + "num_tokens": 87108420.0, + "step": 2886 + }, + { + "epoch": 0.23806382452378988, + "grad_norm": 2.228864908218384, + "learning_rate": 7.930750206100577e-07, + "loss": 0.9712, + "mean_token_accuracy": 0.7114773988723755, + "num_tokens": 87134330.0, + "step": 2887 + }, + { + "epoch": 0.23814628514884142, + "grad_norm": 1.8774727582931519, + "learning_rate": 7.933498213794998e-07, + "loss": 0.8757, + "mean_token_accuracy": 0.7339721918106079, + "num_tokens": 87167714.0, + "step": 2888 + }, + { + "epoch": 0.23822874577389297, + "grad_norm": 1.9976813793182373, + "learning_rate": 7.93624622148942e-07, + "loss": 0.7107, + "mean_token_accuracy": 0.7833136916160583, + "num_tokens": 87197187.0, + "step": 2889 + }, + { + "epoch": 0.23831120639894451, + "grad_norm": 2.3500161170959473, + "learning_rate": 7.938994229183842e-07, + "loss": 0.9407, + "mean_token_accuracy": 0.7152541875839233, + "num_tokens": 87220787.0, + "step": 2890 + }, + { + "epoch": 0.23839366702399603, + "grad_norm": 1.8923481702804565, + "learning_rate": 7.941742236878263e-07, + "loss": 0.7015, + "mean_token_accuracy": 0.7864431738853455, + "num_tokens": 87254423.0, + "step": 2891 + }, + { + "epoch": 0.23847612764904758, + "grad_norm": 1.9980430603027344, + "learning_rate": 7.944490244572684e-07, + "loss": 0.7962, + "mean_token_accuracy": 0.7502663135528564, + "num_tokens": 87283471.0, + "step": 2892 + }, + { + "epoch": 0.23855858827409912, + "grad_norm": 1.8559716939926147, + "learning_rate": 7.947238252267105e-07, + "loss": 0.8342, + "mean_token_accuracy": 0.7395209074020386, + "num_tokens": 87317396.0, + "step": 2893 + }, + { + "epoch": 0.23864104889915067, + "grad_norm": 2.060746192932129, + "learning_rate": 7.949986259961528e-07, + "loss": 0.7435, + "mean_token_accuracy": 0.7718777060508728, + "num_tokens": 87345015.0, + "step": 2894 + }, + { + "epoch": 0.23872350952420218, + "grad_norm": 1.9737757444381714, + "learning_rate": 7.95273426765595e-07, + "loss": 0.8281, + "mean_token_accuracy": 0.7387884259223938, + "num_tokens": 87373772.0, + "step": 2895 + }, + { + "epoch": 0.23880597014925373, + "grad_norm": 1.9953020811080933, + "learning_rate": 7.95548227535037e-07, + "loss": 0.8165, + "mean_token_accuracy": 0.749285101890564, + "num_tokens": 87405404.0, + "step": 2896 + }, + { + "epoch": 0.23888843077430527, + "grad_norm": 1.7661889791488647, + "learning_rate": 7.958230283044792e-07, + "loss": 0.6556, + "mean_token_accuracy": 0.7933279871940613, + "num_tokens": 87440543.0, + "step": 2897 + }, + { + "epoch": 0.23897089139935682, + "grad_norm": 2.123077392578125, + "learning_rate": 7.960978290739215e-07, + "loss": 0.8647, + "mean_token_accuracy": 0.732560932636261, + "num_tokens": 87468813.0, + "step": 2898 + }, + { + "epoch": 0.23905335202440833, + "grad_norm": 1.8664766550064087, + "learning_rate": 7.963726298433635e-07, + "loss": 0.8519, + "mean_token_accuracy": 0.7358191609382629, + "num_tokens": 87505140.0, + "step": 2899 + }, + { + "epoch": 0.23913581264945988, + "grad_norm": 1.7849897146224976, + "learning_rate": 7.966474306128057e-07, + "loss": 0.8569, + "mean_token_accuracy": 0.7384194135665894, + "num_tokens": 87542526.0, + "step": 2900 + }, + { + "epoch": 0.23921827327451142, + "grad_norm": 2.044203996658325, + "learning_rate": 7.969222313822478e-07, + "loss": 0.7646, + "mean_token_accuracy": 0.7573578953742981, + "num_tokens": 87572245.0, + "step": 2901 + }, + { + "epoch": 0.23930073389956297, + "grad_norm": 1.940001130104065, + "learning_rate": 7.971970321516899e-07, + "loss": 0.8844, + "mean_token_accuracy": 0.7296363115310669, + "num_tokens": 87604688.0, + "step": 2902 + }, + { + "epoch": 0.23938319452461448, + "grad_norm": 2.0875611305236816, + "learning_rate": 7.974718329211322e-07, + "loss": 0.7467, + "mean_token_accuracy": 0.7633678317070007, + "num_tokens": 87634689.0, + "step": 2903 + }, + { + "epoch": 0.23946565514966603, + "grad_norm": 2.1851677894592285, + "learning_rate": 7.977466336905743e-07, + "loss": 0.9522, + "mean_token_accuracy": 0.7135165929794312, + "num_tokens": 87664323.0, + "step": 2904 + }, + { + "epoch": 0.23954811577471757, + "grad_norm": 1.9880092144012451, + "learning_rate": 7.980214344600165e-07, + "loss": 0.9807, + "mean_token_accuracy": 0.705793023109436, + "num_tokens": 87698053.0, + "step": 2905 + }, + { + "epoch": 0.23963057639976912, + "grad_norm": 2.0678200721740723, + "learning_rate": 7.982962352294586e-07, + "loss": 0.9032, + "mean_token_accuracy": 0.7180957794189453, + "num_tokens": 87729144.0, + "step": 2906 + }, + { + "epoch": 0.23971303702482064, + "grad_norm": 2.0460104942321777, + "learning_rate": 7.985710359989007e-07, + "loss": 0.9637, + "mean_token_accuracy": 0.7107120156288147, + "num_tokens": 87760402.0, + "step": 2907 + }, + { + "epoch": 0.23979549764987218, + "grad_norm": 1.9576635360717773, + "learning_rate": 7.98845836768343e-07, + "loss": 0.8661, + "mean_token_accuracy": 0.735933780670166, + "num_tokens": 87792086.0, + "step": 2908 + }, + { + "epoch": 0.23987795827492372, + "grad_norm": 2.083163261413574, + "learning_rate": 7.99120637537785e-07, + "loss": 0.8925, + "mean_token_accuracy": 0.7248178720474243, + "num_tokens": 87819620.0, + "step": 2909 + }, + { + "epoch": 0.23996041889997527, + "grad_norm": 2.1692123413085938, + "learning_rate": 7.993954383072272e-07, + "loss": 0.9632, + "mean_token_accuracy": 0.7064695954322815, + "num_tokens": 87849254.0, + "step": 2910 + }, + { + "epoch": 0.2400428795250268, + "grad_norm": 2.0327842235565186, + "learning_rate": 7.996702390766694e-07, + "loss": 0.7176, + "mean_token_accuracy": 0.7803647518157959, + "num_tokens": 87877330.0, + "step": 2911 + }, + { + "epoch": 0.24012534015007833, + "grad_norm": 1.94316828250885, + "learning_rate": 7.999450398461116e-07, + "loss": 0.8414, + "mean_token_accuracy": 0.7413402795791626, + "num_tokens": 87911448.0, + "step": 2912 + }, + { + "epoch": 0.24020780077512988, + "grad_norm": 1.7987457513809204, + "learning_rate": 8.002198406155537e-07, + "loss": 0.9396, + "mean_token_accuracy": 0.7139359712600708, + "num_tokens": 87950243.0, + "step": 2913 + }, + { + "epoch": 0.24029026140018142, + "grad_norm": 2.035637378692627, + "learning_rate": 8.004946413849959e-07, + "loss": 0.8828, + "mean_token_accuracy": 0.7211529016494751, + "num_tokens": 87982619.0, + "step": 2914 + }, + { + "epoch": 0.24037272202523294, + "grad_norm": 2.254378080368042, + "learning_rate": 8.00769442154438e-07, + "loss": 0.9872, + "mean_token_accuracy": 0.6949944496154785, + "num_tokens": 88008801.0, + "step": 2915 + }, + { + "epoch": 0.24045518265028448, + "grad_norm": 1.7944355010986328, + "learning_rate": 8.010442429238801e-07, + "loss": 0.8187, + "mean_token_accuracy": 0.7443000078201294, + "num_tokens": 88043310.0, + "step": 2916 + }, + { + "epoch": 0.24053764327533603, + "grad_norm": 2.05757474899292, + "learning_rate": 8.013190436933223e-07, + "loss": 0.7784, + "mean_token_accuracy": 0.7586668729782104, + "num_tokens": 88070872.0, + "step": 2917 + }, + { + "epoch": 0.24062010390038757, + "grad_norm": 2.1060917377471924, + "learning_rate": 8.015938444627645e-07, + "loss": 0.9705, + "mean_token_accuracy": 0.7159435749053955, + "num_tokens": 88100325.0, + "step": 2918 + }, + { + "epoch": 0.24070256452543912, + "grad_norm": 1.9521071910858154, + "learning_rate": 8.018686452322066e-07, + "loss": 0.7325, + "mean_token_accuracy": 0.7728582620620728, + "num_tokens": 88128089.0, + "step": 2919 + }, + { + "epoch": 0.24078502515049063, + "grad_norm": 1.9957947731018066, + "learning_rate": 8.021434460016487e-07, + "loss": 1.0412, + "mean_token_accuracy": 0.6875209808349609, + "num_tokens": 88158696.0, + "step": 2920 + }, + { + "epoch": 0.24086748577554218, + "grad_norm": 2.868675708770752, + "learning_rate": 8.024182467710909e-07, + "loss": 0.9706, + "mean_token_accuracy": 0.7070162296295166, + "num_tokens": 88194685.0, + "step": 2921 + }, + { + "epoch": 0.24094994640059372, + "grad_norm": 1.9917895793914795, + "learning_rate": 8.026930475405332e-07, + "loss": 0.8524, + "mean_token_accuracy": 0.734796404838562, + "num_tokens": 88224157.0, + "step": 2922 + }, + { + "epoch": 0.24103240702564527, + "grad_norm": 1.9386789798736572, + "learning_rate": 8.029678483099752e-07, + "loss": 0.7364, + "mean_token_accuracy": 0.7635430097579956, + "num_tokens": 88252128.0, + "step": 2923 + }, + { + "epoch": 0.24111486765069678, + "grad_norm": 1.8923815488815308, + "learning_rate": 8.032426490794174e-07, + "loss": 0.8654, + "mean_token_accuracy": 0.7302290797233582, + "num_tokens": 88286399.0, + "step": 2924 + }, + { + "epoch": 0.24119732827574833, + "grad_norm": 2.0652616024017334, + "learning_rate": 8.035174498488595e-07, + "loss": 0.8198, + "mean_token_accuracy": 0.7452307939529419, + "num_tokens": 88317973.0, + "step": 2925 + }, + { + "epoch": 0.24127978890079987, + "grad_norm": 2.0459039211273193, + "learning_rate": 8.037922506183017e-07, + "loss": 0.7828, + "mean_token_accuracy": 0.7510647773742676, + "num_tokens": 88346832.0, + "step": 2926 + }, + { + "epoch": 0.24136224952585142, + "grad_norm": 2.001014471054077, + "learning_rate": 8.040670513877439e-07, + "loss": 0.8798, + "mean_token_accuracy": 0.7440779209136963, + "num_tokens": 88377864.0, + "step": 2927 + }, + { + "epoch": 0.24144471015090294, + "grad_norm": 2.1376800537109375, + "learning_rate": 8.04341852157186e-07, + "loss": 0.8333, + "mean_token_accuracy": 0.7392692565917969, + "num_tokens": 88405180.0, + "step": 2928 + }, + { + "epoch": 0.24152717077595448, + "grad_norm": 1.8436442613601685, + "learning_rate": 8.046166529266281e-07, + "loss": 0.7636, + "mean_token_accuracy": 0.7731835842132568, + "num_tokens": 88438004.0, + "step": 2929 + }, + { + "epoch": 0.24160963140100603, + "grad_norm": 2.1963515281677246, + "learning_rate": 8.048914536960703e-07, + "loss": 1.0228, + "mean_token_accuracy": 0.6869796514511108, + "num_tokens": 88470145.0, + "step": 2930 + }, + { + "epoch": 0.24169209202605757, + "grad_norm": 2.321312665939331, + "learning_rate": 8.051662544655125e-07, + "loss": 0.9327, + "mean_token_accuracy": 0.721723198890686, + "num_tokens": 88493158.0, + "step": 2931 + }, + { + "epoch": 0.2417745526511091, + "grad_norm": 2.032564878463745, + "learning_rate": 8.054410552349547e-07, + "loss": 0.8285, + "mean_token_accuracy": 0.7293152809143066, + "num_tokens": 88522854.0, + "step": 2932 + }, + { + "epoch": 0.24185701327616063, + "grad_norm": 1.7799173593521118, + "learning_rate": 8.057158560043967e-07, + "loss": 0.7889, + "mean_token_accuracy": 0.7598764896392822, + "num_tokens": 88557892.0, + "step": 2933 + }, + { + "epoch": 0.24193947390121218, + "grad_norm": 1.958203673362732, + "learning_rate": 8.059906567738389e-07, + "loss": 0.8168, + "mean_token_accuracy": 0.7512269020080566, + "num_tokens": 88588452.0, + "step": 2934 + }, + { + "epoch": 0.24202193452626372, + "grad_norm": 2.2875137329101562, + "learning_rate": 8.062654575432811e-07, + "loss": 0.8627, + "mean_token_accuracy": 0.7324807047843933, + "num_tokens": 88612819.0, + "step": 2935 + }, + { + "epoch": 0.24210439515131524, + "grad_norm": 2.1076819896698, + "learning_rate": 8.065402583127232e-07, + "loss": 0.8348, + "mean_token_accuracy": 0.7393090724945068, + "num_tokens": 88643687.0, + "step": 2936 + }, + { + "epoch": 0.24218685577636678, + "grad_norm": 2.470517635345459, + "learning_rate": 8.068150590821654e-07, + "loss": 0.7478, + "mean_token_accuracy": 0.7561801671981812, + "num_tokens": 88667201.0, + "step": 2937 + }, + { + "epoch": 0.24226931640141833, + "grad_norm": 1.7993558645248413, + "learning_rate": 8.070898598516076e-07, + "loss": 0.7505, + "mean_token_accuracy": 0.7657331824302673, + "num_tokens": 88702051.0, + "step": 2938 + }, + { + "epoch": 0.24235177702646987, + "grad_norm": 2.2524027824401855, + "learning_rate": 8.073646606210497e-07, + "loss": 0.9424, + "mean_token_accuracy": 0.7112904787063599, + "num_tokens": 88728108.0, + "step": 2939 + }, + { + "epoch": 0.2424342376515214, + "grad_norm": 1.9660896062850952, + "learning_rate": 8.076394613904919e-07, + "loss": 0.8725, + "mean_token_accuracy": 0.7270766496658325, + "num_tokens": 88761858.0, + "step": 2940 + }, + { + "epoch": 0.24251669827657293, + "grad_norm": 2.0890395641326904, + "learning_rate": 8.07914262159934e-07, + "loss": 0.9535, + "mean_token_accuracy": 0.710138201713562, + "num_tokens": 88788551.0, + "step": 2941 + }, + { + "epoch": 0.24259915890162448, + "grad_norm": 2.340116500854492, + "learning_rate": 8.081890629293762e-07, + "loss": 0.9071, + "mean_token_accuracy": 0.7159212231636047, + "num_tokens": 88812300.0, + "step": 2942 + }, + { + "epoch": 0.24268161952667602, + "grad_norm": 2.2279882431030273, + "learning_rate": 8.084638636988183e-07, + "loss": 0.7268, + "mean_token_accuracy": 0.7674252986907959, + "num_tokens": 88836056.0, + "step": 2943 + }, + { + "epoch": 0.24276408015172754, + "grad_norm": 1.8222424983978271, + "learning_rate": 8.087386644682604e-07, + "loss": 0.7493, + "mean_token_accuracy": 0.7661515474319458, + "num_tokens": 88868322.0, + "step": 2944 + }, + { + "epoch": 0.24284654077677909, + "grad_norm": 2.1566271781921387, + "learning_rate": 8.090134652377027e-07, + "loss": 0.896, + "mean_token_accuracy": 0.7333436012268066, + "num_tokens": 88898290.0, + "step": 2945 + }, + { + "epoch": 0.24292900140183063, + "grad_norm": 2.003873586654663, + "learning_rate": 8.092882660071449e-07, + "loss": 0.8804, + "mean_token_accuracy": 0.7257477045059204, + "num_tokens": 88928239.0, + "step": 2946 + }, + { + "epoch": 0.24301146202688217, + "grad_norm": 2.0316758155822754, + "learning_rate": 8.095630667765869e-07, + "loss": 0.8553, + "mean_token_accuracy": 0.7368944883346558, + "num_tokens": 88958781.0, + "step": 2947 + }, + { + "epoch": 0.2430939226519337, + "grad_norm": 2.048036575317383, + "learning_rate": 8.098378675460291e-07, + "loss": 0.8588, + "mean_token_accuracy": 0.738042414188385, + "num_tokens": 88986566.0, + "step": 2948 + }, + { + "epoch": 0.24317638327698524, + "grad_norm": 1.974768042564392, + "learning_rate": 8.101126683154712e-07, + "loss": 0.8793, + "mean_token_accuracy": 0.7291345596313477, + "num_tokens": 89017488.0, + "step": 2949 + }, + { + "epoch": 0.24325884390203678, + "grad_norm": 2.0324020385742188, + "learning_rate": 8.103874690849134e-07, + "loss": 0.9269, + "mean_token_accuracy": 0.7143833637237549, + "num_tokens": 89046372.0, + "step": 2950 + }, + { + "epoch": 0.24334130452708833, + "grad_norm": 2.0707757472991943, + "learning_rate": 8.106622698543556e-07, + "loss": 0.7191, + "mean_token_accuracy": 0.7696067094802856, + "num_tokens": 89074362.0, + "step": 2951 + }, + { + "epoch": 0.24342376515213984, + "grad_norm": 2.152294397354126, + "learning_rate": 8.109370706237977e-07, + "loss": 0.8616, + "mean_token_accuracy": 0.7288621664047241, + "num_tokens": 89102781.0, + "step": 2952 + }, + { + "epoch": 0.2435062257771914, + "grad_norm": 2.011561393737793, + "learning_rate": 8.112118713932398e-07, + "loss": 0.7948, + "mean_token_accuracy": 0.7526154518127441, + "num_tokens": 89134085.0, + "step": 2953 + }, + { + "epoch": 0.24358868640224293, + "grad_norm": 2.0651631355285645, + "learning_rate": 8.114866721626821e-07, + "loss": 1.0, + "mean_token_accuracy": 0.7021774053573608, + "num_tokens": 89165080.0, + "step": 2954 + }, + { + "epoch": 0.24367114702729448, + "grad_norm": 1.7474888563156128, + "learning_rate": 8.117614729321242e-07, + "loss": 0.7583, + "mean_token_accuracy": 0.7654873132705688, + "num_tokens": 89205518.0, + "step": 2955 + }, + { + "epoch": 0.243753607652346, + "grad_norm": 1.86317777633667, + "learning_rate": 8.120362737015664e-07, + "loss": 0.7928, + "mean_token_accuracy": 0.7581067085266113, + "num_tokens": 89239015.0, + "step": 2956 + }, + { + "epoch": 0.24383606827739754, + "grad_norm": 2.090118408203125, + "learning_rate": 8.123110744710084e-07, + "loss": 0.8749, + "mean_token_accuracy": 0.7251877188682556, + "num_tokens": 89269414.0, + "step": 2957 + }, + { + "epoch": 0.24391852890244908, + "grad_norm": 2.0042178630828857, + "learning_rate": 8.125858752404506e-07, + "loss": 0.9506, + "mean_token_accuracy": 0.7113829851150513, + "num_tokens": 89300921.0, + "step": 2958 + }, + { + "epoch": 0.24400098952750063, + "grad_norm": 2.1421382427215576, + "learning_rate": 8.128606760098929e-07, + "loss": 0.9526, + "mean_token_accuracy": 0.7134632468223572, + "num_tokens": 89328972.0, + "step": 2959 + }, + { + "epoch": 0.24408345015255214, + "grad_norm": 1.8107616901397705, + "learning_rate": 8.131354767793349e-07, + "loss": 0.905, + "mean_token_accuracy": 0.7076144218444824, + "num_tokens": 89362446.0, + "step": 2960 + }, + { + "epoch": 0.2441659107776037, + "grad_norm": 2.0198020935058594, + "learning_rate": 8.134102775487771e-07, + "loss": 0.883, + "mean_token_accuracy": 0.7420706748962402, + "num_tokens": 89394190.0, + "step": 2961 + }, + { + "epoch": 0.24424837140265523, + "grad_norm": 2.1493616104125977, + "learning_rate": 8.136850783182193e-07, + "loss": 0.8552, + "mean_token_accuracy": 0.7345677614212036, + "num_tokens": 89420355.0, + "step": 2962 + }, + { + "epoch": 0.24433083202770678, + "grad_norm": 1.9583059549331665, + "learning_rate": 8.139598790876613e-07, + "loss": 0.6899, + "mean_token_accuracy": 0.7816130518913269, + "num_tokens": 89450020.0, + "step": 2963 + }, + { + "epoch": 0.2444132926527583, + "grad_norm": 1.8386695384979248, + "learning_rate": 8.142346798571036e-07, + "loss": 0.879, + "mean_token_accuracy": 0.7361371517181396, + "num_tokens": 89485751.0, + "step": 2964 + }, + { + "epoch": 0.24449575327780984, + "grad_norm": 2.2735538482666016, + "learning_rate": 8.145094806265457e-07, + "loss": 0.925, + "mean_token_accuracy": 0.7220317125320435, + "num_tokens": 89516266.0, + "step": 2965 + }, + { + "epoch": 0.24457821390286139, + "grad_norm": 2.013969659805298, + "learning_rate": 8.147842813959879e-07, + "loss": 0.9763, + "mean_token_accuracy": 0.7004073262214661, + "num_tokens": 89548743.0, + "step": 2966 + }, + { + "epoch": 0.24466067452791293, + "grad_norm": 2.043588161468506, + "learning_rate": 8.1505908216543e-07, + "loss": 0.8907, + "mean_token_accuracy": 0.7265464067459106, + "num_tokens": 89579974.0, + "step": 2967 + }, + { + "epoch": 0.24474313515296445, + "grad_norm": 2.3736178874969482, + "learning_rate": 8.153338829348722e-07, + "loss": 0.9032, + "mean_token_accuracy": 0.7328037619590759, + "num_tokens": 89602361.0, + "step": 2968 + }, + { + "epoch": 0.244825595778016, + "grad_norm": 2.1572275161743164, + "learning_rate": 8.156086837043144e-07, + "loss": 0.8216, + "mean_token_accuracy": 0.7485895156860352, + "num_tokens": 89627266.0, + "step": 2969 + }, + { + "epoch": 0.24490805640306754, + "grad_norm": 1.868584156036377, + "learning_rate": 8.158834844737565e-07, + "loss": 0.7505, + "mean_token_accuracy": 0.7562406659126282, + "num_tokens": 89657255.0, + "step": 2970 + }, + { + "epoch": 0.24499051702811908, + "grad_norm": 2.2761590480804443, + "learning_rate": 8.161582852431986e-07, + "loss": 0.8449, + "mean_token_accuracy": 0.7389310598373413, + "num_tokens": 89683516.0, + "step": 2971 + }, + { + "epoch": 0.2450729776531706, + "grad_norm": 2.2489285469055176, + "learning_rate": 8.164330860126408e-07, + "loss": 0.8537, + "mean_token_accuracy": 0.7373940944671631, + "num_tokens": 89708584.0, + "step": 2972 + }, + { + "epoch": 0.24515543827822214, + "grad_norm": 1.9846010208129883, + "learning_rate": 8.16707886782083e-07, + "loss": 0.805, + "mean_token_accuracy": 0.7479632496833801, + "num_tokens": 89742636.0, + "step": 2973 + }, + { + "epoch": 0.2452378989032737, + "grad_norm": 1.7491360902786255, + "learning_rate": 8.169826875515251e-07, + "loss": 0.6738, + "mean_token_accuracy": 0.7829608917236328, + "num_tokens": 89782558.0, + "step": 2974 + }, + { + "epoch": 0.24532035952832523, + "grad_norm": 1.9093660116195679, + "learning_rate": 8.172574883209673e-07, + "loss": 0.7565, + "mean_token_accuracy": 0.7653811573982239, + "num_tokens": 89812521.0, + "step": 2975 + }, + { + "epoch": 0.24540282015337675, + "grad_norm": 2.0878469944000244, + "learning_rate": 8.175322890904094e-07, + "loss": 0.9442, + "mean_token_accuracy": 0.704322099685669, + "num_tokens": 89841995.0, + "step": 2976 + }, + { + "epoch": 0.2454852807784283, + "grad_norm": 2.277451753616333, + "learning_rate": 8.178070898598515e-07, + "loss": 0.7497, + "mean_token_accuracy": 0.7583543658256531, + "num_tokens": 89864650.0, + "step": 2977 + }, + { + "epoch": 0.24556774140347984, + "grad_norm": 1.8394113779067993, + "learning_rate": 8.180818906292938e-07, + "loss": 0.8145, + "mean_token_accuracy": 0.7548269033432007, + "num_tokens": 89899846.0, + "step": 2978 + }, + { + "epoch": 0.24565020202853138, + "grad_norm": 2.1205391883850098, + "learning_rate": 8.183566913987359e-07, + "loss": 0.7301, + "mean_token_accuracy": 0.758974552154541, + "num_tokens": 89927241.0, + "step": 2979 + }, + { + "epoch": 0.2457326626535829, + "grad_norm": 1.831144094467163, + "learning_rate": 8.18631492168178e-07, + "loss": 0.7315, + "mean_token_accuracy": 0.7606890797615051, + "num_tokens": 89963320.0, + "step": 2980 + }, + { + "epoch": 0.24581512327863445, + "grad_norm": 2.123741388320923, + "learning_rate": 8.189062929376201e-07, + "loss": 0.928, + "mean_token_accuracy": 0.7209801077842712, + "num_tokens": 89992092.0, + "step": 2981 + }, + { + "epoch": 0.245897583903686, + "grad_norm": 2.073413133621216, + "learning_rate": 8.191810937070624e-07, + "loss": 0.8096, + "mean_token_accuracy": 0.733586311340332, + "num_tokens": 90019773.0, + "step": 2982 + }, + { + "epoch": 0.24598004452873753, + "grad_norm": 2.136913776397705, + "learning_rate": 8.194558944765046e-07, + "loss": 0.7933, + "mean_token_accuracy": 0.7600803375244141, + "num_tokens": 90047171.0, + "step": 2983 + }, + { + "epoch": 0.24606250515378908, + "grad_norm": 2.003831386566162, + "learning_rate": 8.197306952459466e-07, + "loss": 0.8364, + "mean_token_accuracy": 0.732869029045105, + "num_tokens": 90078175.0, + "step": 2984 + }, + { + "epoch": 0.2461449657788406, + "grad_norm": 1.7845295667648315, + "learning_rate": 8.200054960153888e-07, + "loss": 0.9364, + "mean_token_accuracy": 0.715692400932312, + "num_tokens": 90120877.0, + "step": 2985 + }, + { + "epoch": 0.24622742640389214, + "grad_norm": 2.0061066150665283, + "learning_rate": 8.202802967848309e-07, + "loss": 0.9461, + "mean_token_accuracy": 0.7065956592559814, + "num_tokens": 90154778.0, + "step": 2986 + }, + { + "epoch": 0.24630988702894369, + "grad_norm": 2.207115888595581, + "learning_rate": 8.205550975542731e-07, + "loss": 0.7524, + "mean_token_accuracy": 0.7598283886909485, + "num_tokens": 90180857.0, + "step": 2987 + }, + { + "epoch": 0.24639234765399523, + "grad_norm": 2.136551856994629, + "learning_rate": 8.208298983237153e-07, + "loss": 0.7807, + "mean_token_accuracy": 0.7561821341514587, + "num_tokens": 90208583.0, + "step": 2988 + }, + { + "epoch": 0.24647480827904675, + "grad_norm": 2.190495252609253, + "learning_rate": 8.211046990931574e-07, + "loss": 0.7206, + "mean_token_accuracy": 0.7700384259223938, + "num_tokens": 90235144.0, + "step": 2989 + }, + { + "epoch": 0.2465572689040983, + "grad_norm": 2.3031957149505615, + "learning_rate": 8.213794998625996e-07, + "loss": 0.9228, + "mean_token_accuracy": 0.7210859060287476, + "num_tokens": 90260722.0, + "step": 2990 + }, + { + "epoch": 0.24663972952914984, + "grad_norm": 1.808118224143982, + "learning_rate": 8.216543006320417e-07, + "loss": 0.7727, + "mean_token_accuracy": 0.7583757638931274, + "num_tokens": 90293862.0, + "step": 2991 + }, + { + "epoch": 0.24672219015420138, + "grad_norm": 1.9749183654785156, + "learning_rate": 8.219291014014839e-07, + "loss": 0.8572, + "mean_token_accuracy": 0.7365995645523071, + "num_tokens": 90325275.0, + "step": 2992 + }, + { + "epoch": 0.2468046507792529, + "grad_norm": 2.3790252208709717, + "learning_rate": 8.222039021709261e-07, + "loss": 0.9817, + "mean_token_accuracy": 0.6995612382888794, + "num_tokens": 90349014.0, + "step": 2993 + }, + { + "epoch": 0.24688711140430444, + "grad_norm": 1.826681137084961, + "learning_rate": 8.224787029403681e-07, + "loss": 0.8581, + "mean_token_accuracy": 0.7421201467514038, + "num_tokens": 90384074.0, + "step": 2994 + }, + { + "epoch": 0.246969572029356, + "grad_norm": 2.0961203575134277, + "learning_rate": 8.227535037098103e-07, + "loss": 0.9071, + "mean_token_accuracy": 0.7365866899490356, + "num_tokens": 90412285.0, + "step": 2995 + }, + { + "epoch": 0.24705203265440753, + "grad_norm": 1.874550461769104, + "learning_rate": 8.230283044792526e-07, + "loss": 0.829, + "mean_token_accuracy": 0.7448141574859619, + "num_tokens": 90447404.0, + "step": 2996 + }, + { + "epoch": 0.24713449327945905, + "grad_norm": 2.3108959197998047, + "learning_rate": 8.233031052486947e-07, + "loss": 0.9032, + "mean_token_accuracy": 0.7243342399597168, + "num_tokens": 90472140.0, + "step": 2997 + }, + { + "epoch": 0.2472169539045106, + "grad_norm": 2.2289130687713623, + "learning_rate": 8.235779060181368e-07, + "loss": 0.8458, + "mean_token_accuracy": 0.745411217212677, + "num_tokens": 90497602.0, + "step": 2998 + }, + { + "epoch": 0.24729941452956214, + "grad_norm": 1.988171100616455, + "learning_rate": 8.23852706787579e-07, + "loss": 0.7878, + "mean_token_accuracy": 0.7567859292030334, + "num_tokens": 90530148.0, + "step": 2999 + }, + { + "epoch": 0.24738187515461368, + "grad_norm": 2.2071139812469482, + "learning_rate": 8.241275075570211e-07, + "loss": 0.903, + "mean_token_accuracy": 0.7206310033798218, + "num_tokens": 90559144.0, + "step": 3000 + }, + { + "epoch": 0.2474643357796652, + "grad_norm": 2.366135358810425, + "learning_rate": 8.244023083264633e-07, + "loss": 0.9205, + "mean_token_accuracy": 0.7192155718803406, + "num_tokens": 90582751.0, + "step": 3001 + }, + { + "epoch": 0.24754679640471675, + "grad_norm": 1.99573814868927, + "learning_rate": 8.246771090959054e-07, + "loss": 0.954, + "mean_token_accuracy": 0.7125682830810547, + "num_tokens": 90616041.0, + "step": 3002 + }, + { + "epoch": 0.2476292570297683, + "grad_norm": 2.22422456741333, + "learning_rate": 8.249519098653476e-07, + "loss": 0.8395, + "mean_token_accuracy": 0.7400773763656616, + "num_tokens": 90644070.0, + "step": 3003 + }, + { + "epoch": 0.24771171765481984, + "grad_norm": 1.95534086227417, + "learning_rate": 8.252267106347897e-07, + "loss": 0.7477, + "mean_token_accuracy": 0.7523884773254395, + "num_tokens": 90674623.0, + "step": 3004 + }, + { + "epoch": 0.24779417827987135, + "grad_norm": 2.094066619873047, + "learning_rate": 8.255015114042318e-07, + "loss": 0.9426, + "mean_token_accuracy": 0.7071391940116882, + "num_tokens": 90703837.0, + "step": 3005 + }, + { + "epoch": 0.2478766389049229, + "grad_norm": 1.7083760499954224, + "learning_rate": 8.257763121736741e-07, + "loss": 0.7531, + "mean_token_accuracy": 0.7623559832572937, + "num_tokens": 90739748.0, + "step": 3006 + }, + { + "epoch": 0.24795909952997444, + "grad_norm": 2.190866708755493, + "learning_rate": 8.260511129431163e-07, + "loss": 0.7908, + "mean_token_accuracy": 0.7542363405227661, + "num_tokens": 90763872.0, + "step": 3007 + }, + { + "epoch": 0.248041560155026, + "grad_norm": 2.055711269378662, + "learning_rate": 8.263259137125583e-07, + "loss": 0.8887, + "mean_token_accuracy": 0.7307429313659668, + "num_tokens": 90795502.0, + "step": 3008 + }, + { + "epoch": 0.2481240207800775, + "grad_norm": 2.066826105117798, + "learning_rate": 8.266007144820005e-07, + "loss": 0.8763, + "mean_token_accuracy": 0.7236257791519165, + "num_tokens": 90825730.0, + "step": 3009 + }, + { + "epoch": 0.24820648140512905, + "grad_norm": 1.7512991428375244, + "learning_rate": 8.268755152514426e-07, + "loss": 0.8587, + "mean_token_accuracy": 0.7295424938201904, + "num_tokens": 90862758.0, + "step": 3010 + }, + { + "epoch": 0.2482889420301806, + "grad_norm": 2.038250684738159, + "learning_rate": 8.271503160208848e-07, + "loss": 0.8258, + "mean_token_accuracy": 0.748851478099823, + "num_tokens": 90894790.0, + "step": 3011 + }, + { + "epoch": 0.24837140265523214, + "grad_norm": 1.9390127658843994, + "learning_rate": 8.27425116790327e-07, + "loss": 0.7636, + "mean_token_accuracy": 0.7684917449951172, + "num_tokens": 90929306.0, + "step": 3012 + }, + { + "epoch": 0.24845386328028365, + "grad_norm": 1.8454780578613281, + "learning_rate": 8.276999175597691e-07, + "loss": 0.9785, + "mean_token_accuracy": 0.7144582867622375, + "num_tokens": 90969194.0, + "step": 3013 + }, + { + "epoch": 0.2485363239053352, + "grad_norm": 1.808675765991211, + "learning_rate": 8.279747183292113e-07, + "loss": 0.8477, + "mean_token_accuracy": 0.7320754528045654, + "num_tokens": 91010890.0, + "step": 3014 + }, + { + "epoch": 0.24861878453038674, + "grad_norm": 1.7861812114715576, + "learning_rate": 8.282495190986535e-07, + "loss": 0.7735, + "mean_token_accuracy": 0.7560163736343384, + "num_tokens": 91048345.0, + "step": 3015 + }, + { + "epoch": 0.2487012451554383, + "grad_norm": 2.012373924255371, + "learning_rate": 8.285243198680956e-07, + "loss": 0.7765, + "mean_token_accuracy": 0.7723577618598938, + "num_tokens": 91075945.0, + "step": 3016 + }, + { + "epoch": 0.2487837057804898, + "grad_norm": 1.8420183658599854, + "learning_rate": 8.287991206375378e-07, + "loss": 0.6307, + "mean_token_accuracy": 0.7886710166931152, + "num_tokens": 91105410.0, + "step": 3017 + }, + { + "epoch": 0.24886616640554135, + "grad_norm": 1.9932013750076294, + "learning_rate": 8.290739214069798e-07, + "loss": 0.9245, + "mean_token_accuracy": 0.713261604309082, + "num_tokens": 91136693.0, + "step": 3018 + }, + { + "epoch": 0.2489486270305929, + "grad_norm": 2.3708972930908203, + "learning_rate": 8.29348722176422e-07, + "loss": 0.9024, + "mean_token_accuracy": 0.7230898141860962, + "num_tokens": 91160674.0, + "step": 3019 + }, + { + "epoch": 0.24903108765564444, + "grad_norm": 2.3085341453552246, + "learning_rate": 8.296235229458643e-07, + "loss": 0.912, + "mean_token_accuracy": 0.7228250503540039, + "num_tokens": 91185088.0, + "step": 3020 + }, + { + "epoch": 0.24911354828069596, + "grad_norm": 2.0770156383514404, + "learning_rate": 8.298983237153063e-07, + "loss": 0.7962, + "mean_token_accuracy": 0.7602465748786926, + "num_tokens": 91214113.0, + "step": 3021 + }, + { + "epoch": 0.2491960089057475, + "grad_norm": 2.175837993621826, + "learning_rate": 8.301731244847485e-07, + "loss": 0.9621, + "mean_token_accuracy": 0.7051395177841187, + "num_tokens": 91243646.0, + "step": 3022 + }, + { + "epoch": 0.24927846953079905, + "grad_norm": 2.0468297004699707, + "learning_rate": 8.304479252541907e-07, + "loss": 0.913, + "mean_token_accuracy": 0.7167987823486328, + "num_tokens": 91273779.0, + "step": 3023 + }, + { + "epoch": 0.2493609301558506, + "grad_norm": 2.105182409286499, + "learning_rate": 8.307227260236328e-07, + "loss": 0.6621, + "mean_token_accuracy": 0.7852244973182678, + "num_tokens": 91301143.0, + "step": 3024 + }, + { + "epoch": 0.2494433907809021, + "grad_norm": 1.8738547563552856, + "learning_rate": 8.30997526793075e-07, + "loss": 0.8248, + "mean_token_accuracy": 0.734054446220398, + "num_tokens": 91335557.0, + "step": 3025 + }, + { + "epoch": 0.24952585140595365, + "grad_norm": 1.961143136024475, + "learning_rate": 8.312723275625171e-07, + "loss": 0.8143, + "mean_token_accuracy": 0.7507439255714417, + "num_tokens": 91367657.0, + "step": 3026 + }, + { + "epoch": 0.2496083120310052, + "grad_norm": 2.2152111530303955, + "learning_rate": 8.315471283319593e-07, + "loss": 0.7213, + "mean_token_accuracy": 0.7704465389251709, + "num_tokens": 91392068.0, + "step": 3027 + }, + { + "epoch": 0.24969077265605674, + "grad_norm": 2.063410520553589, + "learning_rate": 8.318219291014014e-07, + "loss": 0.8493, + "mean_token_accuracy": 0.7378103733062744, + "num_tokens": 91421768.0, + "step": 3028 + }, + { + "epoch": 0.24977323328110826, + "grad_norm": 2.1131439208984375, + "learning_rate": 8.320967298708436e-07, + "loss": 0.8445, + "mean_token_accuracy": 0.7489503026008606, + "num_tokens": 91448423.0, + "step": 3029 + }, + { + "epoch": 0.2498556939061598, + "grad_norm": 1.884110689163208, + "learning_rate": 8.323715306402858e-07, + "loss": 0.7503, + "mean_token_accuracy": 0.76736980676651, + "num_tokens": 91482610.0, + "step": 3030 + }, + { + "epoch": 0.24993815453121135, + "grad_norm": 2.2274646759033203, + "learning_rate": 8.32646331409728e-07, + "loss": 1.0447, + "mean_token_accuracy": 0.6858671307563782, + "num_tokens": 91510974.0, + "step": 3031 + }, + { + "epoch": 0.2500206151562629, + "grad_norm": 1.5832812786102295, + "learning_rate": 8.3292113217917e-07, + "loss": 0.676, + "mean_token_accuracy": 0.7903723120689392, + "num_tokens": 91550191.0, + "step": 3032 + }, + { + "epoch": 0.2501030757813144, + "grad_norm": 2.320701837539673, + "learning_rate": 8.331959329486122e-07, + "loss": 0.8878, + "mean_token_accuracy": 0.7322517037391663, + "num_tokens": 91576773.0, + "step": 3033 + }, + { + "epoch": 0.250185536406366, + "grad_norm": 1.674655795097351, + "learning_rate": 8.334707337180544e-07, + "loss": 0.7709, + "mean_token_accuracy": 0.7601493000984192, + "num_tokens": 91616645.0, + "step": 3034 + }, + { + "epoch": 0.2502679970314175, + "grad_norm": 2.04620623588562, + "learning_rate": 8.337455344874965e-07, + "loss": 0.9092, + "mean_token_accuracy": 0.7187224626541138, + "num_tokens": 91647497.0, + "step": 3035 + }, + { + "epoch": 0.250350457656469, + "grad_norm": 1.9821335077285767, + "learning_rate": 8.340203352569387e-07, + "loss": 0.8652, + "mean_token_accuracy": 0.7318099737167358, + "num_tokens": 91683136.0, + "step": 3036 + }, + { + "epoch": 0.2504329182815206, + "grad_norm": 2.0288491249084473, + "learning_rate": 8.342951360263808e-07, + "loss": 0.8476, + "mean_token_accuracy": 0.7319749593734741, + "num_tokens": 91714216.0, + "step": 3037 + }, + { + "epoch": 0.2505153789065721, + "grad_norm": 2.3975751399993896, + "learning_rate": 8.345699367958229e-07, + "loss": 0.9342, + "mean_token_accuracy": 0.7115978002548218, + "num_tokens": 91740030.0, + "step": 3038 + }, + { + "epoch": 0.2505978395316236, + "grad_norm": 2.0795509815216064, + "learning_rate": 8.348447375652652e-07, + "loss": 0.9358, + "mean_token_accuracy": 0.7126806974411011, + "num_tokens": 91770645.0, + "step": 3039 + }, + { + "epoch": 0.2506803001566752, + "grad_norm": 2.43854022026062, + "learning_rate": 8.351195383347073e-07, + "loss": 0.8787, + "mean_token_accuracy": 0.7408115863800049, + "num_tokens": 91795649.0, + "step": 3040 + }, + { + "epoch": 0.2507627607817267, + "grad_norm": 2.220132827758789, + "learning_rate": 8.353943391041495e-07, + "loss": 0.9571, + "mean_token_accuracy": 0.7046298980712891, + "num_tokens": 91822659.0, + "step": 3041 + }, + { + "epoch": 0.2508452214067783, + "grad_norm": 2.1006717681884766, + "learning_rate": 8.356691398735915e-07, + "loss": 0.7845, + "mean_token_accuracy": 0.7538868188858032, + "num_tokens": 91850970.0, + "step": 3042 + }, + { + "epoch": 0.2509276820318298, + "grad_norm": 1.9179459810256958, + "learning_rate": 8.359439406430338e-07, + "loss": 0.8236, + "mean_token_accuracy": 0.7491475343704224, + "num_tokens": 91882276.0, + "step": 3043 + }, + { + "epoch": 0.2510101426568813, + "grad_norm": 1.9477980136871338, + "learning_rate": 8.36218741412476e-07, + "loss": 0.7712, + "mean_token_accuracy": 0.7616086006164551, + "num_tokens": 91914722.0, + "step": 3044 + }, + { + "epoch": 0.2510926032819329, + "grad_norm": 1.808280110359192, + "learning_rate": 8.36493542181918e-07, + "loss": 0.7481, + "mean_token_accuracy": 0.7627720236778259, + "num_tokens": 91952122.0, + "step": 3045 + }, + { + "epoch": 0.2511750639069844, + "grad_norm": 2.023289442062378, + "learning_rate": 8.367683429513602e-07, + "loss": 0.7193, + "mean_token_accuracy": 0.7686412334442139, + "num_tokens": 91982199.0, + "step": 3046 + }, + { + "epoch": 0.2512575245320359, + "grad_norm": 2.1203744411468506, + "learning_rate": 8.370431437208024e-07, + "loss": 0.7236, + "mean_token_accuracy": 0.7721759080886841, + "num_tokens": 92010549.0, + "step": 3047 + }, + { + "epoch": 0.2513399851570875, + "grad_norm": 1.9856619834899902, + "learning_rate": 8.373179444902446e-07, + "loss": 0.8466, + "mean_token_accuracy": 0.7366390824317932, + "num_tokens": 92039768.0, + "step": 3048 + }, + { + "epoch": 0.251422445782139, + "grad_norm": 1.8108749389648438, + "learning_rate": 8.375927452596867e-07, + "loss": 0.8686, + "mean_token_accuracy": 0.7352802753448486, + "num_tokens": 92073143.0, + "step": 3049 + }, + { + "epoch": 0.2515049064071906, + "grad_norm": 2.2210192680358887, + "learning_rate": 8.378675460291288e-07, + "loss": 0.7003, + "mean_token_accuracy": 0.7666305303573608, + "num_tokens": 92098520.0, + "step": 3050 + }, + { + "epoch": 0.2515873670322421, + "grad_norm": 1.9149284362792969, + "learning_rate": 8.38142346798571e-07, + "loss": 0.8724, + "mean_token_accuracy": 0.7265274524688721, + "num_tokens": 92131368.0, + "step": 3051 + }, + { + "epoch": 0.2516698276572936, + "grad_norm": 2.013077735900879, + "learning_rate": 8.384171475680131e-07, + "loss": 0.8595, + "mean_token_accuracy": 0.7416258454322815, + "num_tokens": 92159853.0, + "step": 3052 + }, + { + "epoch": 0.2517522882823452, + "grad_norm": 1.9315828084945679, + "learning_rate": 8.386919483374553e-07, + "loss": 0.8902, + "mean_token_accuracy": 0.7318017482757568, + "num_tokens": 92192954.0, + "step": 3053 + }, + { + "epoch": 0.2518347489073967, + "grad_norm": 2.2145328521728516, + "learning_rate": 8.389667491068975e-07, + "loss": 1.0735, + "mean_token_accuracy": 0.6817210912704468, + "num_tokens": 92223863.0, + "step": 3054 + }, + { + "epoch": 0.2519172095324483, + "grad_norm": 2.474628210067749, + "learning_rate": 8.392415498763396e-07, + "loss": 1.0338, + "mean_token_accuracy": 0.6882857084274292, + "num_tokens": 92244907.0, + "step": 3055 + }, + { + "epoch": 0.2519996701574998, + "grad_norm": 2.0933001041412354, + "learning_rate": 8.395163506457817e-07, + "loss": 0.7971, + "mean_token_accuracy": 0.7528295516967773, + "num_tokens": 92272261.0, + "step": 3056 + }, + { + "epoch": 0.2520821307825513, + "grad_norm": 2.0433037281036377, + "learning_rate": 8.39791151415224e-07, + "loss": 0.7303, + "mean_token_accuracy": 0.7614318132400513, + "num_tokens": 92297638.0, + "step": 3057 + }, + { + "epoch": 0.2521645914076029, + "grad_norm": 1.8987923860549927, + "learning_rate": 8.400659521846661e-07, + "loss": 0.8346, + "mean_token_accuracy": 0.7366838455200195, + "num_tokens": 92328745.0, + "step": 3058 + }, + { + "epoch": 0.2522470520326544, + "grad_norm": 1.9177418947219849, + "learning_rate": 8.403407529541082e-07, + "loss": 0.5991, + "mean_token_accuracy": 0.7995885014533997, + "num_tokens": 92356193.0, + "step": 3059 + }, + { + "epoch": 0.2523295126577059, + "grad_norm": 1.9610967636108398, + "learning_rate": 8.406155537235504e-07, + "loss": 0.8254, + "mean_token_accuracy": 0.7399632930755615, + "num_tokens": 92389868.0, + "step": 3060 + }, + { + "epoch": 0.2524119732827575, + "grad_norm": 1.8996096849441528, + "learning_rate": 8.408903544929925e-07, + "loss": 0.7767, + "mean_token_accuracy": 0.7641260623931885, + "num_tokens": 92421859.0, + "step": 3061 + }, + { + "epoch": 0.252494433907809, + "grad_norm": 2.0031657218933105, + "learning_rate": 8.411651552624347e-07, + "loss": 0.7999, + "mean_token_accuracy": 0.7418282628059387, + "num_tokens": 92452003.0, + "step": 3062 + }, + { + "epoch": 0.2525768945328606, + "grad_norm": 2.27546763420105, + "learning_rate": 8.414399560318769e-07, + "loss": 0.9114, + "mean_token_accuracy": 0.7146966457366943, + "num_tokens": 92478698.0, + "step": 3063 + }, + { + "epoch": 0.2526593551579121, + "grad_norm": 2.0043790340423584, + "learning_rate": 8.41714756801319e-07, + "loss": 0.8413, + "mean_token_accuracy": 0.744383692741394, + "num_tokens": 92505494.0, + "step": 3064 + }, + { + "epoch": 0.2527418157829636, + "grad_norm": 2.110668659210205, + "learning_rate": 8.419895575707612e-07, + "loss": 0.9183, + "mean_token_accuracy": 0.717710018157959, + "num_tokens": 92535638.0, + "step": 3065 + }, + { + "epoch": 0.2528242764080152, + "grad_norm": 2.053677797317505, + "learning_rate": 8.422643583402032e-07, + "loss": 0.7724, + "mean_token_accuracy": 0.7602529525756836, + "num_tokens": 92565214.0, + "step": 3066 + }, + { + "epoch": 0.2529067370330667, + "grad_norm": 1.8978722095489502, + "learning_rate": 8.425391591096455e-07, + "loss": 0.8285, + "mean_token_accuracy": 0.746672511100769, + "num_tokens": 92596558.0, + "step": 3067 + }, + { + "epoch": 0.2529891976581182, + "grad_norm": 2.05185866355896, + "learning_rate": 8.428139598790877e-07, + "loss": 0.7144, + "mean_token_accuracy": 0.7717475891113281, + "num_tokens": 92622519.0, + "step": 3068 + }, + { + "epoch": 0.2530716582831698, + "grad_norm": 1.730835199356079, + "learning_rate": 8.430887606485297e-07, + "loss": 0.7907, + "mean_token_accuracy": 0.740986168384552, + "num_tokens": 92659727.0, + "step": 3069 + }, + { + "epoch": 0.2531541189082213, + "grad_norm": 2.1833369731903076, + "learning_rate": 8.433635614179719e-07, + "loss": 0.9809, + "mean_token_accuracy": 0.6994920969009399, + "num_tokens": 92688266.0, + "step": 3070 + }, + { + "epoch": 0.2532365795332729, + "grad_norm": 1.72648024559021, + "learning_rate": 8.436383621874142e-07, + "loss": 0.8491, + "mean_token_accuracy": 0.7360501289367676, + "num_tokens": 92725385.0, + "step": 3071 + }, + { + "epoch": 0.2533190401583244, + "grad_norm": 2.2043206691741943, + "learning_rate": 8.439131629568562e-07, + "loss": 0.8564, + "mean_token_accuracy": 0.7397028207778931, + "num_tokens": 92751591.0, + "step": 3072 + }, + { + "epoch": 0.2534015007833759, + "grad_norm": 2.1310572624206543, + "learning_rate": 8.441879637262984e-07, + "loss": 1.0266, + "mean_token_accuracy": 0.6907410025596619, + "num_tokens": 92779787.0, + "step": 3073 + }, + { + "epoch": 0.2534839614084275, + "grad_norm": 1.8657054901123047, + "learning_rate": 8.444627644957405e-07, + "loss": 0.8795, + "mean_token_accuracy": 0.7322306632995605, + "num_tokens": 92813133.0, + "step": 3074 + }, + { + "epoch": 0.253566422033479, + "grad_norm": 2.2133948802948, + "learning_rate": 8.447375652651827e-07, + "loss": 0.8757, + "mean_token_accuracy": 0.7295893430709839, + "num_tokens": 92837283.0, + "step": 3075 + }, + { + "epoch": 0.25364888265853053, + "grad_norm": 1.9656383991241455, + "learning_rate": 8.450123660346249e-07, + "loss": 0.8321, + "mean_token_accuracy": 0.7447705268859863, + "num_tokens": 92868565.0, + "step": 3076 + }, + { + "epoch": 0.2537313432835821, + "grad_norm": 2.146174430847168, + "learning_rate": 8.45287166804067e-07, + "loss": 0.9149, + "mean_token_accuracy": 0.7194024920463562, + "num_tokens": 92895295.0, + "step": 3077 + }, + { + "epoch": 0.2538138039086336, + "grad_norm": 1.8637559413909912, + "learning_rate": 8.455619675735092e-07, + "loss": 0.8502, + "mean_token_accuracy": 0.7398658990859985, + "num_tokens": 92928832.0, + "step": 3078 + }, + { + "epoch": 0.2538962645336852, + "grad_norm": 2.145946502685547, + "learning_rate": 8.458367683429513e-07, + "loss": 0.8633, + "mean_token_accuracy": 0.7348580360412598, + "num_tokens": 92958199.0, + "step": 3079 + }, + { + "epoch": 0.2539787251587367, + "grad_norm": 1.9894176721572876, + "learning_rate": 8.461115691123934e-07, + "loss": 0.9033, + "mean_token_accuracy": 0.7179148197174072, + "num_tokens": 92991871.0, + "step": 3080 + }, + { + "epoch": 0.2540611857837882, + "grad_norm": 2.0617988109588623, + "learning_rate": 8.463863698818357e-07, + "loss": 0.807, + "mean_token_accuracy": 0.7391984462738037, + "num_tokens": 93021043.0, + "step": 3081 + }, + { + "epoch": 0.2541436464088398, + "grad_norm": 2.3763115406036377, + "learning_rate": 8.466611706512778e-07, + "loss": 0.9696, + "mean_token_accuracy": 0.7153676748275757, + "num_tokens": 93045726.0, + "step": 3082 + }, + { + "epoch": 0.2542261070338913, + "grad_norm": 2.1198525428771973, + "learning_rate": 8.469359714207199e-07, + "loss": 0.8333, + "mean_token_accuracy": 0.7375245094299316, + "num_tokens": 93073173.0, + "step": 3083 + }, + { + "epoch": 0.25430856765894283, + "grad_norm": 1.890522837638855, + "learning_rate": 8.472107721901621e-07, + "loss": 0.8685, + "mean_token_accuracy": 0.7276787161827087, + "num_tokens": 93104316.0, + "step": 3084 + }, + { + "epoch": 0.2543910282839944, + "grad_norm": 2.0811727046966553, + "learning_rate": 8.474855729596042e-07, + "loss": 0.905, + "mean_token_accuracy": 0.7247037887573242, + "num_tokens": 93134060.0, + "step": 3085 + }, + { + "epoch": 0.2544734889090459, + "grad_norm": 1.9072777032852173, + "learning_rate": 8.477603737290464e-07, + "loss": 0.9219, + "mean_token_accuracy": 0.7189697027206421, + "num_tokens": 93165788.0, + "step": 3086 + }, + { + "epoch": 0.2545559495340975, + "grad_norm": 1.8539695739746094, + "learning_rate": 8.480351744984886e-07, + "loss": 0.9494, + "mean_token_accuracy": 0.7135511040687561, + "num_tokens": 93199622.0, + "step": 3087 + }, + { + "epoch": 0.254638410159149, + "grad_norm": 1.8082363605499268, + "learning_rate": 8.483099752679307e-07, + "loss": 0.7853, + "mean_token_accuracy": 0.755982518196106, + "num_tokens": 93236075.0, + "step": 3088 + }, + { + "epoch": 0.2547208707842005, + "grad_norm": 1.786314845085144, + "learning_rate": 8.485847760373728e-07, + "loss": 0.6438, + "mean_token_accuracy": 0.7871204018592834, + "num_tokens": 93269442.0, + "step": 3089 + }, + { + "epoch": 0.2548033314092521, + "grad_norm": 2.209010124206543, + "learning_rate": 8.48859576806815e-07, + "loss": 0.9449, + "mean_token_accuracy": 0.7094565629959106, + "num_tokens": 93298439.0, + "step": 3090 + }, + { + "epoch": 0.2548857920343036, + "grad_norm": 2.4193778038024902, + "learning_rate": 8.491343775762572e-07, + "loss": 0.8889, + "mean_token_accuracy": 0.7222721576690674, + "num_tokens": 93320652.0, + "step": 3091 + }, + { + "epoch": 0.25496825265935513, + "grad_norm": 1.797629475593567, + "learning_rate": 8.494091783456994e-07, + "loss": 0.7746, + "mean_token_accuracy": 0.7585779428482056, + "num_tokens": 93355444.0, + "step": 3092 + }, + { + "epoch": 0.2550507132844067, + "grad_norm": 1.7926361560821533, + "learning_rate": 8.496839791151414e-07, + "loss": 0.7834, + "mean_token_accuracy": 0.7517762184143066, + "num_tokens": 93393181.0, + "step": 3093 + }, + { + "epoch": 0.2551331739094582, + "grad_norm": 1.875672698020935, + "learning_rate": 8.499587798845836e-07, + "loss": 0.8714, + "mean_token_accuracy": 0.7189695239067078, + "num_tokens": 93427419.0, + "step": 3094 + }, + { + "epoch": 0.2552156345345098, + "grad_norm": 1.7599111795425415, + "learning_rate": 8.502335806540259e-07, + "loss": 0.675, + "mean_token_accuracy": 0.7903441786766052, + "num_tokens": 93463004.0, + "step": 3095 + }, + { + "epoch": 0.2552980951595613, + "grad_norm": 2.0071067810058594, + "learning_rate": 8.505083814234679e-07, + "loss": 0.7727, + "mean_token_accuracy": 0.7589031457901001, + "num_tokens": 93492876.0, + "step": 3096 + }, + { + "epoch": 0.25538055578461283, + "grad_norm": 2.0021910667419434, + "learning_rate": 8.507831821929101e-07, + "loss": 0.9213, + "mean_token_accuracy": 0.7126666307449341, + "num_tokens": 93527296.0, + "step": 3097 + }, + { + "epoch": 0.2554630164096644, + "grad_norm": 2.1439573764801025, + "learning_rate": 8.510579829623522e-07, + "loss": 0.9974, + "mean_token_accuracy": 0.6971936225891113, + "num_tokens": 93555415.0, + "step": 3098 + }, + { + "epoch": 0.2555454770347159, + "grad_norm": 1.918052315711975, + "learning_rate": 8.513327837317944e-07, + "loss": 0.8647, + "mean_token_accuracy": 0.7357863783836365, + "num_tokens": 93590481.0, + "step": 3099 + }, + { + "epoch": 0.25562793765976743, + "grad_norm": 2.0781946182250977, + "learning_rate": 8.516075845012366e-07, + "loss": 0.8672, + "mean_token_accuracy": 0.7361669540405273, + "num_tokens": 93619023.0, + "step": 3100 + }, + { + "epoch": 0.255710398284819, + "grad_norm": 2.459477424621582, + "learning_rate": 8.518823852706787e-07, + "loss": 1.0189, + "mean_token_accuracy": 0.6928310394287109, + "num_tokens": 93642413.0, + "step": 3101 + }, + { + "epoch": 0.2557928589098705, + "grad_norm": 1.9313536882400513, + "learning_rate": 8.521571860401209e-07, + "loss": 0.7431, + "mean_token_accuracy": 0.7659434080123901, + "num_tokens": 93671321.0, + "step": 3102 + }, + { + "epoch": 0.2558753195349221, + "grad_norm": 2.012812376022339, + "learning_rate": 8.52431986809563e-07, + "loss": 0.7989, + "mean_token_accuracy": 0.7435516715049744, + "num_tokens": 93702343.0, + "step": 3103 + }, + { + "epoch": 0.2559577801599736, + "grad_norm": 2.106571912765503, + "learning_rate": 8.527067875790052e-07, + "loss": 0.7831, + "mean_token_accuracy": 0.7521770000457764, + "num_tokens": 93729042.0, + "step": 3104 + }, + { + "epoch": 0.25604024078502513, + "grad_norm": 1.9127641916275024, + "learning_rate": 8.529815883484474e-07, + "loss": 0.894, + "mean_token_accuracy": 0.7371978759765625, + "num_tokens": 93765022.0, + "step": 3105 + }, + { + "epoch": 0.2561227014100767, + "grad_norm": 2.3173420429229736, + "learning_rate": 8.532563891178894e-07, + "loss": 0.9017, + "mean_token_accuracy": 0.7137678861618042, + "num_tokens": 93790552.0, + "step": 3106 + }, + { + "epoch": 0.2562051620351282, + "grad_norm": 1.8616008758544922, + "learning_rate": 8.535311898873316e-07, + "loss": 0.8411, + "mean_token_accuracy": 0.7407013177871704, + "num_tokens": 93824735.0, + "step": 3107 + }, + { + "epoch": 0.25628762266017974, + "grad_norm": 1.814001202583313, + "learning_rate": 8.538059906567738e-07, + "loss": 0.89, + "mean_token_accuracy": 0.7296326160430908, + "num_tokens": 93861859.0, + "step": 3108 + }, + { + "epoch": 0.2563700832852313, + "grad_norm": 2.4100115299224854, + "learning_rate": 8.54080791426216e-07, + "loss": 1.0287, + "mean_token_accuracy": 0.691307783126831, + "num_tokens": 93883447.0, + "step": 3109 + }, + { + "epoch": 0.2564525439102828, + "grad_norm": 1.8575873374938965, + "learning_rate": 8.543555921956581e-07, + "loss": 0.8479, + "mean_token_accuracy": 0.7406107187271118, + "num_tokens": 93917236.0, + "step": 3110 + }, + { + "epoch": 0.2565350045353344, + "grad_norm": 2.248439073562622, + "learning_rate": 8.546303929651003e-07, + "loss": 0.918, + "mean_token_accuracy": 0.7193465232849121, + "num_tokens": 93944040.0, + "step": 3111 + }, + { + "epoch": 0.2566174651603859, + "grad_norm": 2.459545135498047, + "learning_rate": 8.549051937345424e-07, + "loss": 0.8447, + "mean_token_accuracy": 0.7405740022659302, + "num_tokens": 93966811.0, + "step": 3112 + }, + { + "epoch": 0.25669992578543743, + "grad_norm": 1.9269535541534424, + "learning_rate": 8.551799945039845e-07, + "loss": 0.8547, + "mean_token_accuracy": 0.7366177439689636, + "num_tokens": 93999703.0, + "step": 3113 + }, + { + "epoch": 0.256782386410489, + "grad_norm": 2.0864417552948, + "learning_rate": 8.554547952734267e-07, + "loss": 0.7967, + "mean_token_accuracy": 0.7525461316108704, + "num_tokens": 94030946.0, + "step": 3114 + }, + { + "epoch": 0.2568648470355405, + "grad_norm": 1.862740397453308, + "learning_rate": 8.557295960428689e-07, + "loss": 0.7865, + "mean_token_accuracy": 0.7558249235153198, + "num_tokens": 94062887.0, + "step": 3115 + }, + { + "epoch": 0.2569473076605921, + "grad_norm": 2.1154837608337402, + "learning_rate": 8.560043968123111e-07, + "loss": 0.8036, + "mean_token_accuracy": 0.748548150062561, + "num_tokens": 94089814.0, + "step": 3116 + }, + { + "epoch": 0.2570297682856436, + "grad_norm": 2.150820255279541, + "learning_rate": 8.562791975817531e-07, + "loss": 0.9285, + "mean_token_accuracy": 0.7139182686805725, + "num_tokens": 94118102.0, + "step": 3117 + }, + { + "epoch": 0.25711222891069513, + "grad_norm": 1.9042291641235352, + "learning_rate": 8.565539983511954e-07, + "loss": 0.8163, + "mean_token_accuracy": 0.7406887412071228, + "num_tokens": 94150824.0, + "step": 3118 + }, + { + "epoch": 0.2571946895357467, + "grad_norm": 2.0153090953826904, + "learning_rate": 8.568287991206376e-07, + "loss": 1.0719, + "mean_token_accuracy": 0.6880002617835999, + "num_tokens": 94183850.0, + "step": 3119 + }, + { + "epoch": 0.2572771501607982, + "grad_norm": 1.8449487686157227, + "learning_rate": 8.571035998900796e-07, + "loss": 0.6829, + "mean_token_accuracy": 0.7761650681495667, + "num_tokens": 94221318.0, + "step": 3120 + }, + { + "epoch": 0.25735961078584974, + "grad_norm": 1.876462459564209, + "learning_rate": 8.573784006595218e-07, + "loss": 0.9237, + "mean_token_accuracy": 0.7291480302810669, + "num_tokens": 94257662.0, + "step": 3121 + }, + { + "epoch": 0.2574420714109013, + "grad_norm": 2.0561184883117676, + "learning_rate": 8.576532014289639e-07, + "loss": 0.8681, + "mean_token_accuracy": 0.7451661229133606, + "num_tokens": 94286368.0, + "step": 3122 + }, + { + "epoch": 0.2575245320359528, + "grad_norm": 2.134568452835083, + "learning_rate": 8.579280021984062e-07, + "loss": 0.9203, + "mean_token_accuracy": 0.7222712635993958, + "num_tokens": 94314691.0, + "step": 3123 + }, + { + "epoch": 0.2576069926610044, + "grad_norm": 2.1188747882843018, + "learning_rate": 8.582028029678483e-07, + "loss": 0.645, + "mean_token_accuracy": 0.7828097343444824, + "num_tokens": 94338927.0, + "step": 3124 + }, + { + "epoch": 0.2576894532860559, + "grad_norm": 2.0810723304748535, + "learning_rate": 8.584776037372904e-07, + "loss": 0.9549, + "mean_token_accuracy": 0.7123960852622986, + "num_tokens": 94373466.0, + "step": 3125 + }, + { + "epoch": 0.25777191391110743, + "grad_norm": 1.9636297225952148, + "learning_rate": 8.587524045067326e-07, + "loss": 0.6873, + "mean_token_accuracy": 0.7604292631149292, + "num_tokens": 94402779.0, + "step": 3126 + }, + { + "epoch": 0.257854374536159, + "grad_norm": 1.8095510005950928, + "learning_rate": 8.590272052761747e-07, + "loss": 0.7688, + "mean_token_accuracy": 0.7543262243270874, + "num_tokens": 94436056.0, + "step": 3127 + }, + { + "epoch": 0.2579368351612105, + "grad_norm": 1.8411157131195068, + "learning_rate": 8.593020060456169e-07, + "loss": 0.6807, + "mean_token_accuracy": 0.7789371013641357, + "num_tokens": 94466884.0, + "step": 3128 + }, + { + "epoch": 0.25801929578626204, + "grad_norm": 1.8740954399108887, + "learning_rate": 8.595768068150591e-07, + "loss": 0.8429, + "mean_token_accuracy": 0.7358429431915283, + "num_tokens": 94501085.0, + "step": 3129 + }, + { + "epoch": 0.2581017564113136, + "grad_norm": 2.067730188369751, + "learning_rate": 8.598516075845011e-07, + "loss": 0.9562, + "mean_token_accuracy": 0.7156294584274292, + "num_tokens": 94531195.0, + "step": 3130 + }, + { + "epoch": 0.2581842170363651, + "grad_norm": 2.2524983882904053, + "learning_rate": 8.601264083539433e-07, + "loss": 0.932, + "mean_token_accuracy": 0.7135071754455566, + "num_tokens": 94558616.0, + "step": 3131 + }, + { + "epoch": 0.2582666776614167, + "grad_norm": 1.983063817024231, + "learning_rate": 8.604012091233856e-07, + "loss": 0.8052, + "mean_token_accuracy": 0.7540422081947327, + "num_tokens": 94589656.0, + "step": 3132 + }, + { + "epoch": 0.2583491382864682, + "grad_norm": 2.0193610191345215, + "learning_rate": 8.606760098928277e-07, + "loss": 0.7981, + "mean_token_accuracy": 0.7538420557975769, + "num_tokens": 94617738.0, + "step": 3133 + }, + { + "epoch": 0.25843159891151973, + "grad_norm": 1.8394525051116943, + "learning_rate": 8.609508106622698e-07, + "loss": 0.7244, + "mean_token_accuracy": 0.7655331492424011, + "num_tokens": 94650695.0, + "step": 3134 + }, + { + "epoch": 0.2585140595365713, + "grad_norm": 1.774170160293579, + "learning_rate": 8.61225611431712e-07, + "loss": 0.7131, + "mean_token_accuracy": 0.7729222774505615, + "num_tokens": 94683431.0, + "step": 3135 + }, + { + "epoch": 0.2585965201616228, + "grad_norm": 2.1006643772125244, + "learning_rate": 8.615004122011541e-07, + "loss": 0.7375, + "mean_token_accuracy": 0.7646832466125488, + "num_tokens": 94710848.0, + "step": 3136 + }, + { + "epoch": 0.25867898078667434, + "grad_norm": 2.151634454727173, + "learning_rate": 8.617752129705963e-07, + "loss": 0.9777, + "mean_token_accuracy": 0.7078948020935059, + "num_tokens": 94740366.0, + "step": 3137 + }, + { + "epoch": 0.2587614414117259, + "grad_norm": 1.9837359189987183, + "learning_rate": 8.620500137400384e-07, + "loss": 0.7622, + "mean_token_accuracy": 0.7646562457084656, + "num_tokens": 94771203.0, + "step": 3138 + }, + { + "epoch": 0.25884390203677743, + "grad_norm": 2.0300486087799072, + "learning_rate": 8.623248145094806e-07, + "loss": 0.7848, + "mean_token_accuracy": 0.7593695521354675, + "num_tokens": 94799797.0, + "step": 3139 + }, + { + "epoch": 0.258926362661829, + "grad_norm": 2.1205408573150635, + "learning_rate": 8.625996152789227e-07, + "loss": 0.8315, + "mean_token_accuracy": 0.7334384918212891, + "num_tokens": 94828603.0, + "step": 3140 + }, + { + "epoch": 0.2590088232868805, + "grad_norm": 1.8761473894119263, + "learning_rate": 8.628744160483648e-07, + "loss": 0.6834, + "mean_token_accuracy": 0.7767688035964966, + "num_tokens": 94862795.0, + "step": 3141 + }, + { + "epoch": 0.25909128391193204, + "grad_norm": 2.060708999633789, + "learning_rate": 8.631492168178071e-07, + "loss": 0.947, + "mean_token_accuracy": 0.7133203744888306, + "num_tokens": 94893684.0, + "step": 3142 + }, + { + "epoch": 0.2591737445369836, + "grad_norm": 2.0406248569488525, + "learning_rate": 8.634240175872493e-07, + "loss": 0.7322, + "mean_token_accuracy": 0.7634773254394531, + "num_tokens": 94919668.0, + "step": 3143 + }, + { + "epoch": 0.2592562051620351, + "grad_norm": 2.0616509914398193, + "learning_rate": 8.636988183566913e-07, + "loss": 0.8104, + "mean_token_accuracy": 0.7533336281776428, + "num_tokens": 94948206.0, + "step": 3144 + }, + { + "epoch": 0.25933866578708664, + "grad_norm": 2.0703399181365967, + "learning_rate": 8.639736191261335e-07, + "loss": 0.9667, + "mean_token_accuracy": 0.719044029712677, + "num_tokens": 94976780.0, + "step": 3145 + }, + { + "epoch": 0.2594211264121382, + "grad_norm": 2.0622057914733887, + "learning_rate": 8.642484198955756e-07, + "loss": 0.8241, + "mean_token_accuracy": 0.7477367520332336, + "num_tokens": 95004836.0, + "step": 3146 + }, + { + "epoch": 0.25950358703718973, + "grad_norm": 2.091670274734497, + "learning_rate": 8.645232206650178e-07, + "loss": 0.9069, + "mean_token_accuracy": 0.7188698649406433, + "num_tokens": 95035365.0, + "step": 3147 + }, + { + "epoch": 0.2595860476622413, + "grad_norm": 2.1611263751983643, + "learning_rate": 8.6479802143446e-07, + "loss": 0.8355, + "mean_token_accuracy": 0.7341094613075256, + "num_tokens": 95061243.0, + "step": 3148 + }, + { + "epoch": 0.2596685082872928, + "grad_norm": 2.010342597961426, + "learning_rate": 8.650728222039021e-07, + "loss": 0.8644, + "mean_token_accuracy": 0.7389361262321472, + "num_tokens": 95092353.0, + "step": 3149 + }, + { + "epoch": 0.25975096891234434, + "grad_norm": 2.3021440505981445, + "learning_rate": 8.653476229733443e-07, + "loss": 0.9658, + "mean_token_accuracy": 0.7002250552177429, + "num_tokens": 95117062.0, + "step": 3150 + }, + { + "epoch": 0.2598334295373959, + "grad_norm": 1.9237192869186401, + "learning_rate": 8.656224237427865e-07, + "loss": 0.8207, + "mean_token_accuracy": 0.7458497285842896, + "num_tokens": 95149262.0, + "step": 3151 + }, + { + "epoch": 0.2599158901624474, + "grad_norm": 1.8462311029434204, + "learning_rate": 8.658972245122286e-07, + "loss": 0.9262, + "mean_token_accuracy": 0.721125602722168, + "num_tokens": 95187148.0, + "step": 3152 + }, + { + "epoch": 0.25999835078749894, + "grad_norm": 1.9195948839187622, + "learning_rate": 8.661720252816708e-07, + "loss": 0.7094, + "mean_token_accuracy": 0.7750893235206604, + "num_tokens": 95219044.0, + "step": 3153 + }, + { + "epoch": 0.2600808114125505, + "grad_norm": 1.9845154285430908, + "learning_rate": 8.664468260511128e-07, + "loss": 1.0338, + "mean_token_accuracy": 0.6830523014068604, + "num_tokens": 95250918.0, + "step": 3154 + }, + { + "epoch": 0.26016327203760203, + "grad_norm": 1.628007173538208, + "learning_rate": 8.66721626820555e-07, + "loss": 0.7501, + "mean_token_accuracy": 0.7794972658157349, + "num_tokens": 95292657.0, + "step": 3155 + }, + { + "epoch": 0.2602457326626536, + "grad_norm": 2.1128454208374023, + "learning_rate": 8.669964275899973e-07, + "loss": 0.9748, + "mean_token_accuracy": 0.7061673402786255, + "num_tokens": 95322383.0, + "step": 3156 + }, + { + "epoch": 0.2603281932877051, + "grad_norm": 2.0181517601013184, + "learning_rate": 8.672712283594393e-07, + "loss": 0.7902, + "mean_token_accuracy": 0.7557519674301147, + "num_tokens": 95353333.0, + "step": 3157 + }, + { + "epoch": 0.26041065391275664, + "grad_norm": 1.954195499420166, + "learning_rate": 8.675460291288815e-07, + "loss": 0.9351, + "mean_token_accuracy": 0.7177387475967407, + "num_tokens": 95383351.0, + "step": 3158 + }, + { + "epoch": 0.2604931145378082, + "grad_norm": 2.3050923347473145, + "learning_rate": 8.678208298983237e-07, + "loss": 0.8666, + "mean_token_accuracy": 0.7367119789123535, + "num_tokens": 95409194.0, + "step": 3159 + }, + { + "epoch": 0.26057557516285973, + "grad_norm": 1.9873194694519043, + "learning_rate": 8.680956306677659e-07, + "loss": 0.8761, + "mean_token_accuracy": 0.7277313470840454, + "num_tokens": 95439495.0, + "step": 3160 + }, + { + "epoch": 0.26065803578791125, + "grad_norm": 2.0233616828918457, + "learning_rate": 8.68370431437208e-07, + "loss": 0.8743, + "mean_token_accuracy": 0.7335236668586731, + "num_tokens": 95468733.0, + "step": 3161 + }, + { + "epoch": 0.2607404964129628, + "grad_norm": 2.157245397567749, + "learning_rate": 8.686452322066501e-07, + "loss": 0.8971, + "mean_token_accuracy": 0.722415566444397, + "num_tokens": 95495324.0, + "step": 3162 + }, + { + "epoch": 0.26082295703801434, + "grad_norm": 2.0392932891845703, + "learning_rate": 8.689200329760923e-07, + "loss": 0.8666, + "mean_token_accuracy": 0.7313351631164551, + "num_tokens": 95526505.0, + "step": 3163 + }, + { + "epoch": 0.2609054176630659, + "grad_norm": 1.9193083047866821, + "learning_rate": 8.691948337455344e-07, + "loss": 0.9862, + "mean_token_accuracy": 0.6983226537704468, + "num_tokens": 95561932.0, + "step": 3164 + }, + { + "epoch": 0.2609878782881174, + "grad_norm": 1.894473671913147, + "learning_rate": 8.694696345149766e-07, + "loss": 0.7968, + "mean_token_accuracy": 0.7483813166618347, + "num_tokens": 95597473.0, + "step": 3165 + }, + { + "epoch": 0.26107033891316894, + "grad_norm": 2.413506507873535, + "learning_rate": 8.697444352844188e-07, + "loss": 0.7843, + "mean_token_accuracy": 0.7543096542358398, + "num_tokens": 95622668.0, + "step": 3166 + }, + { + "epoch": 0.2611527995382205, + "grad_norm": 1.9496800899505615, + "learning_rate": 8.70019236053861e-07, + "loss": 0.7698, + "mean_token_accuracy": 0.7633106112480164, + "num_tokens": 95653260.0, + "step": 3167 + }, + { + "epoch": 0.26123526016327203, + "grad_norm": 1.9899439811706543, + "learning_rate": 8.70294036823303e-07, + "loss": 0.7074, + "mean_token_accuracy": 0.7671296000480652, + "num_tokens": 95682936.0, + "step": 3168 + }, + { + "epoch": 0.26131772078832355, + "grad_norm": 2.1182429790496826, + "learning_rate": 8.705688375927452e-07, + "loss": 0.7052, + "mean_token_accuracy": 0.7668545246124268, + "num_tokens": 95709748.0, + "step": 3169 + }, + { + "epoch": 0.2614001814133751, + "grad_norm": 2.1117284297943115, + "learning_rate": 8.708436383621874e-07, + "loss": 0.7875, + "mean_token_accuracy": 0.7555755972862244, + "num_tokens": 95738683.0, + "step": 3170 + }, + { + "epoch": 0.26148264203842664, + "grad_norm": 2.0109434127807617, + "learning_rate": 8.711184391316295e-07, + "loss": 0.8551, + "mean_token_accuracy": 0.7267714738845825, + "num_tokens": 95773614.0, + "step": 3171 + }, + { + "epoch": 0.2615651026634782, + "grad_norm": 2.1241915225982666, + "learning_rate": 8.713932399010717e-07, + "loss": 0.8268, + "mean_token_accuracy": 0.7388721108436584, + "num_tokens": 95801747.0, + "step": 3172 + }, + { + "epoch": 0.2616475632885297, + "grad_norm": 1.8295646905899048, + "learning_rate": 8.716680406705138e-07, + "loss": 0.8304, + "mean_token_accuracy": 0.7467583417892456, + "num_tokens": 95840615.0, + "step": 3173 + }, + { + "epoch": 0.26173002391358124, + "grad_norm": 2.2794854640960693, + "learning_rate": 8.719428414399561e-07, + "loss": 0.9628, + "mean_token_accuracy": 0.708662211894989, + "num_tokens": 95867987.0, + "step": 3174 + }, + { + "epoch": 0.2618124845386328, + "grad_norm": 2.3390085697174072, + "learning_rate": 8.722176422093982e-07, + "loss": 0.6991, + "mean_token_accuracy": 0.7727514505386353, + "num_tokens": 95892822.0, + "step": 3175 + }, + { + "epoch": 0.26189494516368433, + "grad_norm": 2.017695426940918, + "learning_rate": 8.724924429788403e-07, + "loss": 0.8624, + "mean_token_accuracy": 0.7440187335014343, + "num_tokens": 95924362.0, + "step": 3176 + }, + { + "epoch": 0.26197740578873585, + "grad_norm": 2.170856237411499, + "learning_rate": 8.727672437482825e-07, + "loss": 0.7622, + "mean_token_accuracy": 0.7599581480026245, + "num_tokens": 95954147.0, + "step": 3177 + }, + { + "epoch": 0.2620598664137874, + "grad_norm": 2.248199701309204, + "learning_rate": 8.730420445177245e-07, + "loss": 0.9474, + "mean_token_accuracy": 0.711846113204956, + "num_tokens": 95981744.0, + "step": 3178 + }, + { + "epoch": 0.26214232703883894, + "grad_norm": 2.065809726715088, + "learning_rate": 8.733168452871668e-07, + "loss": 0.6746, + "mean_token_accuracy": 0.7794921398162842, + "num_tokens": 96010919.0, + "step": 3179 + }, + { + "epoch": 0.2622247876638905, + "grad_norm": 2.085622787475586, + "learning_rate": 8.73591646056609e-07, + "loss": 0.7855, + "mean_token_accuracy": 0.754935085773468, + "num_tokens": 96035834.0, + "step": 3180 + }, + { + "epoch": 0.26230724828894203, + "grad_norm": 2.113497257232666, + "learning_rate": 8.73866446826051e-07, + "loss": 0.817, + "mean_token_accuracy": 0.7392144203186035, + "num_tokens": 96065640.0, + "step": 3181 + }, + { + "epoch": 0.26238970891399355, + "grad_norm": 1.8999645709991455, + "learning_rate": 8.741412475954932e-07, + "loss": 0.7997, + "mean_token_accuracy": 0.7626833319664001, + "num_tokens": 96099279.0, + "step": 3182 + }, + { + "epoch": 0.2624721695390451, + "grad_norm": 1.9568872451782227, + "learning_rate": 8.744160483649354e-07, + "loss": 0.7724, + "mean_token_accuracy": 0.7589555978775024, + "num_tokens": 96128419.0, + "step": 3183 + }, + { + "epoch": 0.26255463016409664, + "grad_norm": 2.0010221004486084, + "learning_rate": 8.746908491343776e-07, + "loss": 0.8882, + "mean_token_accuracy": 0.7215757369995117, + "num_tokens": 96157715.0, + "step": 3184 + }, + { + "epoch": 0.2626370907891482, + "grad_norm": 2.176116466522217, + "learning_rate": 8.749656499038197e-07, + "loss": 0.8811, + "mean_token_accuracy": 0.7291514277458191, + "num_tokens": 96183418.0, + "step": 3185 + }, + { + "epoch": 0.2627195514141997, + "grad_norm": 2.0924861431121826, + "learning_rate": 8.752404506732618e-07, + "loss": 0.9031, + "mean_token_accuracy": 0.7190884947776794, + "num_tokens": 96214137.0, + "step": 3186 + }, + { + "epoch": 0.26280201203925124, + "grad_norm": 1.9172916412353516, + "learning_rate": 8.75515251442704e-07, + "loss": 0.7275, + "mean_token_accuracy": 0.7691915035247803, + "num_tokens": 96247533.0, + "step": 3187 + }, + { + "epoch": 0.2628844726643028, + "grad_norm": 1.9557331800460815, + "learning_rate": 8.757900522121462e-07, + "loss": 0.7161, + "mean_token_accuracy": 0.7803199291229248, + "num_tokens": 96277687.0, + "step": 3188 + }, + { + "epoch": 0.26296693328935433, + "grad_norm": 1.8210846185684204, + "learning_rate": 8.760648529815883e-07, + "loss": 0.7455, + "mean_token_accuracy": 0.7632904052734375, + "num_tokens": 96315556.0, + "step": 3189 + }, + { + "epoch": 0.26304939391440585, + "grad_norm": 2.0489728450775146, + "learning_rate": 8.763396537510305e-07, + "loss": 0.766, + "mean_token_accuracy": 0.7649458646774292, + "num_tokens": 96347974.0, + "step": 3190 + }, + { + "epoch": 0.2631318545394574, + "grad_norm": 1.8621560335159302, + "learning_rate": 8.766144545204727e-07, + "loss": 0.8271, + "mean_token_accuracy": 0.7426567077636719, + "num_tokens": 96384268.0, + "step": 3191 + }, + { + "epoch": 0.26321431516450894, + "grad_norm": 2.0059337615966797, + "learning_rate": 8.768892552899147e-07, + "loss": 0.8582, + "mean_token_accuracy": 0.7341989278793335, + "num_tokens": 96412605.0, + "step": 3192 + }, + { + "epoch": 0.2632967757895605, + "grad_norm": 1.9042437076568604, + "learning_rate": 8.77164056059357e-07, + "loss": 0.92, + "mean_token_accuracy": 0.7216432094573975, + "num_tokens": 96446850.0, + "step": 3193 + }, + { + "epoch": 0.26337923641461203, + "grad_norm": 2.164095163345337, + "learning_rate": 8.774388568287991e-07, + "loss": 0.9388, + "mean_token_accuracy": 0.7039585113525391, + "num_tokens": 96473861.0, + "step": 3194 + }, + { + "epoch": 0.26346169703966354, + "grad_norm": 1.9718228578567505, + "learning_rate": 8.777136575982412e-07, + "loss": 0.9426, + "mean_token_accuracy": 0.7097271084785461, + "num_tokens": 96507623.0, + "step": 3195 + }, + { + "epoch": 0.2635441576647151, + "grad_norm": 1.8548147678375244, + "learning_rate": 8.779884583676834e-07, + "loss": 0.7757, + "mean_token_accuracy": 0.7589925527572632, + "num_tokens": 96540423.0, + "step": 3196 + }, + { + "epoch": 0.26362661828976663, + "grad_norm": 2.0343873500823975, + "learning_rate": 8.782632591371255e-07, + "loss": 0.89, + "mean_token_accuracy": 0.722806453704834, + "num_tokens": 96571171.0, + "step": 3197 + }, + { + "epoch": 0.26370907891481815, + "grad_norm": 2.1517834663391113, + "learning_rate": 8.785380599065677e-07, + "loss": 0.8075, + "mean_token_accuracy": 0.7428122162818909, + "num_tokens": 96598283.0, + "step": 3198 + }, + { + "epoch": 0.2637915395398697, + "grad_norm": 2.3075551986694336, + "learning_rate": 8.788128606760099e-07, + "loss": 0.7783, + "mean_token_accuracy": 0.7602455615997314, + "num_tokens": 96620916.0, + "step": 3199 + }, + { + "epoch": 0.26387400016492124, + "grad_norm": 2.0498199462890625, + "learning_rate": 8.79087661445452e-07, + "loss": 0.857, + "mean_token_accuracy": 0.7371186017990112, + "num_tokens": 96648724.0, + "step": 3200 + }, + { + "epoch": 0.2639564607899728, + "grad_norm": 2.0284626483917236, + "learning_rate": 8.793624622148942e-07, + "loss": 0.7859, + "mean_token_accuracy": 0.7523159980773926, + "num_tokens": 96676378.0, + "step": 3201 + }, + { + "epoch": 0.26403892141502433, + "grad_norm": 2.00900936126709, + "learning_rate": 8.796372629843362e-07, + "loss": 0.9248, + "mean_token_accuracy": 0.7171025276184082, + "num_tokens": 96705522.0, + "step": 3202 + }, + { + "epoch": 0.26412138204007585, + "grad_norm": 2.6571123600006104, + "learning_rate": 8.799120637537785e-07, + "loss": 1.0041, + "mean_token_accuracy": 0.7053487300872803, + "num_tokens": 96728899.0, + "step": 3203 + }, + { + "epoch": 0.2642038426651274, + "grad_norm": 2.060237407684326, + "learning_rate": 8.801868645232207e-07, + "loss": 0.8576, + "mean_token_accuracy": 0.7239236235618591, + "num_tokens": 96759700.0, + "step": 3204 + }, + { + "epoch": 0.26428630329017894, + "grad_norm": 2.3124496936798096, + "learning_rate": 8.804616652926627e-07, + "loss": 0.9272, + "mean_token_accuracy": 0.7213441133499146, + "num_tokens": 96783896.0, + "step": 3205 + }, + { + "epoch": 0.26436876391523045, + "grad_norm": 2.3121602535247803, + "learning_rate": 8.807364660621049e-07, + "loss": 0.8635, + "mean_token_accuracy": 0.7350187301635742, + "num_tokens": 96806579.0, + "step": 3206 + }, + { + "epoch": 0.264451224540282, + "grad_norm": 1.947806715965271, + "learning_rate": 8.810112668315472e-07, + "loss": 0.7464, + "mean_token_accuracy": 0.7604238390922546, + "num_tokens": 96838811.0, + "step": 3207 + }, + { + "epoch": 0.26453368516533354, + "grad_norm": 1.9682660102844238, + "learning_rate": 8.812860676009893e-07, + "loss": 0.855, + "mean_token_accuracy": 0.7485986351966858, + "num_tokens": 96867761.0, + "step": 3208 + }, + { + "epoch": 0.2646161457903851, + "grad_norm": 2.0094423294067383, + "learning_rate": 8.815608683704314e-07, + "loss": 0.8039, + "mean_token_accuracy": 0.7463339567184448, + "num_tokens": 96899732.0, + "step": 3209 + }, + { + "epoch": 0.26469860641543663, + "grad_norm": 2.273207187652588, + "learning_rate": 8.818356691398735e-07, + "loss": 0.7799, + "mean_token_accuracy": 0.751299262046814, + "num_tokens": 96924840.0, + "step": 3210 + }, + { + "epoch": 0.26478106704048815, + "grad_norm": 2.1656315326690674, + "learning_rate": 8.821104699093157e-07, + "loss": 0.8514, + "mean_token_accuracy": 0.7347451448440552, + "num_tokens": 96954192.0, + "step": 3211 + }, + { + "epoch": 0.2648635276655397, + "grad_norm": 2.02591609954834, + "learning_rate": 8.823852706787579e-07, + "loss": 0.9408, + "mean_token_accuracy": 0.7167550921440125, + "num_tokens": 96986600.0, + "step": 3212 + }, + { + "epoch": 0.26494598829059124, + "grad_norm": 2.2030160427093506, + "learning_rate": 8.826600714482e-07, + "loss": 0.8824, + "mean_token_accuracy": 0.7204549908638, + "num_tokens": 97014493.0, + "step": 3213 + }, + { + "epoch": 0.26502844891564276, + "grad_norm": 1.8461546897888184, + "learning_rate": 8.829348722176422e-07, + "loss": 0.7377, + "mean_token_accuracy": 0.7675585746765137, + "num_tokens": 97050380.0, + "step": 3214 + }, + { + "epoch": 0.26511090954069433, + "grad_norm": 1.8903871774673462, + "learning_rate": 8.832096729870843e-07, + "loss": 0.7835, + "mean_token_accuracy": 0.7530006766319275, + "num_tokens": 97084005.0, + "step": 3215 + }, + { + "epoch": 0.26519337016574585, + "grad_norm": 2.341266393661499, + "learning_rate": 8.834844737565264e-07, + "loss": 0.8365, + "mean_token_accuracy": 0.745089054107666, + "num_tokens": 97109604.0, + "step": 3216 + }, + { + "epoch": 0.2652758307907974, + "grad_norm": 2.313816785812378, + "learning_rate": 8.837592745259687e-07, + "loss": 0.9275, + "mean_token_accuracy": 0.7087475061416626, + "num_tokens": 97136955.0, + "step": 3217 + }, + { + "epoch": 0.26535829141584893, + "grad_norm": 1.8715238571166992, + "learning_rate": 8.840340752954108e-07, + "loss": 0.7757, + "mean_token_accuracy": 0.7566240429878235, + "num_tokens": 97172183.0, + "step": 3218 + }, + { + "epoch": 0.26544075204090045, + "grad_norm": 2.2839627265930176, + "learning_rate": 8.843088760648529e-07, + "loss": 0.8625, + "mean_token_accuracy": 0.7354557514190674, + "num_tokens": 97199278.0, + "step": 3219 + }, + { + "epoch": 0.265523212665952, + "grad_norm": 2.144487142562866, + "learning_rate": 8.845836768342951e-07, + "loss": 0.9688, + "mean_token_accuracy": 0.7055472731590271, + "num_tokens": 97226241.0, + "step": 3220 + }, + { + "epoch": 0.26560567329100354, + "grad_norm": 2.1685709953308105, + "learning_rate": 8.848584776037373e-07, + "loss": 0.8018, + "mean_token_accuracy": 0.7503498792648315, + "num_tokens": 97253742.0, + "step": 3221 + }, + { + "epoch": 0.26568813391605506, + "grad_norm": 1.6790274381637573, + "learning_rate": 8.851332783731794e-07, + "loss": 0.77, + "mean_token_accuracy": 0.7603166699409485, + "num_tokens": 97294311.0, + "step": 3222 + }, + { + "epoch": 0.26577059454110663, + "grad_norm": 1.9802292585372925, + "learning_rate": 8.854080791426216e-07, + "loss": 0.9106, + "mean_token_accuracy": 0.7283381223678589, + "num_tokens": 97327315.0, + "step": 3223 + }, + { + "epoch": 0.26585305516615815, + "grad_norm": 2.129884719848633, + "learning_rate": 8.856828799120637e-07, + "loss": 0.8959, + "mean_token_accuracy": 0.7200907468795776, + "num_tokens": 97354488.0, + "step": 3224 + }, + { + "epoch": 0.2659355157912097, + "grad_norm": 1.9978355169296265, + "learning_rate": 8.859576806815059e-07, + "loss": 0.8269, + "mean_token_accuracy": 0.7451276183128357, + "num_tokens": 97384656.0, + "step": 3225 + }, + { + "epoch": 0.26601797641626124, + "grad_norm": 2.040438652038574, + "learning_rate": 8.86232481450948e-07, + "loss": 0.7361, + "mean_token_accuracy": 0.7751886248588562, + "num_tokens": 97417517.0, + "step": 3226 + }, + { + "epoch": 0.26610043704131275, + "grad_norm": 1.9389303922653198, + "learning_rate": 8.865072822203902e-07, + "loss": 0.8048, + "mean_token_accuracy": 0.7454460859298706, + "num_tokens": 97447673.0, + "step": 3227 + }, + { + "epoch": 0.2661828976663643, + "grad_norm": 1.9512428045272827, + "learning_rate": 8.867820829898324e-07, + "loss": 1.0031, + "mean_token_accuracy": 0.7004824876785278, + "num_tokens": 97484173.0, + "step": 3228 + }, + { + "epoch": 0.26626535829141584, + "grad_norm": 1.9562654495239258, + "learning_rate": 8.870568837592744e-07, + "loss": 0.8742, + "mean_token_accuracy": 0.734378457069397, + "num_tokens": 97517292.0, + "step": 3229 + }, + { + "epoch": 0.26634781891646736, + "grad_norm": 2.4040403366088867, + "learning_rate": 8.873316845287166e-07, + "loss": 0.7645, + "mean_token_accuracy": 0.7589819431304932, + "num_tokens": 97541354.0, + "step": 3230 + }, + { + "epoch": 0.26643027954151893, + "grad_norm": 2.2436234951019287, + "learning_rate": 8.876064852981589e-07, + "loss": 0.9374, + "mean_token_accuracy": 0.7245005369186401, + "num_tokens": 97569383.0, + "step": 3231 + }, + { + "epoch": 0.26651274016657045, + "grad_norm": 2.107173204421997, + "learning_rate": 8.878812860676009e-07, + "loss": 0.822, + "mean_token_accuracy": 0.7447227239608765, + "num_tokens": 97595913.0, + "step": 3232 + }, + { + "epoch": 0.266595200791622, + "grad_norm": 2.116095542907715, + "learning_rate": 8.881560868370431e-07, + "loss": 0.9114, + "mean_token_accuracy": 0.7258272171020508, + "num_tokens": 97623811.0, + "step": 3233 + }, + { + "epoch": 0.26667766141667354, + "grad_norm": 2.1260924339294434, + "learning_rate": 8.884308876064852e-07, + "loss": 0.8609, + "mean_token_accuracy": 0.7298499345779419, + "num_tokens": 97651142.0, + "step": 3234 + }, + { + "epoch": 0.26676012204172506, + "grad_norm": 2.229644775390625, + "learning_rate": 8.887056883759275e-07, + "loss": 0.8156, + "mean_token_accuracy": 0.7423775792121887, + "num_tokens": 97676331.0, + "step": 3235 + }, + { + "epoch": 0.26684258266677663, + "grad_norm": 1.8452274799346924, + "learning_rate": 8.889804891453696e-07, + "loss": 0.7334, + "mean_token_accuracy": 0.7671387195587158, + "num_tokens": 97712980.0, + "step": 3236 + }, + { + "epoch": 0.26692504329182815, + "grad_norm": 1.9767673015594482, + "learning_rate": 8.892552899148117e-07, + "loss": 0.7329, + "mean_token_accuracy": 0.7734564542770386, + "num_tokens": 97740421.0, + "step": 3237 + }, + { + "epoch": 0.26700750391687966, + "grad_norm": 2.375680923461914, + "learning_rate": 8.895300906842539e-07, + "loss": 0.8222, + "mean_token_accuracy": 0.7475848197937012, + "num_tokens": 97763132.0, + "step": 3238 + }, + { + "epoch": 0.26708996454193124, + "grad_norm": 2.1944613456726074, + "learning_rate": 8.89804891453696e-07, + "loss": 0.9468, + "mean_token_accuracy": 0.7180581092834473, + "num_tokens": 97790716.0, + "step": 3239 + }, + { + "epoch": 0.26717242516698275, + "grad_norm": 1.9439724683761597, + "learning_rate": 8.900796922231382e-07, + "loss": 0.8558, + "mean_token_accuracy": 0.7236964702606201, + "num_tokens": 97825214.0, + "step": 3240 + }, + { + "epoch": 0.2672548857920343, + "grad_norm": 1.9740978479385376, + "learning_rate": 8.903544929925804e-07, + "loss": 0.7645, + "mean_token_accuracy": 0.764146089553833, + "num_tokens": 97855847.0, + "step": 3241 + }, + { + "epoch": 0.26733734641708584, + "grad_norm": 2.501742362976074, + "learning_rate": 8.906292937620224e-07, + "loss": 0.8762, + "mean_token_accuracy": 0.7314049005508423, + "num_tokens": 97878301.0, + "step": 3242 + }, + { + "epoch": 0.26741980704213736, + "grad_norm": 2.2916276454925537, + "learning_rate": 8.909040945314646e-07, + "loss": 0.8472, + "mean_token_accuracy": 0.7364094257354736, + "num_tokens": 97904284.0, + "step": 3243 + }, + { + "epoch": 0.26750226766718893, + "grad_norm": 2.316596746444702, + "learning_rate": 8.911788953009068e-07, + "loss": 0.9391, + "mean_token_accuracy": 0.714890718460083, + "num_tokens": 97929717.0, + "step": 3244 + }, + { + "epoch": 0.26758472829224045, + "grad_norm": 1.8922133445739746, + "learning_rate": 8.91453696070349e-07, + "loss": 0.7405, + "mean_token_accuracy": 0.7709904909133911, + "num_tokens": 97964211.0, + "step": 3245 + }, + { + "epoch": 0.267667188917292, + "grad_norm": 1.791792631149292, + "learning_rate": 8.917284968397911e-07, + "loss": 0.7889, + "mean_token_accuracy": 0.7472127676010132, + "num_tokens": 97999956.0, + "step": 3246 + }, + { + "epoch": 0.26774964954234354, + "grad_norm": 2.113999843597412, + "learning_rate": 8.920032976092333e-07, + "loss": 0.711, + "mean_token_accuracy": 0.772797703742981, + "num_tokens": 98026044.0, + "step": 3247 + }, + { + "epoch": 0.26783211016739505, + "grad_norm": 2.013232707977295, + "learning_rate": 8.922780983786754e-07, + "loss": 0.8186, + "mean_token_accuracy": 0.7468981742858887, + "num_tokens": 98054065.0, + "step": 3248 + }, + { + "epoch": 0.2679145707924466, + "grad_norm": 1.8383134603500366, + "learning_rate": 8.925528991481176e-07, + "loss": 0.8268, + "mean_token_accuracy": 0.7393255233764648, + "num_tokens": 98085085.0, + "step": 3249 + }, + { + "epoch": 0.26799703141749814, + "grad_norm": 2.406935691833496, + "learning_rate": 8.928276999175597e-07, + "loss": 0.6298, + "mean_token_accuracy": 0.7830695509910583, + "num_tokens": 98111455.0, + "step": 3250 + }, + { + "epoch": 0.26807949204254966, + "grad_norm": 1.921833872795105, + "learning_rate": 8.931025006870019e-07, + "loss": 0.8458, + "mean_token_accuracy": 0.7537633776664734, + "num_tokens": 98141276.0, + "step": 3251 + }, + { + "epoch": 0.26816195266760123, + "grad_norm": 2.119011163711548, + "learning_rate": 8.933773014564441e-07, + "loss": 0.8303, + "mean_token_accuracy": 0.7357996702194214, + "num_tokens": 98169733.0, + "step": 3252 + }, + { + "epoch": 0.26824441329265275, + "grad_norm": 2.2865960597991943, + "learning_rate": 8.936521022258861e-07, + "loss": 0.7586, + "mean_token_accuracy": 0.7593629360198975, + "num_tokens": 98193440.0, + "step": 3253 + }, + { + "epoch": 0.2683268739177043, + "grad_norm": 1.8276115655899048, + "learning_rate": 8.939269029953284e-07, + "loss": 0.9033, + "mean_token_accuracy": 0.7247439622879028, + "num_tokens": 98231321.0, + "step": 3254 + }, + { + "epoch": 0.26840933454275584, + "grad_norm": 2.397026777267456, + "learning_rate": 8.942017037647706e-07, + "loss": 0.9163, + "mean_token_accuracy": 0.719527006149292, + "num_tokens": 98255737.0, + "step": 3255 + }, + { + "epoch": 0.26849179516780736, + "grad_norm": 1.7795360088348389, + "learning_rate": 8.944765045342126e-07, + "loss": 0.7426, + "mean_token_accuracy": 0.76338130235672, + "num_tokens": 98290442.0, + "step": 3256 + }, + { + "epoch": 0.26857425579285893, + "grad_norm": 1.9333624839782715, + "learning_rate": 8.947513053036548e-07, + "loss": 0.7813, + "mean_token_accuracy": 0.7561880946159363, + "num_tokens": 98323193.0, + "step": 3257 + }, + { + "epoch": 0.26865671641791045, + "grad_norm": 1.7795459032058716, + "learning_rate": 8.950261060730969e-07, + "loss": 0.8789, + "mean_token_accuracy": 0.7357226610183716, + "num_tokens": 98362498.0, + "step": 3258 + }, + { + "epoch": 0.26873917704296196, + "grad_norm": 2.0264201164245605, + "learning_rate": 8.953009068425392e-07, + "loss": 0.8821, + "mean_token_accuracy": 0.7231096029281616, + "num_tokens": 98393507.0, + "step": 3259 + }, + { + "epoch": 0.26882163766801354, + "grad_norm": 1.9881235361099243, + "learning_rate": 8.955757076119813e-07, + "loss": 0.6884, + "mean_token_accuracy": 0.7676137089729309, + "num_tokens": 98425520.0, + "step": 3260 + }, + { + "epoch": 0.26890409829306505, + "grad_norm": 2.0076119899749756, + "learning_rate": 8.958505083814234e-07, + "loss": 0.8654, + "mean_token_accuracy": 0.727433443069458, + "num_tokens": 98457461.0, + "step": 3261 + }, + { + "epoch": 0.2689865589181166, + "grad_norm": 2.0121092796325684, + "learning_rate": 8.961253091508656e-07, + "loss": 0.8054, + "mean_token_accuracy": 0.7401853799819946, + "num_tokens": 98489033.0, + "step": 3262 + }, + { + "epoch": 0.26906901954316814, + "grad_norm": 2.216907262802124, + "learning_rate": 8.964001099203078e-07, + "loss": 0.8785, + "mean_token_accuracy": 0.7170020341873169, + "num_tokens": 98518807.0, + "step": 3263 + }, + { + "epoch": 0.26915148016821966, + "grad_norm": 2.2799975872039795, + "learning_rate": 8.966749106897499e-07, + "loss": 0.802, + "mean_token_accuracy": 0.7456587553024292, + "num_tokens": 98544444.0, + "step": 3264 + }, + { + "epoch": 0.26923394079327123, + "grad_norm": 2.1219844818115234, + "learning_rate": 8.969497114591921e-07, + "loss": 0.9268, + "mean_token_accuracy": 0.7150046825408936, + "num_tokens": 98574151.0, + "step": 3265 + }, + { + "epoch": 0.26931640141832275, + "grad_norm": 2.4560744762420654, + "learning_rate": 8.972245122286341e-07, + "loss": 0.77, + "mean_token_accuracy": 0.750756025314331, + "num_tokens": 98594052.0, + "step": 3266 + }, + { + "epoch": 0.26939886204337427, + "grad_norm": 1.8970348834991455, + "learning_rate": 8.974993129980763e-07, + "loss": 0.7357, + "mean_token_accuracy": 0.7616891860961914, + "num_tokens": 98625213.0, + "step": 3267 + }, + { + "epoch": 0.26948132266842584, + "grad_norm": 2.061037063598633, + "learning_rate": 8.977741137675186e-07, + "loss": 0.8671, + "mean_token_accuracy": 0.7276613116264343, + "num_tokens": 98653904.0, + "step": 3268 + }, + { + "epoch": 0.26956378329347735, + "grad_norm": 2.083249092102051, + "learning_rate": 8.980489145369607e-07, + "loss": 0.8785, + "mean_token_accuracy": 0.7394304275512695, + "num_tokens": 98682215.0, + "step": 3269 + }, + { + "epoch": 0.2696462439185289, + "grad_norm": 2.1829092502593994, + "learning_rate": 8.983237153064028e-07, + "loss": 0.9744, + "mean_token_accuracy": 0.7050272226333618, + "num_tokens": 98709221.0, + "step": 3270 + }, + { + "epoch": 0.26972870454358044, + "grad_norm": 2.083782434463501, + "learning_rate": 8.98598516075845e-07, + "loss": 0.8371, + "mean_token_accuracy": 0.7508925199508667, + "num_tokens": 98737843.0, + "step": 3271 + }, + { + "epoch": 0.26981116516863196, + "grad_norm": 2.1614532470703125, + "learning_rate": 8.988733168452871e-07, + "loss": 0.8785, + "mean_token_accuracy": 0.7213435173034668, + "num_tokens": 98764235.0, + "step": 3272 + }, + { + "epoch": 0.26989362579368353, + "grad_norm": 1.9868241548538208, + "learning_rate": 8.991481176147293e-07, + "loss": 0.8529, + "mean_token_accuracy": 0.7357844114303589, + "num_tokens": 98793284.0, + "step": 3273 + }, + { + "epoch": 0.26997608641873505, + "grad_norm": 1.934643268585205, + "learning_rate": 8.994229183841714e-07, + "loss": 0.763, + "mean_token_accuracy": 0.7709333300590515, + "num_tokens": 98828144.0, + "step": 3274 + }, + { + "epoch": 0.27005854704378657, + "grad_norm": 2.2543587684631348, + "learning_rate": 8.996977191536136e-07, + "loss": 0.791, + "mean_token_accuracy": 0.7355688214302063, + "num_tokens": 98856370.0, + "step": 3275 + }, + { + "epoch": 0.27014100766883814, + "grad_norm": 2.151214122772217, + "learning_rate": 8.999725199230558e-07, + "loss": 0.8897, + "mean_token_accuracy": 0.725283682346344, + "num_tokens": 98883616.0, + "step": 3276 + }, + { + "epoch": 0.27022346829388966, + "grad_norm": 2.278592348098755, + "learning_rate": 9.002473206924978e-07, + "loss": 0.849, + "mean_token_accuracy": 0.7398346662521362, + "num_tokens": 98910135.0, + "step": 3277 + }, + { + "epoch": 0.27030592891894123, + "grad_norm": 2.025714874267578, + "learning_rate": 9.005221214619401e-07, + "loss": 0.7239, + "mean_token_accuracy": 0.778698205947876, + "num_tokens": 98937516.0, + "step": 3278 + }, + { + "epoch": 0.27038838954399275, + "grad_norm": 1.879880428314209, + "learning_rate": 9.007969222313823e-07, + "loss": 0.8606, + "mean_token_accuracy": 0.7360262870788574, + "num_tokens": 98972052.0, + "step": 3279 + }, + { + "epoch": 0.27047085016904426, + "grad_norm": 1.8540706634521484, + "learning_rate": 9.010717230008243e-07, + "loss": 0.7716, + "mean_token_accuracy": 0.7605283260345459, + "num_tokens": 99007579.0, + "step": 3280 + }, + { + "epoch": 0.27055331079409584, + "grad_norm": 1.9802801609039307, + "learning_rate": 9.013465237702665e-07, + "loss": 0.7152, + "mean_token_accuracy": 0.7576450705528259, + "num_tokens": 99032796.0, + "step": 3281 + }, + { + "epoch": 0.27063577141914735, + "grad_norm": 1.6806200742721558, + "learning_rate": 9.016213245397087e-07, + "loss": 0.6567, + "mean_token_accuracy": 0.7795151472091675, + "num_tokens": 99068646.0, + "step": 3282 + }, + { + "epoch": 0.27071823204419887, + "grad_norm": 1.9928580522537231, + "learning_rate": 9.018961253091508e-07, + "loss": 0.8433, + "mean_token_accuracy": 0.7321860790252686, + "num_tokens": 99098262.0, + "step": 3283 + }, + { + "epoch": 0.27080069266925044, + "grad_norm": 1.9509896039962769, + "learning_rate": 9.02170926078593e-07, + "loss": 0.7956, + "mean_token_accuracy": 0.7464627027511597, + "num_tokens": 99128704.0, + "step": 3284 + }, + { + "epoch": 0.27088315329430196, + "grad_norm": 2.1418793201446533, + "learning_rate": 9.024457268480351e-07, + "loss": 0.9078, + "mean_token_accuracy": 0.7237647175788879, + "num_tokens": 99155352.0, + "step": 3285 + }, + { + "epoch": 0.27096561391935353, + "grad_norm": 1.9688622951507568, + "learning_rate": 9.027205276174773e-07, + "loss": 0.9172, + "mean_token_accuracy": 0.7207881212234497, + "num_tokens": 99189174.0, + "step": 3286 + }, + { + "epoch": 0.27104807454440505, + "grad_norm": 1.7712681293487549, + "learning_rate": 9.029953283869195e-07, + "loss": 0.7845, + "mean_token_accuracy": 0.753121554851532, + "num_tokens": 99227297.0, + "step": 3287 + }, + { + "epoch": 0.27113053516945657, + "grad_norm": 2.005197763442993, + "learning_rate": 9.032701291563616e-07, + "loss": 0.7435, + "mean_token_accuracy": 0.7577853202819824, + "num_tokens": 99256486.0, + "step": 3288 + }, + { + "epoch": 0.27121299579450814, + "grad_norm": 2.050677537918091, + "learning_rate": 9.035449299258038e-07, + "loss": 0.8065, + "mean_token_accuracy": 0.7400500774383545, + "num_tokens": 99283401.0, + "step": 3289 + }, + { + "epoch": 0.27129545641955966, + "grad_norm": 2.0389816761016846, + "learning_rate": 9.038197306952458e-07, + "loss": 0.9199, + "mean_token_accuracy": 0.7201019525527954, + "num_tokens": 99312299.0, + "step": 3290 + }, + { + "epoch": 0.27137791704461117, + "grad_norm": 2.0426485538482666, + "learning_rate": 9.04094531464688e-07, + "loss": 0.7574, + "mean_token_accuracy": 0.7638643980026245, + "num_tokens": 99342948.0, + "step": 3291 + }, + { + "epoch": 0.27146037766966274, + "grad_norm": 2.087374448776245, + "learning_rate": 9.043693322341303e-07, + "loss": 0.9174, + "mean_token_accuracy": 0.7204259634017944, + "num_tokens": 99376624.0, + "step": 3292 + }, + { + "epoch": 0.27154283829471426, + "grad_norm": 1.9008737802505493, + "learning_rate": 9.046441330035724e-07, + "loss": 0.8623, + "mean_token_accuracy": 0.7373234629631042, + "num_tokens": 99410079.0, + "step": 3293 + }, + { + "epoch": 0.27162529891976583, + "grad_norm": 2.2513837814331055, + "learning_rate": 9.049189337730145e-07, + "loss": 0.9377, + "mean_token_accuracy": 0.7027024030685425, + "num_tokens": 99436199.0, + "step": 3294 + }, + { + "epoch": 0.27170775954481735, + "grad_norm": 2.1749017238616943, + "learning_rate": 9.051937345424567e-07, + "loss": 0.7886, + "mean_token_accuracy": 0.7480636835098267, + "num_tokens": 99462986.0, + "step": 3295 + }, + { + "epoch": 0.27179022016986887, + "grad_norm": 1.8287562131881714, + "learning_rate": 9.054685353118989e-07, + "loss": 0.744, + "mean_token_accuracy": 0.7694936990737915, + "num_tokens": 99500579.0, + "step": 3296 + }, + { + "epoch": 0.27187268079492044, + "grad_norm": 1.9717360734939575, + "learning_rate": 9.05743336081341e-07, + "loss": 0.9876, + "mean_token_accuracy": 0.6988987922668457, + "num_tokens": 99535206.0, + "step": 3297 + }, + { + "epoch": 0.27195514141997196, + "grad_norm": 2.2934930324554443, + "learning_rate": 9.060181368507831e-07, + "loss": 0.9543, + "mean_token_accuracy": 0.7081037759780884, + "num_tokens": 99558754.0, + "step": 3298 + }, + { + "epoch": 0.2720376020450235, + "grad_norm": 2.120272636413574, + "learning_rate": 9.062929376202253e-07, + "loss": 0.8556, + "mean_token_accuracy": 0.7350133657455444, + "num_tokens": 99588521.0, + "step": 3299 + }, + { + "epoch": 0.27212006267007505, + "grad_norm": 2.312058925628662, + "learning_rate": 9.065677383896674e-07, + "loss": 0.8762, + "mean_token_accuracy": 0.7276522517204285, + "num_tokens": 99613269.0, + "step": 3300 + }, + { + "epoch": 0.27220252329512656, + "grad_norm": 1.8364750146865845, + "learning_rate": 9.068425391591096e-07, + "loss": 0.9345, + "mean_token_accuracy": 0.7200896739959717, + "num_tokens": 99652662.0, + "step": 3301 + }, + { + "epoch": 0.27228498392017814, + "grad_norm": 1.895487666130066, + "learning_rate": 9.071173399285518e-07, + "loss": 0.7908, + "mean_token_accuracy": 0.7488327026367188, + "num_tokens": 99685111.0, + "step": 3302 + }, + { + "epoch": 0.27236744454522965, + "grad_norm": 2.121530055999756, + "learning_rate": 9.07392140697994e-07, + "loss": 0.923, + "mean_token_accuracy": 0.7124078273773193, + "num_tokens": 99715047.0, + "step": 3303 + }, + { + "epoch": 0.27244990517028117, + "grad_norm": 2.001588821411133, + "learning_rate": 9.07666941467436e-07, + "loss": 0.7993, + "mean_token_accuracy": 0.7506362795829773, + "num_tokens": 99743725.0, + "step": 3304 + }, + { + "epoch": 0.27253236579533274, + "grad_norm": 1.9569891691207886, + "learning_rate": 9.079417422368782e-07, + "loss": 0.8241, + "mean_token_accuracy": 0.740118145942688, + "num_tokens": 99776536.0, + "step": 3305 + }, + { + "epoch": 0.27261482642038426, + "grad_norm": 2.0332863330841064, + "learning_rate": 9.082165430063204e-07, + "loss": 0.7301, + "mean_token_accuracy": 0.761360764503479, + "num_tokens": 99806530.0, + "step": 3306 + }, + { + "epoch": 0.2726972870454358, + "grad_norm": 2.0490763187408447, + "learning_rate": 9.084913437757625e-07, + "loss": 0.9363, + "mean_token_accuracy": 0.7161049842834473, + "num_tokens": 99837311.0, + "step": 3307 + }, + { + "epoch": 0.27277974767048735, + "grad_norm": 1.9954112768173218, + "learning_rate": 9.087661445452047e-07, + "loss": 0.8883, + "mean_token_accuracy": 0.7326468229293823, + "num_tokens": 99866664.0, + "step": 3308 + }, + { + "epoch": 0.27286220829553887, + "grad_norm": 2.1145904064178467, + "learning_rate": 9.090409453146468e-07, + "loss": 0.8723, + "mean_token_accuracy": 0.7229611873626709, + "num_tokens": 99896083.0, + "step": 3309 + }, + { + "epoch": 0.27294466892059044, + "grad_norm": 2.023787498474121, + "learning_rate": 9.093157460840891e-07, + "loss": 0.915, + "mean_token_accuracy": 0.7201815843582153, + "num_tokens": 99925165.0, + "step": 3310 + }, + { + "epoch": 0.27302712954564196, + "grad_norm": 1.8593977689743042, + "learning_rate": 9.095905468535311e-07, + "loss": 0.8423, + "mean_token_accuracy": 0.7296943664550781, + "num_tokens": 99958847.0, + "step": 3311 + }, + { + "epoch": 0.2731095901706935, + "grad_norm": 1.898210048675537, + "learning_rate": 9.098653476229733e-07, + "loss": 0.7983, + "mean_token_accuracy": 0.7586281299591064, + "num_tokens": 99995459.0, + "step": 3312 + }, + { + "epoch": 0.27319205079574505, + "grad_norm": 2.174145221710205, + "learning_rate": 9.101401483924155e-07, + "loss": 0.7339, + "mean_token_accuracy": 0.7541102170944214, + "num_tokens": 100019831.0, + "step": 3313 + }, + { + "epoch": 0.27327451142079656, + "grad_norm": 2.3053269386291504, + "learning_rate": 9.104149491618575e-07, + "loss": 0.8121, + "mean_token_accuracy": 0.7412045001983643, + "num_tokens": 100043078.0, + "step": 3314 + }, + { + "epoch": 0.27335697204584813, + "grad_norm": 2.0691230297088623, + "learning_rate": 9.106897499312998e-07, + "loss": 0.7513, + "mean_token_accuracy": 0.7633155584335327, + "num_tokens": 100072572.0, + "step": 3315 + }, + { + "epoch": 0.27343943267089965, + "grad_norm": 2.1438190937042236, + "learning_rate": 9.10964550700742e-07, + "loss": 0.8623, + "mean_token_accuracy": 0.7317243814468384, + "num_tokens": 100100234.0, + "step": 3316 + }, + { + "epoch": 0.27352189329595117, + "grad_norm": 2.4285404682159424, + "learning_rate": 9.11239351470184e-07, + "loss": 0.8726, + "mean_token_accuracy": 0.7307536602020264, + "num_tokens": 100122832.0, + "step": 3317 + }, + { + "epoch": 0.27360435392100274, + "grad_norm": 2.1607718467712402, + "learning_rate": 9.115141522396262e-07, + "loss": 0.8181, + "mean_token_accuracy": 0.7394397258758545, + "num_tokens": 100148401.0, + "step": 3318 + }, + { + "epoch": 0.27368681454605426, + "grad_norm": 2.246742010116577, + "learning_rate": 9.117889530090683e-07, + "loss": 0.9891, + "mean_token_accuracy": 0.7006712555885315, + "num_tokens": 100174532.0, + "step": 3319 + }, + { + "epoch": 0.2737692751711058, + "grad_norm": 2.387838363647461, + "learning_rate": 9.120637537785106e-07, + "loss": 0.826, + "mean_token_accuracy": 0.7380138039588928, + "num_tokens": 100196094.0, + "step": 3320 + }, + { + "epoch": 0.27385173579615735, + "grad_norm": 1.8141734600067139, + "learning_rate": 9.123385545479527e-07, + "loss": 0.8871, + "mean_token_accuracy": 0.7355643510818481, + "num_tokens": 100233815.0, + "step": 3321 + }, + { + "epoch": 0.27393419642120886, + "grad_norm": 2.0845773220062256, + "learning_rate": 9.126133553173948e-07, + "loss": 0.8807, + "mean_token_accuracy": 0.7280584573745728, + "num_tokens": 100264019.0, + "step": 3322 + }, + { + "epoch": 0.27401665704626044, + "grad_norm": 1.8865699768066406, + "learning_rate": 9.12888156086837e-07, + "loss": 0.7141, + "mean_token_accuracy": 0.7652730345726013, + "num_tokens": 100299106.0, + "step": 3323 + }, + { + "epoch": 0.27409911767131195, + "grad_norm": 2.366583824157715, + "learning_rate": 9.131629568562792e-07, + "loss": 0.8073, + "mean_token_accuracy": 0.750011682510376, + "num_tokens": 100320255.0, + "step": 3324 + }, + { + "epoch": 0.27418157829636347, + "grad_norm": 2.0194408893585205, + "learning_rate": 9.134377576257213e-07, + "loss": 0.9259, + "mean_token_accuracy": 0.7154312133789062, + "num_tokens": 100352975.0, + "step": 3325 + }, + { + "epoch": 0.27426403892141504, + "grad_norm": 2.1497859954833984, + "learning_rate": 9.137125583951635e-07, + "loss": 0.7107, + "mean_token_accuracy": 0.7751524448394775, + "num_tokens": 100377890.0, + "step": 3326 + }, + { + "epoch": 0.27434649954646656, + "grad_norm": 2.090444803237915, + "learning_rate": 9.139873591646055e-07, + "loss": 0.7775, + "mean_token_accuracy": 0.7492703795433044, + "num_tokens": 100407158.0, + "step": 3327 + }, + { + "epoch": 0.2744289601715181, + "grad_norm": 2.3309733867645264, + "learning_rate": 9.142621599340477e-07, + "loss": 0.8899, + "mean_token_accuracy": 0.7245233058929443, + "num_tokens": 100431966.0, + "step": 3328 + }, + { + "epoch": 0.27451142079656965, + "grad_norm": 2.2696902751922607, + "learning_rate": 9.1453696070349e-07, + "loss": 0.7864, + "mean_token_accuracy": 0.7477949261665344, + "num_tokens": 100455570.0, + "step": 3329 + }, + { + "epoch": 0.27459388142162117, + "grad_norm": 2.0773367881774902, + "learning_rate": 9.148117614729321e-07, + "loss": 0.9407, + "mean_token_accuracy": 0.7068178653717041, + "num_tokens": 100485302.0, + "step": 3330 + }, + { + "epoch": 0.27467634204667274, + "grad_norm": 2.065391778945923, + "learning_rate": 9.150865622423742e-07, + "loss": 0.8441, + "mean_token_accuracy": 0.7363696098327637, + "num_tokens": 100512974.0, + "step": 3331 + }, + { + "epoch": 0.27475880267172426, + "grad_norm": 2.352229595184326, + "learning_rate": 9.153613630118164e-07, + "loss": 0.8129, + "mean_token_accuracy": 0.7485784292221069, + "num_tokens": 100537674.0, + "step": 3332 + }, + { + "epoch": 0.2748412632967758, + "grad_norm": 2.6067709922790527, + "learning_rate": 9.156361637812585e-07, + "loss": 0.9335, + "mean_token_accuracy": 0.7108598947525024, + "num_tokens": 100556635.0, + "step": 3333 + }, + { + "epoch": 0.27492372392182735, + "grad_norm": 1.9634952545166016, + "learning_rate": 9.159109645507008e-07, + "loss": 0.8019, + "mean_token_accuracy": 0.7533135414123535, + "num_tokens": 100590381.0, + "step": 3334 + }, + { + "epoch": 0.27500618454687886, + "grad_norm": 2.0365161895751953, + "learning_rate": 9.161857653201428e-07, + "loss": 0.9431, + "mean_token_accuracy": 0.7151817083358765, + "num_tokens": 100621513.0, + "step": 3335 + }, + { + "epoch": 0.2750886451719304, + "grad_norm": 1.7772204875946045, + "learning_rate": 9.16460566089585e-07, + "loss": 0.6688, + "mean_token_accuracy": 0.789897620677948, + "num_tokens": 100654585.0, + "step": 3336 + }, + { + "epoch": 0.27517110579698195, + "grad_norm": 1.9658600091934204, + "learning_rate": 9.167353668590272e-07, + "loss": 0.7434, + "mean_token_accuracy": 0.7700791358947754, + "num_tokens": 100684930.0, + "step": 3337 + }, + { + "epoch": 0.27525356642203347, + "grad_norm": 1.9333585500717163, + "learning_rate": 9.170101676284692e-07, + "loss": 0.7875, + "mean_token_accuracy": 0.7547301054000854, + "num_tokens": 100717209.0, + "step": 3338 + }, + { + "epoch": 0.27533602704708504, + "grad_norm": 2.331674098968506, + "learning_rate": 9.172849683979115e-07, + "loss": 0.9079, + "mean_token_accuracy": 0.7178543210029602, + "num_tokens": 100740799.0, + "step": 3339 + }, + { + "epoch": 0.27541848767213656, + "grad_norm": 1.9635252952575684, + "learning_rate": 9.175597691673537e-07, + "loss": 0.7081, + "mean_token_accuracy": 0.77532559633255, + "num_tokens": 100768485.0, + "step": 3340 + }, + { + "epoch": 0.2755009482971881, + "grad_norm": 2.2727134227752686, + "learning_rate": 9.178345699367957e-07, + "loss": 0.9464, + "mean_token_accuracy": 0.7184421420097351, + "num_tokens": 100794558.0, + "step": 3341 + }, + { + "epoch": 0.27558340892223965, + "grad_norm": 2.0027382373809814, + "learning_rate": 9.181093707062379e-07, + "loss": 0.8336, + "mean_token_accuracy": 0.7321264743804932, + "num_tokens": 100825671.0, + "step": 3342 + }, + { + "epoch": 0.27566586954729116, + "grad_norm": 1.7408868074417114, + "learning_rate": 9.183841714756801e-07, + "loss": 0.7892, + "mean_token_accuracy": 0.7531566619873047, + "num_tokens": 100861416.0, + "step": 3343 + }, + { + "epoch": 0.2757483301723427, + "grad_norm": 1.810213327407837, + "learning_rate": 9.186589722451223e-07, + "loss": 0.8583, + "mean_token_accuracy": 0.7293412089347839, + "num_tokens": 100899339.0, + "step": 3344 + }, + { + "epoch": 0.27583079079739425, + "grad_norm": 1.905657410621643, + "learning_rate": 9.189337730145644e-07, + "loss": 0.8207, + "mean_token_accuracy": 0.7513031959533691, + "num_tokens": 100931615.0, + "step": 3345 + }, + { + "epoch": 0.27591325142244577, + "grad_norm": 2.042693614959717, + "learning_rate": 9.192085737840065e-07, + "loss": 0.8605, + "mean_token_accuracy": 0.7309791445732117, + "num_tokens": 100960206.0, + "step": 3346 + }, + { + "epoch": 0.27599571204749734, + "grad_norm": 1.808808445930481, + "learning_rate": 9.194833745534487e-07, + "loss": 0.6435, + "mean_token_accuracy": 0.7913779616355896, + "num_tokens": 100993319.0, + "step": 3347 + }, + { + "epoch": 0.27607817267254886, + "grad_norm": 1.9717248678207397, + "learning_rate": 9.197581753228909e-07, + "loss": 0.7209, + "mean_token_accuracy": 0.7659690976142883, + "num_tokens": 101020991.0, + "step": 3348 + }, + { + "epoch": 0.2761606332976004, + "grad_norm": 2.117260456085205, + "learning_rate": 9.20032976092333e-07, + "loss": 0.8185, + "mean_token_accuracy": 0.7474434971809387, + "num_tokens": 101049458.0, + "step": 3349 + }, + { + "epoch": 0.27624309392265195, + "grad_norm": 1.9175139665603638, + "learning_rate": 9.203077768617752e-07, + "loss": 0.8774, + "mean_token_accuracy": 0.7317588329315186, + "num_tokens": 101085349.0, + "step": 3350 + }, + { + "epoch": 0.27632555454770347, + "grad_norm": 2.123823404312134, + "learning_rate": 9.205825776312172e-07, + "loss": 0.9768, + "mean_token_accuracy": 0.707622230052948, + "num_tokens": 101116396.0, + "step": 3351 + }, + { + "epoch": 0.276408015172755, + "grad_norm": 1.9607850313186646, + "learning_rate": 9.208573784006595e-07, + "loss": 0.8557, + "mean_token_accuracy": 0.7372555136680603, + "num_tokens": 101147988.0, + "step": 3352 + }, + { + "epoch": 0.27649047579780656, + "grad_norm": 2.0567948818206787, + "learning_rate": 9.211321791701017e-07, + "loss": 0.8784, + "mean_token_accuracy": 0.718108057975769, + "num_tokens": 101176850.0, + "step": 3353 + }, + { + "epoch": 0.2765729364228581, + "grad_norm": 2.2118637561798096, + "learning_rate": 9.214069799395438e-07, + "loss": 0.8735, + "mean_token_accuracy": 0.7368901968002319, + "num_tokens": 101202217.0, + "step": 3354 + }, + { + "epoch": 0.27665539704790965, + "grad_norm": 2.2856149673461914, + "learning_rate": 9.216817807089859e-07, + "loss": 0.8138, + "mean_token_accuracy": 0.7486339807510376, + "num_tokens": 101226378.0, + "step": 3355 + }, + { + "epoch": 0.27673785767296116, + "grad_norm": 2.0941452980041504, + "learning_rate": 9.219565814784281e-07, + "loss": 0.6912, + "mean_token_accuracy": 0.778224527835846, + "num_tokens": 101252847.0, + "step": 3356 + }, + { + "epoch": 0.2768203182980127, + "grad_norm": 2.0373106002807617, + "learning_rate": 9.222313822478703e-07, + "loss": 0.8573, + "mean_token_accuracy": 0.729921817779541, + "num_tokens": 101281242.0, + "step": 3357 + }, + { + "epoch": 0.27690277892306425, + "grad_norm": 2.029207229614258, + "learning_rate": 9.225061830173124e-07, + "loss": 0.7919, + "mean_token_accuracy": 0.7544115781784058, + "num_tokens": 101309514.0, + "step": 3358 + }, + { + "epoch": 0.27698523954811577, + "grad_norm": 1.8109281063079834, + "learning_rate": 9.227809837867545e-07, + "loss": 0.778, + "mean_token_accuracy": 0.7514623403549194, + "num_tokens": 101346588.0, + "step": 3359 + }, + { + "epoch": 0.2770677001731673, + "grad_norm": 1.9919005632400513, + "learning_rate": 9.230557845561967e-07, + "loss": 0.7195, + "mean_token_accuracy": 0.7618420720100403, + "num_tokens": 101371963.0, + "step": 3360 + }, + { + "epoch": 0.27715016079821886, + "grad_norm": 2.2420365810394287, + "learning_rate": 9.233305853256389e-07, + "loss": 0.9601, + "mean_token_accuracy": 0.7117656469345093, + "num_tokens": 101395895.0, + "step": 3361 + }, + { + "epoch": 0.2772326214232704, + "grad_norm": 2.085367441177368, + "learning_rate": 9.23605386095081e-07, + "loss": 1.0428, + "mean_token_accuracy": 0.688142716884613, + "num_tokens": 101425393.0, + "step": 3362 + }, + { + "epoch": 0.27731508204832195, + "grad_norm": 2.059443712234497, + "learning_rate": 9.238801868645232e-07, + "loss": 0.9225, + "mean_token_accuracy": 0.7195301055908203, + "num_tokens": 101456892.0, + "step": 3363 + }, + { + "epoch": 0.27739754267337347, + "grad_norm": 2.00109601020813, + "learning_rate": 9.241549876339654e-07, + "loss": 0.7809, + "mean_token_accuracy": 0.7553586959838867, + "num_tokens": 101485931.0, + "step": 3364 + }, + { + "epoch": 0.277480003298425, + "grad_norm": 1.9718395471572876, + "learning_rate": 9.244297884034074e-07, + "loss": 0.8763, + "mean_token_accuracy": 0.7316547632217407, + "num_tokens": 101517768.0, + "step": 3365 + }, + { + "epoch": 0.27756246392347655, + "grad_norm": 2.228802442550659, + "learning_rate": 9.247045891728497e-07, + "loss": 0.8955, + "mean_token_accuracy": 0.7244762182235718, + "num_tokens": 101542989.0, + "step": 3366 + }, + { + "epoch": 0.27764492454852807, + "grad_norm": 1.8218719959259033, + "learning_rate": 9.249793899422918e-07, + "loss": 0.8254, + "mean_token_accuracy": 0.7389166355133057, + "num_tokens": 101582011.0, + "step": 3367 + }, + { + "epoch": 0.2777273851735796, + "grad_norm": 1.9143002033233643, + "learning_rate": 9.25254190711734e-07, + "loss": 0.7653, + "mean_token_accuracy": 0.7609286308288574, + "num_tokens": 101613017.0, + "step": 3368 + }, + { + "epoch": 0.27780984579863116, + "grad_norm": 1.9964336156845093, + "learning_rate": 9.255289914811761e-07, + "loss": 0.84, + "mean_token_accuracy": 0.7401570081710815, + "num_tokens": 101642350.0, + "step": 3369 + }, + { + "epoch": 0.2778923064236827, + "grad_norm": 2.070493459701538, + "learning_rate": 9.258037922506182e-07, + "loss": 0.9757, + "mean_token_accuracy": 0.7107448577880859, + "num_tokens": 101672230.0, + "step": 3370 + }, + { + "epoch": 0.27797476704873425, + "grad_norm": 2.161222219467163, + "learning_rate": 9.260785930200605e-07, + "loss": 0.8209, + "mean_token_accuracy": 0.7451426982879639, + "num_tokens": 101699182.0, + "step": 3371 + }, + { + "epoch": 0.27805722767378577, + "grad_norm": 1.8785502910614014, + "learning_rate": 9.263533937895026e-07, + "loss": 0.7102, + "mean_token_accuracy": 0.7792341709136963, + "num_tokens": 101730022.0, + "step": 3372 + }, + { + "epoch": 0.2781396882988373, + "grad_norm": 2.3639135360717773, + "learning_rate": 9.266281945589447e-07, + "loss": 0.8367, + "mean_token_accuracy": 0.7466566562652588, + "num_tokens": 101751842.0, + "step": 3373 + }, + { + "epoch": 0.27822214892388886, + "grad_norm": 2.155630588531494, + "learning_rate": 9.269029953283869e-07, + "loss": 0.8148, + "mean_token_accuracy": 0.7401145696640015, + "num_tokens": 101779196.0, + "step": 3374 + }, + { + "epoch": 0.2783046095489404, + "grad_norm": 1.8385827541351318, + "learning_rate": 9.271777960978289e-07, + "loss": 0.7256, + "mean_token_accuracy": 0.7724084854125977, + "num_tokens": 101810881.0, + "step": 3375 + }, + { + "epoch": 0.27838707017399195, + "grad_norm": 1.8392454385757446, + "learning_rate": 9.274525968672712e-07, + "loss": 0.7354, + "mean_token_accuracy": 0.7731077671051025, + "num_tokens": 101847930.0, + "step": 3376 + }, + { + "epoch": 0.27846953079904346, + "grad_norm": 2.069774627685547, + "learning_rate": 9.277273976367134e-07, + "loss": 0.9768, + "mean_token_accuracy": 0.7024157047271729, + "num_tokens": 101879002.0, + "step": 3377 + }, + { + "epoch": 0.278551991424095, + "grad_norm": 1.907010793685913, + "learning_rate": 9.280021984061555e-07, + "loss": 0.7906, + "mean_token_accuracy": 0.7395802140235901, + "num_tokens": 101916419.0, + "step": 3378 + }, + { + "epoch": 0.27863445204914655, + "grad_norm": 2.003038167953491, + "learning_rate": 9.282769991755976e-07, + "loss": 1.0201, + "mean_token_accuracy": 0.6907599568367004, + "num_tokens": 101951449.0, + "step": 3379 + }, + { + "epoch": 0.27871691267419807, + "grad_norm": 1.9157871007919312, + "learning_rate": 9.285517999450399e-07, + "loss": 0.8341, + "mean_token_accuracy": 0.7455794811248779, + "num_tokens": 101986038.0, + "step": 3380 + }, + { + "epoch": 0.2787993732992496, + "grad_norm": 2.329394817352295, + "learning_rate": 9.28826600714482e-07, + "loss": 0.9691, + "mean_token_accuracy": 0.7140483260154724, + "num_tokens": 102013564.0, + "step": 3381 + }, + { + "epoch": 0.27888183392430116, + "grad_norm": 1.9593298435211182, + "learning_rate": 9.291014014839241e-07, + "loss": 0.8509, + "mean_token_accuracy": 0.74046790599823, + "num_tokens": 102045542.0, + "step": 3382 + }, + { + "epoch": 0.2789642945493527, + "grad_norm": 1.9488855600357056, + "learning_rate": 9.293762022533662e-07, + "loss": 0.7953, + "mean_token_accuracy": 0.7549806833267212, + "num_tokens": 102076787.0, + "step": 3383 + }, + { + "epoch": 0.27904675517440425, + "grad_norm": 2.0318124294281006, + "learning_rate": 9.296510030228084e-07, + "loss": 0.7997, + "mean_token_accuracy": 0.7450388669967651, + "num_tokens": 102108061.0, + "step": 3384 + }, + { + "epoch": 0.27912921579945577, + "grad_norm": 1.8765435218811035, + "learning_rate": 9.299258037922507e-07, + "loss": 0.9295, + "mean_token_accuracy": 0.7153228521347046, + "num_tokens": 102143908.0, + "step": 3385 + }, + { + "epoch": 0.2792116764245073, + "grad_norm": 2.140669107437134, + "learning_rate": 9.302006045616927e-07, + "loss": 0.7608, + "mean_token_accuracy": 0.7627434730529785, + "num_tokens": 102173042.0, + "step": 3386 + }, + { + "epoch": 0.27929413704955885, + "grad_norm": 1.9802920818328857, + "learning_rate": 9.304754053311349e-07, + "loss": 0.7287, + "mean_token_accuracy": 0.7724685072898865, + "num_tokens": 102200997.0, + "step": 3387 + }, + { + "epoch": 0.27937659767461037, + "grad_norm": 2.0470235347747803, + "learning_rate": 9.307502061005771e-07, + "loss": 0.8022, + "mean_token_accuracy": 0.7545878291130066, + "num_tokens": 102230806.0, + "step": 3388 + }, + { + "epoch": 0.2794590582996619, + "grad_norm": 2.073193311691284, + "learning_rate": 9.310250068700191e-07, + "loss": 0.9111, + "mean_token_accuracy": 0.7151457667350769, + "num_tokens": 102257894.0, + "step": 3389 + }, + { + "epoch": 0.27954151892471346, + "grad_norm": 1.913663387298584, + "learning_rate": 9.312998076394614e-07, + "loss": 0.8725, + "mean_token_accuracy": 0.7355152368545532, + "num_tokens": 102291222.0, + "step": 3390 + }, + { + "epoch": 0.279623979549765, + "grad_norm": 1.8987704515457153, + "learning_rate": 9.315746084089035e-07, + "loss": 0.7683, + "mean_token_accuracy": 0.7634218335151672, + "num_tokens": 102323135.0, + "step": 3391 + }, + { + "epoch": 0.27970644017481655, + "grad_norm": 2.192244052886963, + "learning_rate": 9.318494091783456e-07, + "loss": 0.9726, + "mean_token_accuracy": 0.7068772912025452, + "num_tokens": 102352688.0, + "step": 3392 + }, + { + "epoch": 0.27978890079986807, + "grad_norm": 2.089430809020996, + "learning_rate": 9.321242099477878e-07, + "loss": 0.7523, + "mean_token_accuracy": 0.7738741040229797, + "num_tokens": 102384379.0, + "step": 3393 + }, + { + "epoch": 0.2798713614249196, + "grad_norm": 1.9759858846664429, + "learning_rate": 9.323990107172299e-07, + "loss": 0.8661, + "mean_token_accuracy": 0.7452719211578369, + "num_tokens": 102413483.0, + "step": 3394 + }, + { + "epoch": 0.27995382204997116, + "grad_norm": 1.841789960861206, + "learning_rate": 9.326738114866722e-07, + "loss": 0.7311, + "mean_token_accuracy": 0.7577238082885742, + "num_tokens": 102444392.0, + "step": 3395 + }, + { + "epoch": 0.2800362826750227, + "grad_norm": 1.8971583843231201, + "learning_rate": 9.329486122561143e-07, + "loss": 0.7156, + "mean_token_accuracy": 0.7669875025749207, + "num_tokens": 102476981.0, + "step": 3396 + }, + { + "epoch": 0.2801187433000742, + "grad_norm": 1.8965916633605957, + "learning_rate": 9.332234130255564e-07, + "loss": 0.8078, + "mean_token_accuracy": 0.7490403652191162, + "num_tokens": 102508443.0, + "step": 3397 + }, + { + "epoch": 0.28020120392512576, + "grad_norm": 2.215663433074951, + "learning_rate": 9.334982137949986e-07, + "loss": 0.8036, + "mean_token_accuracy": 0.746660590171814, + "num_tokens": 102533590.0, + "step": 3398 + }, + { + "epoch": 0.2802836645501773, + "grad_norm": 1.733292579650879, + "learning_rate": 9.337730145644407e-07, + "loss": 0.8119, + "mean_token_accuracy": 0.7379790544509888, + "num_tokens": 102573304.0, + "step": 3399 + }, + { + "epoch": 0.28036612517522885, + "grad_norm": 1.8545690774917603, + "learning_rate": 9.340478153338829e-07, + "loss": 0.7877, + "mean_token_accuracy": 0.7615256309509277, + "num_tokens": 102604815.0, + "step": 3400 + }, + { + "epoch": 0.28044858580028037, + "grad_norm": 2.144550085067749, + "learning_rate": 9.343226161033251e-07, + "loss": 0.8752, + "mean_token_accuracy": 0.7312405109405518, + "num_tokens": 102630518.0, + "step": 3401 + }, + { + "epoch": 0.2805310464253319, + "grad_norm": 2.192601203918457, + "learning_rate": 9.345974168727671e-07, + "loss": 0.7999, + "mean_token_accuracy": 0.7468761205673218, + "num_tokens": 102655615.0, + "step": 3402 + }, + { + "epoch": 0.28061350705038346, + "grad_norm": 1.8708328008651733, + "learning_rate": 9.348722176422093e-07, + "loss": 0.8923, + "mean_token_accuracy": 0.7248790264129639, + "num_tokens": 102690275.0, + "step": 3403 + }, + { + "epoch": 0.280695967675435, + "grad_norm": 1.9030873775482178, + "learning_rate": 9.351470184116516e-07, + "loss": 0.7706, + "mean_token_accuracy": 0.765373945236206, + "num_tokens": 102720470.0, + "step": 3404 + }, + { + "epoch": 0.2807784283004865, + "grad_norm": 2.068902015686035, + "learning_rate": 9.354218191810937e-07, + "loss": 0.8866, + "mean_token_accuracy": 0.7384343147277832, + "num_tokens": 102746786.0, + "step": 3405 + }, + { + "epoch": 0.28086088892553807, + "grad_norm": 2.237548351287842, + "learning_rate": 9.356966199505358e-07, + "loss": 0.9413, + "mean_token_accuracy": 0.7167080640792847, + "num_tokens": 102773249.0, + "step": 3406 + }, + { + "epoch": 0.2809433495505896, + "grad_norm": 2.0277907848358154, + "learning_rate": 9.359714207199779e-07, + "loss": 0.8809, + "mean_token_accuracy": 0.7332684993743896, + "num_tokens": 102802252.0, + "step": 3407 + }, + { + "epoch": 0.28102581017564116, + "grad_norm": 1.8530004024505615, + "learning_rate": 9.362462214894201e-07, + "loss": 0.7947, + "mean_token_accuracy": 0.745830774307251, + "num_tokens": 102836940.0, + "step": 3408 + }, + { + "epoch": 0.2811082708006927, + "grad_norm": 2.1666438579559326, + "learning_rate": 9.365210222588623e-07, + "loss": 0.8793, + "mean_token_accuracy": 0.7255091667175293, + "num_tokens": 102865117.0, + "step": 3409 + }, + { + "epoch": 0.2811907314257442, + "grad_norm": 2.0195508003234863, + "learning_rate": 9.367958230283044e-07, + "loss": 0.9309, + "mean_token_accuracy": 0.7235363721847534, + "num_tokens": 102895490.0, + "step": 3410 + }, + { + "epoch": 0.28127319205079576, + "grad_norm": 2.1878089904785156, + "learning_rate": 9.370706237977466e-07, + "loss": 0.9188, + "mean_token_accuracy": 0.7236096858978271, + "num_tokens": 102920397.0, + "step": 3411 + }, + { + "epoch": 0.2813556526758473, + "grad_norm": 1.8995438814163208, + "learning_rate": 9.373454245671888e-07, + "loss": 0.7601, + "mean_token_accuracy": 0.7706877589225769, + "num_tokens": 102952235.0, + "step": 3412 + }, + { + "epoch": 0.2814381133008988, + "grad_norm": 2.272592544555664, + "learning_rate": 9.376202253366309e-07, + "loss": 0.8177, + "mean_token_accuracy": 0.7422740459442139, + "num_tokens": 102975448.0, + "step": 3413 + }, + { + "epoch": 0.28152057392595037, + "grad_norm": 2.018277168273926, + "learning_rate": 9.378950261060731e-07, + "loss": 0.7999, + "mean_token_accuracy": 0.7539885640144348, + "num_tokens": 103003699.0, + "step": 3414 + }, + { + "epoch": 0.2816030345510019, + "grad_norm": 1.9750571250915527, + "learning_rate": 9.381698268755152e-07, + "loss": 0.6556, + "mean_token_accuracy": 0.787733256816864, + "num_tokens": 103030078.0, + "step": 3415 + }, + { + "epoch": 0.28168549517605346, + "grad_norm": 1.9112244844436646, + "learning_rate": 9.384446276449573e-07, + "loss": 0.7886, + "mean_token_accuracy": 0.7458517551422119, + "num_tokens": 103062949.0, + "step": 3416 + }, + { + "epoch": 0.281767955801105, + "grad_norm": 2.2850499153137207, + "learning_rate": 9.387194284143995e-07, + "loss": 0.8593, + "mean_token_accuracy": 0.7226912975311279, + "num_tokens": 103088827.0, + "step": 3417 + }, + { + "epoch": 0.2818504164261565, + "grad_norm": 1.995815396308899, + "learning_rate": 9.389942291838417e-07, + "loss": 0.9201, + "mean_token_accuracy": 0.7103736400604248, + "num_tokens": 103118906.0, + "step": 3418 + }, + { + "epoch": 0.28193287705120806, + "grad_norm": 1.8778741359710693, + "learning_rate": 9.392690299532839e-07, + "loss": 0.9176, + "mean_token_accuracy": 0.7226558923721313, + "num_tokens": 103155702.0, + "step": 3419 + }, + { + "epoch": 0.2820153376762596, + "grad_norm": 2.3001976013183594, + "learning_rate": 9.39543830722726e-07, + "loss": 0.8865, + "mean_token_accuracy": 0.7239390015602112, + "num_tokens": 103180998.0, + "step": 3420 + }, + { + "epoch": 0.2820977983013111, + "grad_norm": 1.9919737577438354, + "learning_rate": 9.398186314921681e-07, + "loss": 0.7508, + "mean_token_accuracy": 0.7752147912979126, + "num_tokens": 103208961.0, + "step": 3421 + }, + { + "epoch": 0.28218025892636267, + "grad_norm": 1.8720450401306152, + "learning_rate": 9.400934322616103e-07, + "loss": 0.7897, + "mean_token_accuracy": 0.7601596713066101, + "num_tokens": 103239236.0, + "step": 3422 + }, + { + "epoch": 0.2822627195514142, + "grad_norm": 2.1112310886383057, + "learning_rate": 9.403682330310524e-07, + "loss": 0.7592, + "mean_token_accuracy": 0.7626118659973145, + "num_tokens": 103270282.0, + "step": 3423 + }, + { + "epoch": 0.28234518017646576, + "grad_norm": 2.1413156986236572, + "learning_rate": 9.406430338004946e-07, + "loss": 0.9591, + "mean_token_accuracy": 0.7072203755378723, + "num_tokens": 103297517.0, + "step": 3424 + }, + { + "epoch": 0.2824276408015173, + "grad_norm": 1.8819235563278198, + "learning_rate": 9.409178345699368e-07, + "loss": 0.8197, + "mean_token_accuracy": 0.7617201805114746, + "num_tokens": 103331283.0, + "step": 3425 + }, + { + "epoch": 0.2825101014265688, + "grad_norm": 1.7571736574172974, + "learning_rate": 9.411926353393788e-07, + "loss": 0.77, + "mean_token_accuracy": 0.7612042427062988, + "num_tokens": 103368704.0, + "step": 3426 + }, + { + "epoch": 0.28259256205162037, + "grad_norm": 2.173692464828491, + "learning_rate": 9.414674361088211e-07, + "loss": 0.8224, + "mean_token_accuracy": 0.7475183010101318, + "num_tokens": 103396347.0, + "step": 3427 + }, + { + "epoch": 0.2826750226766719, + "grad_norm": 2.1664879322052, + "learning_rate": 9.417422368782633e-07, + "loss": 0.7703, + "mean_token_accuracy": 0.7584713697433472, + "num_tokens": 103422253.0, + "step": 3428 + }, + { + "epoch": 0.2827574833017234, + "grad_norm": 2.1531763076782227, + "learning_rate": 9.420170376477054e-07, + "loss": 1.0391, + "mean_token_accuracy": 0.6875049471855164, + "num_tokens": 103449696.0, + "step": 3429 + }, + { + "epoch": 0.282839943926775, + "grad_norm": 2.0089099407196045, + "learning_rate": 9.422918384171475e-07, + "loss": 0.7108, + "mean_token_accuracy": 0.7797251343727112, + "num_tokens": 103479782.0, + "step": 3430 + }, + { + "epoch": 0.2829224045518265, + "grad_norm": 2.0664124488830566, + "learning_rate": 9.425666391865896e-07, + "loss": 0.9789, + "mean_token_accuracy": 0.6999763250350952, + "num_tokens": 103508714.0, + "step": 3431 + }, + { + "epoch": 0.28300486517687806, + "grad_norm": 1.8697116374969482, + "learning_rate": 9.428414399560319e-07, + "loss": 0.7647, + "mean_token_accuracy": 0.7546308040618896, + "num_tokens": 103540225.0, + "step": 3432 + }, + { + "epoch": 0.2830873258019296, + "grad_norm": 1.9359345436096191, + "learning_rate": 9.43116240725474e-07, + "loss": 0.8939, + "mean_token_accuracy": 0.7245751023292542, + "num_tokens": 103571264.0, + "step": 3433 + }, + { + "epoch": 0.2831697864269811, + "grad_norm": 1.895424246788025, + "learning_rate": 9.433910414949161e-07, + "loss": 0.789, + "mean_token_accuracy": 0.7574657201766968, + "num_tokens": 103605351.0, + "step": 3434 + }, + { + "epoch": 0.28325224705203267, + "grad_norm": 1.7335301637649536, + "learning_rate": 9.436658422643583e-07, + "loss": 0.982, + "mean_token_accuracy": 0.7134857773780823, + "num_tokens": 103643196.0, + "step": 3435 + }, + { + "epoch": 0.2833347076770842, + "grad_norm": 2.1107659339904785, + "learning_rate": 9.439406430338004e-07, + "loss": 0.8923, + "mean_token_accuracy": 0.7260717749595642, + "num_tokens": 103672483.0, + "step": 3436 + }, + { + "epoch": 0.2834171683021357, + "grad_norm": 2.1720633506774902, + "learning_rate": 9.442154438032426e-07, + "loss": 0.8113, + "mean_token_accuracy": 0.744574248790741, + "num_tokens": 103699806.0, + "step": 3437 + }, + { + "epoch": 0.2834996289271873, + "grad_norm": 1.771225094795227, + "learning_rate": 9.444902445726848e-07, + "loss": 0.713, + "mean_token_accuracy": 0.7755175828933716, + "num_tokens": 103733188.0, + "step": 3438 + }, + { + "epoch": 0.2835820895522388, + "grad_norm": 1.9224004745483398, + "learning_rate": 9.447650453421269e-07, + "loss": 0.8441, + "mean_token_accuracy": 0.742536187171936, + "num_tokens": 103764887.0, + "step": 3439 + }, + { + "epoch": 0.28366455017729036, + "grad_norm": 2.0804622173309326, + "learning_rate": 9.45039846111569e-07, + "loss": 0.8198, + "mean_token_accuracy": 0.7387181520462036, + "num_tokens": 103793285.0, + "step": 3440 + }, + { + "epoch": 0.2837470108023419, + "grad_norm": 1.7528808116912842, + "learning_rate": 9.453146468810113e-07, + "loss": 0.7444, + "mean_token_accuracy": 0.7679877281188965, + "num_tokens": 103832912.0, + "step": 3441 + }, + { + "epoch": 0.2838294714273934, + "grad_norm": 1.8460496664047241, + "learning_rate": 9.455894476504534e-07, + "loss": 0.7318, + "mean_token_accuracy": 0.7723982334136963, + "num_tokens": 103864142.0, + "step": 3442 + }, + { + "epoch": 0.28391193205244497, + "grad_norm": 2.201979637145996, + "learning_rate": 9.458642484198955e-07, + "loss": 0.8877, + "mean_token_accuracy": 0.7204164862632751, + "num_tokens": 103891753.0, + "step": 3443 + }, + { + "epoch": 0.2839943926774965, + "grad_norm": 1.788680076599121, + "learning_rate": 9.461390491893377e-07, + "loss": 0.8122, + "mean_token_accuracy": 0.7469525933265686, + "num_tokens": 103926227.0, + "step": 3444 + }, + { + "epoch": 0.28407685330254806, + "grad_norm": 1.877071738243103, + "learning_rate": 9.464138499587798e-07, + "loss": 0.8331, + "mean_token_accuracy": 0.7446708679199219, + "num_tokens": 103958049.0, + "step": 3445 + }, + { + "epoch": 0.2841593139275996, + "grad_norm": 2.332493543624878, + "learning_rate": 9.466886507282221e-07, + "loss": 0.9245, + "mean_token_accuracy": 0.7276257276535034, + "num_tokens": 103986412.0, + "step": 3446 + }, + { + "epoch": 0.2842417745526511, + "grad_norm": 1.7740821838378906, + "learning_rate": 9.469634514976641e-07, + "loss": 0.801, + "mean_token_accuracy": 0.7500151991844177, + "num_tokens": 104026468.0, + "step": 3447 + }, + { + "epoch": 0.28432423517770267, + "grad_norm": 2.0270674228668213, + "learning_rate": 9.472382522671063e-07, + "loss": 0.8188, + "mean_token_accuracy": 0.7656428813934326, + "num_tokens": 104054352.0, + "step": 3448 + }, + { + "epoch": 0.2844066958027542, + "grad_norm": 2.0569350719451904, + "learning_rate": 9.475130530365485e-07, + "loss": 0.8022, + "mean_token_accuracy": 0.742710292339325, + "num_tokens": 104082447.0, + "step": 3449 + }, + { + "epoch": 0.2844891564278057, + "grad_norm": 1.9567502737045288, + "learning_rate": 9.477878538059905e-07, + "loss": 0.7468, + "mean_token_accuracy": 0.7648622989654541, + "num_tokens": 104112898.0, + "step": 3450 + }, + { + "epoch": 0.2845716170528573, + "grad_norm": 1.9633539915084839, + "learning_rate": 9.480626545754328e-07, + "loss": 0.7978, + "mean_token_accuracy": 0.755652666091919, + "num_tokens": 104143661.0, + "step": 3451 + }, + { + "epoch": 0.2846540776779088, + "grad_norm": 1.8108588457107544, + "learning_rate": 9.48337455344875e-07, + "loss": 0.7647, + "mean_token_accuracy": 0.7655800580978394, + "num_tokens": 104177595.0, + "step": 3452 + }, + { + "epoch": 0.28473653830296036, + "grad_norm": 2.0643506050109863, + "learning_rate": 9.48612256114317e-07, + "loss": 0.844, + "mean_token_accuracy": 0.7425796389579773, + "num_tokens": 104207053.0, + "step": 3453 + }, + { + "epoch": 0.2848189989280119, + "grad_norm": 1.9849424362182617, + "learning_rate": 9.488870568837592e-07, + "loss": 0.7355, + "mean_token_accuracy": 0.762323260307312, + "num_tokens": 104237876.0, + "step": 3454 + }, + { + "epoch": 0.2849014595530634, + "grad_norm": 2.114447593688965, + "learning_rate": 9.491618576532013e-07, + "loss": 1.0171, + "mean_token_accuracy": 0.6957997679710388, + "num_tokens": 104270575.0, + "step": 3455 + }, + { + "epoch": 0.28498392017811497, + "grad_norm": 1.9653904438018799, + "learning_rate": 9.494366584226436e-07, + "loss": 0.8537, + "mean_token_accuracy": 0.7201881408691406, + "num_tokens": 104300235.0, + "step": 3456 + }, + { + "epoch": 0.2850663808031665, + "grad_norm": 1.8945956230163574, + "learning_rate": 9.497114591920857e-07, + "loss": 0.7183, + "mean_token_accuracy": 0.7738735675811768, + "num_tokens": 104334443.0, + "step": 3457 + }, + { + "epoch": 0.285148841428218, + "grad_norm": 2.3082284927368164, + "learning_rate": 9.499862599615278e-07, + "loss": 0.7943, + "mean_token_accuracy": 0.754926323890686, + "num_tokens": 104361795.0, + "step": 3458 + }, + { + "epoch": 0.2852313020532696, + "grad_norm": 2.191685676574707, + "learning_rate": 9.5026106073097e-07, + "loss": 0.8059, + "mean_token_accuracy": 0.736355185508728, + "num_tokens": 104386123.0, + "step": 3459 + }, + { + "epoch": 0.2853137626783211, + "grad_norm": 1.8466300964355469, + "learning_rate": 9.505358615004122e-07, + "loss": 0.7466, + "mean_token_accuracy": 0.7630093693733215, + "num_tokens": 104418432.0, + "step": 3460 + }, + { + "epoch": 0.28539622330337266, + "grad_norm": 1.94961678981781, + "learning_rate": 9.508106622698543e-07, + "loss": 0.8191, + "mean_token_accuracy": 0.749405026435852, + "num_tokens": 104453185.0, + "step": 3461 + }, + { + "epoch": 0.2854786839284242, + "grad_norm": 2.4562172889709473, + "learning_rate": 9.510854630392965e-07, + "loss": 0.799, + "mean_token_accuracy": 0.7487127780914307, + "num_tokens": 104474796.0, + "step": 3462 + }, + { + "epoch": 0.2855611445534757, + "grad_norm": 2.160019874572754, + "learning_rate": 9.513602638087386e-07, + "loss": 0.771, + "mean_token_accuracy": 0.7521620988845825, + "num_tokens": 104504728.0, + "step": 3463 + }, + { + "epoch": 0.28564360517852727, + "grad_norm": 2.145951986312866, + "learning_rate": 9.516350645781807e-07, + "loss": 0.811, + "mean_token_accuracy": 0.749767541885376, + "num_tokens": 104531216.0, + "step": 3464 + }, + { + "epoch": 0.2857260658035788, + "grad_norm": 2.1406314373016357, + "learning_rate": 9.51909865347623e-07, + "loss": 0.8641, + "mean_token_accuracy": 0.7295753955841064, + "num_tokens": 104558250.0, + "step": 3465 + }, + { + "epoch": 0.2858085264286303, + "grad_norm": 2.0548136234283447, + "learning_rate": 9.521846661170651e-07, + "loss": 0.8163, + "mean_token_accuracy": 0.7442847490310669, + "num_tokens": 104583353.0, + "step": 3466 + }, + { + "epoch": 0.2858909870536819, + "grad_norm": 1.9489409923553467, + "learning_rate": 9.524594668865072e-07, + "loss": 0.8075, + "mean_token_accuracy": 0.738213062286377, + "num_tokens": 104619826.0, + "step": 3467 + }, + { + "epoch": 0.2859734476787334, + "grad_norm": 2.2219319343566895, + "learning_rate": 9.527342676559494e-07, + "loss": 0.8311, + "mean_token_accuracy": 0.745000958442688, + "num_tokens": 104644783.0, + "step": 3468 + }, + { + "epoch": 0.28605590830378497, + "grad_norm": 1.8154704570770264, + "learning_rate": 9.530090684253915e-07, + "loss": 0.8062, + "mean_token_accuracy": 0.7461264729499817, + "num_tokens": 104679081.0, + "step": 3469 + }, + { + "epoch": 0.2861383689288365, + "grad_norm": 1.8817236423492432, + "learning_rate": 9.532838691948338e-07, + "loss": 0.9796, + "mean_token_accuracy": 0.7056002616882324, + "num_tokens": 104716423.0, + "step": 3470 + }, + { + "epoch": 0.286220829553888, + "grad_norm": 1.9213566780090332, + "learning_rate": 9.535586699642758e-07, + "loss": 0.627, + "mean_token_accuracy": 0.7962247729301453, + "num_tokens": 104744545.0, + "step": 3471 + }, + { + "epoch": 0.2863032901789396, + "grad_norm": 2.0326502323150635, + "learning_rate": 9.53833470733718e-07, + "loss": 0.7303, + "mean_token_accuracy": 0.7531157732009888, + "num_tokens": 104777434.0, + "step": 3472 + }, + { + "epoch": 0.2863857508039911, + "grad_norm": 1.831417202949524, + "learning_rate": 9.541082715031602e-07, + "loss": 0.9022, + "mean_token_accuracy": 0.7326239347457886, + "num_tokens": 104813200.0, + "step": 3473 + }, + { + "epoch": 0.2864682114290426, + "grad_norm": 1.9463986158370972, + "learning_rate": 9.543830722726022e-07, + "loss": 0.7869, + "mean_token_accuracy": 0.7474774122238159, + "num_tokens": 104844847.0, + "step": 3474 + }, + { + "epoch": 0.2865506720540942, + "grad_norm": 2.1393682956695557, + "learning_rate": 9.546578730420445e-07, + "loss": 0.8635, + "mean_token_accuracy": 0.7346956729888916, + "num_tokens": 104873025.0, + "step": 3475 + }, + { + "epoch": 0.2866331326791457, + "grad_norm": 1.7235467433929443, + "learning_rate": 9.549326738114868e-07, + "loss": 0.8414, + "mean_token_accuracy": 0.7418675422668457, + "num_tokens": 104912024.0, + "step": 3476 + }, + { + "epoch": 0.28671559330419727, + "grad_norm": 1.9067851305007935, + "learning_rate": 9.552074745809288e-07, + "loss": 0.8428, + "mean_token_accuracy": 0.7461115121841431, + "num_tokens": 104942053.0, + "step": 3477 + }, + { + "epoch": 0.2867980539292488, + "grad_norm": 1.9886629581451416, + "learning_rate": 9.55482275350371e-07, + "loss": 0.8338, + "mean_token_accuracy": 0.7446297407150269, + "num_tokens": 104973320.0, + "step": 3478 + }, + { + "epoch": 0.2868805145543003, + "grad_norm": 2.4546313285827637, + "learning_rate": 9.55757076119813e-07, + "loss": 0.9838, + "mean_token_accuracy": 0.6998498439788818, + "num_tokens": 104996708.0, + "step": 3479 + }, + { + "epoch": 0.2869629751793519, + "grad_norm": 2.22269868850708, + "learning_rate": 9.560318768892553e-07, + "loss": 0.8596, + "mean_token_accuracy": 0.7410118579864502, + "num_tokens": 105024724.0, + "step": 3480 + }, + { + "epoch": 0.2870454358044034, + "grad_norm": 2.0097720623016357, + "learning_rate": 9.563066776586975e-07, + "loss": 0.8362, + "mean_token_accuracy": 0.7311499118804932, + "num_tokens": 105054348.0, + "step": 3481 + }, + { + "epoch": 0.2871278964294549, + "grad_norm": 1.9088574647903442, + "learning_rate": 9.565814784281396e-07, + "loss": 0.7276, + "mean_token_accuracy": 0.7656928300857544, + "num_tokens": 105084691.0, + "step": 3482 + }, + { + "epoch": 0.2872103570545065, + "grad_norm": 2.0472252368927, + "learning_rate": 9.568562791975817e-07, + "loss": 0.9131, + "mean_token_accuracy": 0.7228103876113892, + "num_tokens": 105115266.0, + "step": 3483 + }, + { + "epoch": 0.287292817679558, + "grad_norm": 1.8814424276351929, + "learning_rate": 9.57131079967024e-07, + "loss": 0.7297, + "mean_token_accuracy": 0.7766274809837341, + "num_tokens": 105145883.0, + "step": 3484 + }, + { + "epoch": 0.28737527830460957, + "grad_norm": 2.0939059257507324, + "learning_rate": 9.57405880736466e-07, + "loss": 0.7864, + "mean_token_accuracy": 0.7423545122146606, + "num_tokens": 105175302.0, + "step": 3485 + }, + { + "epoch": 0.2874577389296611, + "grad_norm": 1.9045723676681519, + "learning_rate": 9.576806815059083e-07, + "loss": 0.7859, + "mean_token_accuracy": 0.7477977275848389, + "num_tokens": 105207983.0, + "step": 3486 + }, + { + "epoch": 0.2875401995547126, + "grad_norm": 1.9000849723815918, + "learning_rate": 9.579554822753504e-07, + "loss": 0.8893, + "mean_token_accuracy": 0.7179543972015381, + "num_tokens": 105239998.0, + "step": 3487 + }, + { + "epoch": 0.2876226601797642, + "grad_norm": 2.2739813327789307, + "learning_rate": 9.582302830447924e-07, + "loss": 0.6758, + "mean_token_accuracy": 0.7830590605735779, + "num_tokens": 105261852.0, + "step": 3488 + }, + { + "epoch": 0.2877051208048157, + "grad_norm": 1.8852896690368652, + "learning_rate": 9.585050838142347e-07, + "loss": 0.8017, + "mean_token_accuracy": 0.747856855392456, + "num_tokens": 105295147.0, + "step": 3489 + }, + { + "epoch": 0.2877875814298672, + "grad_norm": 2.162334442138672, + "learning_rate": 9.587798845836768e-07, + "loss": 0.7314, + "mean_token_accuracy": 0.7573023438453674, + "num_tokens": 105319986.0, + "step": 3490 + }, + { + "epoch": 0.2878700420549188, + "grad_norm": 2.0623247623443604, + "learning_rate": 9.59054685353119e-07, + "loss": 0.8924, + "mean_token_accuracy": 0.7324053049087524, + "num_tokens": 105353340.0, + "step": 3491 + }, + { + "epoch": 0.2879525026799703, + "grad_norm": 2.067582130432129, + "learning_rate": 9.59329486122561e-07, + "loss": 0.9617, + "mean_token_accuracy": 0.7037709355354309, + "num_tokens": 105384277.0, + "step": 3492 + }, + { + "epoch": 0.2880349633050219, + "grad_norm": 1.9091891050338745, + "learning_rate": 9.596042868920032e-07, + "loss": 0.8728, + "mean_token_accuracy": 0.727238655090332, + "num_tokens": 105420430.0, + "step": 3493 + }, + { + "epoch": 0.2881174239300734, + "grad_norm": 1.9371625185012817, + "learning_rate": 9.598790876614454e-07, + "loss": 0.9258, + "mean_token_accuracy": 0.7242939472198486, + "num_tokens": 105449913.0, + "step": 3494 + }, + { + "epoch": 0.2881998845551249, + "grad_norm": 2.219723701477051, + "learning_rate": 9.601538884308875e-07, + "loss": 0.9016, + "mean_token_accuracy": 0.7215821743011475, + "num_tokens": 105475825.0, + "step": 3495 + }, + { + "epoch": 0.2882823451801765, + "grad_norm": 1.860459804534912, + "learning_rate": 9.604286892003298e-07, + "loss": 0.8545, + "mean_token_accuracy": 0.741461455821991, + "num_tokens": 105510446.0, + "step": 3496 + }, + { + "epoch": 0.288364805805228, + "grad_norm": 2.3386664390563965, + "learning_rate": 9.607034899697719e-07, + "loss": 0.9825, + "mean_token_accuracy": 0.7158508896827698, + "num_tokens": 105533382.0, + "step": 3497 + }, + { + "epoch": 0.2884472664302795, + "grad_norm": 2.082653760910034, + "learning_rate": 9.60978290739214e-07, + "loss": 0.8629, + "mean_token_accuracy": 0.7334731221199036, + "num_tokens": 105563668.0, + "step": 3498 + }, + { + "epoch": 0.2885297270553311, + "grad_norm": 2.139597177505493, + "learning_rate": 9.612530915086562e-07, + "loss": 0.8133, + "mean_token_accuracy": 0.7331383228302002, + "num_tokens": 105590677.0, + "step": 3499 + }, + { + "epoch": 0.2886121876803826, + "grad_norm": 1.9755008220672607, + "learning_rate": 9.615278922780985e-07, + "loss": 0.8727, + "mean_token_accuracy": 0.7345382571220398, + "num_tokens": 105622995.0, + "step": 3500 + }, + { + "epoch": 0.2886946483054342, + "grad_norm": 2.359290599822998, + "learning_rate": 9.618026930475405e-07, + "loss": 0.9385, + "mean_token_accuracy": 0.7120428681373596, + "num_tokens": 105649704.0, + "step": 3501 + }, + { + "epoch": 0.2887771089304857, + "grad_norm": 1.9571683406829834, + "learning_rate": 9.620774938169826e-07, + "loss": 0.8444, + "mean_token_accuracy": 0.7385286688804626, + "num_tokens": 105679438.0, + "step": 3502 + }, + { + "epoch": 0.2888595695555372, + "grad_norm": 2.399341344833374, + "learning_rate": 9.623522945864247e-07, + "loss": 0.8969, + "mean_token_accuracy": 0.7255152463912964, + "num_tokens": 105702950.0, + "step": 3503 + }, + { + "epoch": 0.2889420301805888, + "grad_norm": 1.899446964263916, + "learning_rate": 9.62627095355867e-07, + "loss": 0.7823, + "mean_token_accuracy": 0.7555561065673828, + "num_tokens": 105733956.0, + "step": 3504 + }, + { + "epoch": 0.2890244908056403, + "grad_norm": 1.9568355083465576, + "learning_rate": 9.629018961253092e-07, + "loss": 0.908, + "mean_token_accuracy": 0.7173283100128174, + "num_tokens": 105765942.0, + "step": 3505 + }, + { + "epoch": 0.28910695143069187, + "grad_norm": 2.2626235485076904, + "learning_rate": 9.631766968947513e-07, + "loss": 0.6514, + "mean_token_accuracy": 0.7883256673812866, + "num_tokens": 105788158.0, + "step": 3506 + }, + { + "epoch": 0.2891894120557434, + "grad_norm": 2.082063674926758, + "learning_rate": 9.634514976641934e-07, + "loss": 0.9125, + "mean_token_accuracy": 0.7342755794525146, + "num_tokens": 105817412.0, + "step": 3507 + }, + { + "epoch": 0.2892718726807949, + "grad_norm": 2.067624568939209, + "learning_rate": 9.637262984336356e-07, + "loss": 0.9388, + "mean_token_accuracy": 0.7108260989189148, + "num_tokens": 105844556.0, + "step": 3508 + }, + { + "epoch": 0.2893543333058465, + "grad_norm": 1.8989248275756836, + "learning_rate": 9.640010992030777e-07, + "loss": 0.8093, + "mean_token_accuracy": 0.7487156391143799, + "num_tokens": 105877877.0, + "step": 3509 + }, + { + "epoch": 0.289436793930898, + "grad_norm": 2.0970652103424072, + "learning_rate": 9.6427589997252e-07, + "loss": 0.8988, + "mean_token_accuracy": 0.7248483896255493, + "num_tokens": 105907861.0, + "step": 3510 + }, + { + "epoch": 0.2895192545559495, + "grad_norm": 1.850939393043518, + "learning_rate": 9.64550700741962e-07, + "loss": 0.8682, + "mean_token_accuracy": 0.7471941113471985, + "num_tokens": 105943483.0, + "step": 3511 + }, + { + "epoch": 0.2896017151810011, + "grad_norm": 2.297703504562378, + "learning_rate": 9.648255015114041e-07, + "loss": 0.8994, + "mean_token_accuracy": 0.7271174192428589, + "num_tokens": 105967279.0, + "step": 3512 + }, + { + "epoch": 0.2896841758060526, + "grad_norm": 1.9293698072433472, + "learning_rate": 9.651003022808464e-07, + "loss": 0.7914, + "mean_token_accuracy": 0.7482333779335022, + "num_tokens": 105999192.0, + "step": 3513 + }, + { + "epoch": 0.2897666364311042, + "grad_norm": 1.8606679439544678, + "learning_rate": 9.653751030502885e-07, + "loss": 0.9019, + "mean_token_accuracy": 0.7195981740951538, + "num_tokens": 106033127.0, + "step": 3514 + }, + { + "epoch": 0.2898490970561557, + "grad_norm": 2.1847434043884277, + "learning_rate": 9.656499038197307e-07, + "loss": 0.8656, + "mean_token_accuracy": 0.7358213067054749, + "num_tokens": 106057839.0, + "step": 3515 + }, + { + "epoch": 0.2899315576812072, + "grad_norm": 2.043161630630493, + "learning_rate": 9.659247045891728e-07, + "loss": 0.8878, + "mean_token_accuracy": 0.7281414270401001, + "num_tokens": 106088476.0, + "step": 3516 + }, + { + "epoch": 0.2900140183062588, + "grad_norm": 1.8871300220489502, + "learning_rate": 9.661995053586149e-07, + "loss": 0.7459, + "mean_token_accuracy": 0.7637680768966675, + "num_tokens": 106120845.0, + "step": 3517 + }, + { + "epoch": 0.2900964789313103, + "grad_norm": 2.012073278427124, + "learning_rate": 9.664743061280571e-07, + "loss": 0.8462, + "mean_token_accuracy": 0.7407709956169128, + "num_tokens": 106153348.0, + "step": 3518 + }, + { + "epoch": 0.2901789395563618, + "grad_norm": 2.2622580528259277, + "learning_rate": 9.667491068974992e-07, + "loss": 0.9989, + "mean_token_accuracy": 0.7002020478248596, + "num_tokens": 106180600.0, + "step": 3519 + }, + { + "epoch": 0.2902614001814134, + "grad_norm": 2.1646311283111572, + "learning_rate": 9.670239076669415e-07, + "loss": 0.6646, + "mean_token_accuracy": 0.7797020077705383, + "num_tokens": 106203850.0, + "step": 3520 + }, + { + "epoch": 0.2903438608064649, + "grad_norm": 2.2649641036987305, + "learning_rate": 9.672987084363836e-07, + "loss": 0.9271, + "mean_token_accuracy": 0.7161029577255249, + "num_tokens": 106231315.0, + "step": 3521 + }, + { + "epoch": 0.2904263214315165, + "grad_norm": 2.0318984985351562, + "learning_rate": 9.675735092058256e-07, + "loss": 0.923, + "mean_token_accuracy": 0.7263293266296387, + "num_tokens": 106262272.0, + "step": 3522 + }, + { + "epoch": 0.290508782056568, + "grad_norm": 2.014782190322876, + "learning_rate": 9.678483099752679e-07, + "loss": 0.8961, + "mean_token_accuracy": 0.7413232326507568, + "num_tokens": 106292985.0, + "step": 3523 + }, + { + "epoch": 0.2905912426816195, + "grad_norm": 1.985487461090088, + "learning_rate": 9.681231107447102e-07, + "loss": 0.94, + "mean_token_accuracy": 0.7141196727752686, + "num_tokens": 106322687.0, + "step": 3524 + }, + { + "epoch": 0.2906737033066711, + "grad_norm": 1.8942961692810059, + "learning_rate": 9.683979115141522e-07, + "loss": 0.8442, + "mean_token_accuracy": 0.7229042053222656, + "num_tokens": 106355980.0, + "step": 3525 + }, + { + "epoch": 0.2907561639317226, + "grad_norm": 2.281071901321411, + "learning_rate": 9.686727122835943e-07, + "loss": 0.9157, + "mean_token_accuracy": 0.7284045219421387, + "num_tokens": 106380666.0, + "step": 3526 + }, + { + "epoch": 0.2908386245567741, + "grad_norm": 2.2643463611602783, + "learning_rate": 9.689475130530364e-07, + "loss": 0.8713, + "mean_token_accuracy": 0.7181956768035889, + "num_tokens": 106403835.0, + "step": 3527 + }, + { + "epoch": 0.2909210851818257, + "grad_norm": 1.9205636978149414, + "learning_rate": 9.692223138224786e-07, + "loss": 0.6932, + "mean_token_accuracy": 0.7708650827407837, + "num_tokens": 106434009.0, + "step": 3528 + }, + { + "epoch": 0.2910035458068772, + "grad_norm": 2.0924713611602783, + "learning_rate": 9.69497114591921e-07, + "loss": 0.9863, + "mean_token_accuracy": 0.6968698501586914, + "num_tokens": 106467144.0, + "step": 3529 + }, + { + "epoch": 0.2910860064319288, + "grad_norm": 1.9818600416183472, + "learning_rate": 9.69771915361363e-07, + "loss": 0.7744, + "mean_token_accuracy": 0.7521306276321411, + "num_tokens": 106498698.0, + "step": 3530 + }, + { + "epoch": 0.2911684670569803, + "grad_norm": 1.9249128103256226, + "learning_rate": 9.70046716130805e-07, + "loss": 0.8601, + "mean_token_accuracy": 0.7312670946121216, + "num_tokens": 106531359.0, + "step": 3531 + }, + { + "epoch": 0.2912509276820318, + "grad_norm": 1.789476752281189, + "learning_rate": 9.703215169002473e-07, + "loss": 0.8051, + "mean_token_accuracy": 0.7512993216514587, + "num_tokens": 106563894.0, + "step": 3532 + }, + { + "epoch": 0.2913333883070834, + "grad_norm": 2.044813394546509, + "learning_rate": 9.705963176696894e-07, + "loss": 0.8041, + "mean_token_accuracy": 0.7529511451721191, + "num_tokens": 106591504.0, + "step": 3533 + }, + { + "epoch": 0.2914158489321349, + "grad_norm": 2.1441490650177, + "learning_rate": 9.708711184391317e-07, + "loss": 0.8753, + "mean_token_accuracy": 0.7287518978118896, + "num_tokens": 106617129.0, + "step": 3534 + }, + { + "epoch": 0.2914983095571864, + "grad_norm": 1.9883077144622803, + "learning_rate": 9.711459192085737e-07, + "loss": 0.8051, + "mean_token_accuracy": 0.7430095076560974, + "num_tokens": 106645552.0, + "step": 3535 + }, + { + "epoch": 0.291580770182238, + "grad_norm": 1.8874882459640503, + "learning_rate": 9.714207199780158e-07, + "loss": 0.8517, + "mean_token_accuracy": 0.7398183345794678, + "num_tokens": 106678126.0, + "step": 3536 + }, + { + "epoch": 0.2916632308072895, + "grad_norm": 2.0140185356140137, + "learning_rate": 9.71695520747458e-07, + "loss": 0.8953, + "mean_token_accuracy": 0.7221956253051758, + "num_tokens": 106708203.0, + "step": 3537 + }, + { + "epoch": 0.2917456914323411, + "grad_norm": 2.3968985080718994, + "learning_rate": 9.719703215169001e-07, + "loss": 0.8283, + "mean_token_accuracy": 0.7433472871780396, + "num_tokens": 106730606.0, + "step": 3538 + }, + { + "epoch": 0.2918281520573926, + "grad_norm": 2.1645829677581787, + "learning_rate": 9.722451222863424e-07, + "loss": 0.8725, + "mean_token_accuracy": 0.7259324789047241, + "num_tokens": 106756146.0, + "step": 3539 + }, + { + "epoch": 0.2919106126824441, + "grad_norm": 1.942824125289917, + "learning_rate": 9.725199230557845e-07, + "loss": 0.6875, + "mean_token_accuracy": 0.7606462240219116, + "num_tokens": 106786327.0, + "step": 3540 + }, + { + "epoch": 0.2919930733074957, + "grad_norm": 2.0894737243652344, + "learning_rate": 9.727947238252266e-07, + "loss": 0.77, + "mean_token_accuracy": 0.7450107336044312, + "num_tokens": 106814188.0, + "step": 3541 + }, + { + "epoch": 0.2920755339325472, + "grad_norm": 1.973772644996643, + "learning_rate": 9.730695245946688e-07, + "loss": 0.6734, + "mean_token_accuracy": 0.7813385725021362, + "num_tokens": 106842133.0, + "step": 3542 + }, + { + "epoch": 0.2921579945575987, + "grad_norm": 2.2085635662078857, + "learning_rate": 9.73344325364111e-07, + "loss": 0.8827, + "mean_token_accuracy": 0.7263038158416748, + "num_tokens": 106867184.0, + "step": 3543 + }, + { + "epoch": 0.2922404551826503, + "grad_norm": 2.312108039855957, + "learning_rate": 9.736191261335532e-07, + "loss": 0.808, + "mean_token_accuracy": 0.7420276403427124, + "num_tokens": 106888843.0, + "step": 3544 + }, + { + "epoch": 0.2923229158077018, + "grad_norm": 2.083799123764038, + "learning_rate": 9.738939269029952e-07, + "loss": 0.8509, + "mean_token_accuracy": 0.732949435710907, + "num_tokens": 106917736.0, + "step": 3545 + }, + { + "epoch": 0.2924053764327534, + "grad_norm": 1.8753029108047485, + "learning_rate": 9.741687276724373e-07, + "loss": 0.7789, + "mean_token_accuracy": 0.7574050426483154, + "num_tokens": 106951731.0, + "step": 3546 + }, + { + "epoch": 0.2924878370578049, + "grad_norm": 2.2182743549346924, + "learning_rate": 9.744435284418796e-07, + "loss": 0.884, + "mean_token_accuracy": 0.7262222766876221, + "num_tokens": 106977215.0, + "step": 3547 + }, + { + "epoch": 0.2925702976828564, + "grad_norm": 2.1482913494110107, + "learning_rate": 9.747183292113219e-07, + "loss": 0.8645, + "mean_token_accuracy": 0.7243857383728027, + "num_tokens": 107001960.0, + "step": 3548 + }, + { + "epoch": 0.292652758307908, + "grad_norm": 1.9366905689239502, + "learning_rate": 9.74993129980764e-07, + "loss": 0.7994, + "mean_token_accuracy": 0.7535085678100586, + "num_tokens": 107034498.0, + "step": 3549 + }, + { + "epoch": 0.2927352189329595, + "grad_norm": 2.155486822128296, + "learning_rate": 9.75267930750206e-07, + "loss": 0.8629, + "mean_token_accuracy": 0.7298109531402588, + "num_tokens": 107065134.0, + "step": 3550 + }, + { + "epoch": 0.292817679558011, + "grad_norm": 1.9476730823516846, + "learning_rate": 9.75542731519648e-07, + "loss": 0.7929, + "mean_token_accuracy": 0.7399736642837524, + "num_tokens": 107094405.0, + "step": 3551 + }, + { + "epoch": 0.2929001401830626, + "grad_norm": 2.219102621078491, + "learning_rate": 9.758175322890903e-07, + "loss": 0.8348, + "mean_token_accuracy": 0.7347302436828613, + "num_tokens": 107120100.0, + "step": 3552 + }, + { + "epoch": 0.2929826008081141, + "grad_norm": 1.9909672737121582, + "learning_rate": 9.760923330585326e-07, + "loss": 0.7831, + "mean_token_accuracy": 0.7562850713729858, + "num_tokens": 107149437.0, + "step": 3553 + }, + { + "epoch": 0.2930650614331657, + "grad_norm": 2.1861653327941895, + "learning_rate": 9.763671338279747e-07, + "loss": 0.9138, + "mean_token_accuracy": 0.7217181324958801, + "num_tokens": 107174351.0, + "step": 3554 + }, + { + "epoch": 0.2931475220582172, + "grad_norm": 1.9715057611465454, + "learning_rate": 9.766419345974167e-07, + "loss": 0.8433, + "mean_token_accuracy": 0.7419024705886841, + "num_tokens": 107205853.0, + "step": 3555 + }, + { + "epoch": 0.2932299826832687, + "grad_norm": 1.9689645767211914, + "learning_rate": 9.76916735366859e-07, + "loss": 0.8637, + "mean_token_accuracy": 0.7428945899009705, + "num_tokens": 107237357.0, + "step": 3556 + }, + { + "epoch": 0.2933124433083203, + "grad_norm": 2.06335711479187, + "learning_rate": 9.77191536136301e-07, + "loss": 0.7625, + "mean_token_accuracy": 0.7578216791152954, + "num_tokens": 107267292.0, + "step": 3557 + }, + { + "epoch": 0.2933949039333718, + "grad_norm": 1.9851826429367065, + "learning_rate": 9.774663369057434e-07, + "loss": 0.7487, + "mean_token_accuracy": 0.7661941051483154, + "num_tokens": 107295425.0, + "step": 3558 + }, + { + "epoch": 0.2934773645584233, + "grad_norm": 2.1616649627685547, + "learning_rate": 9.777411376751854e-07, + "loss": 0.7706, + "mean_token_accuracy": 0.7454187870025635, + "num_tokens": 107322641.0, + "step": 3559 + }, + { + "epoch": 0.2935598251834749, + "grad_norm": 1.87538743019104, + "learning_rate": 9.780159384446275e-07, + "loss": 1.0063, + "mean_token_accuracy": 0.6967849135398865, + "num_tokens": 107360282.0, + "step": 3560 + }, + { + "epoch": 0.2936422858085264, + "grad_norm": 1.9835537672042847, + "learning_rate": 9.782907392140698e-07, + "loss": 0.9235, + "mean_token_accuracy": 0.72099769115448, + "num_tokens": 107392214.0, + "step": 3561 + }, + { + "epoch": 0.293724746433578, + "grad_norm": 2.138956069946289, + "learning_rate": 9.785655399835118e-07, + "loss": 0.7207, + "mean_token_accuracy": 0.7717907428741455, + "num_tokens": 107418148.0, + "step": 3562 + }, + { + "epoch": 0.2938072070586295, + "grad_norm": 1.7426329851150513, + "learning_rate": 9.788403407529541e-07, + "loss": 0.6796, + "mean_token_accuracy": 0.7801547050476074, + "num_tokens": 107452295.0, + "step": 3563 + }, + { + "epoch": 0.293889667683681, + "grad_norm": 1.965420126914978, + "learning_rate": 9.791151415223962e-07, + "loss": 0.7579, + "mean_token_accuracy": 0.7590221166610718, + "num_tokens": 107485718.0, + "step": 3564 + }, + { + "epoch": 0.2939721283087326, + "grad_norm": 2.024590015411377, + "learning_rate": 9.793899422918383e-07, + "loss": 0.7399, + "mean_token_accuracy": 0.7534557580947876, + "num_tokens": 107513636.0, + "step": 3565 + }, + { + "epoch": 0.2940545889337841, + "grad_norm": 2.0738723278045654, + "learning_rate": 9.796647430612805e-07, + "loss": 0.7261, + "mean_token_accuracy": 0.7661572694778442, + "num_tokens": 107542712.0, + "step": 3566 + }, + { + "epoch": 0.2941370495588357, + "grad_norm": 1.8452224731445312, + "learning_rate": 9.799395438307226e-07, + "loss": 0.7083, + "mean_token_accuracy": 0.7777346968650818, + "num_tokens": 107575324.0, + "step": 3567 + }, + { + "epoch": 0.2942195101838872, + "grad_norm": 1.925792932510376, + "learning_rate": 9.802143446001649e-07, + "loss": 0.8202, + "mean_token_accuracy": 0.7472277283668518, + "num_tokens": 107606723.0, + "step": 3568 + }, + { + "epoch": 0.2943019708089387, + "grad_norm": 1.937127709388733, + "learning_rate": 9.80489145369607e-07, + "loss": 0.8798, + "mean_token_accuracy": 0.7279180884361267, + "num_tokens": 107638267.0, + "step": 3569 + }, + { + "epoch": 0.2943844314339903, + "grad_norm": 2.2412192821502686, + "learning_rate": 9.807639461390492e-07, + "loss": 0.8449, + "mean_token_accuracy": 0.7309749126434326, + "num_tokens": 107661982.0, + "step": 3570 + }, + { + "epoch": 0.2944668920590418, + "grad_norm": 2.1964118480682373, + "learning_rate": 9.810387469084913e-07, + "loss": 0.9411, + "mean_token_accuracy": 0.7301391363143921, + "num_tokens": 107687472.0, + "step": 3571 + }, + { + "epoch": 0.2945493526840933, + "grad_norm": 1.9919003248214722, + "learning_rate": 9.813135476779336e-07, + "loss": 0.8868, + "mean_token_accuracy": 0.72476726770401, + "num_tokens": 107720107.0, + "step": 3572 + }, + { + "epoch": 0.2946318133091449, + "grad_norm": 1.7898783683776855, + "learning_rate": 9.815883484473756e-07, + "loss": 0.8235, + "mean_token_accuracy": 0.7478089332580566, + "num_tokens": 107757172.0, + "step": 3573 + }, + { + "epoch": 0.2947142739341964, + "grad_norm": 1.7673383951187134, + "learning_rate": 9.818631492168177e-07, + "loss": 0.7229, + "mean_token_accuracy": 0.7743195295333862, + "num_tokens": 107793416.0, + "step": 3574 + }, + { + "epoch": 0.294796734559248, + "grad_norm": 2.0971362590789795, + "learning_rate": 9.8213794998626e-07, + "loss": 0.8919, + "mean_token_accuracy": 0.7297149896621704, + "num_tokens": 107825476.0, + "step": 3575 + }, + { + "epoch": 0.2948791951842995, + "grad_norm": 2.0257246494293213, + "learning_rate": 9.82412750755702e-07, + "loss": 0.7646, + "mean_token_accuracy": 0.7582464814186096, + "num_tokens": 107857530.0, + "step": 3576 + }, + { + "epoch": 0.294961655809351, + "grad_norm": 2.129324436187744, + "learning_rate": 9.826875515251443e-07, + "loss": 0.7709, + "mean_token_accuracy": 0.7546970844268799, + "num_tokens": 107883680.0, + "step": 3577 + }, + { + "epoch": 0.2950441164344026, + "grad_norm": 2.3600656986236572, + "learning_rate": 9.829623522945864e-07, + "loss": 0.9747, + "mean_token_accuracy": 0.7032366991043091, + "num_tokens": 107908443.0, + "step": 3578 + }, + { + "epoch": 0.2951265770594541, + "grad_norm": 2.003351926803589, + "learning_rate": 9.832371530640284e-07, + "loss": 0.7764, + "mean_token_accuracy": 0.749669075012207, + "num_tokens": 107935737.0, + "step": 3579 + }, + { + "epoch": 0.2952090376845056, + "grad_norm": 2.2429332733154297, + "learning_rate": 9.835119538334707e-07, + "loss": 0.9056, + "mean_token_accuracy": 0.7184003591537476, + "num_tokens": 107960425.0, + "step": 3580 + }, + { + "epoch": 0.2952914983095572, + "grad_norm": 1.9199758768081665, + "learning_rate": 9.837867546029128e-07, + "loss": 0.7544, + "mean_token_accuracy": 0.7459332942962646, + "num_tokens": 107990956.0, + "step": 3581 + }, + { + "epoch": 0.2953739589346087, + "grad_norm": 2.018376111984253, + "learning_rate": 9.84061555372355e-07, + "loss": 0.9659, + "mean_token_accuracy": 0.7048344612121582, + "num_tokens": 108023344.0, + "step": 3582 + }, + { + "epoch": 0.2954564195596603, + "grad_norm": 1.8727736473083496, + "learning_rate": 9.843363561417971e-07, + "loss": 0.9312, + "mean_token_accuracy": 0.7118246555328369, + "num_tokens": 108060538.0, + "step": 3583 + }, + { + "epoch": 0.2955388801847118, + "grad_norm": 1.778761625289917, + "learning_rate": 9.846111569112394e-07, + "loss": 0.7087, + "mean_token_accuracy": 0.7809760570526123, + "num_tokens": 108093450.0, + "step": 3584 + }, + { + "epoch": 0.2956213408097633, + "grad_norm": 2.1710963249206543, + "learning_rate": 9.848859576806815e-07, + "loss": 1.0753, + "mean_token_accuracy": 0.6813348531723022, + "num_tokens": 108123230.0, + "step": 3585 + }, + { + "epoch": 0.2957038014348149, + "grad_norm": 2.2330493927001953, + "learning_rate": 9.851607584501235e-07, + "loss": 0.8682, + "mean_token_accuracy": 0.7305002212524414, + "num_tokens": 108149898.0, + "step": 3586 + }, + { + "epoch": 0.2957862620598664, + "grad_norm": 1.7892383337020874, + "learning_rate": 9.854355592195658e-07, + "loss": 0.8486, + "mean_token_accuracy": 0.740654706954956, + "num_tokens": 108184949.0, + "step": 3587 + }, + { + "epoch": 0.29586872268491793, + "grad_norm": 2.345804452896118, + "learning_rate": 9.857103599890079e-07, + "loss": 0.9717, + "mean_token_accuracy": 0.7080364227294922, + "num_tokens": 108210158.0, + "step": 3588 + }, + { + "epoch": 0.2959511833099695, + "grad_norm": 2.141871690750122, + "learning_rate": 9.859851607584502e-07, + "loss": 0.7328, + "mean_token_accuracy": 0.7636886835098267, + "num_tokens": 108238332.0, + "step": 3589 + }, + { + "epoch": 0.296033643935021, + "grad_norm": 1.9518096446990967, + "learning_rate": 9.862599615278922e-07, + "loss": 0.7723, + "mean_token_accuracy": 0.749313235282898, + "num_tokens": 108271031.0, + "step": 3590 + }, + { + "epoch": 0.2961161045600726, + "grad_norm": 1.9326200485229492, + "learning_rate": 9.865347622973343e-07, + "loss": 0.7648, + "mean_token_accuracy": 0.7642083168029785, + "num_tokens": 108300612.0, + "step": 3591 + }, + { + "epoch": 0.2961985651851241, + "grad_norm": 1.958447813987732, + "learning_rate": 9.868095630667766e-07, + "loss": 0.7769, + "mean_token_accuracy": 0.7630866765975952, + "num_tokens": 108331371.0, + "step": 3592 + }, + { + "epoch": 0.2962810258101756, + "grad_norm": 2.2455427646636963, + "learning_rate": 9.870843638362186e-07, + "loss": 0.9812, + "mean_token_accuracy": 0.7086679935455322, + "num_tokens": 108356754.0, + "step": 3593 + }, + { + "epoch": 0.2963634864352272, + "grad_norm": 1.980473279953003, + "learning_rate": 9.87359164605661e-07, + "loss": 0.7909, + "mean_token_accuracy": 0.7494762539863586, + "num_tokens": 108384111.0, + "step": 3594 + }, + { + "epoch": 0.2964459470602787, + "grad_norm": 2.02925443649292, + "learning_rate": 9.87633965375103e-07, + "loss": 0.8138, + "mean_token_accuracy": 0.7486258745193481, + "num_tokens": 108413508.0, + "step": 3595 + }, + { + "epoch": 0.29652840768533023, + "grad_norm": 1.9812729358673096, + "learning_rate": 9.879087661445453e-07, + "loss": 0.8729, + "mean_token_accuracy": 0.7233079671859741, + "num_tokens": 108445923.0, + "step": 3596 + }, + { + "epoch": 0.2966108683103818, + "grad_norm": 1.922451376914978, + "learning_rate": 9.881835669139873e-07, + "loss": 0.9671, + "mean_token_accuracy": 0.7110087871551514, + "num_tokens": 108480428.0, + "step": 3597 + }, + { + "epoch": 0.2966933289354333, + "grad_norm": 2.1522741317749023, + "learning_rate": 9.884583676834296e-07, + "loss": 0.7143, + "mean_token_accuracy": 0.7706809639930725, + "num_tokens": 108503342.0, + "step": 3598 + }, + { + "epoch": 0.2967757895604849, + "grad_norm": 2.100708246231079, + "learning_rate": 9.887331684528717e-07, + "loss": 0.8881, + "mean_token_accuracy": 0.7155201435089111, + "num_tokens": 108531180.0, + "step": 3599 + }, + { + "epoch": 0.2968582501855364, + "grad_norm": 2.2472634315490723, + "learning_rate": 9.890079692223137e-07, + "loss": 0.9062, + "mean_token_accuracy": 0.7226040959358215, + "num_tokens": 108557325.0, + "step": 3600 + }, + { + "epoch": 0.2969407108105879, + "grad_norm": 2.3233559131622314, + "learning_rate": 9.89282769991756e-07, + "loss": 0.9629, + "mean_token_accuracy": 0.7073513865470886, + "num_tokens": 108582034.0, + "step": 3601 + }, + { + "epoch": 0.2970231714356395, + "grad_norm": 2.3457515239715576, + "learning_rate": 9.89557570761198e-07, + "loss": 0.9039, + "mean_token_accuracy": 0.7205300331115723, + "num_tokens": 108604570.0, + "step": 3602 + }, + { + "epoch": 0.297105632060691, + "grad_norm": 1.8653526306152344, + "learning_rate": 9.898323715306403e-07, + "loss": 0.7962, + "mean_token_accuracy": 0.7489622235298157, + "num_tokens": 108638916.0, + "step": 3603 + }, + { + "epoch": 0.29718809268574253, + "grad_norm": 2.0911922454833984, + "learning_rate": 9.901071723000824e-07, + "loss": 0.8598, + "mean_token_accuracy": 0.7353463172912598, + "num_tokens": 108666564.0, + "step": 3604 + }, + { + "epoch": 0.2972705533107941, + "grad_norm": 1.940688133239746, + "learning_rate": 9.903819730695245e-07, + "loss": 0.8479, + "mean_token_accuracy": 0.7304019331932068, + "num_tokens": 108696411.0, + "step": 3605 + }, + { + "epoch": 0.2973530139358456, + "grad_norm": 2.504828453063965, + "learning_rate": 9.906567738389668e-07, + "loss": 0.8281, + "mean_token_accuracy": 0.7362768650054932, + "num_tokens": 108716656.0, + "step": 3606 + }, + { + "epoch": 0.2974354745608972, + "grad_norm": 1.910306692123413, + "learning_rate": 9.909315746084088e-07, + "loss": 0.7914, + "mean_token_accuracy": 0.7496538162231445, + "num_tokens": 108748303.0, + "step": 3607 + }, + { + "epoch": 0.2975179351859487, + "grad_norm": 2.1398701667785645, + "learning_rate": 9.91206375377851e-07, + "loss": 0.9115, + "mean_token_accuracy": 0.717930018901825, + "num_tokens": 108779361.0, + "step": 3608 + }, + { + "epoch": 0.29760039581100023, + "grad_norm": 1.8484599590301514, + "learning_rate": 9.914811761472932e-07, + "loss": 0.8911, + "mean_token_accuracy": 0.7232581377029419, + "num_tokens": 108813124.0, + "step": 3609 + }, + { + "epoch": 0.2976828564360518, + "grad_norm": 1.912327527999878, + "learning_rate": 9.917559769167352e-07, + "loss": 0.6952, + "mean_token_accuracy": 0.7802586555480957, + "num_tokens": 108843956.0, + "step": 3610 + }, + { + "epoch": 0.2977653170611033, + "grad_norm": 2.040463447570801, + "learning_rate": 9.920307776861775e-07, + "loss": 1.0169, + "mean_token_accuracy": 0.6922284364700317, + "num_tokens": 108874427.0, + "step": 3611 + }, + { + "epoch": 0.29784777768615484, + "grad_norm": 1.8748754262924194, + "learning_rate": 9.923055784556198e-07, + "loss": 0.8771, + "mean_token_accuracy": 0.7317531108856201, + "num_tokens": 108907462.0, + "step": 3612 + }, + { + "epoch": 0.2979302383112064, + "grad_norm": 1.7722505331039429, + "learning_rate": 9.925803792250619e-07, + "loss": 0.7021, + "mean_token_accuracy": 0.779238760471344, + "num_tokens": 108941652.0, + "step": 3613 + }, + { + "epoch": 0.2980126989362579, + "grad_norm": 2.140223979949951, + "learning_rate": 9.92855179994504e-07, + "loss": 0.8988, + "mean_token_accuracy": 0.7374316453933716, + "num_tokens": 108969760.0, + "step": 3614 + }, + { + "epoch": 0.2980951595613095, + "grad_norm": 1.9852995872497559, + "learning_rate": 9.93129980763946e-07, + "loss": 1.0113, + "mean_token_accuracy": 0.693295419216156, + "num_tokens": 109004091.0, + "step": 3615 + }, + { + "epoch": 0.298177620186361, + "grad_norm": 1.9991655349731445, + "learning_rate": 9.934047815333883e-07, + "loss": 0.8432, + "mean_token_accuracy": 0.7388409376144409, + "num_tokens": 109035714.0, + "step": 3616 + }, + { + "epoch": 0.29826008081141253, + "grad_norm": 1.7924764156341553, + "learning_rate": 9.936795823028305e-07, + "loss": 0.7987, + "mean_token_accuracy": 0.7447245121002197, + "num_tokens": 109072984.0, + "step": 3617 + }, + { + "epoch": 0.2983425414364641, + "grad_norm": 1.9026620388031006, + "learning_rate": 9.939543830722726e-07, + "loss": 0.9074, + "mean_token_accuracy": 0.7205452919006348, + "num_tokens": 109105784.0, + "step": 3618 + }, + { + "epoch": 0.2984250020615156, + "grad_norm": 2.0040700435638428, + "learning_rate": 9.942291838417147e-07, + "loss": 0.8483, + "mean_token_accuracy": 0.7332347631454468, + "num_tokens": 109138527.0, + "step": 3619 + }, + { + "epoch": 0.29850746268656714, + "grad_norm": 2.1055335998535156, + "learning_rate": 9.94503984611157e-07, + "loss": 0.793, + "mean_token_accuracy": 0.7553372383117676, + "num_tokens": 109164033.0, + "step": 3620 + }, + { + "epoch": 0.2985899233116187, + "grad_norm": 1.8071280717849731, + "learning_rate": 9.94778785380599e-07, + "loss": 0.7944, + "mean_token_accuracy": 0.7500289082527161, + "num_tokens": 109198428.0, + "step": 3621 + }, + { + "epoch": 0.2986723839366702, + "grad_norm": 1.7982912063598633, + "learning_rate": 9.950535861500413e-07, + "loss": 0.75, + "mean_token_accuracy": 0.757532000541687, + "num_tokens": 109232465.0, + "step": 3622 + }, + { + "epoch": 0.2987548445617218, + "grad_norm": 1.896878719329834, + "learning_rate": 9.953283869194834e-07, + "loss": 0.7764, + "mean_token_accuracy": 0.7528367638587952, + "num_tokens": 109262764.0, + "step": 3623 + }, + { + "epoch": 0.2988373051867733, + "grad_norm": 2.1902525424957275, + "learning_rate": 9.956031876889254e-07, + "loss": 0.8473, + "mean_token_accuracy": 0.7351592183113098, + "num_tokens": 109290938.0, + "step": 3624 + }, + { + "epoch": 0.29891976581182483, + "grad_norm": 1.749280333518982, + "learning_rate": 9.958779884583677e-07, + "loss": 0.8627, + "mean_token_accuracy": 0.7386779189109802, + "num_tokens": 109330159.0, + "step": 3625 + }, + { + "epoch": 0.2990022264368764, + "grad_norm": 2.020094156265259, + "learning_rate": 9.961527892278098e-07, + "loss": 0.8837, + "mean_token_accuracy": 0.7326391935348511, + "num_tokens": 109364482.0, + "step": 3626 + }, + { + "epoch": 0.2990846870619279, + "grad_norm": 1.9190152883529663, + "learning_rate": 9.96427589997252e-07, + "loss": 0.7203, + "mean_token_accuracy": 0.7731541991233826, + "num_tokens": 109393984.0, + "step": 3627 + }, + { + "epoch": 0.29916714768697944, + "grad_norm": 1.8553717136383057, + "learning_rate": 9.967023907666941e-07, + "loss": 0.9129, + "mean_token_accuracy": 0.7247699499130249, + "num_tokens": 109430396.0, + "step": 3628 + }, + { + "epoch": 0.299249608312031, + "grad_norm": 1.8238773345947266, + "learning_rate": 9.969771915361362e-07, + "loss": 0.8018, + "mean_token_accuracy": 0.7526684999465942, + "num_tokens": 109462915.0, + "step": 3629 + }, + { + "epoch": 0.29933206893708253, + "grad_norm": 1.9847322702407837, + "learning_rate": 9.972519923055785e-07, + "loss": 0.8592, + "mean_token_accuracy": 0.7347280979156494, + "num_tokens": 109494335.0, + "step": 3630 + }, + { + "epoch": 0.2994145295621341, + "grad_norm": 1.9882482290267944, + "learning_rate": 9.975267930750205e-07, + "loss": 0.922, + "mean_token_accuracy": 0.7216251492500305, + "num_tokens": 109523322.0, + "step": 3631 + }, + { + "epoch": 0.2994969901871856, + "grad_norm": 1.8132081031799316, + "learning_rate": 9.978015938444628e-07, + "loss": 0.7424, + "mean_token_accuracy": 0.7661364078521729, + "num_tokens": 109558564.0, + "step": 3632 + }, + { + "epoch": 0.29957945081223714, + "grad_norm": 1.9976584911346436, + "learning_rate": 9.980763946139049e-07, + "loss": 0.8815, + "mean_token_accuracy": 0.7294429540634155, + "num_tokens": 109592410.0, + "step": 3633 + }, + { + "epoch": 0.2996619114372887, + "grad_norm": 1.9525116682052612, + "learning_rate": 9.98351195383347e-07, + "loss": 0.6853, + "mean_token_accuracy": 0.779949426651001, + "num_tokens": 109623029.0, + "step": 3634 + }, + { + "epoch": 0.2997443720623402, + "grad_norm": 2.0284202098846436, + "learning_rate": 9.986259961527892e-07, + "loss": 0.6356, + "mean_token_accuracy": 0.798624575138092, + "num_tokens": 109647934.0, + "step": 3635 + }, + { + "epoch": 0.2998268326873918, + "grad_norm": 2.1312413215637207, + "learning_rate": 9.989007969222315e-07, + "loss": 0.8257, + "mean_token_accuracy": 0.7426664233207703, + "num_tokens": 109673970.0, + "step": 3636 + }, + { + "epoch": 0.2999092933124433, + "grad_norm": 1.904624104499817, + "learning_rate": 9.991755976916735e-07, + "loss": 0.6909, + "mean_token_accuracy": 0.7828778028488159, + "num_tokens": 109703707.0, + "step": 3637 + }, + { + "epoch": 0.29999175393749483, + "grad_norm": 1.870707392692566, + "learning_rate": 9.994503984611156e-07, + "loss": 0.7735, + "mean_token_accuracy": 0.756840705871582, + "num_tokens": 109739637.0, + "step": 3638 + }, + { + "epoch": 0.3000742145625464, + "grad_norm": 2.247462272644043, + "learning_rate": 9.997251992305577e-07, + "loss": 0.7511, + "mean_token_accuracy": 0.7616466283798218, + "num_tokens": 109764952.0, + "step": 3639 + }, + { + "epoch": 0.3001566751875979, + "grad_norm": 1.8842356204986572, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7597370147705078, + "num_tokens": 33843.0, + "step": 3640 + }, + { + "epoch": 0.30023913581264944, + "grad_norm": 2.1849284172058105, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7352315187454224, + "num_tokens": 59123.0, + "step": 3641 + }, + { + "epoch": 0.300321596437701, + "grad_norm": 1.9716740846633911, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7467107772827148, + "num_tokens": 88110.0, + "step": 3642 + }, + { + "epoch": 0.3004040570627525, + "grad_norm": 1.8005963563919067, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7455406188964844, + "num_tokens": 125776.0, + "step": 3643 + }, + { + "epoch": 0.3004865176878041, + "grad_norm": 1.9453797340393066, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7430020570755005, + "num_tokens": 157971.0, + "step": 3644 + }, + { + "epoch": 0.3005689783128556, + "grad_norm": 1.8227527141571045, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7511814832687378, + "num_tokens": 194462.0, + "step": 3645 + }, + { + "epoch": 0.30065143893790713, + "grad_norm": 2.0596814155578613, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7541197538375854, + "num_tokens": 220971.0, + "step": 3646 + }, + { + "epoch": 0.3007338995629587, + "grad_norm": 2.2837564945220947, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7751246690750122, + "num_tokens": 245585.0, + "step": 3647 + }, + { + "epoch": 0.3008163601880102, + "grad_norm": 2.096754789352417, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7321406602859497, + "num_tokens": 274997.0, + "step": 3648 + }, + { + "epoch": 0.30089882081306174, + "grad_norm": 2.050673246383667, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7319486141204834, + "num_tokens": 308344.0, + "step": 3649 + }, + { + "epoch": 0.3009812814381133, + "grad_norm": 2.095172643661499, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7325762510299683, + "num_tokens": 337722.0, + "step": 3650 + }, + { + "epoch": 0.30106374206316483, + "grad_norm": 1.9119607210159302, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.739046037197113, + "num_tokens": 369751.0, + "step": 3651 + }, + { + "epoch": 0.3011462026882164, + "grad_norm": 2.0800912380218506, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7368601560592651, + "num_tokens": 398467.0, + "step": 3652 + }, + { + "epoch": 0.3012286633132679, + "grad_norm": 1.9598705768585205, + "learning_rate": 1e-06, + "loss": 0.7619, + "mean_token_accuracy": 0.7646501064300537, + "num_tokens": 428121.0, + "step": 3653 + }, + { + "epoch": 0.30131112393831944, + "grad_norm": 1.9940065145492554, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7566954493522644, + "num_tokens": 457888.0, + "step": 3654 + }, + { + "epoch": 0.301393584563371, + "grad_norm": 1.81178617477417, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7536726593971252, + "num_tokens": 493348.0, + "step": 3655 + }, + { + "epoch": 0.3014760451884225, + "grad_norm": 1.861708402633667, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7308299541473389, + "num_tokens": 526455.0, + "step": 3656 + }, + { + "epoch": 0.30155850581347404, + "grad_norm": 2.0426125526428223, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7102196216583252, + "num_tokens": 560451.0, + "step": 3657 + }, + { + "epoch": 0.3016409664385256, + "grad_norm": 1.901859164237976, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7392282485961914, + "num_tokens": 592345.0, + "step": 3658 + }, + { + "epoch": 0.30172342706357713, + "grad_norm": 1.9883655309677124, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7479121685028076, + "num_tokens": 624490.0, + "step": 3659 + }, + { + "epoch": 0.3018058876886287, + "grad_norm": 2.4088006019592285, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.730799674987793, + "num_tokens": 648246.0, + "step": 3660 + }, + { + "epoch": 0.3018883483136802, + "grad_norm": 2.2434725761413574, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7081153988838196, + "num_tokens": 675955.0, + "step": 3661 + }, + { + "epoch": 0.30197080893873174, + "grad_norm": 1.91006338596344, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7506875991821289, + "num_tokens": 711257.0, + "step": 3662 + }, + { + "epoch": 0.3020532695637833, + "grad_norm": 1.9546469449996948, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7445193529129028, + "num_tokens": 739572.0, + "step": 3663 + }, + { + "epoch": 0.30213573018883483, + "grad_norm": 1.8199481964111328, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7887895703315735, + "num_tokens": 769755.0, + "step": 3664 + }, + { + "epoch": 0.30221819081388634, + "grad_norm": 1.9852681159973145, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7440911531448364, + "num_tokens": 799851.0, + "step": 3665 + }, + { + "epoch": 0.3023006514389379, + "grad_norm": 1.8378455638885498, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7557593584060669, + "num_tokens": 833294.0, + "step": 3666 + }, + { + "epoch": 0.30238311206398943, + "grad_norm": 2.074833631515503, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7277027368545532, + "num_tokens": 863766.0, + "step": 3667 + }, + { + "epoch": 0.302465572689041, + "grad_norm": 2.0357532501220703, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7303552627563477, + "num_tokens": 891706.0, + "step": 3668 + }, + { + "epoch": 0.3025480333140925, + "grad_norm": 1.9373624324798584, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7252712845802307, + "num_tokens": 923118.0, + "step": 3669 + }, + { + "epoch": 0.30263049393914404, + "grad_norm": 1.8209928274154663, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7631462812423706, + "num_tokens": 955302.0, + "step": 3670 + }, + { + "epoch": 0.3027129545641956, + "grad_norm": 1.8359671831130981, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.748580813407898, + "num_tokens": 992199.0, + "step": 3671 + }, + { + "epoch": 0.30279541518924713, + "grad_norm": 2.043437957763672, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7655417919158936, + "num_tokens": 1019782.0, + "step": 3672 + }, + { + "epoch": 0.30287787581429865, + "grad_norm": 1.857621669769287, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.72258460521698, + "num_tokens": 1055833.0, + "step": 3673 + }, + { + "epoch": 0.3029603364393502, + "grad_norm": 2.040107488632202, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7301097512245178, + "num_tokens": 1085004.0, + "step": 3674 + }, + { + "epoch": 0.30304279706440174, + "grad_norm": 2.048471450805664, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7288694381713867, + "num_tokens": 1115698.0, + "step": 3675 + }, + { + "epoch": 0.3031252576894533, + "grad_norm": 1.8816710710525513, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7630732655525208, + "num_tokens": 1147147.0, + "step": 3676 + }, + { + "epoch": 0.3032077183145048, + "grad_norm": 1.9074443578720093, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.743056058883667, + "num_tokens": 1180906.0, + "step": 3677 + }, + { + "epoch": 0.30329017893955634, + "grad_norm": 1.986318826675415, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7825441360473633, + "num_tokens": 1209076.0, + "step": 3678 + }, + { + "epoch": 0.3033726395646079, + "grad_norm": 2.1739299297332764, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7441509962081909, + "num_tokens": 1237434.0, + "step": 3679 + }, + { + "epoch": 0.30345510018965943, + "grad_norm": 1.8689061403274536, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7340567111968994, + "num_tokens": 1272276.0, + "step": 3680 + }, + { + "epoch": 0.30353756081471095, + "grad_norm": 1.9679241180419922, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7421730756759644, + "num_tokens": 1304004.0, + "step": 3681 + }, + { + "epoch": 0.3036200214397625, + "grad_norm": 1.860766887664795, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7540766000747681, + "num_tokens": 1334629.0, + "step": 3682 + }, + { + "epoch": 0.30370248206481404, + "grad_norm": 2.1334195137023926, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7238177061080933, + "num_tokens": 1360642.0, + "step": 3683 + }, + { + "epoch": 0.3037849426898656, + "grad_norm": 1.9584946632385254, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7493449449539185, + "num_tokens": 1391866.0, + "step": 3684 + }, + { + "epoch": 0.30386740331491713, + "grad_norm": 1.9728416204452515, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7540110349655151, + "num_tokens": 1420441.0, + "step": 3685 + }, + { + "epoch": 0.30394986393996865, + "grad_norm": 2.2267072200775146, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.733078122138977, + "num_tokens": 1451301.0, + "step": 3686 + }, + { + "epoch": 0.3040323245650202, + "grad_norm": 2.0004284381866455, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7575545907020569, + "num_tokens": 1481399.0, + "step": 3687 + }, + { + "epoch": 0.30411478519007173, + "grad_norm": 1.9497146606445312, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.726146936416626, + "num_tokens": 1513991.0, + "step": 3688 + }, + { + "epoch": 0.30419724581512325, + "grad_norm": 1.7632638216018677, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7476407289505005, + "num_tokens": 1550181.0, + "step": 3689 + }, + { + "epoch": 0.3042797064401748, + "grad_norm": 2.2138185501098633, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7274078726768494, + "num_tokens": 1576374.0, + "step": 3690 + }, + { + "epoch": 0.30436216706522634, + "grad_norm": 2.0989179611206055, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7420644760131836, + "num_tokens": 1605695.0, + "step": 3691 + }, + { + "epoch": 0.3044446276902779, + "grad_norm": 2.2244114875793457, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7309451103210449, + "num_tokens": 1634880.0, + "step": 3692 + }, + { + "epoch": 0.30452708831532943, + "grad_norm": 1.9759633541107178, + "learning_rate": 1e-06, + "loss": 0.7706, + "mean_token_accuracy": 0.7647573947906494, + "num_tokens": 1666507.0, + "step": 3693 + }, + { + "epoch": 0.30460954894038095, + "grad_norm": 1.9119527339935303, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7741615772247314, + "num_tokens": 1699433.0, + "step": 3694 + }, + { + "epoch": 0.3046920095654325, + "grad_norm": 2.1207382678985596, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7352824211120605, + "num_tokens": 1725877.0, + "step": 3695 + }, + { + "epoch": 0.30477447019048404, + "grad_norm": 2.342317819595337, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7379180192947388, + "num_tokens": 1748679.0, + "step": 3696 + }, + { + "epoch": 0.3048569308155356, + "grad_norm": 2.485978841781616, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.740363359451294, + "num_tokens": 1772249.0, + "step": 3697 + }, + { + "epoch": 0.3049393914405871, + "grad_norm": 2.1530606746673584, + "learning_rate": 1e-06, + "loss": 1.0445, + "mean_token_accuracy": 0.6907535791397095, + "num_tokens": 1800419.0, + "step": 3698 + }, + { + "epoch": 0.30502185206563864, + "grad_norm": 2.052964687347412, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7516158819198608, + "num_tokens": 1828716.0, + "step": 3699 + }, + { + "epoch": 0.3051043126906902, + "grad_norm": 1.7875573635101318, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7244407534599304, + "num_tokens": 1866812.0, + "step": 3700 + }, + { + "epoch": 0.30518677331574173, + "grad_norm": 2.390165090560913, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7277554273605347, + "num_tokens": 1889456.0, + "step": 3701 + }, + { + "epoch": 0.30526923394079325, + "grad_norm": 2.0432188510894775, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7499783635139465, + "num_tokens": 1917838.0, + "step": 3702 + }, + { + "epoch": 0.3053516945658448, + "grad_norm": 1.9674382209777832, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7352811098098755, + "num_tokens": 1948454.0, + "step": 3703 + }, + { + "epoch": 0.30543415519089634, + "grad_norm": 2.3412656784057617, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7587760090827942, + "num_tokens": 1975168.0, + "step": 3704 + }, + { + "epoch": 0.3055166158159479, + "grad_norm": 1.9216325283050537, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7285763025283813, + "num_tokens": 2009476.0, + "step": 3705 + }, + { + "epoch": 0.30559907644099943, + "grad_norm": 2.154304027557373, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6978458166122437, + "num_tokens": 2037413.0, + "step": 3706 + }, + { + "epoch": 0.30568153706605095, + "grad_norm": 2.2582950592041016, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7220449447631836, + "num_tokens": 2060698.0, + "step": 3707 + }, + { + "epoch": 0.3057639976911025, + "grad_norm": 2.3022513389587402, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7572014331817627, + "num_tokens": 2083161.0, + "step": 3708 + }, + { + "epoch": 0.30584645831615404, + "grad_norm": 1.9138606786727905, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7579524517059326, + "num_tokens": 2114887.0, + "step": 3709 + }, + { + "epoch": 0.30592891894120555, + "grad_norm": 2.0290346145629883, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7407361268997192, + "num_tokens": 2144437.0, + "step": 3710 + }, + { + "epoch": 0.3060113795662571, + "grad_norm": 2.1472928524017334, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7322024703025818, + "num_tokens": 2171619.0, + "step": 3711 + }, + { + "epoch": 0.30609384019130864, + "grad_norm": 2.1993019580841064, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.727810263633728, + "num_tokens": 2197872.0, + "step": 3712 + }, + { + "epoch": 0.3061763008163602, + "grad_norm": 1.9539101123809814, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7353215217590332, + "num_tokens": 2229572.0, + "step": 3713 + }, + { + "epoch": 0.30625876144141173, + "grad_norm": 2.2956533432006836, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7200296521186829, + "num_tokens": 2254198.0, + "step": 3714 + }, + { + "epoch": 0.30634122206646325, + "grad_norm": 1.944934606552124, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7284152507781982, + "num_tokens": 2292278.0, + "step": 3715 + }, + { + "epoch": 0.3064236826915148, + "grad_norm": 1.9839348793029785, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7475219368934631, + "num_tokens": 2325779.0, + "step": 3716 + }, + { + "epoch": 0.30650614331656634, + "grad_norm": 2.67592191696167, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.7043592929840088, + "num_tokens": 2345780.0, + "step": 3717 + }, + { + "epoch": 0.30658860394161785, + "grad_norm": 1.805325984954834, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7777072787284851, + "num_tokens": 2378952.0, + "step": 3718 + }, + { + "epoch": 0.3066710645666694, + "grad_norm": 1.7965279817581177, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7339339852333069, + "num_tokens": 2414187.0, + "step": 3719 + }, + { + "epoch": 0.30675352519172094, + "grad_norm": 1.8692166805267334, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7914050817489624, + "num_tokens": 2445627.0, + "step": 3720 + }, + { + "epoch": 0.3068359858167725, + "grad_norm": 2.0682878494262695, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7470029592514038, + "num_tokens": 2472293.0, + "step": 3721 + }, + { + "epoch": 0.30691844644182403, + "grad_norm": 1.9866732358932495, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7273128032684326, + "num_tokens": 2505613.0, + "step": 3722 + }, + { + "epoch": 0.30700090706687555, + "grad_norm": 1.915789008140564, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7572077512741089, + "num_tokens": 2538374.0, + "step": 3723 + }, + { + "epoch": 0.3070833676919271, + "grad_norm": 2.235896587371826, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7859774827957153, + "num_tokens": 2561298.0, + "step": 3724 + }, + { + "epoch": 0.30716582831697864, + "grad_norm": 2.0777275562286377, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7397556304931641, + "num_tokens": 2588111.0, + "step": 3725 + }, + { + "epoch": 0.30724828894203016, + "grad_norm": 1.7320480346679688, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7426984906196594, + "num_tokens": 2624646.0, + "step": 3726 + }, + { + "epoch": 0.30733074956708173, + "grad_norm": 1.871181607246399, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7465428709983826, + "num_tokens": 2659576.0, + "step": 3727 + }, + { + "epoch": 0.30741321019213325, + "grad_norm": 2.1681039333343506, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7070010304450989, + "num_tokens": 2689141.0, + "step": 3728 + }, + { + "epoch": 0.3074956708171848, + "grad_norm": 2.1913180351257324, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7208320498466492, + "num_tokens": 2716205.0, + "step": 3729 + }, + { + "epoch": 0.30757813144223634, + "grad_norm": 2.0155446529388428, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7558353543281555, + "num_tokens": 2746264.0, + "step": 3730 + }, + { + "epoch": 0.30766059206728785, + "grad_norm": 2.0128579139709473, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7692233324050903, + "num_tokens": 2774929.0, + "step": 3731 + }, + { + "epoch": 0.3077430526923394, + "grad_norm": 2.4618608951568604, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.703942060470581, + "num_tokens": 2798903.0, + "step": 3732 + }, + { + "epoch": 0.30782551331739094, + "grad_norm": 2.2398083209991455, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7352204322814941, + "num_tokens": 2825534.0, + "step": 3733 + }, + { + "epoch": 0.30790797394244246, + "grad_norm": 2.088123321533203, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7702237367630005, + "num_tokens": 2849752.0, + "step": 3734 + }, + { + "epoch": 0.30799043456749403, + "grad_norm": 2.0373265743255615, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7481151223182678, + "num_tokens": 2879239.0, + "step": 3735 + }, + { + "epoch": 0.30807289519254555, + "grad_norm": 1.9423695802688599, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7096207141876221, + "num_tokens": 2914190.0, + "step": 3736 + }, + { + "epoch": 0.3081553558175971, + "grad_norm": 1.7761187553405762, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7582372426986694, + "num_tokens": 2949404.0, + "step": 3737 + }, + { + "epoch": 0.30823781644264864, + "grad_norm": 1.8573102951049805, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.748720109462738, + "num_tokens": 2983579.0, + "step": 3738 + }, + { + "epoch": 0.30832027706770015, + "grad_norm": 2.094503402709961, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7304658889770508, + "num_tokens": 3009355.0, + "step": 3739 + }, + { + "epoch": 0.3084027376927517, + "grad_norm": 1.9800788164138794, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7373561859130859, + "num_tokens": 3041024.0, + "step": 3740 + }, + { + "epoch": 0.30848519831780324, + "grad_norm": 2.0198166370391846, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7279165983200073, + "num_tokens": 3071114.0, + "step": 3741 + }, + { + "epoch": 0.30856765894285476, + "grad_norm": 2.037078857421875, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7439541220664978, + "num_tokens": 3100778.0, + "step": 3742 + }, + { + "epoch": 0.30865011956790633, + "grad_norm": 1.920167088508606, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7573603391647339, + "num_tokens": 3132752.0, + "step": 3743 + }, + { + "epoch": 0.30873258019295785, + "grad_norm": 1.7918602228164673, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7485084533691406, + "num_tokens": 3165025.0, + "step": 3744 + }, + { + "epoch": 0.3088150408180094, + "grad_norm": 1.8761333227157593, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7507412433624268, + "num_tokens": 3197862.0, + "step": 3745 + }, + { + "epoch": 0.30889750144306094, + "grad_norm": 2.1295673847198486, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7201513051986694, + "num_tokens": 3225893.0, + "step": 3746 + }, + { + "epoch": 0.30897996206811246, + "grad_norm": 2.247880220413208, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.704914927482605, + "num_tokens": 3251329.0, + "step": 3747 + }, + { + "epoch": 0.30906242269316403, + "grad_norm": 1.8357874155044556, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7432219982147217, + "num_tokens": 3286774.0, + "step": 3748 + }, + { + "epoch": 0.30914488331821555, + "grad_norm": 2.045423746109009, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7455403804779053, + "num_tokens": 3314327.0, + "step": 3749 + }, + { + "epoch": 0.30922734394326706, + "grad_norm": 1.8189644813537598, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7325396537780762, + "num_tokens": 3349759.0, + "step": 3750 + }, + { + "epoch": 0.30930980456831864, + "grad_norm": 2.1251614093780518, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7170610427856445, + "num_tokens": 3378901.0, + "step": 3751 + }, + { + "epoch": 0.30939226519337015, + "grad_norm": 1.9460482597351074, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.718578577041626, + "num_tokens": 3409783.0, + "step": 3752 + }, + { + "epoch": 0.3094747258184217, + "grad_norm": 1.8686186075210571, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7655102014541626, + "num_tokens": 3442569.0, + "step": 3753 + }, + { + "epoch": 0.30955718644347324, + "grad_norm": 1.8482156991958618, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7792892456054688, + "num_tokens": 3474984.0, + "step": 3754 + }, + { + "epoch": 0.30963964706852476, + "grad_norm": 2.2468066215515137, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7396957278251648, + "num_tokens": 3497347.0, + "step": 3755 + }, + { + "epoch": 0.30972210769357633, + "grad_norm": 2.0483694076538086, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7387337684631348, + "num_tokens": 3525303.0, + "step": 3756 + }, + { + "epoch": 0.30980456831862785, + "grad_norm": 2.1433987617492676, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7779860496520996, + "num_tokens": 3550821.0, + "step": 3757 + }, + { + "epoch": 0.30988702894367937, + "grad_norm": 2.0404484272003174, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7637655735015869, + "num_tokens": 3580659.0, + "step": 3758 + }, + { + "epoch": 0.30996948956873094, + "grad_norm": 2.0426340103149414, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.729568362236023, + "num_tokens": 3608761.0, + "step": 3759 + }, + { + "epoch": 0.31005195019378246, + "grad_norm": 2.2593982219696045, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7010422348976135, + "num_tokens": 3635956.0, + "step": 3760 + }, + { + "epoch": 0.310134410818834, + "grad_norm": 1.89535653591156, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7127860188484192, + "num_tokens": 3671597.0, + "step": 3761 + }, + { + "epoch": 0.31021687144388554, + "grad_norm": 2.2283923625946045, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7040255665779114, + "num_tokens": 3696465.0, + "step": 3762 + }, + { + "epoch": 0.31029933206893706, + "grad_norm": 1.949639081954956, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7334362268447876, + "num_tokens": 3727970.0, + "step": 3763 + }, + { + "epoch": 0.31038179269398863, + "grad_norm": 1.9234485626220703, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7284959554672241, + "num_tokens": 3759784.0, + "step": 3764 + }, + { + "epoch": 0.31046425331904015, + "grad_norm": 1.9302834272384644, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7566215395927429, + "num_tokens": 3793074.0, + "step": 3765 + }, + { + "epoch": 0.3105467139440917, + "grad_norm": 1.9245880842208862, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7022607922554016, + "num_tokens": 3828176.0, + "step": 3766 + }, + { + "epoch": 0.31062917456914324, + "grad_norm": 1.9993482828140259, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7630615830421448, + "num_tokens": 3857544.0, + "step": 3767 + }, + { + "epoch": 0.31071163519419476, + "grad_norm": 2.045661449432373, + "learning_rate": 1e-06, + "loss": 0.7724, + "mean_token_accuracy": 0.7582721710205078, + "num_tokens": 3887801.0, + "step": 3768 + }, + { + "epoch": 0.31079409581924633, + "grad_norm": 1.9872199296951294, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7444652318954468, + "num_tokens": 3920756.0, + "step": 3769 + }, + { + "epoch": 0.31087655644429785, + "grad_norm": 2.0403552055358887, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7381477952003479, + "num_tokens": 3951162.0, + "step": 3770 + }, + { + "epoch": 0.31095901706934936, + "grad_norm": 1.8310214281082153, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7540984153747559, + "num_tokens": 3986781.0, + "step": 3771 + }, + { + "epoch": 0.31104147769440094, + "grad_norm": 2.1073153018951416, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7446639537811279, + "num_tokens": 4013683.0, + "step": 3772 + }, + { + "epoch": 0.31112393831945245, + "grad_norm": 2.060582399368286, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.792610764503479, + "num_tokens": 4040065.0, + "step": 3773 + }, + { + "epoch": 0.311206398944504, + "grad_norm": 1.7747735977172852, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7801741361618042, + "num_tokens": 4073736.0, + "step": 3774 + }, + { + "epoch": 0.31128885956955554, + "grad_norm": 1.9791359901428223, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7764635682106018, + "num_tokens": 4100626.0, + "step": 3775 + }, + { + "epoch": 0.31137132019460706, + "grad_norm": 2.0706160068511963, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7306696176528931, + "num_tokens": 4128218.0, + "step": 3776 + }, + { + "epoch": 0.31145378081965863, + "grad_norm": 1.9420331716537476, + "learning_rate": 1e-06, + "loss": 0.7868, + "mean_token_accuracy": 0.7491839528083801, + "num_tokens": 4156599.0, + "step": 3777 + }, + { + "epoch": 0.31153624144471015, + "grad_norm": 2.0010178089141846, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7514368295669556, + "num_tokens": 4185830.0, + "step": 3778 + }, + { + "epoch": 0.31161870206976167, + "grad_norm": 2.02107834815979, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.765063464641571, + "num_tokens": 4211505.0, + "step": 3779 + }, + { + "epoch": 0.31170116269481324, + "grad_norm": 1.7544934749603271, + "learning_rate": 1e-06, + "loss": 0.7799, + "mean_token_accuracy": 0.7530567049980164, + "num_tokens": 4249133.0, + "step": 3780 + }, + { + "epoch": 0.31178362331986476, + "grad_norm": 1.7928321361541748, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7300780415534973, + "num_tokens": 4286779.0, + "step": 3781 + }, + { + "epoch": 0.31186608394491633, + "grad_norm": 2.0014877319335938, + "learning_rate": 1e-06, + "loss": 0.7531, + "mean_token_accuracy": 0.7621219158172607, + "num_tokens": 4319527.0, + "step": 3782 + }, + { + "epoch": 0.31194854456996784, + "grad_norm": 1.8086159229278564, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.7988383769989014, + "num_tokens": 4351312.0, + "step": 3783 + }, + { + "epoch": 0.31203100519501936, + "grad_norm": 1.8314013481140137, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7360820174217224, + "num_tokens": 4383075.0, + "step": 3784 + }, + { + "epoch": 0.31211346582007093, + "grad_norm": 2.0880043506622314, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7494421005249023, + "num_tokens": 4407316.0, + "step": 3785 + }, + { + "epoch": 0.31219592644512245, + "grad_norm": 1.968425989151001, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.7536141276359558, + "num_tokens": 4435353.0, + "step": 3786 + }, + { + "epoch": 0.31227838707017397, + "grad_norm": 2.5197479724884033, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7214097380638123, + "num_tokens": 4457384.0, + "step": 3787 + }, + { + "epoch": 0.31236084769522554, + "grad_norm": 1.954231858253479, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.73271644115448, + "num_tokens": 4486114.0, + "step": 3788 + }, + { + "epoch": 0.31244330832027706, + "grad_norm": 2.30228590965271, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7076109647750854, + "num_tokens": 4510976.0, + "step": 3789 + }, + { + "epoch": 0.31252576894532863, + "grad_norm": 1.9978604316711426, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.718865692615509, + "num_tokens": 4542788.0, + "step": 3790 + }, + { + "epoch": 0.31260822957038015, + "grad_norm": 1.6538695096969604, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7807643413543701, + "num_tokens": 4584449.0, + "step": 3791 + }, + { + "epoch": 0.31269069019543166, + "grad_norm": 1.8659344911575317, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7468979358673096, + "num_tokens": 4615329.0, + "step": 3792 + }, + { + "epoch": 0.31277315082048324, + "grad_norm": 2.2920422554016113, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7417510151863098, + "num_tokens": 4640094.0, + "step": 3793 + }, + { + "epoch": 0.31285561144553475, + "grad_norm": 2.1679351329803467, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7386348247528076, + "num_tokens": 4663630.0, + "step": 3794 + }, + { + "epoch": 0.31293807207058627, + "grad_norm": 2.1149425506591797, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7429594993591309, + "num_tokens": 4692288.0, + "step": 3795 + }, + { + "epoch": 0.31302053269563784, + "grad_norm": 2.1769626140594482, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.7481927871704102, + "num_tokens": 4718708.0, + "step": 3796 + }, + { + "epoch": 0.31310299332068936, + "grad_norm": 2.1726810932159424, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.753033459186554, + "num_tokens": 4745243.0, + "step": 3797 + }, + { + "epoch": 0.31318545394574093, + "grad_norm": 2.1204235553741455, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.725692868232727, + "num_tokens": 4771885.0, + "step": 3798 + }, + { + "epoch": 0.31326791457079245, + "grad_norm": 2.0060720443725586, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.7946050763130188, + "num_tokens": 4800925.0, + "step": 3799 + }, + { + "epoch": 0.31335037519584397, + "grad_norm": 2.00108003616333, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7442731857299805, + "num_tokens": 4829673.0, + "step": 3800 + }, + { + "epoch": 0.31343283582089554, + "grad_norm": 1.8527019023895264, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7359033823013306, + "num_tokens": 4862122.0, + "step": 3801 + }, + { + "epoch": 0.31351529644594706, + "grad_norm": 2.116382122039795, + "learning_rate": 1e-06, + "loss": 0.7886, + "mean_token_accuracy": 0.7554967403411865, + "num_tokens": 4887500.0, + "step": 3802 + }, + { + "epoch": 0.3135977570709986, + "grad_norm": 2.2935590744018555, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7448104023933411, + "num_tokens": 4909315.0, + "step": 3803 + }, + { + "epoch": 0.31368021769605015, + "grad_norm": 1.886888027191162, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7359769940376282, + "num_tokens": 4941942.0, + "step": 3804 + }, + { + "epoch": 0.31376267832110166, + "grad_norm": 1.927808403968811, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7423949837684631, + "num_tokens": 4974391.0, + "step": 3805 + }, + { + "epoch": 0.31384513894615323, + "grad_norm": 1.8283467292785645, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.760047972202301, + "num_tokens": 5011637.0, + "step": 3806 + }, + { + "epoch": 0.31392759957120475, + "grad_norm": 2.066118001937866, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7122411727905273, + "num_tokens": 5042744.0, + "step": 3807 + }, + { + "epoch": 0.31401006019625627, + "grad_norm": 2.0112271308898926, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7297424077987671, + "num_tokens": 5072034.0, + "step": 3808 + }, + { + "epoch": 0.31409252082130784, + "grad_norm": 1.8467389345169067, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7590497136116028, + "num_tokens": 5105461.0, + "step": 3809 + }, + { + "epoch": 0.31417498144635936, + "grad_norm": 2.1277456283569336, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7215736508369446, + "num_tokens": 5136877.0, + "step": 3810 + }, + { + "epoch": 0.3142574420714109, + "grad_norm": 2.205496072769165, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6997703313827515, + "num_tokens": 5164464.0, + "step": 3811 + }, + { + "epoch": 0.31433990269646245, + "grad_norm": 1.8118627071380615, + "learning_rate": 1e-06, + "loss": 0.5622, + "mean_token_accuracy": 0.8133808970451355, + "num_tokens": 5196524.0, + "step": 3812 + }, + { + "epoch": 0.31442236332151396, + "grad_norm": 2.1256110668182373, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.6994789242744446, + "num_tokens": 5225362.0, + "step": 3813 + }, + { + "epoch": 0.31450482394656554, + "grad_norm": 2.0293004512786865, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7274559736251831, + "num_tokens": 5253423.0, + "step": 3814 + }, + { + "epoch": 0.31458728457161705, + "grad_norm": 2.1330182552337646, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7756778001785278, + "num_tokens": 5279222.0, + "step": 3815 + }, + { + "epoch": 0.31466974519666857, + "grad_norm": 1.8987687826156616, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7921447157859802, + "num_tokens": 5307931.0, + "step": 3816 + }, + { + "epoch": 0.31475220582172014, + "grad_norm": 2.0463314056396484, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.7936269044876099, + "num_tokens": 5335348.0, + "step": 3817 + }, + { + "epoch": 0.31483466644677166, + "grad_norm": 1.8898065090179443, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7387897968292236, + "num_tokens": 5366708.0, + "step": 3818 + }, + { + "epoch": 0.3149171270718232, + "grad_norm": 1.9617937803268433, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7083714008331299, + "num_tokens": 5402569.0, + "step": 3819 + }, + { + "epoch": 0.31499958769687475, + "grad_norm": 2.14192271232605, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7487760782241821, + "num_tokens": 5429273.0, + "step": 3820 + }, + { + "epoch": 0.31508204832192627, + "grad_norm": 1.9455987215042114, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7447083592414856, + "num_tokens": 5460831.0, + "step": 3821 + }, + { + "epoch": 0.31516450894697784, + "grad_norm": 1.8847765922546387, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.794082760810852, + "num_tokens": 5492913.0, + "step": 3822 + }, + { + "epoch": 0.31524696957202936, + "grad_norm": 1.996525526046753, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7108299136161804, + "num_tokens": 5524206.0, + "step": 3823 + }, + { + "epoch": 0.3153294301970809, + "grad_norm": 2.0726046562194824, + "learning_rate": 1e-06, + "loss": 0.7623, + "mean_token_accuracy": 0.7643494606018066, + "num_tokens": 5551598.0, + "step": 3824 + }, + { + "epoch": 0.31541189082213245, + "grad_norm": 2.309962272644043, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6988654136657715, + "num_tokens": 5576892.0, + "step": 3825 + }, + { + "epoch": 0.31549435144718396, + "grad_norm": 1.9474228620529175, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.7584053874015808, + "num_tokens": 5609618.0, + "step": 3826 + }, + { + "epoch": 0.31557681207223554, + "grad_norm": 2.222792625427246, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7363598346710205, + "num_tokens": 5637848.0, + "step": 3827 + }, + { + "epoch": 0.31565927269728705, + "grad_norm": 2.0258758068084717, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7422075271606445, + "num_tokens": 5668602.0, + "step": 3828 + }, + { + "epoch": 0.31574173332233857, + "grad_norm": 2.0268867015838623, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7232049703598022, + "num_tokens": 5700439.0, + "step": 3829 + }, + { + "epoch": 0.31582419394739014, + "grad_norm": 2.1138980388641357, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7609144449234009, + "num_tokens": 5725511.0, + "step": 3830 + }, + { + "epoch": 0.31590665457244166, + "grad_norm": 2.311509609222412, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7644915580749512, + "num_tokens": 5746024.0, + "step": 3831 + }, + { + "epoch": 0.3159891151974932, + "grad_norm": 2.088895559310913, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7415950298309326, + "num_tokens": 5772033.0, + "step": 3832 + }, + { + "epoch": 0.31607157582254475, + "grad_norm": 1.9354392290115356, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7466644048690796, + "num_tokens": 5801600.0, + "step": 3833 + }, + { + "epoch": 0.31615403644759627, + "grad_norm": 1.8960424661636353, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7530913949012756, + "num_tokens": 5836208.0, + "step": 3834 + }, + { + "epoch": 0.31623649707264784, + "grad_norm": 1.7237358093261719, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7344334721565247, + "num_tokens": 5875732.0, + "step": 3835 + }, + { + "epoch": 0.31631895769769935, + "grad_norm": 2.4226162433624268, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7470066547393799, + "num_tokens": 5896575.0, + "step": 3836 + }, + { + "epoch": 0.31640141832275087, + "grad_norm": 1.98225736618042, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7249211072921753, + "num_tokens": 5928141.0, + "step": 3837 + }, + { + "epoch": 0.31648387894780244, + "grad_norm": 2.0659027099609375, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7074651718139648, + "num_tokens": 5956869.0, + "step": 3838 + }, + { + "epoch": 0.31656633957285396, + "grad_norm": 2.120894432067871, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7314443588256836, + "num_tokens": 5983931.0, + "step": 3839 + }, + { + "epoch": 0.3166488001979055, + "grad_norm": 1.8338618278503418, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7921333312988281, + "num_tokens": 6014278.0, + "step": 3840 + }, + { + "epoch": 0.31673126082295705, + "grad_norm": 1.974876880645752, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7234081029891968, + "num_tokens": 6045777.0, + "step": 3841 + }, + { + "epoch": 0.31681372144800857, + "grad_norm": 1.9094994068145752, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7767298221588135, + "num_tokens": 6074869.0, + "step": 3842 + }, + { + "epoch": 0.31689618207306014, + "grad_norm": 1.9159342050552368, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7479335069656372, + "num_tokens": 6110090.0, + "step": 3843 + }, + { + "epoch": 0.31697864269811166, + "grad_norm": 2.1014158725738525, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7524838447570801, + "num_tokens": 6139652.0, + "step": 3844 + }, + { + "epoch": 0.3170611033231632, + "grad_norm": 2.10077166557312, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7153602838516235, + "num_tokens": 6170653.0, + "step": 3845 + }, + { + "epoch": 0.31714356394821475, + "grad_norm": 1.9600459337234497, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.772797167301178, + "num_tokens": 6203256.0, + "step": 3846 + }, + { + "epoch": 0.31722602457326626, + "grad_norm": 2.0451135635375977, + "learning_rate": 1e-06, + "loss": 0.7698, + "mean_token_accuracy": 0.7531367540359497, + "num_tokens": 6230934.0, + "step": 3847 + }, + { + "epoch": 0.3173084851983178, + "grad_norm": 1.872362732887268, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.7890498042106628, + "num_tokens": 6257838.0, + "step": 3848 + }, + { + "epoch": 0.31739094582336935, + "grad_norm": 2.0447134971618652, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7437859773635864, + "num_tokens": 6283787.0, + "step": 3849 + }, + { + "epoch": 0.31747340644842087, + "grad_norm": 2.1802988052368164, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6996738314628601, + "num_tokens": 6310709.0, + "step": 3850 + }, + { + "epoch": 0.31755586707347244, + "grad_norm": 1.9613666534423828, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7461618185043335, + "num_tokens": 6340729.0, + "step": 3851 + }, + { + "epoch": 0.31763832769852396, + "grad_norm": 2.0603396892547607, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7788166403770447, + "num_tokens": 6365768.0, + "step": 3852 + }, + { + "epoch": 0.3177207883235755, + "grad_norm": 1.9242901802062988, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7380998134613037, + "num_tokens": 6398456.0, + "step": 3853 + }, + { + "epoch": 0.31780324894862705, + "grad_norm": 1.9606214761734009, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.7814944982528687, + "num_tokens": 6426944.0, + "step": 3854 + }, + { + "epoch": 0.31788570957367857, + "grad_norm": 2.0192532539367676, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7220592498779297, + "num_tokens": 6459468.0, + "step": 3855 + }, + { + "epoch": 0.3179681701987301, + "grad_norm": 1.917561650276184, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7572729587554932, + "num_tokens": 6491794.0, + "step": 3856 + }, + { + "epoch": 0.31805063082378165, + "grad_norm": 2.466506004333496, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.743188738822937, + "num_tokens": 6515717.0, + "step": 3857 + }, + { + "epoch": 0.31813309144883317, + "grad_norm": 2.0747017860412598, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7322570085525513, + "num_tokens": 6545010.0, + "step": 3858 + }, + { + "epoch": 0.31821555207388474, + "grad_norm": 1.855939507484436, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7381485104560852, + "num_tokens": 6579690.0, + "step": 3859 + }, + { + "epoch": 0.31829801269893626, + "grad_norm": 1.8218673467636108, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7767870426177979, + "num_tokens": 6611998.0, + "step": 3860 + }, + { + "epoch": 0.3183804733239878, + "grad_norm": 1.8313695192337036, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7455226182937622, + "num_tokens": 6647734.0, + "step": 3861 + }, + { + "epoch": 0.31846293394903935, + "grad_norm": 2.063638925552368, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7411248087882996, + "num_tokens": 6674612.0, + "step": 3862 + }, + { + "epoch": 0.31854539457409087, + "grad_norm": 2.1366913318634033, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7662995457649231, + "num_tokens": 6700559.0, + "step": 3863 + }, + { + "epoch": 0.3186278551991424, + "grad_norm": 2.0132198333740234, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6944736838340759, + "num_tokens": 6730658.0, + "step": 3864 + }, + { + "epoch": 0.31871031582419396, + "grad_norm": 2.048405170440674, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.721550703048706, + "num_tokens": 6759505.0, + "step": 3865 + }, + { + "epoch": 0.3187927764492455, + "grad_norm": 2.4072487354278564, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7066985368728638, + "num_tokens": 6781870.0, + "step": 3866 + }, + { + "epoch": 0.31887523707429705, + "grad_norm": 1.9795629978179932, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7436891794204712, + "num_tokens": 6812882.0, + "step": 3867 + }, + { + "epoch": 0.31895769769934856, + "grad_norm": 2.095250368118286, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7276009321212769, + "num_tokens": 6840974.0, + "step": 3868 + }, + { + "epoch": 0.3190401583244001, + "grad_norm": 2.248283863067627, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7082087993621826, + "num_tokens": 6867672.0, + "step": 3869 + }, + { + "epoch": 0.31912261894945165, + "grad_norm": 1.8063308000564575, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7760117053985596, + "num_tokens": 6899291.0, + "step": 3870 + }, + { + "epoch": 0.31920507957450317, + "grad_norm": 2.308016538619995, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7302367687225342, + "num_tokens": 6920247.0, + "step": 3871 + }, + { + "epoch": 0.3192875401995547, + "grad_norm": 2.0573763847351074, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7290729284286499, + "num_tokens": 6946224.0, + "step": 3872 + }, + { + "epoch": 0.31937000082460626, + "grad_norm": 1.8444607257843018, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7212400436401367, + "num_tokens": 6981946.0, + "step": 3873 + }, + { + "epoch": 0.3194524614496578, + "grad_norm": 2.4056124687194824, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7561632990837097, + "num_tokens": 7003932.0, + "step": 3874 + }, + { + "epoch": 0.31953492207470935, + "grad_norm": 1.9518142938613892, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7842501997947693, + "num_tokens": 7032551.0, + "step": 3875 + }, + { + "epoch": 0.31961738269976087, + "grad_norm": 1.8532907962799072, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7321995496749878, + "num_tokens": 7065974.0, + "step": 3876 + }, + { + "epoch": 0.3196998433248124, + "grad_norm": 2.010160446166992, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6966251730918884, + "num_tokens": 7099995.0, + "step": 3877 + }, + { + "epoch": 0.31978230394986396, + "grad_norm": 2.159323215484619, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7071279287338257, + "num_tokens": 7127871.0, + "step": 3878 + }, + { + "epoch": 0.31986476457491547, + "grad_norm": 1.9030588865280151, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7726849317550659, + "num_tokens": 7159827.0, + "step": 3879 + }, + { + "epoch": 0.319947225199967, + "grad_norm": 2.0373289585113525, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7770544290542603, + "num_tokens": 7190351.0, + "step": 3880 + }, + { + "epoch": 0.32002968582501856, + "grad_norm": 2.479774236679077, + "learning_rate": 1e-06, + "loss": 0.7669, + "mean_token_accuracy": 0.7537849545478821, + "num_tokens": 7211707.0, + "step": 3881 + }, + { + "epoch": 0.3201121464500701, + "grad_norm": 2.2048707008361816, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7285856008529663, + "num_tokens": 7238147.0, + "step": 3882 + }, + { + "epoch": 0.32019460707512165, + "grad_norm": 1.9729251861572266, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7684816122055054, + "num_tokens": 7263632.0, + "step": 3883 + }, + { + "epoch": 0.32027706770017317, + "grad_norm": 1.673832893371582, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7869398593902588, + "num_tokens": 7302221.0, + "step": 3884 + }, + { + "epoch": 0.3203595283252247, + "grad_norm": 1.8382983207702637, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7660797834396362, + "num_tokens": 7336522.0, + "step": 3885 + }, + { + "epoch": 0.32044198895027626, + "grad_norm": 2.076829671859741, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7708205580711365, + "num_tokens": 7365116.0, + "step": 3886 + }, + { + "epoch": 0.3205244495753278, + "grad_norm": 1.8960425853729248, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7619786262512207, + "num_tokens": 7397068.0, + "step": 3887 + }, + { + "epoch": 0.3206069102003793, + "grad_norm": 1.9762449264526367, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7400699853897095, + "num_tokens": 7429186.0, + "step": 3888 + }, + { + "epoch": 0.32068937082543086, + "grad_norm": 1.95919668674469, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7704327702522278, + "num_tokens": 7456876.0, + "step": 3889 + }, + { + "epoch": 0.3207718314504824, + "grad_norm": 2.034503698348999, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7158238887786865, + "num_tokens": 7487327.0, + "step": 3890 + }, + { + "epoch": 0.32085429207553395, + "grad_norm": 2.2114758491516113, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7640717029571533, + "num_tokens": 7508629.0, + "step": 3891 + }, + { + "epoch": 0.32093675270058547, + "grad_norm": 1.8778091669082642, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7436689138412476, + "num_tokens": 7541466.0, + "step": 3892 + }, + { + "epoch": 0.321019213325637, + "grad_norm": 1.8102465867996216, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7768726348876953, + "num_tokens": 7575056.0, + "step": 3893 + }, + { + "epoch": 0.32110167395068856, + "grad_norm": 2.0136301517486572, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7557604312896729, + "num_tokens": 7603401.0, + "step": 3894 + }, + { + "epoch": 0.3211841345757401, + "grad_norm": 2.3107898235321045, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7367240786552429, + "num_tokens": 7626944.0, + "step": 3895 + }, + { + "epoch": 0.32126659520079165, + "grad_norm": 1.9215224981307983, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7495021820068359, + "num_tokens": 7662971.0, + "step": 3896 + }, + { + "epoch": 0.32134905582584317, + "grad_norm": 1.865915060043335, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7380287647247314, + "num_tokens": 7699123.0, + "step": 3897 + }, + { + "epoch": 0.3214315164508947, + "grad_norm": 2.1434760093688965, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7487018704414368, + "num_tokens": 7725201.0, + "step": 3898 + }, + { + "epoch": 0.32151397707594626, + "grad_norm": 2.1222076416015625, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7542369365692139, + "num_tokens": 7751221.0, + "step": 3899 + }, + { + "epoch": 0.3215964377009978, + "grad_norm": 2.0186243057250977, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.7849947214126587, + "num_tokens": 7777499.0, + "step": 3900 + }, + { + "epoch": 0.3216788983260493, + "grad_norm": 1.9062081575393677, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7319275736808777, + "num_tokens": 7811257.0, + "step": 3901 + }, + { + "epoch": 0.32176135895110086, + "grad_norm": 1.881614089012146, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7492464780807495, + "num_tokens": 7842330.0, + "step": 3902 + }, + { + "epoch": 0.3218438195761524, + "grad_norm": 1.984208583831787, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7440301775932312, + "num_tokens": 7874579.0, + "step": 3903 + }, + { + "epoch": 0.32192628020120395, + "grad_norm": 2.213068723678589, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7515230178833008, + "num_tokens": 7899929.0, + "step": 3904 + }, + { + "epoch": 0.32200874082625547, + "grad_norm": 1.9858647584915161, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7285038232803345, + "num_tokens": 7929567.0, + "step": 3905 + }, + { + "epoch": 0.322091201451307, + "grad_norm": 1.9053186178207397, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7284529209136963, + "num_tokens": 7967689.0, + "step": 3906 + }, + { + "epoch": 0.32217366207635856, + "grad_norm": 2.3529248237609863, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6950441598892212, + "num_tokens": 7991250.0, + "step": 3907 + }, + { + "epoch": 0.3222561227014101, + "grad_norm": 2.294607400894165, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7197155952453613, + "num_tokens": 8016769.0, + "step": 3908 + }, + { + "epoch": 0.3223385833264616, + "grad_norm": 2.01660418510437, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7811692953109741, + "num_tokens": 8045161.0, + "step": 3909 + }, + { + "epoch": 0.32242104395151316, + "grad_norm": 1.896224856376648, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.7850252389907837, + "num_tokens": 8075602.0, + "step": 3910 + }, + { + "epoch": 0.3225035045765647, + "grad_norm": 2.078686475753784, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.7052046656608582, + "num_tokens": 8103942.0, + "step": 3911 + }, + { + "epoch": 0.32258596520161625, + "grad_norm": 2.159860134124756, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7355878353118896, + "num_tokens": 8130320.0, + "step": 3912 + }, + { + "epoch": 0.32266842582666777, + "grad_norm": 2.035555601119995, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7359832525253296, + "num_tokens": 8157689.0, + "step": 3913 + }, + { + "epoch": 0.3227508864517193, + "grad_norm": 2.3042173385620117, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7549659609794617, + "num_tokens": 8181658.0, + "step": 3914 + }, + { + "epoch": 0.32283334707677086, + "grad_norm": 1.8308879137039185, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7735520005226135, + "num_tokens": 8218200.0, + "step": 3915 + }, + { + "epoch": 0.3229158077018224, + "grad_norm": 2.2288403511047363, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7575210332870483, + "num_tokens": 8242594.0, + "step": 3916 + }, + { + "epoch": 0.3229982683268739, + "grad_norm": 1.8935949802398682, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7679232358932495, + "num_tokens": 8274716.0, + "step": 3917 + }, + { + "epoch": 0.32308072895192547, + "grad_norm": 2.0577313899993896, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7315925359725952, + "num_tokens": 8305129.0, + "step": 3918 + }, + { + "epoch": 0.323163189576977, + "grad_norm": 2.035198211669922, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7551435232162476, + "num_tokens": 8335620.0, + "step": 3919 + }, + { + "epoch": 0.32324565020202856, + "grad_norm": 1.9703242778778076, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7322726845741272, + "num_tokens": 8368068.0, + "step": 3920 + }, + { + "epoch": 0.3233281108270801, + "grad_norm": 2.081925868988037, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.7968993782997131, + "num_tokens": 8394047.0, + "step": 3921 + }, + { + "epoch": 0.3234105714521316, + "grad_norm": 1.8365380764007568, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.74323970079422, + "num_tokens": 8426592.0, + "step": 3922 + }, + { + "epoch": 0.32349303207718316, + "grad_norm": 2.177177906036377, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7422196865081787, + "num_tokens": 8450234.0, + "step": 3923 + }, + { + "epoch": 0.3235754927022347, + "grad_norm": 1.8540326356887817, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7624560594558716, + "num_tokens": 8482126.0, + "step": 3924 + }, + { + "epoch": 0.3236579533272862, + "grad_norm": 2.288113594055176, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7460343241691589, + "num_tokens": 8507674.0, + "step": 3925 + }, + { + "epoch": 0.32374041395233777, + "grad_norm": 1.8006820678710938, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7816990613937378, + "num_tokens": 8543759.0, + "step": 3926 + }, + { + "epoch": 0.3238228745773893, + "grad_norm": 1.8746094703674316, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7509195804595947, + "num_tokens": 8578922.0, + "step": 3927 + }, + { + "epoch": 0.32390533520244086, + "grad_norm": 1.988287091255188, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7299875020980835, + "num_tokens": 8609551.0, + "step": 3928 + }, + { + "epoch": 0.3239877958274924, + "grad_norm": 1.851364016532898, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7872809767723083, + "num_tokens": 8641956.0, + "step": 3929 + }, + { + "epoch": 0.3240702564525439, + "grad_norm": 1.7885204553604126, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7826301455497742, + "num_tokens": 8673311.0, + "step": 3930 + }, + { + "epoch": 0.32415271707759546, + "grad_norm": 1.9667505025863647, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7647690773010254, + "num_tokens": 8702384.0, + "step": 3931 + }, + { + "epoch": 0.324235177702647, + "grad_norm": 1.8785574436187744, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7312200665473938, + "num_tokens": 8736488.0, + "step": 3932 + }, + { + "epoch": 0.3243176383276985, + "grad_norm": 1.7949937582015991, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.77816241979599, + "num_tokens": 8771836.0, + "step": 3933 + }, + { + "epoch": 0.32440009895275007, + "grad_norm": 2.1692771911621094, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7672282457351685, + "num_tokens": 8796992.0, + "step": 3934 + }, + { + "epoch": 0.3244825595778016, + "grad_norm": 2.0827200412750244, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7476427555084229, + "num_tokens": 8825949.0, + "step": 3935 + }, + { + "epoch": 0.32456502020285316, + "grad_norm": 1.9199074506759644, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7437449097633362, + "num_tokens": 8856717.0, + "step": 3936 + }, + { + "epoch": 0.3246474808279047, + "grad_norm": 1.9815268516540527, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.767699122428894, + "num_tokens": 8886467.0, + "step": 3937 + }, + { + "epoch": 0.3247299414529562, + "grad_norm": 1.7507922649383545, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7699713706970215, + "num_tokens": 8926843.0, + "step": 3938 + }, + { + "epoch": 0.32481240207800777, + "grad_norm": 2.018862009048462, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7399011850357056, + "num_tokens": 8953900.0, + "step": 3939 + }, + { + "epoch": 0.3248948627030593, + "grad_norm": 1.953092336654663, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7640489339828491, + "num_tokens": 8982577.0, + "step": 3940 + }, + { + "epoch": 0.3249773233281108, + "grad_norm": 2.051731586456299, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7340607643127441, + "num_tokens": 9011950.0, + "step": 3941 + }, + { + "epoch": 0.3250597839531624, + "grad_norm": 1.7879053354263306, + "learning_rate": 1e-06, + "loss": 0.5762, + "mean_token_accuracy": 0.8059567213058472, + "num_tokens": 9043054.0, + "step": 3942 + }, + { + "epoch": 0.3251422445782139, + "grad_norm": 1.9766114950180054, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7649542093276978, + "num_tokens": 9072666.0, + "step": 3943 + }, + { + "epoch": 0.32522470520326546, + "grad_norm": 2.0666534900665283, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7618064880371094, + "num_tokens": 9101226.0, + "step": 3944 + }, + { + "epoch": 0.325307165828317, + "grad_norm": 1.6999220848083496, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7710031867027283, + "num_tokens": 9138687.0, + "step": 3945 + }, + { + "epoch": 0.3253896264533685, + "grad_norm": 1.8375029563903809, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7497374415397644, + "num_tokens": 9173160.0, + "step": 3946 + }, + { + "epoch": 0.32547208707842007, + "grad_norm": 1.9278546571731567, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7155352830886841, + "num_tokens": 9206363.0, + "step": 3947 + }, + { + "epoch": 0.3255545477034716, + "grad_norm": 1.9301526546478271, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7329927086830139, + "num_tokens": 9237417.0, + "step": 3948 + }, + { + "epoch": 0.3256370083285231, + "grad_norm": 2.2764804363250732, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7439137697219849, + "num_tokens": 9261354.0, + "step": 3949 + }, + { + "epoch": 0.3257194689535747, + "grad_norm": 1.7619894742965698, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.770005464553833, + "num_tokens": 9295307.0, + "step": 3950 + }, + { + "epoch": 0.3258019295786262, + "grad_norm": 1.8232622146606445, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7499121427536011, + "num_tokens": 9331572.0, + "step": 3951 + }, + { + "epoch": 0.32588439020367777, + "grad_norm": 2.150500535964966, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7529920935630798, + "num_tokens": 9359668.0, + "step": 3952 + }, + { + "epoch": 0.3259668508287293, + "grad_norm": 2.9653754234313965, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7494689226150513, + "num_tokens": 9382620.0, + "step": 3953 + }, + { + "epoch": 0.3260493114537808, + "grad_norm": 2.128683090209961, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.731182336807251, + "num_tokens": 9409913.0, + "step": 3954 + }, + { + "epoch": 0.32613177207883237, + "grad_norm": 1.9618200063705444, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7453949451446533, + "num_tokens": 9440529.0, + "step": 3955 + }, + { + "epoch": 0.3262142327038839, + "grad_norm": 2.040809392929077, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7497292757034302, + "num_tokens": 9471468.0, + "step": 3956 + }, + { + "epoch": 0.32629669332893546, + "grad_norm": 1.824594497680664, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.7452186346054077, + "num_tokens": 9505640.0, + "step": 3957 + }, + { + "epoch": 0.326379153953987, + "grad_norm": 2.038320302963257, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.723342776298523, + "num_tokens": 9536741.0, + "step": 3958 + }, + { + "epoch": 0.3264616145790385, + "grad_norm": 1.9300479888916016, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7187933921813965, + "num_tokens": 9568742.0, + "step": 3959 + }, + { + "epoch": 0.32654407520409007, + "grad_norm": 2.2478301525115967, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7581198215484619, + "num_tokens": 9591017.0, + "step": 3960 + }, + { + "epoch": 0.3266265358291416, + "grad_norm": 2.28362774848938, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7381759881973267, + "num_tokens": 9614423.0, + "step": 3961 + }, + { + "epoch": 0.3267089964541931, + "grad_norm": 2.0392556190490723, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7475807070732117, + "num_tokens": 9641622.0, + "step": 3962 + }, + { + "epoch": 0.3267914570792447, + "grad_norm": 2.2182843685150146, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7011305093765259, + "num_tokens": 9666339.0, + "step": 3963 + }, + { + "epoch": 0.3268739177042962, + "grad_norm": 1.9305700063705444, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.7829408645629883, + "num_tokens": 9696790.0, + "step": 3964 + }, + { + "epoch": 0.32695637832934776, + "grad_norm": 1.8314368724822998, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.749432384967804, + "num_tokens": 9729298.0, + "step": 3965 + }, + { + "epoch": 0.3270388389543993, + "grad_norm": 2.129472255706787, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.750821590423584, + "num_tokens": 9758862.0, + "step": 3966 + }, + { + "epoch": 0.3271212995794508, + "grad_norm": 2.0837769508361816, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7176578044891357, + "num_tokens": 9788554.0, + "step": 3967 + }, + { + "epoch": 0.32720376020450237, + "grad_norm": 1.9333289861679077, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7677079439163208, + "num_tokens": 9818118.0, + "step": 3968 + }, + { + "epoch": 0.3272862208295539, + "grad_norm": 2.144516706466675, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7736705541610718, + "num_tokens": 9844355.0, + "step": 3969 + }, + { + "epoch": 0.3273686814546054, + "grad_norm": 1.964416265487671, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7285245656967163, + "num_tokens": 9874192.0, + "step": 3970 + }, + { + "epoch": 0.327451142079657, + "grad_norm": 1.9629193544387817, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7167370319366455, + "num_tokens": 9904687.0, + "step": 3971 + }, + { + "epoch": 0.3275336027047085, + "grad_norm": 1.985535740852356, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7333846092224121, + "num_tokens": 9935819.0, + "step": 3972 + }, + { + "epoch": 0.32761606332976007, + "grad_norm": 2.315551519393921, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.700219988822937, + "num_tokens": 9961546.0, + "step": 3973 + }, + { + "epoch": 0.3276985239548116, + "grad_norm": 1.9284244775772095, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7572168707847595, + "num_tokens": 9992796.0, + "step": 3974 + }, + { + "epoch": 0.3277809845798631, + "grad_norm": 2.1021728515625, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.7022919654846191, + "num_tokens": 10021610.0, + "step": 3975 + }, + { + "epoch": 0.32786344520491467, + "grad_norm": 2.104951858520508, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7447803020477295, + "num_tokens": 10049463.0, + "step": 3976 + }, + { + "epoch": 0.3279459058299662, + "grad_norm": 1.9655400514602661, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7278164625167847, + "num_tokens": 10082914.0, + "step": 3977 + }, + { + "epoch": 0.3280283664550177, + "grad_norm": 2.0524632930755615, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7173606753349304, + "num_tokens": 10111770.0, + "step": 3978 + }, + { + "epoch": 0.3281108270800693, + "grad_norm": 2.583693504333496, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7309171557426453, + "num_tokens": 10131248.0, + "step": 3979 + }, + { + "epoch": 0.3281932877051208, + "grad_norm": 1.7999639511108398, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.749855637550354, + "num_tokens": 10169156.0, + "step": 3980 + }, + { + "epoch": 0.32827574833017237, + "grad_norm": 2.1990952491760254, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7485195994377136, + "num_tokens": 10194805.0, + "step": 3981 + }, + { + "epoch": 0.3283582089552239, + "grad_norm": 2.0454328060150146, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7429209351539612, + "num_tokens": 10226257.0, + "step": 3982 + }, + { + "epoch": 0.3284406695802754, + "grad_norm": 1.9050376415252686, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.780730664730072, + "num_tokens": 10258371.0, + "step": 3983 + }, + { + "epoch": 0.328523130205327, + "grad_norm": 2.047081708908081, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.747511625289917, + "num_tokens": 10287891.0, + "step": 3984 + }, + { + "epoch": 0.3286055908303785, + "grad_norm": 1.7290935516357422, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.7924251556396484, + "num_tokens": 10320758.0, + "step": 3985 + }, + { + "epoch": 0.32868805145543, + "grad_norm": 1.8717405796051025, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7528234720230103, + "num_tokens": 10356430.0, + "step": 3986 + }, + { + "epoch": 0.3287705120804816, + "grad_norm": 1.7881755828857422, + "learning_rate": 1e-06, + "loss": 0.7999, + "mean_token_accuracy": 0.7511268854141235, + "num_tokens": 10391072.0, + "step": 3987 + }, + { + "epoch": 0.3288529727055331, + "grad_norm": 2.145303964614868, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7318838834762573, + "num_tokens": 10418093.0, + "step": 3988 + }, + { + "epoch": 0.32893543333058467, + "grad_norm": 2.095320224761963, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7384164929389954, + "num_tokens": 10443552.0, + "step": 3989 + }, + { + "epoch": 0.3290178939556362, + "grad_norm": 1.8107032775878906, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.7606266736984253, + "num_tokens": 10479018.0, + "step": 3990 + }, + { + "epoch": 0.3291003545806877, + "grad_norm": 2.0925142765045166, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7229466438293457, + "num_tokens": 10509374.0, + "step": 3991 + }, + { + "epoch": 0.3291828152057393, + "grad_norm": 1.9908843040466309, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7447860240936279, + "num_tokens": 10539771.0, + "step": 3992 + }, + { + "epoch": 0.3292652758307908, + "grad_norm": 2.0964291095733643, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.717990517616272, + "num_tokens": 10567336.0, + "step": 3993 + }, + { + "epoch": 0.3293477364558423, + "grad_norm": 2.2330994606018066, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7006275653839111, + "num_tokens": 10592573.0, + "step": 3994 + }, + { + "epoch": 0.3294301970808939, + "grad_norm": 1.911266565322876, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.738807737827301, + "num_tokens": 10625349.0, + "step": 3995 + }, + { + "epoch": 0.3295126577059454, + "grad_norm": 1.9087632894515991, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7812684178352356, + "num_tokens": 10653679.0, + "step": 3996 + }, + { + "epoch": 0.329595118330997, + "grad_norm": 1.9497056007385254, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7692557573318481, + "num_tokens": 10686263.0, + "step": 3997 + }, + { + "epoch": 0.3296775789560485, + "grad_norm": 2.051408052444458, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7534250020980835, + "num_tokens": 10712937.0, + "step": 3998 + }, + { + "epoch": 0.3297600395811, + "grad_norm": 1.9543474912643433, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7826209664344788, + "num_tokens": 10743724.0, + "step": 3999 + }, + { + "epoch": 0.3298425002061516, + "grad_norm": 1.7495604753494263, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.775073766708374, + "num_tokens": 10778846.0, + "step": 4000 + }, + { + "epoch": 0.3299249608312031, + "grad_norm": 2.038379192352295, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7178781032562256, + "num_tokens": 10812320.0, + "step": 4001 + }, + { + "epoch": 0.3300074214562546, + "grad_norm": 1.9938294887542725, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7292412519454956, + "num_tokens": 10843998.0, + "step": 4002 + }, + { + "epoch": 0.3300898820813062, + "grad_norm": 1.9438223838806152, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7059831619262695, + "num_tokens": 10878028.0, + "step": 4003 + }, + { + "epoch": 0.3301723427063577, + "grad_norm": 2.1075239181518555, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7252984046936035, + "num_tokens": 10905648.0, + "step": 4004 + }, + { + "epoch": 0.3302548033314093, + "grad_norm": 2.3285326957702637, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7656584978103638, + "num_tokens": 10927047.0, + "step": 4005 + }, + { + "epoch": 0.3303372639564608, + "grad_norm": 1.9900145530700684, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.744087278842926, + "num_tokens": 10954810.0, + "step": 4006 + }, + { + "epoch": 0.3304197245815123, + "grad_norm": 2.2600913047790527, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7505663633346558, + "num_tokens": 10979138.0, + "step": 4007 + }, + { + "epoch": 0.3305021852065639, + "grad_norm": 2.307072401046753, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7218936681747437, + "num_tokens": 11004727.0, + "step": 4008 + }, + { + "epoch": 0.3305846458316154, + "grad_norm": 1.8772636651992798, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.798126220703125, + "num_tokens": 11036831.0, + "step": 4009 + }, + { + "epoch": 0.3306671064566669, + "grad_norm": 2.017686367034912, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.752736508846283, + "num_tokens": 11067807.0, + "step": 4010 + }, + { + "epoch": 0.3307495670817185, + "grad_norm": 2.142380475997925, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.734381914138794, + "num_tokens": 11092826.0, + "step": 4011 + }, + { + "epoch": 0.33083202770677, + "grad_norm": 1.94721519947052, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.721815288066864, + "num_tokens": 11125071.0, + "step": 4012 + }, + { + "epoch": 0.3309144883318216, + "grad_norm": 2.0104403495788574, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7575482130050659, + "num_tokens": 11155189.0, + "step": 4013 + }, + { + "epoch": 0.3309969489568731, + "grad_norm": 2.1337790489196777, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.755220890045166, + "num_tokens": 11180570.0, + "step": 4014 + }, + { + "epoch": 0.3310794095819246, + "grad_norm": 2.0681729316711426, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.750646710395813, + "num_tokens": 11206372.0, + "step": 4015 + }, + { + "epoch": 0.3311618702069762, + "grad_norm": 1.9428188800811768, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7558819651603699, + "num_tokens": 11236169.0, + "step": 4016 + }, + { + "epoch": 0.3312443308320277, + "grad_norm": 2.254377841949463, + "learning_rate": 1e-06, + "loss": 0.7821, + "mean_token_accuracy": 0.7541542649269104, + "num_tokens": 11261547.0, + "step": 4017 + }, + { + "epoch": 0.3313267914570792, + "grad_norm": 2.4606215953826904, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7308735847473145, + "num_tokens": 11285667.0, + "step": 4018 + }, + { + "epoch": 0.3314092520821308, + "grad_norm": 2.01735782623291, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.736382246017456, + "num_tokens": 11315756.0, + "step": 4019 + }, + { + "epoch": 0.3314917127071823, + "grad_norm": 2.0448715686798096, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.756756603717804, + "num_tokens": 11344511.0, + "step": 4020 + }, + { + "epoch": 0.3315741733322339, + "grad_norm": 1.9650089740753174, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7562752962112427, + "num_tokens": 11375111.0, + "step": 4021 + }, + { + "epoch": 0.3316566339572854, + "grad_norm": 2.253781795501709, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.778867781162262, + "num_tokens": 11398395.0, + "step": 4022 + }, + { + "epoch": 0.3317390945823369, + "grad_norm": 1.811751365661621, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.774734377861023, + "num_tokens": 11431972.0, + "step": 4023 + }, + { + "epoch": 0.3318215552073885, + "grad_norm": 1.89224374294281, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7368186712265015, + "num_tokens": 11465776.0, + "step": 4024 + }, + { + "epoch": 0.33190401583244, + "grad_norm": 1.9867500066757202, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7607898712158203, + "num_tokens": 11494800.0, + "step": 4025 + }, + { + "epoch": 0.3319864764574916, + "grad_norm": 2.5629916191101074, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7392488718032837, + "num_tokens": 11517377.0, + "step": 4026 + }, + { + "epoch": 0.3320689370825431, + "grad_norm": 2.2181615829467773, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.745865523815155, + "num_tokens": 11541346.0, + "step": 4027 + }, + { + "epoch": 0.3321513977075946, + "grad_norm": 1.8969086408615112, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7676001787185669, + "num_tokens": 11571211.0, + "step": 4028 + }, + { + "epoch": 0.3322338583326462, + "grad_norm": 1.938571810722351, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7362245321273804, + "num_tokens": 11601096.0, + "step": 4029 + }, + { + "epoch": 0.3323163189576977, + "grad_norm": 2.098196268081665, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7198977470397949, + "num_tokens": 11630341.0, + "step": 4030 + }, + { + "epoch": 0.3323987795827492, + "grad_norm": 1.899074912071228, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7650678753852844, + "num_tokens": 11664643.0, + "step": 4031 + }, + { + "epoch": 0.3324812402078008, + "grad_norm": 2.029996633529663, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7430342435836792, + "num_tokens": 11694848.0, + "step": 4032 + }, + { + "epoch": 0.3325637008328523, + "grad_norm": 2.0819122791290283, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7274166345596313, + "num_tokens": 11723087.0, + "step": 4033 + }, + { + "epoch": 0.3326461614579039, + "grad_norm": 1.8878746032714844, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7482284307479858, + "num_tokens": 11756998.0, + "step": 4034 + }, + { + "epoch": 0.3327286220829554, + "grad_norm": 2.157585382461548, + "learning_rate": 1e-06, + "loss": 1.0379, + "mean_token_accuracy": 0.6872779130935669, + "num_tokens": 11786982.0, + "step": 4035 + }, + { + "epoch": 0.3328110827080069, + "grad_norm": 2.257641553878784, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.728771448135376, + "num_tokens": 11814086.0, + "step": 4036 + }, + { + "epoch": 0.3328935433330585, + "grad_norm": 1.969354510307312, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.6893610954284668, + "num_tokens": 11847744.0, + "step": 4037 + }, + { + "epoch": 0.33297600395811, + "grad_norm": 2.1508066654205322, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7550672292709351, + "num_tokens": 11873389.0, + "step": 4038 + }, + { + "epoch": 0.3330584645831615, + "grad_norm": 1.941347360610962, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7664608359336853, + "num_tokens": 11905717.0, + "step": 4039 + }, + { + "epoch": 0.3331409252082131, + "grad_norm": 1.896828055381775, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7619099617004395, + "num_tokens": 11936563.0, + "step": 4040 + }, + { + "epoch": 0.3332233858332646, + "grad_norm": 1.9985239505767822, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7481629252433777, + "num_tokens": 11971054.0, + "step": 4041 + }, + { + "epoch": 0.3333058464583162, + "grad_norm": 1.8795918226242065, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7788034677505493, + "num_tokens": 12002139.0, + "step": 4042 + }, + { + "epoch": 0.3333883070833677, + "grad_norm": 2.244438648223877, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7471082210540771, + "num_tokens": 12026289.0, + "step": 4043 + }, + { + "epoch": 0.3334707677084192, + "grad_norm": 1.9206440448760986, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7369288206100464, + "num_tokens": 12056585.0, + "step": 4044 + }, + { + "epoch": 0.3335532283334708, + "grad_norm": 2.1891822814941406, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.6903305053710938, + "num_tokens": 12084607.0, + "step": 4045 + }, + { + "epoch": 0.3336356889585223, + "grad_norm": 1.8581093549728394, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7287032008171082, + "num_tokens": 12119568.0, + "step": 4046 + }, + { + "epoch": 0.3337181495835738, + "grad_norm": 2.1636648178100586, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7329208850860596, + "num_tokens": 12149706.0, + "step": 4047 + }, + { + "epoch": 0.3338006102086254, + "grad_norm": 1.9548377990722656, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7552106976509094, + "num_tokens": 12180736.0, + "step": 4048 + }, + { + "epoch": 0.3338830708336769, + "grad_norm": 1.8372159004211426, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7388728260993958, + "num_tokens": 12218168.0, + "step": 4049 + }, + { + "epoch": 0.3339655314587285, + "grad_norm": 1.8345706462860107, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7679308652877808, + "num_tokens": 12250032.0, + "step": 4050 + }, + { + "epoch": 0.33404799208378, + "grad_norm": 1.8540751934051514, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7649985551834106, + "num_tokens": 12284549.0, + "step": 4051 + }, + { + "epoch": 0.3341304527088315, + "grad_norm": 1.861732006072998, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7613484263420105, + "num_tokens": 12316324.0, + "step": 4052 + }, + { + "epoch": 0.3342129133338831, + "grad_norm": 2.129441976547241, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7595076560974121, + "num_tokens": 12340672.0, + "step": 4053 + }, + { + "epoch": 0.3342953739589346, + "grad_norm": 1.9184017181396484, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7600898742675781, + "num_tokens": 12373471.0, + "step": 4054 + }, + { + "epoch": 0.3343778345839861, + "grad_norm": 2.063305377960205, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.759627103805542, + "num_tokens": 12400821.0, + "step": 4055 + }, + { + "epoch": 0.3344602952090377, + "grad_norm": 2.3187313079833984, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7084968090057373, + "num_tokens": 12424210.0, + "step": 4056 + }, + { + "epoch": 0.3345427558340892, + "grad_norm": 2.743034839630127, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7390572428703308, + "num_tokens": 12445319.0, + "step": 4057 + }, + { + "epoch": 0.3346252164591408, + "grad_norm": 1.805346131324768, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7479986548423767, + "num_tokens": 12477761.0, + "step": 4058 + }, + { + "epoch": 0.3347076770841923, + "grad_norm": 2.053650140762329, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7587631940841675, + "num_tokens": 12509433.0, + "step": 4059 + }, + { + "epoch": 0.3347901377092438, + "grad_norm": 2.141835927963257, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7151229977607727, + "num_tokens": 12539406.0, + "step": 4060 + }, + { + "epoch": 0.3348725983342954, + "grad_norm": 2.0684592723846436, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7094858884811401, + "num_tokens": 12570841.0, + "step": 4061 + }, + { + "epoch": 0.3349550589593469, + "grad_norm": 1.8167959451675415, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.781599760055542, + "num_tokens": 12603780.0, + "step": 4062 + }, + { + "epoch": 0.3350375195843984, + "grad_norm": 2.2640416622161865, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7441198229789734, + "num_tokens": 12627479.0, + "step": 4063 + }, + { + "epoch": 0.33511998020945, + "grad_norm": 2.038583993911743, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7139056324958801, + "num_tokens": 12654837.0, + "step": 4064 + }, + { + "epoch": 0.3352024408345015, + "grad_norm": 2.2687180042266846, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7146283984184265, + "num_tokens": 12678494.0, + "step": 4065 + }, + { + "epoch": 0.3352849014595531, + "grad_norm": 1.800903081893921, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7652379870414734, + "num_tokens": 12712995.0, + "step": 4066 + }, + { + "epoch": 0.3353673620846046, + "grad_norm": 2.185084342956543, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7698259353637695, + "num_tokens": 12737090.0, + "step": 4067 + }, + { + "epoch": 0.3354498227096561, + "grad_norm": 1.8789516687393188, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7516540288925171, + "num_tokens": 12770010.0, + "step": 4068 + }, + { + "epoch": 0.3355322833347077, + "grad_norm": 2.003032684326172, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.74651038646698, + "num_tokens": 12797268.0, + "step": 4069 + }, + { + "epoch": 0.3356147439597592, + "grad_norm": 2.069265842437744, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.7971212863922119, + "num_tokens": 12830813.0, + "step": 4070 + }, + { + "epoch": 0.3356972045848107, + "grad_norm": 2.096254825592041, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7412519454956055, + "num_tokens": 12859678.0, + "step": 4071 + }, + { + "epoch": 0.3357796652098623, + "grad_norm": 2.042412281036377, + "learning_rate": 1e-06, + "loss": 0.7825, + "mean_token_accuracy": 0.753158688545227, + "num_tokens": 12887899.0, + "step": 4072 + }, + { + "epoch": 0.3358621258349138, + "grad_norm": 1.8887172937393188, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.741570234298706, + "num_tokens": 12920981.0, + "step": 4073 + }, + { + "epoch": 0.3359445864599654, + "grad_norm": 1.9116636514663696, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7780712842941284, + "num_tokens": 12951656.0, + "step": 4074 + }, + { + "epoch": 0.3360270470850169, + "grad_norm": 1.7910356521606445, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7129010558128357, + "num_tokens": 12987813.0, + "step": 4075 + }, + { + "epoch": 0.3361095077100684, + "grad_norm": 1.9522265195846558, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7624092698097229, + "num_tokens": 13018981.0, + "step": 4076 + }, + { + "epoch": 0.33619196833512, + "grad_norm": 2.048224687576294, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.745771050453186, + "num_tokens": 13048749.0, + "step": 4077 + }, + { + "epoch": 0.3362744289601715, + "grad_norm": 2.0577478408813477, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7473903894424438, + "num_tokens": 13078405.0, + "step": 4078 + }, + { + "epoch": 0.33635688958522303, + "grad_norm": 1.950852394104004, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7229878306388855, + "num_tokens": 13109559.0, + "step": 4079 + }, + { + "epoch": 0.3364393502102746, + "grad_norm": 1.7446469068527222, + "learning_rate": 1e-06, + "loss": 0.7688, + "mean_token_accuracy": 0.7586297988891602, + "num_tokens": 13143551.0, + "step": 4080 + }, + { + "epoch": 0.3365218108353261, + "grad_norm": 1.7449628114700317, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7571842670440674, + "num_tokens": 13181367.0, + "step": 4081 + }, + { + "epoch": 0.3366042714603777, + "grad_norm": 1.7816014289855957, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7610442638397217, + "num_tokens": 13215468.0, + "step": 4082 + }, + { + "epoch": 0.3366867320854292, + "grad_norm": 2.0308966636657715, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7525909543037415, + "num_tokens": 13242376.0, + "step": 4083 + }, + { + "epoch": 0.3367691927104807, + "grad_norm": 1.9466171264648438, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7316173315048218, + "num_tokens": 13273605.0, + "step": 4084 + }, + { + "epoch": 0.3368516533355323, + "grad_norm": 1.8946094512939453, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7418158054351807, + "num_tokens": 13309336.0, + "step": 4085 + }, + { + "epoch": 0.3369341139605838, + "grad_norm": 1.947002649307251, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7464495897293091, + "num_tokens": 13339837.0, + "step": 4086 + }, + { + "epoch": 0.3370165745856354, + "grad_norm": 1.9088871479034424, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7423200607299805, + "num_tokens": 13373699.0, + "step": 4087 + }, + { + "epoch": 0.3370990352106869, + "grad_norm": 1.954121470451355, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7332146167755127, + "num_tokens": 13404674.0, + "step": 4088 + }, + { + "epoch": 0.3371814958357384, + "grad_norm": 1.8814873695373535, + "learning_rate": 1e-06, + "loss": 0.5303, + "mean_token_accuracy": 0.8223472833633423, + "num_tokens": 13432586.0, + "step": 4089 + }, + { + "epoch": 0.33726395646079, + "grad_norm": 2.1387979984283447, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7298080921173096, + "num_tokens": 13462591.0, + "step": 4090 + }, + { + "epoch": 0.3373464170858415, + "grad_norm": 1.943923830986023, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6871616840362549, + "num_tokens": 13497520.0, + "step": 4091 + }, + { + "epoch": 0.337428877710893, + "grad_norm": 2.403717041015625, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7537134885787964, + "num_tokens": 13519149.0, + "step": 4092 + }, + { + "epoch": 0.3375113383359446, + "grad_norm": 1.7860132455825806, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7364039421081543, + "num_tokens": 13557920.0, + "step": 4093 + }, + { + "epoch": 0.3375937989609961, + "grad_norm": 2.052780866622925, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.732620358467102, + "num_tokens": 13587363.0, + "step": 4094 + }, + { + "epoch": 0.3376762595860477, + "grad_norm": 1.9687026739120483, + "learning_rate": 1e-06, + "loss": 1.0388, + "mean_token_accuracy": 0.6910146474838257, + "num_tokens": 13621221.0, + "step": 4095 + }, + { + "epoch": 0.3377587202110992, + "grad_norm": 2.118025779724121, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7655901908874512, + "num_tokens": 13645852.0, + "step": 4096 + }, + { + "epoch": 0.3378411808361507, + "grad_norm": 2.005115270614624, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7377333641052246, + "num_tokens": 13678093.0, + "step": 4097 + }, + { + "epoch": 0.3379236414612023, + "grad_norm": 2.1365408897399902, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7653416395187378, + "num_tokens": 13703584.0, + "step": 4098 + }, + { + "epoch": 0.3380061020862538, + "grad_norm": 1.9707075357437134, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7454112768173218, + "num_tokens": 13732643.0, + "step": 4099 + }, + { + "epoch": 0.33808856271130533, + "grad_norm": 1.9663591384887695, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7293632626533508, + "num_tokens": 13762438.0, + "step": 4100 + }, + { + "epoch": 0.3381710233363569, + "grad_norm": 2.002201557159424, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7161230444908142, + "num_tokens": 13794935.0, + "step": 4101 + }, + { + "epoch": 0.3382534839614084, + "grad_norm": 2.2014219760894775, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.708838701248169, + "num_tokens": 13822163.0, + "step": 4102 + }, + { + "epoch": 0.33833594458646, + "grad_norm": 2.0027949810028076, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.78421550989151, + "num_tokens": 13847166.0, + "step": 4103 + }, + { + "epoch": 0.3384184052115115, + "grad_norm": 2.040842056274414, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7578505277633667, + "num_tokens": 13875716.0, + "step": 4104 + }, + { + "epoch": 0.338500865836563, + "grad_norm": 1.8644169569015503, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.7667796611785889, + "num_tokens": 13907760.0, + "step": 4105 + }, + { + "epoch": 0.3385833264616146, + "grad_norm": 2.22159481048584, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7483310699462891, + "num_tokens": 13932026.0, + "step": 4106 + }, + { + "epoch": 0.3386657870866661, + "grad_norm": 2.0660359859466553, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7709512710571289, + "num_tokens": 13957989.0, + "step": 4107 + }, + { + "epoch": 0.33874824771171763, + "grad_norm": 1.9799630641937256, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7386093139648438, + "num_tokens": 13990816.0, + "step": 4108 + }, + { + "epoch": 0.3388307083367692, + "grad_norm": 2.3832387924194336, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7495872974395752, + "num_tokens": 14010949.0, + "step": 4109 + }, + { + "epoch": 0.3389131689618207, + "grad_norm": 1.881340503692627, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7543914318084717, + "num_tokens": 14041875.0, + "step": 4110 + }, + { + "epoch": 0.3389956295868723, + "grad_norm": 2.515925884246826, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7318582534790039, + "num_tokens": 14063242.0, + "step": 4111 + }, + { + "epoch": 0.3390780902119238, + "grad_norm": 1.8935762643814087, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7098196148872375, + "num_tokens": 14097520.0, + "step": 4112 + }, + { + "epoch": 0.3391605508369753, + "grad_norm": 1.9487656354904175, + "learning_rate": 1e-06, + "loss": 0.7798, + "mean_token_accuracy": 0.7600928544998169, + "num_tokens": 14128097.0, + "step": 4113 + }, + { + "epoch": 0.3392430114620269, + "grad_norm": 2.0186662673950195, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7537108063697815, + "num_tokens": 14158652.0, + "step": 4114 + }, + { + "epoch": 0.3393254720870784, + "grad_norm": 2.1141319274902344, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7481704950332642, + "num_tokens": 14187070.0, + "step": 4115 + }, + { + "epoch": 0.33940793271212993, + "grad_norm": 2.1588783264160156, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7446889281272888, + "num_tokens": 14215717.0, + "step": 4116 + }, + { + "epoch": 0.3394903933371815, + "grad_norm": 1.9130825996398926, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7760100960731506, + "num_tokens": 14248955.0, + "step": 4117 + }, + { + "epoch": 0.339572853962233, + "grad_norm": 2.3025169372558594, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7480683326721191, + "num_tokens": 14270111.0, + "step": 4118 + }, + { + "epoch": 0.3396553145872846, + "grad_norm": 1.964429259300232, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7639222145080566, + "num_tokens": 14302025.0, + "step": 4119 + }, + { + "epoch": 0.3397377752123361, + "grad_norm": 1.9456753730773926, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7439035177230835, + "num_tokens": 14331484.0, + "step": 4120 + }, + { + "epoch": 0.33982023583738763, + "grad_norm": 1.8247720003128052, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7483526468276978, + "num_tokens": 14364469.0, + "step": 4121 + }, + { + "epoch": 0.3399026964624392, + "grad_norm": 1.954756498336792, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7148516178131104, + "num_tokens": 14397948.0, + "step": 4122 + }, + { + "epoch": 0.3399851570874907, + "grad_norm": 1.8398274183273315, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7447561025619507, + "num_tokens": 14431743.0, + "step": 4123 + }, + { + "epoch": 0.34006761771254224, + "grad_norm": 1.7602463960647583, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7743332386016846, + "num_tokens": 14466799.0, + "step": 4124 + }, + { + "epoch": 0.3401500783375938, + "grad_norm": 1.8997173309326172, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.758050799369812, + "num_tokens": 14499615.0, + "step": 4125 + }, + { + "epoch": 0.3402325389626453, + "grad_norm": 1.8907283544540405, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.6997354030609131, + "num_tokens": 14530940.0, + "step": 4126 + }, + { + "epoch": 0.3403149995876969, + "grad_norm": 2.1855037212371826, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7067452669143677, + "num_tokens": 14559824.0, + "step": 4127 + }, + { + "epoch": 0.3403974602127484, + "grad_norm": 2.128713369369507, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7714755535125732, + "num_tokens": 14584819.0, + "step": 4128 + }, + { + "epoch": 0.34047992083779993, + "grad_norm": 2.4533886909484863, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7048160433769226, + "num_tokens": 14605694.0, + "step": 4129 + }, + { + "epoch": 0.3405623814628515, + "grad_norm": 1.8982486724853516, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.724634051322937, + "num_tokens": 14640387.0, + "step": 4130 + }, + { + "epoch": 0.340644842087903, + "grad_norm": 1.8287242650985718, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7548524737358093, + "num_tokens": 14676657.0, + "step": 4131 + }, + { + "epoch": 0.34072730271295454, + "grad_norm": 1.9892454147338867, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7596617937088013, + "num_tokens": 14707455.0, + "step": 4132 + }, + { + "epoch": 0.3408097633380061, + "grad_norm": 1.7428369522094727, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.7968963384628296, + "num_tokens": 14741786.0, + "step": 4133 + }, + { + "epoch": 0.34089222396305763, + "grad_norm": 1.906328797340393, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7782208919525146, + "num_tokens": 14770732.0, + "step": 4134 + }, + { + "epoch": 0.3409746845881092, + "grad_norm": 2.062516450881958, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7493890523910522, + "num_tokens": 14800897.0, + "step": 4135 + }, + { + "epoch": 0.3410571452131607, + "grad_norm": 1.9516024589538574, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7411471605300903, + "num_tokens": 14832014.0, + "step": 4136 + }, + { + "epoch": 0.34113960583821223, + "grad_norm": 2.431234359741211, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7443424463272095, + "num_tokens": 14852710.0, + "step": 4137 + }, + { + "epoch": 0.3412220664632638, + "grad_norm": 2.0788962841033936, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7716089487075806, + "num_tokens": 14883969.0, + "step": 4138 + }, + { + "epoch": 0.3413045270883153, + "grad_norm": 1.911133050918579, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7407208681106567, + "num_tokens": 14917317.0, + "step": 4139 + }, + { + "epoch": 0.34138698771336684, + "grad_norm": 2.173417806625366, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.729099690914154, + "num_tokens": 14943677.0, + "step": 4140 + }, + { + "epoch": 0.3414694483384184, + "grad_norm": 1.8784160614013672, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7523708343505859, + "num_tokens": 14974657.0, + "step": 4141 + }, + { + "epoch": 0.34155190896346993, + "grad_norm": 2.2023279666900635, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.6979787349700928, + "num_tokens": 15002061.0, + "step": 4142 + }, + { + "epoch": 0.3416343695885215, + "grad_norm": 2.07110333442688, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7141896486282349, + "num_tokens": 15029638.0, + "step": 4143 + }, + { + "epoch": 0.341716830213573, + "grad_norm": 1.810882568359375, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.756630003452301, + "num_tokens": 15061982.0, + "step": 4144 + }, + { + "epoch": 0.34179929083862454, + "grad_norm": 2.0300679206848145, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7673635482788086, + "num_tokens": 15088546.0, + "step": 4145 + }, + { + "epoch": 0.3418817514636761, + "grad_norm": 1.9118789434432983, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7811881303787231, + "num_tokens": 15116508.0, + "step": 4146 + }, + { + "epoch": 0.3419642120887276, + "grad_norm": 1.8956800699234009, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.7878227829933167, + "num_tokens": 15149280.0, + "step": 4147 + }, + { + "epoch": 0.34204667271377914, + "grad_norm": 2.148411989212036, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7234569787979126, + "num_tokens": 15178872.0, + "step": 4148 + }, + { + "epoch": 0.3421291333388307, + "grad_norm": 2.290855646133423, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7254047989845276, + "num_tokens": 15203453.0, + "step": 4149 + }, + { + "epoch": 0.34221159396388223, + "grad_norm": 2.201277256011963, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7217464447021484, + "num_tokens": 15228960.0, + "step": 4150 + }, + { + "epoch": 0.3422940545889338, + "grad_norm": 1.9227771759033203, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7442954778671265, + "num_tokens": 15262815.0, + "step": 4151 + }, + { + "epoch": 0.3423765152139853, + "grad_norm": 2.154203176498413, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7544879913330078, + "num_tokens": 15286254.0, + "step": 4152 + }, + { + "epoch": 0.34245897583903684, + "grad_norm": 1.9549741744995117, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.743559718132019, + "num_tokens": 15316719.0, + "step": 4153 + }, + { + "epoch": 0.3425414364640884, + "grad_norm": 2.0775437355041504, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6887039542198181, + "num_tokens": 15346773.0, + "step": 4154 + }, + { + "epoch": 0.34262389708913993, + "grad_norm": 1.7854558229446411, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7598472833633423, + "num_tokens": 15381823.0, + "step": 4155 + }, + { + "epoch": 0.3427063577141915, + "grad_norm": 2.032078266143799, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7747058868408203, + "num_tokens": 15408025.0, + "step": 4156 + }, + { + "epoch": 0.342788818339243, + "grad_norm": 2.0124549865722656, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.7587926387786865, + "num_tokens": 15436815.0, + "step": 4157 + }, + { + "epoch": 0.34287127896429453, + "grad_norm": 2.0511367321014404, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7621563673019409, + "num_tokens": 15464422.0, + "step": 4158 + }, + { + "epoch": 0.3429537395893461, + "grad_norm": 1.8261021375656128, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7301152944564819, + "num_tokens": 15500376.0, + "step": 4159 + }, + { + "epoch": 0.3430362002143976, + "grad_norm": 1.939175009727478, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7115721702575684, + "num_tokens": 15532730.0, + "step": 4160 + }, + { + "epoch": 0.34311866083944914, + "grad_norm": 2.0174031257629395, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7396074533462524, + "num_tokens": 15560773.0, + "step": 4161 + }, + { + "epoch": 0.3432011214645007, + "grad_norm": 2.1684794425964355, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7169982194900513, + "num_tokens": 15587840.0, + "step": 4162 + }, + { + "epoch": 0.34328358208955223, + "grad_norm": 1.8242802619934082, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7535927891731262, + "num_tokens": 15621182.0, + "step": 4163 + }, + { + "epoch": 0.3433660427146038, + "grad_norm": 1.8226038217544556, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7331154942512512, + "num_tokens": 15657258.0, + "step": 4164 + }, + { + "epoch": 0.3434485033396553, + "grad_norm": 2.0192768573760986, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7899566292762756, + "num_tokens": 15683446.0, + "step": 4165 + }, + { + "epoch": 0.34353096396470684, + "grad_norm": 1.8919003009796143, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7529643177986145, + "num_tokens": 15717235.0, + "step": 4166 + }, + { + "epoch": 0.3436134245897584, + "grad_norm": 1.7100356817245483, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.771395206451416, + "num_tokens": 15753732.0, + "step": 4167 + }, + { + "epoch": 0.3436958852148099, + "grad_norm": 2.045948028564453, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.7450893521308899, + "num_tokens": 15782714.0, + "step": 4168 + }, + { + "epoch": 0.34377834583986144, + "grad_norm": 2.0972065925598145, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7279316186904907, + "num_tokens": 15810051.0, + "step": 4169 + }, + { + "epoch": 0.343860806464913, + "grad_norm": 1.9707622528076172, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7296735048294067, + "num_tokens": 15839975.0, + "step": 4170 + }, + { + "epoch": 0.34394326708996453, + "grad_norm": 1.8484843969345093, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.765365719795227, + "num_tokens": 15870404.0, + "step": 4171 + }, + { + "epoch": 0.3440257277150161, + "grad_norm": 1.9931796789169312, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7372852563858032, + "num_tokens": 15901245.0, + "step": 4172 + }, + { + "epoch": 0.3441081883400676, + "grad_norm": 1.813077688217163, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7523853778839111, + "num_tokens": 15937905.0, + "step": 4173 + }, + { + "epoch": 0.34419064896511914, + "grad_norm": 1.9944877624511719, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7622108459472656, + "num_tokens": 15967467.0, + "step": 4174 + }, + { + "epoch": 0.3442731095901707, + "grad_norm": 1.767818808555603, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7308640480041504, + "num_tokens": 16003683.0, + "step": 4175 + }, + { + "epoch": 0.34435557021522223, + "grad_norm": 2.499929904937744, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.748325765132904, + "num_tokens": 16022218.0, + "step": 4176 + }, + { + "epoch": 0.34443803084027375, + "grad_norm": 2.0942084789276123, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.75907963514328, + "num_tokens": 16048548.0, + "step": 4177 + }, + { + "epoch": 0.3445204914653253, + "grad_norm": 2.078951120376587, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7236821055412292, + "num_tokens": 16076026.0, + "step": 4178 + }, + { + "epoch": 0.34460295209037684, + "grad_norm": 1.9843205213546753, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7546802759170532, + "num_tokens": 16106917.0, + "step": 4179 + }, + { + "epoch": 0.3446854127154284, + "grad_norm": 2.0451345443725586, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7678638696670532, + "num_tokens": 16132477.0, + "step": 4180 + }, + { + "epoch": 0.3447678733404799, + "grad_norm": 2.0714094638824463, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7089669704437256, + "num_tokens": 16161881.0, + "step": 4181 + }, + { + "epoch": 0.34485033396553144, + "grad_norm": 1.7436785697937012, + "learning_rate": 1e-06, + "loss": 0.7619, + "mean_token_accuracy": 0.7587804794311523, + "num_tokens": 16202633.0, + "step": 4182 + }, + { + "epoch": 0.344932794590583, + "grad_norm": 1.8053559064865112, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7665276527404785, + "num_tokens": 16238636.0, + "step": 4183 + }, + { + "epoch": 0.34501525521563453, + "grad_norm": 2.198134422302246, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7368192076683044, + "num_tokens": 16265937.0, + "step": 4184 + }, + { + "epoch": 0.34509771584068605, + "grad_norm": 1.8612613677978516, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.755900502204895, + "num_tokens": 16298512.0, + "step": 4185 + }, + { + "epoch": 0.3451801764657376, + "grad_norm": 1.8741439580917358, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7438681125640869, + "num_tokens": 16334026.0, + "step": 4186 + }, + { + "epoch": 0.34526263709078914, + "grad_norm": 1.9771825075149536, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7783638834953308, + "num_tokens": 16364228.0, + "step": 4187 + }, + { + "epoch": 0.3453450977158407, + "grad_norm": 2.193296432495117, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7434567213058472, + "num_tokens": 16389742.0, + "step": 4188 + }, + { + "epoch": 0.3454275583408922, + "grad_norm": 1.8526643514633179, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7860519886016846, + "num_tokens": 16424804.0, + "step": 4189 + }, + { + "epoch": 0.34551001896594374, + "grad_norm": 1.817051887512207, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7711822390556335, + "num_tokens": 16460945.0, + "step": 4190 + }, + { + "epoch": 0.3455924795909953, + "grad_norm": 1.9808769226074219, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7600809931755066, + "num_tokens": 16493431.0, + "step": 4191 + }, + { + "epoch": 0.34567494021604683, + "grad_norm": 2.2169103622436523, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7312208414077759, + "num_tokens": 16518691.0, + "step": 4192 + }, + { + "epoch": 0.34575740084109835, + "grad_norm": 2.1210761070251465, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7220889329910278, + "num_tokens": 16547378.0, + "step": 4193 + }, + { + "epoch": 0.3458398614661499, + "grad_norm": 2.0040106773376465, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7603229284286499, + "num_tokens": 16574159.0, + "step": 4194 + }, + { + "epoch": 0.34592232209120144, + "grad_norm": 1.940409541130066, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.735562801361084, + "num_tokens": 16604180.0, + "step": 4195 + }, + { + "epoch": 0.346004782716253, + "grad_norm": 1.9400080442428589, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7119576930999756, + "num_tokens": 16636011.0, + "step": 4196 + }, + { + "epoch": 0.34608724334130453, + "grad_norm": 2.0281291007995605, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7110536694526672, + "num_tokens": 16666349.0, + "step": 4197 + }, + { + "epoch": 0.34616970396635605, + "grad_norm": 1.8996763229370117, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7440813183784485, + "num_tokens": 16696906.0, + "step": 4198 + }, + { + "epoch": 0.3462521645914076, + "grad_norm": 2.365812301635742, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7502322793006897, + "num_tokens": 16719953.0, + "step": 4199 + }, + { + "epoch": 0.34633462521645914, + "grad_norm": 2.0002188682556152, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.7678103446960449, + "num_tokens": 16748919.0, + "step": 4200 + }, + { + "epoch": 0.34641708584151065, + "grad_norm": 1.8971620798110962, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7238109111785889, + "num_tokens": 16779961.0, + "step": 4201 + }, + { + "epoch": 0.3464995464665622, + "grad_norm": 2.0048928260803223, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7003642320632935, + "num_tokens": 16810911.0, + "step": 4202 + }, + { + "epoch": 0.34658200709161374, + "grad_norm": 2.085932731628418, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7543622255325317, + "num_tokens": 16837623.0, + "step": 4203 + }, + { + "epoch": 0.3466644677166653, + "grad_norm": 2.3010036945343018, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7154780626296997, + "num_tokens": 16861492.0, + "step": 4204 + }, + { + "epoch": 0.34674692834171683, + "grad_norm": 2.197550058364868, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7386809587478638, + "num_tokens": 16890105.0, + "step": 4205 + }, + { + "epoch": 0.34682938896676835, + "grad_norm": 1.9687825441360474, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7441574335098267, + "num_tokens": 16918528.0, + "step": 4206 + }, + { + "epoch": 0.3469118495918199, + "grad_norm": 2.18466854095459, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7470184564590454, + "num_tokens": 16945312.0, + "step": 4207 + }, + { + "epoch": 0.34699431021687144, + "grad_norm": 1.9383143186569214, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7347932457923889, + "num_tokens": 16978320.0, + "step": 4208 + }, + { + "epoch": 0.34707677084192295, + "grad_norm": 2.547818899154663, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7122806310653687, + "num_tokens": 17001487.0, + "step": 4209 + }, + { + "epoch": 0.3471592314669745, + "grad_norm": 2.060460329055786, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7605123519897461, + "num_tokens": 17029416.0, + "step": 4210 + }, + { + "epoch": 0.34724169209202604, + "grad_norm": 2.1247379779815674, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7123477458953857, + "num_tokens": 17058060.0, + "step": 4211 + }, + { + "epoch": 0.3473241527170776, + "grad_norm": 2.042602062225342, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7522091269493103, + "num_tokens": 17086612.0, + "step": 4212 + }, + { + "epoch": 0.34740661334212913, + "grad_norm": 1.996091365814209, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7536786198616028, + "num_tokens": 17116073.0, + "step": 4213 + }, + { + "epoch": 0.34748907396718065, + "grad_norm": 2.023385763168335, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7638709545135498, + "num_tokens": 17145590.0, + "step": 4214 + }, + { + "epoch": 0.3475715345922322, + "grad_norm": 2.061302423477173, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.744481086730957, + "num_tokens": 17176056.0, + "step": 4215 + }, + { + "epoch": 0.34765399521728374, + "grad_norm": 1.8936336040496826, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7302868366241455, + "num_tokens": 17212349.0, + "step": 4216 + }, + { + "epoch": 0.3477364558423353, + "grad_norm": 1.942803978919983, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7452256679534912, + "num_tokens": 17242078.0, + "step": 4217 + }, + { + "epoch": 0.34781891646738683, + "grad_norm": 1.8618196249008179, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7327569723129272, + "num_tokens": 17275730.0, + "step": 4218 + }, + { + "epoch": 0.34790137709243835, + "grad_norm": 2.010922431945801, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7225664854049683, + "num_tokens": 17308266.0, + "step": 4219 + }, + { + "epoch": 0.3479838377174899, + "grad_norm": 2.4664227962493896, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6938610076904297, + "num_tokens": 17331850.0, + "step": 4220 + }, + { + "epoch": 0.34806629834254144, + "grad_norm": 1.8488194942474365, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7278721332550049, + "num_tokens": 17367702.0, + "step": 4221 + }, + { + "epoch": 0.34814875896759295, + "grad_norm": 1.9696425199508667, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7772297859191895, + "num_tokens": 17397419.0, + "step": 4222 + }, + { + "epoch": 0.3482312195926445, + "grad_norm": 1.8293529748916626, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7695722579956055, + "num_tokens": 17430727.0, + "step": 4223 + }, + { + "epoch": 0.34831368021769604, + "grad_norm": 2.1020541191101074, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7491695880889893, + "num_tokens": 17457599.0, + "step": 4224 + }, + { + "epoch": 0.3483961408427476, + "grad_norm": 1.7934426069259644, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7407142519950867, + "num_tokens": 17494970.0, + "step": 4225 + }, + { + "epoch": 0.34847860146779913, + "grad_norm": 2.2281904220581055, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7267153263092041, + "num_tokens": 17520716.0, + "step": 4226 + }, + { + "epoch": 0.34856106209285065, + "grad_norm": 1.7356189489364624, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7532022595405579, + "num_tokens": 17558343.0, + "step": 4227 + }, + { + "epoch": 0.3486435227179022, + "grad_norm": 2.033177375793457, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7017377614974976, + "num_tokens": 17587772.0, + "step": 4228 + }, + { + "epoch": 0.34872598334295374, + "grad_norm": 1.8780310153961182, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7496581673622131, + "num_tokens": 17621514.0, + "step": 4229 + }, + { + "epoch": 0.34880844396800526, + "grad_norm": 2.0253281593322754, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7717608213424683, + "num_tokens": 17649037.0, + "step": 4230 + }, + { + "epoch": 0.3488909045930568, + "grad_norm": 1.9988839626312256, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.754417896270752, + "num_tokens": 17678390.0, + "step": 4231 + }, + { + "epoch": 0.34897336521810834, + "grad_norm": 2.139881134033203, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7504613399505615, + "num_tokens": 17704181.0, + "step": 4232 + }, + { + "epoch": 0.3490558258431599, + "grad_norm": 1.8591639995574951, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7723305821418762, + "num_tokens": 17736336.0, + "step": 4233 + }, + { + "epoch": 0.34913828646821143, + "grad_norm": 1.896045207977295, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7368184924125671, + "num_tokens": 17765364.0, + "step": 4234 + }, + { + "epoch": 0.34922074709326295, + "grad_norm": 1.9258774518966675, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7383290529251099, + "num_tokens": 17798731.0, + "step": 4235 + }, + { + "epoch": 0.3493032077183145, + "grad_norm": 2.353001356124878, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7135419845581055, + "num_tokens": 17824054.0, + "step": 4236 + }, + { + "epoch": 0.34938566834336604, + "grad_norm": 1.7367898225784302, + "learning_rate": 1e-06, + "loss": 0.5597, + "mean_token_accuracy": 0.7966752052307129, + "num_tokens": 17858008.0, + "step": 4237 + }, + { + "epoch": 0.34946812896841756, + "grad_norm": 1.91355562210083, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7674567103385925, + "num_tokens": 17891428.0, + "step": 4238 + }, + { + "epoch": 0.34955058959346913, + "grad_norm": 2.1037609577178955, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7649412155151367, + "num_tokens": 17916802.0, + "step": 4239 + }, + { + "epoch": 0.34963305021852065, + "grad_norm": 1.919386386871338, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7247952222824097, + "num_tokens": 17950252.0, + "step": 4240 + }, + { + "epoch": 0.3497155108435722, + "grad_norm": 2.226016044616699, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7335644960403442, + "num_tokens": 17973337.0, + "step": 4241 + }, + { + "epoch": 0.34979797146862374, + "grad_norm": 2.26694917678833, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7502262592315674, + "num_tokens": 17997230.0, + "step": 4242 + }, + { + "epoch": 0.34988043209367525, + "grad_norm": 2.0365796089172363, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7101060748100281, + "num_tokens": 18028699.0, + "step": 4243 + }, + { + "epoch": 0.3499628927187268, + "grad_norm": 2.2598869800567627, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7191507816314697, + "num_tokens": 18052244.0, + "step": 4244 + }, + { + "epoch": 0.35004535334377834, + "grad_norm": 1.9570231437683105, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7598515748977661, + "num_tokens": 18085501.0, + "step": 4245 + }, + { + "epoch": 0.35012781396882986, + "grad_norm": 1.8510774374008179, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7477656602859497, + "num_tokens": 18119091.0, + "step": 4246 + }, + { + "epoch": 0.35021027459388143, + "grad_norm": 2.0759856700897217, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7388136386871338, + "num_tokens": 18147030.0, + "step": 4247 + }, + { + "epoch": 0.35029273521893295, + "grad_norm": 1.9701811075210571, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.775196373462677, + "num_tokens": 18175917.0, + "step": 4248 + }, + { + "epoch": 0.3503751958439845, + "grad_norm": 2.2051053047180176, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7680213451385498, + "num_tokens": 18199889.0, + "step": 4249 + }, + { + "epoch": 0.35045765646903604, + "grad_norm": 1.9754234552383423, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.7972632646560669, + "num_tokens": 18229179.0, + "step": 4250 + }, + { + "epoch": 0.35054011709408756, + "grad_norm": 1.886425256729126, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.753153920173645, + "num_tokens": 18261125.0, + "step": 4251 + }, + { + "epoch": 0.35062257771913913, + "grad_norm": 2.1680798530578613, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7359156608581543, + "num_tokens": 18287711.0, + "step": 4252 + }, + { + "epoch": 0.35070503834419064, + "grad_norm": 1.7009645700454712, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.800686240196228, + "num_tokens": 18328128.0, + "step": 4253 + }, + { + "epoch": 0.35078749896924216, + "grad_norm": 2.1485233306884766, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7037627100944519, + "num_tokens": 18357401.0, + "step": 4254 + }, + { + "epoch": 0.35086995959429373, + "grad_norm": 1.8325307369232178, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7881468534469604, + "num_tokens": 18389873.0, + "step": 4255 + }, + { + "epoch": 0.35095242021934525, + "grad_norm": 2.1662068367004395, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.74064701795578, + "num_tokens": 18416596.0, + "step": 4256 + }, + { + "epoch": 0.3510348808443968, + "grad_norm": 1.976177453994751, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7443533539772034, + "num_tokens": 18446412.0, + "step": 4257 + }, + { + "epoch": 0.35111734146944834, + "grad_norm": 1.9187946319580078, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.732848584651947, + "num_tokens": 18477652.0, + "step": 4258 + }, + { + "epoch": 0.35119980209449986, + "grad_norm": 2.160705804824829, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7494865655899048, + "num_tokens": 18504050.0, + "step": 4259 + }, + { + "epoch": 0.35128226271955143, + "grad_norm": 1.7423869371414185, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7693935632705688, + "num_tokens": 18536787.0, + "step": 4260 + }, + { + "epoch": 0.35136472334460295, + "grad_norm": 2.253708600997925, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7287690043449402, + "num_tokens": 18562092.0, + "step": 4261 + }, + { + "epoch": 0.35144718396965446, + "grad_norm": 2.187295913696289, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7248693704605103, + "num_tokens": 18590905.0, + "step": 4262 + }, + { + "epoch": 0.35152964459470604, + "grad_norm": 1.9648069143295288, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7179433107376099, + "num_tokens": 18620906.0, + "step": 4263 + }, + { + "epoch": 0.35161210521975755, + "grad_norm": 2.071974039077759, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7420120239257812, + "num_tokens": 18649215.0, + "step": 4264 + }, + { + "epoch": 0.3516945658448091, + "grad_norm": 1.8933303356170654, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.743389368057251, + "num_tokens": 18682839.0, + "step": 4265 + }, + { + "epoch": 0.35177702646986064, + "grad_norm": 1.9396034479141235, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7664414644241333, + "num_tokens": 18713867.0, + "step": 4266 + }, + { + "epoch": 0.35185948709491216, + "grad_norm": 2.10311222076416, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7074747085571289, + "num_tokens": 18744384.0, + "step": 4267 + }, + { + "epoch": 0.35194194771996373, + "grad_norm": 1.9129586219787598, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7668238282203674, + "num_tokens": 18772806.0, + "step": 4268 + }, + { + "epoch": 0.35202440834501525, + "grad_norm": 1.6612013578414917, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.792216956615448, + "num_tokens": 18808171.0, + "step": 4269 + }, + { + "epoch": 0.35210686897006677, + "grad_norm": 1.874272346496582, + "learning_rate": 1e-06, + "loss": 0.7704, + "mean_token_accuracy": 0.7530257105827332, + "num_tokens": 18840468.0, + "step": 4270 + }, + { + "epoch": 0.35218932959511834, + "grad_norm": 2.2095582485198975, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7370610237121582, + "num_tokens": 18865254.0, + "step": 4271 + }, + { + "epoch": 0.35227179022016986, + "grad_norm": 2.100565195083618, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7304604053497314, + "num_tokens": 18893072.0, + "step": 4272 + }, + { + "epoch": 0.35235425084522143, + "grad_norm": 1.7874027490615845, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7849964499473572, + "num_tokens": 18927765.0, + "step": 4273 + }, + { + "epoch": 0.35243671147027295, + "grad_norm": 1.9966883659362793, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7105520963668823, + "num_tokens": 18962322.0, + "step": 4274 + }, + { + "epoch": 0.35251917209532446, + "grad_norm": 2.049320697784424, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.723524808883667, + "num_tokens": 18991291.0, + "step": 4275 + }, + { + "epoch": 0.35260163272037603, + "grad_norm": 1.990423560142517, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7568526268005371, + "num_tokens": 19018509.0, + "step": 4276 + }, + { + "epoch": 0.35268409334542755, + "grad_norm": 1.9906853437423706, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7061295509338379, + "num_tokens": 19050452.0, + "step": 4277 + }, + { + "epoch": 0.35276655397047907, + "grad_norm": 1.918666124343872, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7693122625350952, + "num_tokens": 19080724.0, + "step": 4278 + }, + { + "epoch": 0.35284901459553064, + "grad_norm": 1.8607591390609741, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7595651745796204, + "num_tokens": 19112356.0, + "step": 4279 + }, + { + "epoch": 0.35293147522058216, + "grad_norm": 2.253277063369751, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7029975652694702, + "num_tokens": 19140198.0, + "step": 4280 + }, + { + "epoch": 0.35301393584563373, + "grad_norm": 1.8419480323791504, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7362322807312012, + "num_tokens": 19174939.0, + "step": 4281 + }, + { + "epoch": 0.35309639647068525, + "grad_norm": 2.596924066543579, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.71384596824646, + "num_tokens": 19197520.0, + "step": 4282 + }, + { + "epoch": 0.35317885709573676, + "grad_norm": 1.889517068862915, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7587549686431885, + "num_tokens": 19227583.0, + "step": 4283 + }, + { + "epoch": 0.35326131772078834, + "grad_norm": 2.094043493270874, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7255658507347107, + "num_tokens": 19257015.0, + "step": 4284 + }, + { + "epoch": 0.35334377834583985, + "grad_norm": 1.9544116258621216, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7399964332580566, + "num_tokens": 19290612.0, + "step": 4285 + }, + { + "epoch": 0.3534262389708914, + "grad_norm": 1.6897324323654175, + "learning_rate": 1e-06, + "loss": 0.7746, + "mean_token_accuracy": 0.7550992965698242, + "num_tokens": 19327783.0, + "step": 4286 + }, + { + "epoch": 0.35350869959594294, + "grad_norm": 2.044617176055908, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7132371068000793, + "num_tokens": 19357175.0, + "step": 4287 + }, + { + "epoch": 0.35359116022099446, + "grad_norm": 1.8097398281097412, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7090556621551514, + "num_tokens": 19395215.0, + "step": 4288 + }, + { + "epoch": 0.35367362084604603, + "grad_norm": 2.247114896774292, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7725646495819092, + "num_tokens": 19420444.0, + "step": 4289 + }, + { + "epoch": 0.35375608147109755, + "grad_norm": 1.9009983539581299, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7234050035476685, + "num_tokens": 19454897.0, + "step": 4290 + }, + { + "epoch": 0.35383854209614907, + "grad_norm": 1.8743857145309448, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7302147150039673, + "num_tokens": 19485989.0, + "step": 4291 + }, + { + "epoch": 0.35392100272120064, + "grad_norm": 2.029127597808838, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7359235286712646, + "num_tokens": 19514995.0, + "step": 4292 + }, + { + "epoch": 0.35400346334625216, + "grad_norm": 2.019270658493042, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.730999767780304, + "num_tokens": 19544124.0, + "step": 4293 + }, + { + "epoch": 0.35408592397130373, + "grad_norm": 2.0577914714813232, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7450721263885498, + "num_tokens": 19575485.0, + "step": 4294 + }, + { + "epoch": 0.35416838459635525, + "grad_norm": 1.8530412912368774, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7539930939674377, + "num_tokens": 19607662.0, + "step": 4295 + }, + { + "epoch": 0.35425084522140676, + "grad_norm": 1.887044072151184, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7327812910079956, + "num_tokens": 19640823.0, + "step": 4296 + }, + { + "epoch": 0.35433330584645834, + "grad_norm": 1.9714423418045044, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7738327980041504, + "num_tokens": 19668317.0, + "step": 4297 + }, + { + "epoch": 0.35441576647150985, + "grad_norm": 1.9274507761001587, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7620831727981567, + "num_tokens": 19700017.0, + "step": 4298 + }, + { + "epoch": 0.35449822709656137, + "grad_norm": 1.8117287158966064, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.743236780166626, + "num_tokens": 19734113.0, + "step": 4299 + }, + { + "epoch": 0.35458068772161294, + "grad_norm": 1.8130451440811157, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7624690532684326, + "num_tokens": 19767414.0, + "step": 4300 + }, + { + "epoch": 0.35466314834666446, + "grad_norm": 1.9221569299697876, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7484270334243774, + "num_tokens": 19800196.0, + "step": 4301 + }, + { + "epoch": 0.35474560897171603, + "grad_norm": 1.8112123012542725, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7179707288742065, + "num_tokens": 19837960.0, + "step": 4302 + }, + { + "epoch": 0.35482806959676755, + "grad_norm": 2.0769264698028564, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7358995079994202, + "num_tokens": 19866755.0, + "step": 4303 + }, + { + "epoch": 0.35491053022181906, + "grad_norm": 2.0627357959747314, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7309679388999939, + "num_tokens": 19896095.0, + "step": 4304 + }, + { + "epoch": 0.35499299084687064, + "grad_norm": 1.9124253988265991, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7811604738235474, + "num_tokens": 19925691.0, + "step": 4305 + }, + { + "epoch": 0.35507545147192215, + "grad_norm": 2.1194264888763428, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7286962270736694, + "num_tokens": 19953739.0, + "step": 4306 + }, + { + "epoch": 0.35515791209697367, + "grad_norm": 2.1709463596343994, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7246153950691223, + "num_tokens": 19982422.0, + "step": 4307 + }, + { + "epoch": 0.35524037272202524, + "grad_norm": 1.9890868663787842, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7573543190956116, + "num_tokens": 20013367.0, + "step": 4308 + }, + { + "epoch": 0.35532283334707676, + "grad_norm": 2.1488821506500244, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7706975936889648, + "num_tokens": 20036225.0, + "step": 4309 + }, + { + "epoch": 0.35540529397212833, + "grad_norm": 2.017958164215088, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7706112861633301, + "num_tokens": 20066157.0, + "step": 4310 + }, + { + "epoch": 0.35548775459717985, + "grad_norm": 1.8704586029052734, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7455239295959473, + "num_tokens": 20099460.0, + "step": 4311 + }, + { + "epoch": 0.35557021522223137, + "grad_norm": 1.910558819770813, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7678547501564026, + "num_tokens": 20131813.0, + "step": 4312 + }, + { + "epoch": 0.35565267584728294, + "grad_norm": 2.0942437648773193, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7344822883605957, + "num_tokens": 20159471.0, + "step": 4313 + }, + { + "epoch": 0.35573513647233446, + "grad_norm": 1.9825305938720703, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7177747488021851, + "num_tokens": 20192235.0, + "step": 4314 + }, + { + "epoch": 0.355817597097386, + "grad_norm": 2.151466131210327, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7333836555480957, + "num_tokens": 20218757.0, + "step": 4315 + }, + { + "epoch": 0.35590005772243755, + "grad_norm": 1.8988512754440308, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7590590119361877, + "num_tokens": 20252229.0, + "step": 4316 + }, + { + "epoch": 0.35598251834748906, + "grad_norm": 2.004206895828247, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7291209697723389, + "num_tokens": 20280762.0, + "step": 4317 + }, + { + "epoch": 0.35606497897254064, + "grad_norm": 2.1956260204315186, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7318029999732971, + "num_tokens": 20305574.0, + "step": 4318 + }, + { + "epoch": 0.35614743959759215, + "grad_norm": 1.8862892389297485, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7306212186813354, + "num_tokens": 20338049.0, + "step": 4319 + }, + { + "epoch": 0.35622990022264367, + "grad_norm": 1.9128872156143188, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7390120029449463, + "num_tokens": 20372128.0, + "step": 4320 + }, + { + "epoch": 0.35631236084769524, + "grad_norm": 1.9073189496994019, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7634201645851135, + "num_tokens": 20405098.0, + "step": 4321 + }, + { + "epoch": 0.35639482147274676, + "grad_norm": 1.9019485712051392, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7667742967605591, + "num_tokens": 20433942.0, + "step": 4322 + }, + { + "epoch": 0.3564772820977983, + "grad_norm": 1.9020085334777832, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7171604633331299, + "num_tokens": 20467706.0, + "step": 4323 + }, + { + "epoch": 0.35655974272284985, + "grad_norm": 1.8890444040298462, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7647403478622437, + "num_tokens": 20500756.0, + "step": 4324 + }, + { + "epoch": 0.35664220334790137, + "grad_norm": 1.8295602798461914, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7144463062286377, + "num_tokens": 20534397.0, + "step": 4325 + }, + { + "epoch": 0.35672466397295294, + "grad_norm": 2.337528944015503, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7377806901931763, + "num_tokens": 20556208.0, + "step": 4326 + }, + { + "epoch": 0.35680712459800445, + "grad_norm": 2.50561785697937, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7335819005966187, + "num_tokens": 20576854.0, + "step": 4327 + }, + { + "epoch": 0.35688958522305597, + "grad_norm": 2.3808891773223877, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7386393547058105, + "num_tokens": 20601014.0, + "step": 4328 + }, + { + "epoch": 0.35697204584810754, + "grad_norm": 1.9900493621826172, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7384819388389587, + "num_tokens": 20629823.0, + "step": 4329 + }, + { + "epoch": 0.35705450647315906, + "grad_norm": 2.0451159477233887, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7209982872009277, + "num_tokens": 20656083.0, + "step": 4330 + }, + { + "epoch": 0.3571369670982106, + "grad_norm": 1.82675039768219, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7482456564903259, + "num_tokens": 20695134.0, + "step": 4331 + }, + { + "epoch": 0.35721942772326215, + "grad_norm": 3.2290186882019043, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.767061710357666, + "num_tokens": 20731938.0, + "step": 4332 + }, + { + "epoch": 0.35730188834831367, + "grad_norm": 1.8359792232513428, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7751194834709167, + "num_tokens": 20764122.0, + "step": 4333 + }, + { + "epoch": 0.35738434897336524, + "grad_norm": 1.9704989194869995, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7326961755752563, + "num_tokens": 20794562.0, + "step": 4334 + }, + { + "epoch": 0.35746680959841676, + "grad_norm": 2.0203278064727783, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7373857498168945, + "num_tokens": 20821202.0, + "step": 4335 + }, + { + "epoch": 0.3575492702234683, + "grad_norm": 1.9953285455703735, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.745546817779541, + "num_tokens": 20853432.0, + "step": 4336 + }, + { + "epoch": 0.35763173084851985, + "grad_norm": 1.8140525817871094, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7316727638244629, + "num_tokens": 20890857.0, + "step": 4337 + }, + { + "epoch": 0.35771419147357136, + "grad_norm": 1.7721387147903442, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7544053792953491, + "num_tokens": 20926924.0, + "step": 4338 + }, + { + "epoch": 0.3577966520986229, + "grad_norm": 1.8376483917236328, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7541728019714355, + "num_tokens": 20959992.0, + "step": 4339 + }, + { + "epoch": 0.35787911272367445, + "grad_norm": 1.7621982097625732, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7783464193344116, + "num_tokens": 20996829.0, + "step": 4340 + }, + { + "epoch": 0.35796157334872597, + "grad_norm": 2.028176784515381, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7471706867218018, + "num_tokens": 21025405.0, + "step": 4341 + }, + { + "epoch": 0.35804403397377754, + "grad_norm": 1.8654481172561646, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.7923035621643066, + "num_tokens": 21053088.0, + "step": 4342 + }, + { + "epoch": 0.35812649459882906, + "grad_norm": 1.9842464923858643, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7225470542907715, + "num_tokens": 21082348.0, + "step": 4343 + }, + { + "epoch": 0.3582089552238806, + "grad_norm": 2.2559330463409424, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7431381344795227, + "num_tokens": 21104109.0, + "step": 4344 + }, + { + "epoch": 0.35829141584893215, + "grad_norm": 2.067864179611206, + "learning_rate": 1e-06, + "loss": 0.6184, + "mean_token_accuracy": 0.7931959629058838, + "num_tokens": 21132328.0, + "step": 4345 + }, + { + "epoch": 0.35837387647398367, + "grad_norm": 2.019895553588867, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7718733549118042, + "num_tokens": 21156964.0, + "step": 4346 + }, + { + "epoch": 0.35845633709903524, + "grad_norm": 1.9152086973190308, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7652807235717773, + "num_tokens": 21187480.0, + "step": 4347 + }, + { + "epoch": 0.35853879772408676, + "grad_norm": 2.2792131900787354, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.757952868938446, + "num_tokens": 21210743.0, + "step": 4348 + }, + { + "epoch": 0.35862125834913827, + "grad_norm": 1.7392023801803589, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7415200471878052, + "num_tokens": 21246352.0, + "step": 4349 + }, + { + "epoch": 0.35870371897418984, + "grad_norm": 2.080324649810791, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7185118198394775, + "num_tokens": 21273989.0, + "step": 4350 + }, + { + "epoch": 0.35878617959924136, + "grad_norm": 2.23160719871521, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7135932445526123, + "num_tokens": 21298243.0, + "step": 4351 + }, + { + "epoch": 0.3588686402242929, + "grad_norm": 1.9543874263763428, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7435179948806763, + "num_tokens": 21330156.0, + "step": 4352 + }, + { + "epoch": 0.35895110084934445, + "grad_norm": 2.0753204822540283, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7592312693595886, + "num_tokens": 21356699.0, + "step": 4353 + }, + { + "epoch": 0.35903356147439597, + "grad_norm": 2.036489725112915, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7113775610923767, + "num_tokens": 21386706.0, + "step": 4354 + }, + { + "epoch": 0.35911602209944754, + "grad_norm": 1.9769339561462402, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7700846195220947, + "num_tokens": 21414949.0, + "step": 4355 + }, + { + "epoch": 0.35919848272449906, + "grad_norm": 1.8316833972930908, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7573271989822388, + "num_tokens": 21448652.0, + "step": 4356 + }, + { + "epoch": 0.3592809433495506, + "grad_norm": 2.154664993286133, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7189260721206665, + "num_tokens": 21477930.0, + "step": 4357 + }, + { + "epoch": 0.35936340397460215, + "grad_norm": 2.223198413848877, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7209771871566772, + "num_tokens": 21507689.0, + "step": 4358 + }, + { + "epoch": 0.35944586459965366, + "grad_norm": 2.118274450302124, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7077970504760742, + "num_tokens": 21538927.0, + "step": 4359 + }, + { + "epoch": 0.3595283252247052, + "grad_norm": 2.3014492988586426, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7631490230560303, + "num_tokens": 21561387.0, + "step": 4360 + }, + { + "epoch": 0.35961078584975675, + "grad_norm": 2.017456293106079, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.7426648139953613, + "num_tokens": 21592166.0, + "step": 4361 + }, + { + "epoch": 0.35969324647480827, + "grad_norm": 2.076568841934204, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7343654632568359, + "num_tokens": 21621073.0, + "step": 4362 + }, + { + "epoch": 0.35977570709985984, + "grad_norm": 2.0900635719299316, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7440634965896606, + "num_tokens": 21650426.0, + "step": 4363 + }, + { + "epoch": 0.35985816772491136, + "grad_norm": 2.0280587673187256, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7458400726318359, + "num_tokens": 21680691.0, + "step": 4364 + }, + { + "epoch": 0.3599406283499629, + "grad_norm": 2.0227644443511963, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7741501331329346, + "num_tokens": 21706959.0, + "step": 4365 + }, + { + "epoch": 0.36002308897501445, + "grad_norm": 1.9974877834320068, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7776656746864319, + "num_tokens": 21739237.0, + "step": 4366 + }, + { + "epoch": 0.36010554960006597, + "grad_norm": 1.7544647455215454, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7716035842895508, + "num_tokens": 21774860.0, + "step": 4367 + }, + { + "epoch": 0.3601880102251175, + "grad_norm": 1.8892459869384766, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7473273277282715, + "num_tokens": 21807427.0, + "step": 4368 + }, + { + "epoch": 0.36027047085016906, + "grad_norm": 1.9831901788711548, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7466470003128052, + "num_tokens": 21834298.0, + "step": 4369 + }, + { + "epoch": 0.3603529314752206, + "grad_norm": 1.9621559381484985, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7849304676055908, + "num_tokens": 21863530.0, + "step": 4370 + }, + { + "epoch": 0.36043539210027215, + "grad_norm": 1.8248332738876343, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.740675151348114, + "num_tokens": 21898612.0, + "step": 4371 + }, + { + "epoch": 0.36051785272532366, + "grad_norm": 2.2042629718780518, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.722247302532196, + "num_tokens": 21926737.0, + "step": 4372 + }, + { + "epoch": 0.3606003133503752, + "grad_norm": 2.076540470123291, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7105898261070251, + "num_tokens": 21955806.0, + "step": 4373 + }, + { + "epoch": 0.36068277397542675, + "grad_norm": 1.7841089963912964, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.7843146324157715, + "num_tokens": 21989156.0, + "step": 4374 + }, + { + "epoch": 0.36076523460047827, + "grad_norm": 2.0854532718658447, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7471230030059814, + "num_tokens": 22015243.0, + "step": 4375 + }, + { + "epoch": 0.3608476952255298, + "grad_norm": 1.9057002067565918, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7485908269882202, + "num_tokens": 22048488.0, + "step": 4376 + }, + { + "epoch": 0.36093015585058136, + "grad_norm": 2.1063523292541504, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7075827717781067, + "num_tokens": 22076189.0, + "step": 4377 + }, + { + "epoch": 0.3610126164756329, + "grad_norm": 1.9961122274398804, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.7527461051940918, + "num_tokens": 22103823.0, + "step": 4378 + }, + { + "epoch": 0.36109507710068445, + "grad_norm": 1.9708279371261597, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.722918689250946, + "num_tokens": 22135666.0, + "step": 4379 + }, + { + "epoch": 0.36117753772573596, + "grad_norm": 2.1578195095062256, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7151279449462891, + "num_tokens": 22162888.0, + "step": 4380 + }, + { + "epoch": 0.3612599983507875, + "grad_norm": 2.017705202102661, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7363004684448242, + "num_tokens": 22191346.0, + "step": 4381 + }, + { + "epoch": 0.36134245897583905, + "grad_norm": 2.5157508850097656, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7295169234275818, + "num_tokens": 22212015.0, + "step": 4382 + }, + { + "epoch": 0.36142491960089057, + "grad_norm": 2.080869197845459, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7369694709777832, + "num_tokens": 22239676.0, + "step": 4383 + }, + { + "epoch": 0.3615073802259421, + "grad_norm": 2.023834466934204, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7055644989013672, + "num_tokens": 22269899.0, + "step": 4384 + }, + { + "epoch": 0.36158984085099366, + "grad_norm": 2.065140962600708, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7828671932220459, + "num_tokens": 22297339.0, + "step": 4385 + }, + { + "epoch": 0.3616723014760452, + "grad_norm": 2.0753018856048584, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7400762438774109, + "num_tokens": 22325027.0, + "step": 4386 + }, + { + "epoch": 0.36175476210109675, + "grad_norm": 1.871286153793335, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.780550479888916, + "num_tokens": 22358351.0, + "step": 4387 + }, + { + "epoch": 0.36183722272614827, + "grad_norm": 1.9152905941009521, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.783346951007843, + "num_tokens": 22387620.0, + "step": 4388 + }, + { + "epoch": 0.3619196833511998, + "grad_norm": 2.3243744373321533, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7208185791969299, + "num_tokens": 22410208.0, + "step": 4389 + }, + { + "epoch": 0.36200214397625136, + "grad_norm": 2.0715394020080566, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7776641249656677, + "num_tokens": 22437128.0, + "step": 4390 + }, + { + "epoch": 0.3620846046013029, + "grad_norm": 2.239332675933838, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.7894363403320312, + "num_tokens": 22460669.0, + "step": 4391 + }, + { + "epoch": 0.3621670652263544, + "grad_norm": 2.0674538612365723, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7516976594924927, + "num_tokens": 22488047.0, + "step": 4392 + }, + { + "epoch": 0.36224952585140596, + "grad_norm": 2.3102571964263916, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7379344701766968, + "num_tokens": 22513139.0, + "step": 4393 + }, + { + "epoch": 0.3623319864764575, + "grad_norm": 1.8619533777236938, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7701462507247925, + "num_tokens": 22548163.0, + "step": 4394 + }, + { + "epoch": 0.36241444710150905, + "grad_norm": 1.9105786085128784, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7778967022895813, + "num_tokens": 22578332.0, + "step": 4395 + }, + { + "epoch": 0.36249690772656057, + "grad_norm": 2.049544334411621, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7523454427719116, + "num_tokens": 22609921.0, + "step": 4396 + }, + { + "epoch": 0.3625793683516121, + "grad_norm": 1.933559536933899, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7212772369384766, + "num_tokens": 22643184.0, + "step": 4397 + }, + { + "epoch": 0.36266182897666366, + "grad_norm": 1.8487049341201782, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.7710655927658081, + "num_tokens": 22676549.0, + "step": 4398 + }, + { + "epoch": 0.3627442896017152, + "grad_norm": 1.9859942197799683, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.7803990840911865, + "num_tokens": 22703942.0, + "step": 4399 + }, + { + "epoch": 0.3628267502267667, + "grad_norm": 1.8126928806304932, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7831672430038452, + "num_tokens": 22735521.0, + "step": 4400 + }, + { + "epoch": 0.36290921085181826, + "grad_norm": 1.8652393817901611, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7413725852966309, + "num_tokens": 22768183.0, + "step": 4401 + }, + { + "epoch": 0.3629916714768698, + "grad_norm": 1.9571619033813477, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7307050228118896, + "num_tokens": 22801145.0, + "step": 4402 + }, + { + "epoch": 0.36307413210192135, + "grad_norm": 1.6411546468734741, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7790459394454956, + "num_tokens": 22844086.0, + "step": 4403 + }, + { + "epoch": 0.36315659272697287, + "grad_norm": 2.0700900554656982, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.757479190826416, + "num_tokens": 22875199.0, + "step": 4404 + }, + { + "epoch": 0.3632390533520244, + "grad_norm": 1.9229382276535034, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7316349148750305, + "num_tokens": 22909619.0, + "step": 4405 + }, + { + "epoch": 0.36332151397707596, + "grad_norm": 1.8852132558822632, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7435899376869202, + "num_tokens": 22941663.0, + "step": 4406 + }, + { + "epoch": 0.3634039746021275, + "grad_norm": 1.8049545288085938, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7550292015075684, + "num_tokens": 22980429.0, + "step": 4407 + }, + { + "epoch": 0.363486435227179, + "grad_norm": 1.8594934940338135, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7353895902633667, + "num_tokens": 23016173.0, + "step": 4408 + }, + { + "epoch": 0.36356889585223057, + "grad_norm": 2.004136323928833, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7346680760383606, + "num_tokens": 23047121.0, + "step": 4409 + }, + { + "epoch": 0.3636513564772821, + "grad_norm": 1.9969515800476074, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7578710317611694, + "num_tokens": 23077256.0, + "step": 4410 + }, + { + "epoch": 0.36373381710233366, + "grad_norm": 2.1755363941192627, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7071742415428162, + "num_tokens": 23103339.0, + "step": 4411 + }, + { + "epoch": 0.3638162777273852, + "grad_norm": 1.8700835704803467, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.7622292041778564, + "num_tokens": 23136434.0, + "step": 4412 + }, + { + "epoch": 0.3638987383524367, + "grad_norm": 2.11427640914917, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7497960925102234, + "num_tokens": 23164946.0, + "step": 4413 + }, + { + "epoch": 0.36398119897748826, + "grad_norm": 1.9031463861465454, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7225439548492432, + "num_tokens": 23197001.0, + "step": 4414 + }, + { + "epoch": 0.3640636596025398, + "grad_norm": 1.8347156047821045, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7701746225357056, + "num_tokens": 23228944.0, + "step": 4415 + }, + { + "epoch": 0.36414612022759135, + "grad_norm": 1.731986165046692, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7467414140701294, + "num_tokens": 23264682.0, + "step": 4416 + }, + { + "epoch": 0.36422858085264287, + "grad_norm": 2.079319953918457, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7382156848907471, + "num_tokens": 23291399.0, + "step": 4417 + }, + { + "epoch": 0.3643110414776944, + "grad_norm": 2.392873764038086, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7305630445480347, + "num_tokens": 23315001.0, + "step": 4418 + }, + { + "epoch": 0.36439350210274596, + "grad_norm": 2.226578712463379, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7376251220703125, + "num_tokens": 23341967.0, + "step": 4419 + }, + { + "epoch": 0.3644759627277975, + "grad_norm": 2.192854404449463, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7433857917785645, + "num_tokens": 23368341.0, + "step": 4420 + }, + { + "epoch": 0.364558423352849, + "grad_norm": 2.1421875953674316, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7784019708633423, + "num_tokens": 23395301.0, + "step": 4421 + }, + { + "epoch": 0.36464088397790057, + "grad_norm": 1.9894769191741943, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7451278567314148, + "num_tokens": 23424815.0, + "step": 4422 + }, + { + "epoch": 0.3647233446029521, + "grad_norm": 1.9531766176223755, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7844141721725464, + "num_tokens": 23451989.0, + "step": 4423 + }, + { + "epoch": 0.36480580522800365, + "grad_norm": 2.288290023803711, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7301594018936157, + "num_tokens": 23474840.0, + "step": 4424 + }, + { + "epoch": 0.36488826585305517, + "grad_norm": 2.0785627365112305, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7726584672927856, + "num_tokens": 23500045.0, + "step": 4425 + }, + { + "epoch": 0.3649707264781067, + "grad_norm": 1.9481010437011719, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.740807294845581, + "num_tokens": 23533821.0, + "step": 4426 + }, + { + "epoch": 0.36505318710315826, + "grad_norm": 2.0187244415283203, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7330710887908936, + "num_tokens": 23563490.0, + "step": 4427 + }, + { + "epoch": 0.3651356477282098, + "grad_norm": 2.0675535202026367, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7511628866195679, + "num_tokens": 23590739.0, + "step": 4428 + }, + { + "epoch": 0.3652181083532613, + "grad_norm": 2.273343086242676, + "learning_rate": 1e-06, + "loss": 1.0508, + "mean_token_accuracy": 0.6820851564407349, + "num_tokens": 23616256.0, + "step": 4429 + }, + { + "epoch": 0.36530056897831287, + "grad_norm": 2.1157917976379395, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7051343321800232, + "num_tokens": 23645967.0, + "step": 4430 + }, + { + "epoch": 0.3653830296033644, + "grad_norm": 1.9273542165756226, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7555322647094727, + "num_tokens": 23679428.0, + "step": 4431 + }, + { + "epoch": 0.36546549022841596, + "grad_norm": 1.8709967136383057, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.7558026313781738, + "num_tokens": 23712864.0, + "step": 4432 + }, + { + "epoch": 0.3655479508534675, + "grad_norm": 1.9382274150848389, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7438878417015076, + "num_tokens": 23744314.0, + "step": 4433 + }, + { + "epoch": 0.365630411478519, + "grad_norm": 2.264054775238037, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.69508957862854, + "num_tokens": 23771012.0, + "step": 4434 + }, + { + "epoch": 0.36571287210357056, + "grad_norm": 1.6997040510177612, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7489914894104004, + "num_tokens": 23810340.0, + "step": 4435 + }, + { + "epoch": 0.3657953327286221, + "grad_norm": 1.8943361043930054, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7712359428405762, + "num_tokens": 23840939.0, + "step": 4436 + }, + { + "epoch": 0.3658777933536736, + "grad_norm": 1.8394001722335815, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7472519874572754, + "num_tokens": 23872316.0, + "step": 4437 + }, + { + "epoch": 0.36596025397872517, + "grad_norm": 2.0055313110351562, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7701534032821655, + "num_tokens": 23902342.0, + "step": 4438 + }, + { + "epoch": 0.3660427146037767, + "grad_norm": 1.7499430179595947, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7499094605445862, + "num_tokens": 23938705.0, + "step": 4439 + }, + { + "epoch": 0.36612517522882826, + "grad_norm": 1.9717215299606323, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7339510321617126, + "num_tokens": 23969314.0, + "step": 4440 + }, + { + "epoch": 0.3662076358538798, + "grad_norm": 2.137444019317627, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.76493239402771, + "num_tokens": 23992987.0, + "step": 4441 + }, + { + "epoch": 0.3662900964789313, + "grad_norm": 1.9344807863235474, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7178971767425537, + "num_tokens": 24024888.0, + "step": 4442 + }, + { + "epoch": 0.36637255710398287, + "grad_norm": 1.85343599319458, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7597968578338623, + "num_tokens": 24058044.0, + "step": 4443 + }, + { + "epoch": 0.3664550177290344, + "grad_norm": 1.7660396099090576, + "learning_rate": 1e-06, + "loss": 0.5815, + "mean_token_accuracy": 0.8055717945098877, + "num_tokens": 24088961.0, + "step": 4444 + }, + { + "epoch": 0.3665374783540859, + "grad_norm": 2.015920639038086, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7044057846069336, + "num_tokens": 24122109.0, + "step": 4445 + }, + { + "epoch": 0.36661993897913747, + "grad_norm": 2.2850091457366943, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7187957763671875, + "num_tokens": 24145248.0, + "step": 4446 + }, + { + "epoch": 0.366702399604189, + "grad_norm": 1.9528566598892212, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7373639345169067, + "num_tokens": 24175992.0, + "step": 4447 + }, + { + "epoch": 0.36678486022924056, + "grad_norm": 2.0211000442504883, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.764824628829956, + "num_tokens": 24204264.0, + "step": 4448 + }, + { + "epoch": 0.3668673208542921, + "grad_norm": 1.9846707582473755, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7333006858825684, + "num_tokens": 24234303.0, + "step": 4449 + }, + { + "epoch": 0.3669497814793436, + "grad_norm": 2.1329543590545654, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7554373145103455, + "num_tokens": 24260726.0, + "step": 4450 + }, + { + "epoch": 0.36703224210439517, + "grad_norm": 2.0013091564178467, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7114300727844238, + "num_tokens": 24292991.0, + "step": 4451 + }, + { + "epoch": 0.3671147027294467, + "grad_norm": 1.7977944612503052, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.71920245885849, + "num_tokens": 24328421.0, + "step": 4452 + }, + { + "epoch": 0.3671971633544982, + "grad_norm": 1.7395864725112915, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7387138605117798, + "num_tokens": 24365157.0, + "step": 4453 + }, + { + "epoch": 0.3672796239795498, + "grad_norm": 1.945522665977478, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7252031564712524, + "num_tokens": 24395955.0, + "step": 4454 + }, + { + "epoch": 0.3673620846046013, + "grad_norm": 1.919259786605835, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7510992884635925, + "num_tokens": 24427776.0, + "step": 4455 + }, + { + "epoch": 0.36744454522965286, + "grad_norm": 1.8850843906402588, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7373652458190918, + "num_tokens": 24458987.0, + "step": 4456 + }, + { + "epoch": 0.3675270058547044, + "grad_norm": 2.071052312850952, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.752228856086731, + "num_tokens": 24484552.0, + "step": 4457 + }, + { + "epoch": 0.3676094664797559, + "grad_norm": 1.7868189811706543, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.757215142250061, + "num_tokens": 24518286.0, + "step": 4458 + }, + { + "epoch": 0.36769192710480747, + "grad_norm": 2.08207631111145, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7649340033531189, + "num_tokens": 24543257.0, + "step": 4459 + }, + { + "epoch": 0.367774387729859, + "grad_norm": 2.285642385482788, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7313967347145081, + "num_tokens": 24567806.0, + "step": 4460 + }, + { + "epoch": 0.3678568483549105, + "grad_norm": 1.8667768239974976, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7640255093574524, + "num_tokens": 24598302.0, + "step": 4461 + }, + { + "epoch": 0.3679393089799621, + "grad_norm": 2.1750214099884033, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7353963851928711, + "num_tokens": 24622364.0, + "step": 4462 + }, + { + "epoch": 0.3680217696050136, + "grad_norm": 2.0480520725250244, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7875163555145264, + "num_tokens": 24647237.0, + "step": 4463 + }, + { + "epoch": 0.36810423023006517, + "grad_norm": 2.2314493656158447, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7508134841918945, + "num_tokens": 24672149.0, + "step": 4464 + }, + { + "epoch": 0.3681866908551167, + "grad_norm": 2.145028829574585, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7393441200256348, + "num_tokens": 24706821.0, + "step": 4465 + }, + { + "epoch": 0.3682691514801682, + "grad_norm": 2.1708929538726807, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7331571578979492, + "num_tokens": 24734118.0, + "step": 4466 + }, + { + "epoch": 0.3683516121052198, + "grad_norm": 2.277280330657959, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7310365438461304, + "num_tokens": 24760350.0, + "step": 4467 + }, + { + "epoch": 0.3684340727302713, + "grad_norm": 1.9229216575622559, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.734631359577179, + "num_tokens": 24794493.0, + "step": 4468 + }, + { + "epoch": 0.3685165333553228, + "grad_norm": 1.8122062683105469, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.755382776260376, + "num_tokens": 24825984.0, + "step": 4469 + }, + { + "epoch": 0.3685989939803744, + "grad_norm": 1.896575927734375, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7723005414009094, + "num_tokens": 24857326.0, + "step": 4470 + }, + { + "epoch": 0.3686814546054259, + "grad_norm": 1.8416833877563477, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7662485837936401, + "num_tokens": 24890448.0, + "step": 4471 + }, + { + "epoch": 0.36876391523047747, + "grad_norm": 1.9610410928726196, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.7816551923751831, + "num_tokens": 24918112.0, + "step": 4472 + }, + { + "epoch": 0.368846375855529, + "grad_norm": 2.189591884613037, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7275477647781372, + "num_tokens": 24943635.0, + "step": 4473 + }, + { + "epoch": 0.3689288364805805, + "grad_norm": 2.2176616191864014, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7378946542739868, + "num_tokens": 24967419.0, + "step": 4474 + }, + { + "epoch": 0.3690112971056321, + "grad_norm": 2.0454180240631104, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7498830556869507, + "num_tokens": 25002433.0, + "step": 4475 + }, + { + "epoch": 0.3690937577306836, + "grad_norm": 1.9449950456619263, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7545838356018066, + "num_tokens": 25032760.0, + "step": 4476 + }, + { + "epoch": 0.36917621835573516, + "grad_norm": 1.932058334350586, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7002803087234497, + "num_tokens": 25064156.0, + "step": 4477 + }, + { + "epoch": 0.3692586789807867, + "grad_norm": 2.2965006828308105, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.742169201374054, + "num_tokens": 25086258.0, + "step": 4478 + }, + { + "epoch": 0.3693411396058382, + "grad_norm": 2.047398805618286, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.749742865562439, + "num_tokens": 25112456.0, + "step": 4479 + }, + { + "epoch": 0.36942360023088977, + "grad_norm": 2.1339306831359863, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7694918513298035, + "num_tokens": 25139613.0, + "step": 4480 + }, + { + "epoch": 0.3695060608559413, + "grad_norm": 1.7437506914138794, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7739567756652832, + "num_tokens": 25173823.0, + "step": 4481 + }, + { + "epoch": 0.3695885214809928, + "grad_norm": 2.06762957572937, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7453402876853943, + "num_tokens": 25200482.0, + "step": 4482 + }, + { + "epoch": 0.3696709821060444, + "grad_norm": 1.7370468378067017, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7129122018814087, + "num_tokens": 25238794.0, + "step": 4483 + }, + { + "epoch": 0.3697534427310959, + "grad_norm": 1.9528234004974365, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7599396705627441, + "num_tokens": 25269975.0, + "step": 4484 + }, + { + "epoch": 0.36983590335614747, + "grad_norm": 2.0007760524749756, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7697752118110657, + "num_tokens": 25298099.0, + "step": 4485 + }, + { + "epoch": 0.369918363981199, + "grad_norm": 2.0401220321655273, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7443779706954956, + "num_tokens": 25326988.0, + "step": 4486 + }, + { + "epoch": 0.3700008246062505, + "grad_norm": 2.061401605606079, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7649828195571899, + "num_tokens": 25352926.0, + "step": 4487 + }, + { + "epoch": 0.3700832852313021, + "grad_norm": 2.0171773433685303, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7340955138206482, + "num_tokens": 25383745.0, + "step": 4488 + }, + { + "epoch": 0.3701657458563536, + "grad_norm": 2.161029577255249, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7515557408332825, + "num_tokens": 25409415.0, + "step": 4489 + }, + { + "epoch": 0.3702482064814051, + "grad_norm": 2.0237035751342773, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7539688944816589, + "num_tokens": 25439789.0, + "step": 4490 + }, + { + "epoch": 0.3703306671064567, + "grad_norm": 1.9034749269485474, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7159236073493958, + "num_tokens": 25472806.0, + "step": 4491 + }, + { + "epoch": 0.3704131277315082, + "grad_norm": 1.855351209640503, + "learning_rate": 1e-06, + "loss": 0.567, + "mean_token_accuracy": 0.8200533390045166, + "num_tokens": 25503665.0, + "step": 4492 + }, + { + "epoch": 0.37049558835655977, + "grad_norm": 2.077183961868286, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7218749523162842, + "num_tokens": 25534105.0, + "step": 4493 + }, + { + "epoch": 0.3705780489816113, + "grad_norm": 2.1862905025482178, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7433492541313171, + "num_tokens": 25559340.0, + "step": 4494 + }, + { + "epoch": 0.3706605096066628, + "grad_norm": 2.007873296737671, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7811962962150574, + "num_tokens": 25585326.0, + "step": 4495 + }, + { + "epoch": 0.3707429702317144, + "grad_norm": 1.9903382062911987, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7455147504806519, + "num_tokens": 25615288.0, + "step": 4496 + }, + { + "epoch": 0.3708254308567659, + "grad_norm": 2.0182504653930664, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7368838787078857, + "num_tokens": 25643777.0, + "step": 4497 + }, + { + "epoch": 0.3709078914818174, + "grad_norm": 1.8566192388534546, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7804079055786133, + "num_tokens": 25672789.0, + "step": 4498 + }, + { + "epoch": 0.370990352106869, + "grad_norm": 1.9339534044265747, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7607661485671997, + "num_tokens": 25703792.0, + "step": 4499 + }, + { + "epoch": 0.3710728127319205, + "grad_norm": 2.05306077003479, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7200382351875305, + "num_tokens": 25733868.0, + "step": 4500 + }, + { + "epoch": 0.37115527335697207, + "grad_norm": 1.758000373840332, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7514623403549194, + "num_tokens": 25769246.0, + "step": 4501 + }, + { + "epoch": 0.3712377339820236, + "grad_norm": 2.1432859897613525, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7355334162712097, + "num_tokens": 25797164.0, + "step": 4502 + }, + { + "epoch": 0.3713201946070751, + "grad_norm": 2.0291390419006348, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7630016803741455, + "num_tokens": 25827167.0, + "step": 4503 + }, + { + "epoch": 0.3714026552321267, + "grad_norm": 1.9141809940338135, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7415590882301331, + "num_tokens": 25859348.0, + "step": 4504 + }, + { + "epoch": 0.3714851158571782, + "grad_norm": 1.92379891872406, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7370331287384033, + "num_tokens": 25891875.0, + "step": 4505 + }, + { + "epoch": 0.3715675764822297, + "grad_norm": 2.0033745765686035, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.72785884141922, + "num_tokens": 25920168.0, + "step": 4506 + }, + { + "epoch": 0.3716500371072813, + "grad_norm": 2.292644739151001, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7323430776596069, + "num_tokens": 25947186.0, + "step": 4507 + }, + { + "epoch": 0.3717324977323328, + "grad_norm": 1.8151658773422241, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7630524635314941, + "num_tokens": 25981628.0, + "step": 4508 + }, + { + "epoch": 0.3718149583573844, + "grad_norm": 1.8144062757492065, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7309158444404602, + "num_tokens": 26017500.0, + "step": 4509 + }, + { + "epoch": 0.3718974189824359, + "grad_norm": 2.1905710697174072, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7469007968902588, + "num_tokens": 26043880.0, + "step": 4510 + }, + { + "epoch": 0.3719798796074874, + "grad_norm": 1.9428718090057373, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7842645645141602, + "num_tokens": 26073556.0, + "step": 4511 + }, + { + "epoch": 0.372062340232539, + "grad_norm": 2.192657709121704, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7421797513961792, + "num_tokens": 26099495.0, + "step": 4512 + }, + { + "epoch": 0.3721448008575905, + "grad_norm": 1.952086329460144, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6956668496131897, + "num_tokens": 26133189.0, + "step": 4513 + }, + { + "epoch": 0.372227261482642, + "grad_norm": 1.9383277893066406, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7376172542572021, + "num_tokens": 26167383.0, + "step": 4514 + }, + { + "epoch": 0.3723097221076936, + "grad_norm": 1.8385668992996216, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7670361399650574, + "num_tokens": 26200082.0, + "step": 4515 + }, + { + "epoch": 0.3723921827327451, + "grad_norm": 2.251129388809204, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7458999156951904, + "num_tokens": 26222894.0, + "step": 4516 + }, + { + "epoch": 0.3724746433577967, + "grad_norm": 2.226231098175049, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7314611673355103, + "num_tokens": 26248111.0, + "step": 4517 + }, + { + "epoch": 0.3725571039828482, + "grad_norm": 1.8438985347747803, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7355101108551025, + "num_tokens": 26280473.0, + "step": 4518 + }, + { + "epoch": 0.3726395646078997, + "grad_norm": 1.6764923334121704, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.7868480682373047, + "num_tokens": 26316227.0, + "step": 4519 + }, + { + "epoch": 0.3727220252329513, + "grad_norm": 2.2838683128356934, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7337328791618347, + "num_tokens": 26341583.0, + "step": 4520 + }, + { + "epoch": 0.3728044858580028, + "grad_norm": 2.2519443035125732, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7287853360176086, + "num_tokens": 26372658.0, + "step": 4521 + }, + { + "epoch": 0.3728869464830543, + "grad_norm": 1.9127064943313599, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7576667070388794, + "num_tokens": 26402045.0, + "step": 4522 + }, + { + "epoch": 0.3729694071081059, + "grad_norm": 1.8323420286178589, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7484387159347534, + "num_tokens": 26436537.0, + "step": 4523 + }, + { + "epoch": 0.3730518677331574, + "grad_norm": 1.9307339191436768, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7507308721542358, + "num_tokens": 26465191.0, + "step": 4524 + }, + { + "epoch": 0.373134328358209, + "grad_norm": 1.8704431056976318, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7451093196868896, + "num_tokens": 26498019.0, + "step": 4525 + }, + { + "epoch": 0.3732167889832605, + "grad_norm": 1.983024001121521, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7812119126319885, + "num_tokens": 26524016.0, + "step": 4526 + }, + { + "epoch": 0.373299249608312, + "grad_norm": 2.2634775638580322, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.746557354927063, + "num_tokens": 26548330.0, + "step": 4527 + }, + { + "epoch": 0.3733817102333636, + "grad_norm": 1.8787676095962524, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7658941745758057, + "num_tokens": 26578692.0, + "step": 4528 + }, + { + "epoch": 0.3734641708584151, + "grad_norm": 1.860780119895935, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7625554203987122, + "num_tokens": 26613966.0, + "step": 4529 + }, + { + "epoch": 0.3735466314834666, + "grad_norm": 1.8874411582946777, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7462220191955566, + "num_tokens": 26646702.0, + "step": 4530 + }, + { + "epoch": 0.3736290921085182, + "grad_norm": 1.8154098987579346, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7754780054092407, + "num_tokens": 26680839.0, + "step": 4531 + }, + { + "epoch": 0.3737115527335697, + "grad_norm": 1.725246548652649, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7558463215827942, + "num_tokens": 26714903.0, + "step": 4532 + }, + { + "epoch": 0.3737940133586213, + "grad_norm": 2.0839550495147705, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7527979612350464, + "num_tokens": 26742392.0, + "step": 4533 + }, + { + "epoch": 0.3738764739836728, + "grad_norm": 2.1132915019989014, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7672244310379028, + "num_tokens": 26766940.0, + "step": 4534 + }, + { + "epoch": 0.3739589346087243, + "grad_norm": 2.0224413871765137, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7602239847183228, + "num_tokens": 26795799.0, + "step": 4535 + }, + { + "epoch": 0.3740413952337759, + "grad_norm": 2.2745139598846436, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7256149053573608, + "num_tokens": 26820894.0, + "step": 4536 + }, + { + "epoch": 0.3741238558588274, + "grad_norm": 1.9265180826187134, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7648664712905884, + "num_tokens": 26850426.0, + "step": 4537 + }, + { + "epoch": 0.3742063164838789, + "grad_norm": 2.0105032920837402, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.7048971652984619, + "num_tokens": 26883494.0, + "step": 4538 + }, + { + "epoch": 0.3742887771089305, + "grad_norm": 2.3736021518707275, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7351033687591553, + "num_tokens": 26907688.0, + "step": 4539 + }, + { + "epoch": 0.374371237733982, + "grad_norm": 2.0841403007507324, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7477086782455444, + "num_tokens": 26936251.0, + "step": 4540 + }, + { + "epoch": 0.3744536983590336, + "grad_norm": 2.096081256866455, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7108627557754517, + "num_tokens": 26965809.0, + "step": 4541 + }, + { + "epoch": 0.3745361589840851, + "grad_norm": 1.9795458316802979, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7590646743774414, + "num_tokens": 26996707.0, + "step": 4542 + }, + { + "epoch": 0.3746186196091366, + "grad_norm": 1.9525816440582275, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.742437481880188, + "num_tokens": 27028393.0, + "step": 4543 + }, + { + "epoch": 0.3747010802341882, + "grad_norm": 1.9257235527038574, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7277208566665649, + "num_tokens": 27061378.0, + "step": 4544 + }, + { + "epoch": 0.3747835408592397, + "grad_norm": 2.149503469467163, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7352092266082764, + "num_tokens": 27086803.0, + "step": 4545 + }, + { + "epoch": 0.3748660014842913, + "grad_norm": 1.9606956243515015, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7181287407875061, + "num_tokens": 27118282.0, + "step": 4546 + }, + { + "epoch": 0.3749484621093428, + "grad_norm": 2.0617384910583496, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7289538383483887, + "num_tokens": 27149030.0, + "step": 4547 + }, + { + "epoch": 0.3750309227343943, + "grad_norm": 1.9217767715454102, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7180819511413574, + "num_tokens": 27183744.0, + "step": 4548 + }, + { + "epoch": 0.3751133833594459, + "grad_norm": 1.9770809412002563, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7367875576019287, + "num_tokens": 27214364.0, + "step": 4549 + }, + { + "epoch": 0.3751958439844974, + "grad_norm": 1.7831881046295166, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.755841851234436, + "num_tokens": 27249829.0, + "step": 4550 + }, + { + "epoch": 0.3752783046095489, + "grad_norm": 2.2460174560546875, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7639776468276978, + "num_tokens": 27273367.0, + "step": 4551 + }, + { + "epoch": 0.3753607652346005, + "grad_norm": 1.913699746131897, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.7672393918037415, + "num_tokens": 27304063.0, + "step": 4552 + }, + { + "epoch": 0.375443225859652, + "grad_norm": 2.00262713432312, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7734737396240234, + "num_tokens": 27334382.0, + "step": 4553 + }, + { + "epoch": 0.3755256864847036, + "grad_norm": 1.8926613330841064, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.761488139629364, + "num_tokens": 27365032.0, + "step": 4554 + }, + { + "epoch": 0.3756081471097551, + "grad_norm": 1.9999395608901978, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7560417056083679, + "num_tokens": 27392150.0, + "step": 4555 + }, + { + "epoch": 0.3756906077348066, + "grad_norm": 1.9830634593963623, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7455068230628967, + "num_tokens": 27420127.0, + "step": 4556 + }, + { + "epoch": 0.3757730683598582, + "grad_norm": 1.805201768875122, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7454792261123657, + "num_tokens": 27453935.0, + "step": 4557 + }, + { + "epoch": 0.3758555289849097, + "grad_norm": 2.1897103786468506, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7389399409294128, + "num_tokens": 27480104.0, + "step": 4558 + }, + { + "epoch": 0.3759379896099612, + "grad_norm": 2.0178301334381104, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7596884965896606, + "num_tokens": 27509374.0, + "step": 4559 + }, + { + "epoch": 0.3760204502350128, + "grad_norm": 1.948156476020813, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7203220129013062, + "num_tokens": 27540339.0, + "step": 4560 + }, + { + "epoch": 0.3761029108600643, + "grad_norm": 1.9885417222976685, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7322150468826294, + "num_tokens": 27569337.0, + "step": 4561 + }, + { + "epoch": 0.3761853714851159, + "grad_norm": 1.9664885997772217, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7370713949203491, + "num_tokens": 27599122.0, + "step": 4562 + }, + { + "epoch": 0.3762678321101674, + "grad_norm": 2.112748622894287, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7372516393661499, + "num_tokens": 27627359.0, + "step": 4563 + }, + { + "epoch": 0.3763502927352189, + "grad_norm": 2.222492218017578, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7459456920623779, + "num_tokens": 27654465.0, + "step": 4564 + }, + { + "epoch": 0.3764327533602705, + "grad_norm": 2.025904655456543, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7182238101959229, + "num_tokens": 27687422.0, + "step": 4565 + }, + { + "epoch": 0.376515213985322, + "grad_norm": 2.0755839347839355, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.726049542427063, + "num_tokens": 27716861.0, + "step": 4566 + }, + { + "epoch": 0.3765976746103735, + "grad_norm": 2.001230478286743, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7429283857345581, + "num_tokens": 27745018.0, + "step": 4567 + }, + { + "epoch": 0.3766801352354251, + "grad_norm": 2.0020365715026855, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7563735842704773, + "num_tokens": 27774361.0, + "step": 4568 + }, + { + "epoch": 0.3767625958604766, + "grad_norm": 2.3174784183502197, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.7104043960571289, + "num_tokens": 27797899.0, + "step": 4569 + }, + { + "epoch": 0.3768450564855282, + "grad_norm": 1.7841764688491821, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.7724944353103638, + "num_tokens": 27830824.0, + "step": 4570 + }, + { + "epoch": 0.3769275171105797, + "grad_norm": 1.900155782699585, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7579259872436523, + "num_tokens": 27864771.0, + "step": 4571 + }, + { + "epoch": 0.3770099777356312, + "grad_norm": 2.070185899734497, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7142552137374878, + "num_tokens": 27895854.0, + "step": 4572 + }, + { + "epoch": 0.3770924383606828, + "grad_norm": 1.8987607955932617, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7548670172691345, + "num_tokens": 27927099.0, + "step": 4573 + }, + { + "epoch": 0.3771748989857343, + "grad_norm": 2.0167737007141113, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.6996945142745972, + "num_tokens": 27962414.0, + "step": 4574 + }, + { + "epoch": 0.3772573596107858, + "grad_norm": 2.192502498626709, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7334610223770142, + "num_tokens": 27989137.0, + "step": 4575 + }, + { + "epoch": 0.3773398202358374, + "grad_norm": 2.3171699047088623, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7107908725738525, + "num_tokens": 28013309.0, + "step": 4576 + }, + { + "epoch": 0.3774222808608889, + "grad_norm": 1.7452472448349, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7657808065414429, + "num_tokens": 28049056.0, + "step": 4577 + }, + { + "epoch": 0.3775047414859405, + "grad_norm": 2.0792298316955566, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7334122061729431, + "num_tokens": 28074667.0, + "step": 4578 + }, + { + "epoch": 0.377587202110992, + "grad_norm": 1.9130879640579224, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7579200267791748, + "num_tokens": 28106104.0, + "step": 4579 + }, + { + "epoch": 0.3776696627360435, + "grad_norm": 2.1228790283203125, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7840638756752014, + "num_tokens": 28131745.0, + "step": 4580 + }, + { + "epoch": 0.3777521233610951, + "grad_norm": 1.9331145286560059, + "learning_rate": 1e-06, + "loss": 0.5624, + "mean_token_accuracy": 0.8124430775642395, + "num_tokens": 28160332.0, + "step": 4581 + }, + { + "epoch": 0.3778345839861466, + "grad_norm": 2.098942279815674, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7361599206924438, + "num_tokens": 28189203.0, + "step": 4582 + }, + { + "epoch": 0.3779170446111981, + "grad_norm": 1.8297998905181885, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7616162300109863, + "num_tokens": 28220144.0, + "step": 4583 + }, + { + "epoch": 0.3779995052362497, + "grad_norm": 2.02095365524292, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.71617591381073, + "num_tokens": 28249617.0, + "step": 4584 + }, + { + "epoch": 0.3780819658613012, + "grad_norm": 2.0008013248443604, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7417540550231934, + "num_tokens": 28280046.0, + "step": 4585 + }, + { + "epoch": 0.3781644264863528, + "grad_norm": 1.917118787765503, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7481143474578857, + "num_tokens": 28310116.0, + "step": 4586 + }, + { + "epoch": 0.3782468871114043, + "grad_norm": 1.9427497386932373, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7386835813522339, + "num_tokens": 28341891.0, + "step": 4587 + }, + { + "epoch": 0.3783293477364558, + "grad_norm": 2.003192186355591, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7229310870170593, + "num_tokens": 28374895.0, + "step": 4588 + }, + { + "epoch": 0.3784118083615074, + "grad_norm": 2.0022103786468506, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7565262913703918, + "num_tokens": 28406261.0, + "step": 4589 + }, + { + "epoch": 0.3784942689865589, + "grad_norm": 2.1203794479370117, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7385028004646301, + "num_tokens": 28432801.0, + "step": 4590 + }, + { + "epoch": 0.37857672961161043, + "grad_norm": 2.2295610904693604, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.747117280960083, + "num_tokens": 28456328.0, + "step": 4591 + }, + { + "epoch": 0.378659190236662, + "grad_norm": 2.0101282596588135, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7418285608291626, + "num_tokens": 28485912.0, + "step": 4592 + }, + { + "epoch": 0.3787416508617135, + "grad_norm": 2.108944892883301, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7146422266960144, + "num_tokens": 28515208.0, + "step": 4593 + }, + { + "epoch": 0.3788241114867651, + "grad_norm": 2.0539658069610596, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.746546745300293, + "num_tokens": 28543293.0, + "step": 4594 + }, + { + "epoch": 0.3789065721118166, + "grad_norm": 2.1058268547058105, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7376825213432312, + "num_tokens": 28570393.0, + "step": 4595 + }, + { + "epoch": 0.3789890327368681, + "grad_norm": 1.951715350151062, + "learning_rate": 1e-06, + "loss": 0.5799, + "mean_token_accuracy": 0.8024791479110718, + "num_tokens": 28594539.0, + "step": 4596 + }, + { + "epoch": 0.3790714933619197, + "grad_norm": 1.9831081628799438, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7458193302154541, + "num_tokens": 28623675.0, + "step": 4597 + }, + { + "epoch": 0.3791539539869712, + "grad_norm": 1.9380871057510376, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7639247179031372, + "num_tokens": 28652193.0, + "step": 4598 + }, + { + "epoch": 0.37923641461202273, + "grad_norm": 1.9061191082000732, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7540990114212036, + "num_tokens": 28687010.0, + "step": 4599 + }, + { + "epoch": 0.3793188752370743, + "grad_norm": 2.267674446105957, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7194674611091614, + "num_tokens": 28712606.0, + "step": 4600 + }, + { + "epoch": 0.3794013358621258, + "grad_norm": 1.898414969444275, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.707633912563324, + "num_tokens": 28745823.0, + "step": 4601 + }, + { + "epoch": 0.3794837964871774, + "grad_norm": 1.9277890920639038, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7703983783721924, + "num_tokens": 28777550.0, + "step": 4602 + }, + { + "epoch": 0.3795662571122289, + "grad_norm": 2.022573947906494, + "learning_rate": 1e-06, + "loss": 0.7821, + "mean_token_accuracy": 0.7560428977012634, + "num_tokens": 28808619.0, + "step": 4603 + }, + { + "epoch": 0.37964871773728043, + "grad_norm": 2.0250966548919678, + "learning_rate": 1e-06, + "loss": 0.7852, + "mean_token_accuracy": 0.7521641254425049, + "num_tokens": 28838806.0, + "step": 4604 + }, + { + "epoch": 0.379731178362332, + "grad_norm": 2.002000093460083, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6999561190605164, + "num_tokens": 28870214.0, + "step": 4605 + }, + { + "epoch": 0.3798136389873835, + "grad_norm": 2.339951992034912, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7353300452232361, + "num_tokens": 28894959.0, + "step": 4606 + }, + { + "epoch": 0.3798960996124351, + "grad_norm": 2.249295711517334, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7307213544845581, + "num_tokens": 28920929.0, + "step": 4607 + }, + { + "epoch": 0.3799785602374866, + "grad_norm": 1.9642667770385742, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7243261337280273, + "num_tokens": 28950465.0, + "step": 4608 + }, + { + "epoch": 0.3800610208625381, + "grad_norm": 2.005486249923706, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7144589424133301, + "num_tokens": 28979611.0, + "step": 4609 + }, + { + "epoch": 0.3801434814875897, + "grad_norm": 1.7760493755340576, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.7795543074607849, + "num_tokens": 29014609.0, + "step": 4610 + }, + { + "epoch": 0.3802259421126412, + "grad_norm": 1.875472068786621, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.7882363796234131, + "num_tokens": 29047414.0, + "step": 4611 + }, + { + "epoch": 0.38030840273769273, + "grad_norm": 1.7523367404937744, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7259569764137268, + "num_tokens": 29085605.0, + "step": 4612 + }, + { + "epoch": 0.3803908633627443, + "grad_norm": 2.010977029800415, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7530871629714966, + "num_tokens": 29116018.0, + "step": 4613 + }, + { + "epoch": 0.3804733239877958, + "grad_norm": 1.8489981889724731, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.763943076133728, + "num_tokens": 29149206.0, + "step": 4614 + }, + { + "epoch": 0.3805557846128474, + "grad_norm": 2.111332416534424, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6942394971847534, + "num_tokens": 29181295.0, + "step": 4615 + }, + { + "epoch": 0.3806382452378989, + "grad_norm": 1.9380793571472168, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7554754018783569, + "num_tokens": 29212735.0, + "step": 4616 + }, + { + "epoch": 0.3807207058629504, + "grad_norm": 2.100800037384033, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7265846729278564, + "num_tokens": 29244165.0, + "step": 4617 + }, + { + "epoch": 0.380803166488002, + "grad_norm": 2.2211875915527344, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7328354120254517, + "num_tokens": 29269504.0, + "step": 4618 + }, + { + "epoch": 0.3808856271130535, + "grad_norm": 2.2033021450042725, + "learning_rate": 1e-06, + "loss": 1.0401, + "mean_token_accuracy": 0.690268874168396, + "num_tokens": 29296129.0, + "step": 4619 + }, + { + "epoch": 0.38096808773810503, + "grad_norm": 1.9975051879882812, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7785893678665161, + "num_tokens": 29324032.0, + "step": 4620 + }, + { + "epoch": 0.3810505483631566, + "grad_norm": 1.9234179258346558, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7174566984176636, + "num_tokens": 29356624.0, + "step": 4621 + }, + { + "epoch": 0.3811330089882081, + "grad_norm": 2.3281545639038086, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7305277585983276, + "num_tokens": 29377632.0, + "step": 4622 + }, + { + "epoch": 0.3812154696132597, + "grad_norm": 2.027345895767212, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7375071048736572, + "num_tokens": 29405100.0, + "step": 4623 + }, + { + "epoch": 0.3812979302383112, + "grad_norm": 1.69904625415802, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7445082068443298, + "num_tokens": 29443358.0, + "step": 4624 + }, + { + "epoch": 0.38138039086336273, + "grad_norm": 2.0926530361175537, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7145123481750488, + "num_tokens": 29473116.0, + "step": 4625 + }, + { + "epoch": 0.3814628514884143, + "grad_norm": 2.1047916412353516, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.75541090965271, + "num_tokens": 29499840.0, + "step": 4626 + }, + { + "epoch": 0.3815453121134658, + "grad_norm": 2.197848081588745, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7282119989395142, + "num_tokens": 29525993.0, + "step": 4627 + }, + { + "epoch": 0.38162777273851733, + "grad_norm": 1.993560552597046, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7304750680923462, + "num_tokens": 29558049.0, + "step": 4628 + }, + { + "epoch": 0.3817102333635689, + "grad_norm": 1.9700548648834229, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7554898262023926, + "num_tokens": 29587666.0, + "step": 4629 + }, + { + "epoch": 0.3817926939886204, + "grad_norm": 1.756661295890808, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7433983087539673, + "num_tokens": 29622024.0, + "step": 4630 + }, + { + "epoch": 0.381875154613672, + "grad_norm": 2.1111583709716797, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7235243916511536, + "num_tokens": 29651257.0, + "step": 4631 + }, + { + "epoch": 0.3819576152387235, + "grad_norm": 1.9305092096328735, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.7544317245483398, + "num_tokens": 29680558.0, + "step": 4632 + }, + { + "epoch": 0.38204007586377503, + "grad_norm": 1.8508824110031128, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7129182815551758, + "num_tokens": 29716886.0, + "step": 4633 + }, + { + "epoch": 0.3821225364888266, + "grad_norm": 1.7950340509414673, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7663015127182007, + "num_tokens": 29748907.0, + "step": 4634 + }, + { + "epoch": 0.3822049971138781, + "grad_norm": 1.8919657468795776, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.7634133100509644, + "num_tokens": 29781857.0, + "step": 4635 + }, + { + "epoch": 0.38228745773892964, + "grad_norm": 2.1781575679779053, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7254475355148315, + "num_tokens": 29810151.0, + "step": 4636 + }, + { + "epoch": 0.3823699183639812, + "grad_norm": 2.2571640014648438, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7067461013793945, + "num_tokens": 29838318.0, + "step": 4637 + }, + { + "epoch": 0.3824523789890327, + "grad_norm": 1.8828436136245728, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.745656430721283, + "num_tokens": 29871755.0, + "step": 4638 + }, + { + "epoch": 0.3825348396140843, + "grad_norm": 2.319344997406006, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7176414728164673, + "num_tokens": 29896664.0, + "step": 4639 + }, + { + "epoch": 0.3826173002391358, + "grad_norm": 1.975599765777588, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7622745633125305, + "num_tokens": 29924990.0, + "step": 4640 + }, + { + "epoch": 0.38269976086418733, + "grad_norm": 31.592702865600586, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7407729625701904, + "num_tokens": 29951725.0, + "step": 4641 + }, + { + "epoch": 0.3827822214892389, + "grad_norm": 2.0163888931274414, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7346246242523193, + "num_tokens": 29980814.0, + "step": 4642 + }, + { + "epoch": 0.3828646821142904, + "grad_norm": 2.2413418292999268, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7297836542129517, + "num_tokens": 30008441.0, + "step": 4643 + }, + { + "epoch": 0.38294714273934194, + "grad_norm": 1.8069736957550049, + "learning_rate": 1e-06, + "loss": 0.7656, + "mean_token_accuracy": 0.7531492710113525, + "num_tokens": 30043807.0, + "step": 4644 + }, + { + "epoch": 0.3830296033643935, + "grad_norm": 2.1002495288848877, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7245186567306519, + "num_tokens": 30074734.0, + "step": 4645 + }, + { + "epoch": 0.38311206398944503, + "grad_norm": 1.8728628158569336, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7847704887390137, + "num_tokens": 30105293.0, + "step": 4646 + }, + { + "epoch": 0.3831945246144966, + "grad_norm": 1.8029797077178955, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7537751197814941, + "num_tokens": 30141665.0, + "step": 4647 + }, + { + "epoch": 0.3832769852395481, + "grad_norm": 2.03238844871521, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7249271869659424, + "num_tokens": 30172403.0, + "step": 4648 + }, + { + "epoch": 0.38335944586459964, + "grad_norm": 1.8939540386199951, + "learning_rate": 1e-06, + "loss": 0.7702, + "mean_token_accuracy": 0.7538276314735413, + "num_tokens": 30201463.0, + "step": 4649 + }, + { + "epoch": 0.3834419064896512, + "grad_norm": 1.9369274377822876, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7444561719894409, + "num_tokens": 30231920.0, + "step": 4650 + }, + { + "epoch": 0.3835243671147027, + "grad_norm": 2.1414384841918945, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7191413640975952, + "num_tokens": 30260954.0, + "step": 4651 + }, + { + "epoch": 0.38360682773975424, + "grad_norm": 1.851272463798523, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7280865907669067, + "num_tokens": 30295609.0, + "step": 4652 + }, + { + "epoch": 0.3836892883648058, + "grad_norm": 2.518264055252075, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7398408651351929, + "num_tokens": 30315135.0, + "step": 4653 + }, + { + "epoch": 0.38377174898985733, + "grad_norm": 1.9873425960540771, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.738339900970459, + "num_tokens": 30343804.0, + "step": 4654 + }, + { + "epoch": 0.3838542096149089, + "grad_norm": 2.0697083473205566, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7395287752151489, + "num_tokens": 30373862.0, + "step": 4655 + }, + { + "epoch": 0.3839366702399604, + "grad_norm": 1.9161821603775024, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7608855962753296, + "num_tokens": 30405713.0, + "step": 4656 + }, + { + "epoch": 0.38401913086501194, + "grad_norm": 1.8838165998458862, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7255673408508301, + "num_tokens": 30437379.0, + "step": 4657 + }, + { + "epoch": 0.3841015914900635, + "grad_norm": 1.9774224758148193, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.755003809928894, + "num_tokens": 30468179.0, + "step": 4658 + }, + { + "epoch": 0.384184052115115, + "grad_norm": 2.0902745723724365, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7356013059616089, + "num_tokens": 30495670.0, + "step": 4659 + }, + { + "epoch": 0.38426651274016654, + "grad_norm": 1.7986745834350586, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7279335260391235, + "num_tokens": 30532430.0, + "step": 4660 + }, + { + "epoch": 0.3843489733652181, + "grad_norm": 2.024184465408325, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7181563377380371, + "num_tokens": 30562875.0, + "step": 4661 + }, + { + "epoch": 0.38443143399026963, + "grad_norm": 1.8500335216522217, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7533389329910278, + "num_tokens": 30595734.0, + "step": 4662 + }, + { + "epoch": 0.3845138946153212, + "grad_norm": 1.861739158630371, + "learning_rate": 1e-06, + "loss": 0.6178, + "mean_token_accuracy": 0.8022891283035278, + "num_tokens": 30627208.0, + "step": 4663 + }, + { + "epoch": 0.3845963552403727, + "grad_norm": 1.9231138229370117, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7304214239120483, + "num_tokens": 30659195.0, + "step": 4664 + }, + { + "epoch": 0.38467881586542424, + "grad_norm": 2.2432713508605957, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7618604302406311, + "num_tokens": 30684992.0, + "step": 4665 + }, + { + "epoch": 0.3847612764904758, + "grad_norm": 1.9582908153533936, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7134020328521729, + "num_tokens": 30715600.0, + "step": 4666 + }, + { + "epoch": 0.38484373711552733, + "grad_norm": 1.926503300666809, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7482888698577881, + "num_tokens": 30748678.0, + "step": 4667 + }, + { + "epoch": 0.38492619774057885, + "grad_norm": 2.062764883041382, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7350173592567444, + "num_tokens": 30779441.0, + "step": 4668 + }, + { + "epoch": 0.3850086583656304, + "grad_norm": 1.9051235914230347, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7249318957328796, + "num_tokens": 30814223.0, + "step": 4669 + }, + { + "epoch": 0.38509111899068194, + "grad_norm": 2.114589214324951, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7333818078041077, + "num_tokens": 30840767.0, + "step": 4670 + }, + { + "epoch": 0.3851735796157335, + "grad_norm": 1.8917815685272217, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7607065439224243, + "num_tokens": 30871179.0, + "step": 4671 + }, + { + "epoch": 0.385256040240785, + "grad_norm": 2.0446670055389404, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7102648019790649, + "num_tokens": 30899584.0, + "step": 4672 + }, + { + "epoch": 0.38533850086583654, + "grad_norm": 1.9456932544708252, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7375727295875549, + "num_tokens": 30931223.0, + "step": 4673 + }, + { + "epoch": 0.3854209614908881, + "grad_norm": 1.7021375894546509, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7606769800186157, + "num_tokens": 30966874.0, + "step": 4674 + }, + { + "epoch": 0.38550342211593963, + "grad_norm": 2.092817783355713, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7090487480163574, + "num_tokens": 30995146.0, + "step": 4675 + }, + { + "epoch": 0.3855858827409912, + "grad_norm": 1.9247103929519653, + "learning_rate": 1e-06, + "loss": 0.6147, + "mean_token_accuracy": 0.8045984506607056, + "num_tokens": 31023336.0, + "step": 4676 + }, + { + "epoch": 0.3856683433660427, + "grad_norm": 2.131647825241089, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7519718408584595, + "num_tokens": 31046927.0, + "step": 4677 + }, + { + "epoch": 0.38575080399109424, + "grad_norm": 1.9804779291152954, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7443377375602722, + "num_tokens": 31075730.0, + "step": 4678 + }, + { + "epoch": 0.3858332646161458, + "grad_norm": 1.9070154428482056, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7203273773193359, + "num_tokens": 31110621.0, + "step": 4679 + }, + { + "epoch": 0.3859157252411973, + "grad_norm": 1.8936618566513062, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7429822683334351, + "num_tokens": 31142928.0, + "step": 4680 + }, + { + "epoch": 0.38599818586624884, + "grad_norm": 2.1443428993225098, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.717485249042511, + "num_tokens": 31170301.0, + "step": 4681 + }, + { + "epoch": 0.3860806464913004, + "grad_norm": 2.420623779296875, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7497883439064026, + "num_tokens": 31193194.0, + "step": 4682 + }, + { + "epoch": 0.38616310711635193, + "grad_norm": 1.8341310024261475, + "learning_rate": 1e-06, + "loss": 0.6085, + "mean_token_accuracy": 0.8001194596290588, + "num_tokens": 31225156.0, + "step": 4683 + }, + { + "epoch": 0.3862455677414035, + "grad_norm": 1.944283366203308, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7491215467453003, + "num_tokens": 31255063.0, + "step": 4684 + }, + { + "epoch": 0.386328028366455, + "grad_norm": 2.334425926208496, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7674793004989624, + "num_tokens": 31279647.0, + "step": 4685 + }, + { + "epoch": 0.38641048899150654, + "grad_norm": 1.7532674074172974, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7707276940345764, + "num_tokens": 31315716.0, + "step": 4686 + }, + { + "epoch": 0.3864929496165581, + "grad_norm": 2.0829110145568848, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7555167078971863, + "num_tokens": 31344292.0, + "step": 4687 + }, + { + "epoch": 0.38657541024160963, + "grad_norm": 2.137223243713379, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7413880825042725, + "num_tokens": 31368975.0, + "step": 4688 + }, + { + "epoch": 0.38665787086666115, + "grad_norm": 1.894829273223877, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.793419361114502, + "num_tokens": 31401215.0, + "step": 4689 + }, + { + "epoch": 0.3867403314917127, + "grad_norm": 2.062098264694214, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7322686910629272, + "num_tokens": 31429505.0, + "step": 4690 + }, + { + "epoch": 0.38682279211676424, + "grad_norm": 2.081202983856201, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.749840497970581, + "num_tokens": 31457247.0, + "step": 4691 + }, + { + "epoch": 0.3869052527418158, + "grad_norm": 1.8913809061050415, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7454809546470642, + "num_tokens": 31492820.0, + "step": 4692 + }, + { + "epoch": 0.3869877133668673, + "grad_norm": 1.8383009433746338, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7455903887748718, + "num_tokens": 31525601.0, + "step": 4693 + }, + { + "epoch": 0.38707017399191884, + "grad_norm": 1.8498680591583252, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.738849937915802, + "num_tokens": 31559538.0, + "step": 4694 + }, + { + "epoch": 0.3871526346169704, + "grad_norm": 1.995460033416748, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7358648777008057, + "num_tokens": 31590600.0, + "step": 4695 + }, + { + "epoch": 0.38723509524202193, + "grad_norm": 1.79999577999115, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7668783664703369, + "num_tokens": 31622545.0, + "step": 4696 + }, + { + "epoch": 0.38731755586707345, + "grad_norm": 1.9023988246917725, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.7700457572937012, + "num_tokens": 31652920.0, + "step": 4697 + }, + { + "epoch": 0.387400016492125, + "grad_norm": 1.995336651802063, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7571783065795898, + "num_tokens": 31680031.0, + "step": 4698 + }, + { + "epoch": 0.38748247711717654, + "grad_norm": 2.2553086280822754, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7583456635475159, + "num_tokens": 31703896.0, + "step": 4699 + }, + { + "epoch": 0.3875649377422281, + "grad_norm": 1.7879117727279663, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.74267578125, + "num_tokens": 31738378.0, + "step": 4700 + }, + { + "epoch": 0.3876473983672796, + "grad_norm": 1.7038319110870361, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7210584878921509, + "num_tokens": 31778380.0, + "step": 4701 + }, + { + "epoch": 0.38772985899233114, + "grad_norm": 1.8303569555282593, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.7822405695915222, + "num_tokens": 31812058.0, + "step": 4702 + }, + { + "epoch": 0.3878123196173827, + "grad_norm": 2.4381942749023438, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7326833605766296, + "num_tokens": 31834112.0, + "step": 4703 + }, + { + "epoch": 0.38789478024243423, + "grad_norm": 2.18963360786438, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.7016298770904541, + "num_tokens": 31861723.0, + "step": 4704 + }, + { + "epoch": 0.38797724086748575, + "grad_norm": 2.0671660900115967, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7020082473754883, + "num_tokens": 31891050.0, + "step": 4705 + }, + { + "epoch": 0.3880597014925373, + "grad_norm": 2.0148799419403076, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7123139500617981, + "num_tokens": 31921334.0, + "step": 4706 + }, + { + "epoch": 0.38814216211758884, + "grad_norm": 1.9650936126708984, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.774810791015625, + "num_tokens": 31952546.0, + "step": 4707 + }, + { + "epoch": 0.3882246227426404, + "grad_norm": 2.0219340324401855, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7804961800575256, + "num_tokens": 31980308.0, + "step": 4708 + }, + { + "epoch": 0.38830708336769193, + "grad_norm": 2.1453726291656494, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.72931307554245, + "num_tokens": 32008555.0, + "step": 4709 + }, + { + "epoch": 0.38838954399274345, + "grad_norm": 1.9784305095672607, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7491269111633301, + "num_tokens": 32038571.0, + "step": 4710 + }, + { + "epoch": 0.388472004617795, + "grad_norm": 2.094987154006958, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7200003862380981, + "num_tokens": 32067706.0, + "step": 4711 + }, + { + "epoch": 0.38855446524284654, + "grad_norm": 2.037431001663208, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7788447141647339, + "num_tokens": 32097369.0, + "step": 4712 + }, + { + "epoch": 0.38863692586789805, + "grad_norm": 2.1848278045654297, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6937326192855835, + "num_tokens": 32125872.0, + "step": 4713 + }, + { + "epoch": 0.3887193864929496, + "grad_norm": 2.440030336380005, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7114760875701904, + "num_tokens": 32148751.0, + "step": 4714 + }, + { + "epoch": 0.38880184711800114, + "grad_norm": 2.0825653076171875, + "learning_rate": 1e-06, + "loss": 1.0405, + "mean_token_accuracy": 0.6918720006942749, + "num_tokens": 32179459.0, + "step": 4715 + }, + { + "epoch": 0.3888843077430527, + "grad_norm": 2.029078960418701, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7122461795806885, + "num_tokens": 32209077.0, + "step": 4716 + }, + { + "epoch": 0.38896676836810423, + "grad_norm": 1.9242243766784668, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7306705713272095, + "num_tokens": 32243674.0, + "step": 4717 + }, + { + "epoch": 0.38904922899315575, + "grad_norm": 2.3763887882232666, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7376128435134888, + "num_tokens": 32265536.0, + "step": 4718 + }, + { + "epoch": 0.3891316896182073, + "grad_norm": 2.119072198867798, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.717300534248352, + "num_tokens": 32294523.0, + "step": 4719 + }, + { + "epoch": 0.38921415024325884, + "grad_norm": 2.000788450241089, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.714819610118866, + "num_tokens": 32325643.0, + "step": 4720 + }, + { + "epoch": 0.38929661086831036, + "grad_norm": 1.9944480657577515, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7391152381896973, + "num_tokens": 32353553.0, + "step": 4721 + }, + { + "epoch": 0.38937907149336193, + "grad_norm": 2.068188190460205, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7485332489013672, + "num_tokens": 32382034.0, + "step": 4722 + }, + { + "epoch": 0.38946153211841344, + "grad_norm": 1.9640593528747559, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7379359006881714, + "num_tokens": 32411474.0, + "step": 4723 + }, + { + "epoch": 0.389543992743465, + "grad_norm": 2.044137954711914, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7431162595748901, + "num_tokens": 32442019.0, + "step": 4724 + }, + { + "epoch": 0.38962645336851653, + "grad_norm": 1.8789944648742676, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7778118848800659, + "num_tokens": 32472867.0, + "step": 4725 + }, + { + "epoch": 0.38970891399356805, + "grad_norm": 2.094635486602783, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7292905449867249, + "num_tokens": 32500847.0, + "step": 4726 + }, + { + "epoch": 0.3897913746186196, + "grad_norm": 1.858967661857605, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7483763694763184, + "num_tokens": 32531935.0, + "step": 4727 + }, + { + "epoch": 0.38987383524367114, + "grad_norm": 2.1289775371551514, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7115002870559692, + "num_tokens": 32558889.0, + "step": 4728 + }, + { + "epoch": 0.38995629586872266, + "grad_norm": 1.8833131790161133, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7853403091430664, + "num_tokens": 32591243.0, + "step": 4729 + }, + { + "epoch": 0.39003875649377423, + "grad_norm": 1.836316704750061, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7431550025939941, + "num_tokens": 32625033.0, + "step": 4730 + }, + { + "epoch": 0.39012121711882575, + "grad_norm": 2.2705423831939697, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7441075444221497, + "num_tokens": 32649737.0, + "step": 4731 + }, + { + "epoch": 0.3902036777438773, + "grad_norm": 2.1647348403930664, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7451775670051575, + "num_tokens": 32678428.0, + "step": 4732 + }, + { + "epoch": 0.39028613836892884, + "grad_norm": 2.1051056385040283, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7167147994041443, + "num_tokens": 32704555.0, + "step": 4733 + }, + { + "epoch": 0.39036859899398035, + "grad_norm": 2.122670888900757, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7482857704162598, + "num_tokens": 32730794.0, + "step": 4734 + }, + { + "epoch": 0.3904510596190319, + "grad_norm": 2.304353952407837, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7342342138290405, + "num_tokens": 32754777.0, + "step": 4735 + }, + { + "epoch": 0.39053352024408344, + "grad_norm": 1.9820888042449951, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7723689079284668, + "num_tokens": 32784476.0, + "step": 4736 + }, + { + "epoch": 0.390615980869135, + "grad_norm": 1.8541373014450073, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7447221279144287, + "num_tokens": 32817880.0, + "step": 4737 + }, + { + "epoch": 0.39069844149418653, + "grad_norm": 2.140700101852417, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7342825531959534, + "num_tokens": 32844563.0, + "step": 4738 + }, + { + "epoch": 0.39078090211923805, + "grad_norm": 1.8930901288986206, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7807731628417969, + "num_tokens": 32873901.0, + "step": 4739 + }, + { + "epoch": 0.3908633627442896, + "grad_norm": 2.1589157581329346, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7334563732147217, + "num_tokens": 32900097.0, + "step": 4740 + }, + { + "epoch": 0.39094582336934114, + "grad_norm": 2.236053705215454, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7553642988204956, + "num_tokens": 32933381.0, + "step": 4741 + }, + { + "epoch": 0.39102828399439266, + "grad_norm": 2.4606943130493164, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7390953302383423, + "num_tokens": 32957341.0, + "step": 4742 + }, + { + "epoch": 0.39111074461944423, + "grad_norm": 2.5063397884368896, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7150247097015381, + "num_tokens": 32978594.0, + "step": 4743 + }, + { + "epoch": 0.39119320524449575, + "grad_norm": 1.8747164011001587, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.755915641784668, + "num_tokens": 33012003.0, + "step": 4744 + }, + { + "epoch": 0.3912756658695473, + "grad_norm": 2.062865734100342, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.738632321357727, + "num_tokens": 33041335.0, + "step": 4745 + }, + { + "epoch": 0.39135812649459883, + "grad_norm": 2.0468978881835938, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7647340297698975, + "num_tokens": 33070828.0, + "step": 4746 + }, + { + "epoch": 0.39144058711965035, + "grad_norm": 1.9593720436096191, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7163698673248291, + "num_tokens": 33106253.0, + "step": 4747 + }, + { + "epoch": 0.3915230477447019, + "grad_norm": 1.7656795978546143, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.731935441493988, + "num_tokens": 33143630.0, + "step": 4748 + }, + { + "epoch": 0.39160550836975344, + "grad_norm": 2.060593843460083, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7562808990478516, + "num_tokens": 33170486.0, + "step": 4749 + }, + { + "epoch": 0.39168796899480496, + "grad_norm": 1.9546575546264648, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7687931060791016, + "num_tokens": 33200236.0, + "step": 4750 + }, + { + "epoch": 0.39177042961985653, + "grad_norm": 2.0765380859375, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7131472826004028, + "num_tokens": 33227161.0, + "step": 4751 + }, + { + "epoch": 0.39185289024490805, + "grad_norm": 2.106558084487915, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7101096510887146, + "num_tokens": 33258874.0, + "step": 4752 + }, + { + "epoch": 0.3919353508699596, + "grad_norm": 2.039761781692505, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7365208864212036, + "num_tokens": 33290649.0, + "step": 4753 + }, + { + "epoch": 0.39201781149501114, + "grad_norm": 2.2037248611450195, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.77143394947052, + "num_tokens": 33313403.0, + "step": 4754 + }, + { + "epoch": 0.39210027212006265, + "grad_norm": 2.002230405807495, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7324739694595337, + "num_tokens": 33344812.0, + "step": 4755 + }, + { + "epoch": 0.3921827327451142, + "grad_norm": 1.9220178127288818, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7263708710670471, + "num_tokens": 33379474.0, + "step": 4756 + }, + { + "epoch": 0.39226519337016574, + "grad_norm": 1.8664902448654175, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.737745463848114, + "num_tokens": 33411346.0, + "step": 4757 + }, + { + "epoch": 0.39234765399521726, + "grad_norm": 2.0149242877960205, + "learning_rate": 1e-06, + "loss": 0.7498, + "mean_token_accuracy": 0.7525638937950134, + "num_tokens": 33438426.0, + "step": 4758 + }, + { + "epoch": 0.39243011462026883, + "grad_norm": 1.8130145072937012, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.7889053821563721, + "num_tokens": 33472035.0, + "step": 4759 + }, + { + "epoch": 0.39251257524532035, + "grad_norm": 1.8528814315795898, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.7887135744094849, + "num_tokens": 33501508.0, + "step": 4760 + }, + { + "epoch": 0.3925950358703719, + "grad_norm": 2.390892267227173, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7310328483581543, + "num_tokens": 33524854.0, + "step": 4761 + }, + { + "epoch": 0.39267749649542344, + "grad_norm": 2.0679728984832764, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7003654837608337, + "num_tokens": 33553909.0, + "step": 4762 + }, + { + "epoch": 0.39275995712047496, + "grad_norm": 1.8053058385849, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7482006549835205, + "num_tokens": 33591447.0, + "step": 4763 + }, + { + "epoch": 0.39284241774552653, + "grad_norm": 1.9962519407272339, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7282788753509521, + "num_tokens": 33621812.0, + "step": 4764 + }, + { + "epoch": 0.39292487837057805, + "grad_norm": 2.1403069496154785, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7270894646644592, + "num_tokens": 33648958.0, + "step": 4765 + }, + { + "epoch": 0.39300733899562956, + "grad_norm": 1.7160738706588745, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7587138414382935, + "num_tokens": 33689362.0, + "step": 4766 + }, + { + "epoch": 0.39308979962068114, + "grad_norm": 2.1200828552246094, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7350976467132568, + "num_tokens": 33718789.0, + "step": 4767 + }, + { + "epoch": 0.39317226024573265, + "grad_norm": 2.1562774181365967, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7208235263824463, + "num_tokens": 33746355.0, + "step": 4768 + }, + { + "epoch": 0.3932547208707842, + "grad_norm": 2.3543126583099365, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7325792908668518, + "num_tokens": 33770092.0, + "step": 4769 + }, + { + "epoch": 0.39333718149583574, + "grad_norm": 1.9154658317565918, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7373000383377075, + "num_tokens": 33799683.0, + "step": 4770 + }, + { + "epoch": 0.39341964212088726, + "grad_norm": 1.9942786693572998, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7513654232025146, + "num_tokens": 33827184.0, + "step": 4771 + }, + { + "epoch": 0.39350210274593883, + "grad_norm": 2.125243902206421, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7400095462799072, + "num_tokens": 33852812.0, + "step": 4772 + }, + { + "epoch": 0.39358456337099035, + "grad_norm": 2.013519763946533, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7272387742996216, + "num_tokens": 33882231.0, + "step": 4773 + }, + { + "epoch": 0.39366702399604186, + "grad_norm": 2.0326130390167236, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7303756475448608, + "num_tokens": 33909226.0, + "step": 4774 + }, + { + "epoch": 0.39374948462109344, + "grad_norm": 1.7300028800964355, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.7958398461341858, + "num_tokens": 33944259.0, + "step": 4775 + }, + { + "epoch": 0.39383194524614495, + "grad_norm": 2.081695079803467, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7510980367660522, + "num_tokens": 33970232.0, + "step": 4776 + }, + { + "epoch": 0.3939144058711965, + "grad_norm": 2.1391963958740234, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7580349445343018, + "num_tokens": 33994904.0, + "step": 4777 + }, + { + "epoch": 0.39399686649624804, + "grad_norm": 2.2225863933563232, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.745233416557312, + "num_tokens": 34019022.0, + "step": 4778 + }, + { + "epoch": 0.39407932712129956, + "grad_norm": 2.28289532661438, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7155452966690063, + "num_tokens": 34043537.0, + "step": 4779 + }, + { + "epoch": 0.39416178774635113, + "grad_norm": 1.9298195838928223, + "learning_rate": 1e-06, + "loss": 0.7746, + "mean_token_accuracy": 0.7588956356048584, + "num_tokens": 34076374.0, + "step": 4780 + }, + { + "epoch": 0.39424424837140265, + "grad_norm": 2.0342161655426025, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7661839127540588, + "num_tokens": 34109918.0, + "step": 4781 + }, + { + "epoch": 0.39432670899645417, + "grad_norm": 1.869322657585144, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7440736293792725, + "num_tokens": 34141253.0, + "step": 4782 + }, + { + "epoch": 0.39440916962150574, + "grad_norm": 1.89442777633667, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7629787921905518, + "num_tokens": 34174241.0, + "step": 4783 + }, + { + "epoch": 0.39449163024655726, + "grad_norm": 2.1132421493530273, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7291316390037537, + "num_tokens": 34203638.0, + "step": 4784 + }, + { + "epoch": 0.39457409087160883, + "grad_norm": 1.9730461835861206, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7532762289047241, + "num_tokens": 34230689.0, + "step": 4785 + }, + { + "epoch": 0.39465655149666035, + "grad_norm": 2.2007415294647217, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7224946022033691, + "num_tokens": 34255756.0, + "step": 4786 + }, + { + "epoch": 0.39473901212171186, + "grad_norm": 1.960597276687622, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.774621307849884, + "num_tokens": 34283151.0, + "step": 4787 + }, + { + "epoch": 0.39482147274676344, + "grad_norm": 1.9853254556655884, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7294734716415405, + "num_tokens": 34312377.0, + "step": 4788 + }, + { + "epoch": 0.39490393337181495, + "grad_norm": 1.9069892168045044, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7113522291183472, + "num_tokens": 34346945.0, + "step": 4789 + }, + { + "epoch": 0.39498639399686647, + "grad_norm": 2.036534070968628, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7162709832191467, + "num_tokens": 34380145.0, + "step": 4790 + }, + { + "epoch": 0.39506885462191804, + "grad_norm": 1.9244499206542969, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7384217977523804, + "num_tokens": 34413442.0, + "step": 4791 + }, + { + "epoch": 0.39515131524696956, + "grad_norm": 1.7946077585220337, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7918186187744141, + "num_tokens": 34443948.0, + "step": 4792 + }, + { + "epoch": 0.39523377587202113, + "grad_norm": 1.9527684450149536, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7528209090232849, + "num_tokens": 34475582.0, + "step": 4793 + }, + { + "epoch": 0.39531623649707265, + "grad_norm": 2.090386152267456, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.7851240038871765, + "num_tokens": 34499401.0, + "step": 4794 + }, + { + "epoch": 0.39539869712212417, + "grad_norm": 2.0204334259033203, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7655645608901978, + "num_tokens": 34526474.0, + "step": 4795 + }, + { + "epoch": 0.39548115774717574, + "grad_norm": 1.8491612672805786, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7748217582702637, + "num_tokens": 34557905.0, + "step": 4796 + }, + { + "epoch": 0.39556361837222725, + "grad_norm": 1.948344111442566, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7226786017417908, + "num_tokens": 34586018.0, + "step": 4797 + }, + { + "epoch": 0.39564607899727877, + "grad_norm": 2.0870401859283447, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7203958630561829, + "num_tokens": 34613441.0, + "step": 4798 + }, + { + "epoch": 0.39572853962233034, + "grad_norm": 1.9021579027175903, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7564452290534973, + "num_tokens": 34644525.0, + "step": 4799 + }, + { + "epoch": 0.39581100024738186, + "grad_norm": 1.7334403991699219, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7471296787261963, + "num_tokens": 34682027.0, + "step": 4800 + }, + { + "epoch": 0.39589346087243343, + "grad_norm": 1.940330147743225, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7149233222007751, + "num_tokens": 34711470.0, + "step": 4801 + }, + { + "epoch": 0.39597592149748495, + "grad_norm": 2.01761531829834, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7548286318778992, + "num_tokens": 34738995.0, + "step": 4802 + }, + { + "epoch": 0.39605838212253647, + "grad_norm": 1.803003191947937, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7569157481193542, + "num_tokens": 34774540.0, + "step": 4803 + }, + { + "epoch": 0.39614084274758804, + "grad_norm": 2.1757540702819824, + "learning_rate": 1e-06, + "loss": 0.7825, + "mean_token_accuracy": 0.749458909034729, + "num_tokens": 34799792.0, + "step": 4804 + }, + { + "epoch": 0.39622330337263956, + "grad_norm": 1.8232883214950562, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7796304225921631, + "num_tokens": 34830649.0, + "step": 4805 + }, + { + "epoch": 0.39630576399769113, + "grad_norm": 1.9865926504135132, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.751478910446167, + "num_tokens": 34860319.0, + "step": 4806 + }, + { + "epoch": 0.39638822462274265, + "grad_norm": 2.174506664276123, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7279781699180603, + "num_tokens": 34887180.0, + "step": 4807 + }, + { + "epoch": 0.39647068524779416, + "grad_norm": 1.8771766424179077, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7497434616088867, + "num_tokens": 34918241.0, + "step": 4808 + }, + { + "epoch": 0.39655314587284574, + "grad_norm": 2.0563437938690186, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7546722888946533, + "num_tokens": 34945483.0, + "step": 4809 + }, + { + "epoch": 0.39663560649789725, + "grad_norm": 1.9215056896209717, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7348563075065613, + "num_tokens": 34976818.0, + "step": 4810 + }, + { + "epoch": 0.39671806712294877, + "grad_norm": 1.9379382133483887, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7793420553207397, + "num_tokens": 35006898.0, + "step": 4811 + }, + { + "epoch": 0.39680052774800034, + "grad_norm": 1.927120566368103, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7419614791870117, + "num_tokens": 35038217.0, + "step": 4812 + }, + { + "epoch": 0.39688298837305186, + "grad_norm": 2.162815570831299, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7522236704826355, + "num_tokens": 35064618.0, + "step": 4813 + }, + { + "epoch": 0.39696544899810343, + "grad_norm": 2.107034206390381, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7559242248535156, + "num_tokens": 35089642.0, + "step": 4814 + }, + { + "epoch": 0.39704790962315495, + "grad_norm": 1.9030230045318604, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7741509079933167, + "num_tokens": 35119417.0, + "step": 4815 + }, + { + "epoch": 0.39713037024820647, + "grad_norm": 1.849588394165039, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.751887321472168, + "num_tokens": 35154515.0, + "step": 4816 + }, + { + "epoch": 0.39721283087325804, + "grad_norm": 2.1462361812591553, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.740298330783844, + "num_tokens": 35179815.0, + "step": 4817 + }, + { + "epoch": 0.39729529149830956, + "grad_norm": 2.115405321121216, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7332195043563843, + "num_tokens": 35205035.0, + "step": 4818 + }, + { + "epoch": 0.39737775212336107, + "grad_norm": 1.8622841835021973, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7369082570075989, + "num_tokens": 35239570.0, + "step": 4819 + }, + { + "epoch": 0.39746021274841264, + "grad_norm": 1.9035340547561646, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7770811319351196, + "num_tokens": 35269226.0, + "step": 4820 + }, + { + "epoch": 0.39754267337346416, + "grad_norm": 2.086223602294922, + "learning_rate": 1e-06, + "loss": 1.1034, + "mean_token_accuracy": 0.6853667497634888, + "num_tokens": 35299700.0, + "step": 4821 + }, + { + "epoch": 0.39762513399851573, + "grad_norm": 2.0660910606384277, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7467740774154663, + "num_tokens": 35325961.0, + "step": 4822 + }, + { + "epoch": 0.39770759462356725, + "grad_norm": 1.9315063953399658, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7440351843833923, + "num_tokens": 35355804.0, + "step": 4823 + }, + { + "epoch": 0.39779005524861877, + "grad_norm": 1.968891978263855, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7259936928749084, + "num_tokens": 35385531.0, + "step": 4824 + }, + { + "epoch": 0.39787251587367034, + "grad_norm": 2.2672009468078613, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7482925057411194, + "num_tokens": 35409765.0, + "step": 4825 + }, + { + "epoch": 0.39795497649872186, + "grad_norm": 1.6972216367721558, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.765426516532898, + "num_tokens": 35445956.0, + "step": 4826 + }, + { + "epoch": 0.3980374371237734, + "grad_norm": 1.9999035596847534, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.79414302110672, + "num_tokens": 35473749.0, + "step": 4827 + }, + { + "epoch": 0.39811989774882495, + "grad_norm": 1.883862018585205, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.757470965385437, + "num_tokens": 35504998.0, + "step": 4828 + }, + { + "epoch": 0.39820235837387646, + "grad_norm": 2.1387956142425537, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7238589525222778, + "num_tokens": 35533433.0, + "step": 4829 + }, + { + "epoch": 0.39828481899892804, + "grad_norm": 2.0148510932922363, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7511082887649536, + "num_tokens": 35566656.0, + "step": 4830 + }, + { + "epoch": 0.39836727962397955, + "grad_norm": 2.1742031574249268, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7447035312652588, + "num_tokens": 35591522.0, + "step": 4831 + }, + { + "epoch": 0.39844974024903107, + "grad_norm": 1.9083094596862793, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7203015685081482, + "num_tokens": 35621434.0, + "step": 4832 + }, + { + "epoch": 0.39853220087408264, + "grad_norm": 2.1748156547546387, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6855899095535278, + "num_tokens": 35648015.0, + "step": 4833 + }, + { + "epoch": 0.39861466149913416, + "grad_norm": 1.8709020614624023, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.772824227809906, + "num_tokens": 35676935.0, + "step": 4834 + }, + { + "epoch": 0.3986971221241857, + "grad_norm": 1.6928399801254272, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7894711494445801, + "num_tokens": 35713252.0, + "step": 4835 + }, + { + "epoch": 0.39877958274923725, + "grad_norm": 2.3024117946624756, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.735826849937439, + "num_tokens": 35738039.0, + "step": 4836 + }, + { + "epoch": 0.39886204337428877, + "grad_norm": 2.139671564102173, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7234209179878235, + "num_tokens": 35766044.0, + "step": 4837 + }, + { + "epoch": 0.39894450399934034, + "grad_norm": 1.7819654941558838, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7424275875091553, + "num_tokens": 35801281.0, + "step": 4838 + }, + { + "epoch": 0.39902696462439186, + "grad_norm": 2.037968397140503, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7380225658416748, + "num_tokens": 35829164.0, + "step": 4839 + }, + { + "epoch": 0.3991094252494434, + "grad_norm": 1.9977126121520996, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7361551523208618, + "num_tokens": 35858587.0, + "step": 4840 + }, + { + "epoch": 0.39919188587449494, + "grad_norm": 2.0307044982910156, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.6976809501647949, + "num_tokens": 35890224.0, + "step": 4841 + }, + { + "epoch": 0.39927434649954646, + "grad_norm": 2.3109421730041504, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7339115142822266, + "num_tokens": 35913062.0, + "step": 4842 + }, + { + "epoch": 0.399356807124598, + "grad_norm": 2.0670371055603027, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7340554594993591, + "num_tokens": 35942692.0, + "step": 4843 + }, + { + "epoch": 0.39943926774964955, + "grad_norm": 1.8203763961791992, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.791070818901062, + "num_tokens": 35973319.0, + "step": 4844 + }, + { + "epoch": 0.39952172837470107, + "grad_norm": 1.98651123046875, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7361698746681213, + "num_tokens": 36003507.0, + "step": 4845 + }, + { + "epoch": 0.39960418899975264, + "grad_norm": 2.2273504734039307, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.729674220085144, + "num_tokens": 36029462.0, + "step": 4846 + }, + { + "epoch": 0.39968664962480416, + "grad_norm": 1.921826958656311, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7239298820495605, + "num_tokens": 36063107.0, + "step": 4847 + }, + { + "epoch": 0.3997691102498557, + "grad_norm": 1.7753254175186157, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.7066829800605774, + "num_tokens": 36100596.0, + "step": 4848 + }, + { + "epoch": 0.39985157087490725, + "grad_norm": 1.7774430513381958, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7550427913665771, + "num_tokens": 36139137.0, + "step": 4849 + }, + { + "epoch": 0.39993403149995876, + "grad_norm": 2.047417640686035, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7073537111282349, + "num_tokens": 36170038.0, + "step": 4850 + }, + { + "epoch": 0.4000164921250103, + "grad_norm": 1.8016200065612793, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7548480033874512, + "num_tokens": 36203014.0, + "step": 4851 + }, + { + "epoch": 0.40009895275006185, + "grad_norm": 1.9168429374694824, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7509693503379822, + "num_tokens": 36234974.0, + "step": 4852 + }, + { + "epoch": 0.40018141337511337, + "grad_norm": 1.9584513902664185, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.72743821144104, + "num_tokens": 36267809.0, + "step": 4853 + }, + { + "epoch": 0.40026387400016494, + "grad_norm": 1.9269410371780396, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7478642463684082, + "num_tokens": 36299628.0, + "step": 4854 + }, + { + "epoch": 0.40034633462521646, + "grad_norm": 1.6499565839767456, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7481173276901245, + "num_tokens": 36336687.0, + "step": 4855 + }, + { + "epoch": 0.400428795250268, + "grad_norm": 2.1839051246643066, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.7055517435073853, + "num_tokens": 36365066.0, + "step": 4856 + }, + { + "epoch": 0.40051125587531955, + "grad_norm": 2.0090932846069336, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.726213812828064, + "num_tokens": 36393937.0, + "step": 4857 + }, + { + "epoch": 0.40059371650037107, + "grad_norm": 2.0912981033325195, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7573845982551575, + "num_tokens": 36419737.0, + "step": 4858 + }, + { + "epoch": 0.4006761771254226, + "grad_norm": 1.9726459980010986, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7562243938446045, + "num_tokens": 36451154.0, + "step": 4859 + }, + { + "epoch": 0.40075863775047416, + "grad_norm": 2.208547353744507, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.752475917339325, + "num_tokens": 36476659.0, + "step": 4860 + }, + { + "epoch": 0.4008410983755257, + "grad_norm": 1.8700087070465088, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7257656455039978, + "num_tokens": 36511048.0, + "step": 4861 + }, + { + "epoch": 0.40092355900057725, + "grad_norm": 2.044353723526001, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7503829598426819, + "num_tokens": 36538669.0, + "step": 4862 + }, + { + "epoch": 0.40100601962562876, + "grad_norm": 2.0453298091888428, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7220590114593506, + "num_tokens": 36567974.0, + "step": 4863 + }, + { + "epoch": 0.4010884802506803, + "grad_norm": 2.0899555683135986, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7301527261734009, + "num_tokens": 36595961.0, + "step": 4864 + }, + { + "epoch": 0.40117094087573185, + "grad_norm": 2.114708423614502, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7571377158164978, + "num_tokens": 36622464.0, + "step": 4865 + }, + { + "epoch": 0.40125340150078337, + "grad_norm": 2.226426601409912, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7734203338623047, + "num_tokens": 36647207.0, + "step": 4866 + }, + { + "epoch": 0.40133586212583494, + "grad_norm": 1.903608798980713, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7715582847595215, + "num_tokens": 36677619.0, + "step": 4867 + }, + { + "epoch": 0.40141832275088646, + "grad_norm": 1.755793571472168, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7541069984436035, + "num_tokens": 36713271.0, + "step": 4868 + }, + { + "epoch": 0.401500783375938, + "grad_norm": 1.9542350769042969, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7794507741928101, + "num_tokens": 36739572.0, + "step": 4869 + }, + { + "epoch": 0.40158324400098955, + "grad_norm": 1.8850284814834595, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7625811100006104, + "num_tokens": 36766552.0, + "step": 4870 + }, + { + "epoch": 0.40166570462604106, + "grad_norm": 1.9544538259506226, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7423995733261108, + "num_tokens": 36800982.0, + "step": 4871 + }, + { + "epoch": 0.4017481652510926, + "grad_norm": 2.2107815742492676, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7171512842178345, + "num_tokens": 36828389.0, + "step": 4872 + }, + { + "epoch": 0.40183062587614415, + "grad_norm": 1.6191588640213013, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.7985659837722778, + "num_tokens": 36866355.0, + "step": 4873 + }, + { + "epoch": 0.40191308650119567, + "grad_norm": 1.900464415550232, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7479264736175537, + "num_tokens": 36901523.0, + "step": 4874 + }, + { + "epoch": 0.40199554712624724, + "grad_norm": 2.122009038925171, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7232440710067749, + "num_tokens": 36927885.0, + "step": 4875 + }, + { + "epoch": 0.40207800775129876, + "grad_norm": 2.286975622177124, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.7109808325767517, + "num_tokens": 36951467.0, + "step": 4876 + }, + { + "epoch": 0.4021604683763503, + "grad_norm": 2.2669119834899902, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7208433151245117, + "num_tokens": 36975692.0, + "step": 4877 + }, + { + "epoch": 0.40224292900140185, + "grad_norm": 2.0388383865356445, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7153432369232178, + "num_tokens": 37006831.0, + "step": 4878 + }, + { + "epoch": 0.40232538962645337, + "grad_norm": 2.0162460803985596, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7425116300582886, + "num_tokens": 37037139.0, + "step": 4879 + }, + { + "epoch": 0.4024078502515049, + "grad_norm": 1.9680293798446655, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.74711012840271, + "num_tokens": 37068390.0, + "step": 4880 + }, + { + "epoch": 0.40249031087655646, + "grad_norm": 1.8536896705627441, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7250627279281616, + "num_tokens": 37102341.0, + "step": 4881 + }, + { + "epoch": 0.402572771501608, + "grad_norm": 2.2997171878814697, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7399842739105225, + "num_tokens": 37124373.0, + "step": 4882 + }, + { + "epoch": 0.40265523212665955, + "grad_norm": 1.9548736810684204, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.760418176651001, + "num_tokens": 37154997.0, + "step": 4883 + }, + { + "epoch": 0.40273769275171106, + "grad_norm": 2.267390012741089, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7482845187187195, + "num_tokens": 37178719.0, + "step": 4884 + }, + { + "epoch": 0.4028201533767626, + "grad_norm": 2.1871228218078613, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7463036775588989, + "num_tokens": 37202652.0, + "step": 4885 + }, + { + "epoch": 0.40290261400181415, + "grad_norm": 2.2676258087158203, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7314867973327637, + "num_tokens": 37226319.0, + "step": 4886 + }, + { + "epoch": 0.40298507462686567, + "grad_norm": 1.8779609203338623, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7358749508857727, + "num_tokens": 37259285.0, + "step": 4887 + }, + { + "epoch": 0.4030675352519172, + "grad_norm": 1.9794470071792603, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7441792488098145, + "num_tokens": 37287708.0, + "step": 4888 + }, + { + "epoch": 0.40314999587696876, + "grad_norm": 1.948408842086792, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7166763544082642, + "num_tokens": 37318134.0, + "step": 4889 + }, + { + "epoch": 0.4032324565020203, + "grad_norm": 1.7932982444763184, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7444523572921753, + "num_tokens": 37353598.0, + "step": 4890 + }, + { + "epoch": 0.40331491712707185, + "grad_norm": 1.8124070167541504, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.723397970199585, + "num_tokens": 37388783.0, + "step": 4891 + }, + { + "epoch": 0.40339737775212337, + "grad_norm": 1.8365328311920166, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7391633987426758, + "num_tokens": 37423163.0, + "step": 4892 + }, + { + "epoch": 0.4034798383771749, + "grad_norm": 1.891722321510315, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7780554294586182, + "num_tokens": 37452266.0, + "step": 4893 + }, + { + "epoch": 0.40356229900222645, + "grad_norm": 1.7972970008850098, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.729860246181488, + "num_tokens": 37488328.0, + "step": 4894 + }, + { + "epoch": 0.40364475962727797, + "grad_norm": 2.0115978717803955, + "learning_rate": 1e-06, + "loss": 0.7543, + "mean_token_accuracy": 0.7597514390945435, + "num_tokens": 37516707.0, + "step": 4895 + }, + { + "epoch": 0.4037272202523295, + "grad_norm": 2.03291654586792, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7447653412818909, + "num_tokens": 37543562.0, + "step": 4896 + }, + { + "epoch": 0.40380968087738106, + "grad_norm": 2.03328800201416, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7500888109207153, + "num_tokens": 37571438.0, + "step": 4897 + }, + { + "epoch": 0.4038921415024326, + "grad_norm": 1.9466289281845093, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7372661232948303, + "num_tokens": 37602399.0, + "step": 4898 + }, + { + "epoch": 0.40397460212748415, + "grad_norm": 1.910505771636963, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7302652597427368, + "num_tokens": 37635977.0, + "step": 4899 + }, + { + "epoch": 0.40405706275253567, + "grad_norm": 2.2190024852752686, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.709434986114502, + "num_tokens": 37663804.0, + "step": 4900 + }, + { + "epoch": 0.4041395233775872, + "grad_norm": 2.10955548286438, + "learning_rate": 1e-06, + "loss": 1.0558, + "mean_token_accuracy": 0.6878460049629211, + "num_tokens": 37693592.0, + "step": 4901 + }, + { + "epoch": 0.40422198400263876, + "grad_norm": 2.027547597885132, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7240767478942871, + "num_tokens": 37722513.0, + "step": 4902 + }, + { + "epoch": 0.4043044446276903, + "grad_norm": 2.033268690109253, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7657009363174438, + "num_tokens": 37751795.0, + "step": 4903 + }, + { + "epoch": 0.4043869052527418, + "grad_norm": 2.194394826889038, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7154462337493896, + "num_tokens": 37777012.0, + "step": 4904 + }, + { + "epoch": 0.40446936587779336, + "grad_norm": 1.9133720397949219, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7497453093528748, + "num_tokens": 37808172.0, + "step": 4905 + }, + { + "epoch": 0.4045518265028449, + "grad_norm": 1.949729084968567, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7424225211143494, + "num_tokens": 37841818.0, + "step": 4906 + }, + { + "epoch": 0.40463428712789645, + "grad_norm": 2.3268487453460693, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7186077237129211, + "num_tokens": 37864766.0, + "step": 4907 + }, + { + "epoch": 0.40471674775294797, + "grad_norm": 1.757849097251892, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7513167858123779, + "num_tokens": 37901507.0, + "step": 4908 + }, + { + "epoch": 0.4047992083779995, + "grad_norm": 2.0536985397338867, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7857710123062134, + "num_tokens": 37930546.0, + "step": 4909 + }, + { + "epoch": 0.40488166900305106, + "grad_norm": 2.042771577835083, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7272023558616638, + "num_tokens": 37961746.0, + "step": 4910 + }, + { + "epoch": 0.4049641296281026, + "grad_norm": 2.058711528778076, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7121095657348633, + "num_tokens": 37992030.0, + "step": 4911 + }, + { + "epoch": 0.4050465902531541, + "grad_norm": 1.8608803749084473, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7406865358352661, + "num_tokens": 38025158.0, + "step": 4912 + }, + { + "epoch": 0.40512905087820567, + "grad_norm": 1.95320725440979, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7410194873809814, + "num_tokens": 38052820.0, + "step": 4913 + }, + { + "epoch": 0.4052115115032572, + "grad_norm": 2.4011504650115967, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.705621600151062, + "num_tokens": 38078220.0, + "step": 4914 + }, + { + "epoch": 0.40529397212830875, + "grad_norm": 1.9377444982528687, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7605293989181519, + "num_tokens": 38109051.0, + "step": 4915 + }, + { + "epoch": 0.40537643275336027, + "grad_norm": 1.6706068515777588, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7300413846969604, + "num_tokens": 38152434.0, + "step": 4916 + }, + { + "epoch": 0.4054588933784118, + "grad_norm": 2.1066489219665527, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7263809442520142, + "num_tokens": 38179466.0, + "step": 4917 + }, + { + "epoch": 0.40554135400346336, + "grad_norm": 1.8086910247802734, + "learning_rate": 1e-06, + "loss": 0.5764, + "mean_token_accuracy": 0.8088952302932739, + "num_tokens": 38211174.0, + "step": 4918 + }, + { + "epoch": 0.4056238146285149, + "grad_norm": 2.002253532409668, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7625337243080139, + "num_tokens": 38239102.0, + "step": 4919 + }, + { + "epoch": 0.4057062752535664, + "grad_norm": 2.0581820011138916, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7398645877838135, + "num_tokens": 38265259.0, + "step": 4920 + }, + { + "epoch": 0.40578873587861797, + "grad_norm": 1.933851718902588, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7194807529449463, + "num_tokens": 38299780.0, + "step": 4921 + }, + { + "epoch": 0.4058711965036695, + "grad_norm": 1.9683512449264526, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7572340369224548, + "num_tokens": 38328045.0, + "step": 4922 + }, + { + "epoch": 0.40595365712872106, + "grad_norm": 2.033745765686035, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7288084030151367, + "num_tokens": 38356443.0, + "step": 4923 + }, + { + "epoch": 0.4060361177537726, + "grad_norm": 1.9352787733078003, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7360168099403381, + "num_tokens": 38387950.0, + "step": 4924 + }, + { + "epoch": 0.4061185783788241, + "grad_norm": 1.7238885164260864, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7768499851226807, + "num_tokens": 38421169.0, + "step": 4925 + }, + { + "epoch": 0.40620103900387566, + "grad_norm": 1.9221404790878296, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7620443105697632, + "num_tokens": 38448210.0, + "step": 4926 + }, + { + "epoch": 0.4062834996289272, + "grad_norm": 1.882592797279358, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7218165993690491, + "num_tokens": 38481822.0, + "step": 4927 + }, + { + "epoch": 0.4063659602539787, + "grad_norm": 2.0117318630218506, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7577946782112122, + "num_tokens": 38509832.0, + "step": 4928 + }, + { + "epoch": 0.40644842087903027, + "grad_norm": 1.9146873950958252, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7491365075111389, + "num_tokens": 38542824.0, + "step": 4929 + }, + { + "epoch": 0.4065308815040818, + "grad_norm": 2.2617392539978027, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7337105870246887, + "num_tokens": 38569081.0, + "step": 4930 + }, + { + "epoch": 0.40661334212913336, + "grad_norm": 1.6580777168273926, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7714757919311523, + "num_tokens": 38608427.0, + "step": 4931 + }, + { + "epoch": 0.4066958027541849, + "grad_norm": 2.2311720848083496, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7275344133377075, + "num_tokens": 38634963.0, + "step": 4932 + }, + { + "epoch": 0.4067782633792364, + "grad_norm": 1.75791597366333, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7765436172485352, + "num_tokens": 38667636.0, + "step": 4933 + }, + { + "epoch": 0.40686072400428797, + "grad_norm": 1.8480089902877808, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7319474220275879, + "num_tokens": 38706148.0, + "step": 4934 + }, + { + "epoch": 0.4069431846293395, + "grad_norm": 2.3344838619232178, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7660537958145142, + "num_tokens": 38728249.0, + "step": 4935 + }, + { + "epoch": 0.40702564525439106, + "grad_norm": 1.7757081985473633, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7565475702285767, + "num_tokens": 38764033.0, + "step": 4936 + }, + { + "epoch": 0.40710810587944257, + "grad_norm": 1.9554482698440552, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.7748876810073853, + "num_tokens": 38790949.0, + "step": 4937 + }, + { + "epoch": 0.4071905665044941, + "grad_norm": 1.8139798641204834, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7780744433403015, + "num_tokens": 38822932.0, + "step": 4938 + }, + { + "epoch": 0.40727302712954566, + "grad_norm": 1.9146689176559448, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7732171416282654, + "num_tokens": 38854335.0, + "step": 4939 + }, + { + "epoch": 0.4073554877545972, + "grad_norm": 2.0441503524780273, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7357265949249268, + "num_tokens": 38883501.0, + "step": 4940 + }, + { + "epoch": 0.4074379483796487, + "grad_norm": 1.9116415977478027, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7349887490272522, + "num_tokens": 38915944.0, + "step": 4941 + }, + { + "epoch": 0.40752040900470027, + "grad_norm": 2.1111955642700195, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7411761283874512, + "num_tokens": 38942396.0, + "step": 4942 + }, + { + "epoch": 0.4076028696297518, + "grad_norm": 2.304516553878784, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.698074460029602, + "num_tokens": 38967108.0, + "step": 4943 + }, + { + "epoch": 0.40768533025480336, + "grad_norm": 1.838616967201233, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7434892654418945, + "num_tokens": 39000207.0, + "step": 4944 + }, + { + "epoch": 0.4077677908798549, + "grad_norm": 2.0160470008850098, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7301053404808044, + "num_tokens": 39028027.0, + "step": 4945 + }, + { + "epoch": 0.4078502515049064, + "grad_norm": 1.7817999124526978, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7599952816963196, + "num_tokens": 39063914.0, + "step": 4946 + }, + { + "epoch": 0.40793271212995796, + "grad_norm": 1.9178156852722168, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7662963271141052, + "num_tokens": 39093914.0, + "step": 4947 + }, + { + "epoch": 0.4080151727550095, + "grad_norm": 2.110071897506714, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7451433539390564, + "num_tokens": 39121153.0, + "step": 4948 + }, + { + "epoch": 0.408097633380061, + "grad_norm": 2.1728250980377197, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7056580185890198, + "num_tokens": 39146795.0, + "step": 4949 + }, + { + "epoch": 0.40818009400511257, + "grad_norm": 1.8529788255691528, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7564151287078857, + "num_tokens": 39178803.0, + "step": 4950 + }, + { + "epoch": 0.4082625546301641, + "grad_norm": 1.866890549659729, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7768792510032654, + "num_tokens": 39209420.0, + "step": 4951 + }, + { + "epoch": 0.40834501525521566, + "grad_norm": 1.8907378911972046, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.7923376560211182, + "num_tokens": 39238020.0, + "step": 4952 + }, + { + "epoch": 0.4084274758802672, + "grad_norm": 2.2126688957214355, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7030402421951294, + "num_tokens": 39265675.0, + "step": 4953 + }, + { + "epoch": 0.4085099365053187, + "grad_norm": 1.8719463348388672, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.755481481552124, + "num_tokens": 39301927.0, + "step": 4954 + }, + { + "epoch": 0.40859239713037027, + "grad_norm": 1.9863743782043457, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7452967166900635, + "num_tokens": 39334295.0, + "step": 4955 + }, + { + "epoch": 0.4086748577554218, + "grad_norm": 2.091714859008789, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7256187796592712, + "num_tokens": 39363184.0, + "step": 4956 + }, + { + "epoch": 0.4087573183804733, + "grad_norm": 1.8463817834854126, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.782191276550293, + "num_tokens": 39392065.0, + "step": 4957 + }, + { + "epoch": 0.4088397790055249, + "grad_norm": 1.6899877786636353, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7589779496192932, + "num_tokens": 39429762.0, + "step": 4958 + }, + { + "epoch": 0.4089222396305764, + "grad_norm": 2.1995694637298584, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7441554069519043, + "num_tokens": 39452105.0, + "step": 4959 + }, + { + "epoch": 0.40900470025562796, + "grad_norm": 1.9080288410186768, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7564181089401245, + "num_tokens": 39485219.0, + "step": 4960 + }, + { + "epoch": 0.4090871608806795, + "grad_norm": 2.176060438156128, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7608376741409302, + "num_tokens": 39511230.0, + "step": 4961 + }, + { + "epoch": 0.409169621505731, + "grad_norm": 2.0682084560394287, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.743532121181488, + "num_tokens": 39537321.0, + "step": 4962 + }, + { + "epoch": 0.40925208213078257, + "grad_norm": 1.9643861055374146, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7584197521209717, + "num_tokens": 39565929.0, + "step": 4963 + }, + { + "epoch": 0.4093345427558341, + "grad_norm": 2.160026788711548, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7314344644546509, + "num_tokens": 39594606.0, + "step": 4964 + }, + { + "epoch": 0.4094170033808856, + "grad_norm": 1.8336478471755981, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.759490966796875, + "num_tokens": 39629569.0, + "step": 4965 + }, + { + "epoch": 0.4094994640059372, + "grad_norm": 2.332385778427124, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7163909673690796, + "num_tokens": 39653306.0, + "step": 4966 + }, + { + "epoch": 0.4095819246309887, + "grad_norm": 2.1030454635620117, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7235895395278931, + "num_tokens": 39685417.0, + "step": 4967 + }, + { + "epoch": 0.40966438525604026, + "grad_norm": 2.0664849281311035, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7182135581970215, + "num_tokens": 39715486.0, + "step": 4968 + }, + { + "epoch": 0.4097468458810918, + "grad_norm": 2.0284080505371094, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7495650053024292, + "num_tokens": 39746137.0, + "step": 4969 + }, + { + "epoch": 0.4098293065061433, + "grad_norm": 1.9466898441314697, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7543602585792542, + "num_tokens": 39780425.0, + "step": 4970 + }, + { + "epoch": 0.40991176713119487, + "grad_norm": 2.1817595958709717, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7286605834960938, + "num_tokens": 39805843.0, + "step": 4971 + }, + { + "epoch": 0.4099942277562464, + "grad_norm": 1.8272974491119385, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7263333201408386, + "num_tokens": 39840453.0, + "step": 4972 + }, + { + "epoch": 0.4100766883812979, + "grad_norm": 2.1158106327056885, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7099374532699585, + "num_tokens": 39869210.0, + "step": 4973 + }, + { + "epoch": 0.4101591490063495, + "grad_norm": 1.8079668283462524, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7780954837799072, + "num_tokens": 39903867.0, + "step": 4974 + }, + { + "epoch": 0.410241609631401, + "grad_norm": 1.9836505651474, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7510276436805725, + "num_tokens": 39934715.0, + "step": 4975 + }, + { + "epoch": 0.41032407025645257, + "grad_norm": 2.0720436573028564, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7255260944366455, + "num_tokens": 39963072.0, + "step": 4976 + }, + { + "epoch": 0.4104065308815041, + "grad_norm": 1.8135048151016235, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7569526433944702, + "num_tokens": 39995255.0, + "step": 4977 + }, + { + "epoch": 0.4104889915065556, + "grad_norm": 2.0004336833953857, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7528127431869507, + "num_tokens": 40024828.0, + "step": 4978 + }, + { + "epoch": 0.4105714521316072, + "grad_norm": 1.7766125202178955, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7376714944839478, + "num_tokens": 40059906.0, + "step": 4979 + }, + { + "epoch": 0.4106539127566587, + "grad_norm": 2.0748112201690674, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7151703834533691, + "num_tokens": 40088900.0, + "step": 4980 + }, + { + "epoch": 0.4107363733817102, + "grad_norm": 1.9430230855941772, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.7531150579452515, + "num_tokens": 40121695.0, + "step": 4981 + }, + { + "epoch": 0.4108188340067618, + "grad_norm": 2.5109970569610596, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7277321219444275, + "num_tokens": 40142571.0, + "step": 4982 + }, + { + "epoch": 0.4109012946318133, + "grad_norm": 2.0061981678009033, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7134396433830261, + "num_tokens": 40173040.0, + "step": 4983 + }, + { + "epoch": 0.41098375525686487, + "grad_norm": 1.8207716941833496, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7687342166900635, + "num_tokens": 40205216.0, + "step": 4984 + }, + { + "epoch": 0.4110662158819164, + "grad_norm": 2.2189950942993164, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7486532926559448, + "num_tokens": 40228365.0, + "step": 4985 + }, + { + "epoch": 0.4111486765069679, + "grad_norm": 1.9733911752700806, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7390083074569702, + "num_tokens": 40260710.0, + "step": 4986 + }, + { + "epoch": 0.4112311371320195, + "grad_norm": 1.7580360174179077, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.734338104724884, + "num_tokens": 40295510.0, + "step": 4987 + }, + { + "epoch": 0.411313597757071, + "grad_norm": 2.03863787651062, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7460377812385559, + "num_tokens": 40323618.0, + "step": 4988 + }, + { + "epoch": 0.4113960583821225, + "grad_norm": 1.8284608125686646, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7372397780418396, + "num_tokens": 40359469.0, + "step": 4989 + }, + { + "epoch": 0.4114785190071741, + "grad_norm": 2.131707191467285, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7283637523651123, + "num_tokens": 40387740.0, + "step": 4990 + }, + { + "epoch": 0.4115609796322256, + "grad_norm": 1.7291889190673828, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7587264776229858, + "num_tokens": 40425525.0, + "step": 4991 + }, + { + "epoch": 0.41164344025727717, + "grad_norm": 1.9875974655151367, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7270612716674805, + "num_tokens": 40458256.0, + "step": 4992 + }, + { + "epoch": 0.4117259008823287, + "grad_norm": 1.9769549369812012, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7518202066421509, + "num_tokens": 40488888.0, + "step": 4993 + }, + { + "epoch": 0.4118083615073802, + "grad_norm": 1.8912384510040283, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7633985877037048, + "num_tokens": 40522717.0, + "step": 4994 + }, + { + "epoch": 0.4118908221324318, + "grad_norm": 2.034665107727051, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7559305429458618, + "num_tokens": 40553273.0, + "step": 4995 + }, + { + "epoch": 0.4119732827574833, + "grad_norm": 1.961632251739502, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7444666028022766, + "num_tokens": 40582625.0, + "step": 4996 + }, + { + "epoch": 0.41205574338253487, + "grad_norm": 2.11440372467041, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7816239595413208, + "num_tokens": 40608875.0, + "step": 4997 + }, + { + "epoch": 0.4121382040075864, + "grad_norm": 1.7874407768249512, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7434860467910767, + "num_tokens": 40645131.0, + "step": 4998 + }, + { + "epoch": 0.4122206646326379, + "grad_norm": 1.9408676624298096, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.733647882938385, + "num_tokens": 40679127.0, + "step": 4999 + }, + { + "epoch": 0.4123031252576895, + "grad_norm": 2.0600297451019287, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7343599796295166, + "num_tokens": 40706197.0, + "step": 5000 + }, + { + "epoch": 0.412385585882741, + "grad_norm": 1.8755587339401245, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.7984674572944641, + "num_tokens": 40739243.0, + "step": 5001 + }, + { + "epoch": 0.4124680465077925, + "grad_norm": 2.142613410949707, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7666746377944946, + "num_tokens": 40766149.0, + "step": 5002 + }, + { + "epoch": 0.4125505071328441, + "grad_norm": 2.010129928588867, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7477931380271912, + "num_tokens": 40796498.0, + "step": 5003 + }, + { + "epoch": 0.4126329677578956, + "grad_norm": 1.6554882526397705, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7627188563346863, + "num_tokens": 40835071.0, + "step": 5004 + }, + { + "epoch": 0.41271542838294717, + "grad_norm": 1.7320466041564941, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7714179754257202, + "num_tokens": 40869305.0, + "step": 5005 + }, + { + "epoch": 0.4127978890079987, + "grad_norm": 1.9563359022140503, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7619506120681763, + "num_tokens": 40899404.0, + "step": 5006 + }, + { + "epoch": 0.4128803496330502, + "grad_norm": 2.2249796390533447, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7590392231941223, + "num_tokens": 40922246.0, + "step": 5007 + }, + { + "epoch": 0.4129628102581018, + "grad_norm": 1.8017520904541016, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.763451874256134, + "num_tokens": 40955525.0, + "step": 5008 + }, + { + "epoch": 0.4130452708831533, + "grad_norm": 1.711232304573059, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.7556203007698059, + "num_tokens": 40991563.0, + "step": 5009 + }, + { + "epoch": 0.4131277315082048, + "grad_norm": 2.1515579223632812, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7612487077713013, + "num_tokens": 41015360.0, + "step": 5010 + }, + { + "epoch": 0.4132101921332564, + "grad_norm": 1.7630032300949097, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7787631154060364, + "num_tokens": 41048168.0, + "step": 5011 + }, + { + "epoch": 0.4132926527583079, + "grad_norm": 1.992686152458191, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7415657043457031, + "num_tokens": 41077455.0, + "step": 5012 + }, + { + "epoch": 0.41337511338335947, + "grad_norm": 2.031308650970459, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7698718905448914, + "num_tokens": 41103534.0, + "step": 5013 + }, + { + "epoch": 0.413457574008411, + "grad_norm": 2.0353753566741943, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7502965927124023, + "num_tokens": 41130810.0, + "step": 5014 + }, + { + "epoch": 0.4135400346334625, + "grad_norm": 2.2619171142578125, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7693333625793457, + "num_tokens": 41163140.0, + "step": 5015 + }, + { + "epoch": 0.4136224952585141, + "grad_norm": 2.0462663173675537, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7352303862571716, + "num_tokens": 41193738.0, + "step": 5016 + }, + { + "epoch": 0.4137049558835656, + "grad_norm": 2.042590618133545, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7112839221954346, + "num_tokens": 41222432.0, + "step": 5017 + }, + { + "epoch": 0.4137874165086171, + "grad_norm": 1.926853060722351, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.698786735534668, + "num_tokens": 41254870.0, + "step": 5018 + }, + { + "epoch": 0.4138698771336687, + "grad_norm": 1.8497605323791504, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7469762563705444, + "num_tokens": 41288922.0, + "step": 5019 + }, + { + "epoch": 0.4139523377587202, + "grad_norm": 1.9963148832321167, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7172583341598511, + "num_tokens": 41318314.0, + "step": 5020 + }, + { + "epoch": 0.4140347983837718, + "grad_norm": 2.06022047996521, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7163741588592529, + "num_tokens": 41347299.0, + "step": 5021 + }, + { + "epoch": 0.4141172590088233, + "grad_norm": 2.133890151977539, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7460587024688721, + "num_tokens": 41373777.0, + "step": 5022 + }, + { + "epoch": 0.4141997196338748, + "grad_norm": 2.139892816543579, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7416876554489136, + "num_tokens": 41401187.0, + "step": 5023 + }, + { + "epoch": 0.4142821802589264, + "grad_norm": 2.165482521057129, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.73252272605896, + "num_tokens": 41429719.0, + "step": 5024 + }, + { + "epoch": 0.4143646408839779, + "grad_norm": 2.0049288272857666, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7243518233299255, + "num_tokens": 41458826.0, + "step": 5025 + }, + { + "epoch": 0.4144471015090294, + "grad_norm": 1.7032476663589478, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7758027911186218, + "num_tokens": 41493242.0, + "step": 5026 + }, + { + "epoch": 0.414529562134081, + "grad_norm": 1.9033408164978027, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.7727687358856201, + "num_tokens": 41525837.0, + "step": 5027 + }, + { + "epoch": 0.4146120227591325, + "grad_norm": 2.208104372024536, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7697710990905762, + "num_tokens": 41551059.0, + "step": 5028 + }, + { + "epoch": 0.4146944833841841, + "grad_norm": 2.0612435340881348, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7615121603012085, + "num_tokens": 41580309.0, + "step": 5029 + }, + { + "epoch": 0.4147769440092356, + "grad_norm": 2.0549514293670654, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7305961847305298, + "num_tokens": 41611809.0, + "step": 5030 + }, + { + "epoch": 0.4148594046342871, + "grad_norm": 1.8894776105880737, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.77718186378479, + "num_tokens": 41640717.0, + "step": 5031 + }, + { + "epoch": 0.4149418652593387, + "grad_norm": 2.0514557361602783, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7822551727294922, + "num_tokens": 41668678.0, + "step": 5032 + }, + { + "epoch": 0.4150243258843902, + "grad_norm": 2.0199923515319824, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.71622234582901, + "num_tokens": 41699850.0, + "step": 5033 + }, + { + "epoch": 0.4151067865094417, + "grad_norm": 1.8052916526794434, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7567325830459595, + "num_tokens": 41733795.0, + "step": 5034 + }, + { + "epoch": 0.4151892471344933, + "grad_norm": 2.060530662536621, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6880789399147034, + "num_tokens": 41763239.0, + "step": 5035 + }, + { + "epoch": 0.4152717077595448, + "grad_norm": 2.2698845863342285, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7150774002075195, + "num_tokens": 41789856.0, + "step": 5036 + }, + { + "epoch": 0.4153541683845964, + "grad_norm": 2.017674446105957, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7272886037826538, + "num_tokens": 41816767.0, + "step": 5037 + }, + { + "epoch": 0.4154366290096479, + "grad_norm": 1.9509835243225098, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7565629482269287, + "num_tokens": 41846483.0, + "step": 5038 + }, + { + "epoch": 0.4155190896346994, + "grad_norm": 1.8777750730514526, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7405439615249634, + "num_tokens": 41882092.0, + "step": 5039 + }, + { + "epoch": 0.415601550259751, + "grad_norm": 2.0169596672058105, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.774519681930542, + "num_tokens": 41911199.0, + "step": 5040 + }, + { + "epoch": 0.4156840108848025, + "grad_norm": 2.1569128036499023, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7751551866531372, + "num_tokens": 41939046.0, + "step": 5041 + }, + { + "epoch": 0.415766471509854, + "grad_norm": 1.9642277956008911, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7487348318099976, + "num_tokens": 41969048.0, + "step": 5042 + }, + { + "epoch": 0.4158489321349056, + "grad_norm": 2.12691330909729, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7437440156936646, + "num_tokens": 41996953.0, + "step": 5043 + }, + { + "epoch": 0.4159313927599571, + "grad_norm": 1.9391039609909058, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7477179765701294, + "num_tokens": 42024560.0, + "step": 5044 + }, + { + "epoch": 0.4160138533850087, + "grad_norm": 1.8424861431121826, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7384485006332397, + "num_tokens": 42061111.0, + "step": 5045 + }, + { + "epoch": 0.4160963140100602, + "grad_norm": 2.2960939407348633, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7545759677886963, + "num_tokens": 42083363.0, + "step": 5046 + }, + { + "epoch": 0.4161787746351117, + "grad_norm": 2.282890558242798, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7629011869430542, + "num_tokens": 42108676.0, + "step": 5047 + }, + { + "epoch": 0.4162612352601633, + "grad_norm": 1.7751368284225464, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7711930274963379, + "num_tokens": 42146069.0, + "step": 5048 + }, + { + "epoch": 0.4163436958852148, + "grad_norm": 1.8004145622253418, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7125146985054016, + "num_tokens": 42182801.0, + "step": 5049 + }, + { + "epoch": 0.4164261565102663, + "grad_norm": 1.959862470626831, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7183471918106079, + "num_tokens": 42213197.0, + "step": 5050 + }, + { + "epoch": 0.4165086171353179, + "grad_norm": 1.7277191877365112, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7790191173553467, + "num_tokens": 42254177.0, + "step": 5051 + }, + { + "epoch": 0.4165910777603694, + "grad_norm": 2.24637508392334, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7511454820632935, + "num_tokens": 42276837.0, + "step": 5052 + }, + { + "epoch": 0.416673538385421, + "grad_norm": 2.1262853145599365, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7316673994064331, + "num_tokens": 42305233.0, + "step": 5053 + }, + { + "epoch": 0.4167559990104725, + "grad_norm": 2.0323657989501953, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7176448106765747, + "num_tokens": 42339096.0, + "step": 5054 + }, + { + "epoch": 0.416838459635524, + "grad_norm": 2.1403260231018066, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.757715106010437, + "num_tokens": 42364853.0, + "step": 5055 + }, + { + "epoch": 0.4169209202605756, + "grad_norm": 2.3393049240112305, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7420583963394165, + "num_tokens": 42387990.0, + "step": 5056 + }, + { + "epoch": 0.4170033808856271, + "grad_norm": 1.8846306800842285, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7876774072647095, + "num_tokens": 42416863.0, + "step": 5057 + }, + { + "epoch": 0.4170858415106786, + "grad_norm": 2.073241710662842, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7235122919082642, + "num_tokens": 42443083.0, + "step": 5058 + }, + { + "epoch": 0.4171683021357302, + "grad_norm": 1.967836856842041, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7436686158180237, + "num_tokens": 42474834.0, + "step": 5059 + }, + { + "epoch": 0.4172507627607817, + "grad_norm": 2.206045389175415, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7532414197921753, + "num_tokens": 42499061.0, + "step": 5060 + }, + { + "epoch": 0.4173332233858333, + "grad_norm": 1.8930702209472656, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.766425609588623, + "num_tokens": 42530182.0, + "step": 5061 + }, + { + "epoch": 0.4174156840108848, + "grad_norm": 1.8892076015472412, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7190884351730347, + "num_tokens": 42564480.0, + "step": 5062 + }, + { + "epoch": 0.4174981446359363, + "grad_norm": 1.8201768398284912, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7647662162780762, + "num_tokens": 42596414.0, + "step": 5063 + }, + { + "epoch": 0.4175806052609879, + "grad_norm": 1.9595999717712402, + "learning_rate": 1e-06, + "loss": 0.765, + "mean_token_accuracy": 0.7580921649932861, + "num_tokens": 42624819.0, + "step": 5064 + }, + { + "epoch": 0.4176630658860394, + "grad_norm": 2.1183128356933594, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7206474542617798, + "num_tokens": 42652015.0, + "step": 5065 + }, + { + "epoch": 0.417745526511091, + "grad_norm": 2.1058058738708496, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7250678539276123, + "num_tokens": 42681751.0, + "step": 5066 + }, + { + "epoch": 0.4178279871361425, + "grad_norm": 1.9074918031692505, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7406352162361145, + "num_tokens": 42717907.0, + "step": 5067 + }, + { + "epoch": 0.417910447761194, + "grad_norm": 1.811510443687439, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7396502494812012, + "num_tokens": 42758486.0, + "step": 5068 + }, + { + "epoch": 0.4179929083862456, + "grad_norm": 1.9445877075195312, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.7033926844596863, + "num_tokens": 42791013.0, + "step": 5069 + }, + { + "epoch": 0.4180753690112971, + "grad_norm": 1.8818727731704712, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7297863364219666, + "num_tokens": 42822212.0, + "step": 5070 + }, + { + "epoch": 0.4181578296363486, + "grad_norm": 1.8591006994247437, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.752763032913208, + "num_tokens": 42856207.0, + "step": 5071 + }, + { + "epoch": 0.4182402902614002, + "grad_norm": 2.0063223838806152, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7522110342979431, + "num_tokens": 42885988.0, + "step": 5072 + }, + { + "epoch": 0.4183227508864517, + "grad_norm": 1.8257853984832764, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.765424907207489, + "num_tokens": 42919847.0, + "step": 5073 + }, + { + "epoch": 0.4184052115115033, + "grad_norm": 1.7595200538635254, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7510713338851929, + "num_tokens": 42955348.0, + "step": 5074 + }, + { + "epoch": 0.4184876721365548, + "grad_norm": 1.9992289543151855, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.732956051826477, + "num_tokens": 42990941.0, + "step": 5075 + }, + { + "epoch": 0.4185701327616063, + "grad_norm": 1.9233436584472656, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7799650430679321, + "num_tokens": 43019459.0, + "step": 5076 + }, + { + "epoch": 0.4186525933866579, + "grad_norm": 2.030085325241089, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7510391473770142, + "num_tokens": 43045991.0, + "step": 5077 + }, + { + "epoch": 0.4187350540117094, + "grad_norm": 1.8865655660629272, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7758121490478516, + "num_tokens": 43076289.0, + "step": 5078 + }, + { + "epoch": 0.4188175146367609, + "grad_norm": 2.0114645957946777, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7370978593826294, + "num_tokens": 43105901.0, + "step": 5079 + }, + { + "epoch": 0.4188999752618125, + "grad_norm": 1.9998550415039062, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7488840818405151, + "num_tokens": 43137268.0, + "step": 5080 + }, + { + "epoch": 0.418982435886864, + "grad_norm": 2.3233749866485596, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7283860445022583, + "num_tokens": 43161922.0, + "step": 5081 + }, + { + "epoch": 0.4190648965119156, + "grad_norm": 2.0907390117645264, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7216967344284058, + "num_tokens": 43190937.0, + "step": 5082 + }, + { + "epoch": 0.4191473571369671, + "grad_norm": 1.763257384300232, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7260223031044006, + "num_tokens": 43229022.0, + "step": 5083 + }, + { + "epoch": 0.4192298177620186, + "grad_norm": 1.7222421169281006, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.7888379096984863, + "num_tokens": 43266805.0, + "step": 5084 + }, + { + "epoch": 0.4193122783870702, + "grad_norm": 2.149970293045044, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7166469693183899, + "num_tokens": 43294976.0, + "step": 5085 + }, + { + "epoch": 0.4193947390121217, + "grad_norm": 2.1742138862609863, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7382441759109497, + "num_tokens": 43319212.0, + "step": 5086 + }, + { + "epoch": 0.4194771996371732, + "grad_norm": 2.2089786529541016, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7639554738998413, + "num_tokens": 43346263.0, + "step": 5087 + }, + { + "epoch": 0.4195596602622248, + "grad_norm": 1.7925693988800049, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7560867071151733, + "num_tokens": 43380078.0, + "step": 5088 + }, + { + "epoch": 0.4196421208872763, + "grad_norm": 1.9280389547348022, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7359272241592407, + "num_tokens": 43413386.0, + "step": 5089 + }, + { + "epoch": 0.4197245815123279, + "grad_norm": 1.928929090499878, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7766206860542297, + "num_tokens": 43439132.0, + "step": 5090 + }, + { + "epoch": 0.4198070421373794, + "grad_norm": 1.9758973121643066, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7406921982765198, + "num_tokens": 43469521.0, + "step": 5091 + }, + { + "epoch": 0.4198895027624309, + "grad_norm": 1.9162721633911133, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7584339380264282, + "num_tokens": 43498768.0, + "step": 5092 + }, + { + "epoch": 0.4199719633874825, + "grad_norm": 1.8388949632644653, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7776328325271606, + "num_tokens": 43529529.0, + "step": 5093 + }, + { + "epoch": 0.420054424012534, + "grad_norm": 2.2993922233581543, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7042093276977539, + "num_tokens": 43555038.0, + "step": 5094 + }, + { + "epoch": 0.42013688463758553, + "grad_norm": 1.7297035455703735, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7625138163566589, + "num_tokens": 43591025.0, + "step": 5095 + }, + { + "epoch": 0.4202193452626371, + "grad_norm": 2.0078277587890625, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7795995473861694, + "num_tokens": 43616981.0, + "step": 5096 + }, + { + "epoch": 0.4203018058876886, + "grad_norm": 1.9211630821228027, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7812143564224243, + "num_tokens": 43648505.0, + "step": 5097 + }, + { + "epoch": 0.4203842665127402, + "grad_norm": 1.8887416124343872, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7224084138870239, + "num_tokens": 43679864.0, + "step": 5098 + }, + { + "epoch": 0.4204667271377917, + "grad_norm": 1.9582839012145996, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7705327272415161, + "num_tokens": 43710983.0, + "step": 5099 + }, + { + "epoch": 0.4205491877628432, + "grad_norm": 1.8594287633895874, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7422604560852051, + "num_tokens": 43744797.0, + "step": 5100 + }, + { + "epoch": 0.4206316483878948, + "grad_norm": 1.893600344657898, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.770022988319397, + "num_tokens": 43774331.0, + "step": 5101 + }, + { + "epoch": 0.4207141090129463, + "grad_norm": 2.0449252128601074, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7534319162368774, + "num_tokens": 43799969.0, + "step": 5102 + }, + { + "epoch": 0.42079656963799783, + "grad_norm": 1.7962507009506226, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7780505418777466, + "num_tokens": 43834149.0, + "step": 5103 + }, + { + "epoch": 0.4208790302630494, + "grad_norm": 1.9750229120254517, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7522175312042236, + "num_tokens": 43863962.0, + "step": 5104 + }, + { + "epoch": 0.4209614908881009, + "grad_norm": 1.7158328294754028, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.794442892074585, + "num_tokens": 43894788.0, + "step": 5105 + }, + { + "epoch": 0.4210439515131525, + "grad_norm": 2.064582586288452, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7503699660301208, + "num_tokens": 43922168.0, + "step": 5106 + }, + { + "epoch": 0.421126412138204, + "grad_norm": 1.9630887508392334, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7214320302009583, + "num_tokens": 43953252.0, + "step": 5107 + }, + { + "epoch": 0.4212088727632555, + "grad_norm": 1.9434890747070312, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7150447368621826, + "num_tokens": 43986842.0, + "step": 5108 + }, + { + "epoch": 0.4212913333883071, + "grad_norm": 2.121095895767212, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7205819487571716, + "num_tokens": 44016016.0, + "step": 5109 + }, + { + "epoch": 0.4213737940133586, + "grad_norm": 1.970452070236206, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7274931073188782, + "num_tokens": 44046697.0, + "step": 5110 + }, + { + "epoch": 0.42145625463841013, + "grad_norm": 1.7480274438858032, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.7849849462509155, + "num_tokens": 44079555.0, + "step": 5111 + }, + { + "epoch": 0.4215387152634617, + "grad_norm": 2.0750041007995605, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7607656717300415, + "num_tokens": 44107841.0, + "step": 5112 + }, + { + "epoch": 0.4216211758885132, + "grad_norm": 1.8865479230880737, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7389705181121826, + "num_tokens": 44139862.0, + "step": 5113 + }, + { + "epoch": 0.4217036365135648, + "grad_norm": 1.790891170501709, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.7449805736541748, + "num_tokens": 44176174.0, + "step": 5114 + }, + { + "epoch": 0.4217860971386163, + "grad_norm": 2.3433544635772705, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7402704954147339, + "num_tokens": 44197132.0, + "step": 5115 + }, + { + "epoch": 0.42186855776366783, + "grad_norm": 2.4314610958099365, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7229564189910889, + "num_tokens": 44220971.0, + "step": 5116 + }, + { + "epoch": 0.4219510183887194, + "grad_norm": 1.8459150791168213, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.730935275554657, + "num_tokens": 44257160.0, + "step": 5117 + }, + { + "epoch": 0.4220334790137709, + "grad_norm": 2.1199748516082764, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7199386358261108, + "num_tokens": 44282928.0, + "step": 5118 + }, + { + "epoch": 0.42211593963882243, + "grad_norm": 2.129558801651001, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7677876353263855, + "num_tokens": 44308654.0, + "step": 5119 + }, + { + "epoch": 0.422198400263874, + "grad_norm": 1.8853873014450073, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7790964245796204, + "num_tokens": 44339789.0, + "step": 5120 + }, + { + "epoch": 0.4222808608889255, + "grad_norm": 1.9410772323608398, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7122402191162109, + "num_tokens": 44372063.0, + "step": 5121 + }, + { + "epoch": 0.4223633215139771, + "grad_norm": 1.8028265237808228, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.730028510093689, + "num_tokens": 44408412.0, + "step": 5122 + }, + { + "epoch": 0.4224457821390286, + "grad_norm": 1.850942611694336, + "learning_rate": 1e-06, + "loss": 0.5909, + "mean_token_accuracy": 0.8095663189888, + "num_tokens": 44437672.0, + "step": 5123 + }, + { + "epoch": 0.42252824276408013, + "grad_norm": 1.856384038925171, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7456051111221313, + "num_tokens": 44472556.0, + "step": 5124 + }, + { + "epoch": 0.4226107033891317, + "grad_norm": 1.9678376913070679, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.759427011013031, + "num_tokens": 44502894.0, + "step": 5125 + }, + { + "epoch": 0.4226931640141832, + "grad_norm": 2.3047642707824707, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7144054174423218, + "num_tokens": 44528739.0, + "step": 5126 + }, + { + "epoch": 0.4227756246392348, + "grad_norm": 1.8483500480651855, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7338892817497253, + "num_tokens": 44562806.0, + "step": 5127 + }, + { + "epoch": 0.4228580852642863, + "grad_norm": 2.0741682052612305, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.734103262424469, + "num_tokens": 44592375.0, + "step": 5128 + }, + { + "epoch": 0.4229405458893378, + "grad_norm": 2.2018308639526367, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7461598515510559, + "num_tokens": 44618772.0, + "step": 5129 + }, + { + "epoch": 0.4230230065143894, + "grad_norm": 1.9350367784500122, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7119570970535278, + "num_tokens": 44654908.0, + "step": 5130 + }, + { + "epoch": 0.4231054671394409, + "grad_norm": 2.057535171508789, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7638044953346252, + "num_tokens": 44683636.0, + "step": 5131 + }, + { + "epoch": 0.42318792776449243, + "grad_norm": 1.8081114292144775, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7629755139350891, + "num_tokens": 44716299.0, + "step": 5132 + }, + { + "epoch": 0.423270388389544, + "grad_norm": 1.8858013153076172, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7375497817993164, + "num_tokens": 44749494.0, + "step": 5133 + }, + { + "epoch": 0.4233528490145955, + "grad_norm": 2.2263457775115967, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7289974689483643, + "num_tokens": 44777236.0, + "step": 5134 + }, + { + "epoch": 0.4234353096396471, + "grad_norm": 2.0233521461486816, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7141762971878052, + "num_tokens": 44805867.0, + "step": 5135 + }, + { + "epoch": 0.4235177702646986, + "grad_norm": 1.9605962038040161, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.728347897529602, + "num_tokens": 44836733.0, + "step": 5136 + }, + { + "epoch": 0.42360023088975013, + "grad_norm": 2.0760817527770996, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7372808456420898, + "num_tokens": 44864564.0, + "step": 5137 + }, + { + "epoch": 0.4236826915148017, + "grad_norm": 2.0995805263519287, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7622878551483154, + "num_tokens": 44893422.0, + "step": 5138 + }, + { + "epoch": 0.4237651521398532, + "grad_norm": 1.8525198698043823, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.779077410697937, + "num_tokens": 44922371.0, + "step": 5139 + }, + { + "epoch": 0.42384761276490474, + "grad_norm": 2.0956928730010986, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7440165281295776, + "num_tokens": 44948907.0, + "step": 5140 + }, + { + "epoch": 0.4239300733899563, + "grad_norm": 1.8446581363677979, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7587829232215881, + "num_tokens": 44982918.0, + "step": 5141 + }, + { + "epoch": 0.4240125340150078, + "grad_norm": 2.0123472213745117, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7157458066940308, + "num_tokens": 45012153.0, + "step": 5142 + }, + { + "epoch": 0.4240949946400594, + "grad_norm": 1.9667601585388184, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7432296872138977, + "num_tokens": 45042302.0, + "step": 5143 + }, + { + "epoch": 0.4241774552651109, + "grad_norm": 2.0071523189544678, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7617993354797363, + "num_tokens": 45071119.0, + "step": 5144 + }, + { + "epoch": 0.42425991589016243, + "grad_norm": 1.5677841901779175, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7845156788825989, + "num_tokens": 45116158.0, + "step": 5145 + }, + { + "epoch": 0.424342376515214, + "grad_norm": 1.8812975883483887, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7367792129516602, + "num_tokens": 45148462.0, + "step": 5146 + }, + { + "epoch": 0.4244248371402655, + "grad_norm": 2.1431527137756348, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7101880311965942, + "num_tokens": 45176073.0, + "step": 5147 + }, + { + "epoch": 0.42450729776531704, + "grad_norm": 2.3719332218170166, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7212799787521362, + "num_tokens": 45202046.0, + "step": 5148 + }, + { + "epoch": 0.4245897583903686, + "grad_norm": 2.1400015354156494, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7267593145370483, + "num_tokens": 45230177.0, + "step": 5149 + }, + { + "epoch": 0.4246722190154201, + "grad_norm": 2.129913568496704, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6996873617172241, + "num_tokens": 45260460.0, + "step": 5150 + }, + { + "epoch": 0.4247546796404717, + "grad_norm": 2.240433692932129, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7651684284210205, + "num_tokens": 45285650.0, + "step": 5151 + }, + { + "epoch": 0.4248371402655232, + "grad_norm": 1.8093738555908203, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.762608528137207, + "num_tokens": 45318000.0, + "step": 5152 + }, + { + "epoch": 0.42491960089057473, + "grad_norm": 2.0135385990142822, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7337926030158997, + "num_tokens": 45346067.0, + "step": 5153 + }, + { + "epoch": 0.4250020615156263, + "grad_norm": 1.8578448295593262, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7481783628463745, + "num_tokens": 45379913.0, + "step": 5154 + }, + { + "epoch": 0.4250845221406778, + "grad_norm": 2.0400664806365967, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7540473937988281, + "num_tokens": 45406079.0, + "step": 5155 + }, + { + "epoch": 0.42516698276572934, + "grad_norm": 2.0927040576934814, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7302887439727783, + "num_tokens": 45433173.0, + "step": 5156 + }, + { + "epoch": 0.4252494433907809, + "grad_norm": 1.9619637727737427, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7592146396636963, + "num_tokens": 45459878.0, + "step": 5157 + }, + { + "epoch": 0.42533190401583243, + "grad_norm": 2.026141881942749, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7100794911384583, + "num_tokens": 45489756.0, + "step": 5158 + }, + { + "epoch": 0.425414364640884, + "grad_norm": 1.8607691526412964, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7556626200675964, + "num_tokens": 45522205.0, + "step": 5159 + }, + { + "epoch": 0.4254968252659355, + "grad_norm": 2.2186944484710693, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7390636205673218, + "num_tokens": 45550407.0, + "step": 5160 + }, + { + "epoch": 0.42557928589098704, + "grad_norm": 1.9552562236785889, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7693682909011841, + "num_tokens": 45579733.0, + "step": 5161 + }, + { + "epoch": 0.4256617465160386, + "grad_norm": 2.0594091415405273, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7457704544067383, + "num_tokens": 45607591.0, + "step": 5162 + }, + { + "epoch": 0.4257442071410901, + "grad_norm": 1.7425130605697632, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7702999114990234, + "num_tokens": 45642388.0, + "step": 5163 + }, + { + "epoch": 0.42582666776614164, + "grad_norm": 2.2357256412506104, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7008762359619141, + "num_tokens": 45668609.0, + "step": 5164 + }, + { + "epoch": 0.4259091283911932, + "grad_norm": 2.1496031284332275, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7164676785469055, + "num_tokens": 45694224.0, + "step": 5165 + }, + { + "epoch": 0.42599158901624473, + "grad_norm": 2.1362199783325195, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.7457095384597778, + "num_tokens": 45720793.0, + "step": 5166 + }, + { + "epoch": 0.4260740496412963, + "grad_norm": 1.786208152770996, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.749201774597168, + "num_tokens": 45753616.0, + "step": 5167 + }, + { + "epoch": 0.4261565102663478, + "grad_norm": 2.0533087253570557, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.75975102186203, + "num_tokens": 45779703.0, + "step": 5168 + }, + { + "epoch": 0.42623897089139934, + "grad_norm": 2.1416351795196533, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.737467885017395, + "num_tokens": 45804921.0, + "step": 5169 + }, + { + "epoch": 0.4263214315164509, + "grad_norm": 1.6810041666030884, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7404648065567017, + "num_tokens": 45845604.0, + "step": 5170 + }, + { + "epoch": 0.4264038921415024, + "grad_norm": 1.8483331203460693, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7768057584762573, + "num_tokens": 45877105.0, + "step": 5171 + }, + { + "epoch": 0.42648635276655394, + "grad_norm": 2.029245138168335, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.711841344833374, + "num_tokens": 45905486.0, + "step": 5172 + }, + { + "epoch": 0.4265688133916055, + "grad_norm": 2.088865041732788, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7421700954437256, + "num_tokens": 45931847.0, + "step": 5173 + }, + { + "epoch": 0.42665127401665703, + "grad_norm": 2.0087685585021973, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7704073786735535, + "num_tokens": 45958429.0, + "step": 5174 + }, + { + "epoch": 0.4267337346417086, + "grad_norm": 2.1086080074310303, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7261021733283997, + "num_tokens": 45986458.0, + "step": 5175 + }, + { + "epoch": 0.4268161952667601, + "grad_norm": 2.2186784744262695, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7176268100738525, + "num_tokens": 46011926.0, + "step": 5176 + }, + { + "epoch": 0.42689865589181164, + "grad_norm": 1.87601900100708, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7586446404457092, + "num_tokens": 46042979.0, + "step": 5177 + }, + { + "epoch": 0.4269811165168632, + "grad_norm": 2.01043963432312, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7325592637062073, + "num_tokens": 46071438.0, + "step": 5178 + }, + { + "epoch": 0.42706357714191473, + "grad_norm": 2.1457276344299316, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7430823445320129, + "num_tokens": 46100716.0, + "step": 5179 + }, + { + "epoch": 0.42714603776696625, + "grad_norm": 1.6809808015823364, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7559193968772888, + "num_tokens": 46141765.0, + "step": 5180 + }, + { + "epoch": 0.4272284983920178, + "grad_norm": 2.118516206741333, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7397193908691406, + "num_tokens": 46168175.0, + "step": 5181 + }, + { + "epoch": 0.42731095901706934, + "grad_norm": 1.93500816822052, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7444538474082947, + "num_tokens": 46198457.0, + "step": 5182 + }, + { + "epoch": 0.4273934196421209, + "grad_norm": 1.9439685344696045, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7459890246391296, + "num_tokens": 46229990.0, + "step": 5183 + }, + { + "epoch": 0.4274758802671724, + "grad_norm": 1.8559763431549072, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7506659626960754, + "num_tokens": 46263771.0, + "step": 5184 + }, + { + "epoch": 0.42755834089222394, + "grad_norm": 2.0021793842315674, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7100566625595093, + "num_tokens": 46291623.0, + "step": 5185 + }, + { + "epoch": 0.4276408015172755, + "grad_norm": 2.077425003051758, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7522116899490356, + "num_tokens": 46317816.0, + "step": 5186 + }, + { + "epoch": 0.42772326214232703, + "grad_norm": 1.7196263074874878, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7564620971679688, + "num_tokens": 46354991.0, + "step": 5187 + }, + { + "epoch": 0.42780572276737855, + "grad_norm": 1.724700927734375, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7269132733345032, + "num_tokens": 46395847.0, + "step": 5188 + }, + { + "epoch": 0.4278881833924301, + "grad_norm": 1.7242032289505005, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7721847295761108, + "num_tokens": 46431306.0, + "step": 5189 + }, + { + "epoch": 0.42797064401748164, + "grad_norm": 2.1479790210723877, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.745017409324646, + "num_tokens": 46460208.0, + "step": 5190 + }, + { + "epoch": 0.4280531046425332, + "grad_norm": 2.1492466926574707, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7320882081985474, + "num_tokens": 46488090.0, + "step": 5191 + }, + { + "epoch": 0.42813556526758473, + "grad_norm": 1.9973888397216797, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7452734708786011, + "num_tokens": 46517209.0, + "step": 5192 + }, + { + "epoch": 0.42821802589263624, + "grad_norm": 2.0057599544525146, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7637339234352112, + "num_tokens": 46545997.0, + "step": 5193 + }, + { + "epoch": 0.4283004865176878, + "grad_norm": 1.8925000429153442, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.7984384298324585, + "num_tokens": 46575102.0, + "step": 5194 + }, + { + "epoch": 0.42838294714273933, + "grad_norm": 2.169827699661255, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7080528736114502, + "num_tokens": 46602366.0, + "step": 5195 + }, + { + "epoch": 0.4284654077677909, + "grad_norm": 2.0241191387176514, + "learning_rate": 1e-06, + "loss": 0.7723, + "mean_token_accuracy": 0.7563056945800781, + "num_tokens": 46628897.0, + "step": 5196 + }, + { + "epoch": 0.4285478683928424, + "grad_norm": 2.071218252182007, + "learning_rate": 1e-06, + "loss": 1.0746, + "mean_token_accuracy": 0.6766285300254822, + "num_tokens": 46659956.0, + "step": 5197 + }, + { + "epoch": 0.42863032901789394, + "grad_norm": 1.9658057689666748, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7522488832473755, + "num_tokens": 46690965.0, + "step": 5198 + }, + { + "epoch": 0.4287127896429455, + "grad_norm": 1.9855698347091675, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7451657056808472, + "num_tokens": 46719459.0, + "step": 5199 + }, + { + "epoch": 0.42879525026799703, + "grad_norm": 1.9411700963974, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7362045049667358, + "num_tokens": 46750710.0, + "step": 5200 + }, + { + "epoch": 0.42887771089304855, + "grad_norm": 2.3636314868927, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7451157569885254, + "num_tokens": 46772098.0, + "step": 5201 + }, + { + "epoch": 0.4289601715181001, + "grad_norm": 1.9251723289489746, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7493492960929871, + "num_tokens": 46802581.0, + "step": 5202 + }, + { + "epoch": 0.42904263214315164, + "grad_norm": 2.0887653827667236, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7357105016708374, + "num_tokens": 46832381.0, + "step": 5203 + }, + { + "epoch": 0.4291250927682032, + "grad_norm": 2.142971992492676, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7639378905296326, + "num_tokens": 46856920.0, + "step": 5204 + }, + { + "epoch": 0.4292075533932547, + "grad_norm": 2.0337073802948, + "learning_rate": 1e-06, + "loss": 1.0536, + "mean_token_accuracy": 0.688947856426239, + "num_tokens": 46886635.0, + "step": 5205 + }, + { + "epoch": 0.42929001401830624, + "grad_norm": 1.8397687673568726, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7594543695449829, + "num_tokens": 46916632.0, + "step": 5206 + }, + { + "epoch": 0.4293724746433578, + "grad_norm": 1.9210140705108643, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7380797863006592, + "num_tokens": 46946940.0, + "step": 5207 + }, + { + "epoch": 0.42945493526840933, + "grad_norm": 1.7509428262710571, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7919662594795227, + "num_tokens": 46982888.0, + "step": 5208 + }, + { + "epoch": 0.42953739589346085, + "grad_norm": 1.8058478832244873, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.782188892364502, + "num_tokens": 47017340.0, + "step": 5209 + }, + { + "epoch": 0.4296198565185124, + "grad_norm": 1.855716347694397, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7406027317047119, + "num_tokens": 47050786.0, + "step": 5210 + }, + { + "epoch": 0.42970231714356394, + "grad_norm": 2.013205051422119, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7350976467132568, + "num_tokens": 47080402.0, + "step": 5211 + }, + { + "epoch": 0.4297847777686155, + "grad_norm": 2.2922208309173584, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7546249032020569, + "num_tokens": 47102701.0, + "step": 5212 + }, + { + "epoch": 0.42986723839366703, + "grad_norm": 1.969589352607727, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7666870355606079, + "num_tokens": 47130137.0, + "step": 5213 + }, + { + "epoch": 0.42994969901871855, + "grad_norm": 1.7010548114776611, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7614116072654724, + "num_tokens": 47166403.0, + "step": 5214 + }, + { + "epoch": 0.4300321596437701, + "grad_norm": 1.9679821729660034, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7531700730323792, + "num_tokens": 47198405.0, + "step": 5215 + }, + { + "epoch": 0.43011462026882163, + "grad_norm": 2.0836050510406494, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6958316564559937, + "num_tokens": 47228667.0, + "step": 5216 + }, + { + "epoch": 0.43019708089387315, + "grad_norm": 2.133244276046753, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.76472008228302, + "num_tokens": 47254493.0, + "step": 5217 + }, + { + "epoch": 0.4302795415189247, + "grad_norm": 1.9070736169815063, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7335067987442017, + "num_tokens": 47288612.0, + "step": 5218 + }, + { + "epoch": 0.43036200214397624, + "grad_norm": 2.144876480102539, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.7736223936080933, + "num_tokens": 47314684.0, + "step": 5219 + }, + { + "epoch": 0.4304444627690278, + "grad_norm": 2.267434597015381, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7512036561965942, + "num_tokens": 47338752.0, + "step": 5220 + }, + { + "epoch": 0.43052692339407933, + "grad_norm": 2.185819387435913, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7386310696601868, + "num_tokens": 47361974.0, + "step": 5221 + }, + { + "epoch": 0.43060938401913085, + "grad_norm": 1.8469873666763306, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.781354546546936, + "num_tokens": 47391782.0, + "step": 5222 + }, + { + "epoch": 0.4306918446441824, + "grad_norm": 1.8415324687957764, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7351125478744507, + "num_tokens": 47425305.0, + "step": 5223 + }, + { + "epoch": 0.43077430526923394, + "grad_norm": 2.0748250484466553, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7350814342498779, + "num_tokens": 47451534.0, + "step": 5224 + }, + { + "epoch": 0.43085676589428545, + "grad_norm": 1.8041059970855713, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7204250693321228, + "num_tokens": 47484605.0, + "step": 5225 + }, + { + "epoch": 0.430939226519337, + "grad_norm": 1.9934064149856567, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7574950456619263, + "num_tokens": 47513514.0, + "step": 5226 + }, + { + "epoch": 0.43102168714438854, + "grad_norm": 1.7781702280044556, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7382332682609558, + "num_tokens": 47545731.0, + "step": 5227 + }, + { + "epoch": 0.4311041477694401, + "grad_norm": 1.7197952270507812, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7653941512107849, + "num_tokens": 47583329.0, + "step": 5228 + }, + { + "epoch": 0.43118660839449163, + "grad_norm": 1.522331953048706, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.7948647737503052, + "num_tokens": 47626139.0, + "step": 5229 + }, + { + "epoch": 0.43126906901954315, + "grad_norm": 2.1187076568603516, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.71142578125, + "num_tokens": 47656156.0, + "step": 5230 + }, + { + "epoch": 0.4313515296445947, + "grad_norm": 2.148130416870117, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.734581470489502, + "num_tokens": 47681298.0, + "step": 5231 + }, + { + "epoch": 0.43143399026964624, + "grad_norm": 2.1708316802978516, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7149816751480103, + "num_tokens": 47709614.0, + "step": 5232 + }, + { + "epoch": 0.43151645089469776, + "grad_norm": 2.0072314739227295, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7249294519424438, + "num_tokens": 47738437.0, + "step": 5233 + }, + { + "epoch": 0.43159891151974933, + "grad_norm": 2.1608784198760986, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7286990880966187, + "num_tokens": 47765756.0, + "step": 5234 + }, + { + "epoch": 0.43168137214480085, + "grad_norm": 1.8813788890838623, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7722289562225342, + "num_tokens": 47796642.0, + "step": 5235 + }, + { + "epoch": 0.4317638327698524, + "grad_norm": 1.9411500692367554, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7402280569076538, + "num_tokens": 47828883.0, + "step": 5236 + }, + { + "epoch": 0.43184629339490394, + "grad_norm": 1.9190579652786255, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7222490310668945, + "num_tokens": 47861482.0, + "step": 5237 + }, + { + "epoch": 0.43192875401995545, + "grad_norm": 1.8735767602920532, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7683882713317871, + "num_tokens": 47891828.0, + "step": 5238 + }, + { + "epoch": 0.432011214645007, + "grad_norm": 2.149568557739258, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.738002359867096, + "num_tokens": 47917309.0, + "step": 5239 + }, + { + "epoch": 0.43209367527005854, + "grad_norm": 1.9760996103286743, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7359827756881714, + "num_tokens": 47950362.0, + "step": 5240 + }, + { + "epoch": 0.43217613589511006, + "grad_norm": 2.2749476432800293, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7468309998512268, + "num_tokens": 47975091.0, + "step": 5241 + }, + { + "epoch": 0.43225859652016163, + "grad_norm": 2.0054855346679688, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7325761914253235, + "num_tokens": 48006783.0, + "step": 5242 + }, + { + "epoch": 0.43234105714521315, + "grad_norm": 1.9655009508132935, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7691669464111328, + "num_tokens": 48038744.0, + "step": 5243 + }, + { + "epoch": 0.4324235177702647, + "grad_norm": 2.3461806774139404, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7386735677719116, + "num_tokens": 48061085.0, + "step": 5244 + }, + { + "epoch": 0.43250597839531624, + "grad_norm": 1.8758102655410767, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7683637738227844, + "num_tokens": 48094975.0, + "step": 5245 + }, + { + "epoch": 0.43258843902036775, + "grad_norm": 2.220418930053711, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7499906420707703, + "num_tokens": 48120621.0, + "step": 5246 + }, + { + "epoch": 0.4326708996454193, + "grad_norm": 2.3319509029388428, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7416884303092957, + "num_tokens": 48144616.0, + "step": 5247 + }, + { + "epoch": 0.43275336027047084, + "grad_norm": 2.063166856765747, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7605398893356323, + "num_tokens": 48171129.0, + "step": 5248 + }, + { + "epoch": 0.43283582089552236, + "grad_norm": 2.018273115158081, + "learning_rate": 1e-06, + "loss": 1.0568, + "mean_token_accuracy": 0.6894298791885376, + "num_tokens": 48201239.0, + "step": 5249 + }, + { + "epoch": 0.43291828152057393, + "grad_norm": 1.7986431121826172, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7385442852973938, + "num_tokens": 48236042.0, + "step": 5250 + }, + { + "epoch": 0.43300074214562545, + "grad_norm": 1.890419363975525, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.754967451095581, + "num_tokens": 48264661.0, + "step": 5251 + }, + { + "epoch": 0.433083202770677, + "grad_norm": 2.2323455810546875, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7616842985153198, + "num_tokens": 48287075.0, + "step": 5252 + }, + { + "epoch": 0.43316566339572854, + "grad_norm": 2.5101850032806396, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7229515314102173, + "num_tokens": 48306720.0, + "step": 5253 + }, + { + "epoch": 0.43324812402078006, + "grad_norm": 2.219623565673828, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7221652865409851, + "num_tokens": 48331541.0, + "step": 5254 + }, + { + "epoch": 0.43333058464583163, + "grad_norm": 2.080806255340576, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7292604446411133, + "num_tokens": 48359412.0, + "step": 5255 + }, + { + "epoch": 0.43341304527088315, + "grad_norm": 1.968802809715271, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7777429819107056, + "num_tokens": 48387339.0, + "step": 5256 + }, + { + "epoch": 0.4334955058959347, + "grad_norm": 2.151268243789673, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7304670810699463, + "num_tokens": 48414229.0, + "step": 5257 + }, + { + "epoch": 0.43357796652098624, + "grad_norm": 2.010136127471924, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7587468028068542, + "num_tokens": 48442983.0, + "step": 5258 + }, + { + "epoch": 0.43366042714603775, + "grad_norm": 1.8121376037597656, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.7788292169570923, + "num_tokens": 48474158.0, + "step": 5259 + }, + { + "epoch": 0.4337428877710893, + "grad_norm": 1.8879138231277466, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.769298255443573, + "num_tokens": 48504529.0, + "step": 5260 + }, + { + "epoch": 0.43382534839614084, + "grad_norm": 1.8747071027755737, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7511639595031738, + "num_tokens": 48537071.0, + "step": 5261 + }, + { + "epoch": 0.43390780902119236, + "grad_norm": 2.2141709327697754, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7183882594108582, + "num_tokens": 48561753.0, + "step": 5262 + }, + { + "epoch": 0.43399026964624393, + "grad_norm": 1.7740592956542969, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7340131998062134, + "num_tokens": 48595590.0, + "step": 5263 + }, + { + "epoch": 0.43407273027129545, + "grad_norm": 1.6547874212265015, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7579532265663147, + "num_tokens": 48634667.0, + "step": 5264 + }, + { + "epoch": 0.434155190896347, + "grad_norm": 2.0817418098449707, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7604790329933167, + "num_tokens": 48659730.0, + "step": 5265 + }, + { + "epoch": 0.43423765152139854, + "grad_norm": 2.0518696308135986, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.749005913734436, + "num_tokens": 48689002.0, + "step": 5266 + }, + { + "epoch": 0.43432011214645005, + "grad_norm": 2.2642722129821777, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.707987904548645, + "num_tokens": 48716730.0, + "step": 5267 + }, + { + "epoch": 0.4344025727715016, + "grad_norm": 1.8261505365371704, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7647707462310791, + "num_tokens": 48749820.0, + "step": 5268 + }, + { + "epoch": 0.43448503339655314, + "grad_norm": 2.1927273273468018, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.7022002935409546, + "num_tokens": 48777550.0, + "step": 5269 + }, + { + "epoch": 0.43456749402160466, + "grad_norm": 2.2190704345703125, + "learning_rate": 1e-06, + "loss": 1.0465, + "mean_token_accuracy": 0.6910527944564819, + "num_tokens": 48804863.0, + "step": 5270 + }, + { + "epoch": 0.43464995464665623, + "grad_norm": 2.1354081630706787, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.7834485769271851, + "num_tokens": 48829785.0, + "step": 5271 + }, + { + "epoch": 0.43473241527170775, + "grad_norm": 1.8131028413772583, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7455413937568665, + "num_tokens": 48862608.0, + "step": 5272 + }, + { + "epoch": 0.4348148758967593, + "grad_norm": 1.9595606327056885, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7568830847740173, + "num_tokens": 48892335.0, + "step": 5273 + }, + { + "epoch": 0.43489733652181084, + "grad_norm": 2.0670385360717773, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7178248763084412, + "num_tokens": 48921006.0, + "step": 5274 + }, + { + "epoch": 0.43497979714686236, + "grad_norm": 1.8505685329437256, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7210409641265869, + "num_tokens": 48957191.0, + "step": 5275 + }, + { + "epoch": 0.43506225777191393, + "grad_norm": 2.1171774864196777, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7729434370994568, + "num_tokens": 48982501.0, + "step": 5276 + }, + { + "epoch": 0.43514471839696545, + "grad_norm": 2.028994560241699, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7636009454727173, + "num_tokens": 49008464.0, + "step": 5277 + }, + { + "epoch": 0.43522717902201696, + "grad_norm": 2.098121166229248, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.715502142906189, + "num_tokens": 49035322.0, + "step": 5278 + }, + { + "epoch": 0.43530963964706854, + "grad_norm": 2.074209451675415, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7352250218391418, + "num_tokens": 49063650.0, + "step": 5279 + }, + { + "epoch": 0.43539210027212005, + "grad_norm": 2.003711700439453, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7171967625617981, + "num_tokens": 49097138.0, + "step": 5280 + }, + { + "epoch": 0.4354745608971716, + "grad_norm": 2.0529661178588867, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7353031039237976, + "num_tokens": 49123995.0, + "step": 5281 + }, + { + "epoch": 0.43555702152222314, + "grad_norm": 1.7309412956237793, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7824743986129761, + "num_tokens": 49160268.0, + "step": 5282 + }, + { + "epoch": 0.43563948214727466, + "grad_norm": 1.821319341659546, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7767132520675659, + "num_tokens": 49191955.0, + "step": 5283 + }, + { + "epoch": 0.43572194277232623, + "grad_norm": 2.000302791595459, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7433735132217407, + "num_tokens": 49221530.0, + "step": 5284 + }, + { + "epoch": 0.43580440339737775, + "grad_norm": 2.4227652549743652, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7162008285522461, + "num_tokens": 49245674.0, + "step": 5285 + }, + { + "epoch": 0.43588686402242927, + "grad_norm": 2.1540329456329346, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7350777983665466, + "num_tokens": 49273595.0, + "step": 5286 + }, + { + "epoch": 0.43596932464748084, + "grad_norm": 2.108299493789673, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7152732610702515, + "num_tokens": 49301367.0, + "step": 5287 + }, + { + "epoch": 0.43605178527253236, + "grad_norm": 1.997848629951477, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7112312912940979, + "num_tokens": 49332719.0, + "step": 5288 + }, + { + "epoch": 0.4361342458975839, + "grad_norm": 1.9420536756515503, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.739534854888916, + "num_tokens": 49362882.0, + "step": 5289 + }, + { + "epoch": 0.43621670652263544, + "grad_norm": 2.2677364349365234, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7250022292137146, + "num_tokens": 49388134.0, + "step": 5290 + }, + { + "epoch": 0.43629916714768696, + "grad_norm": 1.8759537935256958, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7382490634918213, + "num_tokens": 49420588.0, + "step": 5291 + }, + { + "epoch": 0.43638162777273853, + "grad_norm": 2.162659168243408, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7421096563339233, + "num_tokens": 49446389.0, + "step": 5292 + }, + { + "epoch": 0.43646408839779005, + "grad_norm": 1.8961422443389893, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7715438604354858, + "num_tokens": 49478350.0, + "step": 5293 + }, + { + "epoch": 0.43654654902284157, + "grad_norm": 1.6785393953323364, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7498441338539124, + "num_tokens": 49513986.0, + "step": 5294 + }, + { + "epoch": 0.43662900964789314, + "grad_norm": 1.7602691650390625, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7434672117233276, + "num_tokens": 49549639.0, + "step": 5295 + }, + { + "epoch": 0.43671147027294466, + "grad_norm": 1.9289237260818481, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.7525238990783691, + "num_tokens": 49578556.0, + "step": 5296 + }, + { + "epoch": 0.43679393089799623, + "grad_norm": 1.7532665729522705, + "learning_rate": 1e-06, + "loss": 0.5821, + "mean_token_accuracy": 0.78692227602005, + "num_tokens": 49611641.0, + "step": 5297 + }, + { + "epoch": 0.43687639152304775, + "grad_norm": 2.2390096187591553, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7322278022766113, + "num_tokens": 49634785.0, + "step": 5298 + }, + { + "epoch": 0.43695885214809926, + "grad_norm": 1.9080520868301392, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7485533952713013, + "num_tokens": 49665643.0, + "step": 5299 + }, + { + "epoch": 0.43704131277315084, + "grad_norm": 2.0776591300964355, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.729156494140625, + "num_tokens": 49694916.0, + "step": 5300 + }, + { + "epoch": 0.43712377339820235, + "grad_norm": 1.865470051765442, + "learning_rate": 1e-06, + "loss": 0.6189, + "mean_token_accuracy": 0.7992156147956848, + "num_tokens": 49725993.0, + "step": 5301 + }, + { + "epoch": 0.43720623402325387, + "grad_norm": 1.9098221063613892, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7567477226257324, + "num_tokens": 49755350.0, + "step": 5302 + }, + { + "epoch": 0.43728869464830544, + "grad_norm": 1.7446871995925903, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7535077333450317, + "num_tokens": 49790681.0, + "step": 5303 + }, + { + "epoch": 0.43737115527335696, + "grad_norm": 1.6793066263198853, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.774605393409729, + "num_tokens": 49828719.0, + "step": 5304 + }, + { + "epoch": 0.43745361589840853, + "grad_norm": 1.8490877151489258, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7424752712249756, + "num_tokens": 49859586.0, + "step": 5305 + }, + { + "epoch": 0.43753607652346005, + "grad_norm": 2.156614303588867, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7667925357818604, + "num_tokens": 49886084.0, + "step": 5306 + }, + { + "epoch": 0.43761853714851157, + "grad_norm": 1.919341802597046, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7787477970123291, + "num_tokens": 49919658.0, + "step": 5307 + }, + { + "epoch": 0.43770099777356314, + "grad_norm": 2.1130757331848145, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7185623049736023, + "num_tokens": 49949232.0, + "step": 5308 + }, + { + "epoch": 0.43778345839861466, + "grad_norm": 2.108393907546997, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7430319786071777, + "num_tokens": 49977407.0, + "step": 5309 + }, + { + "epoch": 0.4378659190236662, + "grad_norm": 2.3454480171203613, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7282154560089111, + "num_tokens": 50002194.0, + "step": 5310 + }, + { + "epoch": 0.43794837964871774, + "grad_norm": 2.041707992553711, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7329010963439941, + "num_tokens": 50029969.0, + "step": 5311 + }, + { + "epoch": 0.43803084027376926, + "grad_norm": 2.1719768047332764, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7068794965744019, + "num_tokens": 50057188.0, + "step": 5312 + }, + { + "epoch": 0.43811330089882083, + "grad_norm": 2.074599266052246, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7408547401428223, + "num_tokens": 50087405.0, + "step": 5313 + }, + { + "epoch": 0.43819576152387235, + "grad_norm": 2.285484552383423, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7361218333244324, + "num_tokens": 50108023.0, + "step": 5314 + }, + { + "epoch": 0.43827822214892387, + "grad_norm": 2.1445446014404297, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7481199502944946, + "num_tokens": 50135951.0, + "step": 5315 + }, + { + "epoch": 0.43836068277397544, + "grad_norm": 1.9155917167663574, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.739448606967926, + "num_tokens": 50167922.0, + "step": 5316 + }, + { + "epoch": 0.43844314339902696, + "grad_norm": 1.9186643362045288, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.740542471408844, + "num_tokens": 50197153.0, + "step": 5317 + }, + { + "epoch": 0.43852560402407853, + "grad_norm": 1.8652163743972778, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7576683759689331, + "num_tokens": 50228941.0, + "step": 5318 + }, + { + "epoch": 0.43860806464913005, + "grad_norm": 2.0047197341918945, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7475192546844482, + "num_tokens": 50256317.0, + "step": 5319 + }, + { + "epoch": 0.43869052527418156, + "grad_norm": 1.7318212985992432, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7805685997009277, + "num_tokens": 50289078.0, + "step": 5320 + }, + { + "epoch": 0.43877298589923314, + "grad_norm": 1.8418278694152832, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7059794068336487, + "num_tokens": 50322793.0, + "step": 5321 + }, + { + "epoch": 0.43885544652428465, + "grad_norm": 1.875181794166565, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.71907639503479, + "num_tokens": 50358360.0, + "step": 5322 + }, + { + "epoch": 0.43893790714933617, + "grad_norm": 2.4863827228546143, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7555066347122192, + "num_tokens": 50378571.0, + "step": 5323 + }, + { + "epoch": 0.43902036777438774, + "grad_norm": 2.021669864654541, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7363654375076294, + "num_tokens": 50406354.0, + "step": 5324 + }, + { + "epoch": 0.43910282839943926, + "grad_norm": 2.0352656841278076, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7225514650344849, + "num_tokens": 50438693.0, + "step": 5325 + }, + { + "epoch": 0.43918528902449083, + "grad_norm": 2.070664644241333, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7604913115501404, + "num_tokens": 50463212.0, + "step": 5326 + }, + { + "epoch": 0.43926774964954235, + "grad_norm": 2.2137601375579834, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7609169483184814, + "num_tokens": 50491154.0, + "step": 5327 + }, + { + "epoch": 0.43935021027459387, + "grad_norm": 1.7911955118179321, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7142941355705261, + "num_tokens": 50528950.0, + "step": 5328 + }, + { + "epoch": 0.43943267089964544, + "grad_norm": 1.8990428447723389, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.71607905626297, + "num_tokens": 50560761.0, + "step": 5329 + }, + { + "epoch": 0.43951513152469696, + "grad_norm": 2.028965473175049, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7780855298042297, + "num_tokens": 50586733.0, + "step": 5330 + }, + { + "epoch": 0.4395975921497485, + "grad_norm": 1.8395811319351196, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7656732201576233, + "num_tokens": 50621214.0, + "step": 5331 + }, + { + "epoch": 0.43968005277480005, + "grad_norm": 1.8114537000656128, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7754548192024231, + "num_tokens": 50654474.0, + "step": 5332 + }, + { + "epoch": 0.43976251339985156, + "grad_norm": 1.796183705329895, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7671941518783569, + "num_tokens": 50688936.0, + "step": 5333 + }, + { + "epoch": 0.43984497402490313, + "grad_norm": 2.0896964073181152, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7607895731925964, + "num_tokens": 50716077.0, + "step": 5334 + }, + { + "epoch": 0.43992743464995465, + "grad_norm": 1.9146648645401, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7560491561889648, + "num_tokens": 50747291.0, + "step": 5335 + }, + { + "epoch": 0.44000989527500617, + "grad_norm": 1.8987047672271729, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7185196876525879, + "num_tokens": 50780818.0, + "step": 5336 + }, + { + "epoch": 0.44009235590005774, + "grad_norm": 1.8537499904632568, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.747704029083252, + "num_tokens": 50814342.0, + "step": 5337 + }, + { + "epoch": 0.44017481652510926, + "grad_norm": 1.890824556350708, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7595897912979126, + "num_tokens": 50845925.0, + "step": 5338 + }, + { + "epoch": 0.4402572771501608, + "grad_norm": 1.5885926485061646, + "learning_rate": 1e-06, + "loss": 0.5575, + "mean_token_accuracy": 0.8167436122894287, + "num_tokens": 50884046.0, + "step": 5339 + }, + { + "epoch": 0.44033973777521235, + "grad_norm": 2.0342795848846436, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7510438561439514, + "num_tokens": 50911719.0, + "step": 5340 + }, + { + "epoch": 0.44042219840026386, + "grad_norm": 1.7888710498809814, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7629333138465881, + "num_tokens": 50946128.0, + "step": 5341 + }, + { + "epoch": 0.44050465902531544, + "grad_norm": 2.097663164138794, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7343113422393799, + "num_tokens": 50973741.0, + "step": 5342 + }, + { + "epoch": 0.44058711965036695, + "grad_norm": 2.0448360443115234, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7851612567901611, + "num_tokens": 50999402.0, + "step": 5343 + }, + { + "epoch": 0.44066958027541847, + "grad_norm": 1.9427818059921265, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7563711404800415, + "num_tokens": 51030336.0, + "step": 5344 + }, + { + "epoch": 0.44075204090047004, + "grad_norm": 1.8405272960662842, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7486778497695923, + "num_tokens": 51062241.0, + "step": 5345 + }, + { + "epoch": 0.44083450152552156, + "grad_norm": 2.228921890258789, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7334819436073303, + "num_tokens": 51085049.0, + "step": 5346 + }, + { + "epoch": 0.4409169621505731, + "grad_norm": 1.8378039598464966, + "learning_rate": 1e-06, + "loss": 0.771, + "mean_token_accuracy": 0.7596371173858643, + "num_tokens": 51117079.0, + "step": 5347 + }, + { + "epoch": 0.44099942277562465, + "grad_norm": 2.282318115234375, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7281748652458191, + "num_tokens": 51139751.0, + "step": 5348 + }, + { + "epoch": 0.44108188340067617, + "grad_norm": 2.0732061862945557, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7524670362472534, + "num_tokens": 51164157.0, + "step": 5349 + }, + { + "epoch": 0.44116434402572774, + "grad_norm": 2.1324591636657715, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7382310628890991, + "num_tokens": 51190125.0, + "step": 5350 + }, + { + "epoch": 0.44124680465077926, + "grad_norm": 2.0062310695648193, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7433104515075684, + "num_tokens": 51218533.0, + "step": 5351 + }, + { + "epoch": 0.4413292652758308, + "grad_norm": 2.0503740310668945, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7499026656150818, + "num_tokens": 51246991.0, + "step": 5352 + }, + { + "epoch": 0.44141172590088235, + "grad_norm": 2.0268588066101074, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.781399130821228, + "num_tokens": 51272555.0, + "step": 5353 + }, + { + "epoch": 0.44149418652593386, + "grad_norm": 1.638717770576477, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7796478271484375, + "num_tokens": 51310502.0, + "step": 5354 + }, + { + "epoch": 0.4415766471509854, + "grad_norm": 2.064887285232544, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.726818323135376, + "num_tokens": 51339318.0, + "step": 5355 + }, + { + "epoch": 0.44165910777603695, + "grad_norm": 2.177828550338745, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7514661550521851, + "num_tokens": 51366153.0, + "step": 5356 + }, + { + "epoch": 0.44174156840108847, + "grad_norm": 2.348620653152466, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7395942211151123, + "num_tokens": 51389549.0, + "step": 5357 + }, + { + "epoch": 0.44182402902614004, + "grad_norm": 1.9720693826675415, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7423728704452515, + "num_tokens": 51419695.0, + "step": 5358 + }, + { + "epoch": 0.44190648965119156, + "grad_norm": 1.9176579713821411, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7342891693115234, + "num_tokens": 51450118.0, + "step": 5359 + }, + { + "epoch": 0.4419889502762431, + "grad_norm": 2.002842664718628, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.741852879524231, + "num_tokens": 51477999.0, + "step": 5360 + }, + { + "epoch": 0.44207141090129465, + "grad_norm": 2.0211710929870605, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7040029764175415, + "num_tokens": 51508782.0, + "step": 5361 + }, + { + "epoch": 0.44215387152634616, + "grad_norm": 2.1208913326263428, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7486250400543213, + "num_tokens": 51534742.0, + "step": 5362 + }, + { + "epoch": 0.4422363321513977, + "grad_norm": 1.8896068334579468, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7530476450920105, + "num_tokens": 51567198.0, + "step": 5363 + }, + { + "epoch": 0.44231879277644925, + "grad_norm": 2.090611696243286, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7691665887832642, + "num_tokens": 51593223.0, + "step": 5364 + }, + { + "epoch": 0.44240125340150077, + "grad_norm": 2.173295021057129, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7526464462280273, + "num_tokens": 51616423.0, + "step": 5365 + }, + { + "epoch": 0.44248371402655234, + "grad_norm": 1.7874481678009033, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7483320236206055, + "num_tokens": 51648007.0, + "step": 5366 + }, + { + "epoch": 0.44256617465160386, + "grad_norm": 1.8337057828903198, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7317774295806885, + "num_tokens": 51684508.0, + "step": 5367 + }, + { + "epoch": 0.4426486352766554, + "grad_norm": 1.850691318511963, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.7623387575149536, + "num_tokens": 51716924.0, + "step": 5368 + }, + { + "epoch": 0.44273109590170695, + "grad_norm": 2.026324987411499, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.7867885231971741, + "num_tokens": 51743813.0, + "step": 5369 + }, + { + "epoch": 0.44281355652675847, + "grad_norm": 1.9177082777023315, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7717272043228149, + "num_tokens": 51775754.0, + "step": 5370 + }, + { + "epoch": 0.44289601715181, + "grad_norm": 1.7971336841583252, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7124567031860352, + "num_tokens": 51814092.0, + "step": 5371 + }, + { + "epoch": 0.44297847777686156, + "grad_norm": 1.9339332580566406, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7221372127532959, + "num_tokens": 51847218.0, + "step": 5372 + }, + { + "epoch": 0.4430609384019131, + "grad_norm": 2.127877712249756, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.744736909866333, + "num_tokens": 51873923.0, + "step": 5373 + }, + { + "epoch": 0.44314339902696465, + "grad_norm": 1.9583327770233154, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7649473547935486, + "num_tokens": 51902000.0, + "step": 5374 + }, + { + "epoch": 0.44322585965201616, + "grad_norm": 2.2131781578063965, + "learning_rate": 1e-06, + "loss": 0.7724, + "mean_token_accuracy": 0.7675882577896118, + "num_tokens": 51926623.0, + "step": 5375 + }, + { + "epoch": 0.4433083202770677, + "grad_norm": 1.9595372676849365, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7828444838523865, + "num_tokens": 51955261.0, + "step": 5376 + }, + { + "epoch": 0.44339078090211925, + "grad_norm": 1.9199185371398926, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7365291118621826, + "num_tokens": 51987300.0, + "step": 5377 + }, + { + "epoch": 0.44347324152717077, + "grad_norm": 1.7930335998535156, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7925538420677185, + "num_tokens": 52019025.0, + "step": 5378 + }, + { + "epoch": 0.4435557021522223, + "grad_norm": 1.7302131652832031, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.745598316192627, + "num_tokens": 52056435.0, + "step": 5379 + }, + { + "epoch": 0.44363816277727386, + "grad_norm": 1.9024478197097778, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7124702334403992, + "num_tokens": 52091425.0, + "step": 5380 + }, + { + "epoch": 0.4437206234023254, + "grad_norm": 1.9607133865356445, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7475743293762207, + "num_tokens": 52121331.0, + "step": 5381 + }, + { + "epoch": 0.44380308402737695, + "grad_norm": 1.9670915603637695, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7705355286598206, + "num_tokens": 52150911.0, + "step": 5382 + }, + { + "epoch": 0.44388554465242847, + "grad_norm": 1.8403569459915161, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7113521099090576, + "num_tokens": 52185306.0, + "step": 5383 + }, + { + "epoch": 0.44396800527748, + "grad_norm": 2.145655632019043, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7697435021400452, + "num_tokens": 52210750.0, + "step": 5384 + }, + { + "epoch": 0.44405046590253155, + "grad_norm": 2.0890142917633057, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7751466035842896, + "num_tokens": 52235197.0, + "step": 5385 + }, + { + "epoch": 0.44413292652758307, + "grad_norm": 2.0068976879119873, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7494997382164001, + "num_tokens": 52262536.0, + "step": 5386 + }, + { + "epoch": 0.44421538715263464, + "grad_norm": 1.9126814603805542, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7232266664505005, + "num_tokens": 52298539.0, + "step": 5387 + }, + { + "epoch": 0.44429784777768616, + "grad_norm": 2.1895463466644287, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7211591005325317, + "num_tokens": 52324484.0, + "step": 5388 + }, + { + "epoch": 0.4443803084027377, + "grad_norm": 1.970285415649414, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.775772213935852, + "num_tokens": 52351326.0, + "step": 5389 + }, + { + "epoch": 0.44446276902778925, + "grad_norm": 1.851019263267517, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7484514713287354, + "num_tokens": 52385053.0, + "step": 5390 + }, + { + "epoch": 0.44454522965284077, + "grad_norm": 1.9663974046707153, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7533676624298096, + "num_tokens": 52418278.0, + "step": 5391 + }, + { + "epoch": 0.4446276902778923, + "grad_norm": 1.79032564163208, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.724048376083374, + "num_tokens": 52456640.0, + "step": 5392 + }, + { + "epoch": 0.44471015090294386, + "grad_norm": 2.082643985748291, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.731631338596344, + "num_tokens": 52482851.0, + "step": 5393 + }, + { + "epoch": 0.4447926115279954, + "grad_norm": 1.8842579126358032, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7611925601959229, + "num_tokens": 52514540.0, + "step": 5394 + }, + { + "epoch": 0.44487507215304695, + "grad_norm": 2.055171251296997, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7487733364105225, + "num_tokens": 52542171.0, + "step": 5395 + }, + { + "epoch": 0.44495753277809846, + "grad_norm": 1.866546630859375, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7462817430496216, + "num_tokens": 52577378.0, + "step": 5396 + }, + { + "epoch": 0.44503999340315, + "grad_norm": 2.0792558193206787, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7498554587364197, + "num_tokens": 52602339.0, + "step": 5397 + }, + { + "epoch": 0.44512245402820155, + "grad_norm": 1.9360437393188477, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7336591482162476, + "num_tokens": 52634665.0, + "step": 5398 + }, + { + "epoch": 0.44520491465325307, + "grad_norm": 2.0486533641815186, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7407840490341187, + "num_tokens": 52664270.0, + "step": 5399 + }, + { + "epoch": 0.4452873752783046, + "grad_norm": 1.8841553926467896, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.7597579956054688, + "num_tokens": 52695437.0, + "step": 5400 + }, + { + "epoch": 0.44536983590335616, + "grad_norm": 2.0372602939605713, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7655761241912842, + "num_tokens": 52720848.0, + "step": 5401 + }, + { + "epoch": 0.4454522965284077, + "grad_norm": 2.161041021347046, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7077512741088867, + "num_tokens": 52747525.0, + "step": 5402 + }, + { + "epoch": 0.44553475715345925, + "grad_norm": 2.2091355323791504, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7220190763473511, + "num_tokens": 52772358.0, + "step": 5403 + }, + { + "epoch": 0.44561721777851077, + "grad_norm": 1.7771300077438354, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7248697280883789, + "num_tokens": 52808014.0, + "step": 5404 + }, + { + "epoch": 0.4456996784035623, + "grad_norm": 1.9199613332748413, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7511466145515442, + "num_tokens": 52838270.0, + "step": 5405 + }, + { + "epoch": 0.44578213902861386, + "grad_norm": 2.0276126861572266, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7583718299865723, + "num_tokens": 52868175.0, + "step": 5406 + }, + { + "epoch": 0.44586459965366537, + "grad_norm": 1.8811626434326172, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7869055271148682, + "num_tokens": 52900632.0, + "step": 5407 + }, + { + "epoch": 0.4459470602787169, + "grad_norm": 1.7602587938308716, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7343401908874512, + "num_tokens": 52936235.0, + "step": 5408 + }, + { + "epoch": 0.44602952090376846, + "grad_norm": 1.9835408926010132, + "learning_rate": 1e-06, + "loss": 0.734, + "mean_token_accuracy": 0.7668604254722595, + "num_tokens": 52967098.0, + "step": 5409 + }, + { + "epoch": 0.44611198152882, + "grad_norm": 2.054814577102661, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7756636738777161, + "num_tokens": 52994560.0, + "step": 5410 + }, + { + "epoch": 0.44619444215387155, + "grad_norm": 2.0814011096954346, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7684452533721924, + "num_tokens": 53019711.0, + "step": 5411 + }, + { + "epoch": 0.44627690277892307, + "grad_norm": 1.9750392436981201, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7484602928161621, + "num_tokens": 53049328.0, + "step": 5412 + }, + { + "epoch": 0.4463593634039746, + "grad_norm": 1.9800891876220703, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7133470177650452, + "num_tokens": 53077759.0, + "step": 5413 + }, + { + "epoch": 0.44644182402902616, + "grad_norm": 1.888985276222229, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7530971765518188, + "num_tokens": 53108095.0, + "step": 5414 + }, + { + "epoch": 0.4465242846540777, + "grad_norm": 1.8960773944854736, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7824760675430298, + "num_tokens": 53139181.0, + "step": 5415 + }, + { + "epoch": 0.4466067452791292, + "grad_norm": 1.9493708610534668, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.757229745388031, + "num_tokens": 53169182.0, + "step": 5416 + }, + { + "epoch": 0.44668920590418076, + "grad_norm": 2.2802059650421143, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7383632659912109, + "num_tokens": 53194853.0, + "step": 5417 + }, + { + "epoch": 0.4467716665292323, + "grad_norm": 2.0580663681030273, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7147238850593567, + "num_tokens": 53223111.0, + "step": 5418 + }, + { + "epoch": 0.44685412715428385, + "grad_norm": 1.923762321472168, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7511732578277588, + "num_tokens": 53253875.0, + "step": 5419 + }, + { + "epoch": 0.44693658777933537, + "grad_norm": 1.917873740196228, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7808630466461182, + "num_tokens": 53285189.0, + "step": 5420 + }, + { + "epoch": 0.4470190484043869, + "grad_norm": 2.20885968208313, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7444554567337036, + "num_tokens": 53311201.0, + "step": 5421 + }, + { + "epoch": 0.44710150902943846, + "grad_norm": 1.9192196130752563, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7606123089790344, + "num_tokens": 53342892.0, + "step": 5422 + }, + { + "epoch": 0.44718396965449, + "grad_norm": 2.2033700942993164, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7449394464492798, + "num_tokens": 53366107.0, + "step": 5423 + }, + { + "epoch": 0.4472664302795415, + "grad_norm": 1.6619166135787964, + "learning_rate": 1e-06, + "loss": 0.6068, + "mean_token_accuracy": 0.8007679581642151, + "num_tokens": 53403555.0, + "step": 5424 + }, + { + "epoch": 0.44734889090459307, + "grad_norm": 2.0093424320220947, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7296450138092041, + "num_tokens": 53436400.0, + "step": 5425 + }, + { + "epoch": 0.4474313515296446, + "grad_norm": 2.3788645267486572, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7360981702804565, + "num_tokens": 53457928.0, + "step": 5426 + }, + { + "epoch": 0.44751381215469616, + "grad_norm": 1.8817943334579468, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7696146965026855, + "num_tokens": 53487867.0, + "step": 5427 + }, + { + "epoch": 0.4475962727797477, + "grad_norm": 1.9777193069458008, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7259693145751953, + "num_tokens": 53518655.0, + "step": 5428 + }, + { + "epoch": 0.4476787334047992, + "grad_norm": 2.351644992828369, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7230516672134399, + "num_tokens": 53541180.0, + "step": 5429 + }, + { + "epoch": 0.44776119402985076, + "grad_norm": 1.9347124099731445, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7543151378631592, + "num_tokens": 53570921.0, + "step": 5430 + }, + { + "epoch": 0.4478436546549023, + "grad_norm": 1.770127534866333, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7432586550712585, + "num_tokens": 53608736.0, + "step": 5431 + }, + { + "epoch": 0.4479261152799538, + "grad_norm": 1.8209280967712402, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7676889896392822, + "num_tokens": 53638573.0, + "step": 5432 + }, + { + "epoch": 0.44800857590500537, + "grad_norm": 2.0582613945007324, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.7451748251914978, + "num_tokens": 53671325.0, + "step": 5433 + }, + { + "epoch": 0.4480910365300569, + "grad_norm": 1.8221845626831055, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7595529556274414, + "num_tokens": 53704665.0, + "step": 5434 + }, + { + "epoch": 0.44817349715510846, + "grad_norm": 1.8811368942260742, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.7898303270339966, + "num_tokens": 53735604.0, + "step": 5435 + }, + { + "epoch": 0.44825595778016, + "grad_norm": 1.9730154275894165, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7128555774688721, + "num_tokens": 53765164.0, + "step": 5436 + }, + { + "epoch": 0.4483384184052115, + "grad_norm": 2.2029128074645996, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7096203565597534, + "num_tokens": 53792082.0, + "step": 5437 + }, + { + "epoch": 0.44842087903026306, + "grad_norm": 2.0244061946868896, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7498180270195007, + "num_tokens": 53819550.0, + "step": 5438 + }, + { + "epoch": 0.4485033396553146, + "grad_norm": 2.3388607501983643, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7135568261146545, + "num_tokens": 53841869.0, + "step": 5439 + }, + { + "epoch": 0.4485858002803661, + "grad_norm": 1.9328685998916626, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7388375401496887, + "num_tokens": 53873900.0, + "step": 5440 + }, + { + "epoch": 0.44866826090541767, + "grad_norm": 2.057403802871704, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7300115823745728, + "num_tokens": 53899293.0, + "step": 5441 + }, + { + "epoch": 0.4487507215304692, + "grad_norm": 2.254164457321167, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7373033761978149, + "num_tokens": 53923215.0, + "step": 5442 + }, + { + "epoch": 0.44883318215552076, + "grad_norm": 2.033809185028076, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7262117862701416, + "num_tokens": 53953569.0, + "step": 5443 + }, + { + "epoch": 0.4489156427805723, + "grad_norm": 1.9977043867111206, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7394043207168579, + "num_tokens": 53981487.0, + "step": 5444 + }, + { + "epoch": 0.4489981034056238, + "grad_norm": 1.79768705368042, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7297474145889282, + "num_tokens": 54017300.0, + "step": 5445 + }, + { + "epoch": 0.44908056403067537, + "grad_norm": 2.052542209625244, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7356791496276855, + "num_tokens": 54043455.0, + "step": 5446 + }, + { + "epoch": 0.4491630246557269, + "grad_norm": 1.862065076828003, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7713662385940552, + "num_tokens": 54077923.0, + "step": 5447 + }, + { + "epoch": 0.44924548528077846, + "grad_norm": 1.824554204940796, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.771137535572052, + "num_tokens": 54110415.0, + "step": 5448 + }, + { + "epoch": 0.44932794590583, + "grad_norm": 2.3399815559387207, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.7574899196624756, + "num_tokens": 54130853.0, + "step": 5449 + }, + { + "epoch": 0.4494104065308815, + "grad_norm": 1.9039636850357056, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7442501783370972, + "num_tokens": 54161939.0, + "step": 5450 + }, + { + "epoch": 0.44949286715593306, + "grad_norm": 2.5667431354522705, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.72637540102005, + "num_tokens": 54180783.0, + "step": 5451 + }, + { + "epoch": 0.4495753277809846, + "grad_norm": 1.9903254508972168, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7459998726844788, + "num_tokens": 54209323.0, + "step": 5452 + }, + { + "epoch": 0.4496577884060361, + "grad_norm": 1.874932050704956, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7185803651809692, + "num_tokens": 54241915.0, + "step": 5453 + }, + { + "epoch": 0.44974024903108767, + "grad_norm": 1.882126808166504, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.7867574095726013, + "num_tokens": 54270818.0, + "step": 5454 + }, + { + "epoch": 0.4498227096561392, + "grad_norm": 2.1760683059692383, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7262791991233826, + "num_tokens": 54299382.0, + "step": 5455 + }, + { + "epoch": 0.44990517028119076, + "grad_norm": 2.0332109928131104, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7299174070358276, + "num_tokens": 54330030.0, + "step": 5456 + }, + { + "epoch": 0.4499876309062423, + "grad_norm": 1.9086155891418457, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.727039635181427, + "num_tokens": 54366189.0, + "step": 5457 + }, + { + "epoch": 0.4500700915312938, + "grad_norm": 2.0297141075134277, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7802817821502686, + "num_tokens": 54392860.0, + "step": 5458 + }, + { + "epoch": 0.45015255215634536, + "grad_norm": 1.8860290050506592, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.743283748626709, + "num_tokens": 54425459.0, + "step": 5459 + }, + { + "epoch": 0.4502350127813969, + "grad_norm": 1.748758316040039, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7698719501495361, + "num_tokens": 54457735.0, + "step": 5460 + }, + { + "epoch": 0.4503174734064484, + "grad_norm": 2.1747257709503174, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7161648273468018, + "num_tokens": 54487371.0, + "step": 5461 + }, + { + "epoch": 0.45039993403149997, + "grad_norm": 2.228020668029785, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7228065133094788, + "num_tokens": 54514471.0, + "step": 5462 + }, + { + "epoch": 0.4504823946565515, + "grad_norm": 1.7967170476913452, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7310732007026672, + "num_tokens": 54548680.0, + "step": 5463 + }, + { + "epoch": 0.45056485528160306, + "grad_norm": 2.1400654315948486, + "learning_rate": 1e-06, + "loss": 0.7852, + "mean_token_accuracy": 0.7491245269775391, + "num_tokens": 54573101.0, + "step": 5464 + }, + { + "epoch": 0.4506473159066546, + "grad_norm": 2.0551388263702393, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7489186525344849, + "num_tokens": 54602066.0, + "step": 5465 + }, + { + "epoch": 0.4507297765317061, + "grad_norm": 1.8761768341064453, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7566144466400146, + "num_tokens": 54635519.0, + "step": 5466 + }, + { + "epoch": 0.45081223715675767, + "grad_norm": 1.8390674591064453, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.7964811325073242, + "num_tokens": 54668888.0, + "step": 5467 + }, + { + "epoch": 0.4508946977818092, + "grad_norm": 2.17254638671875, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7360063195228577, + "num_tokens": 54699223.0, + "step": 5468 + }, + { + "epoch": 0.4509771584068607, + "grad_norm": 1.9634451866149902, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7539069056510925, + "num_tokens": 54727541.0, + "step": 5469 + }, + { + "epoch": 0.4510596190319123, + "grad_norm": 2.092942714691162, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7515241503715515, + "num_tokens": 54758268.0, + "step": 5470 + }, + { + "epoch": 0.4511420796569638, + "grad_norm": 2.3761448860168457, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7082996368408203, + "num_tokens": 54781070.0, + "step": 5471 + }, + { + "epoch": 0.45122454028201536, + "grad_norm": 2.0909013748168945, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7500221133232117, + "num_tokens": 54805950.0, + "step": 5472 + }, + { + "epoch": 0.4513070009070669, + "grad_norm": 1.952284812927246, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7256563901901245, + "num_tokens": 54837374.0, + "step": 5473 + }, + { + "epoch": 0.4513894615321184, + "grad_norm": 2.0653984546661377, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7315649390220642, + "num_tokens": 54864149.0, + "step": 5474 + }, + { + "epoch": 0.45147192215716997, + "grad_norm": 1.9231343269348145, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.743712842464447, + "num_tokens": 54896303.0, + "step": 5475 + }, + { + "epoch": 0.4515543827822215, + "grad_norm": 1.6628320217132568, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7558178901672363, + "num_tokens": 54934116.0, + "step": 5476 + }, + { + "epoch": 0.451636843407273, + "grad_norm": 2.133627414703369, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7256473898887634, + "num_tokens": 54962444.0, + "step": 5477 + }, + { + "epoch": 0.4517193040323246, + "grad_norm": 1.7623937129974365, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7651745676994324, + "num_tokens": 54996459.0, + "step": 5478 + }, + { + "epoch": 0.4518017646573761, + "grad_norm": 2.216158151626587, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.775374174118042, + "num_tokens": 55019793.0, + "step": 5479 + }, + { + "epoch": 0.45188422528242767, + "grad_norm": 1.9296071529388428, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7395755648612976, + "num_tokens": 55049159.0, + "step": 5480 + }, + { + "epoch": 0.4519666859074792, + "grad_norm": 1.6860758066177368, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7806679606437683, + "num_tokens": 55084864.0, + "step": 5481 + }, + { + "epoch": 0.4520491465325307, + "grad_norm": 2.1788201332092285, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7155526876449585, + "num_tokens": 55110568.0, + "step": 5482 + }, + { + "epoch": 0.45213160715758227, + "grad_norm": 1.8897514343261719, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7440261840820312, + "num_tokens": 55143378.0, + "step": 5483 + }, + { + "epoch": 0.4522140677826338, + "grad_norm": 1.8998336791992188, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.7867084741592407, + "num_tokens": 55170083.0, + "step": 5484 + }, + { + "epoch": 0.4522965284076853, + "grad_norm": 1.8002023696899414, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7356393337249756, + "num_tokens": 55204479.0, + "step": 5485 + }, + { + "epoch": 0.4523789890327369, + "grad_norm": 1.9311778545379639, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.740489661693573, + "num_tokens": 55236037.0, + "step": 5486 + }, + { + "epoch": 0.4524614496577884, + "grad_norm": 2.208017587661743, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7576110363006592, + "num_tokens": 55261435.0, + "step": 5487 + }, + { + "epoch": 0.45254391028283997, + "grad_norm": 1.8248507976531982, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.7898374795913696, + "num_tokens": 55295492.0, + "step": 5488 + }, + { + "epoch": 0.4526263709078915, + "grad_norm": 2.071226119995117, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7454746961593628, + "num_tokens": 55325369.0, + "step": 5489 + }, + { + "epoch": 0.452708831532943, + "grad_norm": 2.095914602279663, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7138634920120239, + "num_tokens": 55350948.0, + "step": 5490 + }, + { + "epoch": 0.4527912921579946, + "grad_norm": 2.1681888103485107, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7583363056182861, + "num_tokens": 55375471.0, + "step": 5491 + }, + { + "epoch": 0.4528737527830461, + "grad_norm": 2.0967860221862793, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7067242860794067, + "num_tokens": 55406409.0, + "step": 5492 + }, + { + "epoch": 0.4529562134080976, + "grad_norm": 1.9019325971603394, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7645859122276306, + "num_tokens": 55435669.0, + "step": 5493 + }, + { + "epoch": 0.4530386740331492, + "grad_norm": 2.185617446899414, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7351830005645752, + "num_tokens": 55459566.0, + "step": 5494 + }, + { + "epoch": 0.4531211346582007, + "grad_norm": 2.065659523010254, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.745096743106842, + "num_tokens": 55485072.0, + "step": 5495 + }, + { + "epoch": 0.45320359528325227, + "grad_norm": 1.979454517364502, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7433210611343384, + "num_tokens": 55514231.0, + "step": 5496 + }, + { + "epoch": 0.4532860559083038, + "grad_norm": 1.8099206686019897, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.7900486588478088, + "num_tokens": 55545283.0, + "step": 5497 + }, + { + "epoch": 0.4533685165333553, + "grad_norm": 2.198741912841797, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7135806083679199, + "num_tokens": 55570997.0, + "step": 5498 + }, + { + "epoch": 0.4534509771584069, + "grad_norm": 1.844338297843933, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7356983423233032, + "num_tokens": 55603211.0, + "step": 5499 + }, + { + "epoch": 0.4535334377834584, + "grad_norm": 1.890311360359192, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7348911166191101, + "num_tokens": 55637351.0, + "step": 5500 + }, + { + "epoch": 0.4536158984085099, + "grad_norm": 2.0857136249542236, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.756432056427002, + "num_tokens": 55663759.0, + "step": 5501 + }, + { + "epoch": 0.4536983590335615, + "grad_norm": 1.8207323551177979, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7386268377304077, + "num_tokens": 55698228.0, + "step": 5502 + }, + { + "epoch": 0.453780819658613, + "grad_norm": 2.277210235595703, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7661294341087341, + "num_tokens": 55721923.0, + "step": 5503 + }, + { + "epoch": 0.45386328028366457, + "grad_norm": 2.0075223445892334, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7210366129875183, + "num_tokens": 55752301.0, + "step": 5504 + }, + { + "epoch": 0.4539457409087161, + "grad_norm": 1.9749913215637207, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7389191389083862, + "num_tokens": 55781127.0, + "step": 5505 + }, + { + "epoch": 0.4540282015337676, + "grad_norm": 1.8261659145355225, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7275403738021851, + "num_tokens": 55815704.0, + "step": 5506 + }, + { + "epoch": 0.4541106621588192, + "grad_norm": 2.0324742794036865, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.7942206859588623, + "num_tokens": 55840192.0, + "step": 5507 + }, + { + "epoch": 0.4541931227838707, + "grad_norm": 2.2101762294769287, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7423056364059448, + "num_tokens": 55866128.0, + "step": 5508 + }, + { + "epoch": 0.4542755834089222, + "grad_norm": 2.0553367137908936, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7354689240455627, + "num_tokens": 55892831.0, + "step": 5509 + }, + { + "epoch": 0.4543580440339738, + "grad_norm": 1.9089809656143188, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7385495901107788, + "num_tokens": 55926683.0, + "step": 5510 + }, + { + "epoch": 0.4544405046590253, + "grad_norm": 1.833906650543213, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7841549515724182, + "num_tokens": 55957715.0, + "step": 5511 + }, + { + "epoch": 0.4545229652840769, + "grad_norm": 1.906448245048523, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7790646553039551, + "num_tokens": 55988310.0, + "step": 5512 + }, + { + "epoch": 0.4546054259091284, + "grad_norm": 1.9153615236282349, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7326817512512207, + "num_tokens": 56020990.0, + "step": 5513 + }, + { + "epoch": 0.4546878865341799, + "grad_norm": 2.145740270614624, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7632914781570435, + "num_tokens": 56045220.0, + "step": 5514 + }, + { + "epoch": 0.4547703471592315, + "grad_norm": 2.1116840839385986, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7378907203674316, + "num_tokens": 56071141.0, + "step": 5515 + }, + { + "epoch": 0.454852807784283, + "grad_norm": 2.0268990993499756, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7356278896331787, + "num_tokens": 56100256.0, + "step": 5516 + }, + { + "epoch": 0.45493526840933457, + "grad_norm": 1.96145498752594, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7506744265556335, + "num_tokens": 56129369.0, + "step": 5517 + }, + { + "epoch": 0.4550177290343861, + "grad_norm": 2.1019270420074463, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7157965898513794, + "num_tokens": 56155392.0, + "step": 5518 + }, + { + "epoch": 0.4551001896594376, + "grad_norm": 1.9001661539077759, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7678089141845703, + "num_tokens": 56185220.0, + "step": 5519 + }, + { + "epoch": 0.4551826502844892, + "grad_norm": 1.8660635948181152, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7406257390975952, + "num_tokens": 56217283.0, + "step": 5520 + }, + { + "epoch": 0.4552651109095407, + "grad_norm": 1.9264060258865356, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7625468969345093, + "num_tokens": 56246015.0, + "step": 5521 + }, + { + "epoch": 0.4553475715345922, + "grad_norm": 1.923089623451233, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7453161478042603, + "num_tokens": 56277563.0, + "step": 5522 + }, + { + "epoch": 0.4554300321596438, + "grad_norm": 1.8302017450332642, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7748292684555054, + "num_tokens": 56308730.0, + "step": 5523 + }, + { + "epoch": 0.4555124927846953, + "grad_norm": 1.9137746095657349, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7593488097190857, + "num_tokens": 56339000.0, + "step": 5524 + }, + { + "epoch": 0.4555949534097469, + "grad_norm": 2.251424551010132, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7483421564102173, + "num_tokens": 56364712.0, + "step": 5525 + }, + { + "epoch": 0.4556774140347984, + "grad_norm": 1.8536813259124756, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7624801397323608, + "num_tokens": 56397642.0, + "step": 5526 + }, + { + "epoch": 0.4557598746598499, + "grad_norm": 2.4043846130371094, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7304389476776123, + "num_tokens": 56420237.0, + "step": 5527 + }, + { + "epoch": 0.4558423352849015, + "grad_norm": 1.8737767934799194, + "learning_rate": 1e-06, + "loss": 0.5718, + "mean_token_accuracy": 0.8017237782478333, + "num_tokens": 56450508.0, + "step": 5528 + }, + { + "epoch": 0.455924795909953, + "grad_norm": 2.1182219982147217, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7221980690956116, + "num_tokens": 56477861.0, + "step": 5529 + }, + { + "epoch": 0.4560072565350045, + "grad_norm": 1.8586082458496094, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7862601280212402, + "num_tokens": 56506948.0, + "step": 5530 + }, + { + "epoch": 0.4560897171600561, + "grad_norm": 1.8058249950408936, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7774662971496582, + "num_tokens": 56540731.0, + "step": 5531 + }, + { + "epoch": 0.4561721777851076, + "grad_norm": 1.9055676460266113, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7368096709251404, + "num_tokens": 56573335.0, + "step": 5532 + }, + { + "epoch": 0.4562546384101592, + "grad_norm": 2.314300060272217, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7110465168952942, + "num_tokens": 56595257.0, + "step": 5533 + }, + { + "epoch": 0.4563370990352107, + "grad_norm": 1.702242374420166, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7580695748329163, + "num_tokens": 56633819.0, + "step": 5534 + }, + { + "epoch": 0.4564195596602622, + "grad_norm": 1.7512692213058472, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7471094131469727, + "num_tokens": 56667822.0, + "step": 5535 + }, + { + "epoch": 0.4565020202853138, + "grad_norm": 2.0229594707489014, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.748354434967041, + "num_tokens": 56696725.0, + "step": 5536 + }, + { + "epoch": 0.4565844809103653, + "grad_norm": 1.8882479667663574, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.708503782749176, + "num_tokens": 56729689.0, + "step": 5537 + }, + { + "epoch": 0.4566669415354168, + "grad_norm": 2.136444568634033, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7091712355613708, + "num_tokens": 56758387.0, + "step": 5538 + }, + { + "epoch": 0.4567494021604684, + "grad_norm": 1.8917067050933838, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.7788676023483276, + "num_tokens": 56788435.0, + "step": 5539 + }, + { + "epoch": 0.4568318627855199, + "grad_norm": 1.7722333669662476, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.790949821472168, + "num_tokens": 56822666.0, + "step": 5540 + }, + { + "epoch": 0.4569143234105715, + "grad_norm": 2.2411317825317383, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7384427785873413, + "num_tokens": 56847044.0, + "step": 5541 + }, + { + "epoch": 0.456996784035623, + "grad_norm": 1.8348095417022705, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7389373779296875, + "num_tokens": 56879711.0, + "step": 5542 + }, + { + "epoch": 0.4570792446606745, + "grad_norm": 1.9228595495224, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7928982377052307, + "num_tokens": 56908477.0, + "step": 5543 + }, + { + "epoch": 0.4571617052857261, + "grad_norm": 1.9313474893569946, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7405139207839966, + "num_tokens": 56940945.0, + "step": 5544 + }, + { + "epoch": 0.4572441659107776, + "grad_norm": 2.019904136657715, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7541055679321289, + "num_tokens": 56968711.0, + "step": 5545 + }, + { + "epoch": 0.4573266265358291, + "grad_norm": 2.19199800491333, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7483296394348145, + "num_tokens": 56993510.0, + "step": 5546 + }, + { + "epoch": 0.4574090871608807, + "grad_norm": 2.0716068744659424, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.731533408164978, + "num_tokens": 57023730.0, + "step": 5547 + }, + { + "epoch": 0.4574915477859322, + "grad_norm": 2.1070938110351562, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7666114568710327, + "num_tokens": 57054827.0, + "step": 5548 + }, + { + "epoch": 0.4575740084109838, + "grad_norm": 1.7969763278961182, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7858349680900574, + "num_tokens": 57087906.0, + "step": 5549 + }, + { + "epoch": 0.4576564690360353, + "grad_norm": 1.8486782312393188, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7447320222854614, + "num_tokens": 57123999.0, + "step": 5550 + }, + { + "epoch": 0.4577389296610868, + "grad_norm": 1.9442943334579468, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7077620625495911, + "num_tokens": 57157203.0, + "step": 5551 + }, + { + "epoch": 0.4578213902861384, + "grad_norm": 2.2176706790924072, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7538919448852539, + "num_tokens": 57180831.0, + "step": 5552 + }, + { + "epoch": 0.4579038509111899, + "grad_norm": 2.0186164379119873, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7208905220031738, + "num_tokens": 57210534.0, + "step": 5553 + }, + { + "epoch": 0.4579863115362414, + "grad_norm": 1.9538860321044922, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7438640594482422, + "num_tokens": 57243113.0, + "step": 5554 + }, + { + "epoch": 0.458068772161293, + "grad_norm": 2.064724922180176, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7595682144165039, + "num_tokens": 57269668.0, + "step": 5555 + }, + { + "epoch": 0.4581512327863445, + "grad_norm": 2.1066927909851074, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7105938196182251, + "num_tokens": 57298353.0, + "step": 5556 + }, + { + "epoch": 0.4582336934113961, + "grad_norm": 2.0297181606292725, + "learning_rate": 1e-06, + "loss": 0.7706, + "mean_token_accuracy": 0.7579867839813232, + "num_tokens": 57327989.0, + "step": 5557 + }, + { + "epoch": 0.4583161540364476, + "grad_norm": 2.0427913665771484, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7594070434570312, + "num_tokens": 57354140.0, + "step": 5558 + }, + { + "epoch": 0.4583986146614991, + "grad_norm": 1.9723961353302002, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7410333752632141, + "num_tokens": 57387585.0, + "step": 5559 + }, + { + "epoch": 0.4584810752865507, + "grad_norm": 2.0190916061401367, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6996060609817505, + "num_tokens": 57419159.0, + "step": 5560 + }, + { + "epoch": 0.4585635359116022, + "grad_norm": 2.1719021797180176, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.6944925785064697, + "num_tokens": 57446162.0, + "step": 5561 + }, + { + "epoch": 0.4586459965366537, + "grad_norm": 1.8725651502609253, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7425777912139893, + "num_tokens": 57479563.0, + "step": 5562 + }, + { + "epoch": 0.4587284571617053, + "grad_norm": 2.1269872188568115, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7319338321685791, + "num_tokens": 57505519.0, + "step": 5563 + }, + { + "epoch": 0.4588109177867568, + "grad_norm": 1.9564194679260254, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7127622961997986, + "num_tokens": 57537713.0, + "step": 5564 + }, + { + "epoch": 0.4588933784118084, + "grad_norm": 2.038531541824341, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.715485155582428, + "num_tokens": 57567462.0, + "step": 5565 + }, + { + "epoch": 0.4589758390368599, + "grad_norm": 2.135938882827759, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7108250260353088, + "num_tokens": 57594673.0, + "step": 5566 + }, + { + "epoch": 0.4590582996619114, + "grad_norm": 2.0117523670196533, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7208966016769409, + "num_tokens": 57622388.0, + "step": 5567 + }, + { + "epoch": 0.459140760286963, + "grad_norm": 2.0723228454589844, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7428032159805298, + "num_tokens": 57652379.0, + "step": 5568 + }, + { + "epoch": 0.4592232209120145, + "grad_norm": 2.275245428085327, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7310643196105957, + "num_tokens": 57677042.0, + "step": 5569 + }, + { + "epoch": 0.459305681537066, + "grad_norm": 1.9230633974075317, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7358616590499878, + "num_tokens": 57708275.0, + "step": 5570 + }, + { + "epoch": 0.4593881421621176, + "grad_norm": 2.2503106594085693, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7236008644104004, + "num_tokens": 57732169.0, + "step": 5571 + }, + { + "epoch": 0.4594706027871691, + "grad_norm": 1.907885193824768, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7692650556564331, + "num_tokens": 57761274.0, + "step": 5572 + }, + { + "epoch": 0.4595530634122207, + "grad_norm": 2.077664375305176, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7418580055236816, + "num_tokens": 57788930.0, + "step": 5573 + }, + { + "epoch": 0.4596355240372722, + "grad_norm": 2.1830971240997314, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7105146050453186, + "num_tokens": 57820517.0, + "step": 5574 + }, + { + "epoch": 0.4597179846623237, + "grad_norm": 1.8165570497512817, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7759525179862976, + "num_tokens": 57857981.0, + "step": 5575 + }, + { + "epoch": 0.4598004452873753, + "grad_norm": 1.8262907266616821, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7579421997070312, + "num_tokens": 57892883.0, + "step": 5576 + }, + { + "epoch": 0.4598829059124268, + "grad_norm": 2.101477861404419, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7799774408340454, + "num_tokens": 57915874.0, + "step": 5577 + }, + { + "epoch": 0.4599653665374784, + "grad_norm": 1.8674010038375854, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7333505749702454, + "num_tokens": 57949855.0, + "step": 5578 + }, + { + "epoch": 0.4600478271625299, + "grad_norm": 1.9800305366516113, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7589349150657654, + "num_tokens": 57985575.0, + "step": 5579 + }, + { + "epoch": 0.4601302877875814, + "grad_norm": 2.0049076080322266, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7621806859970093, + "num_tokens": 58014200.0, + "step": 5580 + }, + { + "epoch": 0.460212748412633, + "grad_norm": 2.192744255065918, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.72723388671875, + "num_tokens": 58040556.0, + "step": 5581 + }, + { + "epoch": 0.4602952090376845, + "grad_norm": 1.876192569732666, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7493048906326294, + "num_tokens": 58072697.0, + "step": 5582 + }, + { + "epoch": 0.460377669662736, + "grad_norm": 2.0004723072052, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7553136348724365, + "num_tokens": 58101256.0, + "step": 5583 + }, + { + "epoch": 0.4604601302877876, + "grad_norm": 2.2633302211761475, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7221068143844604, + "num_tokens": 58126974.0, + "step": 5584 + }, + { + "epoch": 0.4605425909128391, + "grad_norm": 1.9205517768859863, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7359175682067871, + "num_tokens": 58157397.0, + "step": 5585 + }, + { + "epoch": 0.4606250515378907, + "grad_norm": 2.1497344970703125, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7668508291244507, + "num_tokens": 58187119.0, + "step": 5586 + }, + { + "epoch": 0.4607075121629422, + "grad_norm": 1.9670984745025635, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7395695447921753, + "num_tokens": 58216861.0, + "step": 5587 + }, + { + "epoch": 0.4607899727879937, + "grad_norm": 1.9331343173980713, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7388453483581543, + "num_tokens": 58246321.0, + "step": 5588 + }, + { + "epoch": 0.4608724334130453, + "grad_norm": 1.7602366209030151, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7475183606147766, + "num_tokens": 58282308.0, + "step": 5589 + }, + { + "epoch": 0.4609548940380968, + "grad_norm": 1.6958714723587036, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7580551505088806, + "num_tokens": 58321672.0, + "step": 5590 + }, + { + "epoch": 0.4610373546631483, + "grad_norm": 2.0959982872009277, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7163472771644592, + "num_tokens": 58350786.0, + "step": 5591 + }, + { + "epoch": 0.4611198152881999, + "grad_norm": 2.1048502922058105, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7072362899780273, + "num_tokens": 58379751.0, + "step": 5592 + }, + { + "epoch": 0.4612022759132514, + "grad_norm": 1.8873366117477417, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7650595903396606, + "num_tokens": 58409614.0, + "step": 5593 + }, + { + "epoch": 0.461284736538303, + "grad_norm": 1.858924388885498, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.7840163707733154, + "num_tokens": 58440668.0, + "step": 5594 + }, + { + "epoch": 0.4613671971633545, + "grad_norm": 2.375396966934204, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7153893709182739, + "num_tokens": 58466347.0, + "step": 5595 + }, + { + "epoch": 0.461449657788406, + "grad_norm": 1.9895095825195312, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7606009244918823, + "num_tokens": 58494724.0, + "step": 5596 + }, + { + "epoch": 0.4615321184134576, + "grad_norm": 2.254554271697998, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.726047158241272, + "num_tokens": 58517470.0, + "step": 5597 + }, + { + "epoch": 0.4616145790385091, + "grad_norm": 1.7076495885849, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.7636703252792358, + "num_tokens": 58553983.0, + "step": 5598 + }, + { + "epoch": 0.4616970396635606, + "grad_norm": 2.0265159606933594, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7626378536224365, + "num_tokens": 58581900.0, + "step": 5599 + }, + { + "epoch": 0.4617795002886122, + "grad_norm": 1.9706443548202515, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7657054662704468, + "num_tokens": 58610234.0, + "step": 5600 + }, + { + "epoch": 0.4618619609136637, + "grad_norm": 1.9367706775665283, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7326017618179321, + "num_tokens": 58642376.0, + "step": 5601 + }, + { + "epoch": 0.4619444215387153, + "grad_norm": 1.8908379077911377, + "learning_rate": 1e-06, + "loss": 0.6028, + "mean_token_accuracy": 0.7975777387619019, + "num_tokens": 58670431.0, + "step": 5602 + }, + { + "epoch": 0.4620268821637668, + "grad_norm": 1.7932605743408203, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7606970071792603, + "num_tokens": 58703410.0, + "step": 5603 + }, + { + "epoch": 0.4621093427888183, + "grad_norm": 2.2350728511810303, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7519118189811707, + "num_tokens": 58729043.0, + "step": 5604 + }, + { + "epoch": 0.4621918034138699, + "grad_norm": 2.017711639404297, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7278432846069336, + "num_tokens": 58759464.0, + "step": 5605 + }, + { + "epoch": 0.4622742640389214, + "grad_norm": 2.035060405731201, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7396554350852966, + "num_tokens": 58785827.0, + "step": 5606 + }, + { + "epoch": 0.46235672466397293, + "grad_norm": 2.080095052719116, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.7793285846710205, + "num_tokens": 58810733.0, + "step": 5607 + }, + { + "epoch": 0.4624391852890245, + "grad_norm": 2.069138526916504, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7308996915817261, + "num_tokens": 58840265.0, + "step": 5608 + }, + { + "epoch": 0.462521645914076, + "grad_norm": 1.6698217391967773, + "learning_rate": 1e-06, + "loss": 0.5863, + "mean_token_accuracy": 0.8129216432571411, + "num_tokens": 58876069.0, + "step": 5609 + }, + { + "epoch": 0.4626041065391276, + "grad_norm": 1.9829827547073364, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7553691864013672, + "num_tokens": 58908035.0, + "step": 5610 + }, + { + "epoch": 0.4626865671641791, + "grad_norm": 2.405194044113159, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7469643354415894, + "num_tokens": 58929782.0, + "step": 5611 + }, + { + "epoch": 0.4627690277892306, + "grad_norm": 2.33164119720459, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7399850487709045, + "num_tokens": 58953262.0, + "step": 5612 + }, + { + "epoch": 0.4628514884142822, + "grad_norm": 2.2131662368774414, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7602648735046387, + "num_tokens": 58976915.0, + "step": 5613 + }, + { + "epoch": 0.4629339490393337, + "grad_norm": 1.8159538507461548, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8052668571472168, + "num_tokens": 59007631.0, + "step": 5614 + }, + { + "epoch": 0.46301640966438523, + "grad_norm": 1.876075267791748, + "learning_rate": 1e-06, + "loss": 0.7396, + "mean_token_accuracy": 0.7629776000976562, + "num_tokens": 59038346.0, + "step": 5615 + }, + { + "epoch": 0.4630988702894368, + "grad_norm": 1.545264482498169, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7285650372505188, + "num_tokens": 59081325.0, + "step": 5616 + }, + { + "epoch": 0.4631813309144883, + "grad_norm": 2.1372408866882324, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7432510852813721, + "num_tokens": 59106764.0, + "step": 5617 + }, + { + "epoch": 0.4632637915395399, + "grad_norm": 2.2530648708343506, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7655670642852783, + "num_tokens": 59130416.0, + "step": 5618 + }, + { + "epoch": 0.4633462521645914, + "grad_norm": 1.9384815692901611, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7786092758178711, + "num_tokens": 59162201.0, + "step": 5619 + }, + { + "epoch": 0.4634287127896429, + "grad_norm": 1.692847490310669, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.7963299751281738, + "num_tokens": 59197844.0, + "step": 5620 + }, + { + "epoch": 0.4635111734146945, + "grad_norm": 1.9211952686309814, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7456831336021423, + "num_tokens": 59229231.0, + "step": 5621 + }, + { + "epoch": 0.463593634039746, + "grad_norm": 2.166806221008301, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.754403829574585, + "num_tokens": 59253826.0, + "step": 5622 + }, + { + "epoch": 0.46367609466479753, + "grad_norm": 1.7696810960769653, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7644733190536499, + "num_tokens": 59286248.0, + "step": 5623 + }, + { + "epoch": 0.4637585552898491, + "grad_norm": 1.848523736000061, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7356868982315063, + "num_tokens": 59321425.0, + "step": 5624 + }, + { + "epoch": 0.4638410159149006, + "grad_norm": 1.8644442558288574, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.750286340713501, + "num_tokens": 59352195.0, + "step": 5625 + }, + { + "epoch": 0.4639234765399522, + "grad_norm": 1.7949012517929077, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.736598014831543, + "num_tokens": 59386713.0, + "step": 5626 + }, + { + "epoch": 0.4640059371650037, + "grad_norm": 2.0572071075439453, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7146979570388794, + "num_tokens": 59418563.0, + "step": 5627 + }, + { + "epoch": 0.46408839779005523, + "grad_norm": 1.958385944366455, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7201399803161621, + "num_tokens": 59448555.0, + "step": 5628 + }, + { + "epoch": 0.4641708584151068, + "grad_norm": 1.7767446041107178, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.782602071762085, + "num_tokens": 59481106.0, + "step": 5629 + }, + { + "epoch": 0.4642533190401583, + "grad_norm": 1.9471449851989746, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7408014535903931, + "num_tokens": 59512965.0, + "step": 5630 + }, + { + "epoch": 0.46433577966520984, + "grad_norm": 2.236213207244873, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7346580028533936, + "num_tokens": 59537699.0, + "step": 5631 + }, + { + "epoch": 0.4644182402902614, + "grad_norm": 2.1082231998443604, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.725139856338501, + "num_tokens": 59567536.0, + "step": 5632 + }, + { + "epoch": 0.4645007009153129, + "grad_norm": 1.9422823190689087, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7657033801078796, + "num_tokens": 59597354.0, + "step": 5633 + }, + { + "epoch": 0.4645831615403645, + "grad_norm": 1.852416753768921, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7082189917564392, + "num_tokens": 59631430.0, + "step": 5634 + }, + { + "epoch": 0.464665622165416, + "grad_norm": 2.1125261783599854, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7130521535873413, + "num_tokens": 59658910.0, + "step": 5635 + }, + { + "epoch": 0.46474808279046753, + "grad_norm": 1.8821073770523071, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.7930524349212646, + "num_tokens": 59686756.0, + "step": 5636 + }, + { + "epoch": 0.4648305434155191, + "grad_norm": 2.1020092964172363, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7394810914993286, + "num_tokens": 59716995.0, + "step": 5637 + }, + { + "epoch": 0.4649130040405706, + "grad_norm": 2.0914456844329834, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7505741119384766, + "num_tokens": 59744112.0, + "step": 5638 + }, + { + "epoch": 0.46499546466562214, + "grad_norm": 1.9597269296646118, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7266550660133362, + "num_tokens": 59773936.0, + "step": 5639 + }, + { + "epoch": 0.4650779252906737, + "grad_norm": 1.8040220737457275, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7298164367675781, + "num_tokens": 59808789.0, + "step": 5640 + }, + { + "epoch": 0.4651603859157252, + "grad_norm": 1.7837719917297363, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7538874745368958, + "num_tokens": 59841074.0, + "step": 5641 + }, + { + "epoch": 0.4652428465407768, + "grad_norm": 2.0766897201538086, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7399122714996338, + "num_tokens": 59870328.0, + "step": 5642 + }, + { + "epoch": 0.4653253071658283, + "grad_norm": 2.056640625, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7519038319587708, + "num_tokens": 59899013.0, + "step": 5643 + }, + { + "epoch": 0.46540776779087983, + "grad_norm": 2.217278242111206, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6902846097946167, + "num_tokens": 59926548.0, + "step": 5644 + }, + { + "epoch": 0.4654902284159314, + "grad_norm": 2.1412155628204346, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7198482751846313, + "num_tokens": 59953792.0, + "step": 5645 + }, + { + "epoch": 0.4655726890409829, + "grad_norm": 1.8982619047164917, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7535505294799805, + "num_tokens": 59985519.0, + "step": 5646 + }, + { + "epoch": 0.4656551496660345, + "grad_norm": 2.136565685272217, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7539688944816589, + "num_tokens": 60012124.0, + "step": 5647 + }, + { + "epoch": 0.465737610291086, + "grad_norm": 2.0583386421203613, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7140533328056335, + "num_tokens": 60042249.0, + "step": 5648 + }, + { + "epoch": 0.46582007091613753, + "grad_norm": 1.8127896785736084, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7727524042129517, + "num_tokens": 60075481.0, + "step": 5649 + }, + { + "epoch": 0.4659025315411891, + "grad_norm": 1.987683653831482, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.7380762100219727, + "num_tokens": 60107376.0, + "step": 5650 + }, + { + "epoch": 0.4659849921662406, + "grad_norm": 2.068547248840332, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7567273378372192, + "num_tokens": 60137113.0, + "step": 5651 + }, + { + "epoch": 0.46606745279129214, + "grad_norm": 1.8516178131103516, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7468523979187012, + "num_tokens": 60168382.0, + "step": 5652 + }, + { + "epoch": 0.4661499134163437, + "grad_norm": 2.1455931663513184, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7097899913787842, + "num_tokens": 60195916.0, + "step": 5653 + }, + { + "epoch": 0.4662323740413952, + "grad_norm": 2.0205888748168945, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7183440923690796, + "num_tokens": 60227103.0, + "step": 5654 + }, + { + "epoch": 0.4663148346664468, + "grad_norm": 1.8351105451583862, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7006193399429321, + "num_tokens": 60262443.0, + "step": 5655 + }, + { + "epoch": 0.4663972952914983, + "grad_norm": 1.719571590423584, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7554339170455933, + "num_tokens": 60299406.0, + "step": 5656 + }, + { + "epoch": 0.46647975591654983, + "grad_norm": 1.7230112552642822, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7545312643051147, + "num_tokens": 60335368.0, + "step": 5657 + }, + { + "epoch": 0.4665622165416014, + "grad_norm": 1.7831817865371704, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.784725546836853, + "num_tokens": 60369798.0, + "step": 5658 + }, + { + "epoch": 0.4666446771666529, + "grad_norm": 2.032585620880127, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7253485321998596, + "num_tokens": 60397215.0, + "step": 5659 + }, + { + "epoch": 0.46672713779170444, + "grad_norm": 2.2846174240112305, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7460023164749146, + "num_tokens": 60420361.0, + "step": 5660 + }, + { + "epoch": 0.466809598416756, + "grad_norm": 2.048438787460327, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.744610071182251, + "num_tokens": 60451172.0, + "step": 5661 + }, + { + "epoch": 0.46689205904180753, + "grad_norm": 2.154991865158081, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7203131914138794, + "num_tokens": 60477859.0, + "step": 5662 + }, + { + "epoch": 0.4669745196668591, + "grad_norm": 2.0492823123931885, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7115641236305237, + "num_tokens": 60506245.0, + "step": 5663 + }, + { + "epoch": 0.4670569802919106, + "grad_norm": 2.072176694869995, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7507590055465698, + "num_tokens": 60533738.0, + "step": 5664 + }, + { + "epoch": 0.46713944091696213, + "grad_norm": 2.0167489051818848, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.7841649055480957, + "num_tokens": 60559339.0, + "step": 5665 + }, + { + "epoch": 0.4672219015420137, + "grad_norm": 1.8940283060073853, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7546669244766235, + "num_tokens": 60590788.0, + "step": 5666 + }, + { + "epoch": 0.4673043621670652, + "grad_norm": 1.824859857559204, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7585448622703552, + "num_tokens": 60624668.0, + "step": 5667 + }, + { + "epoch": 0.46738682279211674, + "grad_norm": 2.0557098388671875, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7857401371002197, + "num_tokens": 60650084.0, + "step": 5668 + }, + { + "epoch": 0.4674692834171683, + "grad_norm": 1.7379604578018188, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7593623399734497, + "num_tokens": 60685033.0, + "step": 5669 + }, + { + "epoch": 0.46755174404221983, + "grad_norm": 1.9041553735733032, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7166075706481934, + "num_tokens": 60717212.0, + "step": 5670 + }, + { + "epoch": 0.4676342046672714, + "grad_norm": 1.9971953630447388, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7493611574172974, + "num_tokens": 60746516.0, + "step": 5671 + }, + { + "epoch": 0.4677166652923229, + "grad_norm": 1.7433972358703613, + "learning_rate": 1e-06, + "loss": 0.5955, + "mean_token_accuracy": 0.8093547821044922, + "num_tokens": 60781436.0, + "step": 5672 + }, + { + "epoch": 0.46779912591737444, + "grad_norm": 1.6858144998550415, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7786625027656555, + "num_tokens": 60819861.0, + "step": 5673 + }, + { + "epoch": 0.467881586542426, + "grad_norm": 2.0084638595581055, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7533844709396362, + "num_tokens": 60847186.0, + "step": 5674 + }, + { + "epoch": 0.4679640471674775, + "grad_norm": 2.0710527896881104, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7137124538421631, + "num_tokens": 60876757.0, + "step": 5675 + }, + { + "epoch": 0.46804650779252904, + "grad_norm": 2.143430471420288, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7587063312530518, + "num_tokens": 60905065.0, + "step": 5676 + }, + { + "epoch": 0.4681289684175806, + "grad_norm": 2.0023627281188965, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7301016449928284, + "num_tokens": 60935742.0, + "step": 5677 + }, + { + "epoch": 0.46821142904263213, + "grad_norm": 1.8837695121765137, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7060133814811707, + "num_tokens": 60970246.0, + "step": 5678 + }, + { + "epoch": 0.4682938896676837, + "grad_norm": 2.228101968765259, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7152254581451416, + "num_tokens": 60995114.0, + "step": 5679 + }, + { + "epoch": 0.4683763502927352, + "grad_norm": 2.031273126602173, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7170177698135376, + "num_tokens": 61024782.0, + "step": 5680 + }, + { + "epoch": 0.46845881091778674, + "grad_norm": 1.974310040473938, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.732457160949707, + "num_tokens": 61055900.0, + "step": 5681 + }, + { + "epoch": 0.4685412715428383, + "grad_norm": 1.8191131353378296, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7439448833465576, + "num_tokens": 61091055.0, + "step": 5682 + }, + { + "epoch": 0.46862373216788983, + "grad_norm": 1.8639365434646606, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.72198486328125, + "num_tokens": 61122619.0, + "step": 5683 + }, + { + "epoch": 0.46870619279294135, + "grad_norm": 1.8082973957061768, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7645944952964783, + "num_tokens": 61155251.0, + "step": 5684 + }, + { + "epoch": 0.4687886534179929, + "grad_norm": 1.8585717678070068, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7414162755012512, + "num_tokens": 61190197.0, + "step": 5685 + }, + { + "epoch": 0.46887111404304443, + "grad_norm": 1.8105297088623047, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7418581247329712, + "num_tokens": 61221222.0, + "step": 5686 + }, + { + "epoch": 0.468953574668096, + "grad_norm": 2.1453185081481934, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.7944696545600891, + "num_tokens": 61243191.0, + "step": 5687 + }, + { + "epoch": 0.4690360352931475, + "grad_norm": 1.9244107007980347, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.768993616104126, + "num_tokens": 61273449.0, + "step": 5688 + }, + { + "epoch": 0.46911849591819904, + "grad_norm": 1.8451522588729858, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7746569514274597, + "num_tokens": 61305681.0, + "step": 5689 + }, + { + "epoch": 0.4692009565432506, + "grad_norm": 1.9183441400527954, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7830570936203003, + "num_tokens": 61337378.0, + "step": 5690 + }, + { + "epoch": 0.46928341716830213, + "grad_norm": 1.9644602537155151, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7332048416137695, + "num_tokens": 61366066.0, + "step": 5691 + }, + { + "epoch": 0.46936587779335365, + "grad_norm": 2.194394826889038, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7256667613983154, + "num_tokens": 61391109.0, + "step": 5692 + }, + { + "epoch": 0.4694483384184052, + "grad_norm": 1.733801245689392, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7204011678695679, + "num_tokens": 61431179.0, + "step": 5693 + }, + { + "epoch": 0.46953079904345674, + "grad_norm": 1.6942895650863647, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7784242033958435, + "num_tokens": 61467314.0, + "step": 5694 + }, + { + "epoch": 0.4696132596685083, + "grad_norm": 1.9006938934326172, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7324399948120117, + "num_tokens": 61498943.0, + "step": 5695 + }, + { + "epoch": 0.4696957202935598, + "grad_norm": 2.0814573764801025, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7257084846496582, + "num_tokens": 61525586.0, + "step": 5696 + }, + { + "epoch": 0.46977818091861134, + "grad_norm": 1.9139008522033691, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7552843689918518, + "num_tokens": 61556926.0, + "step": 5697 + }, + { + "epoch": 0.4698606415436629, + "grad_norm": 1.921260952949524, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7460960745811462, + "num_tokens": 61588401.0, + "step": 5698 + }, + { + "epoch": 0.46994310216871443, + "grad_norm": 2.3162097930908203, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7200227975845337, + "num_tokens": 61612763.0, + "step": 5699 + }, + { + "epoch": 0.47002556279376595, + "grad_norm": 1.8432120084762573, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.7545374631881714, + "num_tokens": 61644392.0, + "step": 5700 + }, + { + "epoch": 0.4701080234188175, + "grad_norm": 1.8857002258300781, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.735622763633728, + "num_tokens": 61678680.0, + "step": 5701 + }, + { + "epoch": 0.47019048404386904, + "grad_norm": 1.9923826456069946, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7380777597427368, + "num_tokens": 61706007.0, + "step": 5702 + }, + { + "epoch": 0.4702729446689206, + "grad_norm": 1.8653062582015991, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7558392882347107, + "num_tokens": 61737135.0, + "step": 5703 + }, + { + "epoch": 0.47035540529397213, + "grad_norm": 2.0129268169403076, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7420388460159302, + "num_tokens": 61764701.0, + "step": 5704 + }, + { + "epoch": 0.47043786591902365, + "grad_norm": 2.34175443649292, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.73761385679245, + "num_tokens": 61789022.0, + "step": 5705 + }, + { + "epoch": 0.4705203265440752, + "grad_norm": 1.8941781520843506, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7443658709526062, + "num_tokens": 61822299.0, + "step": 5706 + }, + { + "epoch": 0.47060278716912674, + "grad_norm": 1.984266757965088, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7232967019081116, + "num_tokens": 61852061.0, + "step": 5707 + }, + { + "epoch": 0.4706852477941783, + "grad_norm": 2.0057127475738525, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6853138208389282, + "num_tokens": 61883965.0, + "step": 5708 + }, + { + "epoch": 0.4707677084192298, + "grad_norm": 1.8256276845932007, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7646181583404541, + "num_tokens": 61916656.0, + "step": 5709 + }, + { + "epoch": 0.47085016904428134, + "grad_norm": 1.9616779088974, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7266044616699219, + "num_tokens": 61946803.0, + "step": 5710 + }, + { + "epoch": 0.4709326296693329, + "grad_norm": 1.9856446981430054, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.7657172679901123, + "num_tokens": 61976133.0, + "step": 5711 + }, + { + "epoch": 0.47101509029438443, + "grad_norm": 1.894208550453186, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7331571578979492, + "num_tokens": 62007685.0, + "step": 5712 + }, + { + "epoch": 0.47109755091943595, + "grad_norm": 1.8544012308120728, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7293736934661865, + "num_tokens": 62039031.0, + "step": 5713 + }, + { + "epoch": 0.4711800115444875, + "grad_norm": 1.9704804420471191, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7239876985549927, + "num_tokens": 62070784.0, + "step": 5714 + }, + { + "epoch": 0.47126247216953904, + "grad_norm": 1.853966236114502, + "learning_rate": 1e-06, + "loss": 0.6085, + "mean_token_accuracy": 0.7900158166885376, + "num_tokens": 62102297.0, + "step": 5715 + }, + { + "epoch": 0.4713449327945906, + "grad_norm": 2.3954365253448486, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7379913330078125, + "num_tokens": 62124881.0, + "step": 5716 + }, + { + "epoch": 0.4714273934196421, + "grad_norm": 2.0270538330078125, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7012186050415039, + "num_tokens": 62155234.0, + "step": 5717 + }, + { + "epoch": 0.47150985404469364, + "grad_norm": 2.0027823448181152, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7626864910125732, + "num_tokens": 62185544.0, + "step": 5718 + }, + { + "epoch": 0.4715923146697452, + "grad_norm": 1.6491765975952148, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7449932098388672, + "num_tokens": 62224627.0, + "step": 5719 + }, + { + "epoch": 0.47167477529479673, + "grad_norm": 2.1383614540100098, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7459396123886108, + "num_tokens": 62248673.0, + "step": 5720 + }, + { + "epoch": 0.47175723591984825, + "grad_norm": 2.0784831047058105, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7455391883850098, + "num_tokens": 62273153.0, + "step": 5721 + }, + { + "epoch": 0.4718396965448998, + "grad_norm": 2.076920747756958, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7156084775924683, + "num_tokens": 62301560.0, + "step": 5722 + }, + { + "epoch": 0.47192215716995134, + "grad_norm": 1.7929167747497559, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7559504508972168, + "num_tokens": 62334771.0, + "step": 5723 + }, + { + "epoch": 0.4720046177950029, + "grad_norm": 2.0280492305755615, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7188414335250854, + "num_tokens": 62366966.0, + "step": 5724 + }, + { + "epoch": 0.47208707842005443, + "grad_norm": 1.64546537399292, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7472774982452393, + "num_tokens": 62405296.0, + "step": 5725 + }, + { + "epoch": 0.47216953904510595, + "grad_norm": 1.762863278388977, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7091290950775146, + "num_tokens": 62439770.0, + "step": 5726 + }, + { + "epoch": 0.4722519996701575, + "grad_norm": 1.7495096921920776, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7677906155586243, + "num_tokens": 62473754.0, + "step": 5727 + }, + { + "epoch": 0.47233446029520904, + "grad_norm": 1.8631014823913574, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7359364032745361, + "num_tokens": 62505748.0, + "step": 5728 + }, + { + "epoch": 0.47241692092026055, + "grad_norm": 2.127575397491455, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7323547601699829, + "num_tokens": 62536691.0, + "step": 5729 + }, + { + "epoch": 0.4724993815453121, + "grad_norm": 2.138216495513916, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7508136034011841, + "num_tokens": 62562654.0, + "step": 5730 + }, + { + "epoch": 0.47258184217036364, + "grad_norm": 1.7851758003234863, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7493569850921631, + "num_tokens": 62600093.0, + "step": 5731 + }, + { + "epoch": 0.4726643027954152, + "grad_norm": 1.7310682535171509, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7286257743835449, + "num_tokens": 62639345.0, + "step": 5732 + }, + { + "epoch": 0.47274676342046673, + "grad_norm": 1.9930614233016968, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7691842317581177, + "num_tokens": 62666976.0, + "step": 5733 + }, + { + "epoch": 0.47282922404551825, + "grad_norm": 2.0190515518188477, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7486347556114197, + "num_tokens": 62696227.0, + "step": 5734 + }, + { + "epoch": 0.4729116846705698, + "grad_norm": 1.9545094966888428, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7657290697097778, + "num_tokens": 62723623.0, + "step": 5735 + }, + { + "epoch": 0.47299414529562134, + "grad_norm": 1.8850504159927368, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7669076323509216, + "num_tokens": 62757140.0, + "step": 5736 + }, + { + "epoch": 0.47307660592067285, + "grad_norm": 1.8330316543579102, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7793587446212769, + "num_tokens": 62788734.0, + "step": 5737 + }, + { + "epoch": 0.4731590665457244, + "grad_norm": 2.1036829948425293, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7232677936553955, + "num_tokens": 62819476.0, + "step": 5738 + }, + { + "epoch": 0.47324152717077594, + "grad_norm": 1.870625376701355, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7372540235519409, + "num_tokens": 62853569.0, + "step": 5739 + }, + { + "epoch": 0.4733239877958275, + "grad_norm": 1.901229977607727, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7674533128738403, + "num_tokens": 62882168.0, + "step": 5740 + }, + { + "epoch": 0.47340644842087903, + "grad_norm": 2.205548048019409, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.738689124584198, + "num_tokens": 62907392.0, + "step": 5741 + }, + { + "epoch": 0.47348890904593055, + "grad_norm": 2.0843729972839355, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.77480149269104, + "num_tokens": 62937030.0, + "step": 5742 + }, + { + "epoch": 0.4735713696709821, + "grad_norm": 2.2711668014526367, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7081368565559387, + "num_tokens": 62963211.0, + "step": 5743 + }, + { + "epoch": 0.47365383029603364, + "grad_norm": 1.7783812284469604, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7664557695388794, + "num_tokens": 62996921.0, + "step": 5744 + }, + { + "epoch": 0.47373629092108516, + "grad_norm": 1.803985357284546, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7635047435760498, + "num_tokens": 63030878.0, + "step": 5745 + }, + { + "epoch": 0.47381875154613673, + "grad_norm": 2.320918560028076, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7369731068611145, + "num_tokens": 63053931.0, + "step": 5746 + }, + { + "epoch": 0.47390121217118825, + "grad_norm": 2.0394515991210938, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7637327313423157, + "num_tokens": 63082337.0, + "step": 5747 + }, + { + "epoch": 0.4739836727962398, + "grad_norm": 2.025634765625, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7532905340194702, + "num_tokens": 63110433.0, + "step": 5748 + }, + { + "epoch": 0.47406613342129134, + "grad_norm": 1.7063013315200806, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.7938663363456726, + "num_tokens": 63147885.0, + "step": 5749 + }, + { + "epoch": 0.47414859404634285, + "grad_norm": 1.7709375619888306, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7626532316207886, + "num_tokens": 63184968.0, + "step": 5750 + }, + { + "epoch": 0.4742310546713944, + "grad_norm": 1.6912243366241455, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7595492005348206, + "num_tokens": 63220504.0, + "step": 5751 + }, + { + "epoch": 0.47431351529644594, + "grad_norm": 1.924359917640686, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7408053874969482, + "num_tokens": 63253038.0, + "step": 5752 + }, + { + "epoch": 0.47439597592149746, + "grad_norm": 1.8704252243041992, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7287587523460388, + "num_tokens": 63283981.0, + "step": 5753 + }, + { + "epoch": 0.47447843654654903, + "grad_norm": 2.102231740951538, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.7730845212936401, + "num_tokens": 63309819.0, + "step": 5754 + }, + { + "epoch": 0.47456089717160055, + "grad_norm": 1.9687596559524536, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7384824156761169, + "num_tokens": 63340500.0, + "step": 5755 + }, + { + "epoch": 0.4746433577966521, + "grad_norm": 1.8165628910064697, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7629115581512451, + "num_tokens": 63371170.0, + "step": 5756 + }, + { + "epoch": 0.47472581842170364, + "grad_norm": 2.410592555999756, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7326106429100037, + "num_tokens": 63393668.0, + "step": 5757 + }, + { + "epoch": 0.47480827904675516, + "grad_norm": 2.041689157485962, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7225590944290161, + "num_tokens": 63422239.0, + "step": 5758 + }, + { + "epoch": 0.4748907396718067, + "grad_norm": 2.025813341140747, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7062556743621826, + "num_tokens": 63451255.0, + "step": 5759 + }, + { + "epoch": 0.47497320029685824, + "grad_norm": 1.8103197813034058, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.7959020137786865, + "num_tokens": 63485686.0, + "step": 5760 + }, + { + "epoch": 0.47505566092190976, + "grad_norm": 2.167853355407715, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7165101766586304, + "num_tokens": 63512426.0, + "step": 5761 + }, + { + "epoch": 0.47513812154696133, + "grad_norm": 1.878737211227417, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.725793719291687, + "num_tokens": 63546344.0, + "step": 5762 + }, + { + "epoch": 0.47522058217201285, + "grad_norm": 1.7185331583023071, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7743700742721558, + "num_tokens": 63584584.0, + "step": 5763 + }, + { + "epoch": 0.4753030427970644, + "grad_norm": 1.8396713733673096, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.750312864780426, + "num_tokens": 63616896.0, + "step": 5764 + }, + { + "epoch": 0.47538550342211594, + "grad_norm": 1.9452794790267944, + "learning_rate": 1e-06, + "loss": 0.6103, + "mean_token_accuracy": 0.8008227944374084, + "num_tokens": 63645280.0, + "step": 5765 + }, + { + "epoch": 0.47546796404716746, + "grad_norm": 2.0794403553009033, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7150246500968933, + "num_tokens": 63673682.0, + "step": 5766 + }, + { + "epoch": 0.47555042467221903, + "grad_norm": 2.3065760135650635, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7294657230377197, + "num_tokens": 63696662.0, + "step": 5767 + }, + { + "epoch": 0.47563288529727055, + "grad_norm": 1.9916881322860718, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7845124006271362, + "num_tokens": 63724465.0, + "step": 5768 + }, + { + "epoch": 0.47571534592232206, + "grad_norm": 2.0417702198028564, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7630879282951355, + "num_tokens": 63751704.0, + "step": 5769 + }, + { + "epoch": 0.47579780654737364, + "grad_norm": 2.0791525840759277, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7249000668525696, + "num_tokens": 63778432.0, + "step": 5770 + }, + { + "epoch": 0.47588026717242515, + "grad_norm": 2.3511626720428467, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7294878959655762, + "num_tokens": 63804301.0, + "step": 5771 + }, + { + "epoch": 0.4759627277974767, + "grad_norm": 1.7468029260635376, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7604948282241821, + "num_tokens": 63840604.0, + "step": 5772 + }, + { + "epoch": 0.47604518842252824, + "grad_norm": 2.198219060897827, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7290599942207336, + "num_tokens": 63865626.0, + "step": 5773 + }, + { + "epoch": 0.47612764904757976, + "grad_norm": 2.074037790298462, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7385560274124146, + "num_tokens": 63893050.0, + "step": 5774 + }, + { + "epoch": 0.47621010967263133, + "grad_norm": 1.8394204378128052, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7499444484710693, + "num_tokens": 63924948.0, + "step": 5775 + }, + { + "epoch": 0.47629257029768285, + "grad_norm": 2.043081045150757, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7416021823883057, + "num_tokens": 63953556.0, + "step": 5776 + }, + { + "epoch": 0.4763750309227344, + "grad_norm": 1.7531099319458008, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7158986926078796, + "num_tokens": 63991537.0, + "step": 5777 + }, + { + "epoch": 0.47645749154778594, + "grad_norm": 1.9823930263519287, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7813155651092529, + "num_tokens": 64023444.0, + "step": 5778 + }, + { + "epoch": 0.47653995217283746, + "grad_norm": 2.081480026245117, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.7991374135017395, + "num_tokens": 64050860.0, + "step": 5779 + }, + { + "epoch": 0.47662241279788903, + "grad_norm": 1.5545789003372192, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7685813903808594, + "num_tokens": 64091519.0, + "step": 5780 + }, + { + "epoch": 0.47670487342294054, + "grad_norm": 1.9791014194488525, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7422176599502563, + "num_tokens": 64122377.0, + "step": 5781 + }, + { + "epoch": 0.47678733404799206, + "grad_norm": 1.8001574277877808, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7198395729064941, + "num_tokens": 64159953.0, + "step": 5782 + }, + { + "epoch": 0.47686979467304363, + "grad_norm": 1.8151228427886963, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.7889991998672485, + "num_tokens": 64192147.0, + "step": 5783 + }, + { + "epoch": 0.47695225529809515, + "grad_norm": 1.99579656124115, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.745512843132019, + "num_tokens": 64220854.0, + "step": 5784 + }, + { + "epoch": 0.4770347159231467, + "grad_norm": 1.7421400547027588, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7669607996940613, + "num_tokens": 64257322.0, + "step": 5785 + }, + { + "epoch": 0.47711717654819824, + "grad_norm": 2.0889830589294434, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7487315535545349, + "num_tokens": 64284371.0, + "step": 5786 + }, + { + "epoch": 0.47719963717324976, + "grad_norm": 2.016801357269287, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7417541146278381, + "num_tokens": 64312750.0, + "step": 5787 + }, + { + "epoch": 0.47728209779830133, + "grad_norm": 1.9082717895507812, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7205145359039307, + "num_tokens": 64346576.0, + "step": 5788 + }, + { + "epoch": 0.47736455842335285, + "grad_norm": 2.4440224170684814, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.706606388092041, + "num_tokens": 64369877.0, + "step": 5789 + }, + { + "epoch": 0.47744701904840436, + "grad_norm": 1.8372408151626587, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7339019775390625, + "num_tokens": 64403851.0, + "step": 5790 + }, + { + "epoch": 0.47752947967345594, + "grad_norm": 1.9919301271438599, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7273578643798828, + "num_tokens": 64437854.0, + "step": 5791 + }, + { + "epoch": 0.47761194029850745, + "grad_norm": 1.9279509782791138, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7142692804336548, + "num_tokens": 64467797.0, + "step": 5792 + }, + { + "epoch": 0.477694400923559, + "grad_norm": 1.9342331886291504, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.7812789082527161, + "num_tokens": 64496204.0, + "step": 5793 + }, + { + "epoch": 0.47777686154861054, + "grad_norm": 1.8850513696670532, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7704266905784607, + "num_tokens": 64525328.0, + "step": 5794 + }, + { + "epoch": 0.47785932217366206, + "grad_norm": 1.8503434658050537, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7643690705299377, + "num_tokens": 64557574.0, + "step": 5795 + }, + { + "epoch": 0.47794178279871363, + "grad_norm": 1.9220788478851318, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.710085391998291, + "num_tokens": 64591062.0, + "step": 5796 + }, + { + "epoch": 0.47802424342376515, + "grad_norm": 1.7634942531585693, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7528228759765625, + "num_tokens": 64626047.0, + "step": 5797 + }, + { + "epoch": 0.47810670404881667, + "grad_norm": 1.9001199007034302, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7441278100013733, + "num_tokens": 64659854.0, + "step": 5798 + }, + { + "epoch": 0.47818916467386824, + "grad_norm": 1.8703937530517578, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7815694808959961, + "num_tokens": 64689515.0, + "step": 5799 + }, + { + "epoch": 0.47827162529891976, + "grad_norm": 1.7642598152160645, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.7879178524017334, + "num_tokens": 64727822.0, + "step": 5800 + }, + { + "epoch": 0.47835408592397133, + "grad_norm": 2.1168723106384277, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7773006558418274, + "num_tokens": 64752725.0, + "step": 5801 + }, + { + "epoch": 0.47843654654902285, + "grad_norm": 2.168245315551758, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6958075761795044, + "num_tokens": 64781467.0, + "step": 5802 + }, + { + "epoch": 0.47851900717407436, + "grad_norm": 1.7534520626068115, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.7585051655769348, + "num_tokens": 64816801.0, + "step": 5803 + }, + { + "epoch": 0.47860146779912593, + "grad_norm": 2.2889397144317627, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7064718008041382, + "num_tokens": 64842735.0, + "step": 5804 + }, + { + "epoch": 0.47868392842417745, + "grad_norm": 2.2622478008270264, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7237156629562378, + "num_tokens": 64868190.0, + "step": 5805 + }, + { + "epoch": 0.47876638904922897, + "grad_norm": 2.303422451019287, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7570053339004517, + "num_tokens": 64891880.0, + "step": 5806 + }, + { + "epoch": 0.47884884967428054, + "grad_norm": 2.1761136054992676, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7684809565544128, + "num_tokens": 64916312.0, + "step": 5807 + }, + { + "epoch": 0.47893131029933206, + "grad_norm": 1.8749010562896729, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7297056317329407, + "num_tokens": 64946721.0, + "step": 5808 + }, + { + "epoch": 0.47901377092438363, + "grad_norm": 2.217820644378662, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7599616050720215, + "num_tokens": 64971596.0, + "step": 5809 + }, + { + "epoch": 0.47909623154943515, + "grad_norm": 1.8544515371322632, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7341622710227966, + "num_tokens": 65004869.0, + "step": 5810 + }, + { + "epoch": 0.47917869217448666, + "grad_norm": 1.8451955318450928, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7489444613456726, + "num_tokens": 65036935.0, + "step": 5811 + }, + { + "epoch": 0.47926115279953824, + "grad_norm": 2.0203211307525635, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7458398938179016, + "num_tokens": 65065453.0, + "step": 5812 + }, + { + "epoch": 0.47934361342458975, + "grad_norm": 2.1000936031341553, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7155040502548218, + "num_tokens": 65092669.0, + "step": 5813 + }, + { + "epoch": 0.47942607404964127, + "grad_norm": 2.046429395675659, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7278714179992676, + "num_tokens": 65120034.0, + "step": 5814 + }, + { + "epoch": 0.47950853467469284, + "grad_norm": 2.1055963039398193, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7205771207809448, + "num_tokens": 65149581.0, + "step": 5815 + }, + { + "epoch": 0.47959099529974436, + "grad_norm": 2.0997519493103027, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7782552242279053, + "num_tokens": 65175377.0, + "step": 5816 + }, + { + "epoch": 0.47967345592479593, + "grad_norm": 1.818832278251648, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7539738416671753, + "num_tokens": 65205080.0, + "step": 5817 + }, + { + "epoch": 0.47975591654984745, + "grad_norm": 1.881577491760254, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7220325469970703, + "num_tokens": 65237373.0, + "step": 5818 + }, + { + "epoch": 0.47983837717489897, + "grad_norm": 2.0268750190734863, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.740839958190918, + "num_tokens": 65266497.0, + "step": 5819 + }, + { + "epoch": 0.47992083779995054, + "grad_norm": 2.082716703414917, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7159888744354248, + "num_tokens": 65293609.0, + "step": 5820 + }, + { + "epoch": 0.48000329842500206, + "grad_norm": 2.0084002017974854, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7311389446258545, + "num_tokens": 65324150.0, + "step": 5821 + }, + { + "epoch": 0.4800857590500536, + "grad_norm": 2.079103708267212, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7114652991294861, + "num_tokens": 65352779.0, + "step": 5822 + }, + { + "epoch": 0.48016821967510515, + "grad_norm": 2.1216652393341064, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.733545184135437, + "num_tokens": 65379380.0, + "step": 5823 + }, + { + "epoch": 0.48025068030015666, + "grad_norm": 1.9018486738204956, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7161295413970947, + "num_tokens": 65414441.0, + "step": 5824 + }, + { + "epoch": 0.48033314092520824, + "grad_norm": 2.008247137069702, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7253457903862, + "num_tokens": 65447754.0, + "step": 5825 + }, + { + "epoch": 0.48041560155025975, + "grad_norm": 1.8662612438201904, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7471439242362976, + "num_tokens": 65482243.0, + "step": 5826 + }, + { + "epoch": 0.48049806217531127, + "grad_norm": 1.807371735572815, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7546262145042419, + "num_tokens": 65520153.0, + "step": 5827 + }, + { + "epoch": 0.48058052280036284, + "grad_norm": 2.1969454288482666, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7625553011894226, + "num_tokens": 65542421.0, + "step": 5828 + }, + { + "epoch": 0.48066298342541436, + "grad_norm": 1.834067940711975, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.703798234462738, + "num_tokens": 65576560.0, + "step": 5829 + }, + { + "epoch": 0.4807454440504659, + "grad_norm": 1.650518774986267, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7668850421905518, + "num_tokens": 65614228.0, + "step": 5830 + }, + { + "epoch": 0.48082790467551745, + "grad_norm": 1.7582515478134155, + "learning_rate": 1e-06, + "loss": 0.6219, + "mean_token_accuracy": 0.7883901000022888, + "num_tokens": 65646392.0, + "step": 5831 + }, + { + "epoch": 0.48091036530056896, + "grad_norm": 2.2378990650177, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7587687969207764, + "num_tokens": 65668811.0, + "step": 5832 + }, + { + "epoch": 0.48099282592562054, + "grad_norm": 1.7771108150482178, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.742741584777832, + "num_tokens": 65702366.0, + "step": 5833 + }, + { + "epoch": 0.48107528655067205, + "grad_norm": 1.7786717414855957, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7402684092521667, + "num_tokens": 65739367.0, + "step": 5834 + }, + { + "epoch": 0.48115774717572357, + "grad_norm": 1.8126811981201172, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7452142238616943, + "num_tokens": 65776372.0, + "step": 5835 + }, + { + "epoch": 0.48124020780077514, + "grad_norm": 1.9988349676132202, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.72752845287323, + "num_tokens": 65808037.0, + "step": 5836 + }, + { + "epoch": 0.48132266842582666, + "grad_norm": 1.591201663017273, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7866638898849487, + "num_tokens": 65850554.0, + "step": 5837 + }, + { + "epoch": 0.48140512905087823, + "grad_norm": 1.9944528341293335, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7616247534751892, + "num_tokens": 65878200.0, + "step": 5838 + }, + { + "epoch": 0.48148758967592975, + "grad_norm": 1.9530107975006104, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7330440878868103, + "num_tokens": 65907107.0, + "step": 5839 + }, + { + "epoch": 0.48157005030098127, + "grad_norm": 2.3157036304473877, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.7046826481819153, + "num_tokens": 65931401.0, + "step": 5840 + }, + { + "epoch": 0.48165251092603284, + "grad_norm": 1.8817195892333984, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7423826456069946, + "num_tokens": 65965391.0, + "step": 5841 + }, + { + "epoch": 0.48173497155108436, + "grad_norm": 2.0798394680023193, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7459479570388794, + "num_tokens": 65993715.0, + "step": 5842 + }, + { + "epoch": 0.4818174321761359, + "grad_norm": 1.8651025295257568, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7711116671562195, + "num_tokens": 66025502.0, + "step": 5843 + }, + { + "epoch": 0.48189989280118745, + "grad_norm": 1.9353346824645996, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7171418070793152, + "num_tokens": 66057807.0, + "step": 5844 + }, + { + "epoch": 0.48198235342623896, + "grad_norm": 1.7256009578704834, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.8003939390182495, + "num_tokens": 66092212.0, + "step": 5845 + }, + { + "epoch": 0.48206481405129054, + "grad_norm": 1.916205883026123, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7608656287193298, + "num_tokens": 66123081.0, + "step": 5846 + }, + { + "epoch": 0.48214727467634205, + "grad_norm": 1.846527338027954, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7718029618263245, + "num_tokens": 66157099.0, + "step": 5847 + }, + { + "epoch": 0.48222973530139357, + "grad_norm": 2.0156280994415283, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7475766539573669, + "num_tokens": 66184972.0, + "step": 5848 + }, + { + "epoch": 0.48231219592644514, + "grad_norm": 2.0696375370025635, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7679824829101562, + "num_tokens": 66210875.0, + "step": 5849 + }, + { + "epoch": 0.48239465655149666, + "grad_norm": 1.8263572454452515, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7580734491348267, + "num_tokens": 66245877.0, + "step": 5850 + }, + { + "epoch": 0.4824771171765482, + "grad_norm": 2.1031243801116943, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7248561382293701, + "num_tokens": 66274251.0, + "step": 5851 + }, + { + "epoch": 0.48255957780159975, + "grad_norm": 1.8645591735839844, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.726172685623169, + "num_tokens": 66307473.0, + "step": 5852 + }, + { + "epoch": 0.48264203842665127, + "grad_norm": 1.9447673559188843, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7573329210281372, + "num_tokens": 66337707.0, + "step": 5853 + }, + { + "epoch": 0.48272449905170284, + "grad_norm": 1.8775674104690552, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7616744637489319, + "num_tokens": 66369381.0, + "step": 5854 + }, + { + "epoch": 0.48280695967675435, + "grad_norm": 2.177999258041382, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.6974682211875916, + "num_tokens": 66396777.0, + "step": 5855 + }, + { + "epoch": 0.48288942030180587, + "grad_norm": 2.001041889190674, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.729948878288269, + "num_tokens": 66425797.0, + "step": 5856 + }, + { + "epoch": 0.48297188092685744, + "grad_norm": 1.9770565032958984, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7319673299789429, + "num_tokens": 66454111.0, + "step": 5857 + }, + { + "epoch": 0.48305434155190896, + "grad_norm": 2.4129223823547363, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7155399322509766, + "num_tokens": 66475894.0, + "step": 5858 + }, + { + "epoch": 0.4831368021769605, + "grad_norm": 2.2408323287963867, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7283543348312378, + "num_tokens": 66500062.0, + "step": 5859 + }, + { + "epoch": 0.48321926280201205, + "grad_norm": 1.930017352104187, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7674601674079895, + "num_tokens": 66527120.0, + "step": 5860 + }, + { + "epoch": 0.48330172342706357, + "grad_norm": 2.1059937477111816, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7428517937660217, + "num_tokens": 66554579.0, + "step": 5861 + }, + { + "epoch": 0.48338418405211514, + "grad_norm": 1.8458346128463745, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7533098459243774, + "num_tokens": 66584982.0, + "step": 5862 + }, + { + "epoch": 0.48346664467716666, + "grad_norm": 1.921284794807434, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7367774248123169, + "num_tokens": 66615159.0, + "step": 5863 + }, + { + "epoch": 0.4835491053022182, + "grad_norm": 2.1153905391693115, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7646224498748779, + "num_tokens": 66642285.0, + "step": 5864 + }, + { + "epoch": 0.48363156592726975, + "grad_norm": 1.7733477354049683, + "learning_rate": 1e-06, + "loss": 0.7396, + "mean_token_accuracy": 0.762921154499054, + "num_tokens": 66676679.0, + "step": 5865 + }, + { + "epoch": 0.48371402655232126, + "grad_norm": 1.817894697189331, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7482771277427673, + "num_tokens": 66715335.0, + "step": 5866 + }, + { + "epoch": 0.4837964871773728, + "grad_norm": 2.1032257080078125, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7131277322769165, + "num_tokens": 66743849.0, + "step": 5867 + }, + { + "epoch": 0.48387894780242435, + "grad_norm": 1.7041147947311401, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7317097187042236, + "num_tokens": 66779701.0, + "step": 5868 + }, + { + "epoch": 0.48396140842747587, + "grad_norm": 1.9439244270324707, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7437663078308105, + "num_tokens": 66809252.0, + "step": 5869 + }, + { + "epoch": 0.48404386905252744, + "grad_norm": 1.9756876230239868, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7417465448379517, + "num_tokens": 66839522.0, + "step": 5870 + }, + { + "epoch": 0.48412632967757896, + "grad_norm": 2.1694891452789307, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.761680006980896, + "num_tokens": 66865197.0, + "step": 5871 + }, + { + "epoch": 0.4842087903026305, + "grad_norm": 2.1350040435791016, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.785730242729187, + "num_tokens": 66888801.0, + "step": 5872 + }, + { + "epoch": 0.48429125092768205, + "grad_norm": 1.8319947719573975, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7728322744369507, + "num_tokens": 66920512.0, + "step": 5873 + }, + { + "epoch": 0.48437371155273357, + "grad_norm": 2.1406960487365723, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7202386856079102, + "num_tokens": 66948825.0, + "step": 5874 + }, + { + "epoch": 0.4844561721777851, + "grad_norm": 1.8911426067352295, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7577269673347473, + "num_tokens": 66980973.0, + "step": 5875 + }, + { + "epoch": 0.48453863280283666, + "grad_norm": 2.0463366508483887, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7457486391067505, + "num_tokens": 67010310.0, + "step": 5876 + }, + { + "epoch": 0.48462109342788817, + "grad_norm": 2.0300204753875732, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7451465725898743, + "num_tokens": 67038714.0, + "step": 5877 + }, + { + "epoch": 0.48470355405293974, + "grad_norm": 1.8970545530319214, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7744569778442383, + "num_tokens": 67065150.0, + "step": 5878 + }, + { + "epoch": 0.48478601467799126, + "grad_norm": 1.778538703918457, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7327340841293335, + "num_tokens": 67100590.0, + "step": 5879 + }, + { + "epoch": 0.4848684753030428, + "grad_norm": 2.078396797180176, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7624857425689697, + "num_tokens": 67126194.0, + "step": 5880 + }, + { + "epoch": 0.48495093592809435, + "grad_norm": 2.054123640060425, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7304681539535522, + "num_tokens": 67157150.0, + "step": 5881 + }, + { + "epoch": 0.48503339655314587, + "grad_norm": 1.7528518438339233, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7726736664772034, + "num_tokens": 67193187.0, + "step": 5882 + }, + { + "epoch": 0.4851158571781974, + "grad_norm": 1.9626978635787964, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7648482322692871, + "num_tokens": 67223725.0, + "step": 5883 + }, + { + "epoch": 0.48519831780324896, + "grad_norm": 2.0486972332000732, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7339208126068115, + "num_tokens": 67252047.0, + "step": 5884 + }, + { + "epoch": 0.4852807784283005, + "grad_norm": 2.1469674110412598, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7397925853729248, + "num_tokens": 67279674.0, + "step": 5885 + }, + { + "epoch": 0.48536323905335205, + "grad_norm": 1.9176124334335327, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.7915374040603638, + "num_tokens": 67304391.0, + "step": 5886 + }, + { + "epoch": 0.48544569967840356, + "grad_norm": 2.0437145233154297, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7562519311904907, + "num_tokens": 67331997.0, + "step": 5887 + }, + { + "epoch": 0.4855281603034551, + "grad_norm": 2.1302809715270996, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7247927784919739, + "num_tokens": 67358112.0, + "step": 5888 + }, + { + "epoch": 0.48561062092850665, + "grad_norm": 2.203596830368042, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7532190084457397, + "num_tokens": 67384266.0, + "step": 5889 + }, + { + "epoch": 0.48569308155355817, + "grad_norm": 1.9828015565872192, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7455247640609741, + "num_tokens": 67413769.0, + "step": 5890 + }, + { + "epoch": 0.4857755421786097, + "grad_norm": 2.0404722690582275, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.742742121219635, + "num_tokens": 67442143.0, + "step": 5891 + }, + { + "epoch": 0.48585800280366126, + "grad_norm": 2.0386104583740234, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.726280689239502, + "num_tokens": 67468708.0, + "step": 5892 + }, + { + "epoch": 0.4859404634287128, + "grad_norm": 2.0409250259399414, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7513114213943481, + "num_tokens": 67497625.0, + "step": 5893 + }, + { + "epoch": 0.48602292405376435, + "grad_norm": 2.353792667388916, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7092539072036743, + "num_tokens": 67522389.0, + "step": 5894 + }, + { + "epoch": 0.48610538467881587, + "grad_norm": 1.9969234466552734, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7577052712440491, + "num_tokens": 67551284.0, + "step": 5895 + }, + { + "epoch": 0.4861878453038674, + "grad_norm": 1.7441505193710327, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7657657861709595, + "num_tokens": 67583499.0, + "step": 5896 + }, + { + "epoch": 0.48627030592891896, + "grad_norm": 1.8997929096221924, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.720032811164856, + "num_tokens": 67617096.0, + "step": 5897 + }, + { + "epoch": 0.4863527665539705, + "grad_norm": 1.8646084070205688, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.709889829158783, + "num_tokens": 67651735.0, + "step": 5898 + }, + { + "epoch": 0.486435227179022, + "grad_norm": 1.8614555597305298, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.792289137840271, + "num_tokens": 67682320.0, + "step": 5899 + }, + { + "epoch": 0.48651768780407356, + "grad_norm": 2.192213296890259, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.721482515335083, + "num_tokens": 67709930.0, + "step": 5900 + }, + { + "epoch": 0.4866001484291251, + "grad_norm": 1.9785234928131104, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7443832159042358, + "num_tokens": 67735711.0, + "step": 5901 + }, + { + "epoch": 0.48668260905417665, + "grad_norm": 1.8304054737091064, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7744218111038208, + "num_tokens": 67768890.0, + "step": 5902 + }, + { + "epoch": 0.48676506967922817, + "grad_norm": 1.9195070266723633, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7776496410369873, + "num_tokens": 67796666.0, + "step": 5903 + }, + { + "epoch": 0.4868475303042797, + "grad_norm": 2.2193756103515625, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7378170490264893, + "num_tokens": 67820836.0, + "step": 5904 + }, + { + "epoch": 0.48692999092933126, + "grad_norm": 1.9308595657348633, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7672739028930664, + "num_tokens": 67851483.0, + "step": 5905 + }, + { + "epoch": 0.4870124515543828, + "grad_norm": 2.1663243770599365, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7253543138504028, + "num_tokens": 67876877.0, + "step": 5906 + }, + { + "epoch": 0.48709491217943435, + "grad_norm": 2.1025421619415283, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7034733295440674, + "num_tokens": 67907225.0, + "step": 5907 + }, + { + "epoch": 0.48717737280448586, + "grad_norm": 1.8923271894454956, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7429459095001221, + "num_tokens": 67936728.0, + "step": 5908 + }, + { + "epoch": 0.4872598334295374, + "grad_norm": 1.9492896795272827, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7675009965896606, + "num_tokens": 67967767.0, + "step": 5909 + }, + { + "epoch": 0.48734229405458895, + "grad_norm": 2.203490972518921, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7226612567901611, + "num_tokens": 67993660.0, + "step": 5910 + }, + { + "epoch": 0.48742475467964047, + "grad_norm": 1.949357271194458, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7380378842353821, + "num_tokens": 68022907.0, + "step": 5911 + }, + { + "epoch": 0.487507215304692, + "grad_norm": 2.0466508865356445, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7673073410987854, + "num_tokens": 68049576.0, + "step": 5912 + }, + { + "epoch": 0.48758967592974356, + "grad_norm": 1.9377995729446411, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7165365815162659, + "num_tokens": 68083495.0, + "step": 5913 + }, + { + "epoch": 0.4876721365547951, + "grad_norm": 1.9201158285140991, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7162473201751709, + "num_tokens": 68114753.0, + "step": 5914 + }, + { + "epoch": 0.48775459717984665, + "grad_norm": 1.9759613275527954, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7230855226516724, + "num_tokens": 68144902.0, + "step": 5915 + }, + { + "epoch": 0.48783705780489817, + "grad_norm": 1.960808277130127, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7494339346885681, + "num_tokens": 68176755.0, + "step": 5916 + }, + { + "epoch": 0.4879195184299497, + "grad_norm": 1.9456390142440796, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7589268684387207, + "num_tokens": 68206909.0, + "step": 5917 + }, + { + "epoch": 0.48800197905500126, + "grad_norm": 1.890099048614502, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7416849732398987, + "num_tokens": 68238880.0, + "step": 5918 + }, + { + "epoch": 0.4880844396800528, + "grad_norm": 1.9698559045791626, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7740127444267273, + "num_tokens": 68266020.0, + "step": 5919 + }, + { + "epoch": 0.4881669003051043, + "grad_norm": 1.9088249206542969, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.783173680305481, + "num_tokens": 68298059.0, + "step": 5920 + }, + { + "epoch": 0.48824936093015586, + "grad_norm": 2.05019211769104, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7238069772720337, + "num_tokens": 68327466.0, + "step": 5921 + }, + { + "epoch": 0.4883318215552074, + "grad_norm": 2.087067127227783, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7274585962295532, + "num_tokens": 68357298.0, + "step": 5922 + }, + { + "epoch": 0.48841428218025895, + "grad_norm": 2.0724611282348633, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7149829864501953, + "num_tokens": 68388192.0, + "step": 5923 + }, + { + "epoch": 0.48849674280531047, + "grad_norm": 2.1343586444854736, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7088637351989746, + "num_tokens": 68414203.0, + "step": 5924 + }, + { + "epoch": 0.488579203430362, + "grad_norm": 2.3174219131469727, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7221983671188354, + "num_tokens": 68438814.0, + "step": 5925 + }, + { + "epoch": 0.48866166405541356, + "grad_norm": 1.9142075777053833, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7473568320274353, + "num_tokens": 68470085.0, + "step": 5926 + }, + { + "epoch": 0.4887441246804651, + "grad_norm": 1.925071358680725, + "learning_rate": 1e-06, + "loss": 0.5536, + "mean_token_accuracy": 0.8126702308654785, + "num_tokens": 68498585.0, + "step": 5927 + }, + { + "epoch": 0.4888265853055166, + "grad_norm": 1.7026963233947754, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7786060571670532, + "num_tokens": 68536463.0, + "step": 5928 + }, + { + "epoch": 0.48890904593056816, + "grad_norm": 1.9832863807678223, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7410516738891602, + "num_tokens": 68564085.0, + "step": 5929 + }, + { + "epoch": 0.4889915065556197, + "grad_norm": 2.0842084884643555, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7208840847015381, + "num_tokens": 68590806.0, + "step": 5930 + }, + { + "epoch": 0.48907396718067125, + "grad_norm": 2.0385901927948, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7537662386894226, + "num_tokens": 68615938.0, + "step": 5931 + }, + { + "epoch": 0.48915642780572277, + "grad_norm": 2.1909143924713135, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7643082141876221, + "num_tokens": 68639784.0, + "step": 5932 + }, + { + "epoch": 0.4892388884307743, + "grad_norm": 1.9014695882797241, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7479711771011353, + "num_tokens": 68672127.0, + "step": 5933 + }, + { + "epoch": 0.48932134905582586, + "grad_norm": 1.7890580892562866, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7370800971984863, + "num_tokens": 68712409.0, + "step": 5934 + }, + { + "epoch": 0.4894038096808774, + "grad_norm": 1.9646039009094238, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.7811975479125977, + "num_tokens": 68741797.0, + "step": 5935 + }, + { + "epoch": 0.4894862703059289, + "grad_norm": 2.2548365592956543, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7265467643737793, + "num_tokens": 68767826.0, + "step": 5936 + }, + { + "epoch": 0.48956873093098047, + "grad_norm": 1.992192029953003, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7593855857849121, + "num_tokens": 68798884.0, + "step": 5937 + }, + { + "epoch": 0.489651191556032, + "grad_norm": 2.3500101566314697, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.7075494527816772, + "num_tokens": 68822534.0, + "step": 5938 + }, + { + "epoch": 0.48973365218108356, + "grad_norm": 1.790662407875061, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7846384048461914, + "num_tokens": 68854552.0, + "step": 5939 + }, + { + "epoch": 0.4898161128061351, + "grad_norm": 2.1769421100616455, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7114697694778442, + "num_tokens": 68880847.0, + "step": 5940 + }, + { + "epoch": 0.4898985734311866, + "grad_norm": 2.1159698963165283, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7195338010787964, + "num_tokens": 68907023.0, + "step": 5941 + }, + { + "epoch": 0.48998103405623816, + "grad_norm": 1.936276912689209, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7189117074012756, + "num_tokens": 68938670.0, + "step": 5942 + }, + { + "epoch": 0.4900634946812897, + "grad_norm": 1.6631848812103271, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7725107669830322, + "num_tokens": 68974978.0, + "step": 5943 + }, + { + "epoch": 0.4901459553063412, + "grad_norm": 2.0214641094207764, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7270735502243042, + "num_tokens": 69004334.0, + "step": 5944 + }, + { + "epoch": 0.49022841593139277, + "grad_norm": 2.043919801712036, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7539987564086914, + "num_tokens": 69032443.0, + "step": 5945 + }, + { + "epoch": 0.4903108765564443, + "grad_norm": 1.8395144939422607, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.7754201292991638, + "num_tokens": 69063335.0, + "step": 5946 + }, + { + "epoch": 0.49039333718149586, + "grad_norm": 2.057474374771118, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7414859533309937, + "num_tokens": 69089782.0, + "step": 5947 + }, + { + "epoch": 0.4904757978065474, + "grad_norm": 2.1568918228149414, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7647166848182678, + "num_tokens": 69114670.0, + "step": 5948 + }, + { + "epoch": 0.4905582584315989, + "grad_norm": 1.9007772207260132, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7577729821205139, + "num_tokens": 69145845.0, + "step": 5949 + }, + { + "epoch": 0.49064071905665047, + "grad_norm": 2.0074591636657715, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7313156723976135, + "num_tokens": 69175349.0, + "step": 5950 + }, + { + "epoch": 0.490723179681702, + "grad_norm": 1.8926804065704346, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7291380167007446, + "num_tokens": 69206682.0, + "step": 5951 + }, + { + "epoch": 0.4908056403067535, + "grad_norm": 1.9564650058746338, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7513505220413208, + "num_tokens": 69237793.0, + "step": 5952 + }, + { + "epoch": 0.49088810093180507, + "grad_norm": 2.0932841300964355, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7363454103469849, + "num_tokens": 69262843.0, + "step": 5953 + }, + { + "epoch": 0.4909705615568566, + "grad_norm": 1.9013645648956299, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7446030378341675, + "num_tokens": 69293250.0, + "step": 5954 + }, + { + "epoch": 0.49105302218190816, + "grad_norm": 1.8985164165496826, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7444612383842468, + "num_tokens": 69324980.0, + "step": 5955 + }, + { + "epoch": 0.4911354828069597, + "grad_norm": 1.9762601852416992, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7697303295135498, + "num_tokens": 69350505.0, + "step": 5956 + }, + { + "epoch": 0.4912179434320112, + "grad_norm": 1.9946105480194092, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7821696400642395, + "num_tokens": 69381079.0, + "step": 5957 + }, + { + "epoch": 0.49130040405706277, + "grad_norm": 1.8960070610046387, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7587158679962158, + "num_tokens": 69412633.0, + "step": 5958 + }, + { + "epoch": 0.4913828646821143, + "grad_norm": 2.0757198333740234, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7173911929130554, + "num_tokens": 69438541.0, + "step": 5959 + }, + { + "epoch": 0.4914653253071658, + "grad_norm": 1.7041833400726318, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7214696407318115, + "num_tokens": 69477391.0, + "step": 5960 + }, + { + "epoch": 0.4915477859322174, + "grad_norm": 2.035468816757202, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7536465525627136, + "num_tokens": 69504822.0, + "step": 5961 + }, + { + "epoch": 0.4916302465572689, + "grad_norm": 1.997326374053955, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7602018713951111, + "num_tokens": 69534297.0, + "step": 5962 + }, + { + "epoch": 0.49171270718232046, + "grad_norm": 1.8684756755828857, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7783867716789246, + "num_tokens": 69565600.0, + "step": 5963 + }, + { + "epoch": 0.491795167807372, + "grad_norm": 1.9740737676620483, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7183933258056641, + "num_tokens": 69597427.0, + "step": 5964 + }, + { + "epoch": 0.4918776284324235, + "grad_norm": 2.127427816390991, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7212154865264893, + "num_tokens": 69625019.0, + "step": 5965 + }, + { + "epoch": 0.49196008905747507, + "grad_norm": 1.9473717212677002, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7290129661560059, + "num_tokens": 69656923.0, + "step": 5966 + }, + { + "epoch": 0.4920425496825266, + "grad_norm": 1.9315013885498047, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.732499361038208, + "num_tokens": 69689409.0, + "step": 5967 + }, + { + "epoch": 0.49212501030757816, + "grad_norm": 1.955653190612793, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7486608028411865, + "num_tokens": 69718231.0, + "step": 5968 + }, + { + "epoch": 0.4922074709326297, + "grad_norm": 2.0004451274871826, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7299195528030396, + "num_tokens": 69746733.0, + "step": 5969 + }, + { + "epoch": 0.4922899315576812, + "grad_norm": 1.9996907711029053, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7221123576164246, + "num_tokens": 69776517.0, + "step": 5970 + }, + { + "epoch": 0.49237239218273277, + "grad_norm": 1.769266963005066, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7797237634658813, + "num_tokens": 69808558.0, + "step": 5971 + }, + { + "epoch": 0.4924548528077843, + "grad_norm": 1.917352318763733, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7424464821815491, + "num_tokens": 69842172.0, + "step": 5972 + }, + { + "epoch": 0.4925373134328358, + "grad_norm": 2.0561883449554443, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7643934488296509, + "num_tokens": 69869599.0, + "step": 5973 + }, + { + "epoch": 0.49261977405788737, + "grad_norm": 2.236839771270752, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7639252543449402, + "num_tokens": 69894102.0, + "step": 5974 + }, + { + "epoch": 0.4927022346829389, + "grad_norm": 2.044877529144287, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.6891336441040039, + "num_tokens": 69926029.0, + "step": 5975 + }, + { + "epoch": 0.49278469530799046, + "grad_norm": 1.8382279872894287, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7577966451644897, + "num_tokens": 69960278.0, + "step": 5976 + }, + { + "epoch": 0.492867155933042, + "grad_norm": 2.3795359134674072, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7250682711601257, + "num_tokens": 69984207.0, + "step": 5977 + }, + { + "epoch": 0.4929496165580935, + "grad_norm": 1.9343053102493286, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7670538425445557, + "num_tokens": 70013767.0, + "step": 5978 + }, + { + "epoch": 0.49303207718314507, + "grad_norm": 1.8489879369735718, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7552482485771179, + "num_tokens": 70047602.0, + "step": 5979 + }, + { + "epoch": 0.4931145378081966, + "grad_norm": 1.8062148094177246, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7627156972885132, + "num_tokens": 70080384.0, + "step": 5980 + }, + { + "epoch": 0.4931969984332481, + "grad_norm": 2.1503562927246094, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7243865728378296, + "num_tokens": 70106153.0, + "step": 5981 + }, + { + "epoch": 0.4932794590582997, + "grad_norm": 2.2191007137298584, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7605694532394409, + "num_tokens": 70129179.0, + "step": 5982 + }, + { + "epoch": 0.4933619196833512, + "grad_norm": 1.7752139568328857, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.7492296099662781, + "num_tokens": 70165735.0, + "step": 5983 + }, + { + "epoch": 0.49344438030840276, + "grad_norm": 1.7977232933044434, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7714636325836182, + "num_tokens": 70199149.0, + "step": 5984 + }, + { + "epoch": 0.4935268409334543, + "grad_norm": 2.1035513877868652, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.721318244934082, + "num_tokens": 70227516.0, + "step": 5985 + }, + { + "epoch": 0.4936093015585058, + "grad_norm": 1.542299509048462, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7894352674484253, + "num_tokens": 70269167.0, + "step": 5986 + }, + { + "epoch": 0.49369176218355737, + "grad_norm": 1.9520100355148315, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7423669099807739, + "num_tokens": 70298607.0, + "step": 5987 + }, + { + "epoch": 0.4937742228086089, + "grad_norm": 1.7718276977539062, + "learning_rate": 1e-06, + "loss": 0.5781, + "mean_token_accuracy": 0.8189381957054138, + "num_tokens": 70324963.0, + "step": 5988 + }, + { + "epoch": 0.4938566834336604, + "grad_norm": 1.899084448814392, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.7967579364776611, + "num_tokens": 70352298.0, + "step": 5989 + }, + { + "epoch": 0.493939144058712, + "grad_norm": 1.8641926050186157, + "learning_rate": 1e-06, + "loss": 0.7683, + "mean_token_accuracy": 0.7483500242233276, + "num_tokens": 70383916.0, + "step": 5990 + }, + { + "epoch": 0.4940216046837635, + "grad_norm": 2.0140514373779297, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.7968126535415649, + "num_tokens": 70408884.0, + "step": 5991 + }, + { + "epoch": 0.49410406530881507, + "grad_norm": 1.8885308504104614, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7249678373336792, + "num_tokens": 70445191.0, + "step": 5992 + }, + { + "epoch": 0.4941865259338666, + "grad_norm": 1.815057396888733, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7492106556892395, + "num_tokens": 70477828.0, + "step": 5993 + }, + { + "epoch": 0.4942689865589181, + "grad_norm": 2.173731803894043, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7408145666122437, + "num_tokens": 70505118.0, + "step": 5994 + }, + { + "epoch": 0.49435144718396967, + "grad_norm": 1.9000778198242188, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.740196168422699, + "num_tokens": 70537514.0, + "step": 5995 + }, + { + "epoch": 0.4944339078090212, + "grad_norm": 1.9064692258834839, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7590703964233398, + "num_tokens": 70567672.0, + "step": 5996 + }, + { + "epoch": 0.4945163684340727, + "grad_norm": 1.690704584121704, + "learning_rate": 1e-06, + "loss": 0.5601, + "mean_token_accuracy": 0.8083365559577942, + "num_tokens": 70597876.0, + "step": 5997 + }, + { + "epoch": 0.4945988290591243, + "grad_norm": 1.9914441108703613, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7864969968795776, + "num_tokens": 70627799.0, + "step": 5998 + }, + { + "epoch": 0.4946812896841758, + "grad_norm": 1.9347128868103027, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7362359166145325, + "num_tokens": 70659959.0, + "step": 5999 + }, + { + "epoch": 0.49476375030922737, + "grad_norm": 2.3523523807525635, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7159359455108643, + "num_tokens": 70681633.0, + "step": 6000 + }, + { + "epoch": 0.4948462109342789, + "grad_norm": 1.8224191665649414, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7552372217178345, + "num_tokens": 70715712.0, + "step": 6001 + }, + { + "epoch": 0.4949286715593304, + "grad_norm": 2.067680835723877, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.7868701815605164, + "num_tokens": 70742666.0, + "step": 6002 + }, + { + "epoch": 0.495011132184382, + "grad_norm": 1.8649647235870361, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.7689155340194702, + "num_tokens": 70772363.0, + "step": 6003 + }, + { + "epoch": 0.4950935928094335, + "grad_norm": 1.9335631132125854, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7303537130355835, + "num_tokens": 70801813.0, + "step": 6004 + }, + { + "epoch": 0.495176053434485, + "grad_norm": 1.9616895914077759, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.7825765013694763, + "num_tokens": 70828885.0, + "step": 6005 + }, + { + "epoch": 0.4952585140595366, + "grad_norm": 1.9786133766174316, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7247860431671143, + "num_tokens": 70859356.0, + "step": 6006 + }, + { + "epoch": 0.4953409746845881, + "grad_norm": 2.2494308948516846, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7472236156463623, + "num_tokens": 70882024.0, + "step": 6007 + }, + { + "epoch": 0.49542343530963967, + "grad_norm": 2.2543163299560547, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7295030355453491, + "num_tokens": 70907046.0, + "step": 6008 + }, + { + "epoch": 0.4955058959346912, + "grad_norm": 2.1570098400115967, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7019489407539368, + "num_tokens": 70935177.0, + "step": 6009 + }, + { + "epoch": 0.4955883565597427, + "grad_norm": 1.8449597358703613, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7707207202911377, + "num_tokens": 70965145.0, + "step": 6010 + }, + { + "epoch": 0.4956708171847943, + "grad_norm": 2.05670428276062, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.7812792062759399, + "num_tokens": 70992654.0, + "step": 6011 + }, + { + "epoch": 0.4957532778098458, + "grad_norm": 1.6994843482971191, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7457939982414246, + "num_tokens": 71030697.0, + "step": 6012 + }, + { + "epoch": 0.4958357384348973, + "grad_norm": 1.9308604001998901, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7757536172866821, + "num_tokens": 71060479.0, + "step": 6013 + }, + { + "epoch": 0.4959181990599489, + "grad_norm": 1.8329496383666992, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7384645342826843, + "num_tokens": 71094147.0, + "step": 6014 + }, + { + "epoch": 0.4960006596850004, + "grad_norm": 1.9148465394973755, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7473537921905518, + "num_tokens": 71124763.0, + "step": 6015 + }, + { + "epoch": 0.496083120310052, + "grad_norm": 2.3084161281585693, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7125148773193359, + "num_tokens": 71149403.0, + "step": 6016 + }, + { + "epoch": 0.4961655809351035, + "grad_norm": 1.9775067567825317, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7331324815750122, + "num_tokens": 71181569.0, + "step": 6017 + }, + { + "epoch": 0.496248041560155, + "grad_norm": 2.2738730907440186, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6845394372940063, + "num_tokens": 71206459.0, + "step": 6018 + }, + { + "epoch": 0.4963305021852066, + "grad_norm": 2.0097596645355225, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7129385471343994, + "num_tokens": 71236390.0, + "step": 6019 + }, + { + "epoch": 0.4964129628102581, + "grad_norm": 1.986938238143921, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7247382402420044, + "num_tokens": 71265876.0, + "step": 6020 + }, + { + "epoch": 0.4964954234353096, + "grad_norm": 1.8775594234466553, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7293819189071655, + "num_tokens": 71297778.0, + "step": 6021 + }, + { + "epoch": 0.4965778840603612, + "grad_norm": 1.8443464040756226, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7242487668991089, + "num_tokens": 71335351.0, + "step": 6022 + }, + { + "epoch": 0.4966603446854127, + "grad_norm": 1.7846364974975586, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7649912238121033, + "num_tokens": 71371081.0, + "step": 6023 + }, + { + "epoch": 0.4967428053104643, + "grad_norm": 1.9349651336669922, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7439910173416138, + "num_tokens": 71400669.0, + "step": 6024 + }, + { + "epoch": 0.4968252659355158, + "grad_norm": 1.6966501474380493, + "learning_rate": 1e-06, + "loss": 0.7465, + "mean_token_accuracy": 0.7600499391555786, + "num_tokens": 71438690.0, + "step": 6025 + }, + { + "epoch": 0.4969077265605673, + "grad_norm": 1.862173318862915, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7294594645500183, + "num_tokens": 71472107.0, + "step": 6026 + }, + { + "epoch": 0.4969901871856189, + "grad_norm": 1.9167624711990356, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7742125988006592, + "num_tokens": 71503300.0, + "step": 6027 + }, + { + "epoch": 0.4970726478106704, + "grad_norm": 1.9247866868972778, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6929009556770325, + "num_tokens": 71534421.0, + "step": 6028 + }, + { + "epoch": 0.4971551084357219, + "grad_norm": 2.2030375003814697, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.7445038557052612, + "num_tokens": 71558423.0, + "step": 6029 + }, + { + "epoch": 0.4972375690607735, + "grad_norm": 2.0475430488586426, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7140816450119019, + "num_tokens": 71590866.0, + "step": 6030 + }, + { + "epoch": 0.497320029685825, + "grad_norm": 2.2549631595611572, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7774577140808105, + "num_tokens": 71618136.0, + "step": 6031 + }, + { + "epoch": 0.4974024903108766, + "grad_norm": 2.2677717208862305, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7084966897964478, + "num_tokens": 71644161.0, + "step": 6032 + }, + { + "epoch": 0.4974849509359281, + "grad_norm": 2.06990647315979, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7496556043624878, + "num_tokens": 71671985.0, + "step": 6033 + }, + { + "epoch": 0.4975674115609796, + "grad_norm": 2.014451503753662, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7163115739822388, + "num_tokens": 71699947.0, + "step": 6034 + }, + { + "epoch": 0.4976498721860312, + "grad_norm": 1.9337588548660278, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.722028374671936, + "num_tokens": 71731507.0, + "step": 6035 + }, + { + "epoch": 0.4977323328110827, + "grad_norm": 1.9365298748016357, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7619469165802002, + "num_tokens": 71761156.0, + "step": 6036 + }, + { + "epoch": 0.4978147934361343, + "grad_norm": 2.0047385692596436, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7314054369926453, + "num_tokens": 71789278.0, + "step": 6037 + }, + { + "epoch": 0.4978972540611858, + "grad_norm": 2.292189359664917, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7014038562774658, + "num_tokens": 71814169.0, + "step": 6038 + }, + { + "epoch": 0.4979797146862373, + "grad_norm": 1.7118672132492065, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.776188850402832, + "num_tokens": 71849862.0, + "step": 6039 + }, + { + "epoch": 0.4980621753112889, + "grad_norm": 1.9194550514221191, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7253681421279907, + "num_tokens": 71883303.0, + "step": 6040 + }, + { + "epoch": 0.4981446359363404, + "grad_norm": 1.984760046005249, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.733857274055481, + "num_tokens": 71913614.0, + "step": 6041 + }, + { + "epoch": 0.4982270965613919, + "grad_norm": 2.136737823486328, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7382227778434753, + "num_tokens": 71943487.0, + "step": 6042 + }, + { + "epoch": 0.4983095571864435, + "grad_norm": 2.159480571746826, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7316661477088928, + "num_tokens": 71971974.0, + "step": 6043 + }, + { + "epoch": 0.498392017811495, + "grad_norm": 2.134923219680786, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.730689525604248, + "num_tokens": 71998281.0, + "step": 6044 + }, + { + "epoch": 0.4984744784365466, + "grad_norm": 2.1346752643585205, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7203984260559082, + "num_tokens": 72025658.0, + "step": 6045 + }, + { + "epoch": 0.4985569390615981, + "grad_norm": 1.947917103767395, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7864788174629211, + "num_tokens": 72053269.0, + "step": 6046 + }, + { + "epoch": 0.4986393996866496, + "grad_norm": 1.7998526096343994, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.7965505719184875, + "num_tokens": 72085682.0, + "step": 6047 + }, + { + "epoch": 0.4987218603117012, + "grad_norm": 1.693270206451416, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.741835355758667, + "num_tokens": 72124011.0, + "step": 6048 + }, + { + "epoch": 0.4988043209367527, + "grad_norm": 1.661623477935791, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7711498737335205, + "num_tokens": 72161988.0, + "step": 6049 + }, + { + "epoch": 0.4988867815618042, + "grad_norm": 1.748412013053894, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7454094886779785, + "num_tokens": 72198989.0, + "step": 6050 + }, + { + "epoch": 0.4989692421868558, + "grad_norm": 1.785672903060913, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.764026403427124, + "num_tokens": 72232681.0, + "step": 6051 + }, + { + "epoch": 0.4990517028119073, + "grad_norm": 1.7760761976242065, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7679190039634705, + "num_tokens": 72268533.0, + "step": 6052 + }, + { + "epoch": 0.4991341634369589, + "grad_norm": 1.8535221815109253, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.7819181680679321, + "num_tokens": 72299860.0, + "step": 6053 + }, + { + "epoch": 0.4992166240620104, + "grad_norm": 2.2491166591644287, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7231366634368896, + "num_tokens": 72325028.0, + "step": 6054 + }, + { + "epoch": 0.4992990846870619, + "grad_norm": 1.6427894830703735, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7806209921836853, + "num_tokens": 72359816.0, + "step": 6055 + }, + { + "epoch": 0.4993815453121135, + "grad_norm": 1.8370643854141235, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7611743211746216, + "num_tokens": 72391542.0, + "step": 6056 + }, + { + "epoch": 0.499464005937165, + "grad_norm": 1.7617145776748657, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7874809503555298, + "num_tokens": 72426340.0, + "step": 6057 + }, + { + "epoch": 0.4995464665622165, + "grad_norm": 2.0540385246276855, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.714561939239502, + "num_tokens": 72454865.0, + "step": 6058 + }, + { + "epoch": 0.4996289271872681, + "grad_norm": 2.016317844390869, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.733208179473877, + "num_tokens": 72482793.0, + "step": 6059 + }, + { + "epoch": 0.4997113878123196, + "grad_norm": 1.9339536428451538, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7480655908584595, + "num_tokens": 72517046.0, + "step": 6060 + }, + { + "epoch": 0.4997938484373712, + "grad_norm": 2.274383068084717, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7317244410514832, + "num_tokens": 72539832.0, + "step": 6061 + }, + { + "epoch": 0.4998763090624227, + "grad_norm": 1.8783619403839111, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.6995681524276733, + "num_tokens": 72572505.0, + "step": 6062 + }, + { + "epoch": 0.4999587696874742, + "grad_norm": 2.026749610900879, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.79176265001297, + "num_tokens": 72601919.0, + "step": 6063 + }, + { + "epoch": 0.5000412303125258, + "grad_norm": 2.1116387844085693, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.751304030418396, + "num_tokens": 72626891.0, + "step": 6064 + }, + { + "epoch": 0.5001236909375774, + "grad_norm": 1.9376369714736938, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7754606008529663, + "num_tokens": 72655419.0, + "step": 6065 + }, + { + "epoch": 0.5002061515626288, + "grad_norm": 1.8871634006500244, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7409863471984863, + "num_tokens": 72685635.0, + "step": 6066 + }, + { + "epoch": 0.5002886121876804, + "grad_norm": 2.0022623538970947, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7384421825408936, + "num_tokens": 72714050.0, + "step": 6067 + }, + { + "epoch": 0.500371072812732, + "grad_norm": 2.066605567932129, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7539294958114624, + "num_tokens": 72738560.0, + "step": 6068 + }, + { + "epoch": 0.5004535334377834, + "grad_norm": 2.1842703819274902, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7461300492286682, + "num_tokens": 72762589.0, + "step": 6069 + }, + { + "epoch": 0.500535994062835, + "grad_norm": 1.833537220954895, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7474331855773926, + "num_tokens": 72795775.0, + "step": 6070 + }, + { + "epoch": 0.5006184546878866, + "grad_norm": 2.0432991981506348, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7547154426574707, + "num_tokens": 72821601.0, + "step": 6071 + }, + { + "epoch": 0.500700915312938, + "grad_norm": 2.1162357330322266, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7375616431236267, + "num_tokens": 72854294.0, + "step": 6072 + }, + { + "epoch": 0.5007833759379896, + "grad_norm": 1.829269528388977, + "learning_rate": 1e-06, + "loss": 0.6203, + "mean_token_accuracy": 0.7980414032936096, + "num_tokens": 72887954.0, + "step": 6073 + }, + { + "epoch": 0.5008658365630412, + "grad_norm": 1.947442650794983, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7764989733695984, + "num_tokens": 72915411.0, + "step": 6074 + }, + { + "epoch": 0.5009482971880926, + "grad_norm": 1.9323937892913818, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.7846791744232178, + "num_tokens": 72942035.0, + "step": 6075 + }, + { + "epoch": 0.5010307578131442, + "grad_norm": 2.138321876525879, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7653923630714417, + "num_tokens": 72965039.0, + "step": 6076 + }, + { + "epoch": 0.5011132184381958, + "grad_norm": 1.9821412563323975, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7695221900939941, + "num_tokens": 72992265.0, + "step": 6077 + }, + { + "epoch": 0.5011956790632472, + "grad_norm": 1.8355621099472046, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.7565140724182129, + "num_tokens": 73028831.0, + "step": 6078 + }, + { + "epoch": 0.5012781396882988, + "grad_norm": 1.773155689239502, + "learning_rate": 1e-06, + "loss": 0.6252, + "mean_token_accuracy": 0.7985920310020447, + "num_tokens": 73060314.0, + "step": 6079 + }, + { + "epoch": 0.5013606003133504, + "grad_norm": 2.028998851776123, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7374511957168579, + "num_tokens": 73087586.0, + "step": 6080 + }, + { + "epoch": 0.501443060938402, + "grad_norm": 1.9086278676986694, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7517918944358826, + "num_tokens": 73116640.0, + "step": 6081 + }, + { + "epoch": 0.5015255215634534, + "grad_norm": 1.9278398752212524, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.7979059219360352, + "num_tokens": 73147151.0, + "step": 6082 + }, + { + "epoch": 0.501607982188505, + "grad_norm": 2.017152786254883, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7337824106216431, + "num_tokens": 73175202.0, + "step": 6083 + }, + { + "epoch": 0.5016904428135566, + "grad_norm": 1.9693602323532104, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7213853001594543, + "num_tokens": 73206194.0, + "step": 6084 + }, + { + "epoch": 0.501772903438608, + "grad_norm": 1.9036145210266113, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7343605756759644, + "num_tokens": 73237327.0, + "step": 6085 + }, + { + "epoch": 0.5018553640636596, + "grad_norm": 2.2870616912841797, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7294753193855286, + "num_tokens": 73260150.0, + "step": 6086 + }, + { + "epoch": 0.5019378246887112, + "grad_norm": 2.198491096496582, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7092616558074951, + "num_tokens": 73290543.0, + "step": 6087 + }, + { + "epoch": 0.5020202853137626, + "grad_norm": 2.097846269607544, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7756121158599854, + "num_tokens": 73318677.0, + "step": 6088 + }, + { + "epoch": 0.5021027459388142, + "grad_norm": 2.1760237216949463, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7444416284561157, + "num_tokens": 73344090.0, + "step": 6089 + }, + { + "epoch": 0.5021852065638658, + "grad_norm": 1.9006696939468384, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7458467483520508, + "num_tokens": 73374990.0, + "step": 6090 + }, + { + "epoch": 0.5022676671889172, + "grad_norm": 1.8540371656417847, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7757824659347534, + "num_tokens": 73403717.0, + "step": 6091 + }, + { + "epoch": 0.5023501278139688, + "grad_norm": 1.8620649576187134, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7442489266395569, + "num_tokens": 73434798.0, + "step": 6092 + }, + { + "epoch": 0.5024325884390204, + "grad_norm": 1.913874864578247, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7427113056182861, + "num_tokens": 73465023.0, + "step": 6093 + }, + { + "epoch": 0.5025150490640719, + "grad_norm": 2.275442600250244, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7378149032592773, + "num_tokens": 73488742.0, + "step": 6094 + }, + { + "epoch": 0.5025975096891234, + "grad_norm": 1.9288274049758911, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7748187780380249, + "num_tokens": 73520007.0, + "step": 6095 + }, + { + "epoch": 0.502679970314175, + "grad_norm": 2.1452796459198, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7444616556167603, + "num_tokens": 73544622.0, + "step": 6096 + }, + { + "epoch": 0.5027624309392266, + "grad_norm": 1.720741868019104, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7302800416946411, + "num_tokens": 73582700.0, + "step": 6097 + }, + { + "epoch": 0.502844891564278, + "grad_norm": 1.929781436920166, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7365030646324158, + "num_tokens": 73614221.0, + "step": 6098 + }, + { + "epoch": 0.5029273521893296, + "grad_norm": 1.9731231927871704, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7385649085044861, + "num_tokens": 73649246.0, + "step": 6099 + }, + { + "epoch": 0.5030098128143812, + "grad_norm": 1.9736180305480957, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7550560235977173, + "num_tokens": 73677428.0, + "step": 6100 + }, + { + "epoch": 0.5030922734394326, + "grad_norm": 1.7287663221359253, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7459617257118225, + "num_tokens": 73715924.0, + "step": 6101 + }, + { + "epoch": 0.5031747340644842, + "grad_norm": 2.2120344638824463, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7472274303436279, + "num_tokens": 73747535.0, + "step": 6102 + }, + { + "epoch": 0.5032571946895358, + "grad_norm": 2.0897722244262695, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7252600193023682, + "num_tokens": 73776238.0, + "step": 6103 + }, + { + "epoch": 0.5033396553145872, + "grad_norm": 2.1005241870880127, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7342675924301147, + "num_tokens": 73802831.0, + "step": 6104 + }, + { + "epoch": 0.5034221159396388, + "grad_norm": 2.409759521484375, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7221653461456299, + "num_tokens": 73822916.0, + "step": 6105 + }, + { + "epoch": 0.5035045765646904, + "grad_norm": 2.061898946762085, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.763221263885498, + "num_tokens": 73849961.0, + "step": 6106 + }, + { + "epoch": 0.5035870371897418, + "grad_norm": 1.9674993753433228, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7631234526634216, + "num_tokens": 73880788.0, + "step": 6107 + }, + { + "epoch": 0.5036694978147934, + "grad_norm": 2.10080623626709, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7582491636276245, + "num_tokens": 73907859.0, + "step": 6108 + }, + { + "epoch": 0.503751958439845, + "grad_norm": 2.274566173553467, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7399860620498657, + "num_tokens": 73933202.0, + "step": 6109 + }, + { + "epoch": 0.5038344190648966, + "grad_norm": 1.9200024604797363, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7606427669525146, + "num_tokens": 73963972.0, + "step": 6110 + }, + { + "epoch": 0.503916879689948, + "grad_norm": 2.0147130489349365, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7275158762931824, + "num_tokens": 73993418.0, + "step": 6111 + }, + { + "epoch": 0.5039993403149996, + "grad_norm": 1.787607192993164, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7456351518630981, + "num_tokens": 74031022.0, + "step": 6112 + }, + { + "epoch": 0.5040818009400512, + "grad_norm": 2.1211469173431396, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7173950672149658, + "num_tokens": 74058323.0, + "step": 6113 + }, + { + "epoch": 0.5041642615651026, + "grad_norm": 1.8496359586715698, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7161344885826111, + "num_tokens": 74095343.0, + "step": 6114 + }, + { + "epoch": 0.5042467221901542, + "grad_norm": 1.9114757776260376, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7518366575241089, + "num_tokens": 74127233.0, + "step": 6115 + }, + { + "epoch": 0.5043291828152058, + "grad_norm": 2.2530903816223145, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7576775550842285, + "num_tokens": 74150326.0, + "step": 6116 + }, + { + "epoch": 0.5044116434402572, + "grad_norm": 2.166747570037842, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7314649820327759, + "num_tokens": 74178768.0, + "step": 6117 + }, + { + "epoch": 0.5044941040653088, + "grad_norm": 2.257014274597168, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7487924098968506, + "num_tokens": 74201171.0, + "step": 6118 + }, + { + "epoch": 0.5045765646903604, + "grad_norm": 1.9088873863220215, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7449882626533508, + "num_tokens": 74232585.0, + "step": 6119 + }, + { + "epoch": 0.5046590253154118, + "grad_norm": 1.7887349128723145, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7590636014938354, + "num_tokens": 74267154.0, + "step": 6120 + }, + { + "epoch": 0.5047414859404634, + "grad_norm": 2.346196174621582, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7157605290412903, + "num_tokens": 74290596.0, + "step": 6121 + }, + { + "epoch": 0.504823946565515, + "grad_norm": 1.8270598649978638, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7584431767463684, + "num_tokens": 74323776.0, + "step": 6122 + }, + { + "epoch": 0.5049064071905665, + "grad_norm": 1.8763734102249146, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7573012113571167, + "num_tokens": 74354095.0, + "step": 6123 + }, + { + "epoch": 0.504988867815618, + "grad_norm": 1.8461812734603882, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7624526023864746, + "num_tokens": 74387468.0, + "step": 6124 + }, + { + "epoch": 0.5050713284406696, + "grad_norm": 1.8761224746704102, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7508577704429626, + "num_tokens": 74420758.0, + "step": 6125 + }, + { + "epoch": 0.5051537890657212, + "grad_norm": 1.8521796464920044, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7610049247741699, + "num_tokens": 74452622.0, + "step": 6126 + }, + { + "epoch": 0.5052362496907726, + "grad_norm": 2.0671896934509277, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7514858245849609, + "num_tokens": 74481126.0, + "step": 6127 + }, + { + "epoch": 0.5053187103158242, + "grad_norm": 1.9735442399978638, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7111918926239014, + "num_tokens": 74512680.0, + "step": 6128 + }, + { + "epoch": 0.5054011709408758, + "grad_norm": 1.8547248840332031, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7710205912590027, + "num_tokens": 74546871.0, + "step": 6129 + }, + { + "epoch": 0.5054836315659272, + "grad_norm": 1.9138288497924805, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.747846245765686, + "num_tokens": 74580068.0, + "step": 6130 + }, + { + "epoch": 0.5055660921909788, + "grad_norm": 1.9224765300750732, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7204918265342712, + "num_tokens": 74611985.0, + "step": 6131 + }, + { + "epoch": 0.5056485528160304, + "grad_norm": 1.9605990648269653, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7476784586906433, + "num_tokens": 74641952.0, + "step": 6132 + }, + { + "epoch": 0.5057310134410818, + "grad_norm": 2.1041977405548096, + "learning_rate": 1e-06, + "loss": 0.7396, + "mean_token_accuracy": 0.7619775533676147, + "num_tokens": 74669873.0, + "step": 6133 + }, + { + "epoch": 0.5058134740661334, + "grad_norm": 1.9965524673461914, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7417751550674438, + "num_tokens": 74700373.0, + "step": 6134 + }, + { + "epoch": 0.505895934691185, + "grad_norm": 2.317518711090088, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7076281905174255, + "num_tokens": 74724438.0, + "step": 6135 + }, + { + "epoch": 0.5059783953162365, + "grad_norm": 1.8463140726089478, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7577047944068909, + "num_tokens": 74759987.0, + "step": 6136 + }, + { + "epoch": 0.506060855941288, + "grad_norm": 1.876572608947754, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7671599984169006, + "num_tokens": 74788271.0, + "step": 6137 + }, + { + "epoch": 0.5061433165663396, + "grad_norm": 2.168447732925415, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7457570433616638, + "num_tokens": 74814330.0, + "step": 6138 + }, + { + "epoch": 0.5062257771913911, + "grad_norm": 1.9746524095535278, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7434450387954712, + "num_tokens": 74841989.0, + "step": 6139 + }, + { + "epoch": 0.5063082378164426, + "grad_norm": 2.002717971801758, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7267444133758545, + "num_tokens": 74872756.0, + "step": 6140 + }, + { + "epoch": 0.5063906984414942, + "grad_norm": 2.16546630859375, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7300636768341064, + "num_tokens": 74900646.0, + "step": 6141 + }, + { + "epoch": 0.5064731590665458, + "grad_norm": 1.9262564182281494, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7250810861587524, + "num_tokens": 74933327.0, + "step": 6142 + }, + { + "epoch": 0.5065556196915972, + "grad_norm": 2.1802000999450684, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.6979395151138306, + "num_tokens": 74959430.0, + "step": 6143 + }, + { + "epoch": 0.5066380803166488, + "grad_norm": 2.099320650100708, + "learning_rate": 1e-06, + "loss": 0.7614, + "mean_token_accuracy": 0.7518450021743774, + "num_tokens": 74986268.0, + "step": 6144 + }, + { + "epoch": 0.5067205409417004, + "grad_norm": 1.7465829849243164, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7791330218315125, + "num_tokens": 75020132.0, + "step": 6145 + }, + { + "epoch": 0.5068030015667518, + "grad_norm": 2.183425188064575, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7427187561988831, + "num_tokens": 75046108.0, + "step": 6146 + }, + { + "epoch": 0.5068854621918034, + "grad_norm": 2.273298740386963, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7428245544433594, + "num_tokens": 75069641.0, + "step": 6147 + }, + { + "epoch": 0.506967922816855, + "grad_norm": 1.9467058181762695, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7671328783035278, + "num_tokens": 75099449.0, + "step": 6148 + }, + { + "epoch": 0.5070503834419064, + "grad_norm": 1.8602626323699951, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7681555151939392, + "num_tokens": 75130661.0, + "step": 6149 + }, + { + "epoch": 0.507132844066958, + "grad_norm": 1.9143261909484863, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7744312286376953, + "num_tokens": 75160861.0, + "step": 6150 + }, + { + "epoch": 0.5072153046920096, + "grad_norm": 1.857272982597351, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7453818917274475, + "num_tokens": 75194622.0, + "step": 6151 + }, + { + "epoch": 0.5072977653170611, + "grad_norm": 1.9657143354415894, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7164345979690552, + "num_tokens": 75224688.0, + "step": 6152 + }, + { + "epoch": 0.5073802259421126, + "grad_norm": 1.9129798412322998, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7537878155708313, + "num_tokens": 75253847.0, + "step": 6153 + }, + { + "epoch": 0.5074626865671642, + "grad_norm": 2.321208953857422, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7410484552383423, + "num_tokens": 75278241.0, + "step": 6154 + }, + { + "epoch": 0.5075451471922157, + "grad_norm": 1.9213461875915527, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7727593779563904, + "num_tokens": 75306386.0, + "step": 6155 + }, + { + "epoch": 0.5076276078172672, + "grad_norm": 2.0057730674743652, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7783598899841309, + "num_tokens": 75335813.0, + "step": 6156 + }, + { + "epoch": 0.5077100684423188, + "grad_norm": 1.8981939554214478, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7765560150146484, + "num_tokens": 75366781.0, + "step": 6157 + }, + { + "epoch": 0.5077925290673704, + "grad_norm": 1.935124397277832, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.781894862651825, + "num_tokens": 75393543.0, + "step": 6158 + }, + { + "epoch": 0.5078749896924218, + "grad_norm": 2.162803888320923, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7600054144859314, + "num_tokens": 75416746.0, + "step": 6159 + }, + { + "epoch": 0.5079574503174734, + "grad_norm": 1.99228835105896, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7512629628181458, + "num_tokens": 75445009.0, + "step": 6160 + }, + { + "epoch": 0.508039910942525, + "grad_norm": 2.0792746543884277, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7482893466949463, + "num_tokens": 75471570.0, + "step": 6161 + }, + { + "epoch": 0.5081223715675764, + "grad_norm": 1.8423923254013062, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7616755366325378, + "num_tokens": 75503443.0, + "step": 6162 + }, + { + "epoch": 0.508204832192628, + "grad_norm": 1.8828725814819336, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7683228850364685, + "num_tokens": 75533717.0, + "step": 6163 + }, + { + "epoch": 0.5082872928176796, + "grad_norm": 1.940950870513916, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7399694919586182, + "num_tokens": 75567058.0, + "step": 6164 + }, + { + "epoch": 0.508369753442731, + "grad_norm": 1.9034972190856934, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.74712735414505, + "num_tokens": 75599073.0, + "step": 6165 + }, + { + "epoch": 0.5084522140677826, + "grad_norm": 1.9206136465072632, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7453474998474121, + "num_tokens": 75628236.0, + "step": 6166 + }, + { + "epoch": 0.5085346746928342, + "grad_norm": 1.7795976400375366, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.766476571559906, + "num_tokens": 75661998.0, + "step": 6167 + }, + { + "epoch": 0.5086171353178857, + "grad_norm": 1.8289812803268433, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7429128289222717, + "num_tokens": 75697450.0, + "step": 6168 + }, + { + "epoch": 0.5086995959429372, + "grad_norm": 2.324880599975586, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6912832260131836, + "num_tokens": 75721291.0, + "step": 6169 + }, + { + "epoch": 0.5087820565679888, + "grad_norm": 1.7397984266281128, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7781908512115479, + "num_tokens": 75757268.0, + "step": 6170 + }, + { + "epoch": 0.5088645171930404, + "grad_norm": 2.223567247390747, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.704186201095581, + "num_tokens": 75782634.0, + "step": 6171 + }, + { + "epoch": 0.5089469778180918, + "grad_norm": 1.8812819719314575, + "learning_rate": 1e-06, + "loss": 0.6189, + "mean_token_accuracy": 0.8025482296943665, + "num_tokens": 75811309.0, + "step": 6172 + }, + { + "epoch": 0.5090294384431434, + "grad_norm": 1.9683423042297363, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7358694076538086, + "num_tokens": 75841350.0, + "step": 6173 + }, + { + "epoch": 0.509111899068195, + "grad_norm": 1.7041703462600708, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7683414220809937, + "num_tokens": 75878467.0, + "step": 6174 + }, + { + "epoch": 0.5091943596932464, + "grad_norm": 1.8228858709335327, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7662156820297241, + "num_tokens": 75913004.0, + "step": 6175 + }, + { + "epoch": 0.509276820318298, + "grad_norm": 1.8828926086425781, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7514945268630981, + "num_tokens": 75944996.0, + "step": 6176 + }, + { + "epoch": 0.5093592809433496, + "grad_norm": 1.956868290901184, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7404101490974426, + "num_tokens": 75974605.0, + "step": 6177 + }, + { + "epoch": 0.509441741568401, + "grad_norm": 2.0447747707366943, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7134905457496643, + "num_tokens": 76004918.0, + "step": 6178 + }, + { + "epoch": 0.5095242021934526, + "grad_norm": 1.6680923700332642, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7716295719146729, + "num_tokens": 76043958.0, + "step": 6179 + }, + { + "epoch": 0.5096066628185042, + "grad_norm": 2.2621984481811523, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7185187339782715, + "num_tokens": 76068024.0, + "step": 6180 + }, + { + "epoch": 0.5096891234435557, + "grad_norm": 1.890442132949829, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7848767638206482, + "num_tokens": 76102266.0, + "step": 6181 + }, + { + "epoch": 0.5097715840686072, + "grad_norm": 2.4290592670440674, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7176876664161682, + "num_tokens": 76123459.0, + "step": 6182 + }, + { + "epoch": 0.5098540446936588, + "grad_norm": 1.9437973499298096, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.737105131149292, + "num_tokens": 76155630.0, + "step": 6183 + }, + { + "epoch": 0.5099365053187103, + "grad_norm": 1.6651593446731567, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7517476081848145, + "num_tokens": 76200234.0, + "step": 6184 + }, + { + "epoch": 0.5100189659437618, + "grad_norm": 1.971292495727539, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7364901900291443, + "num_tokens": 76231364.0, + "step": 6185 + }, + { + "epoch": 0.5101014265688134, + "grad_norm": 2.026402235031128, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7744340896606445, + "num_tokens": 76259443.0, + "step": 6186 + }, + { + "epoch": 0.510183887193865, + "grad_norm": 2.4463422298431396, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7338802814483643, + "num_tokens": 76281214.0, + "step": 6187 + }, + { + "epoch": 0.5102663478189164, + "grad_norm": 2.1335179805755615, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7870455384254456, + "num_tokens": 76307297.0, + "step": 6188 + }, + { + "epoch": 0.510348808443968, + "grad_norm": 2.206538438796997, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7165158987045288, + "num_tokens": 76334408.0, + "step": 6189 + }, + { + "epoch": 0.5104312690690196, + "grad_norm": 1.9098128080368042, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7623525857925415, + "num_tokens": 76365848.0, + "step": 6190 + }, + { + "epoch": 0.510513729694071, + "grad_norm": 1.9535143375396729, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7299103736877441, + "num_tokens": 76395913.0, + "step": 6191 + }, + { + "epoch": 0.5105961903191226, + "grad_norm": 2.1451871395111084, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7240093946456909, + "num_tokens": 76422759.0, + "step": 6192 + }, + { + "epoch": 0.5106786509441742, + "grad_norm": 1.9056401252746582, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7370433807373047, + "num_tokens": 76454440.0, + "step": 6193 + }, + { + "epoch": 0.5107611115692257, + "grad_norm": 1.9308077096939087, + "learning_rate": 1e-06, + "loss": 0.7303, + "mean_token_accuracy": 0.7739889025688171, + "num_tokens": 76483738.0, + "step": 6194 + }, + { + "epoch": 0.5108435721942772, + "grad_norm": 1.90501070022583, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7254536151885986, + "num_tokens": 76516160.0, + "step": 6195 + }, + { + "epoch": 0.5109260328193288, + "grad_norm": 2.0778799057006836, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7300053834915161, + "num_tokens": 76542515.0, + "step": 6196 + }, + { + "epoch": 0.5110084934443803, + "grad_norm": 2.232534170150757, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.746941328048706, + "num_tokens": 76569620.0, + "step": 6197 + }, + { + "epoch": 0.5110909540694318, + "grad_norm": 1.7046383619308472, + "learning_rate": 1e-06, + "loss": 0.786, + "mean_token_accuracy": 0.7471939921379089, + "num_tokens": 76607923.0, + "step": 6198 + }, + { + "epoch": 0.5111734146944834, + "grad_norm": 2.01566219329834, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7819549441337585, + "num_tokens": 76636517.0, + "step": 6199 + }, + { + "epoch": 0.5112558753195349, + "grad_norm": 1.9778962135314941, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7046869993209839, + "num_tokens": 76670797.0, + "step": 6200 + }, + { + "epoch": 0.5113383359445864, + "grad_norm": 2.0205092430114746, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7764496803283691, + "num_tokens": 76694706.0, + "step": 6201 + }, + { + "epoch": 0.511420796569638, + "grad_norm": 2.0553948879241943, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7614772319793701, + "num_tokens": 76719608.0, + "step": 6202 + }, + { + "epoch": 0.5115032571946896, + "grad_norm": 2.0305962562561035, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7361973524093628, + "num_tokens": 76749418.0, + "step": 6203 + }, + { + "epoch": 0.511585717819741, + "grad_norm": 1.9624947309494019, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7546883821487427, + "num_tokens": 76780870.0, + "step": 6204 + }, + { + "epoch": 0.5116681784447926, + "grad_norm": 2.247767686843872, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.7888816595077515, + "num_tokens": 76806276.0, + "step": 6205 + }, + { + "epoch": 0.5117506390698442, + "grad_norm": 1.9463034868240356, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7759857177734375, + "num_tokens": 76837157.0, + "step": 6206 + }, + { + "epoch": 0.5118330996948957, + "grad_norm": 1.998347520828247, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.732675313949585, + "num_tokens": 76867170.0, + "step": 6207 + }, + { + "epoch": 0.5119155603199472, + "grad_norm": 1.8377193212509155, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.736322820186615, + "num_tokens": 76898522.0, + "step": 6208 + }, + { + "epoch": 0.5119980209449988, + "grad_norm": 2.010181188583374, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7313140034675598, + "num_tokens": 76931230.0, + "step": 6209 + }, + { + "epoch": 0.5120804815700503, + "grad_norm": 1.8457558155059814, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7785284519195557, + "num_tokens": 76961983.0, + "step": 6210 + }, + { + "epoch": 0.5121629421951018, + "grad_norm": 1.964434027671814, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7318755388259888, + "num_tokens": 76992775.0, + "step": 6211 + }, + { + "epoch": 0.5122454028201534, + "grad_norm": 1.843191385269165, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.754623293876648, + "num_tokens": 77028037.0, + "step": 6212 + }, + { + "epoch": 0.5123278634452049, + "grad_norm": 2.086184024810791, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7558377981185913, + "num_tokens": 77056991.0, + "step": 6213 + }, + { + "epoch": 0.5124103240702564, + "grad_norm": 1.8873522281646729, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7473598122596741, + "num_tokens": 77090927.0, + "step": 6214 + }, + { + "epoch": 0.512492784695308, + "grad_norm": 1.8569709062576294, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7156559228897095, + "num_tokens": 77124820.0, + "step": 6215 + }, + { + "epoch": 0.5125752453203595, + "grad_norm": 2.081265926361084, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7597342133522034, + "num_tokens": 77150322.0, + "step": 6216 + }, + { + "epoch": 0.512657705945411, + "grad_norm": 2.045959711074829, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7660170793533325, + "num_tokens": 77176496.0, + "step": 6217 + }, + { + "epoch": 0.5127401665704626, + "grad_norm": 1.9429012537002563, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7522616982460022, + "num_tokens": 77208716.0, + "step": 6218 + }, + { + "epoch": 0.5128226271955142, + "grad_norm": 1.8354068994522095, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.750761866569519, + "num_tokens": 77243261.0, + "step": 6219 + }, + { + "epoch": 0.5129050878205657, + "grad_norm": 1.797897458076477, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7867127656936646, + "num_tokens": 77276324.0, + "step": 6220 + }, + { + "epoch": 0.5129875484456172, + "grad_norm": 2.193986654281616, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7263628244400024, + "num_tokens": 77301163.0, + "step": 6221 + }, + { + "epoch": 0.5130700090706688, + "grad_norm": 1.8244338035583496, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7833315134048462, + "num_tokens": 77337339.0, + "step": 6222 + }, + { + "epoch": 0.5131524696957203, + "grad_norm": 2.087759494781494, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7102830410003662, + "num_tokens": 77365761.0, + "step": 6223 + }, + { + "epoch": 0.5132349303207718, + "grad_norm": 2.308584690093994, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7341204285621643, + "num_tokens": 77391183.0, + "step": 6224 + }, + { + "epoch": 0.5133173909458234, + "grad_norm": 2.2461297512054443, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.72735995054245, + "num_tokens": 77415855.0, + "step": 6225 + }, + { + "epoch": 0.5133998515708749, + "grad_norm": 2.225121021270752, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7481304407119751, + "num_tokens": 77439534.0, + "step": 6226 + }, + { + "epoch": 0.5134823121959264, + "grad_norm": 1.80342435836792, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7703878283500671, + "num_tokens": 77475293.0, + "step": 6227 + }, + { + "epoch": 0.513564772820978, + "grad_norm": 4.060338497161865, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7805421352386475, + "num_tokens": 77500362.0, + "step": 6228 + }, + { + "epoch": 0.5136472334460295, + "grad_norm": 2.1521146297454834, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7320213913917542, + "num_tokens": 77527789.0, + "step": 6229 + }, + { + "epoch": 0.513729694071081, + "grad_norm": 2.011467695236206, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7248102426528931, + "num_tokens": 77558653.0, + "step": 6230 + }, + { + "epoch": 0.5138121546961326, + "grad_norm": 1.9718096256256104, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7352414131164551, + "num_tokens": 77586432.0, + "step": 6231 + }, + { + "epoch": 0.5138946153211842, + "grad_norm": 2.3947176933288574, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7229124307632446, + "num_tokens": 77608374.0, + "step": 6232 + }, + { + "epoch": 0.5139770759462357, + "grad_norm": 1.9306315183639526, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7082833051681519, + "num_tokens": 77641641.0, + "step": 6233 + }, + { + "epoch": 0.5140595365712872, + "grad_norm": 2.030870199203491, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7350193858146667, + "num_tokens": 77669120.0, + "step": 6234 + }, + { + "epoch": 0.5141419971963388, + "grad_norm": 1.9229151010513306, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7333288192749023, + "num_tokens": 77699642.0, + "step": 6235 + }, + { + "epoch": 0.5142244578213903, + "grad_norm": 1.8487625122070312, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7586432695388794, + "num_tokens": 77732290.0, + "step": 6236 + }, + { + "epoch": 0.5143069184464418, + "grad_norm": 2.3267643451690674, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6868712902069092, + "num_tokens": 77756053.0, + "step": 6237 + }, + { + "epoch": 0.5143893790714934, + "grad_norm": 1.8701229095458984, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7646794319152832, + "num_tokens": 77786926.0, + "step": 6238 + }, + { + "epoch": 0.5144718396965449, + "grad_norm": 1.7498657703399658, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.7928675413131714, + "num_tokens": 77818916.0, + "step": 6239 + }, + { + "epoch": 0.5145543003215964, + "grad_norm": 2.270798921585083, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7398501634597778, + "num_tokens": 77842778.0, + "step": 6240 + }, + { + "epoch": 0.514636760946648, + "grad_norm": 1.825783133506775, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7568196058273315, + "num_tokens": 77875258.0, + "step": 6241 + }, + { + "epoch": 0.5147192215716995, + "grad_norm": 1.8301846981048584, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7570908665657043, + "num_tokens": 77906983.0, + "step": 6242 + }, + { + "epoch": 0.514801682196751, + "grad_norm": 2.250105381011963, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7241414785385132, + "num_tokens": 77933781.0, + "step": 6243 + }, + { + "epoch": 0.5148841428218026, + "grad_norm": 2.164273500442505, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7196931838989258, + "num_tokens": 77962323.0, + "step": 6244 + }, + { + "epoch": 0.5149666034468541, + "grad_norm": 2.028425455093384, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7336267828941345, + "num_tokens": 77992500.0, + "step": 6245 + }, + { + "epoch": 0.5150490640719056, + "grad_norm": 1.934815764427185, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7317308187484741, + "num_tokens": 78023293.0, + "step": 6246 + }, + { + "epoch": 0.5151315246969572, + "grad_norm": 1.9802582263946533, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7336334586143494, + "num_tokens": 78054363.0, + "step": 6247 + }, + { + "epoch": 0.5152139853220088, + "grad_norm": 1.7164849042892456, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.7950177192687988, + "num_tokens": 78088031.0, + "step": 6248 + }, + { + "epoch": 0.5152964459470603, + "grad_norm": 1.8247737884521484, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.768480122089386, + "num_tokens": 78120234.0, + "step": 6249 + }, + { + "epoch": 0.5153789065721118, + "grad_norm": 1.9735229015350342, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.7974753379821777, + "num_tokens": 78148859.0, + "step": 6250 + }, + { + "epoch": 0.5154613671971634, + "grad_norm": 2.0165464878082275, + "learning_rate": 1e-06, + "loss": 1.0706, + "mean_token_accuracy": 0.6793438196182251, + "num_tokens": 78180646.0, + "step": 6251 + }, + { + "epoch": 0.5155438278222149, + "grad_norm": 2.043574094772339, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6927650570869446, + "num_tokens": 78210381.0, + "step": 6252 + }, + { + "epoch": 0.5156262884472664, + "grad_norm": 2.071542739868164, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7215734720230103, + "num_tokens": 78239234.0, + "step": 6253 + }, + { + "epoch": 0.515708749072318, + "grad_norm": 2.119779586791992, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7283304929733276, + "num_tokens": 78267884.0, + "step": 6254 + }, + { + "epoch": 0.5157912096973695, + "grad_norm": 2.0502164363861084, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6888298988342285, + "num_tokens": 78297427.0, + "step": 6255 + }, + { + "epoch": 0.515873670322421, + "grad_norm": 2.0010368824005127, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.759418249130249, + "num_tokens": 78325015.0, + "step": 6256 + }, + { + "epoch": 0.5159561309474726, + "grad_norm": 2.071254014968872, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7319387793540955, + "num_tokens": 78353933.0, + "step": 6257 + }, + { + "epoch": 0.5160385915725241, + "grad_norm": 1.854626178741455, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7655830979347229, + "num_tokens": 78384598.0, + "step": 6258 + }, + { + "epoch": 0.5161210521975756, + "grad_norm": 2.0807836055755615, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7457427978515625, + "num_tokens": 78412463.0, + "step": 6259 + }, + { + "epoch": 0.5162035128226272, + "grad_norm": 2.1299755573272705, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7202831506729126, + "num_tokens": 78441573.0, + "step": 6260 + }, + { + "epoch": 0.5162859734476787, + "grad_norm": 2.1587343215942383, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7309315204620361, + "num_tokens": 78467146.0, + "step": 6261 + }, + { + "epoch": 0.5163684340727303, + "grad_norm": 2.0139012336730957, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7388702630996704, + "num_tokens": 78497043.0, + "step": 6262 + }, + { + "epoch": 0.5164508946977818, + "grad_norm": 2.071476697921753, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.755665123462677, + "num_tokens": 78521697.0, + "step": 6263 + }, + { + "epoch": 0.5165333553228334, + "grad_norm": 2.106915235519409, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7095144391059875, + "num_tokens": 78550998.0, + "step": 6264 + }, + { + "epoch": 0.5166158159478849, + "grad_norm": 1.9475303888320923, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.733291506767273, + "num_tokens": 78583997.0, + "step": 6265 + }, + { + "epoch": 0.5166982765729364, + "grad_norm": 1.849058985710144, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7455914616584778, + "num_tokens": 78619962.0, + "step": 6266 + }, + { + "epoch": 0.516780737197988, + "grad_norm": 1.9327422380447388, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7457984685897827, + "num_tokens": 78651323.0, + "step": 6267 + }, + { + "epoch": 0.5168631978230395, + "grad_norm": 2.0699527263641357, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7651441693305969, + "num_tokens": 78679376.0, + "step": 6268 + }, + { + "epoch": 0.516945658448091, + "grad_norm": 1.9679322242736816, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7655521631240845, + "num_tokens": 78708118.0, + "step": 6269 + }, + { + "epoch": 0.5170281190731426, + "grad_norm": 1.769905924797058, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7567542791366577, + "num_tokens": 78742515.0, + "step": 6270 + }, + { + "epoch": 0.5171105796981941, + "grad_norm": 2.0041086673736572, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7561454772949219, + "num_tokens": 78772412.0, + "step": 6271 + }, + { + "epoch": 0.5171930403232456, + "grad_norm": 2.0897932052612305, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.7811453342437744, + "num_tokens": 78798809.0, + "step": 6272 + }, + { + "epoch": 0.5172755009482972, + "grad_norm": 1.7775174379348755, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7421169877052307, + "num_tokens": 78833948.0, + "step": 6273 + }, + { + "epoch": 0.5173579615733487, + "grad_norm": 1.9082483053207397, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7766014933586121, + "num_tokens": 78864368.0, + "step": 6274 + }, + { + "epoch": 0.5174404221984003, + "grad_norm": 1.994828701019287, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7098974585533142, + "num_tokens": 78895140.0, + "step": 6275 + }, + { + "epoch": 0.5175228828234518, + "grad_norm": 2.119417190551758, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7591094970703125, + "num_tokens": 78921720.0, + "step": 6276 + }, + { + "epoch": 0.5176053434485033, + "grad_norm": 2.2538156509399414, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7131773829460144, + "num_tokens": 78945890.0, + "step": 6277 + }, + { + "epoch": 0.5176878040735549, + "grad_norm": 1.8305206298828125, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7269694805145264, + "num_tokens": 78980438.0, + "step": 6278 + }, + { + "epoch": 0.5177702646986064, + "grad_norm": 1.9405654668807983, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7428548336029053, + "num_tokens": 79011922.0, + "step": 6279 + }, + { + "epoch": 0.517852725323658, + "grad_norm": 2.0109057426452637, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7693215608596802, + "num_tokens": 79039649.0, + "step": 6280 + }, + { + "epoch": 0.5179351859487095, + "grad_norm": 1.8828213214874268, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7093492150306702, + "num_tokens": 79071285.0, + "step": 6281 + }, + { + "epoch": 0.518017646573761, + "grad_norm": 1.8247177600860596, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7421038150787354, + "num_tokens": 79105334.0, + "step": 6282 + }, + { + "epoch": 0.5181001071988126, + "grad_norm": 2.1138947010040283, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7308825850486755, + "num_tokens": 79132694.0, + "step": 6283 + }, + { + "epoch": 0.5181825678238641, + "grad_norm": 1.953781247138977, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7545714378356934, + "num_tokens": 79162897.0, + "step": 6284 + }, + { + "epoch": 0.5182650284489156, + "grad_norm": 2.2972569465637207, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7142397165298462, + "num_tokens": 79185443.0, + "step": 6285 + }, + { + "epoch": 0.5183474890739672, + "grad_norm": 1.9478282928466797, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7155637741088867, + "num_tokens": 79216686.0, + "step": 6286 + }, + { + "epoch": 0.5184299496990187, + "grad_norm": 2.0783474445343018, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7236372232437134, + "num_tokens": 79247613.0, + "step": 6287 + }, + { + "epoch": 0.5185124103240702, + "grad_norm": 1.8532583713531494, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7475534677505493, + "num_tokens": 79280259.0, + "step": 6288 + }, + { + "epoch": 0.5185948709491218, + "grad_norm": 2.0296123027801514, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7169491648674011, + "num_tokens": 79310591.0, + "step": 6289 + }, + { + "epoch": 0.5186773315741733, + "grad_norm": 1.9249111413955688, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7513483762741089, + "num_tokens": 79340969.0, + "step": 6290 + }, + { + "epoch": 0.5187597921992249, + "grad_norm": 1.9026377201080322, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7792420983314514, + "num_tokens": 79371027.0, + "step": 6291 + }, + { + "epoch": 0.5188422528242764, + "grad_norm": 2.166156768798828, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.7820247411727905, + "num_tokens": 79403688.0, + "step": 6292 + }, + { + "epoch": 0.518924713449328, + "grad_norm": 1.9778944253921509, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7474381923675537, + "num_tokens": 79432063.0, + "step": 6293 + }, + { + "epoch": 0.5190071740743795, + "grad_norm": 2.043008804321289, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7572941780090332, + "num_tokens": 79458388.0, + "step": 6294 + }, + { + "epoch": 0.519089634699431, + "grad_norm": 2.0408003330230713, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7553644776344299, + "num_tokens": 79485364.0, + "step": 6295 + }, + { + "epoch": 0.5191720953244826, + "grad_norm": 1.9362108707427979, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7532531023025513, + "num_tokens": 79515536.0, + "step": 6296 + }, + { + "epoch": 0.5192545559495341, + "grad_norm": 1.9430309534072876, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.747835636138916, + "num_tokens": 79545013.0, + "step": 6297 + }, + { + "epoch": 0.5193370165745856, + "grad_norm": 1.9674278497695923, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7739956378936768, + "num_tokens": 79575258.0, + "step": 6298 + }, + { + "epoch": 0.5194194771996372, + "grad_norm": 1.980142593383789, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7489920854568481, + "num_tokens": 79607402.0, + "step": 6299 + }, + { + "epoch": 0.5195019378246887, + "grad_norm": 2.2487385272979736, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.685026228427887, + "num_tokens": 79634707.0, + "step": 6300 + }, + { + "epoch": 0.5195843984497402, + "grad_norm": 2.2211196422576904, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7313527464866638, + "num_tokens": 79660289.0, + "step": 6301 + }, + { + "epoch": 0.5196668590747918, + "grad_norm": 1.9267126321792603, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7616695165634155, + "num_tokens": 79688419.0, + "step": 6302 + }, + { + "epoch": 0.5197493196998433, + "grad_norm": 1.8643345832824707, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7622637152671814, + "num_tokens": 79719970.0, + "step": 6303 + }, + { + "epoch": 0.5198317803248949, + "grad_norm": 1.9211605787277222, + "learning_rate": 1e-06, + "loss": 1.086, + "mean_token_accuracy": 0.6786710619926453, + "num_tokens": 79760346.0, + "step": 6304 + }, + { + "epoch": 0.5199142409499464, + "grad_norm": 2.1491055488586426, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7197607755661011, + "num_tokens": 79787755.0, + "step": 6305 + }, + { + "epoch": 0.5199967015749979, + "grad_norm": 1.8944356441497803, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7602876424789429, + "num_tokens": 79818359.0, + "step": 6306 + }, + { + "epoch": 0.5200791622000495, + "grad_norm": 1.9404215812683105, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.724340558052063, + "num_tokens": 79852728.0, + "step": 6307 + }, + { + "epoch": 0.520161622825101, + "grad_norm": 1.823826551437378, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.763655424118042, + "num_tokens": 79885633.0, + "step": 6308 + }, + { + "epoch": 0.5202440834501526, + "grad_norm": 2.0300698280334473, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7101702094078064, + "num_tokens": 79916599.0, + "step": 6309 + }, + { + "epoch": 0.5203265440752041, + "grad_norm": 2.068359375, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7442017197608948, + "num_tokens": 79945365.0, + "step": 6310 + }, + { + "epoch": 0.5204090047002556, + "grad_norm": 1.90785813331604, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7568856477737427, + "num_tokens": 79978175.0, + "step": 6311 + }, + { + "epoch": 0.5204914653253072, + "grad_norm": 2.163803815841675, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7573341131210327, + "num_tokens": 80003038.0, + "step": 6312 + }, + { + "epoch": 0.5205739259503587, + "grad_norm": 1.7523409128189087, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7536003589630127, + "num_tokens": 80038583.0, + "step": 6313 + }, + { + "epoch": 0.5206563865754102, + "grad_norm": 1.8660691976547241, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7671839594841003, + "num_tokens": 80069054.0, + "step": 6314 + }, + { + "epoch": 0.5207388472004618, + "grad_norm": 2.3072710037231445, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.6885727643966675, + "num_tokens": 80097232.0, + "step": 6315 + }, + { + "epoch": 0.5208213078255133, + "grad_norm": 2.086945056915283, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7867310047149658, + "num_tokens": 80124081.0, + "step": 6316 + }, + { + "epoch": 0.5209037684505649, + "grad_norm": 1.9266865253448486, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7757698893547058, + "num_tokens": 80154039.0, + "step": 6317 + }, + { + "epoch": 0.5209862290756164, + "grad_norm": 2.2128477096557617, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7261369228363037, + "num_tokens": 80181012.0, + "step": 6318 + }, + { + "epoch": 0.5210686897006679, + "grad_norm": 1.8027907609939575, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7720637917518616, + "num_tokens": 80213718.0, + "step": 6319 + }, + { + "epoch": 0.5211511503257195, + "grad_norm": 2.0266592502593994, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.758756160736084, + "num_tokens": 80240068.0, + "step": 6320 + }, + { + "epoch": 0.521233610950771, + "grad_norm": 1.8715332746505737, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7144297361373901, + "num_tokens": 80274870.0, + "step": 6321 + }, + { + "epoch": 0.5213160715758225, + "grad_norm": 1.826853632926941, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7639396786689758, + "num_tokens": 80306271.0, + "step": 6322 + }, + { + "epoch": 0.5213985322008741, + "grad_norm": 1.749253273010254, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7437435984611511, + "num_tokens": 80342630.0, + "step": 6323 + }, + { + "epoch": 0.5214809928259256, + "grad_norm": 1.6187236309051514, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7891483902931213, + "num_tokens": 80382819.0, + "step": 6324 + }, + { + "epoch": 0.5215634534509772, + "grad_norm": 1.843355417251587, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8007321357727051, + "num_tokens": 80417400.0, + "step": 6325 + }, + { + "epoch": 0.5216459140760287, + "grad_norm": 1.8148326873779297, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7441510558128357, + "num_tokens": 80451112.0, + "step": 6326 + }, + { + "epoch": 0.5217283747010802, + "grad_norm": 1.9310567378997803, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7472596764564514, + "num_tokens": 80482758.0, + "step": 6327 + }, + { + "epoch": 0.5218108353261318, + "grad_norm": 2.1267800331115723, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7046480774879456, + "num_tokens": 80509823.0, + "step": 6328 + }, + { + "epoch": 0.5218932959511833, + "grad_norm": 2.2248964309692383, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.739001989364624, + "num_tokens": 80534927.0, + "step": 6329 + }, + { + "epoch": 0.5219757565762349, + "grad_norm": 2.0796425342559814, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7377818822860718, + "num_tokens": 80564417.0, + "step": 6330 + }, + { + "epoch": 0.5220582172012864, + "grad_norm": 2.1855099201202393, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7179850935935974, + "num_tokens": 80592334.0, + "step": 6331 + }, + { + "epoch": 0.5221406778263379, + "grad_norm": 1.969911813735962, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7550984621047974, + "num_tokens": 80622386.0, + "step": 6332 + }, + { + "epoch": 0.5222231384513895, + "grad_norm": 1.8396574258804321, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7330251336097717, + "num_tokens": 80657832.0, + "step": 6333 + }, + { + "epoch": 0.522305599076441, + "grad_norm": 1.8945544958114624, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.75898677110672, + "num_tokens": 80691925.0, + "step": 6334 + }, + { + "epoch": 0.5223880597014925, + "grad_norm": 1.9671365022659302, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7650816440582275, + "num_tokens": 80720297.0, + "step": 6335 + }, + { + "epoch": 0.5224705203265441, + "grad_norm": 1.8782998323440552, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7346579432487488, + "num_tokens": 80753283.0, + "step": 6336 + }, + { + "epoch": 0.5225529809515956, + "grad_norm": 2.0617191791534424, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7774770259857178, + "num_tokens": 80780964.0, + "step": 6337 + }, + { + "epoch": 0.5226354415766471, + "grad_norm": 1.832276701927185, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.7530471682548523, + "num_tokens": 80815622.0, + "step": 6338 + }, + { + "epoch": 0.5227179022016987, + "grad_norm": 1.8977034091949463, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.7903663516044617, + "num_tokens": 80842726.0, + "step": 6339 + }, + { + "epoch": 0.5228003628267502, + "grad_norm": 2.2649686336517334, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7346235513687134, + "num_tokens": 80867920.0, + "step": 6340 + }, + { + "epoch": 0.5228828234518018, + "grad_norm": 1.813444972038269, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.753333330154419, + "num_tokens": 80904040.0, + "step": 6341 + }, + { + "epoch": 0.5229652840768533, + "grad_norm": 1.8707780838012695, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7690348029136658, + "num_tokens": 80934029.0, + "step": 6342 + }, + { + "epoch": 0.5230477447019048, + "grad_norm": 1.8499643802642822, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.723872721195221, + "num_tokens": 80968487.0, + "step": 6343 + }, + { + "epoch": 0.5231302053269564, + "grad_norm": 1.865301489830017, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.748072624206543, + "num_tokens": 81001517.0, + "step": 6344 + }, + { + "epoch": 0.5232126659520079, + "grad_norm": 1.9923176765441895, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7414535284042358, + "num_tokens": 81030186.0, + "step": 6345 + }, + { + "epoch": 0.5232951265770595, + "grad_norm": 1.7498944997787476, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.7876284122467041, + "num_tokens": 81061971.0, + "step": 6346 + }, + { + "epoch": 0.523377587202111, + "grad_norm": 1.965938925743103, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.746732234954834, + "num_tokens": 81093884.0, + "step": 6347 + }, + { + "epoch": 0.5234600478271625, + "grad_norm": 1.757299780845642, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7411375045776367, + "num_tokens": 81128950.0, + "step": 6348 + }, + { + "epoch": 0.5235425084522141, + "grad_norm": 1.8314731121063232, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7499114274978638, + "num_tokens": 81160641.0, + "step": 6349 + }, + { + "epoch": 0.5236249690772656, + "grad_norm": 1.938344120979309, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7501883506774902, + "num_tokens": 81193108.0, + "step": 6350 + }, + { + "epoch": 0.5237074297023171, + "grad_norm": 2.0436244010925293, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7138000726699829, + "num_tokens": 81221504.0, + "step": 6351 + }, + { + "epoch": 0.5237898903273687, + "grad_norm": 2.2836530208587646, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.7079384326934814, + "num_tokens": 81247464.0, + "step": 6352 + }, + { + "epoch": 0.5238723509524202, + "grad_norm": 1.8219196796417236, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7821537256240845, + "num_tokens": 81277538.0, + "step": 6353 + }, + { + "epoch": 0.5239548115774717, + "grad_norm": 1.844683051109314, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.735151469707489, + "num_tokens": 81311051.0, + "step": 6354 + }, + { + "epoch": 0.5240372722025233, + "grad_norm": 1.8616893291473389, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7145482301712036, + "num_tokens": 81344988.0, + "step": 6355 + }, + { + "epoch": 0.5241197328275748, + "grad_norm": 1.7660585641860962, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7624281048774719, + "num_tokens": 81378106.0, + "step": 6356 + }, + { + "epoch": 0.5242021934526264, + "grad_norm": 2.1513423919677734, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7386366724967957, + "num_tokens": 81402575.0, + "step": 6357 + }, + { + "epoch": 0.5242846540776779, + "grad_norm": 2.174391269683838, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7576824426651001, + "num_tokens": 81429038.0, + "step": 6358 + }, + { + "epoch": 0.5243671147027295, + "grad_norm": 1.992514729499817, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.760668933391571, + "num_tokens": 81459813.0, + "step": 6359 + }, + { + "epoch": 0.524449575327781, + "grad_norm": 1.7817107439041138, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7762366533279419, + "num_tokens": 81492670.0, + "step": 6360 + }, + { + "epoch": 0.5245320359528325, + "grad_norm": 1.8730018138885498, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7371792793273926, + "num_tokens": 81527170.0, + "step": 6361 + }, + { + "epoch": 0.5246144965778841, + "grad_norm": 1.8898380994796753, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7664512991905212, + "num_tokens": 81558784.0, + "step": 6362 + }, + { + "epoch": 0.5246969572029356, + "grad_norm": 1.9506345987319946, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7356778383255005, + "num_tokens": 81594496.0, + "step": 6363 + }, + { + "epoch": 0.5247794178279871, + "grad_norm": 1.7094933986663818, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7868781089782715, + "num_tokens": 81628243.0, + "step": 6364 + }, + { + "epoch": 0.5248618784530387, + "grad_norm": 1.8019382953643799, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7791585922241211, + "num_tokens": 81659276.0, + "step": 6365 + }, + { + "epoch": 0.5249443390780902, + "grad_norm": 1.7459224462509155, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7447947263717651, + "num_tokens": 81695031.0, + "step": 6366 + }, + { + "epoch": 0.5250267997031417, + "grad_norm": 1.705286979675293, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7711629271507263, + "num_tokens": 81729830.0, + "step": 6367 + }, + { + "epoch": 0.5251092603281933, + "grad_norm": 1.6692405939102173, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7343991994857788, + "num_tokens": 81769415.0, + "step": 6368 + }, + { + "epoch": 0.5251917209532448, + "grad_norm": 2.059774398803711, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7425620555877686, + "num_tokens": 81797382.0, + "step": 6369 + }, + { + "epoch": 0.5252741815782964, + "grad_norm": 1.753511905670166, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7684782147407532, + "num_tokens": 81831963.0, + "step": 6370 + }, + { + "epoch": 0.5253566422033479, + "grad_norm": 2.0375940799713135, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7706649303436279, + "num_tokens": 81860513.0, + "step": 6371 + }, + { + "epoch": 0.5254391028283995, + "grad_norm": 2.078226327896118, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7231430411338806, + "num_tokens": 81888511.0, + "step": 6372 + }, + { + "epoch": 0.525521563453451, + "grad_norm": 1.8980094194412231, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7474503517150879, + "num_tokens": 81919952.0, + "step": 6373 + }, + { + "epoch": 0.5256040240785025, + "grad_norm": 2.004657745361328, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.7589038610458374, + "num_tokens": 81947166.0, + "step": 6374 + }, + { + "epoch": 0.5256864847035541, + "grad_norm": 1.9377851486206055, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7655820846557617, + "num_tokens": 81977349.0, + "step": 6375 + }, + { + "epoch": 0.5257689453286056, + "grad_norm": 1.9866372346878052, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7529343366622925, + "num_tokens": 82010950.0, + "step": 6376 + }, + { + "epoch": 0.5258514059536571, + "grad_norm": 1.881195306777954, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7229015827178955, + "num_tokens": 82046539.0, + "step": 6377 + }, + { + "epoch": 0.5259338665787087, + "grad_norm": 2.082064628601074, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7430038452148438, + "num_tokens": 82074111.0, + "step": 6378 + }, + { + "epoch": 0.5260163272037602, + "grad_norm": 1.8606964349746704, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.766243040561676, + "num_tokens": 82104052.0, + "step": 6379 + }, + { + "epoch": 0.5260987878288117, + "grad_norm": 2.004650831222534, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7604252696037292, + "num_tokens": 82130685.0, + "step": 6380 + }, + { + "epoch": 0.5261812484538633, + "grad_norm": 1.7966861724853516, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.749347448348999, + "num_tokens": 82165181.0, + "step": 6381 + }, + { + "epoch": 0.5262637090789148, + "grad_norm": 2.0447261333465576, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7523491978645325, + "num_tokens": 82192640.0, + "step": 6382 + }, + { + "epoch": 0.5263461697039663, + "grad_norm": 1.9639219045639038, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7774226665496826, + "num_tokens": 82219341.0, + "step": 6383 + }, + { + "epoch": 0.5264286303290179, + "grad_norm": 1.984948992729187, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7451101541519165, + "num_tokens": 82251365.0, + "step": 6384 + }, + { + "epoch": 0.5265110909540694, + "grad_norm": 1.8343557119369507, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7211999297142029, + "num_tokens": 82285909.0, + "step": 6385 + }, + { + "epoch": 0.526593551579121, + "grad_norm": 1.7517173290252686, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7559680342674255, + "num_tokens": 82319285.0, + "step": 6386 + }, + { + "epoch": 0.5266760122041725, + "grad_norm": 1.9241033792495728, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7559717297554016, + "num_tokens": 82349932.0, + "step": 6387 + }, + { + "epoch": 0.5267584728292241, + "grad_norm": 2.1600303649902344, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.732830822467804, + "num_tokens": 82377323.0, + "step": 6388 + }, + { + "epoch": 0.5268409334542756, + "grad_norm": 1.9585072994232178, + "learning_rate": 1e-06, + "loss": 0.6012, + "mean_token_accuracy": 0.804947018623352, + "num_tokens": 82406843.0, + "step": 6389 + }, + { + "epoch": 0.5269233940793271, + "grad_norm": 2.048248529434204, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7192566990852356, + "num_tokens": 82434409.0, + "step": 6390 + }, + { + "epoch": 0.5270058547043787, + "grad_norm": 2.0298097133636475, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7649635076522827, + "num_tokens": 82463193.0, + "step": 6391 + }, + { + "epoch": 0.5270883153294302, + "grad_norm": 2.392601490020752, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7146412134170532, + "num_tokens": 82485899.0, + "step": 6392 + }, + { + "epoch": 0.5271707759544817, + "grad_norm": 2.0347251892089844, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7356398105621338, + "num_tokens": 82514982.0, + "step": 6393 + }, + { + "epoch": 0.5272532365795333, + "grad_norm": 1.9499199390411377, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7681390643119812, + "num_tokens": 82544053.0, + "step": 6394 + }, + { + "epoch": 0.5273356972045848, + "grad_norm": 1.9282127618789673, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7632343173027039, + "num_tokens": 82575691.0, + "step": 6395 + }, + { + "epoch": 0.5274181578296363, + "grad_norm": 1.943275809288025, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.7591216564178467, + "num_tokens": 82604005.0, + "step": 6396 + }, + { + "epoch": 0.5275006184546879, + "grad_norm": 1.9738942384719849, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7463486194610596, + "num_tokens": 82634214.0, + "step": 6397 + }, + { + "epoch": 0.5275830790797394, + "grad_norm": 2.2075986862182617, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.716491162776947, + "num_tokens": 82660735.0, + "step": 6398 + }, + { + "epoch": 0.5276655397047909, + "grad_norm": 2.0258545875549316, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7465803623199463, + "num_tokens": 82690475.0, + "step": 6399 + }, + { + "epoch": 0.5277480003298425, + "grad_norm": 1.869085669517517, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7503823637962341, + "num_tokens": 82721461.0, + "step": 6400 + }, + { + "epoch": 0.527830460954894, + "grad_norm": 2.1700973510742188, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7430868148803711, + "num_tokens": 82747265.0, + "step": 6401 + }, + { + "epoch": 0.5279129215799456, + "grad_norm": 2.0620498657226562, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7376728653907776, + "num_tokens": 82775529.0, + "step": 6402 + }, + { + "epoch": 0.5279953822049971, + "grad_norm": 1.9015384912490845, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7492315173149109, + "num_tokens": 82809115.0, + "step": 6403 + }, + { + "epoch": 0.5280778428300487, + "grad_norm": 1.846545696258545, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7613399028778076, + "num_tokens": 82841381.0, + "step": 6404 + }, + { + "epoch": 0.5281603034551002, + "grad_norm": 2.2704999446868896, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7233465909957886, + "num_tokens": 82867168.0, + "step": 6405 + }, + { + "epoch": 0.5282427640801517, + "grad_norm": 2.2242965698242188, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7258000373840332, + "num_tokens": 82893808.0, + "step": 6406 + }, + { + "epoch": 0.5283252247052033, + "grad_norm": 1.9272511005401611, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.745759129524231, + "num_tokens": 82923735.0, + "step": 6407 + }, + { + "epoch": 0.5284076853302548, + "grad_norm": 2.0553205013275146, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7128283381462097, + "num_tokens": 82953155.0, + "step": 6408 + }, + { + "epoch": 0.5284901459553063, + "grad_norm": 2.1265976428985596, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7207313776016235, + "num_tokens": 82981215.0, + "step": 6409 + }, + { + "epoch": 0.5285726065803579, + "grad_norm": 1.8417561054229736, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7052344083786011, + "num_tokens": 83016412.0, + "step": 6410 + }, + { + "epoch": 0.5286550672054094, + "grad_norm": 2.0066945552825928, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7741991281509399, + "num_tokens": 83043222.0, + "step": 6411 + }, + { + "epoch": 0.5287375278304609, + "grad_norm": 1.8562982082366943, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.735622227191925, + "num_tokens": 83077388.0, + "step": 6412 + }, + { + "epoch": 0.5288199884555125, + "grad_norm": 1.9677486419677734, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7411067485809326, + "num_tokens": 83106352.0, + "step": 6413 + }, + { + "epoch": 0.528902449080564, + "grad_norm": 1.8640514612197876, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7846335768699646, + "num_tokens": 83134170.0, + "step": 6414 + }, + { + "epoch": 0.5289849097056155, + "grad_norm": 1.7634477615356445, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7580169439315796, + "num_tokens": 83168424.0, + "step": 6415 + }, + { + "epoch": 0.5290673703306671, + "grad_norm": 1.9736883640289307, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7102031707763672, + "num_tokens": 83196349.0, + "step": 6416 + }, + { + "epoch": 0.5291498309557187, + "grad_norm": 2.002061605453491, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7734472155570984, + "num_tokens": 83229269.0, + "step": 6417 + }, + { + "epoch": 0.5292322915807702, + "grad_norm": 1.910526156425476, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7300005555152893, + "num_tokens": 83263818.0, + "step": 6418 + }, + { + "epoch": 0.5293147522058217, + "grad_norm": 2.0159659385681152, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7331104874610901, + "num_tokens": 83291681.0, + "step": 6419 + }, + { + "epoch": 0.5293972128308733, + "grad_norm": 1.6140962839126587, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7674147486686707, + "num_tokens": 83329051.0, + "step": 6420 + }, + { + "epoch": 0.5294796734559248, + "grad_norm": 1.9440900087356567, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7159558534622192, + "num_tokens": 83360753.0, + "step": 6421 + }, + { + "epoch": 0.5295621340809763, + "grad_norm": 1.8369182348251343, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7767702341079712, + "num_tokens": 83391208.0, + "step": 6422 + }, + { + "epoch": 0.5296445947060279, + "grad_norm": 2.020277976989746, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7705397009849548, + "num_tokens": 83419219.0, + "step": 6423 + }, + { + "epoch": 0.5297270553310794, + "grad_norm": 1.668599009513855, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7810360193252563, + "num_tokens": 83460556.0, + "step": 6424 + }, + { + "epoch": 0.5298095159561309, + "grad_norm": 1.9330931901931763, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.737693190574646, + "num_tokens": 83490122.0, + "step": 6425 + }, + { + "epoch": 0.5298919765811825, + "grad_norm": 1.985520839691162, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7831159830093384, + "num_tokens": 83517347.0, + "step": 6426 + }, + { + "epoch": 0.529974437206234, + "grad_norm": 1.8620352745056152, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7327624559402466, + "num_tokens": 83550163.0, + "step": 6427 + }, + { + "epoch": 0.5300568978312855, + "grad_norm": 1.9854156970977783, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7615776062011719, + "num_tokens": 83579867.0, + "step": 6428 + }, + { + "epoch": 0.5301393584563371, + "grad_norm": 2.0000112056732178, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7546318769454956, + "num_tokens": 83607151.0, + "step": 6429 + }, + { + "epoch": 0.5302218190813887, + "grad_norm": 1.833945870399475, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7605041265487671, + "num_tokens": 83640459.0, + "step": 6430 + }, + { + "epoch": 0.5303042797064402, + "grad_norm": 1.8795429468154907, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7557393312454224, + "num_tokens": 83675047.0, + "step": 6431 + }, + { + "epoch": 0.5303867403314917, + "grad_norm": 1.8184165954589844, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7190934419631958, + "num_tokens": 83707214.0, + "step": 6432 + }, + { + "epoch": 0.5304692009565433, + "grad_norm": 1.8664318323135376, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7611812353134155, + "num_tokens": 83739988.0, + "step": 6433 + }, + { + "epoch": 0.5305516615815948, + "grad_norm": 1.9514180421829224, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7348916530609131, + "num_tokens": 83769510.0, + "step": 6434 + }, + { + "epoch": 0.5306341222066463, + "grad_norm": 1.992587685585022, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.7457578182220459, + "num_tokens": 83798034.0, + "step": 6435 + }, + { + "epoch": 0.5307165828316979, + "grad_norm": 2.1130192279815674, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7242960333824158, + "num_tokens": 83827482.0, + "step": 6436 + }, + { + "epoch": 0.5307990434567494, + "grad_norm": 2.0118727684020996, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.7441140413284302, + "num_tokens": 83855254.0, + "step": 6437 + }, + { + "epoch": 0.5308815040818009, + "grad_norm": 1.894244909286499, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7519214153289795, + "num_tokens": 83885610.0, + "step": 6438 + }, + { + "epoch": 0.5309639647068525, + "grad_norm": 1.9312288761138916, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7615931630134583, + "num_tokens": 83915206.0, + "step": 6439 + }, + { + "epoch": 0.531046425331904, + "grad_norm": 2.0179648399353027, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7429691553115845, + "num_tokens": 83941900.0, + "step": 6440 + }, + { + "epoch": 0.5311288859569555, + "grad_norm": 2.059523344039917, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7490427494049072, + "num_tokens": 83968690.0, + "step": 6441 + }, + { + "epoch": 0.5312113465820071, + "grad_norm": 2.239624500274658, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.705098032951355, + "num_tokens": 83995475.0, + "step": 6442 + }, + { + "epoch": 0.5312938072070587, + "grad_norm": 2.2146482467651367, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7387468814849854, + "num_tokens": 84018577.0, + "step": 6443 + }, + { + "epoch": 0.5313762678321101, + "grad_norm": 2.301574230194092, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6955813765525818, + "num_tokens": 84042863.0, + "step": 6444 + }, + { + "epoch": 0.5314587284571617, + "grad_norm": 2.0953385829925537, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7376629710197449, + "num_tokens": 84069891.0, + "step": 6445 + }, + { + "epoch": 0.5315411890822133, + "grad_norm": 1.6748770475387573, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7780443429946899, + "num_tokens": 84107000.0, + "step": 6446 + }, + { + "epoch": 0.5316236497072648, + "grad_norm": 1.9271245002746582, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7325407266616821, + "num_tokens": 84140494.0, + "step": 6447 + }, + { + "epoch": 0.5317061103323163, + "grad_norm": 1.8137046098709106, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7461144924163818, + "num_tokens": 84177193.0, + "step": 6448 + }, + { + "epoch": 0.5317885709573679, + "grad_norm": 1.9401887655258179, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7729190587997437, + "num_tokens": 84206285.0, + "step": 6449 + }, + { + "epoch": 0.5318710315824194, + "grad_norm": 1.856331706047058, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.781641960144043, + "num_tokens": 84245506.0, + "step": 6450 + }, + { + "epoch": 0.5319534922074709, + "grad_norm": 1.704032301902771, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7607439756393433, + "num_tokens": 84283094.0, + "step": 6451 + }, + { + "epoch": 0.5320359528325225, + "grad_norm": 1.8703352212905884, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7423070669174194, + "num_tokens": 84313744.0, + "step": 6452 + }, + { + "epoch": 0.532118413457574, + "grad_norm": 1.7229912281036377, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7447032332420349, + "num_tokens": 84352429.0, + "step": 6453 + }, + { + "epoch": 0.5322008740826255, + "grad_norm": 2.1123085021972656, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.751858115196228, + "num_tokens": 84380160.0, + "step": 6454 + }, + { + "epoch": 0.5322833347076771, + "grad_norm": 1.8417624235153198, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.7968573570251465, + "num_tokens": 84409449.0, + "step": 6455 + }, + { + "epoch": 0.5323657953327287, + "grad_norm": 1.9171892404556274, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7709441781044006, + "num_tokens": 84438657.0, + "step": 6456 + }, + { + "epoch": 0.5324482559577801, + "grad_norm": 2.0465638637542725, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7263680696487427, + "num_tokens": 84465028.0, + "step": 6457 + }, + { + "epoch": 0.5325307165828317, + "grad_norm": 1.9838007688522339, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.7719443440437317, + "num_tokens": 84492079.0, + "step": 6458 + }, + { + "epoch": 0.5326131772078833, + "grad_norm": 2.1678617000579834, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7496976852416992, + "num_tokens": 84519016.0, + "step": 6459 + }, + { + "epoch": 0.5326956378329347, + "grad_norm": 1.9146074056625366, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7539163827896118, + "num_tokens": 84552314.0, + "step": 6460 + }, + { + "epoch": 0.5327780984579863, + "grad_norm": 1.7869551181793213, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7646524906158447, + "num_tokens": 84585362.0, + "step": 6461 + }, + { + "epoch": 0.5328605590830379, + "grad_norm": 2.0503392219543457, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7642457485198975, + "num_tokens": 84615538.0, + "step": 6462 + }, + { + "epoch": 0.5329430197080894, + "grad_norm": 1.8730995655059814, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7762410640716553, + "num_tokens": 84647135.0, + "step": 6463 + }, + { + "epoch": 0.5330254803331409, + "grad_norm": 1.9783638715744019, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7540733218193054, + "num_tokens": 84676716.0, + "step": 6464 + }, + { + "epoch": 0.5331079409581925, + "grad_norm": 2.054067373275757, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7320781946182251, + "num_tokens": 84704617.0, + "step": 6465 + }, + { + "epoch": 0.533190401583244, + "grad_norm": 2.1452090740203857, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7709000110626221, + "num_tokens": 84729260.0, + "step": 6466 + }, + { + "epoch": 0.5332728622082955, + "grad_norm": 2.2082581520080566, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7209601402282715, + "num_tokens": 84755768.0, + "step": 6467 + }, + { + "epoch": 0.5333553228333471, + "grad_norm": 2.132174491882324, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7245045900344849, + "num_tokens": 84779586.0, + "step": 6468 + }, + { + "epoch": 0.5334377834583987, + "grad_norm": 2.0227060317993164, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7580433487892151, + "num_tokens": 84807688.0, + "step": 6469 + }, + { + "epoch": 0.5335202440834501, + "grad_norm": 2.175370216369629, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7358682155609131, + "num_tokens": 84833238.0, + "step": 6470 + }, + { + "epoch": 0.5336027047085017, + "grad_norm": 1.85713791847229, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.739640474319458, + "num_tokens": 84865857.0, + "step": 6471 + }, + { + "epoch": 0.5336851653335533, + "grad_norm": 1.913461685180664, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7267119884490967, + "num_tokens": 84897950.0, + "step": 6472 + }, + { + "epoch": 0.5337676259586047, + "grad_norm": 2.0608954429626465, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.7520793676376343, + "num_tokens": 84923101.0, + "step": 6473 + }, + { + "epoch": 0.5338500865836563, + "grad_norm": 1.8677259683609009, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.727446973323822, + "num_tokens": 84958246.0, + "step": 6474 + }, + { + "epoch": 0.5339325472087079, + "grad_norm": 2.428173065185547, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7386651039123535, + "num_tokens": 84977508.0, + "step": 6475 + }, + { + "epoch": 0.5340150078337593, + "grad_norm": 1.6841164827346802, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7686447501182556, + "num_tokens": 85016465.0, + "step": 6476 + }, + { + "epoch": 0.5340974684588109, + "grad_norm": 2.0108935832977295, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7388036251068115, + "num_tokens": 85044289.0, + "step": 6477 + }, + { + "epoch": 0.5341799290838625, + "grad_norm": 2.0075907707214355, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7521748542785645, + "num_tokens": 85070693.0, + "step": 6478 + }, + { + "epoch": 0.534262389708914, + "grad_norm": 2.038158416748047, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6993057727813721, + "num_tokens": 85101798.0, + "step": 6479 + }, + { + "epoch": 0.5343448503339655, + "grad_norm": 2.00213360786438, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.747617781162262, + "num_tokens": 85131067.0, + "step": 6480 + }, + { + "epoch": 0.5344273109590171, + "grad_norm": 1.8497200012207031, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7694827318191528, + "num_tokens": 85165663.0, + "step": 6481 + }, + { + "epoch": 0.5345097715840686, + "grad_norm": 1.8946633338928223, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.76633620262146, + "num_tokens": 85194914.0, + "step": 6482 + }, + { + "epoch": 0.5345922322091201, + "grad_norm": 1.9212820529937744, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7297417521476746, + "num_tokens": 85228309.0, + "step": 6483 + }, + { + "epoch": 0.5346746928341717, + "grad_norm": 1.871412754058838, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7602615356445312, + "num_tokens": 85259666.0, + "step": 6484 + }, + { + "epoch": 0.5347571534592233, + "grad_norm": 1.6403449773788452, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.806955873966217, + "num_tokens": 85298270.0, + "step": 6485 + }, + { + "epoch": 0.5348396140842747, + "grad_norm": 1.8674699068069458, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.7883011102676392, + "num_tokens": 85327147.0, + "step": 6486 + }, + { + "epoch": 0.5349220747093263, + "grad_norm": 1.964572787284851, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7480166554450989, + "num_tokens": 85357175.0, + "step": 6487 + }, + { + "epoch": 0.5350045353343779, + "grad_norm": 1.5432173013687134, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7607727646827698, + "num_tokens": 85401561.0, + "step": 6488 + }, + { + "epoch": 0.5350869959594293, + "grad_norm": 2.066784381866455, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7367957830429077, + "num_tokens": 85427961.0, + "step": 6489 + }, + { + "epoch": 0.5351694565844809, + "grad_norm": 2.001466989517212, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7479864954948425, + "num_tokens": 85459533.0, + "step": 6490 + }, + { + "epoch": 0.5352519172095325, + "grad_norm": 2.1647350788116455, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7670139074325562, + "num_tokens": 85482841.0, + "step": 6491 + }, + { + "epoch": 0.535334377834584, + "grad_norm": 2.1169373989105225, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7606257200241089, + "num_tokens": 85513463.0, + "step": 6492 + }, + { + "epoch": 0.5354168384596355, + "grad_norm": 1.9066791534423828, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.7386242151260376, + "num_tokens": 85544269.0, + "step": 6493 + }, + { + "epoch": 0.5354992990846871, + "grad_norm": 1.953883409500122, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7161523103713989, + "num_tokens": 85576191.0, + "step": 6494 + }, + { + "epoch": 0.5355817597097386, + "grad_norm": 2.0915796756744385, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7227944135665894, + "num_tokens": 85605725.0, + "step": 6495 + }, + { + "epoch": 0.5356642203347901, + "grad_norm": 2.15798282623291, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.738349199295044, + "num_tokens": 85632348.0, + "step": 6496 + }, + { + "epoch": 0.5357466809598417, + "grad_norm": 2.05098557472229, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7483624219894409, + "num_tokens": 85660751.0, + "step": 6497 + }, + { + "epoch": 0.5358291415848933, + "grad_norm": 1.8841748237609863, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.746120274066925, + "num_tokens": 85694230.0, + "step": 6498 + }, + { + "epoch": 0.5359116022099447, + "grad_norm": 1.7725069522857666, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.771889328956604, + "num_tokens": 85733388.0, + "step": 6499 + }, + { + "epoch": 0.5359940628349963, + "grad_norm": 1.7229316234588623, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7731794118881226, + "num_tokens": 85767442.0, + "step": 6500 + }, + { + "epoch": 0.5360765234600479, + "grad_norm": 2.1129775047302246, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7652477025985718, + "num_tokens": 85790163.0, + "step": 6501 + }, + { + "epoch": 0.5361589840850993, + "grad_norm": 1.920458197593689, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7679798603057861, + "num_tokens": 85821220.0, + "step": 6502 + }, + { + "epoch": 0.5362414447101509, + "grad_norm": 1.8373444080352783, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7798467874526978, + "num_tokens": 85853652.0, + "step": 6503 + }, + { + "epoch": 0.5363239053352025, + "grad_norm": 2.305741786956787, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7227861881256104, + "num_tokens": 85876721.0, + "step": 6504 + }, + { + "epoch": 0.5364063659602539, + "grad_norm": 1.9371315240859985, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.75770103931427, + "num_tokens": 85911431.0, + "step": 6505 + }, + { + "epoch": 0.5364888265853055, + "grad_norm": 1.9662792682647705, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7298403978347778, + "num_tokens": 85943753.0, + "step": 6506 + }, + { + "epoch": 0.5365712872103571, + "grad_norm": 1.9236494302749634, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7381159067153931, + "num_tokens": 85975772.0, + "step": 6507 + }, + { + "epoch": 0.5366537478354086, + "grad_norm": 2.012019157409668, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.719696044921875, + "num_tokens": 86004644.0, + "step": 6508 + }, + { + "epoch": 0.5367362084604601, + "grad_norm": 2.0963659286499023, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7418609857559204, + "num_tokens": 86031371.0, + "step": 6509 + }, + { + "epoch": 0.5368186690855117, + "grad_norm": 1.980119228363037, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7502041459083557, + "num_tokens": 86063081.0, + "step": 6510 + }, + { + "epoch": 0.5369011297105633, + "grad_norm": 2.2457897663116455, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7075241804122925, + "num_tokens": 86087159.0, + "step": 6511 + }, + { + "epoch": 0.5369835903356147, + "grad_norm": 1.883124828338623, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7596043348312378, + "num_tokens": 86118634.0, + "step": 6512 + }, + { + "epoch": 0.5370660509606663, + "grad_norm": 2.058319330215454, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7174278497695923, + "num_tokens": 86150457.0, + "step": 6513 + }, + { + "epoch": 0.5371485115857179, + "grad_norm": 1.747995138168335, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7809364795684814, + "num_tokens": 86184707.0, + "step": 6514 + }, + { + "epoch": 0.5372309722107693, + "grad_norm": 1.9189677238464355, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7264826893806458, + "num_tokens": 86215048.0, + "step": 6515 + }, + { + "epoch": 0.5373134328358209, + "grad_norm": 2.1533186435699463, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7473020553588867, + "num_tokens": 86242271.0, + "step": 6516 + }, + { + "epoch": 0.5373958934608725, + "grad_norm": 1.9785292148590088, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7648138999938965, + "num_tokens": 86271777.0, + "step": 6517 + }, + { + "epoch": 0.5374783540859239, + "grad_norm": 1.9449576139450073, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7692967653274536, + "num_tokens": 86302526.0, + "step": 6518 + }, + { + "epoch": 0.5375608147109755, + "grad_norm": 2.099799394607544, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7534588575363159, + "num_tokens": 86329329.0, + "step": 6519 + }, + { + "epoch": 0.5376432753360271, + "grad_norm": 2.3218917846679688, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7642809152603149, + "num_tokens": 86358360.0, + "step": 6520 + }, + { + "epoch": 0.5377257359610785, + "grad_norm": 1.6614739894866943, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.7935276031494141, + "num_tokens": 86395370.0, + "step": 6521 + }, + { + "epoch": 0.5378081965861301, + "grad_norm": 1.7684082984924316, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7598786950111389, + "num_tokens": 86427922.0, + "step": 6522 + }, + { + "epoch": 0.5378906572111817, + "grad_norm": 1.9524757862091064, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7666608691215515, + "num_tokens": 86454944.0, + "step": 6523 + }, + { + "epoch": 0.5379731178362332, + "grad_norm": 1.8349549770355225, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7838895916938782, + "num_tokens": 86486983.0, + "step": 6524 + }, + { + "epoch": 0.5380555784612847, + "grad_norm": 1.7630207538604736, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7167357802391052, + "num_tokens": 86523237.0, + "step": 6525 + }, + { + "epoch": 0.5381380390863363, + "grad_norm": 2.183453321456909, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7659894227981567, + "num_tokens": 86546321.0, + "step": 6526 + }, + { + "epoch": 0.5382204997113879, + "grad_norm": 2.192776679992676, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7495013475418091, + "num_tokens": 86574275.0, + "step": 6527 + }, + { + "epoch": 0.5383029603364393, + "grad_norm": 2.451634645462036, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7211446762084961, + "num_tokens": 86596638.0, + "step": 6528 + }, + { + "epoch": 0.5383854209614909, + "grad_norm": 2.161573886871338, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7495202422142029, + "num_tokens": 86621220.0, + "step": 6529 + }, + { + "epoch": 0.5384678815865425, + "grad_norm": 1.8311697244644165, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7712856531143188, + "num_tokens": 86654605.0, + "step": 6530 + }, + { + "epoch": 0.5385503422115939, + "grad_norm": 2.2283382415771484, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7283090353012085, + "num_tokens": 86678073.0, + "step": 6531 + }, + { + "epoch": 0.5386328028366455, + "grad_norm": 1.9841428995132446, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.716704249382019, + "num_tokens": 86708085.0, + "step": 6532 + }, + { + "epoch": 0.5387152634616971, + "grad_norm": 2.0804851055145264, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.744642972946167, + "num_tokens": 86736155.0, + "step": 6533 + }, + { + "epoch": 0.5387977240867485, + "grad_norm": 1.6499333381652832, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7825880646705627, + "num_tokens": 86775198.0, + "step": 6534 + }, + { + "epoch": 0.5388801847118001, + "grad_norm": 1.8605396747589111, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7492536306381226, + "num_tokens": 86805143.0, + "step": 6535 + }, + { + "epoch": 0.5389626453368517, + "grad_norm": 1.9125370979309082, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7406968474388123, + "num_tokens": 86837819.0, + "step": 6536 + }, + { + "epoch": 0.5390451059619031, + "grad_norm": 2.1228444576263428, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7436200976371765, + "num_tokens": 86863057.0, + "step": 6537 + }, + { + "epoch": 0.5391275665869547, + "grad_norm": 1.8682503700256348, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7477456331253052, + "num_tokens": 86897570.0, + "step": 6538 + }, + { + "epoch": 0.5392100272120063, + "grad_norm": 1.8507614135742188, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7283819913864136, + "num_tokens": 86931263.0, + "step": 6539 + }, + { + "epoch": 0.5392924878370579, + "grad_norm": 2.0551137924194336, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7092581987380981, + "num_tokens": 86960207.0, + "step": 6540 + }, + { + "epoch": 0.5393749484621093, + "grad_norm": 1.9381319284439087, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7585474848747253, + "num_tokens": 86990527.0, + "step": 6541 + }, + { + "epoch": 0.5394574090871609, + "grad_norm": 2.1221659183502197, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7132881879806519, + "num_tokens": 87017365.0, + "step": 6542 + }, + { + "epoch": 0.5395398697122125, + "grad_norm": 1.8543167114257812, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7340999245643616, + "num_tokens": 87052379.0, + "step": 6543 + }, + { + "epoch": 0.5396223303372639, + "grad_norm": 1.7383379936218262, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.782569408416748, + "num_tokens": 87087659.0, + "step": 6544 + }, + { + "epoch": 0.5397047909623155, + "grad_norm": 2.193643808364868, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7344918847084045, + "num_tokens": 87111841.0, + "step": 6545 + }, + { + "epoch": 0.5397872515873671, + "grad_norm": 2.2483959197998047, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7464278936386108, + "num_tokens": 87134489.0, + "step": 6546 + }, + { + "epoch": 0.5398697122124185, + "grad_norm": 2.115553855895996, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7813482284545898, + "num_tokens": 87157965.0, + "step": 6547 + }, + { + "epoch": 0.5399521728374701, + "grad_norm": 2.0291457176208496, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7371432781219482, + "num_tokens": 87187323.0, + "step": 6548 + }, + { + "epoch": 0.5400346334625217, + "grad_norm": 1.8780063390731812, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6965416669845581, + "num_tokens": 87218450.0, + "step": 6549 + }, + { + "epoch": 0.5401170940875731, + "grad_norm": 2.1196200847625732, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7241084575653076, + "num_tokens": 87245546.0, + "step": 6550 + }, + { + "epoch": 0.5401995547126247, + "grad_norm": 2.021287441253662, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7375432252883911, + "num_tokens": 87273183.0, + "step": 6551 + }, + { + "epoch": 0.5402820153376763, + "grad_norm": 1.763289213180542, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6954265832901001, + "num_tokens": 87308878.0, + "step": 6552 + }, + { + "epoch": 0.5403644759627279, + "grad_norm": 2.20139741897583, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7154165506362915, + "num_tokens": 87333779.0, + "step": 6553 + }, + { + "epoch": 0.5404469365877793, + "grad_norm": 2.096226692199707, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.745362401008606, + "num_tokens": 87361926.0, + "step": 6554 + }, + { + "epoch": 0.5405293972128309, + "grad_norm": 2.0396361351013184, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.745530366897583, + "num_tokens": 87388537.0, + "step": 6555 + }, + { + "epoch": 0.5406118578378825, + "grad_norm": 2.4147372245788574, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7441352009773254, + "num_tokens": 87407741.0, + "step": 6556 + }, + { + "epoch": 0.5406943184629339, + "grad_norm": 1.9057751893997192, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7613892555236816, + "num_tokens": 87440974.0, + "step": 6557 + }, + { + "epoch": 0.5407767790879855, + "grad_norm": 1.9230918884277344, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7733809351921082, + "num_tokens": 87474450.0, + "step": 6558 + }, + { + "epoch": 0.5408592397130371, + "grad_norm": 1.900146245956421, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7394587993621826, + "num_tokens": 87506469.0, + "step": 6559 + }, + { + "epoch": 0.5409417003380885, + "grad_norm": 2.0144155025482178, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7394677400588989, + "num_tokens": 87538475.0, + "step": 6560 + }, + { + "epoch": 0.5410241609631401, + "grad_norm": 2.0258190631866455, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7361505031585693, + "num_tokens": 87567709.0, + "step": 6561 + }, + { + "epoch": 0.5411066215881917, + "grad_norm": 2.071287155151367, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7066898345947266, + "num_tokens": 87597433.0, + "step": 6562 + }, + { + "epoch": 0.5411890822132431, + "grad_norm": 1.9078682661056519, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7503583431243896, + "num_tokens": 87630646.0, + "step": 6563 + }, + { + "epoch": 0.5412715428382947, + "grad_norm": 2.2155916690826416, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7106184959411621, + "num_tokens": 87653976.0, + "step": 6564 + }, + { + "epoch": 0.5413540034633463, + "grad_norm": 2.0527617931365967, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7476659417152405, + "num_tokens": 87683203.0, + "step": 6565 + }, + { + "epoch": 0.5414364640883977, + "grad_norm": 1.8044205904006958, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.762894332408905, + "num_tokens": 87716553.0, + "step": 6566 + }, + { + "epoch": 0.5415189247134493, + "grad_norm": 1.7818453311920166, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7636941075325012, + "num_tokens": 87749868.0, + "step": 6567 + }, + { + "epoch": 0.5416013853385009, + "grad_norm": 2.1811585426330566, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7181581258773804, + "num_tokens": 87776610.0, + "step": 6568 + }, + { + "epoch": 0.5416838459635525, + "grad_norm": 1.8829092979431152, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6871185302734375, + "num_tokens": 87810244.0, + "step": 6569 + }, + { + "epoch": 0.5417663065886039, + "grad_norm": 1.9899091720581055, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7433373928070068, + "num_tokens": 87839896.0, + "step": 6570 + }, + { + "epoch": 0.5418487672136555, + "grad_norm": 2.4362332820892334, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7599382400512695, + "num_tokens": 87860822.0, + "step": 6571 + }, + { + "epoch": 0.5419312278387071, + "grad_norm": 1.9074283838272095, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7493761777877808, + "num_tokens": 87889956.0, + "step": 6572 + }, + { + "epoch": 0.5420136884637585, + "grad_norm": 2.199887752532959, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7213469743728638, + "num_tokens": 87915634.0, + "step": 6573 + }, + { + "epoch": 0.5420961490888101, + "grad_norm": 2.1083016395568848, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7359533309936523, + "num_tokens": 87942930.0, + "step": 6574 + }, + { + "epoch": 0.5421786097138617, + "grad_norm": 1.6870720386505127, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7680438756942749, + "num_tokens": 87978819.0, + "step": 6575 + }, + { + "epoch": 0.5422610703389131, + "grad_norm": 2.098193407058716, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7516331672668457, + "num_tokens": 88007301.0, + "step": 6576 + }, + { + "epoch": 0.5423435309639647, + "grad_norm": 1.928040862083435, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7479202747344971, + "num_tokens": 88039180.0, + "step": 6577 + }, + { + "epoch": 0.5424259915890163, + "grad_norm": 1.966715931892395, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7584669589996338, + "num_tokens": 88074396.0, + "step": 6578 + }, + { + "epoch": 0.5425084522140677, + "grad_norm": 1.7954944372177124, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7535984516143799, + "num_tokens": 88106092.0, + "step": 6579 + }, + { + "epoch": 0.5425909128391193, + "grad_norm": 1.7459526062011719, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7461771368980408, + "num_tokens": 88141619.0, + "step": 6580 + }, + { + "epoch": 0.5426733734641709, + "grad_norm": 1.8360965251922607, + "learning_rate": 1e-06, + "loss": 0.7483, + "mean_token_accuracy": 0.7573273181915283, + "num_tokens": 88174963.0, + "step": 6581 + }, + { + "epoch": 0.5427558340892223, + "grad_norm": 2.010695219039917, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7424476146697998, + "num_tokens": 88201585.0, + "step": 6582 + }, + { + "epoch": 0.5428382947142739, + "grad_norm": 1.8727003335952759, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7814934253692627, + "num_tokens": 88234674.0, + "step": 6583 + }, + { + "epoch": 0.5429207553393255, + "grad_norm": 2.2698121070861816, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7592803239822388, + "num_tokens": 88255375.0, + "step": 6584 + }, + { + "epoch": 0.5430032159643771, + "grad_norm": 1.9963722229003906, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7651033401489258, + "num_tokens": 88284492.0, + "step": 6585 + }, + { + "epoch": 0.5430856765894285, + "grad_norm": 2.2787814140319824, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7373565435409546, + "num_tokens": 88307113.0, + "step": 6586 + }, + { + "epoch": 0.5431681372144801, + "grad_norm": 1.7514824867248535, + "learning_rate": 1e-06, + "loss": 0.7808, + "mean_token_accuracy": 0.7575399875640869, + "num_tokens": 88339601.0, + "step": 6587 + }, + { + "epoch": 0.5432505978395317, + "grad_norm": 1.8510574102401733, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7512765526771545, + "num_tokens": 88376316.0, + "step": 6588 + }, + { + "epoch": 0.5433330584645831, + "grad_norm": 2.0586142539978027, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7559430599212646, + "num_tokens": 88403366.0, + "step": 6589 + }, + { + "epoch": 0.5434155190896347, + "grad_norm": 2.0329837799072266, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7387717962265015, + "num_tokens": 88431889.0, + "step": 6590 + }, + { + "epoch": 0.5434979797146863, + "grad_norm": 1.829890251159668, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7758364677429199, + "num_tokens": 88464063.0, + "step": 6591 + }, + { + "epoch": 0.5435804403397377, + "grad_norm": 2.208439588546753, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.762343168258667, + "num_tokens": 88488635.0, + "step": 6592 + }, + { + "epoch": 0.5436629009647893, + "grad_norm": 2.4476418495178223, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7712717056274414, + "num_tokens": 88507236.0, + "step": 6593 + }, + { + "epoch": 0.5437453615898409, + "grad_norm": 1.9440956115722656, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7687585353851318, + "num_tokens": 88535137.0, + "step": 6594 + }, + { + "epoch": 0.5438278222148923, + "grad_norm": 2.0749478340148926, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7514569759368896, + "num_tokens": 88562619.0, + "step": 6595 + }, + { + "epoch": 0.5439102828399439, + "grad_norm": 1.8239898681640625, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.7876501679420471, + "num_tokens": 88593176.0, + "step": 6596 + }, + { + "epoch": 0.5439927434649955, + "grad_norm": 1.7085131406784058, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7683413028717041, + "num_tokens": 88631916.0, + "step": 6597 + }, + { + "epoch": 0.544075204090047, + "grad_norm": 2.128749132156372, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7431536912918091, + "num_tokens": 88659363.0, + "step": 6598 + }, + { + "epoch": 0.5441576647150985, + "grad_norm": 2.106168508529663, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7639070153236389, + "num_tokens": 88683651.0, + "step": 6599 + }, + { + "epoch": 0.5442401253401501, + "grad_norm": 1.848062515258789, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7658730745315552, + "num_tokens": 88713928.0, + "step": 6600 + }, + { + "epoch": 0.5443225859652017, + "grad_norm": 1.98207426071167, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.737032413482666, + "num_tokens": 88743102.0, + "step": 6601 + }, + { + "epoch": 0.5444050465902531, + "grad_norm": 2.0501182079315186, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.716451108455658, + "num_tokens": 88773529.0, + "step": 6602 + }, + { + "epoch": 0.5444875072153047, + "grad_norm": 2.090970516204834, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7418085336685181, + "num_tokens": 88802191.0, + "step": 6603 + }, + { + "epoch": 0.5445699678403563, + "grad_norm": 2.0339972972869873, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7591409683227539, + "num_tokens": 88832903.0, + "step": 6604 + }, + { + "epoch": 0.5446524284654077, + "grad_norm": 1.6924753189086914, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7754397392272949, + "num_tokens": 88871489.0, + "step": 6605 + }, + { + "epoch": 0.5447348890904593, + "grad_norm": 2.238802909851074, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7650058269500732, + "num_tokens": 88891627.0, + "step": 6606 + }, + { + "epoch": 0.5448173497155109, + "grad_norm": 1.8707518577575684, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7429306507110596, + "num_tokens": 88923156.0, + "step": 6607 + }, + { + "epoch": 0.5448998103405623, + "grad_norm": 1.8747448921203613, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7534397840499878, + "num_tokens": 88955045.0, + "step": 6608 + }, + { + "epoch": 0.5449822709656139, + "grad_norm": 2.1976630687713623, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7653074860572815, + "num_tokens": 88980446.0, + "step": 6609 + }, + { + "epoch": 0.5450647315906655, + "grad_norm": 1.998551607131958, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7428864240646362, + "num_tokens": 89009550.0, + "step": 6610 + }, + { + "epoch": 0.545147192215717, + "grad_norm": 1.9272133111953735, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.7769627571105957, + "num_tokens": 89038508.0, + "step": 6611 + }, + { + "epoch": 0.5452296528407685, + "grad_norm": 1.8315894603729248, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7314649224281311, + "num_tokens": 89073401.0, + "step": 6612 + }, + { + "epoch": 0.5453121134658201, + "grad_norm": 2.1374361515045166, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7152873277664185, + "num_tokens": 89101391.0, + "step": 6613 + }, + { + "epoch": 0.5453945740908716, + "grad_norm": 1.9250904321670532, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7266628742218018, + "num_tokens": 89135157.0, + "step": 6614 + }, + { + "epoch": 0.5454770347159231, + "grad_norm": 1.8355649709701538, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7804093360900879, + "num_tokens": 89166052.0, + "step": 6615 + }, + { + "epoch": 0.5455594953409747, + "grad_norm": 2.0790021419525146, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7029860615730286, + "num_tokens": 89195237.0, + "step": 6616 + }, + { + "epoch": 0.5456419559660263, + "grad_norm": 2.220491647720337, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7694061994552612, + "num_tokens": 89220399.0, + "step": 6617 + }, + { + "epoch": 0.5457244165910777, + "grad_norm": 2.1667561531066895, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7492870092391968, + "num_tokens": 89244700.0, + "step": 6618 + }, + { + "epoch": 0.5458068772161293, + "grad_norm": 1.8000415563583374, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7460894584655762, + "num_tokens": 89277698.0, + "step": 6619 + }, + { + "epoch": 0.5458893378411809, + "grad_norm": 2.00447416305542, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7204277515411377, + "num_tokens": 89309743.0, + "step": 6620 + }, + { + "epoch": 0.5459717984662323, + "grad_norm": 1.9392694234848022, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.693575382232666, + "num_tokens": 89341654.0, + "step": 6621 + }, + { + "epoch": 0.5460542590912839, + "grad_norm": 1.994006633758545, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7602702379226685, + "num_tokens": 89371032.0, + "step": 6622 + }, + { + "epoch": 0.5461367197163355, + "grad_norm": 2.0394773483276367, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7548425197601318, + "num_tokens": 89400099.0, + "step": 6623 + }, + { + "epoch": 0.546219180341387, + "grad_norm": 2.0743465423583984, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7147248983383179, + "num_tokens": 89426825.0, + "step": 6624 + }, + { + "epoch": 0.5463016409664385, + "grad_norm": 2.071038007736206, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7513523101806641, + "num_tokens": 89454223.0, + "step": 6625 + }, + { + "epoch": 0.5463841015914901, + "grad_norm": 1.7619928121566772, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.7902840375900269, + "num_tokens": 89484652.0, + "step": 6626 + }, + { + "epoch": 0.5464665622165416, + "grad_norm": 1.799513339996338, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.73795485496521, + "num_tokens": 89518646.0, + "step": 6627 + }, + { + "epoch": 0.5465490228415931, + "grad_norm": 2.051854133605957, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7398495674133301, + "num_tokens": 89548626.0, + "step": 6628 + }, + { + "epoch": 0.5466314834666447, + "grad_norm": 2.1082191467285156, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7597376108169556, + "num_tokens": 89575314.0, + "step": 6629 + }, + { + "epoch": 0.5467139440916963, + "grad_norm": 2.1045937538146973, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7691759467124939, + "num_tokens": 89601876.0, + "step": 6630 + }, + { + "epoch": 0.5467964047167477, + "grad_norm": 2.2729804515838623, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7255723476409912, + "num_tokens": 89627583.0, + "step": 6631 + }, + { + "epoch": 0.5468788653417993, + "grad_norm": 1.9935837984085083, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.7916525602340698, + "num_tokens": 89655153.0, + "step": 6632 + }, + { + "epoch": 0.5469613259668509, + "grad_norm": 2.134126663208008, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.6958444118499756, + "num_tokens": 89683411.0, + "step": 6633 + }, + { + "epoch": 0.5470437865919023, + "grad_norm": 1.9548721313476562, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7574341297149658, + "num_tokens": 89711395.0, + "step": 6634 + }, + { + "epoch": 0.5471262472169539, + "grad_norm": 2.145540714263916, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.733422040939331, + "num_tokens": 89740811.0, + "step": 6635 + }, + { + "epoch": 0.5472087078420055, + "grad_norm": 2.397780418395996, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7030173540115356, + "num_tokens": 89763634.0, + "step": 6636 + }, + { + "epoch": 0.5472911684670569, + "grad_norm": 2.0497617721557617, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7563657760620117, + "num_tokens": 89793918.0, + "step": 6637 + }, + { + "epoch": 0.5473736290921085, + "grad_norm": 1.986038088798523, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7063059210777283, + "num_tokens": 89825974.0, + "step": 6638 + }, + { + "epoch": 0.5474560897171601, + "grad_norm": 1.9230496883392334, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7494596838951111, + "num_tokens": 89857601.0, + "step": 6639 + }, + { + "epoch": 0.5475385503422115, + "grad_norm": 1.906356692314148, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7531423568725586, + "num_tokens": 89889218.0, + "step": 6640 + }, + { + "epoch": 0.5476210109672631, + "grad_norm": 2.165181875228882, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7304255962371826, + "num_tokens": 89916011.0, + "step": 6641 + }, + { + "epoch": 0.5477034715923147, + "grad_norm": 1.9111461639404297, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7247896790504456, + "num_tokens": 89947896.0, + "step": 6642 + }, + { + "epoch": 0.5477859322173662, + "grad_norm": 1.8219484090805054, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7180051803588867, + "num_tokens": 89984077.0, + "step": 6643 + }, + { + "epoch": 0.5478683928424177, + "grad_norm": 1.7911632061004639, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7414343357086182, + "num_tokens": 90020922.0, + "step": 6644 + }, + { + "epoch": 0.5479508534674693, + "grad_norm": 1.947115421295166, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7557429671287537, + "num_tokens": 90052939.0, + "step": 6645 + }, + { + "epoch": 0.5480333140925209, + "grad_norm": 2.001544237136841, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7344021797180176, + "num_tokens": 90081198.0, + "step": 6646 + }, + { + "epoch": 0.5481157747175723, + "grad_norm": 2.0113368034362793, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7634316682815552, + "num_tokens": 90109008.0, + "step": 6647 + }, + { + "epoch": 0.5481982353426239, + "grad_norm": 1.8481377363204956, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7692174911499023, + "num_tokens": 90140401.0, + "step": 6648 + }, + { + "epoch": 0.5482806959676755, + "grad_norm": 1.8365448713302612, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7584412693977356, + "num_tokens": 90172463.0, + "step": 6649 + }, + { + "epoch": 0.5483631565927269, + "grad_norm": 1.8586441278457642, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7676598429679871, + "num_tokens": 90204909.0, + "step": 6650 + }, + { + "epoch": 0.5484456172177785, + "grad_norm": 2.484652519226074, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7291010618209839, + "num_tokens": 90224096.0, + "step": 6651 + }, + { + "epoch": 0.5485280778428301, + "grad_norm": 2.1018097400665283, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7253175973892212, + "num_tokens": 90250568.0, + "step": 6652 + }, + { + "epoch": 0.5486105384678815, + "grad_norm": 1.8341718912124634, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7502276301383972, + "num_tokens": 90284010.0, + "step": 6653 + }, + { + "epoch": 0.5486929990929331, + "grad_norm": 1.8682931661605835, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7664984464645386, + "num_tokens": 90315041.0, + "step": 6654 + }, + { + "epoch": 0.5487754597179847, + "grad_norm": 1.8437628746032715, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.757354736328125, + "num_tokens": 90345779.0, + "step": 6655 + }, + { + "epoch": 0.5488579203430362, + "grad_norm": 2.196054458618164, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7349908351898193, + "num_tokens": 90369951.0, + "step": 6656 + }, + { + "epoch": 0.5489403809680877, + "grad_norm": 2.1490259170532227, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7507524490356445, + "num_tokens": 90396166.0, + "step": 6657 + }, + { + "epoch": 0.5490228415931393, + "grad_norm": 2.0715062618255615, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6828737258911133, + "num_tokens": 90427562.0, + "step": 6658 + }, + { + "epoch": 0.5491053022181908, + "grad_norm": 1.9233043193817139, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7467273473739624, + "num_tokens": 90456832.0, + "step": 6659 + }, + { + "epoch": 0.5491877628432423, + "grad_norm": 2.0256688594818115, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7343869805335999, + "num_tokens": 90488673.0, + "step": 6660 + }, + { + "epoch": 0.5492702234682939, + "grad_norm": 2.123394012451172, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7498035430908203, + "num_tokens": 90513907.0, + "step": 6661 + }, + { + "epoch": 0.5493526840933455, + "grad_norm": 2.0213592052459717, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7673177123069763, + "num_tokens": 90539738.0, + "step": 6662 + }, + { + "epoch": 0.5494351447183969, + "grad_norm": 2.6187100410461426, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7264410257339478, + "num_tokens": 90558277.0, + "step": 6663 + }, + { + "epoch": 0.5495176053434485, + "grad_norm": 2.3407838344573975, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7441200017929077, + "num_tokens": 90580857.0, + "step": 6664 + }, + { + "epoch": 0.5496000659685001, + "grad_norm": 2.1163442134857178, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7370912432670593, + "num_tokens": 90608686.0, + "step": 6665 + }, + { + "epoch": 0.5496825265935515, + "grad_norm": 2.3635337352752686, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7121966481208801, + "num_tokens": 90633396.0, + "step": 6666 + }, + { + "epoch": 0.5497649872186031, + "grad_norm": 2.0308828353881836, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.7049190998077393, + "num_tokens": 90664068.0, + "step": 6667 + }, + { + "epoch": 0.5498474478436547, + "grad_norm": 1.9116010665893555, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.759993851184845, + "num_tokens": 90695808.0, + "step": 6668 + }, + { + "epoch": 0.5499299084687062, + "grad_norm": 1.7597516775131226, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7436836957931519, + "num_tokens": 90730488.0, + "step": 6669 + }, + { + "epoch": 0.5500123690937577, + "grad_norm": 1.6716481447219849, + "learning_rate": 1e-06, + "loss": 0.6135, + "mean_token_accuracy": 0.7974610328674316, + "num_tokens": 90764782.0, + "step": 6670 + }, + { + "epoch": 0.5500948297188093, + "grad_norm": 1.9036359786987305, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7590444087982178, + "num_tokens": 90794557.0, + "step": 6671 + }, + { + "epoch": 0.5501772903438608, + "grad_norm": 2.125506639480591, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7591603994369507, + "num_tokens": 90819624.0, + "step": 6672 + }, + { + "epoch": 0.5502597509689123, + "grad_norm": 1.8896852731704712, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7401827573776245, + "num_tokens": 90851285.0, + "step": 6673 + }, + { + "epoch": 0.5503422115939639, + "grad_norm": 2.1142542362213135, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7142378091812134, + "num_tokens": 90879761.0, + "step": 6674 + }, + { + "epoch": 0.5504246722190154, + "grad_norm": 1.7796865701675415, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.7830448150634766, + "num_tokens": 90911241.0, + "step": 6675 + }, + { + "epoch": 0.5505071328440669, + "grad_norm": 1.8784090280532837, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.712081789970398, + "num_tokens": 90944928.0, + "step": 6676 + }, + { + "epoch": 0.5505895934691185, + "grad_norm": 1.9723284244537354, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7659791707992554, + "num_tokens": 90971951.0, + "step": 6677 + }, + { + "epoch": 0.5506720540941701, + "grad_norm": 2.0305492877960205, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7445800304412842, + "num_tokens": 90998637.0, + "step": 6678 + }, + { + "epoch": 0.5507545147192215, + "grad_norm": 2.000185966491699, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7636159658432007, + "num_tokens": 91027008.0, + "step": 6679 + }, + { + "epoch": 0.5508369753442731, + "grad_norm": 2.0077247619628906, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7064085602760315, + "num_tokens": 91058864.0, + "step": 6680 + }, + { + "epoch": 0.5509194359693247, + "grad_norm": 1.8966941833496094, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7459635734558105, + "num_tokens": 91093309.0, + "step": 6681 + }, + { + "epoch": 0.5510018965943762, + "grad_norm": 1.9753761291503906, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7640442848205566, + "num_tokens": 91124670.0, + "step": 6682 + }, + { + "epoch": 0.5510843572194277, + "grad_norm": 2.175694227218628, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7425333261489868, + "num_tokens": 91150668.0, + "step": 6683 + }, + { + "epoch": 0.5511668178444793, + "grad_norm": 1.865796685218811, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7469093799591064, + "num_tokens": 91183152.0, + "step": 6684 + }, + { + "epoch": 0.5512492784695308, + "grad_norm": 1.641124963760376, + "learning_rate": 1e-06, + "loss": 0.5623, + "mean_token_accuracy": 0.812928318977356, + "num_tokens": 91219700.0, + "step": 6685 + }, + { + "epoch": 0.5513317390945823, + "grad_norm": 2.1596579551696777, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7342572808265686, + "num_tokens": 91246991.0, + "step": 6686 + }, + { + "epoch": 0.5514141997196339, + "grad_norm": 1.8927098512649536, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7182777523994446, + "num_tokens": 91278840.0, + "step": 6687 + }, + { + "epoch": 0.5514966603446854, + "grad_norm": 1.9823883771896362, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7291014194488525, + "num_tokens": 91307397.0, + "step": 6688 + }, + { + "epoch": 0.5515791209697369, + "grad_norm": 1.8770700693130493, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7236557006835938, + "num_tokens": 91342204.0, + "step": 6689 + }, + { + "epoch": 0.5516615815947885, + "grad_norm": 2.011275291442871, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7722374200820923, + "num_tokens": 91371303.0, + "step": 6690 + }, + { + "epoch": 0.5517440422198401, + "grad_norm": 2.282521963119507, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7338541746139526, + "num_tokens": 91396228.0, + "step": 6691 + }, + { + "epoch": 0.5518265028448915, + "grad_norm": 2.0535104274749756, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7458715438842773, + "num_tokens": 91424948.0, + "step": 6692 + }, + { + "epoch": 0.5519089634699431, + "grad_norm": 1.998702883720398, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7205187678337097, + "num_tokens": 91453617.0, + "step": 6693 + }, + { + "epoch": 0.5519914240949947, + "grad_norm": 1.9470359086990356, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7354167699813843, + "num_tokens": 91483889.0, + "step": 6694 + }, + { + "epoch": 0.5520738847200461, + "grad_norm": 2.176745891571045, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7252838015556335, + "num_tokens": 91510643.0, + "step": 6695 + }, + { + "epoch": 0.5521563453450977, + "grad_norm": 1.8594624996185303, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7238726615905762, + "num_tokens": 91544808.0, + "step": 6696 + }, + { + "epoch": 0.5522388059701493, + "grad_norm": 2.2236344814300537, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7158403396606445, + "num_tokens": 91569527.0, + "step": 6697 + }, + { + "epoch": 0.5523212665952008, + "grad_norm": 1.8926680088043213, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7221267819404602, + "num_tokens": 91600862.0, + "step": 6698 + }, + { + "epoch": 0.5524037272202523, + "grad_norm": 2.0696728229522705, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7363848686218262, + "num_tokens": 91630290.0, + "step": 6699 + }, + { + "epoch": 0.5524861878453039, + "grad_norm": 1.8481189012527466, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7394852638244629, + "num_tokens": 91662266.0, + "step": 6700 + }, + { + "epoch": 0.5525686484703554, + "grad_norm": 2.311206340789795, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7090895771980286, + "num_tokens": 91686709.0, + "step": 6701 + }, + { + "epoch": 0.5526511090954069, + "grad_norm": 1.809342384338379, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7718837261199951, + "num_tokens": 91720953.0, + "step": 6702 + }, + { + "epoch": 0.5527335697204585, + "grad_norm": 2.082494020462036, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7383629679679871, + "num_tokens": 91748358.0, + "step": 6703 + }, + { + "epoch": 0.55281603034551, + "grad_norm": 2.1161186695098877, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7679281234741211, + "num_tokens": 91773649.0, + "step": 6704 + }, + { + "epoch": 0.5528984909705615, + "grad_norm": 2.1459286212921143, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.750464677810669, + "num_tokens": 91799533.0, + "step": 6705 + }, + { + "epoch": 0.5529809515956131, + "grad_norm": 2.4364941120147705, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.751838743686676, + "num_tokens": 91819950.0, + "step": 6706 + }, + { + "epoch": 0.5530634122206647, + "grad_norm": 1.8907532691955566, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.7572659254074097, + "num_tokens": 91850295.0, + "step": 6707 + }, + { + "epoch": 0.5531458728457161, + "grad_norm": 2.1585023403167725, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7277176380157471, + "num_tokens": 91876402.0, + "step": 6708 + }, + { + "epoch": 0.5532283334707677, + "grad_norm": 2.134838104248047, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7234921455383301, + "num_tokens": 91905898.0, + "step": 6709 + }, + { + "epoch": 0.5533107940958193, + "grad_norm": 2.0335519313812256, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7650642395019531, + "num_tokens": 91936696.0, + "step": 6710 + }, + { + "epoch": 0.5533932547208708, + "grad_norm": 1.9574118852615356, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7393430471420288, + "num_tokens": 91966306.0, + "step": 6711 + }, + { + "epoch": 0.5534757153459223, + "grad_norm": 2.1683385372161865, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7550409436225891, + "num_tokens": 91993920.0, + "step": 6712 + }, + { + "epoch": 0.5535581759709739, + "grad_norm": 1.7851165533065796, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.7989106774330139, + "num_tokens": 92025748.0, + "step": 6713 + }, + { + "epoch": 0.5536406365960254, + "grad_norm": 2.1302425861358643, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7444808483123779, + "num_tokens": 92049512.0, + "step": 6714 + }, + { + "epoch": 0.5537230972210769, + "grad_norm": 1.956220269203186, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7187817692756653, + "num_tokens": 92077672.0, + "step": 6715 + }, + { + "epoch": 0.5538055578461285, + "grad_norm": 1.9130170345306396, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7683367729187012, + "num_tokens": 92106761.0, + "step": 6716 + }, + { + "epoch": 0.55388801847118, + "grad_norm": 2.033372163772583, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7362805008888245, + "num_tokens": 92134048.0, + "step": 6717 + }, + { + "epoch": 0.5539704790962315, + "grad_norm": 1.7031548023223877, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7603217959403992, + "num_tokens": 92171957.0, + "step": 6718 + }, + { + "epoch": 0.5540529397212831, + "grad_norm": 1.642187237739563, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7750610113143921, + "num_tokens": 92207423.0, + "step": 6719 + }, + { + "epoch": 0.5541354003463346, + "grad_norm": 2.103192090988159, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7275139093399048, + "num_tokens": 92236208.0, + "step": 6720 + }, + { + "epoch": 0.5542178609713861, + "grad_norm": 1.8764139413833618, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7622915506362915, + "num_tokens": 92266318.0, + "step": 6721 + }, + { + "epoch": 0.5543003215964377, + "grad_norm": 2.4392683506011963, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7269418239593506, + "num_tokens": 92289870.0, + "step": 6722 + }, + { + "epoch": 0.5543827822214893, + "grad_norm": 2.056882381439209, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7455100417137146, + "num_tokens": 92315870.0, + "step": 6723 + }, + { + "epoch": 0.5544652428465408, + "grad_norm": 2.25774884223938, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7512248754501343, + "num_tokens": 92338776.0, + "step": 6724 + }, + { + "epoch": 0.5545477034715923, + "grad_norm": 2.082592725753784, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7504403591156006, + "num_tokens": 92368823.0, + "step": 6725 + }, + { + "epoch": 0.5546301640966439, + "grad_norm": 2.023287534713745, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7256393432617188, + "num_tokens": 92397342.0, + "step": 6726 + }, + { + "epoch": 0.5547126247216954, + "grad_norm": 1.9977476596832275, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7730069756507874, + "num_tokens": 92424348.0, + "step": 6727 + }, + { + "epoch": 0.5547950853467469, + "grad_norm": 1.6561819314956665, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.771638035774231, + "num_tokens": 92460638.0, + "step": 6728 + }, + { + "epoch": 0.5548775459717985, + "grad_norm": 1.9218863248825073, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7826846837997437, + "num_tokens": 92487621.0, + "step": 6729 + }, + { + "epoch": 0.55496000659685, + "grad_norm": 1.7995620965957642, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7800207138061523, + "num_tokens": 92521434.0, + "step": 6730 + }, + { + "epoch": 0.5550424672219015, + "grad_norm": 1.782820463180542, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7612950801849365, + "num_tokens": 92555484.0, + "step": 6731 + }, + { + "epoch": 0.5551249278469531, + "grad_norm": 1.8124948740005493, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.7974627017974854, + "num_tokens": 92584817.0, + "step": 6732 + }, + { + "epoch": 0.5552073884720046, + "grad_norm": 1.9368088245391846, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7443318367004395, + "num_tokens": 92615197.0, + "step": 6733 + }, + { + "epoch": 0.5552898490970561, + "grad_norm": 2.312645673751831, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7067458629608154, + "num_tokens": 92638135.0, + "step": 6734 + }, + { + "epoch": 0.5553723097221077, + "grad_norm": 1.8296890258789062, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7836117744445801, + "num_tokens": 92673597.0, + "step": 6735 + }, + { + "epoch": 0.5554547703471592, + "grad_norm": 1.8786782026290894, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7504139542579651, + "num_tokens": 92707261.0, + "step": 6736 + }, + { + "epoch": 0.5555372309722107, + "grad_norm": 1.9437458515167236, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7336419820785522, + "num_tokens": 92738013.0, + "step": 6737 + }, + { + "epoch": 0.5556196915972623, + "grad_norm": 1.8224127292633057, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7816107273101807, + "num_tokens": 92769818.0, + "step": 6738 + }, + { + "epoch": 0.5557021522223139, + "grad_norm": 1.871906042098999, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7719900012016296, + "num_tokens": 92800048.0, + "step": 6739 + }, + { + "epoch": 0.5557846128473654, + "grad_norm": 1.9670898914337158, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7715888023376465, + "num_tokens": 92830210.0, + "step": 6740 + }, + { + "epoch": 0.5558670734724169, + "grad_norm": 1.8653961420059204, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7678582668304443, + "num_tokens": 92861338.0, + "step": 6741 + }, + { + "epoch": 0.5559495340974685, + "grad_norm": 2.0805857181549072, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7598656415939331, + "num_tokens": 92891481.0, + "step": 6742 + }, + { + "epoch": 0.55603199472252, + "grad_norm": 2.0423316955566406, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7479648590087891, + "num_tokens": 92919429.0, + "step": 6743 + }, + { + "epoch": 0.5561144553475715, + "grad_norm": 1.8918495178222656, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7533484101295471, + "num_tokens": 92952540.0, + "step": 6744 + }, + { + "epoch": 0.5561969159726231, + "grad_norm": 1.8189524412155151, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7461158633232117, + "num_tokens": 92989285.0, + "step": 6745 + }, + { + "epoch": 0.5562793765976746, + "grad_norm": 2.045851707458496, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7513988614082336, + "num_tokens": 93019149.0, + "step": 6746 + }, + { + "epoch": 0.5563618372227261, + "grad_norm": 1.7509331703186035, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.768600344657898, + "num_tokens": 93052657.0, + "step": 6747 + }, + { + "epoch": 0.5564442978477777, + "grad_norm": 2.2338860034942627, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7169038653373718, + "num_tokens": 93076844.0, + "step": 6748 + }, + { + "epoch": 0.5565267584728292, + "grad_norm": 1.9128258228302002, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7665014266967773, + "num_tokens": 93107486.0, + "step": 6749 + }, + { + "epoch": 0.5566092190978807, + "grad_norm": 1.858837604522705, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7551698684692383, + "num_tokens": 93140235.0, + "step": 6750 + }, + { + "epoch": 0.5566916797229323, + "grad_norm": 1.896964192390442, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7225580215454102, + "num_tokens": 93175092.0, + "step": 6751 + }, + { + "epoch": 0.5567741403479839, + "grad_norm": 1.8407096862792969, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7511321902275085, + "num_tokens": 93207242.0, + "step": 6752 + }, + { + "epoch": 0.5568566009730354, + "grad_norm": 2.328442096710205, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7100516557693481, + "num_tokens": 93231224.0, + "step": 6753 + }, + { + "epoch": 0.5569390615980869, + "grad_norm": 2.353480339050293, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7203889489173889, + "num_tokens": 93254073.0, + "step": 6754 + }, + { + "epoch": 0.5570215222231385, + "grad_norm": 1.9569168090820312, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7411669492721558, + "num_tokens": 93285081.0, + "step": 6755 + }, + { + "epoch": 0.55710398284819, + "grad_norm": 2.140981912612915, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7308374047279358, + "num_tokens": 93312679.0, + "step": 6756 + }, + { + "epoch": 0.5571864434732415, + "grad_norm": 1.7342591285705566, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7795776128768921, + "num_tokens": 93347625.0, + "step": 6757 + }, + { + "epoch": 0.5572689040982931, + "grad_norm": 1.9527052640914917, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7309383749961853, + "num_tokens": 93379202.0, + "step": 6758 + }, + { + "epoch": 0.5573513647233446, + "grad_norm": 1.8795013427734375, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7501182556152344, + "num_tokens": 93411959.0, + "step": 6759 + }, + { + "epoch": 0.5574338253483961, + "grad_norm": 2.1027908325195312, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.7580084800720215, + "num_tokens": 93438655.0, + "step": 6760 + }, + { + "epoch": 0.5575162859734477, + "grad_norm": 2.2356960773468018, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7171341180801392, + "num_tokens": 93463263.0, + "step": 6761 + }, + { + "epoch": 0.5575987465984992, + "grad_norm": 1.8937095403671265, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7747746109962463, + "num_tokens": 93496123.0, + "step": 6762 + }, + { + "epoch": 0.5576812072235507, + "grad_norm": 1.81587553024292, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7866137623786926, + "num_tokens": 93529421.0, + "step": 6763 + }, + { + "epoch": 0.5577636678486023, + "grad_norm": 2.1568613052368164, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.7695522904396057, + "num_tokens": 93553934.0, + "step": 6764 + }, + { + "epoch": 0.5578461284736538, + "grad_norm": 2.265740156173706, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7413538694381714, + "num_tokens": 93579586.0, + "step": 6765 + }, + { + "epoch": 0.5579285890987054, + "grad_norm": 2.076176166534424, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7360153198242188, + "num_tokens": 93608324.0, + "step": 6766 + }, + { + "epoch": 0.5580110497237569, + "grad_norm": 1.9713265895843506, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7796520590782166, + "num_tokens": 93638846.0, + "step": 6767 + }, + { + "epoch": 0.5580935103488085, + "grad_norm": 1.8137515783309937, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7456499338150024, + "num_tokens": 93671995.0, + "step": 6768 + }, + { + "epoch": 0.55817597097386, + "grad_norm": 1.981110692024231, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7363598346710205, + "num_tokens": 93701171.0, + "step": 6769 + }, + { + "epoch": 0.5582584315989115, + "grad_norm": 1.9200242757797241, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.6964403986930847, + "num_tokens": 93733380.0, + "step": 6770 + }, + { + "epoch": 0.5583408922239631, + "grad_norm": 2.151747703552246, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7760021686553955, + "num_tokens": 93758871.0, + "step": 6771 + }, + { + "epoch": 0.5584233528490146, + "grad_norm": 1.8352106809616089, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7785190343856812, + "num_tokens": 93790768.0, + "step": 6772 + }, + { + "epoch": 0.5585058134740661, + "grad_norm": 1.731239676475525, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7735612392425537, + "num_tokens": 93828883.0, + "step": 6773 + }, + { + "epoch": 0.5585882740991177, + "grad_norm": 1.8843814134597778, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7612791061401367, + "num_tokens": 93859408.0, + "step": 6774 + }, + { + "epoch": 0.5586707347241692, + "grad_norm": 1.7906345129013062, + "learning_rate": 1e-06, + "loss": 0.7738, + "mean_token_accuracy": 0.7552271485328674, + "num_tokens": 93891706.0, + "step": 6775 + }, + { + "epoch": 0.5587531953492207, + "grad_norm": 2.0677497386932373, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.729834258556366, + "num_tokens": 93921369.0, + "step": 6776 + }, + { + "epoch": 0.5588356559742723, + "grad_norm": 2.254556179046631, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7146758437156677, + "num_tokens": 93947157.0, + "step": 6777 + }, + { + "epoch": 0.5589181165993238, + "grad_norm": 1.9154613018035889, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7494671940803528, + "num_tokens": 93977504.0, + "step": 6778 + }, + { + "epoch": 0.5590005772243754, + "grad_norm": 1.780202031135559, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.7878604531288147, + "num_tokens": 94009834.0, + "step": 6779 + }, + { + "epoch": 0.5590830378494269, + "grad_norm": 2.3587820529937744, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.742559552192688, + "num_tokens": 94032038.0, + "step": 6780 + }, + { + "epoch": 0.5591654984744784, + "grad_norm": 2.0105817317962646, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7492252588272095, + "num_tokens": 94060116.0, + "step": 6781 + }, + { + "epoch": 0.55924795909953, + "grad_norm": 2.0584542751312256, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.746425986289978, + "num_tokens": 94086441.0, + "step": 6782 + }, + { + "epoch": 0.5593304197245815, + "grad_norm": 2.1673271656036377, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6883785724639893, + "num_tokens": 94112530.0, + "step": 6783 + }, + { + "epoch": 0.5594128803496331, + "grad_norm": 1.7007380723953247, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7808623313903809, + "num_tokens": 94148205.0, + "step": 6784 + }, + { + "epoch": 0.5594953409746846, + "grad_norm": 1.8926810026168823, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7342845797538757, + "num_tokens": 94178206.0, + "step": 6785 + }, + { + "epoch": 0.5595778015997361, + "grad_norm": 1.9494783878326416, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7761488556861877, + "num_tokens": 94204159.0, + "step": 6786 + }, + { + "epoch": 0.5596602622247877, + "grad_norm": 1.958795428276062, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7561451196670532, + "num_tokens": 94233643.0, + "step": 6787 + }, + { + "epoch": 0.5597427228498392, + "grad_norm": 1.856161117553711, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.747150182723999, + "num_tokens": 94269041.0, + "step": 6788 + }, + { + "epoch": 0.5598251834748907, + "grad_norm": 2.0519790649414062, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.755024790763855, + "num_tokens": 94297816.0, + "step": 6789 + }, + { + "epoch": 0.5599076440999423, + "grad_norm": 2.075183153152466, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7244605422019958, + "num_tokens": 94326169.0, + "step": 6790 + }, + { + "epoch": 0.5599901047249938, + "grad_norm": 1.9215948581695557, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7679780125617981, + "num_tokens": 94355090.0, + "step": 6791 + }, + { + "epoch": 0.5600725653500453, + "grad_norm": 1.7722333669662476, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7636321783065796, + "num_tokens": 94390049.0, + "step": 6792 + }, + { + "epoch": 0.5601550259750969, + "grad_norm": 2.0347278118133545, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7624397873878479, + "num_tokens": 94417020.0, + "step": 6793 + }, + { + "epoch": 0.5602374866001484, + "grad_norm": 1.9287526607513428, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7310587167739868, + "num_tokens": 94448597.0, + "step": 6794 + }, + { + "epoch": 0.5603199472252, + "grad_norm": 1.8649178743362427, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7559074759483337, + "num_tokens": 94478494.0, + "step": 6795 + }, + { + "epoch": 0.5604024078502515, + "grad_norm": 2.299290180206299, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7327685952186584, + "num_tokens": 94501527.0, + "step": 6796 + }, + { + "epoch": 0.560484868475303, + "grad_norm": 1.9552488327026367, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7490662336349487, + "num_tokens": 94531021.0, + "step": 6797 + }, + { + "epoch": 0.5605673291003546, + "grad_norm": 1.819555640220642, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.7553021907806396, + "num_tokens": 94567671.0, + "step": 6798 + }, + { + "epoch": 0.5606497897254061, + "grad_norm": 2.1063928604125977, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7203640937805176, + "num_tokens": 94594495.0, + "step": 6799 + }, + { + "epoch": 0.5607322503504577, + "grad_norm": 1.8870842456817627, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7938356995582581, + "num_tokens": 94622594.0, + "step": 6800 + }, + { + "epoch": 0.5608147109755092, + "grad_norm": 1.951253890991211, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7503652572631836, + "num_tokens": 94650739.0, + "step": 6801 + }, + { + "epoch": 0.5608971716005607, + "grad_norm": 1.9491029977798462, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7326855659484863, + "num_tokens": 94681729.0, + "step": 6802 + }, + { + "epoch": 0.5609796322256123, + "grad_norm": 2.081944227218628, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7240685224533081, + "num_tokens": 94708780.0, + "step": 6803 + }, + { + "epoch": 0.5610620928506638, + "grad_norm": 1.7902023792266846, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7414243221282959, + "num_tokens": 94741947.0, + "step": 6804 + }, + { + "epoch": 0.5611445534757153, + "grad_norm": 1.8633662462234497, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7667344808578491, + "num_tokens": 94773695.0, + "step": 6805 + }, + { + "epoch": 0.5612270141007669, + "grad_norm": 1.9322566986083984, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6940956115722656, + "num_tokens": 94806315.0, + "step": 6806 + }, + { + "epoch": 0.5613094747258184, + "grad_norm": 2.1128530502319336, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7272072434425354, + "num_tokens": 94833849.0, + "step": 6807 + }, + { + "epoch": 0.56139193535087, + "grad_norm": 1.7002034187316895, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7571974992752075, + "num_tokens": 94868129.0, + "step": 6808 + }, + { + "epoch": 0.5614743959759215, + "grad_norm": 2.0373573303222656, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7248657941818237, + "num_tokens": 94897869.0, + "step": 6809 + }, + { + "epoch": 0.561556856600973, + "grad_norm": 1.774814486503601, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7706738710403442, + "num_tokens": 94931132.0, + "step": 6810 + }, + { + "epoch": 0.5616393172260246, + "grad_norm": 1.7825965881347656, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7718201875686646, + "num_tokens": 94960822.0, + "step": 6811 + }, + { + "epoch": 0.5617217778510761, + "grad_norm": 1.752480387687683, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7234471440315247, + "num_tokens": 94995667.0, + "step": 6812 + }, + { + "epoch": 0.5618042384761277, + "grad_norm": 2.269968271255493, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7379993200302124, + "num_tokens": 95020425.0, + "step": 6813 + }, + { + "epoch": 0.5618866991011792, + "grad_norm": 1.9452142715454102, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7529969215393066, + "num_tokens": 95048957.0, + "step": 6814 + }, + { + "epoch": 0.5619691597262307, + "grad_norm": 2.0244998931884766, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7309961915016174, + "num_tokens": 95077581.0, + "step": 6815 + }, + { + "epoch": 0.5620516203512823, + "grad_norm": 2.2568767070770264, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7113121747970581, + "num_tokens": 95103795.0, + "step": 6816 + }, + { + "epoch": 0.5621340809763338, + "grad_norm": 2.2727370262145996, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7527875304222107, + "num_tokens": 95128911.0, + "step": 6817 + }, + { + "epoch": 0.5622165416013853, + "grad_norm": 1.994125485420227, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7625647783279419, + "num_tokens": 95154996.0, + "step": 6818 + }, + { + "epoch": 0.5622990022264369, + "grad_norm": 1.9093703031539917, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7673025727272034, + "num_tokens": 95183120.0, + "step": 6819 + }, + { + "epoch": 0.5623814628514884, + "grad_norm": 1.914419174194336, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.80028235912323, + "num_tokens": 95214765.0, + "step": 6820 + }, + { + "epoch": 0.56246392347654, + "grad_norm": 1.899139165878296, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7787929773330688, + "num_tokens": 95244924.0, + "step": 6821 + }, + { + "epoch": 0.5625463841015915, + "grad_norm": 2.0450849533081055, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.7530422210693359, + "num_tokens": 95270949.0, + "step": 6822 + }, + { + "epoch": 0.562628844726643, + "grad_norm": 1.8837532997131348, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.767035722732544, + "num_tokens": 95300987.0, + "step": 6823 + }, + { + "epoch": 0.5627113053516946, + "grad_norm": 1.921724557876587, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7491520047187805, + "num_tokens": 95330204.0, + "step": 6824 + }, + { + "epoch": 0.5627937659767461, + "grad_norm": 1.954482078552246, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7212954759597778, + "num_tokens": 95361268.0, + "step": 6825 + }, + { + "epoch": 0.5628762266017976, + "grad_norm": 2.2008705139160156, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7328752875328064, + "num_tokens": 95387206.0, + "step": 6826 + }, + { + "epoch": 0.5629586872268492, + "grad_norm": 2.0007641315460205, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7196170091629028, + "num_tokens": 95418387.0, + "step": 6827 + }, + { + "epoch": 0.5630411478519007, + "grad_norm": 1.8353270292282104, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7379672527313232, + "num_tokens": 95451925.0, + "step": 6828 + }, + { + "epoch": 0.5631236084769523, + "grad_norm": 2.0531787872314453, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7333745956420898, + "num_tokens": 95478031.0, + "step": 6829 + }, + { + "epoch": 0.5632060691020038, + "grad_norm": 2.06296443939209, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7600978016853333, + "num_tokens": 95505457.0, + "step": 6830 + }, + { + "epoch": 0.5632885297270553, + "grad_norm": 1.8576618432998657, + "learning_rate": 1e-06, + "loss": 0.7738, + "mean_token_accuracy": 0.752648115158081, + "num_tokens": 95536510.0, + "step": 6831 + }, + { + "epoch": 0.5633709903521069, + "grad_norm": 2.081827402114868, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7406569719314575, + "num_tokens": 95561529.0, + "step": 6832 + }, + { + "epoch": 0.5634534509771584, + "grad_norm": 2.0475144386291504, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.7564573884010315, + "num_tokens": 95589373.0, + "step": 6833 + }, + { + "epoch": 0.56353591160221, + "grad_norm": 1.8545852899551392, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7326096892356873, + "num_tokens": 95622476.0, + "step": 6834 + }, + { + "epoch": 0.5636183722272615, + "grad_norm": 2.1245651245117188, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7308908700942993, + "num_tokens": 95648056.0, + "step": 6835 + }, + { + "epoch": 0.563700832852313, + "grad_norm": 1.9067022800445557, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7450857162475586, + "num_tokens": 95680277.0, + "step": 6836 + }, + { + "epoch": 0.5637832934773646, + "grad_norm": 2.011016607284546, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7273584604263306, + "num_tokens": 95710287.0, + "step": 6837 + }, + { + "epoch": 0.5638657541024161, + "grad_norm": 1.720593810081482, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7596248388290405, + "num_tokens": 95745778.0, + "step": 6838 + }, + { + "epoch": 0.5639482147274676, + "grad_norm": 1.7033857107162476, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7711331844329834, + "num_tokens": 95783183.0, + "step": 6839 + }, + { + "epoch": 0.5640306753525192, + "grad_norm": 2.2245359420776367, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7158645391464233, + "num_tokens": 95811747.0, + "step": 6840 + }, + { + "epoch": 0.5641131359775707, + "grad_norm": 1.9795029163360596, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7841260433197021, + "num_tokens": 95838552.0, + "step": 6841 + }, + { + "epoch": 0.5641955966026222, + "grad_norm": 1.7720621824264526, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7667323350906372, + "num_tokens": 95873929.0, + "step": 6842 + }, + { + "epoch": 0.5642780572276738, + "grad_norm": 1.7314549684524536, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7638253569602966, + "num_tokens": 95909525.0, + "step": 6843 + }, + { + "epoch": 0.5643605178527253, + "grad_norm": 1.8859783411026, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7629210352897644, + "num_tokens": 95940118.0, + "step": 6844 + }, + { + "epoch": 0.5644429784777769, + "grad_norm": 1.8219752311706543, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7615060210227966, + "num_tokens": 95971058.0, + "step": 6845 + }, + { + "epoch": 0.5645254391028284, + "grad_norm": 1.8088674545288086, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7804311513900757, + "num_tokens": 96005277.0, + "step": 6846 + }, + { + "epoch": 0.56460789972788, + "grad_norm": 1.8606795072555542, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7382469177246094, + "num_tokens": 96041274.0, + "step": 6847 + }, + { + "epoch": 0.5646903603529315, + "grad_norm": 2.0218493938446045, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7479525208473206, + "num_tokens": 96069769.0, + "step": 6848 + }, + { + "epoch": 0.564772820977983, + "grad_norm": 2.148000717163086, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7157703042030334, + "num_tokens": 96096595.0, + "step": 6849 + }, + { + "epoch": 0.5648552816030346, + "grad_norm": 2.1065571308135986, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7414124011993408, + "num_tokens": 96123658.0, + "step": 6850 + }, + { + "epoch": 0.5649377422280861, + "grad_norm": 2.1026432514190674, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7539808750152588, + "num_tokens": 96151209.0, + "step": 6851 + }, + { + "epoch": 0.5650202028531376, + "grad_norm": 1.711580753326416, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7809175848960876, + "num_tokens": 96186810.0, + "step": 6852 + }, + { + "epoch": 0.5651026634781892, + "grad_norm": 2.067514657974243, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7378358840942383, + "num_tokens": 96213959.0, + "step": 6853 + }, + { + "epoch": 0.5651851241032407, + "grad_norm": 1.8452614545822144, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7169036865234375, + "num_tokens": 96248249.0, + "step": 6854 + }, + { + "epoch": 0.5652675847282922, + "grad_norm": 2.0983693599700928, + "learning_rate": 1e-06, + "loss": 1.0402, + "mean_token_accuracy": 0.6838090419769287, + "num_tokens": 96277710.0, + "step": 6855 + }, + { + "epoch": 0.5653500453533438, + "grad_norm": 2.068476438522339, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.772545576095581, + "num_tokens": 96305348.0, + "step": 6856 + }, + { + "epoch": 0.5654325059783953, + "grad_norm": 1.8762092590332031, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7441608905792236, + "num_tokens": 96335972.0, + "step": 6857 + }, + { + "epoch": 0.5655149666034468, + "grad_norm": 2.059452772140503, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7516241073608398, + "num_tokens": 96360943.0, + "step": 6858 + }, + { + "epoch": 0.5655974272284984, + "grad_norm": 2.1323983669281006, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7818752527236938, + "num_tokens": 96383027.0, + "step": 6859 + }, + { + "epoch": 0.56567988785355, + "grad_norm": 2.1842448711395264, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7504755258560181, + "num_tokens": 96407366.0, + "step": 6860 + }, + { + "epoch": 0.5657623484786015, + "grad_norm": 1.7491711378097534, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7522972822189331, + "num_tokens": 96442639.0, + "step": 6861 + }, + { + "epoch": 0.565844809103653, + "grad_norm": 1.8286360502243042, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7501780986785889, + "num_tokens": 96475377.0, + "step": 6862 + }, + { + "epoch": 0.5659272697287046, + "grad_norm": 2.1813278198242188, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7417470216751099, + "num_tokens": 96502246.0, + "step": 6863 + }, + { + "epoch": 0.5660097303537561, + "grad_norm": 1.7350355386734009, + "learning_rate": 1e-06, + "loss": 0.5667, + "mean_token_accuracy": 0.819035530090332, + "num_tokens": 96529743.0, + "step": 6864 + }, + { + "epoch": 0.5660921909788076, + "grad_norm": 2.0564393997192383, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7263815402984619, + "num_tokens": 96558472.0, + "step": 6865 + }, + { + "epoch": 0.5661746516038592, + "grad_norm": 2.221723794937134, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7679243683815002, + "num_tokens": 96579767.0, + "step": 6866 + }, + { + "epoch": 0.5662571122289107, + "grad_norm": 1.878143548965454, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7523869872093201, + "num_tokens": 96611578.0, + "step": 6867 + }, + { + "epoch": 0.5663395728539622, + "grad_norm": 1.8406825065612793, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7544007301330566, + "num_tokens": 96647391.0, + "step": 6868 + }, + { + "epoch": 0.5664220334790138, + "grad_norm": 2.260529041290283, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7246242761611938, + "num_tokens": 96672312.0, + "step": 6869 + }, + { + "epoch": 0.5665044941040653, + "grad_norm": 2.0593650341033936, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7484649419784546, + "num_tokens": 96699117.0, + "step": 6870 + }, + { + "epoch": 0.5665869547291168, + "grad_norm": 1.7516601085662842, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.7964866161346436, + "num_tokens": 96730090.0, + "step": 6871 + }, + { + "epoch": 0.5666694153541684, + "grad_norm": 2.0462048053741455, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.734338104724884, + "num_tokens": 96760003.0, + "step": 6872 + }, + { + "epoch": 0.5667518759792199, + "grad_norm": 2.040208578109741, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7590505480766296, + "num_tokens": 96786855.0, + "step": 6873 + }, + { + "epoch": 0.5668343366042714, + "grad_norm": 1.930877447128296, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7291221618652344, + "num_tokens": 96818046.0, + "step": 6874 + }, + { + "epoch": 0.566916797229323, + "grad_norm": 2.0909082889556885, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7554088234901428, + "num_tokens": 96843201.0, + "step": 6875 + }, + { + "epoch": 0.5669992578543745, + "grad_norm": 2.046405553817749, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7835789322853088, + "num_tokens": 96869073.0, + "step": 6876 + }, + { + "epoch": 0.5670817184794261, + "grad_norm": 1.9191728830337524, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7482241988182068, + "num_tokens": 96901807.0, + "step": 6877 + }, + { + "epoch": 0.5671641791044776, + "grad_norm": 1.9722994565963745, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7190313339233398, + "num_tokens": 96932554.0, + "step": 6878 + }, + { + "epoch": 0.5672466397295292, + "grad_norm": 2.0912864208221436, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7550334930419922, + "num_tokens": 96959800.0, + "step": 6879 + }, + { + "epoch": 0.5673291003545807, + "grad_norm": 2.2313170433044434, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7277029752731323, + "num_tokens": 96985691.0, + "step": 6880 + }, + { + "epoch": 0.5674115609796322, + "grad_norm": 1.7223504781723022, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7646007537841797, + "num_tokens": 97021911.0, + "step": 6881 + }, + { + "epoch": 0.5674940216046838, + "grad_norm": 2.330745220184326, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7142122983932495, + "num_tokens": 97044840.0, + "step": 6882 + }, + { + "epoch": 0.5675764822297353, + "grad_norm": 1.8956918716430664, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7845981121063232, + "num_tokens": 97074671.0, + "step": 6883 + }, + { + "epoch": 0.5676589428547868, + "grad_norm": 1.8823002576828003, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7463647723197937, + "num_tokens": 97107454.0, + "step": 6884 + }, + { + "epoch": 0.5677414034798384, + "grad_norm": 2.001525402069092, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7357832789421082, + "num_tokens": 97137638.0, + "step": 6885 + }, + { + "epoch": 0.5678238641048899, + "grad_norm": 2.127958297729492, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.7677293419837952, + "num_tokens": 97167414.0, + "step": 6886 + }, + { + "epoch": 0.5679063247299414, + "grad_norm": 2.1549768447875977, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7680619955062866, + "num_tokens": 97192690.0, + "step": 6887 + }, + { + "epoch": 0.567988785354993, + "grad_norm": 1.857363224029541, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7761844396591187, + "num_tokens": 97224957.0, + "step": 6888 + }, + { + "epoch": 0.5680712459800445, + "grad_norm": 2.267595052719116, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7211856842041016, + "num_tokens": 97246868.0, + "step": 6889 + }, + { + "epoch": 0.5681537066050961, + "grad_norm": 1.978392481803894, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7273886203765869, + "num_tokens": 97279143.0, + "step": 6890 + }, + { + "epoch": 0.5682361672301476, + "grad_norm": 1.5462769269943237, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7645062208175659, + "num_tokens": 97320586.0, + "step": 6891 + }, + { + "epoch": 0.5683186278551992, + "grad_norm": 1.9390980005264282, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7278910279273987, + "num_tokens": 97350460.0, + "step": 6892 + }, + { + "epoch": 0.5684010884802507, + "grad_norm": 1.8689087629318237, + "learning_rate": 1e-06, + "loss": 0.7465, + "mean_token_accuracy": 0.7561585307121277, + "num_tokens": 97379971.0, + "step": 6893 + }, + { + "epoch": 0.5684835491053022, + "grad_norm": 1.8388900756835938, + "learning_rate": 1e-06, + "loss": 0.5907, + "mean_token_accuracy": 0.8098294734954834, + "num_tokens": 97410121.0, + "step": 6894 + }, + { + "epoch": 0.5685660097303538, + "grad_norm": 1.8637737035751343, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7799959778785706, + "num_tokens": 97440658.0, + "step": 6895 + }, + { + "epoch": 0.5686484703554053, + "grad_norm": 2.0763869285583496, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.7492388486862183, + "num_tokens": 97467292.0, + "step": 6896 + }, + { + "epoch": 0.5687309309804568, + "grad_norm": 1.790260910987854, + "learning_rate": 1e-06, + "loss": 0.5901, + "mean_token_accuracy": 0.8098522424697876, + "num_tokens": 97501535.0, + "step": 6897 + }, + { + "epoch": 0.5688133916055084, + "grad_norm": 1.8139034509658813, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7566792964935303, + "num_tokens": 97535400.0, + "step": 6898 + }, + { + "epoch": 0.5688958522305599, + "grad_norm": 1.9730032682418823, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.704738974571228, + "num_tokens": 97566372.0, + "step": 6899 + }, + { + "epoch": 0.5689783128556114, + "grad_norm": 1.8000211715698242, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7331093549728394, + "num_tokens": 97601446.0, + "step": 6900 + }, + { + "epoch": 0.569060773480663, + "grad_norm": 1.9379204511642456, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7494216561317444, + "num_tokens": 97630040.0, + "step": 6901 + }, + { + "epoch": 0.5691432341057145, + "grad_norm": 2.017463445663452, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7375494241714478, + "num_tokens": 97658527.0, + "step": 6902 + }, + { + "epoch": 0.569225694730766, + "grad_norm": 2.1057705879211426, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7361167669296265, + "num_tokens": 97685460.0, + "step": 6903 + }, + { + "epoch": 0.5693081553558176, + "grad_norm": 2.0760915279388428, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7504895925521851, + "num_tokens": 97711084.0, + "step": 6904 + }, + { + "epoch": 0.5693906159808692, + "grad_norm": 1.8629060983657837, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7806950807571411, + "num_tokens": 97740010.0, + "step": 6905 + }, + { + "epoch": 0.5694730766059207, + "grad_norm": 1.7551813125610352, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.7859079241752625, + "num_tokens": 97772928.0, + "step": 6906 + }, + { + "epoch": 0.5695555372309722, + "grad_norm": 2.104393482208252, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7255847454071045, + "num_tokens": 97798918.0, + "step": 6907 + }, + { + "epoch": 0.5696379978560238, + "grad_norm": 1.8331248760223389, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7517526149749756, + "num_tokens": 97832747.0, + "step": 6908 + }, + { + "epoch": 0.5697204584810753, + "grad_norm": 2.0474016666412354, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7827115654945374, + "num_tokens": 97859935.0, + "step": 6909 + }, + { + "epoch": 0.5698029191061268, + "grad_norm": 1.9321155548095703, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.781682014465332, + "num_tokens": 97888631.0, + "step": 6910 + }, + { + "epoch": 0.5698853797311784, + "grad_norm": 1.9084001779556274, + "learning_rate": 1e-06, + "loss": 0.5854, + "mean_token_accuracy": 0.8057321310043335, + "num_tokens": 97914634.0, + "step": 6911 + }, + { + "epoch": 0.5699678403562299, + "grad_norm": 2.3448877334594727, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7405843138694763, + "num_tokens": 97935749.0, + "step": 6912 + }, + { + "epoch": 0.5700503009812814, + "grad_norm": 1.9615825414657593, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.7736972570419312, + "num_tokens": 97963316.0, + "step": 6913 + }, + { + "epoch": 0.570132761606333, + "grad_norm": 2.0344808101654053, + "learning_rate": 1e-06, + "loss": 0.535, + "mean_token_accuracy": 0.8235617876052856, + "num_tokens": 97991037.0, + "step": 6914 + }, + { + "epoch": 0.5702152222313845, + "grad_norm": 2.214156150817871, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7312747240066528, + "num_tokens": 98018474.0, + "step": 6915 + }, + { + "epoch": 0.570297682856436, + "grad_norm": 2.2641196250915527, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7554937601089478, + "num_tokens": 98042376.0, + "step": 6916 + }, + { + "epoch": 0.5703801434814876, + "grad_norm": 1.7432130575180054, + "learning_rate": 1e-06, + "loss": 0.7584, + "mean_token_accuracy": 0.7470363974571228, + "num_tokens": 98078153.0, + "step": 6917 + }, + { + "epoch": 0.5704626041065392, + "grad_norm": 1.8024556636810303, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7725435495376587, + "num_tokens": 98111777.0, + "step": 6918 + }, + { + "epoch": 0.5705450647315906, + "grad_norm": 2.055346727371216, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7092423439025879, + "num_tokens": 98141061.0, + "step": 6919 + }, + { + "epoch": 0.5706275253566422, + "grad_norm": 1.9063199758529663, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.771034300327301, + "num_tokens": 98167786.0, + "step": 6920 + }, + { + "epoch": 0.5707099859816938, + "grad_norm": 2.1460604667663574, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.7605319619178772, + "num_tokens": 98192687.0, + "step": 6921 + }, + { + "epoch": 0.5707924466067453, + "grad_norm": 1.9892067909240723, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7682396769523621, + "num_tokens": 98218137.0, + "step": 6922 + }, + { + "epoch": 0.5708749072317968, + "grad_norm": 2.2778139114379883, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7091888785362244, + "num_tokens": 98243700.0, + "step": 6923 + }, + { + "epoch": 0.5709573678568484, + "grad_norm": 1.925341010093689, + "learning_rate": 1e-06, + "loss": 0.7621, + "mean_token_accuracy": 0.7590267658233643, + "num_tokens": 98273588.0, + "step": 6924 + }, + { + "epoch": 0.5710398284818999, + "grad_norm": 1.9904907941818237, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7326698303222656, + "num_tokens": 98303087.0, + "step": 6925 + }, + { + "epoch": 0.5711222891069514, + "grad_norm": 2.0256900787353516, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7164039611816406, + "num_tokens": 98331302.0, + "step": 6926 + }, + { + "epoch": 0.571204749732003, + "grad_norm": 1.8421024084091187, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7774230241775513, + "num_tokens": 98362509.0, + "step": 6927 + }, + { + "epoch": 0.5712872103570545, + "grad_norm": 1.8613882064819336, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7731839418411255, + "num_tokens": 98396131.0, + "step": 6928 + }, + { + "epoch": 0.571369670982106, + "grad_norm": 2.3192718029022217, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7201566100120544, + "num_tokens": 98419380.0, + "step": 6929 + }, + { + "epoch": 0.5714521316071576, + "grad_norm": 1.9125295877456665, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7618396878242493, + "num_tokens": 98451679.0, + "step": 6930 + }, + { + "epoch": 0.5715345922322091, + "grad_norm": 2.0398480892181396, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7196313738822937, + "num_tokens": 98477446.0, + "step": 6931 + }, + { + "epoch": 0.5716170528572606, + "grad_norm": 1.8002532720565796, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7688696384429932, + "num_tokens": 98512494.0, + "step": 6932 + }, + { + "epoch": 0.5716995134823122, + "grad_norm": 2.2276711463928223, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.748410701751709, + "num_tokens": 98533831.0, + "step": 6933 + }, + { + "epoch": 0.5717819741073638, + "grad_norm": 1.8552112579345703, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.774551510810852, + "num_tokens": 98563755.0, + "step": 6934 + }, + { + "epoch": 0.5718644347324152, + "grad_norm": 1.7469539642333984, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7824530601501465, + "num_tokens": 98596882.0, + "step": 6935 + }, + { + "epoch": 0.5719468953574668, + "grad_norm": 1.9119373559951782, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7274450659751892, + "num_tokens": 98627374.0, + "step": 6936 + }, + { + "epoch": 0.5720293559825184, + "grad_norm": 1.9077192544937134, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7612211108207703, + "num_tokens": 98657861.0, + "step": 6937 + }, + { + "epoch": 0.5721118166075699, + "grad_norm": 2.0556533336639404, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7290351390838623, + "num_tokens": 98685106.0, + "step": 6938 + }, + { + "epoch": 0.5721942772326214, + "grad_norm": 2.0375771522521973, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7608687877655029, + "num_tokens": 98710687.0, + "step": 6939 + }, + { + "epoch": 0.572276737857673, + "grad_norm": 2.026898145675659, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7489960193634033, + "num_tokens": 98741709.0, + "step": 6940 + }, + { + "epoch": 0.5723591984827245, + "grad_norm": 2.132089853286743, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7358567714691162, + "num_tokens": 98767319.0, + "step": 6941 + }, + { + "epoch": 0.572441659107776, + "grad_norm": 1.8972234725952148, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7442190647125244, + "num_tokens": 98798659.0, + "step": 6942 + }, + { + "epoch": 0.5725241197328276, + "grad_norm": 2.2063136100769043, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7368661165237427, + "num_tokens": 98822286.0, + "step": 6943 + }, + { + "epoch": 0.5726065803578791, + "grad_norm": 1.8122270107269287, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7153058052062988, + "num_tokens": 98858557.0, + "step": 6944 + }, + { + "epoch": 0.5726890409829306, + "grad_norm": 2.255506992340088, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7487863302230835, + "num_tokens": 98885294.0, + "step": 6945 + }, + { + "epoch": 0.5727715016079822, + "grad_norm": 1.8445948362350464, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.7858409285545349, + "num_tokens": 98916099.0, + "step": 6946 + }, + { + "epoch": 0.5728539622330338, + "grad_norm": 2.0796432495117188, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7547679543495178, + "num_tokens": 98943163.0, + "step": 6947 + }, + { + "epoch": 0.5729364228580852, + "grad_norm": 1.7402684688568115, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7667961120605469, + "num_tokens": 98975261.0, + "step": 6948 + }, + { + "epoch": 0.5730188834831368, + "grad_norm": 1.867551326751709, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7490823864936829, + "num_tokens": 99005882.0, + "step": 6949 + }, + { + "epoch": 0.5731013441081884, + "grad_norm": 1.9603781700134277, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.7527393698692322, + "num_tokens": 99037369.0, + "step": 6950 + }, + { + "epoch": 0.5731838047332399, + "grad_norm": 2.038120985031128, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7194026708602905, + "num_tokens": 99069329.0, + "step": 6951 + }, + { + "epoch": 0.5732662653582914, + "grad_norm": 2.0318610668182373, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7799603343009949, + "num_tokens": 99094945.0, + "step": 6952 + }, + { + "epoch": 0.573348725983343, + "grad_norm": 1.8391051292419434, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7521578073501587, + "num_tokens": 99128121.0, + "step": 6953 + }, + { + "epoch": 0.5734311866083945, + "grad_norm": 2.0244293212890625, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.729809045791626, + "num_tokens": 99160239.0, + "step": 6954 + }, + { + "epoch": 0.573513647233446, + "grad_norm": 2.1346168518066406, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7400161027908325, + "num_tokens": 99187783.0, + "step": 6955 + }, + { + "epoch": 0.5735961078584976, + "grad_norm": 1.9761539697647095, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7368953227996826, + "num_tokens": 99217051.0, + "step": 6956 + }, + { + "epoch": 0.5736785684835491, + "grad_norm": 2.4458911418914795, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7514868974685669, + "num_tokens": 99244557.0, + "step": 6957 + }, + { + "epoch": 0.5737610291086006, + "grad_norm": 1.9750711917877197, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7402716875076294, + "num_tokens": 99276038.0, + "step": 6958 + }, + { + "epoch": 0.5738434897336522, + "grad_norm": 1.9773802757263184, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7829056978225708, + "num_tokens": 99304590.0, + "step": 6959 + }, + { + "epoch": 0.5739259503587038, + "grad_norm": 1.6952420473098755, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7698222398757935, + "num_tokens": 99340361.0, + "step": 6960 + }, + { + "epoch": 0.5740084109837552, + "grad_norm": 2.4212255477905273, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7363258600234985, + "num_tokens": 99363635.0, + "step": 6961 + }, + { + "epoch": 0.5740908716088068, + "grad_norm": 2.049910306930542, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7496503591537476, + "num_tokens": 99392472.0, + "step": 6962 + }, + { + "epoch": 0.5741733322338584, + "grad_norm": 2.1272048950195312, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7293083667755127, + "num_tokens": 99420248.0, + "step": 6963 + }, + { + "epoch": 0.5742557928589098, + "grad_norm": 1.9111104011535645, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7855507731437683, + "num_tokens": 99450777.0, + "step": 6964 + }, + { + "epoch": 0.5743382534839614, + "grad_norm": 2.0781548023223877, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7500205039978027, + "num_tokens": 99475361.0, + "step": 6965 + }, + { + "epoch": 0.574420714109013, + "grad_norm": 2.044114828109741, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7161114811897278, + "num_tokens": 99504477.0, + "step": 6966 + }, + { + "epoch": 0.5745031747340645, + "grad_norm": 2.2437808513641357, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.723142147064209, + "num_tokens": 99527658.0, + "step": 6967 + }, + { + "epoch": 0.574585635359116, + "grad_norm": 2.0785958766937256, + "learning_rate": 1e-06, + "loss": 0.781, + "mean_token_accuracy": 0.7519200444221497, + "num_tokens": 99553726.0, + "step": 6968 + }, + { + "epoch": 0.5746680959841676, + "grad_norm": 2.0213985443115234, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.756750762462616, + "num_tokens": 99583939.0, + "step": 6969 + }, + { + "epoch": 0.5747505566092191, + "grad_norm": 1.7619482278823853, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7164442539215088, + "num_tokens": 99624252.0, + "step": 6970 + }, + { + "epoch": 0.5748330172342706, + "grad_norm": 2.242011308670044, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7209761738777161, + "num_tokens": 99648447.0, + "step": 6971 + }, + { + "epoch": 0.5749154778593222, + "grad_norm": 1.918781042098999, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7418670654296875, + "num_tokens": 99680541.0, + "step": 6972 + }, + { + "epoch": 0.5749979384843737, + "grad_norm": 2.02238392829895, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7223809957504272, + "num_tokens": 99708207.0, + "step": 6973 + }, + { + "epoch": 0.5750803991094252, + "grad_norm": 1.9029783010482788, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.752372145652771, + "num_tokens": 99739233.0, + "step": 6974 + }, + { + "epoch": 0.5751628597344768, + "grad_norm": 1.8260961771011353, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7197383642196655, + "num_tokens": 99770230.0, + "step": 6975 + }, + { + "epoch": 0.5752453203595284, + "grad_norm": 1.9396991729736328, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7178664207458496, + "num_tokens": 99801710.0, + "step": 6976 + }, + { + "epoch": 0.5753277809845798, + "grad_norm": 1.9798314571380615, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7240667939186096, + "num_tokens": 99828598.0, + "step": 6977 + }, + { + "epoch": 0.5754102416096314, + "grad_norm": 1.9274269342422485, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7659944891929626, + "num_tokens": 99858209.0, + "step": 6978 + }, + { + "epoch": 0.575492702234683, + "grad_norm": 2.140076160430908, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7635669708251953, + "num_tokens": 99880377.0, + "step": 6979 + }, + { + "epoch": 0.5755751628597344, + "grad_norm": 2.050980567932129, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7211145162582397, + "num_tokens": 99908466.0, + "step": 6980 + }, + { + "epoch": 0.575657623484786, + "grad_norm": 1.7773414850234985, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.750678539276123, + "num_tokens": 99942297.0, + "step": 6981 + }, + { + "epoch": 0.5757400841098376, + "grad_norm": 1.873345971107483, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.72409987449646, + "num_tokens": 99976264.0, + "step": 6982 + }, + { + "epoch": 0.5758225447348891, + "grad_norm": 1.8570760488510132, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7404390573501587, + "num_tokens": 100008568.0, + "step": 6983 + }, + { + "epoch": 0.5759050053599406, + "grad_norm": 1.970471739768982, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.779037594795227, + "num_tokens": 100036984.0, + "step": 6984 + }, + { + "epoch": 0.5759874659849922, + "grad_norm": 1.9015945196151733, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7294096946716309, + "num_tokens": 100068821.0, + "step": 6985 + }, + { + "epoch": 0.5760699266100437, + "grad_norm": 1.8675153255462646, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.7847381830215454, + "num_tokens": 100101024.0, + "step": 6986 + }, + { + "epoch": 0.5761523872350952, + "grad_norm": 1.7068356275558472, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7843922972679138, + "num_tokens": 100136341.0, + "step": 6987 + }, + { + "epoch": 0.5762348478601468, + "grad_norm": 2.1171092987060547, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.759830117225647, + "num_tokens": 100164135.0, + "step": 6988 + }, + { + "epoch": 0.5763173084851984, + "grad_norm": 1.977497935295105, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7812784314155579, + "num_tokens": 100194408.0, + "step": 6989 + }, + { + "epoch": 0.5763997691102498, + "grad_norm": 2.298126220703125, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7015761137008667, + "num_tokens": 100219114.0, + "step": 6990 + }, + { + "epoch": 0.5764822297353014, + "grad_norm": 2.1048007011413574, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7330543398857117, + "num_tokens": 100246153.0, + "step": 6991 + }, + { + "epoch": 0.576564690360353, + "grad_norm": 1.960646152496338, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7406337261199951, + "num_tokens": 100275574.0, + "step": 6992 + }, + { + "epoch": 0.5766471509854044, + "grad_norm": 1.9045491218566895, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7642919421195984, + "num_tokens": 100307343.0, + "step": 6993 + }, + { + "epoch": 0.576729611610456, + "grad_norm": 1.9920772314071655, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7471094131469727, + "num_tokens": 100335651.0, + "step": 6994 + }, + { + "epoch": 0.5768120722355076, + "grad_norm": 1.8239970207214355, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7742179036140442, + "num_tokens": 100366813.0, + "step": 6995 + }, + { + "epoch": 0.576894532860559, + "grad_norm": 2.2073514461517334, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7067641615867615, + "num_tokens": 100395308.0, + "step": 6996 + }, + { + "epoch": 0.5769769934856106, + "grad_norm": 2.072084903717041, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7210088968276978, + "num_tokens": 100425518.0, + "step": 6997 + }, + { + "epoch": 0.5770594541106622, + "grad_norm": 1.8877217769622803, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.753786027431488, + "num_tokens": 100456447.0, + "step": 6998 + }, + { + "epoch": 0.5771419147357137, + "grad_norm": 1.6336911916732788, + "learning_rate": 1e-06, + "loss": 0.534, + "mean_token_accuracy": 0.8130033016204834, + "num_tokens": 100492106.0, + "step": 6999 + }, + { + "epoch": 0.5772243753607652, + "grad_norm": 1.9406888484954834, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.7037180066108704, + "num_tokens": 100525849.0, + "step": 7000 + }, + { + "epoch": 0.5773068359858168, + "grad_norm": 1.9602891206741333, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7803802490234375, + "num_tokens": 100554968.0, + "step": 7001 + }, + { + "epoch": 0.5773892966108684, + "grad_norm": 2.030810594558716, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7601372599601746, + "num_tokens": 100581710.0, + "step": 7002 + }, + { + "epoch": 0.5774717572359198, + "grad_norm": 1.9820311069488525, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6838910579681396, + "num_tokens": 100612979.0, + "step": 7003 + }, + { + "epoch": 0.5775542178609714, + "grad_norm": 2.0157926082611084, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.761393666267395, + "num_tokens": 100640647.0, + "step": 7004 + }, + { + "epoch": 0.577636678486023, + "grad_norm": 1.8478461503982544, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.753917932510376, + "num_tokens": 100671720.0, + "step": 7005 + }, + { + "epoch": 0.5777191391110744, + "grad_norm": 1.8089239597320557, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7772289514541626, + "num_tokens": 100705282.0, + "step": 7006 + }, + { + "epoch": 0.577801599736126, + "grad_norm": 2.084429979324341, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7166635990142822, + "num_tokens": 100735002.0, + "step": 7007 + }, + { + "epoch": 0.5778840603611776, + "grad_norm": 1.838754653930664, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7583831548690796, + "num_tokens": 100768075.0, + "step": 7008 + }, + { + "epoch": 0.577966520986229, + "grad_norm": 2.1473724842071533, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7389776706695557, + "num_tokens": 100794915.0, + "step": 7009 + }, + { + "epoch": 0.5780489816112806, + "grad_norm": 1.9343534708023071, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7243378162384033, + "num_tokens": 100824097.0, + "step": 7010 + }, + { + "epoch": 0.5781314422363322, + "grad_norm": 1.9290363788604736, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7393041849136353, + "num_tokens": 100855092.0, + "step": 7011 + }, + { + "epoch": 0.5782139028613837, + "grad_norm": 1.9068564176559448, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6942702531814575, + "num_tokens": 100890586.0, + "step": 7012 + }, + { + "epoch": 0.5782963634864352, + "grad_norm": 1.9041539430618286, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7209088802337646, + "num_tokens": 100925514.0, + "step": 7013 + }, + { + "epoch": 0.5783788241114868, + "grad_norm": 1.8192471265792847, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.7880371809005737, + "num_tokens": 100959104.0, + "step": 7014 + }, + { + "epoch": 0.5784612847365384, + "grad_norm": 2.1286544799804688, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.745812177658081, + "num_tokens": 100987859.0, + "step": 7015 + }, + { + "epoch": 0.5785437453615898, + "grad_norm": 1.7498849630355835, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7372542023658752, + "num_tokens": 101022128.0, + "step": 7016 + }, + { + "epoch": 0.5786262059866414, + "grad_norm": 1.9864883422851562, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7875475883483887, + "num_tokens": 101050595.0, + "step": 7017 + }, + { + "epoch": 0.578708666611693, + "grad_norm": 1.9460481405258179, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7384136915206909, + "num_tokens": 101081415.0, + "step": 7018 + }, + { + "epoch": 0.5787911272367444, + "grad_norm": 1.9304455518722534, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7396018505096436, + "num_tokens": 101114848.0, + "step": 7019 + }, + { + "epoch": 0.578873587861796, + "grad_norm": 2.3113458156585693, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7393597364425659, + "num_tokens": 101141040.0, + "step": 7020 + }, + { + "epoch": 0.5789560484868476, + "grad_norm": 2.1580827236175537, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7456612586975098, + "num_tokens": 101165768.0, + "step": 7021 + }, + { + "epoch": 0.579038509111899, + "grad_norm": 1.9361437559127808, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7529286742210388, + "num_tokens": 101195217.0, + "step": 7022 + }, + { + "epoch": 0.5791209697369506, + "grad_norm": 1.913718581199646, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7149238586425781, + "num_tokens": 101227195.0, + "step": 7023 + }, + { + "epoch": 0.5792034303620022, + "grad_norm": 2.079329252243042, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7373784780502319, + "num_tokens": 101255047.0, + "step": 7024 + }, + { + "epoch": 0.5792858909870536, + "grad_norm": 1.9033143520355225, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.749893069267273, + "num_tokens": 101286552.0, + "step": 7025 + }, + { + "epoch": 0.5793683516121052, + "grad_norm": 1.6751937866210938, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7337632179260254, + "num_tokens": 101324033.0, + "step": 7026 + }, + { + "epoch": 0.5794508122371568, + "grad_norm": 1.9932761192321777, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.766842782497406, + "num_tokens": 101350437.0, + "step": 7027 + }, + { + "epoch": 0.5795332728622083, + "grad_norm": 2.099409818649292, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.754031777381897, + "num_tokens": 101377433.0, + "step": 7028 + }, + { + "epoch": 0.5796157334872598, + "grad_norm": 1.8892987966537476, + "learning_rate": 1e-06, + "loss": 0.587, + "mean_token_accuracy": 0.804198145866394, + "num_tokens": 101406818.0, + "step": 7029 + }, + { + "epoch": 0.5796981941123114, + "grad_norm": 1.89077889919281, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7427651882171631, + "num_tokens": 101435928.0, + "step": 7030 + }, + { + "epoch": 0.579780654737363, + "grad_norm": 1.8881030082702637, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7505878806114197, + "num_tokens": 101465980.0, + "step": 7031 + }, + { + "epoch": 0.5798631153624144, + "grad_norm": 1.8878980875015259, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7692924737930298, + "num_tokens": 101496240.0, + "step": 7032 + }, + { + "epoch": 0.579945575987466, + "grad_norm": 1.8633127212524414, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.7612595558166504, + "num_tokens": 101526511.0, + "step": 7033 + }, + { + "epoch": 0.5800280366125176, + "grad_norm": 2.0928752422332764, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7759531736373901, + "num_tokens": 101550628.0, + "step": 7034 + }, + { + "epoch": 0.580110497237569, + "grad_norm": 1.8755522966384888, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7168156504631042, + "num_tokens": 101582107.0, + "step": 7035 + }, + { + "epoch": 0.5801929578626206, + "grad_norm": 1.8249421119689941, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7235232591629028, + "num_tokens": 101616936.0, + "step": 7036 + }, + { + "epoch": 0.5802754184876722, + "grad_norm": 2.054701566696167, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7451918125152588, + "num_tokens": 101644253.0, + "step": 7037 + }, + { + "epoch": 0.5803578791127236, + "grad_norm": 2.027635335922241, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7525244951248169, + "num_tokens": 101673103.0, + "step": 7038 + }, + { + "epoch": 0.5804403397377752, + "grad_norm": 1.9087982177734375, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7601225972175598, + "num_tokens": 101705024.0, + "step": 7039 + }, + { + "epoch": 0.5805228003628268, + "grad_norm": 1.9471721649169922, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7636536359786987, + "num_tokens": 101734094.0, + "step": 7040 + }, + { + "epoch": 0.5806052609878782, + "grad_norm": 2.0725202560424805, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7656330466270447, + "num_tokens": 101759063.0, + "step": 7041 + }, + { + "epoch": 0.5806877216129298, + "grad_norm": 1.8901656866073608, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.7973648309707642, + "num_tokens": 101790780.0, + "step": 7042 + }, + { + "epoch": 0.5807701822379814, + "grad_norm": 1.9740930795669556, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7329443693161011, + "num_tokens": 101819948.0, + "step": 7043 + }, + { + "epoch": 0.580852642863033, + "grad_norm": 2.0034878253936768, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6953254342079163, + "num_tokens": 101850554.0, + "step": 7044 + }, + { + "epoch": 0.5809351034880844, + "grad_norm": 2.1629931926727295, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.729945957660675, + "num_tokens": 101877688.0, + "step": 7045 + }, + { + "epoch": 0.581017564113136, + "grad_norm": 1.882212519645691, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7521497011184692, + "num_tokens": 101909646.0, + "step": 7046 + }, + { + "epoch": 0.5811000247381876, + "grad_norm": 1.7691304683685303, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.7871409058570862, + "num_tokens": 101941595.0, + "step": 7047 + }, + { + "epoch": 0.581182485363239, + "grad_norm": 1.7279094457626343, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7638431787490845, + "num_tokens": 101978560.0, + "step": 7048 + }, + { + "epoch": 0.5812649459882906, + "grad_norm": 1.836276650428772, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7358577251434326, + "num_tokens": 102014728.0, + "step": 7049 + }, + { + "epoch": 0.5813474066133422, + "grad_norm": 1.9745819568634033, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7742264866828918, + "num_tokens": 102044693.0, + "step": 7050 + }, + { + "epoch": 0.5814298672383936, + "grad_norm": 2.2543222904205322, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.699507474899292, + "num_tokens": 102070398.0, + "step": 7051 + }, + { + "epoch": 0.5815123278634452, + "grad_norm": 1.9797459840774536, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7442218661308289, + "num_tokens": 102105429.0, + "step": 7052 + }, + { + "epoch": 0.5815947884884968, + "grad_norm": 1.8869282007217407, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7389487028121948, + "num_tokens": 102136931.0, + "step": 7053 + }, + { + "epoch": 0.5816772491135482, + "grad_norm": 2.0244712829589844, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7654725909233093, + "num_tokens": 102163723.0, + "step": 7054 + }, + { + "epoch": 0.5817597097385998, + "grad_norm": 2.016935110092163, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7607133984565735, + "num_tokens": 102193236.0, + "step": 7055 + }, + { + "epoch": 0.5818421703636514, + "grad_norm": 1.8778471946716309, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7688178420066833, + "num_tokens": 102224155.0, + "step": 7056 + }, + { + "epoch": 0.5819246309887028, + "grad_norm": 2.1098012924194336, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7357034087181091, + "num_tokens": 102248805.0, + "step": 7057 + }, + { + "epoch": 0.5820070916137544, + "grad_norm": 1.9080266952514648, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.759601354598999, + "num_tokens": 102279973.0, + "step": 7058 + }, + { + "epoch": 0.582089552238806, + "grad_norm": 1.9696755409240723, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7593631744384766, + "num_tokens": 102313258.0, + "step": 7059 + }, + { + "epoch": 0.5821720128638576, + "grad_norm": 1.7139627933502197, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.7919763326644897, + "num_tokens": 102346631.0, + "step": 7060 + }, + { + "epoch": 0.582254473488909, + "grad_norm": 1.9727798700332642, + "learning_rate": 1e-06, + "loss": 0.7598, + "mean_token_accuracy": 0.7535579800605774, + "num_tokens": 102379667.0, + "step": 7061 + }, + { + "epoch": 0.5823369341139606, + "grad_norm": 2.03480863571167, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6936862468719482, + "num_tokens": 102411204.0, + "step": 7062 + }, + { + "epoch": 0.5824193947390122, + "grad_norm": 2.208150863647461, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7305043935775757, + "num_tokens": 102435212.0, + "step": 7063 + }, + { + "epoch": 0.5825018553640636, + "grad_norm": 2.14961576461792, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7635237574577332, + "num_tokens": 102461336.0, + "step": 7064 + }, + { + "epoch": 0.5825843159891152, + "grad_norm": 2.026827573776245, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7652022838592529, + "num_tokens": 102487789.0, + "step": 7065 + }, + { + "epoch": 0.5826667766141668, + "grad_norm": 2.0239017009735107, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7125980257987976, + "num_tokens": 102518494.0, + "step": 7066 + }, + { + "epoch": 0.5827492372392182, + "grad_norm": 2.0646095275878906, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7700091600418091, + "num_tokens": 102546034.0, + "step": 7067 + }, + { + "epoch": 0.5828316978642698, + "grad_norm": 2.0694000720977783, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7301362752914429, + "num_tokens": 102573062.0, + "step": 7068 + }, + { + "epoch": 0.5829141584893214, + "grad_norm": 1.8585753440856934, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7110432386398315, + "num_tokens": 102605816.0, + "step": 7069 + }, + { + "epoch": 0.5829966191143728, + "grad_norm": 2.1417157649993896, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.770260214805603, + "num_tokens": 102632317.0, + "step": 7070 + }, + { + "epoch": 0.5830790797394244, + "grad_norm": 1.8195281028747559, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.742362380027771, + "num_tokens": 102664037.0, + "step": 7071 + }, + { + "epoch": 0.583161540364476, + "grad_norm": 1.9695625305175781, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7408583760261536, + "num_tokens": 102693787.0, + "step": 7072 + }, + { + "epoch": 0.5832440009895276, + "grad_norm": 2.1483023166656494, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7767308950424194, + "num_tokens": 102721140.0, + "step": 7073 + }, + { + "epoch": 0.583326461614579, + "grad_norm": 1.886983871459961, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.727368175983429, + "num_tokens": 102752178.0, + "step": 7074 + }, + { + "epoch": 0.5834089222396306, + "grad_norm": 1.9468342065811157, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7387838363647461, + "num_tokens": 102782204.0, + "step": 7075 + }, + { + "epoch": 0.5834913828646822, + "grad_norm": 2.020989418029785, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7302143573760986, + "num_tokens": 102812162.0, + "step": 7076 + }, + { + "epoch": 0.5835738434897336, + "grad_norm": 2.244189500808716, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7214970588684082, + "num_tokens": 102838505.0, + "step": 7077 + }, + { + "epoch": 0.5836563041147852, + "grad_norm": 1.9203453063964844, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7678706645965576, + "num_tokens": 102867437.0, + "step": 7078 + }, + { + "epoch": 0.5837387647398368, + "grad_norm": 1.972293496131897, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7749286890029907, + "num_tokens": 102894725.0, + "step": 7079 + }, + { + "epoch": 0.5838212253648882, + "grad_norm": 2.1809298992156982, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7349082231521606, + "num_tokens": 102919024.0, + "step": 7080 + }, + { + "epoch": 0.5839036859899398, + "grad_norm": 1.8199952840805054, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7652996182441711, + "num_tokens": 102952868.0, + "step": 7081 + }, + { + "epoch": 0.5839861466149914, + "grad_norm": 1.7760934829711914, + "learning_rate": 1e-06, + "loss": 0.771, + "mean_token_accuracy": 0.7632918953895569, + "num_tokens": 102988869.0, + "step": 7082 + }, + { + "epoch": 0.5840686072400428, + "grad_norm": 2.0570437908172607, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.738303542137146, + "num_tokens": 103015668.0, + "step": 7083 + }, + { + "epoch": 0.5841510678650944, + "grad_norm": 1.8882356882095337, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7239978313446045, + "num_tokens": 103048122.0, + "step": 7084 + }, + { + "epoch": 0.584233528490146, + "grad_norm": 2.0200018882751465, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7340233325958252, + "num_tokens": 103078857.0, + "step": 7085 + }, + { + "epoch": 0.5843159891151974, + "grad_norm": 1.8769508600234985, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7487379312515259, + "num_tokens": 103112848.0, + "step": 7086 + }, + { + "epoch": 0.584398449740249, + "grad_norm": 1.9968024492263794, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7251783609390259, + "num_tokens": 103139624.0, + "step": 7087 + }, + { + "epoch": 0.5844809103653006, + "grad_norm": 1.9779460430145264, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7693547606468201, + "num_tokens": 103167763.0, + "step": 7088 + }, + { + "epoch": 0.5845633709903522, + "grad_norm": 1.9527384042739868, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7776052355766296, + "num_tokens": 103194444.0, + "step": 7089 + }, + { + "epoch": 0.5846458316154036, + "grad_norm": 1.7408778667449951, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7264419794082642, + "num_tokens": 103233646.0, + "step": 7090 + }, + { + "epoch": 0.5847282922404552, + "grad_norm": 2.1142807006835938, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7236872315406799, + "num_tokens": 103261611.0, + "step": 7091 + }, + { + "epoch": 0.5848107528655068, + "grad_norm": 2.028292417526245, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7328611016273499, + "num_tokens": 103290863.0, + "step": 7092 + }, + { + "epoch": 0.5848932134905582, + "grad_norm": 1.8819094896316528, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7760987281799316, + "num_tokens": 103320916.0, + "step": 7093 + }, + { + "epoch": 0.5849756741156098, + "grad_norm": 1.9780654907226562, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7252219915390015, + "num_tokens": 103352225.0, + "step": 7094 + }, + { + "epoch": 0.5850581347406614, + "grad_norm": 2.059790849685669, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7302131056785583, + "num_tokens": 103380339.0, + "step": 7095 + }, + { + "epoch": 0.5851405953657128, + "grad_norm": 1.7709152698516846, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7596499919891357, + "num_tokens": 103415963.0, + "step": 7096 + }, + { + "epoch": 0.5852230559907644, + "grad_norm": 1.7829110622406006, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7580884099006653, + "num_tokens": 103450569.0, + "step": 7097 + }, + { + "epoch": 0.585305516615816, + "grad_norm": 2.1474592685699463, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7305911779403687, + "num_tokens": 103479843.0, + "step": 7098 + }, + { + "epoch": 0.5853879772408674, + "grad_norm": 1.9505348205566406, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7045557498931885, + "num_tokens": 103514678.0, + "step": 7099 + }, + { + "epoch": 0.585470437865919, + "grad_norm": 1.8051002025604248, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7676615715026855, + "num_tokens": 103545615.0, + "step": 7100 + }, + { + "epoch": 0.5855528984909706, + "grad_norm": 1.995316743850708, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7360955476760864, + "num_tokens": 103575139.0, + "step": 7101 + }, + { + "epoch": 0.585635359116022, + "grad_norm": 2.0408740043640137, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7321797609329224, + "num_tokens": 103605423.0, + "step": 7102 + }, + { + "epoch": 0.5857178197410736, + "grad_norm": 1.9659031629562378, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7879864573478699, + "num_tokens": 103632099.0, + "step": 7103 + }, + { + "epoch": 0.5858002803661252, + "grad_norm": 1.8845608234405518, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7292838096618652, + "num_tokens": 103665048.0, + "step": 7104 + }, + { + "epoch": 0.5858827409911768, + "grad_norm": 1.607600212097168, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7689849138259888, + "num_tokens": 103703195.0, + "step": 7105 + }, + { + "epoch": 0.5859652016162282, + "grad_norm": 2.1111929416656494, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7452521324157715, + "num_tokens": 103730788.0, + "step": 7106 + }, + { + "epoch": 0.5860476622412798, + "grad_norm": 2.2068822383880615, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7594025135040283, + "num_tokens": 103755240.0, + "step": 7107 + }, + { + "epoch": 0.5861301228663314, + "grad_norm": 1.8671276569366455, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7193584442138672, + "num_tokens": 103789245.0, + "step": 7108 + }, + { + "epoch": 0.5862125834913828, + "grad_norm": 2.2273502349853516, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.71543288230896, + "num_tokens": 103814227.0, + "step": 7109 + }, + { + "epoch": 0.5862950441164344, + "grad_norm": 2.1616392135620117, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7158417701721191, + "num_tokens": 103840318.0, + "step": 7110 + }, + { + "epoch": 0.586377504741486, + "grad_norm": 2.185046911239624, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7428054809570312, + "num_tokens": 103866892.0, + "step": 7111 + }, + { + "epoch": 0.5864599653665374, + "grad_norm": 1.9366564750671387, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7640331387519836, + "num_tokens": 103898494.0, + "step": 7112 + }, + { + "epoch": 0.586542425991589, + "grad_norm": 2.1270415782928467, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7409947514533997, + "num_tokens": 103923764.0, + "step": 7113 + }, + { + "epoch": 0.5866248866166406, + "grad_norm": 1.8717557191848755, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7805391550064087, + "num_tokens": 103954084.0, + "step": 7114 + }, + { + "epoch": 0.586707347241692, + "grad_norm": 1.7483603954315186, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7501459121704102, + "num_tokens": 103992573.0, + "step": 7115 + }, + { + "epoch": 0.5867898078667436, + "grad_norm": 1.9500608444213867, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7323439121246338, + "num_tokens": 104024386.0, + "step": 7116 + }, + { + "epoch": 0.5868722684917952, + "grad_norm": 2.5424771308898926, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7189723253250122, + "num_tokens": 104045602.0, + "step": 7117 + }, + { + "epoch": 0.5869547291168467, + "grad_norm": 1.9180935621261597, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7588322162628174, + "num_tokens": 104076058.0, + "step": 7118 + }, + { + "epoch": 0.5870371897418982, + "grad_norm": 1.983901858329773, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7183502316474915, + "num_tokens": 104105389.0, + "step": 7119 + }, + { + "epoch": 0.5871196503669498, + "grad_norm": 2.0441408157348633, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7546278238296509, + "num_tokens": 104132943.0, + "step": 7120 + }, + { + "epoch": 0.5872021109920014, + "grad_norm": 2.1673107147216797, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7232933640480042, + "num_tokens": 104158736.0, + "step": 7121 + }, + { + "epoch": 0.5872845716170528, + "grad_norm": 2.075056314468384, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7534376978874207, + "num_tokens": 104185333.0, + "step": 7122 + }, + { + "epoch": 0.5873670322421044, + "grad_norm": 2.188279151916504, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7204347848892212, + "num_tokens": 104211094.0, + "step": 7123 + }, + { + "epoch": 0.587449492867156, + "grad_norm": 1.8535491228103638, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7439484596252441, + "num_tokens": 104242506.0, + "step": 7124 + }, + { + "epoch": 0.5875319534922074, + "grad_norm": 2.0171778202056885, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.704068660736084, + "num_tokens": 104273575.0, + "step": 7125 + }, + { + "epoch": 0.587614414117259, + "grad_norm": 1.950478196144104, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.7868806719779968, + "num_tokens": 104303646.0, + "step": 7126 + }, + { + "epoch": 0.5876968747423106, + "grad_norm": 1.7717753648757935, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7462543845176697, + "num_tokens": 104338178.0, + "step": 7127 + }, + { + "epoch": 0.587779335367362, + "grad_norm": 1.7941758632659912, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7776575088500977, + "num_tokens": 104369970.0, + "step": 7128 + }, + { + "epoch": 0.5878617959924136, + "grad_norm": 1.8746322393417358, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.7485507130622864, + "num_tokens": 104399788.0, + "step": 7129 + }, + { + "epoch": 0.5879442566174652, + "grad_norm": 1.8903793096542358, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7595376968383789, + "num_tokens": 104429086.0, + "step": 7130 + }, + { + "epoch": 0.5880267172425166, + "grad_norm": 1.933733582496643, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7323976755142212, + "num_tokens": 104460715.0, + "step": 7131 + }, + { + "epoch": 0.5881091778675682, + "grad_norm": 1.8253928422927856, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7610167264938354, + "num_tokens": 104491960.0, + "step": 7132 + }, + { + "epoch": 0.5881916384926198, + "grad_norm": 2.0260446071624756, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7683979272842407, + "num_tokens": 104520142.0, + "step": 7133 + }, + { + "epoch": 0.5882740991176714, + "grad_norm": 1.9675266742706299, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7640523910522461, + "num_tokens": 104554004.0, + "step": 7134 + }, + { + "epoch": 0.5883565597427228, + "grad_norm": 2.1809582710266113, + "learning_rate": 1e-06, + "loss": 0.7673, + "mean_token_accuracy": 0.7540104389190674, + "num_tokens": 104581774.0, + "step": 7135 + }, + { + "epoch": 0.5884390203677744, + "grad_norm": 2.2246525287628174, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7694463133811951, + "num_tokens": 104604715.0, + "step": 7136 + }, + { + "epoch": 0.588521480992826, + "grad_norm": 2.014617681503296, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7317970991134644, + "num_tokens": 104631559.0, + "step": 7137 + }, + { + "epoch": 0.5886039416178774, + "grad_norm": 2.095991611480713, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7603612542152405, + "num_tokens": 104661663.0, + "step": 7138 + }, + { + "epoch": 0.588686402242929, + "grad_norm": 2.1983089447021484, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.679253339767456, + "num_tokens": 104688131.0, + "step": 7139 + }, + { + "epoch": 0.5887688628679806, + "grad_norm": 2.0618233680725098, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7669683694839478, + "num_tokens": 104712327.0, + "step": 7140 + }, + { + "epoch": 0.588851323493032, + "grad_norm": 1.8976280689239502, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7646138072013855, + "num_tokens": 104741130.0, + "step": 7141 + }, + { + "epoch": 0.5889337841180836, + "grad_norm": 1.7699289321899414, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7522200345993042, + "num_tokens": 104776281.0, + "step": 7142 + }, + { + "epoch": 0.5890162447431352, + "grad_norm": 1.9635045528411865, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7350513935089111, + "num_tokens": 104809908.0, + "step": 7143 + }, + { + "epoch": 0.5890987053681866, + "grad_norm": 2.048050880432129, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7133215665817261, + "num_tokens": 104837877.0, + "step": 7144 + }, + { + "epoch": 0.5891811659932382, + "grad_norm": 1.78312087059021, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.746696949005127, + "num_tokens": 104871655.0, + "step": 7145 + }, + { + "epoch": 0.5892636266182898, + "grad_norm": 1.8653532266616821, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7342742681503296, + "num_tokens": 104902537.0, + "step": 7146 + }, + { + "epoch": 0.5893460872433413, + "grad_norm": 2.0062813758850098, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.760002613067627, + "num_tokens": 104931332.0, + "step": 7147 + }, + { + "epoch": 0.5894285478683928, + "grad_norm": 1.644979476928711, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7617790699005127, + "num_tokens": 104970431.0, + "step": 7148 + }, + { + "epoch": 0.5895110084934444, + "grad_norm": 1.9869223833084106, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7686094045639038, + "num_tokens": 104998837.0, + "step": 7149 + }, + { + "epoch": 0.589593469118496, + "grad_norm": 1.886116623878479, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7820390462875366, + "num_tokens": 105031214.0, + "step": 7150 + }, + { + "epoch": 0.5896759297435474, + "grad_norm": 2.058593511581421, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7343620657920837, + "num_tokens": 105060352.0, + "step": 7151 + }, + { + "epoch": 0.589758390368599, + "grad_norm": 2.153719902038574, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7277722358703613, + "num_tokens": 105087715.0, + "step": 7152 + }, + { + "epoch": 0.5898408509936506, + "grad_norm": 1.929854393005371, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7273149490356445, + "num_tokens": 105117372.0, + "step": 7153 + }, + { + "epoch": 0.589923311618702, + "grad_norm": 1.8904601335525513, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7337032556533813, + "num_tokens": 105151867.0, + "step": 7154 + }, + { + "epoch": 0.5900057722437536, + "grad_norm": 1.836820363998413, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.7899081707000732, + "num_tokens": 105183949.0, + "step": 7155 + }, + { + "epoch": 0.5900882328688052, + "grad_norm": 1.8275706768035889, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7827243208885193, + "num_tokens": 105216818.0, + "step": 7156 + }, + { + "epoch": 0.5901706934938566, + "grad_norm": 1.8398048877716064, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7209405899047852, + "num_tokens": 105250195.0, + "step": 7157 + }, + { + "epoch": 0.5902531541189082, + "grad_norm": 2.182608127593994, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7498042583465576, + "num_tokens": 105273747.0, + "step": 7158 + }, + { + "epoch": 0.5903356147439598, + "grad_norm": 1.8237320184707642, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7540961503982544, + "num_tokens": 105307792.0, + "step": 7159 + }, + { + "epoch": 0.5904180753690113, + "grad_norm": 1.9080039262771606, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7586023807525635, + "num_tokens": 105338371.0, + "step": 7160 + }, + { + "epoch": 0.5905005359940628, + "grad_norm": 2.0657777786254883, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7324620485305786, + "num_tokens": 105365937.0, + "step": 7161 + }, + { + "epoch": 0.5905829966191144, + "grad_norm": 2.0396854877471924, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7379127144813538, + "num_tokens": 105397788.0, + "step": 7162 + }, + { + "epoch": 0.5906654572441659, + "grad_norm": 1.8275970220565796, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7604712843894958, + "num_tokens": 105429091.0, + "step": 7163 + }, + { + "epoch": 0.5907479178692174, + "grad_norm": 1.9244931936264038, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.7818305492401123, + "num_tokens": 105459238.0, + "step": 7164 + }, + { + "epoch": 0.590830378494269, + "grad_norm": 2.1097657680511475, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7316238880157471, + "num_tokens": 105485034.0, + "step": 7165 + }, + { + "epoch": 0.5909128391193206, + "grad_norm": 2.082304000854492, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7383919954299927, + "num_tokens": 105512782.0, + "step": 7166 + }, + { + "epoch": 0.590995299744372, + "grad_norm": 1.7533200979232788, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7377924919128418, + "num_tokens": 105549995.0, + "step": 7167 + }, + { + "epoch": 0.5910777603694236, + "grad_norm": 1.9235254526138306, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7343117594718933, + "num_tokens": 105582353.0, + "step": 7168 + }, + { + "epoch": 0.5911602209944752, + "grad_norm": 2.072815418243408, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7405771613121033, + "num_tokens": 105612246.0, + "step": 7169 + }, + { + "epoch": 0.5912426816195266, + "grad_norm": 1.6756974458694458, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7736390233039856, + "num_tokens": 105648015.0, + "step": 7170 + }, + { + "epoch": 0.5913251422445782, + "grad_norm": 1.8530690670013428, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7914639711380005, + "num_tokens": 105677564.0, + "step": 7171 + }, + { + "epoch": 0.5914076028696298, + "grad_norm": 2.071709632873535, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7442914843559265, + "num_tokens": 105702850.0, + "step": 7172 + }, + { + "epoch": 0.5914900634946813, + "grad_norm": 2.2402966022491455, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7604700326919556, + "num_tokens": 105725145.0, + "step": 7173 + }, + { + "epoch": 0.5915725241197328, + "grad_norm": 1.9222490787506104, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7543600797653198, + "num_tokens": 105755519.0, + "step": 7174 + }, + { + "epoch": 0.5916549847447844, + "grad_norm": 1.723280429840088, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.8024660348892212, + "num_tokens": 105790676.0, + "step": 7175 + }, + { + "epoch": 0.5917374453698359, + "grad_norm": 2.076845169067383, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7761785387992859, + "num_tokens": 105815092.0, + "step": 7176 + }, + { + "epoch": 0.5918199059948874, + "grad_norm": 1.834500789642334, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7360032796859741, + "num_tokens": 105851085.0, + "step": 7177 + }, + { + "epoch": 0.591902366619939, + "grad_norm": 1.7043994665145874, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7935154438018799, + "num_tokens": 105885336.0, + "step": 7178 + }, + { + "epoch": 0.5919848272449905, + "grad_norm": 2.1058733463287354, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7319110631942749, + "num_tokens": 105914679.0, + "step": 7179 + }, + { + "epoch": 0.592067287870042, + "grad_norm": 1.9173613786697388, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.7848109006881714, + "num_tokens": 105944448.0, + "step": 7180 + }, + { + "epoch": 0.5921497484950936, + "grad_norm": 2.0745832920074463, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7387210130691528, + "num_tokens": 105971174.0, + "step": 7181 + }, + { + "epoch": 0.5922322091201452, + "grad_norm": 1.9483286142349243, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7260177135467529, + "num_tokens": 106003704.0, + "step": 7182 + }, + { + "epoch": 0.5923146697451966, + "grad_norm": 2.0126729011535645, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7232603430747986, + "num_tokens": 106032880.0, + "step": 7183 + }, + { + "epoch": 0.5923971303702482, + "grad_norm": 1.7312934398651123, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7853463888168335, + "num_tokens": 106069917.0, + "step": 7184 + }, + { + "epoch": 0.5924795909952998, + "grad_norm": 1.900301456451416, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7512906789779663, + "num_tokens": 106100423.0, + "step": 7185 + }, + { + "epoch": 0.5925620516203512, + "grad_norm": 1.9873422384262085, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7257766723632812, + "num_tokens": 106131940.0, + "step": 7186 + }, + { + "epoch": 0.5926445122454028, + "grad_norm": 2.2152020931243896, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7487652897834778, + "num_tokens": 106155802.0, + "step": 7187 + }, + { + "epoch": 0.5927269728704544, + "grad_norm": 1.5604541301727295, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7886708974838257, + "num_tokens": 106196511.0, + "step": 7188 + }, + { + "epoch": 0.5928094334955059, + "grad_norm": 2.113483428955078, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7399203777313232, + "num_tokens": 106224312.0, + "step": 7189 + }, + { + "epoch": 0.5928918941205574, + "grad_norm": 2.0521676540374756, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.70467209815979, + "num_tokens": 106258276.0, + "step": 7190 + }, + { + "epoch": 0.592974354745609, + "grad_norm": 2.074674367904663, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7419811487197876, + "num_tokens": 106284586.0, + "step": 7191 + }, + { + "epoch": 0.5930568153706605, + "grad_norm": 2.1563315391540527, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7148746252059937, + "num_tokens": 106310242.0, + "step": 7192 + }, + { + "epoch": 0.593139275995712, + "grad_norm": 1.8114572763442993, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7638643383979797, + "num_tokens": 106342697.0, + "step": 7193 + }, + { + "epoch": 0.5932217366207636, + "grad_norm": 2.0665218830108643, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7465221285820007, + "num_tokens": 106370128.0, + "step": 7194 + }, + { + "epoch": 0.5933041972458151, + "grad_norm": 1.9534525871276855, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.7558199167251587, + "num_tokens": 106400921.0, + "step": 7195 + }, + { + "epoch": 0.5933866578708666, + "grad_norm": 1.8477709293365479, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7447084188461304, + "num_tokens": 106433756.0, + "step": 7196 + }, + { + "epoch": 0.5934691184959182, + "grad_norm": 2.1711623668670654, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7201793193817139, + "num_tokens": 106458399.0, + "step": 7197 + }, + { + "epoch": 0.5935515791209698, + "grad_norm": 2.0971286296844482, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7591328620910645, + "num_tokens": 106485468.0, + "step": 7198 + }, + { + "epoch": 0.5936340397460212, + "grad_norm": 1.9307024478912354, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7378150224685669, + "num_tokens": 106515409.0, + "step": 7199 + }, + { + "epoch": 0.5937165003710728, + "grad_norm": 1.794317603111267, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7474222183227539, + "num_tokens": 106549709.0, + "step": 7200 + }, + { + "epoch": 0.5937989609961244, + "grad_norm": 1.580270528793335, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7841713428497314, + "num_tokens": 106590491.0, + "step": 7201 + }, + { + "epoch": 0.5938814216211759, + "grad_norm": 2.1279635429382324, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7063740491867065, + "num_tokens": 106617246.0, + "step": 7202 + }, + { + "epoch": 0.5939638822462274, + "grad_norm": 2.0429491996765137, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7568393349647522, + "num_tokens": 106643871.0, + "step": 7203 + }, + { + "epoch": 0.594046342871279, + "grad_norm": 1.748488426208496, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.787739098072052, + "num_tokens": 106677275.0, + "step": 7204 + }, + { + "epoch": 0.5941288034963305, + "grad_norm": 2.070216655731201, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7208664417266846, + "num_tokens": 106705433.0, + "step": 7205 + }, + { + "epoch": 0.594211264121382, + "grad_norm": 1.9278476238250732, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7262766361236572, + "num_tokens": 106737619.0, + "step": 7206 + }, + { + "epoch": 0.5942937247464336, + "grad_norm": 1.798683524131775, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.7007250785827637, + "num_tokens": 106772766.0, + "step": 7207 + }, + { + "epoch": 0.5943761853714851, + "grad_norm": 1.9961296319961548, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7575017213821411, + "num_tokens": 106803996.0, + "step": 7208 + }, + { + "epoch": 0.5944586459965366, + "grad_norm": 1.9375754594802856, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7748843431472778, + "num_tokens": 106831942.0, + "step": 7209 + }, + { + "epoch": 0.5945411066215882, + "grad_norm": 2.20308518409729, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7744568586349487, + "num_tokens": 106853909.0, + "step": 7210 + }, + { + "epoch": 0.5946235672466398, + "grad_norm": 2.1550281047821045, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7480970025062561, + "num_tokens": 106879959.0, + "step": 7211 + }, + { + "epoch": 0.5947060278716912, + "grad_norm": 1.8577780723571777, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7470210790634155, + "num_tokens": 106911902.0, + "step": 7212 + }, + { + "epoch": 0.5947884884967428, + "grad_norm": 1.8537455797195435, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7562309503555298, + "num_tokens": 106944322.0, + "step": 7213 + }, + { + "epoch": 0.5948709491217944, + "grad_norm": 2.0912466049194336, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7212441563606262, + "num_tokens": 106974106.0, + "step": 7214 + }, + { + "epoch": 0.5949534097468459, + "grad_norm": 1.5869791507720947, + "learning_rate": 1e-06, + "loss": 0.495, + "mean_token_accuracy": 0.8361612558364868, + "num_tokens": 107009471.0, + "step": 7215 + }, + { + "epoch": 0.5950358703718974, + "grad_norm": 1.8618980646133423, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.743028998374939, + "num_tokens": 107041494.0, + "step": 7216 + }, + { + "epoch": 0.595118330996949, + "grad_norm": 1.8610128164291382, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7398170232772827, + "num_tokens": 107074784.0, + "step": 7217 + }, + { + "epoch": 0.5952007916220005, + "grad_norm": 2.0500593185424805, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.753320574760437, + "num_tokens": 107101207.0, + "step": 7218 + }, + { + "epoch": 0.595283252247052, + "grad_norm": 1.8488173484802246, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7320074439048767, + "num_tokens": 107135049.0, + "step": 7219 + }, + { + "epoch": 0.5953657128721036, + "grad_norm": 1.8710658550262451, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7375372648239136, + "num_tokens": 107168069.0, + "step": 7220 + }, + { + "epoch": 0.5954481734971551, + "grad_norm": 1.852211833000183, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7716464996337891, + "num_tokens": 107200261.0, + "step": 7221 + }, + { + "epoch": 0.5955306341222066, + "grad_norm": 1.9685542583465576, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7415274381637573, + "num_tokens": 107229045.0, + "step": 7222 + }, + { + "epoch": 0.5956130947472582, + "grad_norm": 2.044585943222046, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7449922561645508, + "num_tokens": 107255755.0, + "step": 7223 + }, + { + "epoch": 0.5956955553723097, + "grad_norm": 1.8949289321899414, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7409693598747253, + "num_tokens": 107287744.0, + "step": 7224 + }, + { + "epoch": 0.5957780159973612, + "grad_norm": 2.0114235877990723, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7557530999183655, + "num_tokens": 107316150.0, + "step": 7225 + }, + { + "epoch": 0.5958604766224128, + "grad_norm": 2.152639150619507, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7437751889228821, + "num_tokens": 107340977.0, + "step": 7226 + }, + { + "epoch": 0.5959429372474644, + "grad_norm": 1.9757517576217651, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7547303438186646, + "num_tokens": 107369828.0, + "step": 7227 + }, + { + "epoch": 0.5960253978725158, + "grad_norm": 2.3433377742767334, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.7510777711868286, + "num_tokens": 107390719.0, + "step": 7228 + }, + { + "epoch": 0.5961078584975674, + "grad_norm": 2.172905683517456, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7233401536941528, + "num_tokens": 107416663.0, + "step": 7229 + }, + { + "epoch": 0.596190319122619, + "grad_norm": 1.9247502088546753, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.7952749133110046, + "num_tokens": 107446560.0, + "step": 7230 + }, + { + "epoch": 0.5962727797476705, + "grad_norm": 1.9886149168014526, + "learning_rate": 1e-06, + "loss": 0.783, + "mean_token_accuracy": 0.7547683715820312, + "num_tokens": 107473211.0, + "step": 7231 + }, + { + "epoch": 0.596355240372722, + "grad_norm": 2.165438652038574, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7561034560203552, + "num_tokens": 107500467.0, + "step": 7232 + }, + { + "epoch": 0.5964377009977736, + "grad_norm": 1.7888729572296143, + "learning_rate": 1e-06, + "loss": 0.5899, + "mean_token_accuracy": 0.804298996925354, + "num_tokens": 107530768.0, + "step": 7233 + }, + { + "epoch": 0.5965201616228251, + "grad_norm": 1.9664525985717773, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7206257581710815, + "num_tokens": 107564287.0, + "step": 7234 + }, + { + "epoch": 0.5966026222478766, + "grad_norm": 1.8812847137451172, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7640748023986816, + "num_tokens": 107593914.0, + "step": 7235 + }, + { + "epoch": 0.5966850828729282, + "grad_norm": 1.9307326078414917, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7333146333694458, + "num_tokens": 107627264.0, + "step": 7236 + }, + { + "epoch": 0.5967675434979797, + "grad_norm": 1.9740784168243408, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7279033660888672, + "num_tokens": 107656620.0, + "step": 7237 + }, + { + "epoch": 0.5968500041230312, + "grad_norm": 2.3358445167541504, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.771461546421051, + "num_tokens": 107681782.0, + "step": 7238 + }, + { + "epoch": 0.5969324647480828, + "grad_norm": 1.7629941701889038, + "learning_rate": 1e-06, + "loss": 0.5985, + "mean_token_accuracy": 0.7993032932281494, + "num_tokens": 107716424.0, + "step": 7239 + }, + { + "epoch": 0.5970149253731343, + "grad_norm": 2.0803403854370117, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7544740438461304, + "num_tokens": 107742819.0, + "step": 7240 + }, + { + "epoch": 0.5970973859981858, + "grad_norm": 1.6568435430526733, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7536752223968506, + "num_tokens": 107778056.0, + "step": 7241 + }, + { + "epoch": 0.5971798466232374, + "grad_norm": 1.586936593055725, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7549497485160828, + "num_tokens": 107821955.0, + "step": 7242 + }, + { + "epoch": 0.597262307248289, + "grad_norm": 2.0305159091949463, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7340972423553467, + "num_tokens": 107852899.0, + "step": 7243 + }, + { + "epoch": 0.5973447678733405, + "grad_norm": 1.9436966180801392, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7832816243171692, + "num_tokens": 107881396.0, + "step": 7244 + }, + { + "epoch": 0.597427228498392, + "grad_norm": 1.9108623266220093, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.771934986114502, + "num_tokens": 107913641.0, + "step": 7245 + }, + { + "epoch": 0.5975096891234436, + "grad_norm": 1.9464209079742432, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7535521984100342, + "num_tokens": 107942961.0, + "step": 7246 + }, + { + "epoch": 0.5975921497484951, + "grad_norm": 1.9920414686203003, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7385451793670654, + "num_tokens": 107972146.0, + "step": 7247 + }, + { + "epoch": 0.5976746103735466, + "grad_norm": 2.276871919631958, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7245817184448242, + "num_tokens": 107997144.0, + "step": 7248 + }, + { + "epoch": 0.5977570709985982, + "grad_norm": 1.888719081878662, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7721598148345947, + "num_tokens": 108032048.0, + "step": 7249 + }, + { + "epoch": 0.5978395316236497, + "grad_norm": 1.7203004360198975, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7263414263725281, + "num_tokens": 108068329.0, + "step": 7250 + }, + { + "epoch": 0.5979219922487012, + "grad_norm": 1.838641881942749, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.769668698310852, + "num_tokens": 108099176.0, + "step": 7251 + }, + { + "epoch": 0.5980044528737528, + "grad_norm": 1.8806194067001343, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7642830610275269, + "num_tokens": 108129882.0, + "step": 7252 + }, + { + "epoch": 0.5980869134988043, + "grad_norm": 1.7150377035140991, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.786240816116333, + "num_tokens": 108166901.0, + "step": 7253 + }, + { + "epoch": 0.5981693741238558, + "grad_norm": 1.9879242181777954, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7507942914962769, + "num_tokens": 108195953.0, + "step": 7254 + }, + { + "epoch": 0.5982518347489074, + "grad_norm": 1.9987976551055908, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7414342164993286, + "num_tokens": 108222443.0, + "step": 7255 + }, + { + "epoch": 0.5983342953739589, + "grad_norm": 1.7547672986984253, + "learning_rate": 1e-06, + "loss": 0.6233, + "mean_token_accuracy": 0.799167275428772, + "num_tokens": 108253897.0, + "step": 7256 + }, + { + "epoch": 0.5984167559990105, + "grad_norm": 1.9718669652938843, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.727363109588623, + "num_tokens": 108284619.0, + "step": 7257 + }, + { + "epoch": 0.598499216624062, + "grad_norm": 2.2782933712005615, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7123273611068726, + "num_tokens": 108308899.0, + "step": 7258 + }, + { + "epoch": 0.5985816772491136, + "grad_norm": 2.077631950378418, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7701563835144043, + "num_tokens": 108331528.0, + "step": 7259 + }, + { + "epoch": 0.5986641378741651, + "grad_norm": 2.005077362060547, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7440218925476074, + "num_tokens": 108362029.0, + "step": 7260 + }, + { + "epoch": 0.5987465984992166, + "grad_norm": 1.8174543380737305, + "learning_rate": 1e-06, + "loss": 0.6172, + "mean_token_accuracy": 0.7888766527175903, + "num_tokens": 108393153.0, + "step": 7261 + }, + { + "epoch": 0.5988290591242682, + "grad_norm": 1.9290751218795776, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7442526817321777, + "num_tokens": 108422357.0, + "step": 7262 + }, + { + "epoch": 0.5989115197493197, + "grad_norm": 1.6935738325119019, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7818061709403992, + "num_tokens": 108461198.0, + "step": 7263 + }, + { + "epoch": 0.5989939803743712, + "grad_norm": 1.820156216621399, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7371480464935303, + "num_tokens": 108496897.0, + "step": 7264 + }, + { + "epoch": 0.5990764409994228, + "grad_norm": 2.040100574493408, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7377743721008301, + "num_tokens": 108523913.0, + "step": 7265 + }, + { + "epoch": 0.5991589016244743, + "grad_norm": 2.066131591796875, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7510706186294556, + "num_tokens": 108550837.0, + "step": 7266 + }, + { + "epoch": 0.5992413622495258, + "grad_norm": 2.158364772796631, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7308762073516846, + "num_tokens": 108578294.0, + "step": 7267 + }, + { + "epoch": 0.5993238228745774, + "grad_norm": 1.82631516456604, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.769952654838562, + "num_tokens": 108610872.0, + "step": 7268 + }, + { + "epoch": 0.5994062834996289, + "grad_norm": 1.9640127420425415, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7388677597045898, + "num_tokens": 108641789.0, + "step": 7269 + }, + { + "epoch": 0.5994887441246805, + "grad_norm": 1.9738467931747437, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7034214735031128, + "num_tokens": 108673522.0, + "step": 7270 + }, + { + "epoch": 0.599571204749732, + "grad_norm": 2.1940317153930664, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7162648439407349, + "num_tokens": 108700318.0, + "step": 7271 + }, + { + "epoch": 0.5996536653747836, + "grad_norm": 2.305764675140381, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7448619604110718, + "num_tokens": 108720846.0, + "step": 7272 + }, + { + "epoch": 0.5997361259998351, + "grad_norm": 1.7990765571594238, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7738320827484131, + "num_tokens": 108754777.0, + "step": 7273 + }, + { + "epoch": 0.5998185866248866, + "grad_norm": 2.152100086212158, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7198141813278198, + "num_tokens": 108783617.0, + "step": 7274 + }, + { + "epoch": 0.5999010472499382, + "grad_norm": 2.0401878356933594, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7056969404220581, + "num_tokens": 108812783.0, + "step": 7275 + }, + { + "epoch": 0.5999835078749897, + "grad_norm": 1.9116253852844238, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7542499303817749, + "num_tokens": 108842027.0, + "step": 7276 + }, + { + "epoch": 0.6000659685000412, + "grad_norm": 1.9569990634918213, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7450942993164062, + "num_tokens": 108869507.0, + "step": 7277 + }, + { + "epoch": 0.6001484291250928, + "grad_norm": 2.0278377532958984, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7345743179321289, + "num_tokens": 108896703.0, + "step": 7278 + }, + { + "epoch": 0.6002308897501443, + "grad_norm": 2.050893545150757, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.75337815284729, + "num_tokens": 108925079.0, + "step": 7279 + }, + { + "epoch": 0.6003133503751958, + "grad_norm": 1.963303565979004, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7289532423019409, + "num_tokens": 108956719.0, + "step": 7280 + }, + { + "epoch": 0.6003958110002474, + "grad_norm": 2.181769609451294, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7079348564147949, + "num_tokens": 108983825.0, + "step": 7281 + }, + { + "epoch": 0.6004782716252989, + "grad_norm": 1.8517518043518066, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7451497316360474, + "num_tokens": 109013014.0, + "step": 7282 + }, + { + "epoch": 0.6005607322503504, + "grad_norm": 1.9536495208740234, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7312802672386169, + "num_tokens": 109043637.0, + "step": 7283 + }, + { + "epoch": 0.600643192875402, + "grad_norm": 1.9599210023880005, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7367197275161743, + "num_tokens": 109071771.0, + "step": 7284 + }, + { + "epoch": 0.6007256535004535, + "grad_norm": 1.9073303937911987, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7179449796676636, + "num_tokens": 109103293.0, + "step": 7285 + }, + { + "epoch": 0.600808114125505, + "grad_norm": 2.071448802947998, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7674369812011719, + "num_tokens": 109127752.0, + "step": 7286 + }, + { + "epoch": 0.6008905747505566, + "grad_norm": 1.6718864440917969, + "learning_rate": 1e-06, + "loss": 0.5613, + "mean_token_accuracy": 0.8156169056892395, + "num_tokens": 109165162.0, + "step": 7287 + }, + { + "epoch": 0.6009730353756082, + "grad_norm": 2.088869333267212, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7279566526412964, + "num_tokens": 109191445.0, + "step": 7288 + }, + { + "epoch": 0.6010554960006597, + "grad_norm": 2.048557996749878, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7477859258651733, + "num_tokens": 109218898.0, + "step": 7289 + }, + { + "epoch": 0.6011379566257112, + "grad_norm": 2.0616343021392822, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7421362400054932, + "num_tokens": 109245979.0, + "step": 7290 + }, + { + "epoch": 0.6012204172507628, + "grad_norm": 2.04797101020813, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7381582260131836, + "num_tokens": 109274667.0, + "step": 7291 + }, + { + "epoch": 0.6013028778758143, + "grad_norm": 2.2186501026153564, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6965463161468506, + "num_tokens": 109302373.0, + "step": 7292 + }, + { + "epoch": 0.6013853385008658, + "grad_norm": 1.9346729516983032, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7542386054992676, + "num_tokens": 109333284.0, + "step": 7293 + }, + { + "epoch": 0.6014677991259174, + "grad_norm": 2.30849027633667, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.7065523862838745, + "num_tokens": 109358114.0, + "step": 7294 + }, + { + "epoch": 0.6015502597509689, + "grad_norm": 2.0002853870391846, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.7921543717384338, + "num_tokens": 109384081.0, + "step": 7295 + }, + { + "epoch": 0.6016327203760204, + "grad_norm": 2.1755239963531494, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7341894507408142, + "num_tokens": 109412217.0, + "step": 7296 + }, + { + "epoch": 0.601715181001072, + "grad_norm": 1.9386004209518433, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7501171231269836, + "num_tokens": 109441156.0, + "step": 7297 + }, + { + "epoch": 0.6017976416261235, + "grad_norm": 1.6554949283599854, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.7923819422721863, + "num_tokens": 109478884.0, + "step": 7298 + }, + { + "epoch": 0.601880102251175, + "grad_norm": 1.8854618072509766, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7400492429733276, + "num_tokens": 109510477.0, + "step": 7299 + }, + { + "epoch": 0.6019625628762266, + "grad_norm": 1.8811180591583252, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.767858624458313, + "num_tokens": 109542220.0, + "step": 7300 + }, + { + "epoch": 0.6020450235012781, + "grad_norm": 2.156118631362915, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7403150796890259, + "num_tokens": 109568026.0, + "step": 7301 + }, + { + "epoch": 0.6021274841263297, + "grad_norm": 1.7475155591964722, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7384768724441528, + "num_tokens": 109604234.0, + "step": 7302 + }, + { + "epoch": 0.6022099447513812, + "grad_norm": 1.9836760759353638, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.735878586769104, + "num_tokens": 109631660.0, + "step": 7303 + }, + { + "epoch": 0.6022924053764328, + "grad_norm": 1.9186314344406128, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.753331184387207, + "num_tokens": 109662195.0, + "step": 7304 + }, + { + "epoch": 0.6023748660014843, + "grad_norm": 1.9018025398254395, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7506781220436096, + "num_tokens": 109690870.0, + "step": 7305 + }, + { + "epoch": 0.6024573266265358, + "grad_norm": 2.045461654663086, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7188277244567871, + "num_tokens": 109721360.0, + "step": 7306 + }, + { + "epoch": 0.6025397872515874, + "grad_norm": 1.9709073305130005, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7334861159324646, + "num_tokens": 109753209.0, + "step": 7307 + }, + { + "epoch": 0.6026222478766389, + "grad_norm": 1.850895643234253, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7628284692764282, + "num_tokens": 109784747.0, + "step": 7308 + }, + { + "epoch": 0.6027047085016904, + "grad_norm": 2.1012256145477295, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.759533703327179, + "num_tokens": 109808084.0, + "step": 7309 + }, + { + "epoch": 0.602787169126742, + "grad_norm": 2.1218316555023193, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7420284748077393, + "num_tokens": 109835449.0, + "step": 7310 + }, + { + "epoch": 0.6028696297517935, + "grad_norm": 2.0159432888031006, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7190650701522827, + "num_tokens": 109863967.0, + "step": 7311 + }, + { + "epoch": 0.602952090376845, + "grad_norm": 1.9825973510742188, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7444093227386475, + "num_tokens": 109891359.0, + "step": 7312 + }, + { + "epoch": 0.6030345510018966, + "grad_norm": 1.8945788145065308, + "learning_rate": 1e-06, + "loss": 0.7656, + "mean_token_accuracy": 0.7601290941238403, + "num_tokens": 109921526.0, + "step": 7313 + }, + { + "epoch": 0.6031170116269481, + "grad_norm": 1.9924277067184448, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7723709344863892, + "num_tokens": 109949907.0, + "step": 7314 + }, + { + "epoch": 0.6031994722519997, + "grad_norm": 1.7992640733718872, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7520781755447388, + "num_tokens": 109984588.0, + "step": 7315 + }, + { + "epoch": 0.6032819328770512, + "grad_norm": 2.1995718479156494, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7569522857666016, + "num_tokens": 110008541.0, + "step": 7316 + }, + { + "epoch": 0.6033643935021027, + "grad_norm": 2.016305685043335, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7751027941703796, + "num_tokens": 110035727.0, + "step": 7317 + }, + { + "epoch": 0.6034468541271543, + "grad_norm": 2.0411300659179688, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.727357029914856, + "num_tokens": 110064717.0, + "step": 7318 + }, + { + "epoch": 0.6035293147522058, + "grad_norm": 1.7412930727005005, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7693083882331848, + "num_tokens": 110100184.0, + "step": 7319 + }, + { + "epoch": 0.6036117753772574, + "grad_norm": 2.080620288848877, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7408478260040283, + "num_tokens": 110125873.0, + "step": 7320 + }, + { + "epoch": 0.6036942360023089, + "grad_norm": 1.9595134258270264, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7541557550430298, + "num_tokens": 110154578.0, + "step": 7321 + }, + { + "epoch": 0.6037766966273604, + "grad_norm": 1.8609715700149536, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7611751556396484, + "num_tokens": 110187456.0, + "step": 7322 + }, + { + "epoch": 0.603859157252412, + "grad_norm": 1.9858556985855103, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7643693685531616, + "num_tokens": 110215262.0, + "step": 7323 + }, + { + "epoch": 0.6039416178774635, + "grad_norm": 1.6907607316970825, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.7889648079872131, + "num_tokens": 110249081.0, + "step": 7324 + }, + { + "epoch": 0.604024078502515, + "grad_norm": 1.8795307874679565, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.758326530456543, + "num_tokens": 110281219.0, + "step": 7325 + }, + { + "epoch": 0.6041065391275666, + "grad_norm": 2.158254384994507, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.747829020023346, + "num_tokens": 110306001.0, + "step": 7326 + }, + { + "epoch": 0.6041889997526181, + "grad_norm": 2.147808074951172, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.755584716796875, + "num_tokens": 110330607.0, + "step": 7327 + }, + { + "epoch": 0.6042714603776697, + "grad_norm": 1.9697394371032715, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7448223233222961, + "num_tokens": 110357561.0, + "step": 7328 + }, + { + "epoch": 0.6043539210027212, + "grad_norm": 1.9577609300613403, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7804293632507324, + "num_tokens": 110385539.0, + "step": 7329 + }, + { + "epoch": 0.6044363816277727, + "grad_norm": 1.9650347232818604, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7458076477050781, + "num_tokens": 110414846.0, + "step": 7330 + }, + { + "epoch": 0.6045188422528243, + "grad_norm": 2.06742525100708, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7667220830917358, + "num_tokens": 110442613.0, + "step": 7331 + }, + { + "epoch": 0.6046013028778758, + "grad_norm": 2.297851085662842, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7538057565689087, + "num_tokens": 110465531.0, + "step": 7332 + }, + { + "epoch": 0.6046837635029274, + "grad_norm": 2.0157322883605957, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7552877068519592, + "num_tokens": 110492631.0, + "step": 7333 + }, + { + "epoch": 0.6047662241279789, + "grad_norm": 1.8927319049835205, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7711495757102966, + "num_tokens": 110523821.0, + "step": 7334 + }, + { + "epoch": 0.6048486847530304, + "grad_norm": 1.8131943941116333, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7106258869171143, + "num_tokens": 110559085.0, + "step": 7335 + }, + { + "epoch": 0.604931145378082, + "grad_norm": 1.845856785774231, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7362077236175537, + "num_tokens": 110591211.0, + "step": 7336 + }, + { + "epoch": 0.6050136060031335, + "grad_norm": 1.856343388557434, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7758705019950867, + "num_tokens": 110625056.0, + "step": 7337 + }, + { + "epoch": 0.605096066628185, + "grad_norm": 1.9920729398727417, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.7602057456970215, + "num_tokens": 110654818.0, + "step": 7338 + }, + { + "epoch": 0.6051785272532366, + "grad_norm": 1.9987773895263672, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7153459787368774, + "num_tokens": 110685478.0, + "step": 7339 + }, + { + "epoch": 0.6052609878782881, + "grad_norm": 1.9473230838775635, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7237541079521179, + "num_tokens": 110718508.0, + "step": 7340 + }, + { + "epoch": 0.6053434485033397, + "grad_norm": 1.8546983003616333, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7152350544929504, + "num_tokens": 110751197.0, + "step": 7341 + }, + { + "epoch": 0.6054259091283912, + "grad_norm": 1.8390647172927856, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7178372144699097, + "num_tokens": 110786871.0, + "step": 7342 + }, + { + "epoch": 0.6055083697534427, + "grad_norm": 1.8873553276062012, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7232838273048401, + "num_tokens": 110822125.0, + "step": 7343 + }, + { + "epoch": 0.6055908303784943, + "grad_norm": 1.9606496095657349, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7555757761001587, + "num_tokens": 110850989.0, + "step": 7344 + }, + { + "epoch": 0.6056732910035458, + "grad_norm": 1.8702763319015503, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7244230508804321, + "num_tokens": 110884306.0, + "step": 7345 + }, + { + "epoch": 0.6057557516285973, + "grad_norm": 1.7512266635894775, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.7877135276794434, + "num_tokens": 110916531.0, + "step": 7346 + }, + { + "epoch": 0.6058382122536489, + "grad_norm": 1.7893284559249878, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7849717140197754, + "num_tokens": 110948861.0, + "step": 7347 + }, + { + "epoch": 0.6059206728787004, + "grad_norm": 2.0427191257476807, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.750407338142395, + "num_tokens": 110977581.0, + "step": 7348 + }, + { + "epoch": 0.606003133503752, + "grad_norm": 1.7712595462799072, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7721275091171265, + "num_tokens": 111009856.0, + "step": 7349 + }, + { + "epoch": 0.6060855941288035, + "grad_norm": 1.824668049812317, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.7597193717956543, + "num_tokens": 111044099.0, + "step": 7350 + }, + { + "epoch": 0.606168054753855, + "grad_norm": 2.1406688690185547, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7361545562744141, + "num_tokens": 111070269.0, + "step": 7351 + }, + { + "epoch": 0.6062505153789066, + "grad_norm": 2.298910140991211, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.715954065322876, + "num_tokens": 111092624.0, + "step": 7352 + }, + { + "epoch": 0.6063329760039581, + "grad_norm": 2.0938289165496826, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7484081387519836, + "num_tokens": 111118821.0, + "step": 7353 + }, + { + "epoch": 0.6064154366290097, + "grad_norm": 1.9894353151321411, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7587767243385315, + "num_tokens": 111148274.0, + "step": 7354 + }, + { + "epoch": 0.6064978972540612, + "grad_norm": 2.012157440185547, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7262294292449951, + "num_tokens": 111180087.0, + "step": 7355 + }, + { + "epoch": 0.6065803578791127, + "grad_norm": 1.9801092147827148, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.7905198931694031, + "num_tokens": 111207473.0, + "step": 7356 + }, + { + "epoch": 0.6066628185041643, + "grad_norm": 1.868887186050415, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7839718461036682, + "num_tokens": 111238689.0, + "step": 7357 + }, + { + "epoch": 0.6067452791292158, + "grad_norm": 1.658341407775879, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7526383399963379, + "num_tokens": 111278065.0, + "step": 7358 + }, + { + "epoch": 0.6068277397542673, + "grad_norm": 1.8521466255187988, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7724542617797852, + "num_tokens": 111308800.0, + "step": 7359 + }, + { + "epoch": 0.6069102003793189, + "grad_norm": 1.8413889408111572, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7376495599746704, + "num_tokens": 111340359.0, + "step": 7360 + }, + { + "epoch": 0.6069926610043704, + "grad_norm": 2.2386281490325928, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7272164821624756, + "num_tokens": 111365644.0, + "step": 7361 + }, + { + "epoch": 0.6070751216294219, + "grad_norm": 2.120304822921753, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7394473552703857, + "num_tokens": 111392812.0, + "step": 7362 + }, + { + "epoch": 0.6071575822544735, + "grad_norm": 2.4124791622161865, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7237826585769653, + "num_tokens": 111416379.0, + "step": 7363 + }, + { + "epoch": 0.607240042879525, + "grad_norm": 1.9663761854171753, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7865917086601257, + "num_tokens": 111444102.0, + "step": 7364 + }, + { + "epoch": 0.6073225035045766, + "grad_norm": 1.968363881111145, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7299448251724243, + "num_tokens": 111475709.0, + "step": 7365 + }, + { + "epoch": 0.6074049641296281, + "grad_norm": 1.8854479789733887, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.77503901720047, + "num_tokens": 111508646.0, + "step": 7366 + }, + { + "epoch": 0.6074874247546797, + "grad_norm": 1.931850552558899, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.769460916519165, + "num_tokens": 111538808.0, + "step": 7367 + }, + { + "epoch": 0.6075698853797312, + "grad_norm": 2.008397340774536, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7520493865013123, + "num_tokens": 111567045.0, + "step": 7368 + }, + { + "epoch": 0.6076523460047827, + "grad_norm": 2.0538828372955322, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7478168606758118, + "num_tokens": 111593952.0, + "step": 7369 + }, + { + "epoch": 0.6077348066298343, + "grad_norm": 1.8578605651855469, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7663056254386902, + "num_tokens": 111629795.0, + "step": 7370 + }, + { + "epoch": 0.6078172672548858, + "grad_norm": 1.800521969795227, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7573884129524231, + "num_tokens": 111666974.0, + "step": 7371 + }, + { + "epoch": 0.6078997278799373, + "grad_norm": 1.884435772895813, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7395014762878418, + "num_tokens": 111698761.0, + "step": 7372 + }, + { + "epoch": 0.6079821885049889, + "grad_norm": 2.141671895980835, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7159174680709839, + "num_tokens": 111724656.0, + "step": 7373 + }, + { + "epoch": 0.6080646491300404, + "grad_norm": 2.1067299842834473, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7344582080841064, + "num_tokens": 111754263.0, + "step": 7374 + }, + { + "epoch": 0.6081471097550919, + "grad_norm": 1.96379554271698, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7297327518463135, + "num_tokens": 111784863.0, + "step": 7375 + }, + { + "epoch": 0.6082295703801435, + "grad_norm": 2.0971245765686035, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7440450191497803, + "num_tokens": 111813574.0, + "step": 7376 + }, + { + "epoch": 0.608312031005195, + "grad_norm": 1.8886131048202515, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.773171067237854, + "num_tokens": 111846169.0, + "step": 7377 + }, + { + "epoch": 0.6083944916302465, + "grad_norm": 2.08410382270813, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7152819633483887, + "num_tokens": 111874555.0, + "step": 7378 + }, + { + "epoch": 0.6084769522552981, + "grad_norm": 1.891953945159912, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7735546231269836, + "num_tokens": 111904864.0, + "step": 7379 + }, + { + "epoch": 0.6085594128803496, + "grad_norm": 1.9698317050933838, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7188869714736938, + "num_tokens": 111933733.0, + "step": 7380 + }, + { + "epoch": 0.6086418735054012, + "grad_norm": 1.9262053966522217, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.748956024646759, + "num_tokens": 111963641.0, + "step": 7381 + }, + { + "epoch": 0.6087243341304527, + "grad_norm": 1.9104450941085815, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7071083784103394, + "num_tokens": 111996250.0, + "step": 7382 + }, + { + "epoch": 0.6088067947555043, + "grad_norm": 1.7589607238769531, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7670350074768066, + "num_tokens": 112029499.0, + "step": 7383 + }, + { + "epoch": 0.6088892553805558, + "grad_norm": 1.9638612270355225, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7592393159866333, + "num_tokens": 112061004.0, + "step": 7384 + }, + { + "epoch": 0.6089717160056073, + "grad_norm": 2.0677969455718994, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7314528822898865, + "num_tokens": 112090449.0, + "step": 7385 + }, + { + "epoch": 0.6090541766306589, + "grad_norm": 2.2499587535858154, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7082556486129761, + "num_tokens": 112114509.0, + "step": 7386 + }, + { + "epoch": 0.6091366372557104, + "grad_norm": 1.8753776550292969, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.750169038772583, + "num_tokens": 112146083.0, + "step": 7387 + }, + { + "epoch": 0.6092190978807619, + "grad_norm": 2.378997325897217, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7440177202224731, + "num_tokens": 112167295.0, + "step": 7388 + }, + { + "epoch": 0.6093015585058135, + "grad_norm": 1.8844795227050781, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7251688241958618, + "num_tokens": 112201384.0, + "step": 7389 + }, + { + "epoch": 0.609384019130865, + "grad_norm": 2.0334110260009766, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7357721924781799, + "num_tokens": 112230677.0, + "step": 7390 + }, + { + "epoch": 0.6094664797559165, + "grad_norm": 1.874346137046814, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7389202117919922, + "num_tokens": 112263225.0, + "step": 7391 + }, + { + "epoch": 0.6095489403809681, + "grad_norm": 1.7280733585357666, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7712870240211487, + "num_tokens": 112296916.0, + "step": 7392 + }, + { + "epoch": 0.6096314010060196, + "grad_norm": 1.8147940635681152, + "learning_rate": 1e-06, + "loss": 0.5842, + "mean_token_accuracy": 0.8024020791053772, + "num_tokens": 112327200.0, + "step": 7393 + }, + { + "epoch": 0.6097138616310712, + "grad_norm": 1.8553603887557983, + "learning_rate": 1e-06, + "loss": 0.5873, + "mean_token_accuracy": 0.8041381239891052, + "num_tokens": 112354783.0, + "step": 7394 + }, + { + "epoch": 0.6097963222561227, + "grad_norm": 2.084178924560547, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7522149085998535, + "num_tokens": 112383787.0, + "step": 7395 + }, + { + "epoch": 0.6098787828811743, + "grad_norm": 2.02117657661438, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7341662645339966, + "num_tokens": 112413291.0, + "step": 7396 + }, + { + "epoch": 0.6099612435062258, + "grad_norm": 1.9327400922775269, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7642252445220947, + "num_tokens": 112442631.0, + "step": 7397 + }, + { + "epoch": 0.6100437041312773, + "grad_norm": 1.9549623727798462, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7200379371643066, + "num_tokens": 112476124.0, + "step": 7398 + }, + { + "epoch": 0.6101261647563289, + "grad_norm": 2.38761043548584, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7738146185874939, + "num_tokens": 112497248.0, + "step": 7399 + }, + { + "epoch": 0.6102086253813804, + "grad_norm": 1.8134815692901611, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7286526560783386, + "num_tokens": 112528905.0, + "step": 7400 + }, + { + "epoch": 0.6102910860064319, + "grad_norm": 1.8048856258392334, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.7893505692481995, + "num_tokens": 112564611.0, + "step": 7401 + }, + { + "epoch": 0.6103735466314835, + "grad_norm": 1.830588459968567, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7759806513786316, + "num_tokens": 112598664.0, + "step": 7402 + }, + { + "epoch": 0.610456007256535, + "grad_norm": 2.37799334526062, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7420449256896973, + "num_tokens": 112619411.0, + "step": 7403 + }, + { + "epoch": 0.6105384678815865, + "grad_norm": 1.851920485496521, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7561889290809631, + "num_tokens": 112650988.0, + "step": 7404 + }, + { + "epoch": 0.6106209285066381, + "grad_norm": 2.071887254714966, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7449157238006592, + "num_tokens": 112676672.0, + "step": 7405 + }, + { + "epoch": 0.6107033891316896, + "grad_norm": 2.0575268268585205, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7408784031867981, + "num_tokens": 112705832.0, + "step": 7406 + }, + { + "epoch": 0.6107858497567411, + "grad_norm": 1.8177000284194946, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.7920743227005005, + "num_tokens": 112736943.0, + "step": 7407 + }, + { + "epoch": 0.6108683103817927, + "grad_norm": 1.9145746231079102, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7429543137550354, + "num_tokens": 112769530.0, + "step": 7408 + }, + { + "epoch": 0.6109507710068443, + "grad_norm": 2.2634055614471436, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7448912262916565, + "num_tokens": 112795469.0, + "step": 7409 + }, + { + "epoch": 0.6110332316318958, + "grad_norm": 2.254307985305786, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7473222613334656, + "num_tokens": 112819196.0, + "step": 7410 + }, + { + "epoch": 0.6111156922569473, + "grad_norm": 1.8888965845108032, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7810854911804199, + "num_tokens": 112850263.0, + "step": 7411 + }, + { + "epoch": 0.6111981528819989, + "grad_norm": 1.7394605875015259, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7726250290870667, + "num_tokens": 112885771.0, + "step": 7412 + }, + { + "epoch": 0.6112806135070504, + "grad_norm": 2.073005437850952, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7499366998672485, + "num_tokens": 112914215.0, + "step": 7413 + }, + { + "epoch": 0.6113630741321019, + "grad_norm": 1.8237593173980713, + "learning_rate": 1e-06, + "loss": 0.7617, + "mean_token_accuracy": 0.7530995607376099, + "num_tokens": 112946684.0, + "step": 7414 + }, + { + "epoch": 0.6114455347571535, + "grad_norm": 1.987252950668335, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7642389535903931, + "num_tokens": 112977906.0, + "step": 7415 + }, + { + "epoch": 0.611527995382205, + "grad_norm": 1.9133293628692627, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7351729273796082, + "num_tokens": 113008520.0, + "step": 7416 + }, + { + "epoch": 0.6116104560072565, + "grad_norm": 2.2084169387817383, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.708487868309021, + "num_tokens": 113035335.0, + "step": 7417 + }, + { + "epoch": 0.6116929166323081, + "grad_norm": 1.9856810569763184, + "learning_rate": 1e-06, + "loss": 0.7486, + "mean_token_accuracy": 0.7620813846588135, + "num_tokens": 113064822.0, + "step": 7418 + }, + { + "epoch": 0.6117753772573596, + "grad_norm": 1.9271042346954346, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7196215391159058, + "num_tokens": 113099094.0, + "step": 7419 + }, + { + "epoch": 0.6118578378824111, + "grad_norm": 2.031081438064575, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7145795822143555, + "num_tokens": 113128040.0, + "step": 7420 + }, + { + "epoch": 0.6119402985074627, + "grad_norm": 2.2363204956054688, + "learning_rate": 1e-06, + "loss": 0.5653, + "mean_token_accuracy": 0.8079089522361755, + "num_tokens": 113149880.0, + "step": 7421 + }, + { + "epoch": 0.6120227591325142, + "grad_norm": 1.9191166162490845, + "learning_rate": 1e-06, + "loss": 0.5893, + "mean_token_accuracy": 0.8030591011047363, + "num_tokens": 113180025.0, + "step": 7422 + }, + { + "epoch": 0.6121052197575657, + "grad_norm": 1.8010708093643188, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.762115478515625, + "num_tokens": 113216427.0, + "step": 7423 + }, + { + "epoch": 0.6121876803826173, + "grad_norm": 1.9912337064743042, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7447595596313477, + "num_tokens": 113244196.0, + "step": 7424 + }, + { + "epoch": 0.6122701410076689, + "grad_norm": 2.106826066970825, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7312791347503662, + "num_tokens": 113271566.0, + "step": 7425 + }, + { + "epoch": 0.6123526016327204, + "grad_norm": 1.970001459121704, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7024378180503845, + "num_tokens": 113303597.0, + "step": 7426 + }, + { + "epoch": 0.6124350622577719, + "grad_norm": 2.0977885723114014, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7510864734649658, + "num_tokens": 113330740.0, + "step": 7427 + }, + { + "epoch": 0.6125175228828235, + "grad_norm": 2.0560343265533447, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7175745964050293, + "num_tokens": 113357972.0, + "step": 7428 + }, + { + "epoch": 0.612599983507875, + "grad_norm": 2.241511344909668, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7280759215354919, + "num_tokens": 113380243.0, + "step": 7429 + }, + { + "epoch": 0.6126824441329265, + "grad_norm": 2.015746593475342, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7372465133666992, + "num_tokens": 113410762.0, + "step": 7430 + }, + { + "epoch": 0.6127649047579781, + "grad_norm": 1.6255087852478027, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.8078698515892029, + "num_tokens": 113446015.0, + "step": 7431 + }, + { + "epoch": 0.6128473653830296, + "grad_norm": 1.892835259437561, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7356194257736206, + "num_tokens": 113478474.0, + "step": 7432 + }, + { + "epoch": 0.6129298260080811, + "grad_norm": 1.8936328887939453, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.735363245010376, + "num_tokens": 113513687.0, + "step": 7433 + }, + { + "epoch": 0.6130122866331327, + "grad_norm": 1.9069479703903198, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7656656503677368, + "num_tokens": 113544318.0, + "step": 7434 + }, + { + "epoch": 0.6130947472581842, + "grad_norm": 1.7619494199752808, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7508822083473206, + "num_tokens": 113576918.0, + "step": 7435 + }, + { + "epoch": 0.6131772078832357, + "grad_norm": 1.8757691383361816, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7654903531074524, + "num_tokens": 113608485.0, + "step": 7436 + }, + { + "epoch": 0.6132596685082873, + "grad_norm": 2.0039947032928467, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7103009223937988, + "num_tokens": 113639220.0, + "step": 7437 + }, + { + "epoch": 0.6133421291333389, + "grad_norm": 1.9767414331436157, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7458919286727905, + "num_tokens": 113669434.0, + "step": 7438 + }, + { + "epoch": 0.6134245897583903, + "grad_norm": 1.9066280126571655, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7078856229782104, + "num_tokens": 113701829.0, + "step": 7439 + }, + { + "epoch": 0.6135070503834419, + "grad_norm": 1.9953795671463013, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7045674324035645, + "num_tokens": 113735445.0, + "step": 7440 + }, + { + "epoch": 0.6135895110084935, + "grad_norm": 2.048356294631958, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7184127569198608, + "num_tokens": 113764457.0, + "step": 7441 + }, + { + "epoch": 0.613671971633545, + "grad_norm": 1.951008677482605, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7597781419754028, + "num_tokens": 113794251.0, + "step": 7442 + }, + { + "epoch": 0.6137544322585965, + "grad_norm": 1.8099086284637451, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7724719643592834, + "num_tokens": 113827334.0, + "step": 7443 + }, + { + "epoch": 0.6138368928836481, + "grad_norm": 1.7318435907363892, + "learning_rate": 1e-06, + "loss": 0.7486, + "mean_token_accuracy": 0.75776207447052, + "num_tokens": 113862627.0, + "step": 7444 + }, + { + "epoch": 0.6139193535086996, + "grad_norm": 1.8101197481155396, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.706741452217102, + "num_tokens": 113899906.0, + "step": 7445 + }, + { + "epoch": 0.6140018141337511, + "grad_norm": 1.9435628652572632, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7545358538627625, + "num_tokens": 113929566.0, + "step": 7446 + }, + { + "epoch": 0.6140842747588027, + "grad_norm": 2.0802412033081055, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7416531443595886, + "num_tokens": 113957316.0, + "step": 7447 + }, + { + "epoch": 0.6141667353838542, + "grad_norm": 1.6968392133712769, + "learning_rate": 1e-06, + "loss": 0.6, + "mean_token_accuracy": 0.7959778904914856, + "num_tokens": 113989111.0, + "step": 7448 + }, + { + "epoch": 0.6142491960089057, + "grad_norm": 2.2522342205047607, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7411601543426514, + "num_tokens": 114015250.0, + "step": 7449 + }, + { + "epoch": 0.6143316566339573, + "grad_norm": 1.7711232900619507, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7662385702133179, + "num_tokens": 114049524.0, + "step": 7450 + }, + { + "epoch": 0.6144141172590089, + "grad_norm": 1.887420654296875, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7589743137359619, + "num_tokens": 114081732.0, + "step": 7451 + }, + { + "epoch": 0.6144965778840603, + "grad_norm": 1.9222222566604614, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7633047699928284, + "num_tokens": 114115182.0, + "step": 7452 + }, + { + "epoch": 0.6145790385091119, + "grad_norm": 2.019946575164795, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7478090524673462, + "num_tokens": 114143360.0, + "step": 7453 + }, + { + "epoch": 0.6146614991341635, + "grad_norm": 2.0256776809692383, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.755073070526123, + "num_tokens": 114171612.0, + "step": 7454 + }, + { + "epoch": 0.6147439597592149, + "grad_norm": 1.7874550819396973, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7728672027587891, + "num_tokens": 114204178.0, + "step": 7455 + }, + { + "epoch": 0.6148264203842665, + "grad_norm": 1.9248309135437012, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7175753116607666, + "num_tokens": 114234269.0, + "step": 7456 + }, + { + "epoch": 0.6149088810093181, + "grad_norm": 1.8372946977615356, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7833125591278076, + "num_tokens": 114267630.0, + "step": 7457 + }, + { + "epoch": 0.6149913416343696, + "grad_norm": 1.7821664810180664, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.7836490869522095, + "num_tokens": 114300226.0, + "step": 7458 + }, + { + "epoch": 0.6150738022594211, + "grad_norm": 1.928378701210022, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7550466060638428, + "num_tokens": 114332924.0, + "step": 7459 + }, + { + "epoch": 0.6151562628844727, + "grad_norm": 1.8025919198989868, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7810289859771729, + "num_tokens": 114368030.0, + "step": 7460 + }, + { + "epoch": 0.6152387235095242, + "grad_norm": 2.2799127101898193, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7531844973564148, + "num_tokens": 114390733.0, + "step": 7461 + }, + { + "epoch": 0.6153211841345757, + "grad_norm": 1.9672831296920776, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7363781929016113, + "num_tokens": 114421640.0, + "step": 7462 + }, + { + "epoch": 0.6154036447596273, + "grad_norm": 1.8163561820983887, + "learning_rate": 1e-06, + "loss": 0.5386, + "mean_token_accuracy": 0.8205214738845825, + "num_tokens": 114452894.0, + "step": 7463 + }, + { + "epoch": 0.6154861053846788, + "grad_norm": 2.02412748336792, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.762110710144043, + "num_tokens": 114482697.0, + "step": 7464 + }, + { + "epoch": 0.6155685660097303, + "grad_norm": 2.4616429805755615, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7167576551437378, + "num_tokens": 114504001.0, + "step": 7465 + }, + { + "epoch": 0.6156510266347819, + "grad_norm": 1.9140039682388306, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7504450678825378, + "num_tokens": 114537389.0, + "step": 7466 + }, + { + "epoch": 0.6157334872598335, + "grad_norm": 1.70462965965271, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7748297452926636, + "num_tokens": 114573814.0, + "step": 7467 + }, + { + "epoch": 0.6158159478848849, + "grad_norm": 2.3586862087249756, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7715561985969543, + "num_tokens": 114594673.0, + "step": 7468 + }, + { + "epoch": 0.6158984085099365, + "grad_norm": 1.835602045059204, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7376055121421814, + "num_tokens": 114625197.0, + "step": 7469 + }, + { + "epoch": 0.6159808691349881, + "grad_norm": 1.7561218738555908, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7892000079154968, + "num_tokens": 114658138.0, + "step": 7470 + }, + { + "epoch": 0.6160633297600396, + "grad_norm": 2.004394769668579, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7463142275810242, + "num_tokens": 114689804.0, + "step": 7471 + }, + { + "epoch": 0.6161457903850911, + "grad_norm": 1.7463970184326172, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7758831977844238, + "num_tokens": 114724901.0, + "step": 7472 + }, + { + "epoch": 0.6162282510101427, + "grad_norm": 1.8293033838272095, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.7966969013214111, + "num_tokens": 114756277.0, + "step": 7473 + }, + { + "epoch": 0.6163107116351942, + "grad_norm": 1.9077544212341309, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7243777513504028, + "num_tokens": 114789036.0, + "step": 7474 + }, + { + "epoch": 0.6163931722602457, + "grad_norm": 2.0135512351989746, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7542859315872192, + "num_tokens": 114817524.0, + "step": 7475 + }, + { + "epoch": 0.6164756328852973, + "grad_norm": 1.7748844623565674, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.75290846824646, + "num_tokens": 114848862.0, + "step": 7476 + }, + { + "epoch": 0.6165580935103488, + "grad_norm": 1.8272583484649658, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7333426475524902, + "num_tokens": 114885592.0, + "step": 7477 + }, + { + "epoch": 0.6166405541354003, + "grad_norm": 1.6685229539871216, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.7885688543319702, + "num_tokens": 114924574.0, + "step": 7478 + }, + { + "epoch": 0.6167230147604519, + "grad_norm": 2.006661891937256, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7455739974975586, + "num_tokens": 114956081.0, + "step": 7479 + }, + { + "epoch": 0.6168054753855035, + "grad_norm": 1.8825613260269165, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7516520023345947, + "num_tokens": 114986833.0, + "step": 7480 + }, + { + "epoch": 0.6168879360105549, + "grad_norm": 1.9335880279541016, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7380661368370056, + "num_tokens": 115016997.0, + "step": 7481 + }, + { + "epoch": 0.6169703966356065, + "grad_norm": 2.0747556686401367, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.746610164642334, + "num_tokens": 115046773.0, + "step": 7482 + }, + { + "epoch": 0.6170528572606581, + "grad_norm": 1.9058765172958374, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7329246997833252, + "num_tokens": 115079943.0, + "step": 7483 + }, + { + "epoch": 0.6171353178857095, + "grad_norm": 2.1759536266326904, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7224841117858887, + "num_tokens": 115109863.0, + "step": 7484 + }, + { + "epoch": 0.6172177785107611, + "grad_norm": 1.9118386507034302, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7525581121444702, + "num_tokens": 115139812.0, + "step": 7485 + }, + { + "epoch": 0.6173002391358127, + "grad_norm": 1.965281367301941, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7280569672584534, + "num_tokens": 115170093.0, + "step": 7486 + }, + { + "epoch": 0.6173826997608642, + "grad_norm": 1.7185310125350952, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7690032124519348, + "num_tokens": 115203258.0, + "step": 7487 + }, + { + "epoch": 0.6174651603859157, + "grad_norm": 1.9739384651184082, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7455337047576904, + "num_tokens": 115236101.0, + "step": 7488 + }, + { + "epoch": 0.6175476210109673, + "grad_norm": 2.2917797565460205, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7364226579666138, + "num_tokens": 115258609.0, + "step": 7489 + }, + { + "epoch": 0.6176300816360188, + "grad_norm": 1.877377986907959, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.7953225374221802, + "num_tokens": 115289237.0, + "step": 7490 + }, + { + "epoch": 0.6177125422610703, + "grad_norm": 2.0187108516693115, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.750370442867279, + "num_tokens": 115317121.0, + "step": 7491 + }, + { + "epoch": 0.6177950028861219, + "grad_norm": 2.179041862487793, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7447503805160522, + "num_tokens": 115341178.0, + "step": 7492 + }, + { + "epoch": 0.6178774635111735, + "grad_norm": 2.0238049030303955, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7226362228393555, + "num_tokens": 115370186.0, + "step": 7493 + }, + { + "epoch": 0.6179599241362249, + "grad_norm": 1.9913079738616943, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7548513412475586, + "num_tokens": 115398322.0, + "step": 7494 + }, + { + "epoch": 0.6180423847612765, + "grad_norm": 2.170851469039917, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7742059230804443, + "num_tokens": 115425715.0, + "step": 7495 + }, + { + "epoch": 0.6181248453863281, + "grad_norm": 2.1373603343963623, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.7915686368942261, + "num_tokens": 115451640.0, + "step": 7496 + }, + { + "epoch": 0.6182073060113795, + "grad_norm": 1.7629021406173706, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7776401042938232, + "num_tokens": 115489466.0, + "step": 7497 + }, + { + "epoch": 0.6182897666364311, + "grad_norm": 2.9269721508026123, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7484110593795776, + "num_tokens": 115507295.0, + "step": 7498 + }, + { + "epoch": 0.6183722272614827, + "grad_norm": 1.8691879510879517, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7748436331748962, + "num_tokens": 115543737.0, + "step": 7499 + }, + { + "epoch": 0.6184546878865341, + "grad_norm": 2.1696395874023438, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7638620138168335, + "num_tokens": 115571026.0, + "step": 7500 + }, + { + "epoch": 0.6185371485115857, + "grad_norm": 1.881394624710083, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7360984086990356, + "num_tokens": 115603476.0, + "step": 7501 + }, + { + "epoch": 0.6186196091366373, + "grad_norm": 1.903497338294983, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7861151695251465, + "num_tokens": 115635395.0, + "step": 7502 + }, + { + "epoch": 0.6187020697616888, + "grad_norm": 1.8025825023651123, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7487656474113464, + "num_tokens": 115667365.0, + "step": 7503 + }, + { + "epoch": 0.6187845303867403, + "grad_norm": 1.9434828758239746, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7412688732147217, + "num_tokens": 115696831.0, + "step": 7504 + }, + { + "epoch": 0.6188669910117919, + "grad_norm": 1.753069281578064, + "learning_rate": 1e-06, + "loss": 0.7656, + "mean_token_accuracy": 0.7544028162956238, + "num_tokens": 115733470.0, + "step": 7505 + }, + { + "epoch": 0.6189494516368435, + "grad_norm": 2.2951791286468506, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7164890766143799, + "num_tokens": 115757065.0, + "step": 7506 + }, + { + "epoch": 0.6190319122618949, + "grad_norm": 1.7540329694747925, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.769007682800293, + "num_tokens": 115790721.0, + "step": 7507 + }, + { + "epoch": 0.6191143728869465, + "grad_norm": 2.097414255142212, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7270230054855347, + "num_tokens": 115820978.0, + "step": 7508 + }, + { + "epoch": 0.6191968335119981, + "grad_norm": 2.085710287094116, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7610399723052979, + "num_tokens": 115846395.0, + "step": 7509 + }, + { + "epoch": 0.6192792941370495, + "grad_norm": 1.9181009531021118, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7627512216567993, + "num_tokens": 115876330.0, + "step": 7510 + }, + { + "epoch": 0.6193617547621011, + "grad_norm": 1.976973056793213, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7693255543708801, + "num_tokens": 115906309.0, + "step": 7511 + }, + { + "epoch": 0.6194442153871527, + "grad_norm": 1.9251782894134521, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7290980815887451, + "num_tokens": 115936936.0, + "step": 7512 + }, + { + "epoch": 0.6195266760122041, + "grad_norm": 2.006556272506714, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7420895099639893, + "num_tokens": 115965999.0, + "step": 7513 + }, + { + "epoch": 0.6196091366372557, + "grad_norm": 1.8274215459823608, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7833619117736816, + "num_tokens": 115997852.0, + "step": 7514 + }, + { + "epoch": 0.6196915972623073, + "grad_norm": 2.1335299015045166, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7365207672119141, + "num_tokens": 116026031.0, + "step": 7515 + }, + { + "epoch": 0.6197740578873587, + "grad_norm": 1.9557994604110718, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.764340877532959, + "num_tokens": 116054949.0, + "step": 7516 + }, + { + "epoch": 0.6198565185124103, + "grad_norm": 2.1258575916290283, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7366557121276855, + "num_tokens": 116082352.0, + "step": 7517 + }, + { + "epoch": 0.6199389791374619, + "grad_norm": 1.850287914276123, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7367895245552063, + "num_tokens": 116114606.0, + "step": 7518 + }, + { + "epoch": 0.6200214397625134, + "grad_norm": 1.9954758882522583, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7054442167282104, + "num_tokens": 116145139.0, + "step": 7519 + }, + { + "epoch": 0.6201039003875649, + "grad_norm": 2.3724656105041504, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7310596108436584, + "num_tokens": 116165068.0, + "step": 7520 + }, + { + "epoch": 0.6201863610126165, + "grad_norm": 1.8482049703598022, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7574961185455322, + "num_tokens": 116197934.0, + "step": 7521 + }, + { + "epoch": 0.620268821637668, + "grad_norm": 1.9672828912734985, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7457615733146667, + "num_tokens": 116230869.0, + "step": 7522 + }, + { + "epoch": 0.6203512822627195, + "grad_norm": 1.9488037824630737, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.6981368660926819, + "num_tokens": 116264234.0, + "step": 7523 + }, + { + "epoch": 0.6204337428877711, + "grad_norm": 1.9566529989242554, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.791045069694519, + "num_tokens": 116292211.0, + "step": 7524 + }, + { + "epoch": 0.6205162035128227, + "grad_norm": 2.173750162124634, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7437703609466553, + "num_tokens": 116319325.0, + "step": 7525 + }, + { + "epoch": 0.6205986641378741, + "grad_norm": 1.7860956192016602, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7535709738731384, + "num_tokens": 116353219.0, + "step": 7526 + }, + { + "epoch": 0.6206811247629257, + "grad_norm": 2.0793120861053467, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.750085711479187, + "num_tokens": 116378202.0, + "step": 7527 + }, + { + "epoch": 0.6207635853879773, + "grad_norm": 1.7839280366897583, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7603325843811035, + "num_tokens": 116410914.0, + "step": 7528 + }, + { + "epoch": 0.6208460460130287, + "grad_norm": 1.9275925159454346, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.7857135534286499, + "num_tokens": 116443938.0, + "step": 7529 + }, + { + "epoch": 0.6209285066380803, + "grad_norm": 2.0195164680480957, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7440330386161804, + "num_tokens": 116471277.0, + "step": 7530 + }, + { + "epoch": 0.6210109672631319, + "grad_norm": 1.9105489253997803, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.73566734790802, + "num_tokens": 116502851.0, + "step": 7531 + }, + { + "epoch": 0.6210934278881834, + "grad_norm": 1.8786954879760742, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.7631115913391113, + "num_tokens": 116532488.0, + "step": 7532 + }, + { + "epoch": 0.6211758885132349, + "grad_norm": 2.015852212905884, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7268128395080566, + "num_tokens": 116561746.0, + "step": 7533 + }, + { + "epoch": 0.6212583491382865, + "grad_norm": 1.8477039337158203, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7650304436683655, + "num_tokens": 116591494.0, + "step": 7534 + }, + { + "epoch": 0.621340809763338, + "grad_norm": 1.7895675897598267, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7447795867919922, + "num_tokens": 116625540.0, + "step": 7535 + }, + { + "epoch": 0.6214232703883895, + "grad_norm": 1.8399243354797363, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7282080054283142, + "num_tokens": 116658696.0, + "step": 7536 + }, + { + "epoch": 0.6215057310134411, + "grad_norm": 2.030595302581787, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7280972599983215, + "num_tokens": 116686611.0, + "step": 7537 + }, + { + "epoch": 0.6215881916384927, + "grad_norm": 1.741094708442688, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7439008355140686, + "num_tokens": 116723216.0, + "step": 7538 + }, + { + "epoch": 0.6216706522635441, + "grad_norm": 2.018231153488159, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7290953993797302, + "num_tokens": 116753894.0, + "step": 7539 + }, + { + "epoch": 0.6217531128885957, + "grad_norm": 1.7060377597808838, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.7639464139938354, + "num_tokens": 116788394.0, + "step": 7540 + }, + { + "epoch": 0.6218355735136473, + "grad_norm": 2.0009818077087402, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7230708599090576, + "num_tokens": 116817749.0, + "step": 7541 + }, + { + "epoch": 0.6219180341386987, + "grad_norm": 1.9249123334884644, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7320860028266907, + "num_tokens": 116850969.0, + "step": 7542 + }, + { + "epoch": 0.6220004947637503, + "grad_norm": 1.7333877086639404, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7861868143081665, + "num_tokens": 116884884.0, + "step": 7543 + }, + { + "epoch": 0.6220829553888019, + "grad_norm": 1.825098991394043, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7076457738876343, + "num_tokens": 116921389.0, + "step": 7544 + }, + { + "epoch": 0.6221654160138533, + "grad_norm": 1.640136480331421, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.7991299629211426, + "num_tokens": 116961874.0, + "step": 7545 + }, + { + "epoch": 0.6222478766389049, + "grad_norm": 1.892282485961914, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7606701850891113, + "num_tokens": 116993316.0, + "step": 7546 + }, + { + "epoch": 0.6223303372639565, + "grad_norm": 1.799232840538025, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.724919319152832, + "num_tokens": 117030882.0, + "step": 7547 + }, + { + "epoch": 0.622412797889008, + "grad_norm": 1.8888306617736816, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7769045233726501, + "num_tokens": 117061819.0, + "step": 7548 + }, + { + "epoch": 0.6224952585140595, + "grad_norm": 1.7454723119735718, + "learning_rate": 1e-06, + "loss": 0.786, + "mean_token_accuracy": 0.746335506439209, + "num_tokens": 117098538.0, + "step": 7549 + }, + { + "epoch": 0.6225777191391111, + "grad_norm": 2.008363723754883, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7584409117698669, + "num_tokens": 117126079.0, + "step": 7550 + }, + { + "epoch": 0.6226601797641627, + "grad_norm": 1.7779580354690552, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7470978498458862, + "num_tokens": 117160432.0, + "step": 7551 + }, + { + "epoch": 0.6227426403892141, + "grad_norm": 1.8504810333251953, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7476774454116821, + "num_tokens": 117192702.0, + "step": 7552 + }, + { + "epoch": 0.6228251010142657, + "grad_norm": 1.9102413654327393, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7713722586631775, + "num_tokens": 117222785.0, + "step": 7553 + }, + { + "epoch": 0.6229075616393173, + "grad_norm": 1.8244988918304443, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.7695454359054565, + "num_tokens": 117255383.0, + "step": 7554 + }, + { + "epoch": 0.6229900222643687, + "grad_norm": 1.9294215440750122, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7299875020980835, + "num_tokens": 117291162.0, + "step": 7555 + }, + { + "epoch": 0.6230724828894203, + "grad_norm": 1.7341936826705933, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.7713912725448608, + "num_tokens": 117326593.0, + "step": 7556 + }, + { + "epoch": 0.6231549435144719, + "grad_norm": 2.126678943634033, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7432533502578735, + "num_tokens": 117351581.0, + "step": 7557 + }, + { + "epoch": 0.6232374041395233, + "grad_norm": 1.9639906883239746, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7493207454681396, + "num_tokens": 117381922.0, + "step": 7558 + }, + { + "epoch": 0.6233198647645749, + "grad_norm": 1.9919096231460571, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7305925488471985, + "num_tokens": 117410875.0, + "step": 7559 + }, + { + "epoch": 0.6234023253896265, + "grad_norm": 2.333547830581665, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7387274503707886, + "num_tokens": 117433401.0, + "step": 7560 + }, + { + "epoch": 0.6234847860146779, + "grad_norm": 2.386920213699341, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7301251888275146, + "num_tokens": 117456361.0, + "step": 7561 + }, + { + "epoch": 0.6235672466397295, + "grad_norm": 1.9463201761245728, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7300890684127808, + "num_tokens": 117491952.0, + "step": 7562 + }, + { + "epoch": 0.6236497072647811, + "grad_norm": 1.83407461643219, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.725653886795044, + "num_tokens": 117527237.0, + "step": 7563 + }, + { + "epoch": 0.6237321678898327, + "grad_norm": 1.9606550931930542, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7529670000076294, + "num_tokens": 117556391.0, + "step": 7564 + }, + { + "epoch": 0.6238146285148841, + "grad_norm": 2.2697861194610596, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7446004152297974, + "num_tokens": 117582482.0, + "step": 7565 + }, + { + "epoch": 0.6238970891399357, + "grad_norm": 1.804362177848816, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7754548192024231, + "num_tokens": 117614931.0, + "step": 7566 + }, + { + "epoch": 0.6239795497649873, + "grad_norm": 1.8954721689224243, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7529966831207275, + "num_tokens": 117644127.0, + "step": 7567 + }, + { + "epoch": 0.6240620103900387, + "grad_norm": 1.7567282915115356, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7709585428237915, + "num_tokens": 117677068.0, + "step": 7568 + }, + { + "epoch": 0.6241444710150903, + "grad_norm": 2.1021640300750732, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.756328284740448, + "num_tokens": 117702361.0, + "step": 7569 + }, + { + "epoch": 0.6242269316401419, + "grad_norm": 1.7967000007629395, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7704353332519531, + "num_tokens": 117739784.0, + "step": 7570 + }, + { + "epoch": 0.6243093922651933, + "grad_norm": 1.9704420566558838, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.744956374168396, + "num_tokens": 117768509.0, + "step": 7571 + }, + { + "epoch": 0.6243918528902449, + "grad_norm": 1.7255160808563232, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7743426561355591, + "num_tokens": 117805211.0, + "step": 7572 + }, + { + "epoch": 0.6244743135152965, + "grad_norm": 1.8492039442062378, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7169892191886902, + "num_tokens": 117840236.0, + "step": 7573 + }, + { + "epoch": 0.6245567741403479, + "grad_norm": 2.020171642303467, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7282133102416992, + "num_tokens": 117869630.0, + "step": 7574 + }, + { + "epoch": 0.6246392347653995, + "grad_norm": 2.0413408279418945, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7721908092498779, + "num_tokens": 117899918.0, + "step": 7575 + }, + { + "epoch": 0.6247216953904511, + "grad_norm": 1.8076419830322266, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7751457691192627, + "num_tokens": 117933800.0, + "step": 7576 + }, + { + "epoch": 0.6248041560155025, + "grad_norm": 2.12192440032959, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6950687170028687, + "num_tokens": 117961485.0, + "step": 7577 + }, + { + "epoch": 0.6248866166405541, + "grad_norm": 1.912993311882019, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.759575366973877, + "num_tokens": 117991800.0, + "step": 7578 + }, + { + "epoch": 0.6249690772656057, + "grad_norm": 2.007911205291748, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7643800973892212, + "num_tokens": 118019496.0, + "step": 7579 + }, + { + "epoch": 0.6250515378906573, + "grad_norm": 1.821007490158081, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.7944437265396118, + "num_tokens": 118051876.0, + "step": 7580 + }, + { + "epoch": 0.6251339985157087, + "grad_norm": 1.9373390674591064, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7235047221183777, + "num_tokens": 118084751.0, + "step": 7581 + }, + { + "epoch": 0.6252164591407603, + "grad_norm": 2.095984935760498, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7375461459159851, + "num_tokens": 118111346.0, + "step": 7582 + }, + { + "epoch": 0.6252989197658119, + "grad_norm": 1.856971263885498, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.7839078307151794, + "num_tokens": 118142732.0, + "step": 7583 + }, + { + "epoch": 0.6253813803908633, + "grad_norm": 1.9166672229766846, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7759270071983337, + "num_tokens": 118169096.0, + "step": 7584 + }, + { + "epoch": 0.6254638410159149, + "grad_norm": 1.9159786701202393, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.757788896560669, + "num_tokens": 118198893.0, + "step": 7585 + }, + { + "epoch": 0.6255463016409665, + "grad_norm": 2.05414080619812, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7680965662002563, + "num_tokens": 118224878.0, + "step": 7586 + }, + { + "epoch": 0.6256287622660179, + "grad_norm": 2.135556936264038, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.6973891854286194, + "num_tokens": 118253951.0, + "step": 7587 + }, + { + "epoch": 0.6257112228910695, + "grad_norm": 2.2037153244018555, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7335463762283325, + "num_tokens": 118278310.0, + "step": 7588 + }, + { + "epoch": 0.6257936835161211, + "grad_norm": 2.0961790084838867, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7575893402099609, + "num_tokens": 118304515.0, + "step": 7589 + }, + { + "epoch": 0.6258761441411725, + "grad_norm": 2.121683120727539, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7435483932495117, + "num_tokens": 118329086.0, + "step": 7590 + }, + { + "epoch": 0.6259586047662241, + "grad_norm": 1.875558614730835, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7610296010971069, + "num_tokens": 118361229.0, + "step": 7591 + }, + { + "epoch": 0.6260410653912757, + "grad_norm": 1.9245573282241821, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7360484004020691, + "num_tokens": 118395077.0, + "step": 7592 + }, + { + "epoch": 0.6261235260163273, + "grad_norm": 2.254227876663208, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7518578767776489, + "num_tokens": 118419348.0, + "step": 7593 + }, + { + "epoch": 0.6262059866413787, + "grad_norm": 1.904134750366211, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7476521730422974, + "num_tokens": 118449495.0, + "step": 7594 + }, + { + "epoch": 0.6262884472664303, + "grad_norm": 1.886773705482483, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.772467851638794, + "num_tokens": 118479196.0, + "step": 7595 + }, + { + "epoch": 0.6263709078914819, + "grad_norm": 1.9047337770462036, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7618252038955688, + "num_tokens": 118508538.0, + "step": 7596 + }, + { + "epoch": 0.6264533685165333, + "grad_norm": 2.062401533126831, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.755452573299408, + "num_tokens": 118535627.0, + "step": 7597 + }, + { + "epoch": 0.6265358291415849, + "grad_norm": 1.9519665241241455, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7581132650375366, + "num_tokens": 118566579.0, + "step": 7598 + }, + { + "epoch": 0.6266182897666365, + "grad_norm": 1.6457005739212036, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7643066644668579, + "num_tokens": 118607400.0, + "step": 7599 + }, + { + "epoch": 0.6267007503916879, + "grad_norm": 2.158968687057495, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7650256156921387, + "num_tokens": 118631384.0, + "step": 7600 + }, + { + "epoch": 0.6267832110167395, + "grad_norm": 1.991979956626892, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7362529635429382, + "num_tokens": 118662818.0, + "step": 7601 + }, + { + "epoch": 0.6268656716417911, + "grad_norm": 1.8695971965789795, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.765645444393158, + "num_tokens": 118694547.0, + "step": 7602 + }, + { + "epoch": 0.6269481322668425, + "grad_norm": 2.118856191635132, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7059856653213501, + "num_tokens": 118724431.0, + "step": 7603 + }, + { + "epoch": 0.6270305928918941, + "grad_norm": 1.8924617767333984, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7226365804672241, + "num_tokens": 118755054.0, + "step": 7604 + }, + { + "epoch": 0.6271130535169457, + "grad_norm": 1.6864066123962402, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7738416194915771, + "num_tokens": 118789499.0, + "step": 7605 + }, + { + "epoch": 0.6271955141419971, + "grad_norm": 2.1157045364379883, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7726023197174072, + "num_tokens": 118813497.0, + "step": 7606 + }, + { + "epoch": 0.6272779747670487, + "grad_norm": 1.7167807817459106, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.755818784236908, + "num_tokens": 118849123.0, + "step": 7607 + }, + { + "epoch": 0.6273604353921003, + "grad_norm": 2.0322327613830566, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7271384000778198, + "num_tokens": 118878774.0, + "step": 7608 + }, + { + "epoch": 0.6274428960171519, + "grad_norm": 1.8384883403778076, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7517814040184021, + "num_tokens": 118909726.0, + "step": 7609 + }, + { + "epoch": 0.6275253566422033, + "grad_norm": 2.0187580585479736, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.7577357888221741, + "num_tokens": 118936353.0, + "step": 7610 + }, + { + "epoch": 0.6276078172672549, + "grad_norm": 1.913234829902649, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7571446895599365, + "num_tokens": 118968078.0, + "step": 7611 + }, + { + "epoch": 0.6276902778923065, + "grad_norm": 2.124110460281372, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7358324527740479, + "num_tokens": 118993962.0, + "step": 7612 + }, + { + "epoch": 0.6277727385173579, + "grad_norm": 1.9247125387191772, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.756891667842865, + "num_tokens": 119025456.0, + "step": 7613 + }, + { + "epoch": 0.6278551991424095, + "grad_norm": 2.131979465484619, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7265959978103638, + "num_tokens": 119051819.0, + "step": 7614 + }, + { + "epoch": 0.6279376597674611, + "grad_norm": 2.0510454177856445, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7565211653709412, + "num_tokens": 119078257.0, + "step": 7615 + }, + { + "epoch": 0.6280201203925125, + "grad_norm": 1.7993872165679932, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.7937266230583191, + "num_tokens": 119109968.0, + "step": 7616 + }, + { + "epoch": 0.6281025810175641, + "grad_norm": 2.0848193168640137, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7799400091171265, + "num_tokens": 119135221.0, + "step": 7617 + }, + { + "epoch": 0.6281850416426157, + "grad_norm": 1.9309595823287964, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7055485248565674, + "num_tokens": 119170797.0, + "step": 7618 + }, + { + "epoch": 0.6282675022676671, + "grad_norm": 2.116072654724121, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.718711256980896, + "num_tokens": 119198590.0, + "step": 7619 + }, + { + "epoch": 0.6283499628927187, + "grad_norm": 1.880910038948059, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7554590702056885, + "num_tokens": 119231728.0, + "step": 7620 + }, + { + "epoch": 0.6284324235177703, + "grad_norm": 2.0870115756988525, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7522237300872803, + "num_tokens": 119259172.0, + "step": 7621 + }, + { + "epoch": 0.6285148841428218, + "grad_norm": 2.147193193435669, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7301851511001587, + "num_tokens": 119285717.0, + "step": 7622 + }, + { + "epoch": 0.6285973447678733, + "grad_norm": 2.5421969890594482, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7271245718002319, + "num_tokens": 119312031.0, + "step": 7623 + }, + { + "epoch": 0.6286798053929249, + "grad_norm": 1.858345866203308, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7376265525817871, + "num_tokens": 119345980.0, + "step": 7624 + }, + { + "epoch": 0.6287622660179765, + "grad_norm": 2.152095317840576, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7621324062347412, + "num_tokens": 119370111.0, + "step": 7625 + }, + { + "epoch": 0.6288447266430279, + "grad_norm": 1.7558434009552002, + "learning_rate": 1e-06, + "loss": 0.5473, + "mean_token_accuracy": 0.8184086084365845, + "num_tokens": 119404099.0, + "step": 7626 + }, + { + "epoch": 0.6289271872680795, + "grad_norm": 1.9826442003250122, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7530286312103271, + "num_tokens": 119432774.0, + "step": 7627 + }, + { + "epoch": 0.6290096478931311, + "grad_norm": 1.7202755212783813, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7647935748100281, + "num_tokens": 119466897.0, + "step": 7628 + }, + { + "epoch": 0.6290921085181825, + "grad_norm": 1.8557319641113281, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7326934933662415, + "num_tokens": 119500637.0, + "step": 7629 + }, + { + "epoch": 0.6291745691432341, + "grad_norm": 2.1757256984710693, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7354357242584229, + "num_tokens": 119525601.0, + "step": 7630 + }, + { + "epoch": 0.6292570297682857, + "grad_norm": 2.1409671306610107, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7219016551971436, + "num_tokens": 119551660.0, + "step": 7631 + }, + { + "epoch": 0.6293394903933371, + "grad_norm": 2.1248416900634766, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7501958012580872, + "num_tokens": 119579383.0, + "step": 7632 + }, + { + "epoch": 0.6294219510183887, + "grad_norm": 1.6967079639434814, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7265249490737915, + "num_tokens": 119620370.0, + "step": 7633 + }, + { + "epoch": 0.6295044116434403, + "grad_norm": 1.9075429439544678, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.7907438278198242, + "num_tokens": 119649508.0, + "step": 7634 + }, + { + "epoch": 0.6295868722684917, + "grad_norm": 2.0467875003814697, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7538623809814453, + "num_tokens": 119675815.0, + "step": 7635 + }, + { + "epoch": 0.6296693328935433, + "grad_norm": 2.1407318115234375, + "learning_rate": 1e-06, + "loss": 0.7724, + "mean_token_accuracy": 0.7523244619369507, + "num_tokens": 119702822.0, + "step": 7636 + }, + { + "epoch": 0.6297517935185949, + "grad_norm": 1.7819987535476685, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7548565864562988, + "num_tokens": 119738554.0, + "step": 7637 + }, + { + "epoch": 0.6298342541436464, + "grad_norm": 2.041724920272827, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7138296365737915, + "num_tokens": 119764690.0, + "step": 7638 + }, + { + "epoch": 0.6299167147686979, + "grad_norm": 1.9197200536727905, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7750122547149658, + "num_tokens": 119796782.0, + "step": 7639 + }, + { + "epoch": 0.6299991753937495, + "grad_norm": 1.8367997407913208, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.7475275993347168, + "num_tokens": 119826773.0, + "step": 7640 + }, + { + "epoch": 0.6300816360188011, + "grad_norm": 2.14275860786438, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7656145095825195, + "num_tokens": 119850167.0, + "step": 7641 + }, + { + "epoch": 0.6301640966438525, + "grad_norm": 1.7415509223937988, + "learning_rate": 1e-06, + "loss": 0.7704, + "mean_token_accuracy": 0.761428713798523, + "num_tokens": 119886517.0, + "step": 7642 + }, + { + "epoch": 0.6302465572689041, + "grad_norm": 1.9132846593856812, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7607561945915222, + "num_tokens": 119914066.0, + "step": 7643 + }, + { + "epoch": 0.6303290178939557, + "grad_norm": 2.202958583831787, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7448524236679077, + "num_tokens": 119938200.0, + "step": 7644 + }, + { + "epoch": 0.6304114785190071, + "grad_norm": 2.0166678428649902, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7571985721588135, + "num_tokens": 119965750.0, + "step": 7645 + }, + { + "epoch": 0.6304939391440587, + "grad_norm": 2.1951589584350586, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7007098197937012, + "num_tokens": 119990254.0, + "step": 7646 + }, + { + "epoch": 0.6305763997691103, + "grad_norm": 2.1774678230285645, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7169808745384216, + "num_tokens": 120015488.0, + "step": 7647 + }, + { + "epoch": 0.6306588603941617, + "grad_norm": 1.6040970087051392, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7678655982017517, + "num_tokens": 120051413.0, + "step": 7648 + }, + { + "epoch": 0.6307413210192133, + "grad_norm": 1.8955150842666626, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7299807071685791, + "num_tokens": 120082939.0, + "step": 7649 + }, + { + "epoch": 0.6308237816442649, + "grad_norm": 1.9848076105117798, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7361128330230713, + "num_tokens": 120113962.0, + "step": 7650 + }, + { + "epoch": 0.6309062422693164, + "grad_norm": 2.0134494304656982, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7446353435516357, + "num_tokens": 120139092.0, + "step": 7651 + }, + { + "epoch": 0.6309887028943679, + "grad_norm": 1.9303991794586182, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7471030950546265, + "num_tokens": 120167208.0, + "step": 7652 + }, + { + "epoch": 0.6310711635194195, + "grad_norm": 1.7848526239395142, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7309563755989075, + "num_tokens": 120204049.0, + "step": 7653 + }, + { + "epoch": 0.6311536241444711, + "grad_norm": 1.9064000844955444, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7197747230529785, + "num_tokens": 120234698.0, + "step": 7654 + }, + { + "epoch": 0.6312360847695225, + "grad_norm": 2.0304460525512695, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.706032395362854, + "num_tokens": 120266420.0, + "step": 7655 + }, + { + "epoch": 0.6313185453945741, + "grad_norm": 1.910481572151184, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7685560584068298, + "num_tokens": 120293690.0, + "step": 7656 + }, + { + "epoch": 0.6314010060196257, + "grad_norm": 1.9168078899383545, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7336434125900269, + "num_tokens": 120323472.0, + "step": 7657 + }, + { + "epoch": 0.6314834666446771, + "grad_norm": 1.822344183921814, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7793420553207397, + "num_tokens": 120353888.0, + "step": 7658 + }, + { + "epoch": 0.6315659272697287, + "grad_norm": 1.8517415523529053, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7413365840911865, + "num_tokens": 120389274.0, + "step": 7659 + }, + { + "epoch": 0.6316483878947803, + "grad_norm": 2.4265503883361816, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7253807187080383, + "num_tokens": 120417351.0, + "step": 7660 + }, + { + "epoch": 0.6317308485198317, + "grad_norm": 1.9181640148162842, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7670773863792419, + "num_tokens": 120445170.0, + "step": 7661 + }, + { + "epoch": 0.6318133091448833, + "grad_norm": 1.9569658041000366, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.7737034559249878, + "num_tokens": 120471448.0, + "step": 7662 + }, + { + "epoch": 0.6318957697699349, + "grad_norm": 1.8746494054794312, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7222071290016174, + "num_tokens": 120503972.0, + "step": 7663 + }, + { + "epoch": 0.6319782303949864, + "grad_norm": 2.057324171066284, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7438578605651855, + "num_tokens": 120533185.0, + "step": 7664 + }, + { + "epoch": 0.6320606910200379, + "grad_norm": 2.074230909347534, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7707485556602478, + "num_tokens": 120558240.0, + "step": 7665 + }, + { + "epoch": 0.6321431516450895, + "grad_norm": 1.833315134048462, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7248208522796631, + "num_tokens": 120593138.0, + "step": 7666 + }, + { + "epoch": 0.632225612270141, + "grad_norm": 1.7965013980865479, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7739168405532837, + "num_tokens": 120625552.0, + "step": 7667 + }, + { + "epoch": 0.6323080728951925, + "grad_norm": 2.0914201736450195, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7374664545059204, + "num_tokens": 120652779.0, + "step": 7668 + }, + { + "epoch": 0.6323905335202441, + "grad_norm": 2.0308101177215576, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6958557367324829, + "num_tokens": 120684129.0, + "step": 7669 + }, + { + "epoch": 0.6324729941452957, + "grad_norm": 1.5640226602554321, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7373995184898376, + "num_tokens": 120730979.0, + "step": 7670 + }, + { + "epoch": 0.6325554547703471, + "grad_norm": 2.0260095596313477, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7252218127250671, + "num_tokens": 120760520.0, + "step": 7671 + }, + { + "epoch": 0.6326379153953987, + "grad_norm": 2.070441961288452, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7285721302032471, + "num_tokens": 120788191.0, + "step": 7672 + }, + { + "epoch": 0.6327203760204503, + "grad_norm": 2.078652858734131, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.733810305595398, + "num_tokens": 120815668.0, + "step": 7673 + }, + { + "epoch": 0.6328028366455017, + "grad_norm": 1.6493173837661743, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.7956104874610901, + "num_tokens": 120849383.0, + "step": 7674 + }, + { + "epoch": 0.6328852972705533, + "grad_norm": 1.9721238613128662, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.720011830329895, + "num_tokens": 120880562.0, + "step": 7675 + }, + { + "epoch": 0.6329677578956049, + "grad_norm": 1.919321894645691, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7404962778091431, + "num_tokens": 120912618.0, + "step": 7676 + }, + { + "epoch": 0.6330502185206563, + "grad_norm": 2.080167531967163, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7349241971969604, + "num_tokens": 120940373.0, + "step": 7677 + }, + { + "epoch": 0.6331326791457079, + "grad_norm": 1.991950511932373, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7509099245071411, + "num_tokens": 120968634.0, + "step": 7678 + }, + { + "epoch": 0.6332151397707595, + "grad_norm": 2.1172773838043213, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7434170842170715, + "num_tokens": 120993360.0, + "step": 7679 + }, + { + "epoch": 0.633297600395811, + "grad_norm": 1.8774666786193848, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7689584493637085, + "num_tokens": 121025483.0, + "step": 7680 + }, + { + "epoch": 0.6333800610208625, + "grad_norm": 1.872544527053833, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7655866146087646, + "num_tokens": 121054675.0, + "step": 7681 + }, + { + "epoch": 0.6334625216459141, + "grad_norm": 1.968454360961914, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6972069144248962, + "num_tokens": 121087180.0, + "step": 7682 + }, + { + "epoch": 0.6335449822709656, + "grad_norm": 1.6292909383773804, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.7923187017440796, + "num_tokens": 121126005.0, + "step": 7683 + }, + { + "epoch": 0.6336274428960171, + "grad_norm": 2.0622620582580566, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7248423099517822, + "num_tokens": 121152830.0, + "step": 7684 + }, + { + "epoch": 0.6337099035210687, + "grad_norm": 2.163161039352417, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7645883560180664, + "num_tokens": 121175307.0, + "step": 7685 + }, + { + "epoch": 0.6337923641461203, + "grad_norm": 2.1256680488586426, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7473155856132507, + "num_tokens": 121199895.0, + "step": 7686 + }, + { + "epoch": 0.6338748247711717, + "grad_norm": 1.6649202108383179, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7854644656181335, + "num_tokens": 121233553.0, + "step": 7687 + }, + { + "epoch": 0.6339572853962233, + "grad_norm": 2.0973222255706787, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7256356477737427, + "num_tokens": 121261867.0, + "step": 7688 + }, + { + "epoch": 0.6340397460212749, + "grad_norm": 1.7847968339920044, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7343587875366211, + "num_tokens": 121298739.0, + "step": 7689 + }, + { + "epoch": 0.6341222066463263, + "grad_norm": 2.1343119144439697, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7387630939483643, + "num_tokens": 121326382.0, + "step": 7690 + }, + { + "epoch": 0.6342046672713779, + "grad_norm": 2.033630609512329, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7785971164703369, + "num_tokens": 121351613.0, + "step": 7691 + }, + { + "epoch": 0.6342871278964295, + "grad_norm": 1.8473849296569824, + "learning_rate": 1e-06, + "loss": 0.5599, + "mean_token_accuracy": 0.8136721849441528, + "num_tokens": 121380483.0, + "step": 7692 + }, + { + "epoch": 0.634369588521481, + "grad_norm": 2.0132577419281006, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7639978528022766, + "num_tokens": 121406822.0, + "step": 7693 + }, + { + "epoch": 0.6344520491465325, + "grad_norm": 1.9907567501068115, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7217005491256714, + "num_tokens": 121434830.0, + "step": 7694 + }, + { + "epoch": 0.6345345097715841, + "grad_norm": 2.088150978088379, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7719262838363647, + "num_tokens": 121462673.0, + "step": 7695 + }, + { + "epoch": 0.6346169703966356, + "grad_norm": 1.9866935014724731, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7355596423149109, + "num_tokens": 121491337.0, + "step": 7696 + }, + { + "epoch": 0.6346994310216871, + "grad_norm": 2.092634677886963, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7385138273239136, + "num_tokens": 121523138.0, + "step": 7697 + }, + { + "epoch": 0.6347818916467387, + "grad_norm": 2.01579213142395, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7348073720932007, + "num_tokens": 121554809.0, + "step": 7698 + }, + { + "epoch": 0.6348643522717902, + "grad_norm": 2.2210042476654053, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7170097827911377, + "num_tokens": 121578497.0, + "step": 7699 + }, + { + "epoch": 0.6349468128968417, + "grad_norm": 2.1036720275878906, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7245060205459595, + "num_tokens": 121602599.0, + "step": 7700 + }, + { + "epoch": 0.6350292735218933, + "grad_norm": 1.9829556941986084, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7951831817626953, + "num_tokens": 121632172.0, + "step": 7701 + }, + { + "epoch": 0.6351117341469449, + "grad_norm": 1.8620645999908447, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7713160514831543, + "num_tokens": 121661234.0, + "step": 7702 + }, + { + "epoch": 0.6351941947719963, + "grad_norm": 2.064765691757202, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7542887926101685, + "num_tokens": 121687405.0, + "step": 7703 + }, + { + "epoch": 0.6352766553970479, + "grad_norm": 1.7807731628417969, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7710661292076111, + "num_tokens": 121722105.0, + "step": 7704 + }, + { + "epoch": 0.6353591160220995, + "grad_norm": 2.097188949584961, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7531085014343262, + "num_tokens": 121749667.0, + "step": 7705 + }, + { + "epoch": 0.635441576647151, + "grad_norm": 2.106325626373291, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.751611590385437, + "num_tokens": 121776417.0, + "step": 7706 + }, + { + "epoch": 0.6355240372722025, + "grad_norm": 1.9710148572921753, + "learning_rate": 1e-06, + "loss": 0.5836, + "mean_token_accuracy": 0.7978637218475342, + "num_tokens": 121801304.0, + "step": 7707 + }, + { + "epoch": 0.6356064978972541, + "grad_norm": 1.8865759372711182, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7611774206161499, + "num_tokens": 121833590.0, + "step": 7708 + }, + { + "epoch": 0.6356889585223056, + "grad_norm": 2.125762701034546, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7587820291519165, + "num_tokens": 121859607.0, + "step": 7709 + }, + { + "epoch": 0.6357714191473571, + "grad_norm": 1.7916282415390015, + "learning_rate": 1e-06, + "loss": 0.7642, + "mean_token_accuracy": 0.7660722732543945, + "num_tokens": 121893897.0, + "step": 7710 + }, + { + "epoch": 0.6358538797724087, + "grad_norm": 2.167736053466797, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7660872936248779, + "num_tokens": 121918244.0, + "step": 7711 + }, + { + "epoch": 0.6359363403974602, + "grad_norm": 1.7806804180145264, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7564544677734375, + "num_tokens": 121950896.0, + "step": 7712 + }, + { + "epoch": 0.6360188010225117, + "grad_norm": 1.798951506614685, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.7817943096160889, + "num_tokens": 121982457.0, + "step": 7713 + }, + { + "epoch": 0.6361012616475633, + "grad_norm": 2.1240100860595703, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7207438945770264, + "num_tokens": 122011098.0, + "step": 7714 + }, + { + "epoch": 0.6361837222726148, + "grad_norm": 1.8761229515075684, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7168138027191162, + "num_tokens": 122048016.0, + "step": 7715 + }, + { + "epoch": 0.6362661828976663, + "grad_norm": 1.9498335123062134, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.743443489074707, + "num_tokens": 122078190.0, + "step": 7716 + }, + { + "epoch": 0.6363486435227179, + "grad_norm": 1.9543683528900146, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7538847327232361, + "num_tokens": 122105020.0, + "step": 7717 + }, + { + "epoch": 0.6364311041477695, + "grad_norm": 2.0445923805236816, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7549288272857666, + "num_tokens": 122132714.0, + "step": 7718 + }, + { + "epoch": 0.636513564772821, + "grad_norm": 2.059689521789551, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7385146021842957, + "num_tokens": 122161012.0, + "step": 7719 + }, + { + "epoch": 0.6365960253978725, + "grad_norm": 1.8462313413619995, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7333734035491943, + "num_tokens": 122192348.0, + "step": 7720 + }, + { + "epoch": 0.6366784860229241, + "grad_norm": 2.010780096054077, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7291592359542847, + "num_tokens": 122221874.0, + "step": 7721 + }, + { + "epoch": 0.6367609466479756, + "grad_norm": 1.7667139768600464, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7542919516563416, + "num_tokens": 122255256.0, + "step": 7722 + }, + { + "epoch": 0.6368434072730271, + "grad_norm": 1.9116013050079346, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.7556127309799194, + "num_tokens": 122287521.0, + "step": 7723 + }, + { + "epoch": 0.6369258678980787, + "grad_norm": 1.743723750114441, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7419800162315369, + "num_tokens": 122322525.0, + "step": 7724 + }, + { + "epoch": 0.6370083285231302, + "grad_norm": 2.022313356399536, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7419769763946533, + "num_tokens": 122349655.0, + "step": 7725 + }, + { + "epoch": 0.6370907891481817, + "grad_norm": 1.8282850980758667, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7737158536911011, + "num_tokens": 122383494.0, + "step": 7726 + }, + { + "epoch": 0.6371732497732333, + "grad_norm": 1.9949615001678467, + "learning_rate": 1e-06, + "loss": 0.7954, + "mean_token_accuracy": 0.7518402934074402, + "num_tokens": 122413826.0, + "step": 7727 + }, + { + "epoch": 0.6372557103982848, + "grad_norm": 1.9871906042099, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7050803899765015, + "num_tokens": 122443248.0, + "step": 7728 + }, + { + "epoch": 0.6373381710233363, + "grad_norm": 1.8606377840042114, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7472548484802246, + "num_tokens": 122474123.0, + "step": 7729 + }, + { + "epoch": 0.6374206316483879, + "grad_norm": 1.8143351078033447, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7573544979095459, + "num_tokens": 122507915.0, + "step": 7730 + }, + { + "epoch": 0.6375030922734395, + "grad_norm": 2.1126701831817627, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7606790661811829, + "num_tokens": 122534346.0, + "step": 7731 + }, + { + "epoch": 0.637585552898491, + "grad_norm": 1.9380871057510376, + "learning_rate": 1e-06, + "loss": 0.5963, + "mean_token_accuracy": 0.8091815710067749, + "num_tokens": 122561873.0, + "step": 7732 + }, + { + "epoch": 0.6376680135235425, + "grad_norm": 2.187171697616577, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7036272287368774, + "num_tokens": 122589978.0, + "step": 7733 + }, + { + "epoch": 0.6377504741485941, + "grad_norm": 1.837262511253357, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7290883660316467, + "num_tokens": 122623230.0, + "step": 7734 + }, + { + "epoch": 0.6378329347736456, + "grad_norm": 1.8925946950912476, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7413598895072937, + "num_tokens": 122659517.0, + "step": 7735 + }, + { + "epoch": 0.6379153953986971, + "grad_norm": 2.2283689975738525, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7375199794769287, + "num_tokens": 122684670.0, + "step": 7736 + }, + { + "epoch": 0.6379978560237487, + "grad_norm": 1.986401915550232, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7426379919052124, + "num_tokens": 122716108.0, + "step": 7737 + }, + { + "epoch": 0.6380803166488002, + "grad_norm": 1.787266492843628, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7290971875190735, + "num_tokens": 122750336.0, + "step": 7738 + }, + { + "epoch": 0.6381627772738517, + "grad_norm": 1.586795449256897, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7452179193496704, + "num_tokens": 122793428.0, + "step": 7739 + }, + { + "epoch": 0.6382452378989033, + "grad_norm": 1.8352835178375244, + "learning_rate": 1e-06, + "loss": 0.6062, + "mean_token_accuracy": 0.8056423664093018, + "num_tokens": 122823756.0, + "step": 7740 + }, + { + "epoch": 0.6383276985239548, + "grad_norm": 1.856265664100647, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7324204444885254, + "num_tokens": 122858589.0, + "step": 7741 + }, + { + "epoch": 0.6384101591490063, + "grad_norm": 1.7985541820526123, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7620830535888672, + "num_tokens": 122894660.0, + "step": 7742 + }, + { + "epoch": 0.6384926197740579, + "grad_norm": 2.0194079875946045, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7513593435287476, + "num_tokens": 122922642.0, + "step": 7743 + }, + { + "epoch": 0.6385750803991094, + "grad_norm": 2.3713111877441406, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.768837571144104, + "num_tokens": 122945603.0, + "step": 7744 + }, + { + "epoch": 0.638657541024161, + "grad_norm": 1.9309382438659668, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7703870534896851, + "num_tokens": 122975114.0, + "step": 7745 + }, + { + "epoch": 0.6387400016492125, + "grad_norm": 1.7778840065002441, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.772154688835144, + "num_tokens": 123009148.0, + "step": 7746 + }, + { + "epoch": 0.6388224622742641, + "grad_norm": 1.9612140655517578, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7346570491790771, + "num_tokens": 123038418.0, + "step": 7747 + }, + { + "epoch": 0.6389049228993156, + "grad_norm": 1.9342056512832642, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7533490061759949, + "num_tokens": 123067362.0, + "step": 7748 + }, + { + "epoch": 0.6389873835243671, + "grad_norm": 2.0316531658172607, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7321403622627258, + "num_tokens": 123093071.0, + "step": 7749 + }, + { + "epoch": 0.6390698441494187, + "grad_norm": 1.960471510887146, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7654285430908203, + "num_tokens": 123122898.0, + "step": 7750 + }, + { + "epoch": 0.6391523047744702, + "grad_norm": 2.315225124359131, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7291339635848999, + "num_tokens": 123147711.0, + "step": 7751 + }, + { + "epoch": 0.6392347653995217, + "grad_norm": 2.2407288551330566, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7313612103462219, + "num_tokens": 123175124.0, + "step": 7752 + }, + { + "epoch": 0.6393172260245733, + "grad_norm": 2.0868000984191895, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7162308096885681, + "num_tokens": 123202423.0, + "step": 7753 + }, + { + "epoch": 0.6393996866496248, + "grad_norm": 2.0643293857574463, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7722886800765991, + "num_tokens": 123227572.0, + "step": 7754 + }, + { + "epoch": 0.6394821472746763, + "grad_norm": 1.9340665340423584, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7746496200561523, + "num_tokens": 123255635.0, + "step": 7755 + }, + { + "epoch": 0.6395646078997279, + "grad_norm": 1.8017302751541138, + "learning_rate": 1e-06, + "loss": 0.605, + "mean_token_accuracy": 0.8000836372375488, + "num_tokens": 123287808.0, + "step": 7756 + }, + { + "epoch": 0.6396470685247794, + "grad_norm": 2.0163655281066895, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7047001123428345, + "num_tokens": 123317831.0, + "step": 7757 + }, + { + "epoch": 0.6397295291498309, + "grad_norm": 1.848358154296875, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.761461615562439, + "num_tokens": 123351874.0, + "step": 7758 + }, + { + "epoch": 0.6398119897748825, + "grad_norm": 1.8600139617919922, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.7618616819381714, + "num_tokens": 123381973.0, + "step": 7759 + }, + { + "epoch": 0.639894450399934, + "grad_norm": 2.118685722351074, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7415204048156738, + "num_tokens": 123407628.0, + "step": 7760 + }, + { + "epoch": 0.6399769110249856, + "grad_norm": 1.8040499687194824, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.764528214931488, + "num_tokens": 123441402.0, + "step": 7761 + }, + { + "epoch": 0.6400593716500371, + "grad_norm": 1.9940966367721558, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7467421293258667, + "num_tokens": 123469548.0, + "step": 7762 + }, + { + "epoch": 0.6401418322750887, + "grad_norm": 1.8248144388198853, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.772542417049408, + "num_tokens": 123503075.0, + "step": 7763 + }, + { + "epoch": 0.6402242929001402, + "grad_norm": 1.8754734992980957, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7305919528007507, + "num_tokens": 123539276.0, + "step": 7764 + }, + { + "epoch": 0.6403067535251917, + "grad_norm": 1.8871451616287231, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7395284175872803, + "num_tokens": 123572589.0, + "step": 7765 + }, + { + "epoch": 0.6403892141502433, + "grad_norm": 1.9272222518920898, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7600932121276855, + "num_tokens": 123602561.0, + "step": 7766 + }, + { + "epoch": 0.6404716747752948, + "grad_norm": 1.9142155647277832, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7927325963973999, + "num_tokens": 123631887.0, + "step": 7767 + }, + { + "epoch": 0.6405541354003463, + "grad_norm": 2.024844169616699, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7346482276916504, + "num_tokens": 123659467.0, + "step": 7768 + }, + { + "epoch": 0.6406365960253979, + "grad_norm": 1.9363956451416016, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7415072321891785, + "num_tokens": 123691967.0, + "step": 7769 + }, + { + "epoch": 0.6407190566504494, + "grad_norm": 1.9992091655731201, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7634114027023315, + "num_tokens": 123719622.0, + "step": 7770 + }, + { + "epoch": 0.6408015172755009, + "grad_norm": 1.9057886600494385, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7765834927558899, + "num_tokens": 123748239.0, + "step": 7771 + }, + { + "epoch": 0.6408839779005525, + "grad_norm": 1.8002392053604126, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7306594848632812, + "num_tokens": 123784743.0, + "step": 7772 + }, + { + "epoch": 0.640966438525604, + "grad_norm": 1.8630467653274536, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.7864905595779419, + "num_tokens": 123814553.0, + "step": 7773 + }, + { + "epoch": 0.6410488991506555, + "grad_norm": 2.1223039627075195, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7510075569152832, + "num_tokens": 123840840.0, + "step": 7774 + }, + { + "epoch": 0.6411313597757071, + "grad_norm": 2.2193233966827393, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7384222149848938, + "num_tokens": 123864770.0, + "step": 7775 + }, + { + "epoch": 0.6412138204007586, + "grad_norm": 2.151590347290039, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7102412581443787, + "num_tokens": 123892757.0, + "step": 7776 + }, + { + "epoch": 0.6412962810258102, + "grad_norm": 2.111030101776123, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7358575463294983, + "num_tokens": 123919544.0, + "step": 7777 + }, + { + "epoch": 0.6413787416508617, + "grad_norm": 2.053415298461914, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7167412042617798, + "num_tokens": 123947913.0, + "step": 7778 + }, + { + "epoch": 0.6414612022759133, + "grad_norm": 1.9611845016479492, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7154546976089478, + "num_tokens": 123979633.0, + "step": 7779 + }, + { + "epoch": 0.6415436629009648, + "grad_norm": 2.2819747924804688, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7461612224578857, + "num_tokens": 124003642.0, + "step": 7780 + }, + { + "epoch": 0.6416261235260163, + "grad_norm": 1.7644641399383545, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.790007472038269, + "num_tokens": 124036881.0, + "step": 7781 + }, + { + "epoch": 0.6417085841510679, + "grad_norm": 1.767972469329834, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7687063813209534, + "num_tokens": 124071165.0, + "step": 7782 + }, + { + "epoch": 0.6417910447761194, + "grad_norm": 1.8722721338272095, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.752741277217865, + "num_tokens": 124102217.0, + "step": 7783 + }, + { + "epoch": 0.6418735054011709, + "grad_norm": 2.02405047416687, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7476937174797058, + "num_tokens": 124134121.0, + "step": 7784 + }, + { + "epoch": 0.6419559660262225, + "grad_norm": 1.892659068107605, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7357783913612366, + "num_tokens": 124165697.0, + "step": 7785 + }, + { + "epoch": 0.642038426651274, + "grad_norm": 1.8775811195373535, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7616767883300781, + "num_tokens": 124197540.0, + "step": 7786 + }, + { + "epoch": 0.6421208872763255, + "grad_norm": 2.23256516456604, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7735711336135864, + "num_tokens": 124221818.0, + "step": 7787 + }, + { + "epoch": 0.6422033479013771, + "grad_norm": 1.9016026258468628, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7402558922767639, + "num_tokens": 124253834.0, + "step": 7788 + }, + { + "epoch": 0.6422858085264286, + "grad_norm": 1.7870392799377441, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7726990580558777, + "num_tokens": 124286974.0, + "step": 7789 + }, + { + "epoch": 0.6423682691514802, + "grad_norm": 2.405316114425659, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7452293634414673, + "num_tokens": 124308226.0, + "step": 7790 + }, + { + "epoch": 0.6424507297765317, + "grad_norm": 1.7896898984909058, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7661335468292236, + "num_tokens": 124341682.0, + "step": 7791 + }, + { + "epoch": 0.6425331904015833, + "grad_norm": 2.0429742336273193, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7699484825134277, + "num_tokens": 124369870.0, + "step": 7792 + }, + { + "epoch": 0.6426156510266348, + "grad_norm": 2.113271713256836, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7481991052627563, + "num_tokens": 124396148.0, + "step": 7793 + }, + { + "epoch": 0.6426981116516863, + "grad_norm": 1.949539065361023, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7350074052810669, + "num_tokens": 124426562.0, + "step": 7794 + }, + { + "epoch": 0.6427805722767379, + "grad_norm": 1.842450737953186, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.7494592070579529, + "num_tokens": 124459034.0, + "step": 7795 + }, + { + "epoch": 0.6428630329017894, + "grad_norm": 2.1709818840026855, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.729011595249176, + "num_tokens": 124485333.0, + "step": 7796 + }, + { + "epoch": 0.6429454935268409, + "grad_norm": 1.9404222965240479, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7371227741241455, + "num_tokens": 124516249.0, + "step": 7797 + }, + { + "epoch": 0.6430279541518925, + "grad_norm": 1.910199761390686, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7462955713272095, + "num_tokens": 124544249.0, + "step": 7798 + }, + { + "epoch": 0.643110414776944, + "grad_norm": 1.9021944999694824, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7110550403594971, + "num_tokens": 124578891.0, + "step": 7799 + }, + { + "epoch": 0.6431928754019955, + "grad_norm": 1.7352017164230347, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7763818502426147, + "num_tokens": 124613632.0, + "step": 7800 + }, + { + "epoch": 0.6432753360270471, + "grad_norm": 2.082324504852295, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7414023876190186, + "num_tokens": 124640044.0, + "step": 7801 + }, + { + "epoch": 0.6433577966520986, + "grad_norm": 1.9758678674697876, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7650942206382751, + "num_tokens": 124668135.0, + "step": 7802 + }, + { + "epoch": 0.6434402572771502, + "grad_norm": 2.0755715370178223, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.7088491320610046, + "num_tokens": 124694179.0, + "step": 7803 + }, + { + "epoch": 0.6435227179022017, + "grad_norm": 1.697258472442627, + "learning_rate": 1e-06, + "loss": 0.5923, + "mean_token_accuracy": 0.8022347688674927, + "num_tokens": 124729113.0, + "step": 7804 + }, + { + "epoch": 0.6436051785272532, + "grad_norm": 1.8741075992584229, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.7020349502563477, + "num_tokens": 124764164.0, + "step": 7805 + }, + { + "epoch": 0.6436876391523048, + "grad_norm": 1.9465166330337524, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7213329672813416, + "num_tokens": 124793901.0, + "step": 7806 + }, + { + "epoch": 0.6437700997773563, + "grad_norm": 2.0855653285980225, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7320659756660461, + "num_tokens": 124821170.0, + "step": 7807 + }, + { + "epoch": 0.6438525604024079, + "grad_norm": 1.8159688711166382, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7737717628479004, + "num_tokens": 124853888.0, + "step": 7808 + }, + { + "epoch": 0.6439350210274594, + "grad_norm": 1.8328129053115845, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7527056336402893, + "num_tokens": 124888309.0, + "step": 7809 + }, + { + "epoch": 0.6440174816525109, + "grad_norm": 1.837493896484375, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7597341537475586, + "num_tokens": 124922712.0, + "step": 7810 + }, + { + "epoch": 0.6440999422775625, + "grad_norm": 1.8099550008773804, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7636547088623047, + "num_tokens": 124958439.0, + "step": 7811 + }, + { + "epoch": 0.644182402902614, + "grad_norm": 1.844834566116333, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7226691246032715, + "num_tokens": 124990903.0, + "step": 7812 + }, + { + "epoch": 0.6442648635276655, + "grad_norm": 1.8716907501220703, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.770795464515686, + "num_tokens": 125021259.0, + "step": 7813 + }, + { + "epoch": 0.6443473241527171, + "grad_norm": 1.8496023416519165, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7714952230453491, + "num_tokens": 125051485.0, + "step": 7814 + }, + { + "epoch": 0.6444297847777686, + "grad_norm": 1.861222267150879, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7427397966384888, + "num_tokens": 125083719.0, + "step": 7815 + }, + { + "epoch": 0.6445122454028201, + "grad_norm": 1.8691047430038452, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7249152660369873, + "num_tokens": 125118677.0, + "step": 7816 + }, + { + "epoch": 0.6445947060278717, + "grad_norm": 2.009967565536499, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7286525368690491, + "num_tokens": 125145087.0, + "step": 7817 + }, + { + "epoch": 0.6446771666529232, + "grad_norm": 2.025006055831909, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6944149136543274, + "num_tokens": 125174332.0, + "step": 7818 + }, + { + "epoch": 0.6447596272779748, + "grad_norm": 2.06461238861084, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7548313736915588, + "num_tokens": 125201550.0, + "step": 7819 + }, + { + "epoch": 0.6448420879030263, + "grad_norm": 1.7994616031646729, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7603799104690552, + "num_tokens": 125233664.0, + "step": 7820 + }, + { + "epoch": 0.6449245485280778, + "grad_norm": 1.7176796197891235, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.7959772348403931, + "num_tokens": 125266140.0, + "step": 7821 + }, + { + "epoch": 0.6450070091531294, + "grad_norm": 1.9329839944839478, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7034985423088074, + "num_tokens": 125300449.0, + "step": 7822 + }, + { + "epoch": 0.6450894697781809, + "grad_norm": 1.8489683866500854, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7755968570709229, + "num_tokens": 125331944.0, + "step": 7823 + }, + { + "epoch": 0.6451719304032325, + "grad_norm": 1.9768565893173218, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7437210083007812, + "num_tokens": 125360774.0, + "step": 7824 + }, + { + "epoch": 0.645254391028284, + "grad_norm": 1.859060525894165, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7364033460617065, + "num_tokens": 125393396.0, + "step": 7825 + }, + { + "epoch": 0.6453368516533355, + "grad_norm": 1.7716076374053955, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7128447890281677, + "num_tokens": 125429556.0, + "step": 7826 + }, + { + "epoch": 0.6454193122783871, + "grad_norm": 2.0224289894104004, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7682501077651978, + "num_tokens": 125456228.0, + "step": 7827 + }, + { + "epoch": 0.6455017729034386, + "grad_norm": 1.9911108016967773, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7417322397232056, + "num_tokens": 125484896.0, + "step": 7828 + }, + { + "epoch": 0.6455842335284901, + "grad_norm": 1.7607895135879517, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.758615255355835, + "num_tokens": 125521777.0, + "step": 7829 + }, + { + "epoch": 0.6456666941535417, + "grad_norm": 1.6817883253097534, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7592730522155762, + "num_tokens": 125558908.0, + "step": 7830 + }, + { + "epoch": 0.6457491547785932, + "grad_norm": 1.9163459539413452, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7591934204101562, + "num_tokens": 125588771.0, + "step": 7831 + }, + { + "epoch": 0.6458316154036448, + "grad_norm": 1.7762516736984253, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7837842702865601, + "num_tokens": 125620249.0, + "step": 7832 + }, + { + "epoch": 0.6459140760286963, + "grad_norm": 1.7749557495117188, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7280514240264893, + "num_tokens": 125658031.0, + "step": 7833 + }, + { + "epoch": 0.6459965366537478, + "grad_norm": 2.125143527984619, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.742798924446106, + "num_tokens": 125682440.0, + "step": 7834 + }, + { + "epoch": 0.6460789972787994, + "grad_norm": 2.042600393295288, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7735801935195923, + "num_tokens": 125707949.0, + "step": 7835 + }, + { + "epoch": 0.6461614579038509, + "grad_norm": 2.0045554637908936, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7451302409172058, + "num_tokens": 125735509.0, + "step": 7836 + }, + { + "epoch": 0.6462439185289024, + "grad_norm": 1.721506953239441, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7580009698867798, + "num_tokens": 125774869.0, + "step": 7837 + }, + { + "epoch": 0.646326379153954, + "grad_norm": 2.1602444648742676, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7195791006088257, + "num_tokens": 125800300.0, + "step": 7838 + }, + { + "epoch": 0.6464088397790055, + "grad_norm": 2.020667791366577, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7628995180130005, + "num_tokens": 125825604.0, + "step": 7839 + }, + { + "epoch": 0.6464913004040571, + "grad_norm": 2.3254621028900146, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7269159555435181, + "num_tokens": 125845974.0, + "step": 7840 + }, + { + "epoch": 0.6465737610291086, + "grad_norm": 1.9809892177581787, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7296385765075684, + "num_tokens": 125877500.0, + "step": 7841 + }, + { + "epoch": 0.6466562216541601, + "grad_norm": 2.0706229209899902, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7742440700531006, + "num_tokens": 125902406.0, + "step": 7842 + }, + { + "epoch": 0.6467386822792117, + "grad_norm": 2.0396769046783447, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7443116903305054, + "num_tokens": 125931431.0, + "step": 7843 + }, + { + "epoch": 0.6468211429042632, + "grad_norm": 2.0581793785095215, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7798502445220947, + "num_tokens": 125957406.0, + "step": 7844 + }, + { + "epoch": 0.6469036035293148, + "grad_norm": 2.1347568035125732, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7170853614807129, + "num_tokens": 125982599.0, + "step": 7845 + }, + { + "epoch": 0.6469860641543663, + "grad_norm": 1.7070659399032593, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.8036571145057678, + "num_tokens": 126016270.0, + "step": 7846 + }, + { + "epoch": 0.6470685247794178, + "grad_norm": 2.0324511528015137, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.751282274723053, + "num_tokens": 126045389.0, + "step": 7847 + }, + { + "epoch": 0.6471509854044694, + "grad_norm": 2.189953565597534, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.745262622833252, + "num_tokens": 126069699.0, + "step": 7848 + }, + { + "epoch": 0.6472334460295209, + "grad_norm": 2.0073113441467285, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7420322895050049, + "num_tokens": 126097892.0, + "step": 7849 + }, + { + "epoch": 0.6473159066545724, + "grad_norm": 2.3153467178344727, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7270981669425964, + "num_tokens": 126120343.0, + "step": 7850 + }, + { + "epoch": 0.647398367279624, + "grad_norm": 1.9123072624206543, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7396339178085327, + "num_tokens": 126153118.0, + "step": 7851 + }, + { + "epoch": 0.6474808279046755, + "grad_norm": 2.0388238430023193, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7573549747467041, + "num_tokens": 126180311.0, + "step": 7852 + }, + { + "epoch": 0.6475632885297271, + "grad_norm": 1.9256497621536255, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7450460195541382, + "num_tokens": 126211011.0, + "step": 7853 + }, + { + "epoch": 0.6476457491547786, + "grad_norm": 1.9430723190307617, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.715875506401062, + "num_tokens": 126241492.0, + "step": 7854 + }, + { + "epoch": 0.6477282097798301, + "grad_norm": 1.9028764963150024, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7264463901519775, + "num_tokens": 126274413.0, + "step": 7855 + }, + { + "epoch": 0.6478106704048817, + "grad_norm": 1.8802331686019897, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7772692441940308, + "num_tokens": 126308005.0, + "step": 7856 + }, + { + "epoch": 0.6478931310299332, + "grad_norm": 2.037860870361328, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7252093553543091, + "num_tokens": 126333080.0, + "step": 7857 + }, + { + "epoch": 0.6479755916549848, + "grad_norm": 1.9000439643859863, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7162695527076721, + "num_tokens": 126364225.0, + "step": 7858 + }, + { + "epoch": 0.6480580522800363, + "grad_norm": 1.9439481496810913, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7134484052658081, + "num_tokens": 126396473.0, + "step": 7859 + }, + { + "epoch": 0.6481405129050878, + "grad_norm": 1.8802975416183472, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7692732810974121, + "num_tokens": 126428093.0, + "step": 7860 + }, + { + "epoch": 0.6482229735301394, + "grad_norm": 2.2005012035369873, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6894327998161316, + "num_tokens": 126456190.0, + "step": 7861 + }, + { + "epoch": 0.6483054341551909, + "grad_norm": 1.964882493019104, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7554246187210083, + "num_tokens": 126488464.0, + "step": 7862 + }, + { + "epoch": 0.6483878947802424, + "grad_norm": 1.9855691194534302, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7527591586112976, + "num_tokens": 126518863.0, + "step": 7863 + }, + { + "epoch": 0.648470355405294, + "grad_norm": 2.2337067127227783, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7562682628631592, + "num_tokens": 126542246.0, + "step": 7864 + }, + { + "epoch": 0.6485528160303455, + "grad_norm": 1.7580156326293945, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7209928035736084, + "num_tokens": 126580673.0, + "step": 7865 + }, + { + "epoch": 0.648635276655397, + "grad_norm": 2.1664328575134277, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.738665759563446, + "num_tokens": 126605330.0, + "step": 7866 + }, + { + "epoch": 0.6487177372804486, + "grad_norm": 1.9064881801605225, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7373558282852173, + "num_tokens": 126636308.0, + "step": 7867 + }, + { + "epoch": 0.6488001979055001, + "grad_norm": 1.6776516437530518, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7593623399734497, + "num_tokens": 126677819.0, + "step": 7868 + }, + { + "epoch": 0.6488826585305517, + "grad_norm": 1.8093973398208618, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7203453779220581, + "num_tokens": 126714393.0, + "step": 7869 + }, + { + "epoch": 0.6489651191556032, + "grad_norm": 2.1567015647888184, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7387050986289978, + "num_tokens": 126739111.0, + "step": 7870 + }, + { + "epoch": 0.6490475797806547, + "grad_norm": 2.033801317214966, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.7834630608558655, + "num_tokens": 126764813.0, + "step": 7871 + }, + { + "epoch": 0.6491300404057063, + "grad_norm": 1.686492919921875, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.7937184572219849, + "num_tokens": 126798973.0, + "step": 7872 + }, + { + "epoch": 0.6492125010307578, + "grad_norm": 2.027066469192505, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7608926892280579, + "num_tokens": 126825795.0, + "step": 7873 + }, + { + "epoch": 0.6492949616558094, + "grad_norm": 2.0517783164978027, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7367242574691772, + "num_tokens": 126853758.0, + "step": 7874 + }, + { + "epoch": 0.6493774222808609, + "grad_norm": 1.826006293296814, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7460900545120239, + "num_tokens": 126889877.0, + "step": 7875 + }, + { + "epoch": 0.6494598829059124, + "grad_norm": 1.8910435438156128, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.7919754981994629, + "num_tokens": 126919486.0, + "step": 7876 + }, + { + "epoch": 0.649542343530964, + "grad_norm": 2.3753106594085693, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7449833154678345, + "num_tokens": 126941152.0, + "step": 7877 + }, + { + "epoch": 0.6496248041560155, + "grad_norm": 1.9518253803253174, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7582601308822632, + "num_tokens": 126968477.0, + "step": 7878 + }, + { + "epoch": 0.649707264781067, + "grad_norm": 1.9263420104980469, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7283334732055664, + "num_tokens": 127000467.0, + "step": 7879 + }, + { + "epoch": 0.6497897254061186, + "grad_norm": 1.8038270473480225, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7143687009811401, + "num_tokens": 127037444.0, + "step": 7880 + }, + { + "epoch": 0.6498721860311701, + "grad_norm": 1.8345412015914917, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7727491855621338, + "num_tokens": 127070184.0, + "step": 7881 + }, + { + "epoch": 0.6499546466562216, + "grad_norm": 1.8135062456130981, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7594423890113831, + "num_tokens": 127106170.0, + "step": 7882 + }, + { + "epoch": 0.6500371072812732, + "grad_norm": 2.24238920211792, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7626826763153076, + "num_tokens": 127129150.0, + "step": 7883 + }, + { + "epoch": 0.6501195679063247, + "grad_norm": 1.7378020286560059, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.760566234588623, + "num_tokens": 127163263.0, + "step": 7884 + }, + { + "epoch": 0.6502020285313763, + "grad_norm": 1.9298768043518066, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.769451916217804, + "num_tokens": 127195758.0, + "step": 7885 + }, + { + "epoch": 0.6502844891564278, + "grad_norm": 1.7480875253677368, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7327768802642822, + "num_tokens": 127232322.0, + "step": 7886 + }, + { + "epoch": 0.6503669497814794, + "grad_norm": 1.712975263595581, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.7892981767654419, + "num_tokens": 127267780.0, + "step": 7887 + }, + { + "epoch": 0.6504494104065309, + "grad_norm": 1.8117477893829346, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.718220591545105, + "num_tokens": 127301331.0, + "step": 7888 + }, + { + "epoch": 0.6505318710315824, + "grad_norm": 2.011472225189209, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7324715852737427, + "num_tokens": 127332146.0, + "step": 7889 + }, + { + "epoch": 0.650614331656634, + "grad_norm": 2.042307138442993, + "learning_rate": 1e-06, + "loss": 0.7825, + "mean_token_accuracy": 0.7492400407791138, + "num_tokens": 127359861.0, + "step": 7890 + }, + { + "epoch": 0.6506967922816855, + "grad_norm": 1.765038013458252, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.7791178226470947, + "num_tokens": 127390491.0, + "step": 7891 + }, + { + "epoch": 0.650779252906737, + "grad_norm": 1.9553401470184326, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7452713251113892, + "num_tokens": 127418600.0, + "step": 7892 + }, + { + "epoch": 0.6508617135317886, + "grad_norm": 2.3553173542022705, + "learning_rate": 1e-06, + "loss": 0.734, + "mean_token_accuracy": 0.7633444666862488, + "num_tokens": 127439514.0, + "step": 7893 + }, + { + "epoch": 0.6509441741568401, + "grad_norm": 2.0514190196990967, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7636258602142334, + "num_tokens": 127464498.0, + "step": 7894 + }, + { + "epoch": 0.6510266347818916, + "grad_norm": 1.724165916442871, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7724090814590454, + "num_tokens": 127497563.0, + "step": 7895 + }, + { + "epoch": 0.6511090954069432, + "grad_norm": 2.2320187091827393, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7438826560974121, + "num_tokens": 127521618.0, + "step": 7896 + }, + { + "epoch": 0.6511915560319947, + "grad_norm": 1.9435930252075195, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7545316815376282, + "num_tokens": 127552323.0, + "step": 7897 + }, + { + "epoch": 0.6512740166570462, + "grad_norm": 1.7036280632019043, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7713901996612549, + "num_tokens": 127588827.0, + "step": 7898 + }, + { + "epoch": 0.6513564772820978, + "grad_norm": 2.0599613189697266, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7231513261795044, + "num_tokens": 127621624.0, + "step": 7899 + }, + { + "epoch": 0.6514389379071494, + "grad_norm": 1.9944404363632202, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7539633512496948, + "num_tokens": 127651611.0, + "step": 7900 + }, + { + "epoch": 0.6515213985322009, + "grad_norm": 2.0908029079437256, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.703951358795166, + "num_tokens": 127683443.0, + "step": 7901 + }, + { + "epoch": 0.6516038591572524, + "grad_norm": 1.9238964319229126, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7614232897758484, + "num_tokens": 127711950.0, + "step": 7902 + }, + { + "epoch": 0.651686319782304, + "grad_norm": 1.7746111154556274, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7589701414108276, + "num_tokens": 127746620.0, + "step": 7903 + }, + { + "epoch": 0.6517687804073555, + "grad_norm": 2.177252769470215, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.746590256690979, + "num_tokens": 127770897.0, + "step": 7904 + }, + { + "epoch": 0.651851241032407, + "grad_norm": 2.015714406967163, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7304701805114746, + "num_tokens": 127800358.0, + "step": 7905 + }, + { + "epoch": 0.6519337016574586, + "grad_norm": 1.8304892778396606, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7701629400253296, + "num_tokens": 127832052.0, + "step": 7906 + }, + { + "epoch": 0.6520161622825101, + "grad_norm": 1.8903112411499023, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7378143072128296, + "num_tokens": 127865806.0, + "step": 7907 + }, + { + "epoch": 0.6520986229075616, + "grad_norm": 1.8677952289581299, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7491118311882019, + "num_tokens": 127896504.0, + "step": 7908 + }, + { + "epoch": 0.6521810835326132, + "grad_norm": 2.299861192703247, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7556718587875366, + "num_tokens": 127920923.0, + "step": 7909 + }, + { + "epoch": 0.6522635441576647, + "grad_norm": 1.962813138961792, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.804747462272644, + "num_tokens": 127954375.0, + "step": 7910 + }, + { + "epoch": 0.6523460047827162, + "grad_norm": 1.9591165781021118, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7682109475135803, + "num_tokens": 127982022.0, + "step": 7911 + }, + { + "epoch": 0.6524284654077678, + "grad_norm": 2.0390403270721436, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7287775278091431, + "num_tokens": 128017514.0, + "step": 7912 + }, + { + "epoch": 0.6525109260328193, + "grad_norm": 1.919211506843567, + "learning_rate": 1e-06, + "loss": 0.7768, + "mean_token_accuracy": 0.7531625032424927, + "num_tokens": 128049745.0, + "step": 7913 + }, + { + "epoch": 0.6525933866578709, + "grad_norm": 1.9044201374053955, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.8048573732376099, + "num_tokens": 128079107.0, + "step": 7914 + }, + { + "epoch": 0.6526758472829224, + "grad_norm": 1.9897105693817139, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7892472743988037, + "num_tokens": 128108008.0, + "step": 7915 + }, + { + "epoch": 0.652758307907974, + "grad_norm": 1.926592469215393, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7677024602890015, + "num_tokens": 128137720.0, + "step": 7916 + }, + { + "epoch": 0.6528407685330255, + "grad_norm": 1.8352055549621582, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.759534478187561, + "num_tokens": 128170257.0, + "step": 7917 + }, + { + "epoch": 0.652923229158077, + "grad_norm": 1.7839770317077637, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7751550078392029, + "num_tokens": 128201693.0, + "step": 7918 + }, + { + "epoch": 0.6530056897831286, + "grad_norm": 1.9056071043014526, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7365891933441162, + "num_tokens": 128233771.0, + "step": 7919 + }, + { + "epoch": 0.6530881504081801, + "grad_norm": 2.3911423683166504, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7473528385162354, + "num_tokens": 128255533.0, + "step": 7920 + }, + { + "epoch": 0.6531706110332316, + "grad_norm": 2.012362241744995, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.7863397002220154, + "num_tokens": 128283436.0, + "step": 7921 + }, + { + "epoch": 0.6532530716582832, + "grad_norm": 2.0314083099365234, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.800263524055481, + "num_tokens": 128310747.0, + "step": 7922 + }, + { + "epoch": 0.6533355322833347, + "grad_norm": 1.770401954650879, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.770403265953064, + "num_tokens": 128347261.0, + "step": 7923 + }, + { + "epoch": 0.6534179929083862, + "grad_norm": 2.0094094276428223, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7731591463088989, + "num_tokens": 128375294.0, + "step": 7924 + }, + { + "epoch": 0.6535004535334378, + "grad_norm": 1.8273248672485352, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7676742076873779, + "num_tokens": 128408607.0, + "step": 7925 + }, + { + "epoch": 0.6535829141584893, + "grad_norm": 1.99689519405365, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7256439924240112, + "num_tokens": 128440782.0, + "step": 7926 + }, + { + "epoch": 0.6536653747835408, + "grad_norm": 2.020160436630249, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7428448796272278, + "num_tokens": 128470362.0, + "step": 7927 + }, + { + "epoch": 0.6537478354085924, + "grad_norm": 2.082880735397339, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7329050302505493, + "num_tokens": 128498134.0, + "step": 7928 + }, + { + "epoch": 0.653830296033644, + "grad_norm": 1.9299029111862183, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.724636435508728, + "num_tokens": 128532382.0, + "step": 7929 + }, + { + "epoch": 0.6539127566586955, + "grad_norm": 1.9796913862228394, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7093325853347778, + "num_tokens": 128564300.0, + "step": 7930 + }, + { + "epoch": 0.653995217283747, + "grad_norm": 1.8494760990142822, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7690500020980835, + "num_tokens": 128595665.0, + "step": 7931 + }, + { + "epoch": 0.6540776779087986, + "grad_norm": 2.0868427753448486, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.7548021078109741, + "num_tokens": 128622365.0, + "step": 7932 + }, + { + "epoch": 0.6541601385338501, + "grad_norm": 1.83481764793396, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7566664814949036, + "num_tokens": 128654886.0, + "step": 7933 + }, + { + "epoch": 0.6542425991589016, + "grad_norm": 1.883625864982605, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7831436991691589, + "num_tokens": 128685085.0, + "step": 7934 + }, + { + "epoch": 0.6543250597839532, + "grad_norm": 1.6518505811691284, + "learning_rate": 1e-06, + "loss": 0.6118, + "mean_token_accuracy": 0.7926305532455444, + "num_tokens": 128721265.0, + "step": 7935 + }, + { + "epoch": 0.6544075204090047, + "grad_norm": 2.1094555854797363, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7493264675140381, + "num_tokens": 128749690.0, + "step": 7936 + }, + { + "epoch": 0.6544899810340562, + "grad_norm": 1.9755789041519165, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7175171971321106, + "num_tokens": 128779452.0, + "step": 7937 + }, + { + "epoch": 0.6545724416591078, + "grad_norm": 1.8977586030960083, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7393567562103271, + "num_tokens": 128812138.0, + "step": 7938 + }, + { + "epoch": 0.6546549022841593, + "grad_norm": 2.0539000034332275, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.756106972694397, + "num_tokens": 128836799.0, + "step": 7939 + }, + { + "epoch": 0.6547373629092108, + "grad_norm": 1.9507931470870972, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7813291549682617, + "num_tokens": 128866290.0, + "step": 7940 + }, + { + "epoch": 0.6548198235342624, + "grad_norm": 2.0774526596069336, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7448379397392273, + "num_tokens": 128892130.0, + "step": 7941 + }, + { + "epoch": 0.654902284159314, + "grad_norm": 1.9019757509231567, + "learning_rate": 1e-06, + "loss": 0.6166, + "mean_token_accuracy": 0.7975651025772095, + "num_tokens": 128920808.0, + "step": 7942 + }, + { + "epoch": 0.6549847447843654, + "grad_norm": 1.8791042566299438, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7564190626144409, + "num_tokens": 128954487.0, + "step": 7943 + }, + { + "epoch": 0.655067205409417, + "grad_norm": 1.9425288438796997, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7422851324081421, + "num_tokens": 128984803.0, + "step": 7944 + }, + { + "epoch": 0.6551496660344686, + "grad_norm": 2.25534987449646, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7528877854347229, + "num_tokens": 129008822.0, + "step": 7945 + }, + { + "epoch": 0.6552321266595201, + "grad_norm": 1.9423964023590088, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7306578159332275, + "num_tokens": 129039596.0, + "step": 7946 + }, + { + "epoch": 0.6553145872845716, + "grad_norm": 2.0163462162017822, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7288192510604858, + "num_tokens": 129069166.0, + "step": 7947 + }, + { + "epoch": 0.6553970479096232, + "grad_norm": 2.3056540489196777, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7153661251068115, + "num_tokens": 129091755.0, + "step": 7948 + }, + { + "epoch": 0.6554795085346747, + "grad_norm": 1.9751003980636597, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7603477239608765, + "num_tokens": 129119709.0, + "step": 7949 + }, + { + "epoch": 0.6555619691597262, + "grad_norm": 2.0964291095733643, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.7860668897628784, + "num_tokens": 129143297.0, + "step": 7950 + }, + { + "epoch": 0.6556444297847778, + "grad_norm": 1.7764930725097656, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7224336862564087, + "num_tokens": 129177399.0, + "step": 7951 + }, + { + "epoch": 0.6557268904098293, + "grad_norm": 2.050861358642578, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7504521012306213, + "num_tokens": 129204960.0, + "step": 7952 + }, + { + "epoch": 0.6558093510348808, + "grad_norm": 2.1235806941986084, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7751021385192871, + "num_tokens": 129229117.0, + "step": 7953 + }, + { + "epoch": 0.6558918116599324, + "grad_norm": 2.7198851108551025, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7108463048934937, + "num_tokens": 129258791.0, + "step": 7954 + }, + { + "epoch": 0.655974272284984, + "grad_norm": 1.7465991973876953, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.773212730884552, + "num_tokens": 129296709.0, + "step": 7955 + }, + { + "epoch": 0.6560567329100354, + "grad_norm": 1.8587474822998047, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7639732360839844, + "num_tokens": 129328714.0, + "step": 7956 + }, + { + "epoch": 0.656139193535087, + "grad_norm": 2.232358455657959, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7478494644165039, + "num_tokens": 129353311.0, + "step": 7957 + }, + { + "epoch": 0.6562216541601386, + "grad_norm": 1.9989750385284424, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7392921447753906, + "num_tokens": 129380782.0, + "step": 7958 + }, + { + "epoch": 0.65630411478519, + "grad_norm": 2.0819427967071533, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7688878774642944, + "num_tokens": 129406751.0, + "step": 7959 + }, + { + "epoch": 0.6563865754102416, + "grad_norm": 1.7780942916870117, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.757999062538147, + "num_tokens": 129442888.0, + "step": 7960 + }, + { + "epoch": 0.6564690360352932, + "grad_norm": 2.236510992050171, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7178483009338379, + "num_tokens": 129469461.0, + "step": 7961 + }, + { + "epoch": 0.6565514966603447, + "grad_norm": 2.028029441833496, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7666643857955933, + "num_tokens": 129500494.0, + "step": 7962 + }, + { + "epoch": 0.6566339572853962, + "grad_norm": 1.938481092453003, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7425597906112671, + "num_tokens": 129530495.0, + "step": 7963 + }, + { + "epoch": 0.6567164179104478, + "grad_norm": 2.047349691390991, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7763444185256958, + "num_tokens": 129556137.0, + "step": 7964 + }, + { + "epoch": 0.6567988785354993, + "grad_norm": 1.9318370819091797, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7468876838684082, + "num_tokens": 129587510.0, + "step": 7965 + }, + { + "epoch": 0.6568813391605508, + "grad_norm": 1.9260748624801636, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8024819493293762, + "num_tokens": 129614955.0, + "step": 7966 + }, + { + "epoch": 0.6569637997856024, + "grad_norm": 2.0758657455444336, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7419513463973999, + "num_tokens": 129641317.0, + "step": 7967 + }, + { + "epoch": 0.657046260410654, + "grad_norm": 1.7587649822235107, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7775434255599976, + "num_tokens": 129674303.0, + "step": 7968 + }, + { + "epoch": 0.6571287210357054, + "grad_norm": 2.0786023139953613, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6948932409286499, + "num_tokens": 129705400.0, + "step": 7969 + }, + { + "epoch": 0.657211181660757, + "grad_norm": 2.1389427185058594, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7240207195281982, + "num_tokens": 129732350.0, + "step": 7970 + }, + { + "epoch": 0.6572936422858086, + "grad_norm": 1.8846124410629272, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7248495221138, + "num_tokens": 129763548.0, + "step": 7971 + }, + { + "epoch": 0.65737610291086, + "grad_norm": 2.1339409351348877, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7068421244621277, + "num_tokens": 129789489.0, + "step": 7972 + }, + { + "epoch": 0.6574585635359116, + "grad_norm": 1.8913894891738892, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7351607084274292, + "num_tokens": 129821429.0, + "step": 7973 + }, + { + "epoch": 0.6575410241609632, + "grad_norm": 1.8247898817062378, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7405310869216919, + "num_tokens": 129855381.0, + "step": 7974 + }, + { + "epoch": 0.6576234847860146, + "grad_norm": 1.8859376907348633, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7604049444198608, + "num_tokens": 129885886.0, + "step": 7975 + }, + { + "epoch": 0.6577059454110662, + "grad_norm": 1.751686453819275, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.743910551071167, + "num_tokens": 129922090.0, + "step": 7976 + }, + { + "epoch": 0.6577884060361178, + "grad_norm": 2.1342203617095947, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7113534212112427, + "num_tokens": 129947700.0, + "step": 7977 + }, + { + "epoch": 0.6578708666611693, + "grad_norm": 2.02103328704834, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7432670593261719, + "num_tokens": 129976331.0, + "step": 7978 + }, + { + "epoch": 0.6579533272862208, + "grad_norm": 1.7751213312149048, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7518189549446106, + "num_tokens": 130012710.0, + "step": 7979 + }, + { + "epoch": 0.6580357879112724, + "grad_norm": 2.0285871028900146, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7288416028022766, + "num_tokens": 130038594.0, + "step": 7980 + }, + { + "epoch": 0.658118248536324, + "grad_norm": 1.9016170501708984, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7041895389556885, + "num_tokens": 130071798.0, + "step": 7981 + }, + { + "epoch": 0.6582007091613754, + "grad_norm": 1.8843965530395508, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7633194327354431, + "num_tokens": 130105576.0, + "step": 7982 + }, + { + "epoch": 0.658283169786427, + "grad_norm": 1.85537850856781, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7671762108802795, + "num_tokens": 130135992.0, + "step": 7983 + }, + { + "epoch": 0.6583656304114786, + "grad_norm": 2.0735723972320557, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7538599371910095, + "num_tokens": 130163317.0, + "step": 7984 + }, + { + "epoch": 0.65844809103653, + "grad_norm": 1.7119576930999756, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7653358578681946, + "num_tokens": 130201823.0, + "step": 7985 + }, + { + "epoch": 0.6585305516615816, + "grad_norm": 1.988277554512024, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7490448951721191, + "num_tokens": 130232065.0, + "step": 7986 + }, + { + "epoch": 0.6586130122866332, + "grad_norm": 1.9939805269241333, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7432920932769775, + "num_tokens": 130265914.0, + "step": 7987 + }, + { + "epoch": 0.6586954729116846, + "grad_norm": 1.97765052318573, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7377808094024658, + "num_tokens": 130294665.0, + "step": 7988 + }, + { + "epoch": 0.6587779335367362, + "grad_norm": 2.0788722038269043, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7694950103759766, + "num_tokens": 130321373.0, + "step": 7989 + }, + { + "epoch": 0.6588603941617878, + "grad_norm": 2.3179235458374023, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7414650917053223, + "num_tokens": 130344565.0, + "step": 7990 + }, + { + "epoch": 0.6589428547868393, + "grad_norm": 1.9140068292617798, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7597212791442871, + "num_tokens": 130374472.0, + "step": 7991 + }, + { + "epoch": 0.6590253154118908, + "grad_norm": 2.0801219940185547, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7406975626945496, + "num_tokens": 130402074.0, + "step": 7992 + }, + { + "epoch": 0.6591077760369424, + "grad_norm": 1.7392700910568237, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7090429067611694, + "num_tokens": 130441785.0, + "step": 7993 + }, + { + "epoch": 0.659190236661994, + "grad_norm": 2.21840238571167, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.716023325920105, + "num_tokens": 130466918.0, + "step": 7994 + }, + { + "epoch": 0.6592726972870454, + "grad_norm": 2.0397117137908936, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7688746452331543, + "num_tokens": 130494994.0, + "step": 7995 + }, + { + "epoch": 0.659355157912097, + "grad_norm": 1.8671839237213135, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.744615375995636, + "num_tokens": 130529549.0, + "step": 7996 + }, + { + "epoch": 0.6594376185371486, + "grad_norm": 2.202526569366455, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7491371631622314, + "num_tokens": 130553229.0, + "step": 7997 + }, + { + "epoch": 0.6595200791622, + "grad_norm": 1.7713661193847656, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7681856155395508, + "num_tokens": 130587855.0, + "step": 7998 + }, + { + "epoch": 0.6596025397872516, + "grad_norm": 1.9078327417373657, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7262804508209229, + "num_tokens": 130616869.0, + "step": 7999 + }, + { + "epoch": 0.6596850004123032, + "grad_norm": 1.9904085397720337, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.750846266746521, + "num_tokens": 130643495.0, + "step": 8000 + }, + { + "epoch": 0.6597674610373546, + "grad_norm": 1.7633051872253418, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7705763578414917, + "num_tokens": 130676656.0, + "step": 8001 + }, + { + "epoch": 0.6598499216624062, + "grad_norm": 2.284794569015503, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7780991792678833, + "num_tokens": 130697998.0, + "step": 8002 + }, + { + "epoch": 0.6599323822874578, + "grad_norm": 1.9570029973983765, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7552881240844727, + "num_tokens": 130729387.0, + "step": 8003 + }, + { + "epoch": 0.6600148429125092, + "grad_norm": 1.892027735710144, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.74042809009552, + "num_tokens": 130761313.0, + "step": 8004 + }, + { + "epoch": 0.6600973035375608, + "grad_norm": 2.0123071670532227, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7709354162216187, + "num_tokens": 130791205.0, + "step": 8005 + }, + { + "epoch": 0.6601797641626124, + "grad_norm": 1.8827533721923828, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7658793330192566, + "num_tokens": 130822111.0, + "step": 8006 + }, + { + "epoch": 0.6602622247876639, + "grad_norm": 2.0004942417144775, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7322554588317871, + "num_tokens": 130850805.0, + "step": 8007 + }, + { + "epoch": 0.6603446854127154, + "grad_norm": 2.039661169052124, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7389305830001831, + "num_tokens": 130876349.0, + "step": 8008 + }, + { + "epoch": 0.660427146037767, + "grad_norm": 1.8881655931472778, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.7994455099105835, + "num_tokens": 130904575.0, + "step": 8009 + }, + { + "epoch": 0.6605096066628185, + "grad_norm": 1.6833674907684326, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7712984085083008, + "num_tokens": 130939118.0, + "step": 8010 + }, + { + "epoch": 0.66059206728787, + "grad_norm": 2.019994020462036, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7178444266319275, + "num_tokens": 130968163.0, + "step": 8011 + }, + { + "epoch": 0.6606745279129216, + "grad_norm": 1.9565201997756958, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7144285440444946, + "num_tokens": 131002792.0, + "step": 8012 + }, + { + "epoch": 0.6607569885379732, + "grad_norm": 1.9670416116714478, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7764915227890015, + "num_tokens": 131032561.0, + "step": 8013 + }, + { + "epoch": 0.6608394491630246, + "grad_norm": 1.8701705932617188, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7439708709716797, + "num_tokens": 131062374.0, + "step": 8014 + }, + { + "epoch": 0.6609219097880762, + "grad_norm": 1.8974049091339111, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7661846876144409, + "num_tokens": 131091598.0, + "step": 8015 + }, + { + "epoch": 0.6610043704131278, + "grad_norm": 1.9925029277801514, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7486276626586914, + "num_tokens": 131120949.0, + "step": 8016 + }, + { + "epoch": 0.6610868310381792, + "grad_norm": 1.952480435371399, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.7586559653282166, + "num_tokens": 131153675.0, + "step": 8017 + }, + { + "epoch": 0.6611692916632308, + "grad_norm": 2.1394381523132324, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7341437339782715, + "num_tokens": 131179616.0, + "step": 8018 + }, + { + "epoch": 0.6612517522882824, + "grad_norm": 1.8736329078674316, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7468099594116211, + "num_tokens": 131209783.0, + "step": 8019 + }, + { + "epoch": 0.6613342129133338, + "grad_norm": 1.7665261030197144, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7650769948959351, + "num_tokens": 131242496.0, + "step": 8020 + }, + { + "epoch": 0.6614166735383854, + "grad_norm": 2.2428503036499023, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7558477520942688, + "num_tokens": 131266399.0, + "step": 8021 + }, + { + "epoch": 0.661499134163437, + "grad_norm": 1.8480201959609985, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7518392205238342, + "num_tokens": 131301667.0, + "step": 8022 + }, + { + "epoch": 0.6615815947884885, + "grad_norm": 2.0967109203338623, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7275661826133728, + "num_tokens": 131331017.0, + "step": 8023 + }, + { + "epoch": 0.66166405541354, + "grad_norm": 2.0607855319976807, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7474815845489502, + "num_tokens": 131358457.0, + "step": 8024 + }, + { + "epoch": 0.6617465160385916, + "grad_norm": 2.137129068374634, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7145144939422607, + "num_tokens": 131383092.0, + "step": 8025 + }, + { + "epoch": 0.6618289766636432, + "grad_norm": 1.952837586402893, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7666573524475098, + "num_tokens": 131416316.0, + "step": 8026 + }, + { + "epoch": 0.6619114372886946, + "grad_norm": 1.986443281173706, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7778332233428955, + "num_tokens": 131444058.0, + "step": 8027 + }, + { + "epoch": 0.6619938979137462, + "grad_norm": 1.6348201036453247, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7476016283035278, + "num_tokens": 131484109.0, + "step": 8028 + }, + { + "epoch": 0.6620763585387978, + "grad_norm": 1.9440234899520874, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.767540693283081, + "num_tokens": 131512178.0, + "step": 8029 + }, + { + "epoch": 0.6621588191638492, + "grad_norm": 1.868228554725647, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7618036270141602, + "num_tokens": 131546282.0, + "step": 8030 + }, + { + "epoch": 0.6622412797889008, + "grad_norm": 1.65420401096344, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.8035295605659485, + "num_tokens": 131579469.0, + "step": 8031 + }, + { + "epoch": 0.6623237404139524, + "grad_norm": 1.996931552886963, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.759850025177002, + "num_tokens": 131609190.0, + "step": 8032 + }, + { + "epoch": 0.6624062010390038, + "grad_norm": 2.1357831954956055, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.726476788520813, + "num_tokens": 131633975.0, + "step": 8033 + }, + { + "epoch": 0.6624886616640554, + "grad_norm": 1.8958969116210938, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7443386316299438, + "num_tokens": 131665065.0, + "step": 8034 + }, + { + "epoch": 0.662571122289107, + "grad_norm": 2.1951303482055664, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.740882158279419, + "num_tokens": 131690293.0, + "step": 8035 + }, + { + "epoch": 0.6626535829141584, + "grad_norm": 1.7965620756149292, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7810372114181519, + "num_tokens": 131722000.0, + "step": 8036 + }, + { + "epoch": 0.66273604353921, + "grad_norm": 1.9369721412658691, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7382371425628662, + "num_tokens": 131752100.0, + "step": 8037 + }, + { + "epoch": 0.6628185041642616, + "grad_norm": 2.1320343017578125, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7721001505851746, + "num_tokens": 131779590.0, + "step": 8038 + }, + { + "epoch": 0.6629009647893132, + "grad_norm": 1.8685294389724731, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7262859344482422, + "num_tokens": 131812890.0, + "step": 8039 + }, + { + "epoch": 0.6629834254143646, + "grad_norm": 1.743796467781067, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7631645798683167, + "num_tokens": 131848180.0, + "step": 8040 + }, + { + "epoch": 0.6630658860394162, + "grad_norm": 2.1315534114837646, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7351016998291016, + "num_tokens": 131874579.0, + "step": 8041 + }, + { + "epoch": 0.6631483466644678, + "grad_norm": 1.9399068355560303, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7699596285820007, + "num_tokens": 131905269.0, + "step": 8042 + }, + { + "epoch": 0.6632308072895192, + "grad_norm": 1.8583983182907104, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.7834397554397583, + "num_tokens": 131937245.0, + "step": 8043 + }, + { + "epoch": 0.6633132679145708, + "grad_norm": 1.9561480283737183, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7591758370399475, + "num_tokens": 131967538.0, + "step": 8044 + }, + { + "epoch": 0.6633957285396224, + "grad_norm": 2.0257034301757812, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.774273157119751, + "num_tokens": 131995181.0, + "step": 8045 + }, + { + "epoch": 0.6634781891646738, + "grad_norm": 1.9901176691055298, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7274713516235352, + "num_tokens": 132026057.0, + "step": 8046 + }, + { + "epoch": 0.6635606497897254, + "grad_norm": 2.3557965755462646, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7329365015029907, + "num_tokens": 132047541.0, + "step": 8047 + }, + { + "epoch": 0.663643110414777, + "grad_norm": 1.8278088569641113, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7556750774383545, + "num_tokens": 132079934.0, + "step": 8048 + }, + { + "epoch": 0.6637255710398284, + "grad_norm": 1.8170772790908813, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7358946800231934, + "num_tokens": 132115450.0, + "step": 8049 + }, + { + "epoch": 0.66380803166488, + "grad_norm": 2.024137496948242, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7447793483734131, + "num_tokens": 132146780.0, + "step": 8050 + }, + { + "epoch": 0.6638904922899316, + "grad_norm": 2.0654397010803223, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.7853410840034485, + "num_tokens": 132172740.0, + "step": 8051 + }, + { + "epoch": 0.6639729529149831, + "grad_norm": 1.9443225860595703, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7412855625152588, + "num_tokens": 132201923.0, + "step": 8052 + }, + { + "epoch": 0.6640554135400346, + "grad_norm": 1.7735925912857056, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7615732550621033, + "num_tokens": 132237220.0, + "step": 8053 + }, + { + "epoch": 0.6641378741650862, + "grad_norm": 1.8265374898910522, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7470592260360718, + "num_tokens": 132271913.0, + "step": 8054 + }, + { + "epoch": 0.6642203347901378, + "grad_norm": 1.7997264862060547, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7558170557022095, + "num_tokens": 132307774.0, + "step": 8055 + }, + { + "epoch": 0.6643027954151892, + "grad_norm": 1.8435152769088745, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.7554566860198975, + "num_tokens": 132342974.0, + "step": 8056 + }, + { + "epoch": 0.6643852560402408, + "grad_norm": 2.023138999938965, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7767740488052368, + "num_tokens": 132370097.0, + "step": 8057 + }, + { + "epoch": 0.6644677166652924, + "grad_norm": 1.838591456413269, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7117083668708801, + "num_tokens": 132407721.0, + "step": 8058 + }, + { + "epoch": 0.6645501772903438, + "grad_norm": 1.8698796033859253, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7638691067695618, + "num_tokens": 132439768.0, + "step": 8059 + }, + { + "epoch": 0.6646326379153954, + "grad_norm": 1.8556345701217651, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.735348641872406, + "num_tokens": 132470837.0, + "step": 8060 + }, + { + "epoch": 0.664715098540447, + "grad_norm": 1.68229079246521, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7809720039367676, + "num_tokens": 132505208.0, + "step": 8061 + }, + { + "epoch": 0.6647975591654984, + "grad_norm": 1.8183382749557495, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7937397956848145, + "num_tokens": 132534183.0, + "step": 8062 + }, + { + "epoch": 0.66488001979055, + "grad_norm": 1.925019383430481, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7203491926193237, + "num_tokens": 132565278.0, + "step": 8063 + }, + { + "epoch": 0.6649624804156016, + "grad_norm": 1.977753758430481, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7511953711509705, + "num_tokens": 132595152.0, + "step": 8064 + }, + { + "epoch": 0.665044941040653, + "grad_norm": 1.8609585762023926, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7369766235351562, + "num_tokens": 132627530.0, + "step": 8065 + }, + { + "epoch": 0.6651274016657046, + "grad_norm": 1.917988896369934, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7533349990844727, + "num_tokens": 132657282.0, + "step": 8066 + }, + { + "epoch": 0.6652098622907562, + "grad_norm": 2.0197274684906006, + "learning_rate": 1e-06, + "loss": 0.765, + "mean_token_accuracy": 0.7544877529144287, + "num_tokens": 132685112.0, + "step": 8067 + }, + { + "epoch": 0.6652923229158078, + "grad_norm": 1.7245675325393677, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7561884522438049, + "num_tokens": 132722281.0, + "step": 8068 + }, + { + "epoch": 0.6653747835408592, + "grad_norm": 2.117230176925659, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6980947256088257, + "num_tokens": 132751049.0, + "step": 8069 + }, + { + "epoch": 0.6654572441659108, + "grad_norm": 2.0782153606414795, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7672455906867981, + "num_tokens": 132775979.0, + "step": 8070 + }, + { + "epoch": 0.6655397047909624, + "grad_norm": 1.905659556388855, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7539924383163452, + "num_tokens": 132809898.0, + "step": 8071 + }, + { + "epoch": 0.6656221654160138, + "grad_norm": 1.8878896236419678, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7450783848762512, + "num_tokens": 132840281.0, + "step": 8072 + }, + { + "epoch": 0.6657046260410654, + "grad_norm": 2.122868776321411, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7352346777915955, + "num_tokens": 132865874.0, + "step": 8073 + }, + { + "epoch": 0.665787086666117, + "grad_norm": 1.672960877418518, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.778319239616394, + "num_tokens": 132904001.0, + "step": 8074 + }, + { + "epoch": 0.6658695472911684, + "grad_norm": 2.0604634284973145, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7211560606956482, + "num_tokens": 132932342.0, + "step": 8075 + }, + { + "epoch": 0.66595200791622, + "grad_norm": 1.9003982543945312, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7761427164077759, + "num_tokens": 132961339.0, + "step": 8076 + }, + { + "epoch": 0.6660344685412716, + "grad_norm": 1.8369477987289429, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7547081708908081, + "num_tokens": 132993645.0, + "step": 8077 + }, + { + "epoch": 0.666116929166323, + "grad_norm": 1.8807399272918701, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7297144532203674, + "num_tokens": 133024416.0, + "step": 8078 + }, + { + "epoch": 0.6661993897913746, + "grad_norm": 2.379389524459839, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7207622528076172, + "num_tokens": 133047338.0, + "step": 8079 + }, + { + "epoch": 0.6662818504164262, + "grad_norm": 2.148211717605591, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7361860275268555, + "num_tokens": 133074255.0, + "step": 8080 + }, + { + "epoch": 0.6663643110414776, + "grad_norm": 1.9660110473632812, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.746429443359375, + "num_tokens": 133103403.0, + "step": 8081 + }, + { + "epoch": 0.6664467716665292, + "grad_norm": 1.819805383682251, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.732746958732605, + "num_tokens": 133141994.0, + "step": 8082 + }, + { + "epoch": 0.6665292322915808, + "grad_norm": 2.2434396743774414, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7388092279434204, + "num_tokens": 133166582.0, + "step": 8083 + }, + { + "epoch": 0.6666116929166324, + "grad_norm": 1.8304866552352905, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.7865540981292725, + "num_tokens": 133196194.0, + "step": 8084 + }, + { + "epoch": 0.6666941535416838, + "grad_norm": 1.8433643579483032, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7577721476554871, + "num_tokens": 133229621.0, + "step": 8085 + }, + { + "epoch": 0.6667766141667354, + "grad_norm": 2.3270814418792725, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7330185174942017, + "num_tokens": 133251364.0, + "step": 8086 + }, + { + "epoch": 0.666859074791787, + "grad_norm": 1.8706696033477783, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7553757429122925, + "num_tokens": 133283380.0, + "step": 8087 + }, + { + "epoch": 0.6669415354168384, + "grad_norm": 1.8613423109054565, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7592555284500122, + "num_tokens": 133314130.0, + "step": 8088 + }, + { + "epoch": 0.66702399604189, + "grad_norm": 1.9511733055114746, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7496031522750854, + "num_tokens": 133342222.0, + "step": 8089 + }, + { + "epoch": 0.6671064566669416, + "grad_norm": 1.7936352491378784, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7390596866607666, + "num_tokens": 133377152.0, + "step": 8090 + }, + { + "epoch": 0.667188917291993, + "grad_norm": 2.019472122192383, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.6906703114509583, + "num_tokens": 133410366.0, + "step": 8091 + }, + { + "epoch": 0.6672713779170446, + "grad_norm": 1.924121618270874, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7624167799949646, + "num_tokens": 133440158.0, + "step": 8092 + }, + { + "epoch": 0.6673538385420962, + "grad_norm": 2.0028021335601807, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.721499502658844, + "num_tokens": 133469236.0, + "step": 8093 + }, + { + "epoch": 0.6674362991671476, + "grad_norm": 1.6938204765319824, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7450414896011353, + "num_tokens": 133507424.0, + "step": 8094 + }, + { + "epoch": 0.6675187597921992, + "grad_norm": 2.258650302886963, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7590243816375732, + "num_tokens": 133528503.0, + "step": 8095 + }, + { + "epoch": 0.6676012204172508, + "grad_norm": 1.9086120128631592, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7172993421554565, + "num_tokens": 133561469.0, + "step": 8096 + }, + { + "epoch": 0.6676836810423022, + "grad_norm": 1.9810019731521606, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7499182820320129, + "num_tokens": 133589679.0, + "step": 8097 + }, + { + "epoch": 0.6677661416673538, + "grad_norm": 1.991194248199463, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7579034566879272, + "num_tokens": 133618343.0, + "step": 8098 + }, + { + "epoch": 0.6678486022924054, + "grad_norm": 1.806668758392334, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.743205189704895, + "num_tokens": 133650200.0, + "step": 8099 + }, + { + "epoch": 0.667931062917457, + "grad_norm": 1.7865784168243408, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7734049558639526, + "num_tokens": 133681866.0, + "step": 8100 + }, + { + "epoch": 0.6680135235425084, + "grad_norm": 1.814186453819275, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7746419310569763, + "num_tokens": 133714930.0, + "step": 8101 + }, + { + "epoch": 0.66809598416756, + "grad_norm": 2.3396966457366943, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7474661469459534, + "num_tokens": 133737169.0, + "step": 8102 + }, + { + "epoch": 0.6681784447926116, + "grad_norm": 2.099834680557251, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7461339831352234, + "num_tokens": 133763144.0, + "step": 8103 + }, + { + "epoch": 0.668260905417663, + "grad_norm": 1.788902997970581, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.764606237411499, + "num_tokens": 133796239.0, + "step": 8104 + }, + { + "epoch": 0.6683433660427146, + "grad_norm": 1.922249674797058, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7333767414093018, + "num_tokens": 133829109.0, + "step": 8105 + }, + { + "epoch": 0.6684258266677662, + "grad_norm": 2.200932025909424, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7200644612312317, + "num_tokens": 133856038.0, + "step": 8106 + }, + { + "epoch": 0.6685082872928176, + "grad_norm": 1.8881707191467285, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7572712898254395, + "num_tokens": 133887066.0, + "step": 8107 + }, + { + "epoch": 0.6685907479178692, + "grad_norm": 1.8365505933761597, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.730968713760376, + "num_tokens": 133920059.0, + "step": 8108 + }, + { + "epoch": 0.6686732085429208, + "grad_norm": 1.8536343574523926, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7669486403465271, + "num_tokens": 133952691.0, + "step": 8109 + }, + { + "epoch": 0.6687556691679722, + "grad_norm": 1.8937561511993408, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7447973489761353, + "num_tokens": 133984448.0, + "step": 8110 + }, + { + "epoch": 0.6688381297930238, + "grad_norm": 2.03816819190979, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7215372920036316, + "num_tokens": 134015069.0, + "step": 8111 + }, + { + "epoch": 0.6689205904180754, + "grad_norm": 1.7724695205688477, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.7993596196174622, + "num_tokens": 134050751.0, + "step": 8112 + }, + { + "epoch": 0.669003051043127, + "grad_norm": 1.8442564010620117, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7686006426811218, + "num_tokens": 134079498.0, + "step": 8113 + }, + { + "epoch": 0.6690855116681784, + "grad_norm": 2.0622425079345703, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7452571392059326, + "num_tokens": 134105808.0, + "step": 8114 + }, + { + "epoch": 0.66916797229323, + "grad_norm": 2.15714955329895, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7274760007858276, + "num_tokens": 134131980.0, + "step": 8115 + }, + { + "epoch": 0.6692504329182816, + "grad_norm": 1.6731719970703125, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.8087717890739441, + "num_tokens": 134165026.0, + "step": 8116 + }, + { + "epoch": 0.669332893543333, + "grad_norm": 2.0236315727233887, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7486621141433716, + "num_tokens": 134194591.0, + "step": 8117 + }, + { + "epoch": 0.6694153541683846, + "grad_norm": 2.0005078315734863, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7322157621383667, + "num_tokens": 134224148.0, + "step": 8118 + }, + { + "epoch": 0.6694978147934362, + "grad_norm": 1.8780831098556519, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7397154569625854, + "num_tokens": 134256726.0, + "step": 8119 + }, + { + "epoch": 0.6695802754184876, + "grad_norm": 1.8815234899520874, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7281839847564697, + "num_tokens": 134290371.0, + "step": 8120 + }, + { + "epoch": 0.6696627360435392, + "grad_norm": 1.8373384475708008, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7822616100311279, + "num_tokens": 134324323.0, + "step": 8121 + }, + { + "epoch": 0.6697451966685908, + "grad_norm": 1.8168537616729736, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7479985952377319, + "num_tokens": 134357919.0, + "step": 8122 + }, + { + "epoch": 0.6698276572936422, + "grad_norm": 1.8917367458343506, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.750333309173584, + "num_tokens": 134389512.0, + "step": 8123 + }, + { + "epoch": 0.6699101179186938, + "grad_norm": 1.9131516218185425, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7563828825950623, + "num_tokens": 134420215.0, + "step": 8124 + }, + { + "epoch": 0.6699925785437454, + "grad_norm": 2.1663742065429688, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.721043050289154, + "num_tokens": 134447775.0, + "step": 8125 + }, + { + "epoch": 0.6700750391687968, + "grad_norm": 1.8893238306045532, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.7917590141296387, + "num_tokens": 134477814.0, + "step": 8126 + }, + { + "epoch": 0.6701574997938484, + "grad_norm": 2.031297445297241, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7222666144371033, + "num_tokens": 134509626.0, + "step": 8127 + }, + { + "epoch": 0.6702399604189, + "grad_norm": 1.8896175622940063, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7492106556892395, + "num_tokens": 134541474.0, + "step": 8128 + }, + { + "epoch": 0.6703224210439516, + "grad_norm": 1.751322627067566, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7592755556106567, + "num_tokens": 134577674.0, + "step": 8129 + }, + { + "epoch": 0.670404881669003, + "grad_norm": 1.7825897932052612, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.7557259798049927, + "num_tokens": 134616136.0, + "step": 8130 + }, + { + "epoch": 0.6704873422940546, + "grad_norm": 1.9761755466461182, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7371193170547485, + "num_tokens": 134647201.0, + "step": 8131 + }, + { + "epoch": 0.6705698029191062, + "grad_norm": 1.7427544593811035, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7344020009040833, + "num_tokens": 134683804.0, + "step": 8132 + }, + { + "epoch": 0.6706522635441576, + "grad_norm": 1.9023445844650269, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7804005742073059, + "num_tokens": 134713560.0, + "step": 8133 + }, + { + "epoch": 0.6707347241692092, + "grad_norm": 2.285998582839966, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.778922438621521, + "num_tokens": 134736529.0, + "step": 8134 + }, + { + "epoch": 0.6708171847942608, + "grad_norm": 2.224750518798828, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7586053609848022, + "num_tokens": 134758543.0, + "step": 8135 + }, + { + "epoch": 0.6708996454193122, + "grad_norm": 1.9069997072219849, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7189720273017883, + "num_tokens": 134791127.0, + "step": 8136 + }, + { + "epoch": 0.6709821060443638, + "grad_norm": 2.1024787425994873, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7514124512672424, + "num_tokens": 134819123.0, + "step": 8137 + }, + { + "epoch": 0.6710645666694154, + "grad_norm": 1.9056472778320312, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7303444743156433, + "num_tokens": 134854731.0, + "step": 8138 + }, + { + "epoch": 0.6711470272944668, + "grad_norm": 2.297212839126587, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7501672506332397, + "num_tokens": 134876429.0, + "step": 8139 + }, + { + "epoch": 0.6712294879195184, + "grad_norm": 1.9003095626831055, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.779390275478363, + "num_tokens": 134904346.0, + "step": 8140 + }, + { + "epoch": 0.67131194854457, + "grad_norm": 1.9422987699508667, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7631034851074219, + "num_tokens": 134933522.0, + "step": 8141 + }, + { + "epoch": 0.6713944091696215, + "grad_norm": 1.6980204582214355, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7810283899307251, + "num_tokens": 134967691.0, + "step": 8142 + }, + { + "epoch": 0.671476869794673, + "grad_norm": 1.8773080110549927, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7777068018913269, + "num_tokens": 134998598.0, + "step": 8143 + }, + { + "epoch": 0.6715593304197246, + "grad_norm": 1.9763002395629883, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7787232995033264, + "num_tokens": 135027602.0, + "step": 8144 + }, + { + "epoch": 0.6716417910447762, + "grad_norm": 2.1644375324249268, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7278375029563904, + "num_tokens": 135054904.0, + "step": 8145 + }, + { + "epoch": 0.6717242516698276, + "grad_norm": 1.9650462865829468, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7608046531677246, + "num_tokens": 135083288.0, + "step": 8146 + }, + { + "epoch": 0.6718067122948792, + "grad_norm": 2.0194547176361084, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7761902809143066, + "num_tokens": 135112859.0, + "step": 8147 + }, + { + "epoch": 0.6718891729199308, + "grad_norm": 1.8004381656646729, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7498642206192017, + "num_tokens": 135147689.0, + "step": 8148 + }, + { + "epoch": 0.6719716335449822, + "grad_norm": 1.795536994934082, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7749659419059753, + "num_tokens": 135180845.0, + "step": 8149 + }, + { + "epoch": 0.6720540941700338, + "grad_norm": 1.6322922706604004, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7837877869606018, + "num_tokens": 135222552.0, + "step": 8150 + }, + { + "epoch": 0.6721365547950854, + "grad_norm": 1.7393667697906494, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7399994134902954, + "num_tokens": 135260883.0, + "step": 8151 + }, + { + "epoch": 0.6722190154201368, + "grad_norm": 2.1429858207702637, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7488511800765991, + "num_tokens": 135284762.0, + "step": 8152 + }, + { + "epoch": 0.6723014760451884, + "grad_norm": 1.8761290311813354, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.744385838508606, + "num_tokens": 135316744.0, + "step": 8153 + }, + { + "epoch": 0.67238393667024, + "grad_norm": 2.261960744857788, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.6995136737823486, + "num_tokens": 135341314.0, + "step": 8154 + }, + { + "epoch": 0.6724663972952915, + "grad_norm": 1.9644136428833008, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7303093671798706, + "num_tokens": 135371439.0, + "step": 8155 + }, + { + "epoch": 0.672548857920343, + "grad_norm": 2.1728713512420654, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.714706301689148, + "num_tokens": 135396736.0, + "step": 8156 + }, + { + "epoch": 0.6726313185453946, + "grad_norm": 2.0218114852905273, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7403146028518677, + "num_tokens": 135427321.0, + "step": 8157 + }, + { + "epoch": 0.6727137791704461, + "grad_norm": 2.119990110397339, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.721428632736206, + "num_tokens": 135453367.0, + "step": 8158 + }, + { + "epoch": 0.6727962397954976, + "grad_norm": 1.931156039237976, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.773217499256134, + "num_tokens": 135482629.0, + "step": 8159 + }, + { + "epoch": 0.6728787004205492, + "grad_norm": 1.8741014003753662, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7801083326339722, + "num_tokens": 135511568.0, + "step": 8160 + }, + { + "epoch": 0.6729611610456008, + "grad_norm": 1.8594251871109009, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7537140250205994, + "num_tokens": 135542540.0, + "step": 8161 + }, + { + "epoch": 0.6730436216706522, + "grad_norm": 1.9341025352478027, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7436884641647339, + "num_tokens": 135570399.0, + "step": 8162 + }, + { + "epoch": 0.6731260822957038, + "grad_norm": 2.2088892459869385, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7485357522964478, + "num_tokens": 135595100.0, + "step": 8163 + }, + { + "epoch": 0.6732085429207554, + "grad_norm": 1.8601067066192627, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7623454928398132, + "num_tokens": 135629242.0, + "step": 8164 + }, + { + "epoch": 0.6732910035458068, + "grad_norm": 1.9835675954818726, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7237851619720459, + "num_tokens": 135660470.0, + "step": 8165 + }, + { + "epoch": 0.6733734641708584, + "grad_norm": 1.8025771379470825, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7536627054214478, + "num_tokens": 135694029.0, + "step": 8166 + }, + { + "epoch": 0.67345592479591, + "grad_norm": 2.089655637741089, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7247291803359985, + "num_tokens": 135722177.0, + "step": 8167 + }, + { + "epoch": 0.6735383854209614, + "grad_norm": 2.067323684692383, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7518482804298401, + "num_tokens": 135749276.0, + "step": 8168 + }, + { + "epoch": 0.673620846046013, + "grad_norm": 1.9410258531570435, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7197508215904236, + "num_tokens": 135782585.0, + "step": 8169 + }, + { + "epoch": 0.6737033066710646, + "grad_norm": 2.089243173599243, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.696218729019165, + "num_tokens": 135811716.0, + "step": 8170 + }, + { + "epoch": 0.6737857672961161, + "grad_norm": 1.601475477218628, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6910350322723389, + "num_tokens": 135857666.0, + "step": 8171 + }, + { + "epoch": 0.6738682279211676, + "grad_norm": 2.3221261501312256, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7569015026092529, + "num_tokens": 135878640.0, + "step": 8172 + }, + { + "epoch": 0.6739506885462192, + "grad_norm": 2.0947225093841553, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7538782358169556, + "num_tokens": 135904425.0, + "step": 8173 + }, + { + "epoch": 0.6740331491712708, + "grad_norm": 1.6378768682479858, + "learning_rate": 1e-06, + "loss": 0.57, + "mean_token_accuracy": 0.8156523704528809, + "num_tokens": 135942254.0, + "step": 8174 + }, + { + "epoch": 0.6741156097963222, + "grad_norm": 2.261610984802246, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7256460189819336, + "num_tokens": 135964705.0, + "step": 8175 + }, + { + "epoch": 0.6741980704213738, + "grad_norm": 1.8717559576034546, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7291768789291382, + "num_tokens": 135995489.0, + "step": 8176 + }, + { + "epoch": 0.6742805310464254, + "grad_norm": 2.006056308746338, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7322103977203369, + "num_tokens": 136023947.0, + "step": 8177 + }, + { + "epoch": 0.6743629916714768, + "grad_norm": 1.970447301864624, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7626851797103882, + "num_tokens": 136054023.0, + "step": 8178 + }, + { + "epoch": 0.6744454522965284, + "grad_norm": 1.777991533279419, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7613878846168518, + "num_tokens": 136087873.0, + "step": 8179 + }, + { + "epoch": 0.67452791292158, + "grad_norm": 1.7963626384735107, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7521106004714966, + "num_tokens": 136121061.0, + "step": 8180 + }, + { + "epoch": 0.6746103735466314, + "grad_norm": 2.000704765319824, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7137372493743896, + "num_tokens": 136148774.0, + "step": 8181 + }, + { + "epoch": 0.674692834171683, + "grad_norm": 1.84888756275177, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7680795788764954, + "num_tokens": 136180146.0, + "step": 8182 + }, + { + "epoch": 0.6747752947967346, + "grad_norm": 2.043022871017456, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7458044290542603, + "num_tokens": 136207859.0, + "step": 8183 + }, + { + "epoch": 0.674857755421786, + "grad_norm": 1.925398588180542, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7651429176330566, + "num_tokens": 136239300.0, + "step": 8184 + }, + { + "epoch": 0.6749402160468376, + "grad_norm": 1.7614136934280396, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7450028657913208, + "num_tokens": 136277266.0, + "step": 8185 + }, + { + "epoch": 0.6750226766718892, + "grad_norm": 2.0562989711761475, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7044195532798767, + "num_tokens": 136307596.0, + "step": 8186 + }, + { + "epoch": 0.6751051372969407, + "grad_norm": 1.9598602056503296, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7745522856712341, + "num_tokens": 136336511.0, + "step": 8187 + }, + { + "epoch": 0.6751875979219922, + "grad_norm": 1.9053614139556885, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.762962281703949, + "num_tokens": 136366425.0, + "step": 8188 + }, + { + "epoch": 0.6752700585470438, + "grad_norm": 2.02496075630188, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7213797569274902, + "num_tokens": 136394717.0, + "step": 8189 + }, + { + "epoch": 0.6753525191720954, + "grad_norm": 2.0367543697357178, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7102628350257874, + "num_tokens": 136423917.0, + "step": 8190 + }, + { + "epoch": 0.6754349797971468, + "grad_norm": 1.9047306776046753, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7609275579452515, + "num_tokens": 136453922.0, + "step": 8191 + }, + { + "epoch": 0.6755174404221984, + "grad_norm": 2.005051612854004, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7495419979095459, + "num_tokens": 136481879.0, + "step": 8192 + }, + { + "epoch": 0.67559990104725, + "grad_norm": 1.772012710571289, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7307446002960205, + "num_tokens": 136517504.0, + "step": 8193 + }, + { + "epoch": 0.6756823616723014, + "grad_norm": 2.052386999130249, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7349658608436584, + "num_tokens": 136545778.0, + "step": 8194 + }, + { + "epoch": 0.675764822297353, + "grad_norm": 2.09126615524292, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7253177165985107, + "num_tokens": 136573063.0, + "step": 8195 + }, + { + "epoch": 0.6758472829224046, + "grad_norm": 2.1067967414855957, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.7501818537712097, + "num_tokens": 136603551.0, + "step": 8196 + }, + { + "epoch": 0.675929743547456, + "grad_norm": 2.0542099475860596, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7861591577529907, + "num_tokens": 136630587.0, + "step": 8197 + }, + { + "epoch": 0.6760122041725076, + "grad_norm": 2.2536604404449463, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6999226808547974, + "num_tokens": 136658153.0, + "step": 8198 + }, + { + "epoch": 0.6760946647975592, + "grad_norm": 1.8332924842834473, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7074064612388611, + "num_tokens": 136694147.0, + "step": 8199 + }, + { + "epoch": 0.6761771254226107, + "grad_norm": 1.7464008331298828, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7726865410804749, + "num_tokens": 136730512.0, + "step": 8200 + }, + { + "epoch": 0.6762595860476622, + "grad_norm": 1.9559483528137207, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7503294348716736, + "num_tokens": 136760015.0, + "step": 8201 + }, + { + "epoch": 0.6763420466727138, + "grad_norm": 1.9124712944030762, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7231918573379517, + "num_tokens": 136793526.0, + "step": 8202 + }, + { + "epoch": 0.6764245072977653, + "grad_norm": 1.8584513664245605, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7470813393592834, + "num_tokens": 136826281.0, + "step": 8203 + }, + { + "epoch": 0.6765069679228168, + "grad_norm": 1.8644356727600098, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7851593494415283, + "num_tokens": 136857472.0, + "step": 8204 + }, + { + "epoch": 0.6765894285478684, + "grad_norm": 1.917470932006836, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7332738637924194, + "num_tokens": 136888610.0, + "step": 8205 + }, + { + "epoch": 0.67667188917292, + "grad_norm": 1.938401699066162, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.750769317150116, + "num_tokens": 136920978.0, + "step": 8206 + }, + { + "epoch": 0.6767543497979714, + "grad_norm": 2.090411424636841, + "learning_rate": 1e-06, + "loss": 0.6223, + "mean_token_accuracy": 0.7957587242126465, + "num_tokens": 136953271.0, + "step": 8207 + }, + { + "epoch": 0.676836810423023, + "grad_norm": 1.7383641004562378, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7637630701065063, + "num_tokens": 136987079.0, + "step": 8208 + }, + { + "epoch": 0.6769192710480746, + "grad_norm": 1.9909138679504395, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7178284525871277, + "num_tokens": 137014774.0, + "step": 8209 + }, + { + "epoch": 0.677001731673126, + "grad_norm": 1.900229573249817, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.7889163494110107, + "num_tokens": 137045159.0, + "step": 8210 + }, + { + "epoch": 0.6770841922981776, + "grad_norm": 2.0772361755371094, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7627723813056946, + "num_tokens": 137071290.0, + "step": 8211 + }, + { + "epoch": 0.6771666529232292, + "grad_norm": 2.0896928310394287, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7230637073516846, + "num_tokens": 137097886.0, + "step": 8212 + }, + { + "epoch": 0.6772491135482807, + "grad_norm": 2.040353775024414, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7503679990768433, + "num_tokens": 137124712.0, + "step": 8213 + }, + { + "epoch": 0.6773315741733322, + "grad_norm": 2.0188143253326416, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7336545586585999, + "num_tokens": 137154271.0, + "step": 8214 + }, + { + "epoch": 0.6774140347983838, + "grad_norm": 2.0123116970062256, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7717360854148865, + "num_tokens": 137180520.0, + "step": 8215 + }, + { + "epoch": 0.6774964954234353, + "grad_norm": 2.033292770385742, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7547026872634888, + "num_tokens": 137208586.0, + "step": 8216 + }, + { + "epoch": 0.6775789560484868, + "grad_norm": 1.8413337469100952, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7436807751655579, + "num_tokens": 137244187.0, + "step": 8217 + }, + { + "epoch": 0.6776614166735384, + "grad_norm": 2.1830928325653076, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7316725254058838, + "num_tokens": 137268209.0, + "step": 8218 + }, + { + "epoch": 0.6777438772985899, + "grad_norm": 1.9159760475158691, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7574114799499512, + "num_tokens": 137297192.0, + "step": 8219 + }, + { + "epoch": 0.6778263379236414, + "grad_norm": 2.2686338424682617, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.723139762878418, + "num_tokens": 137322103.0, + "step": 8220 + }, + { + "epoch": 0.677908798548693, + "grad_norm": 1.8252462148666382, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7414028644561768, + "num_tokens": 137354089.0, + "step": 8221 + }, + { + "epoch": 0.6779912591737446, + "grad_norm": 2.0893189907073975, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7402054667472839, + "num_tokens": 137380851.0, + "step": 8222 + }, + { + "epoch": 0.678073719798796, + "grad_norm": 2.033173084259033, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7529364824295044, + "num_tokens": 137408333.0, + "step": 8223 + }, + { + "epoch": 0.6781561804238476, + "grad_norm": 2.105120897293091, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.703831672668457, + "num_tokens": 137438335.0, + "step": 8224 + }, + { + "epoch": 0.6782386410488992, + "grad_norm": 1.8741936683654785, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7577358484268188, + "num_tokens": 137468723.0, + "step": 8225 + }, + { + "epoch": 0.6783211016739507, + "grad_norm": 1.9455199241638184, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.754867434501648, + "num_tokens": 137495791.0, + "step": 8226 + }, + { + "epoch": 0.6784035622990022, + "grad_norm": 1.8422893285751343, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7554130554199219, + "num_tokens": 137527088.0, + "step": 8227 + }, + { + "epoch": 0.6784860229240538, + "grad_norm": 2.0793519020080566, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.7860791683197021, + "num_tokens": 137549821.0, + "step": 8228 + }, + { + "epoch": 0.6785684835491053, + "grad_norm": 1.9082046747207642, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7357937693595886, + "num_tokens": 137583736.0, + "step": 8229 + }, + { + "epoch": 0.6786509441741568, + "grad_norm": 1.8213709592819214, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7643822431564331, + "num_tokens": 137614280.0, + "step": 8230 + }, + { + "epoch": 0.6787334047992084, + "grad_norm": 1.660056233406067, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7510290145874023, + "num_tokens": 137656644.0, + "step": 8231 + }, + { + "epoch": 0.6788158654242599, + "grad_norm": 2.1009936332702637, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.749627411365509, + "num_tokens": 137685118.0, + "step": 8232 + }, + { + "epoch": 0.6788983260493114, + "grad_norm": 1.79847252368927, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7348887324333191, + "num_tokens": 137716692.0, + "step": 8233 + }, + { + "epoch": 0.678980786674363, + "grad_norm": 1.966391921043396, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7473217844963074, + "num_tokens": 137748029.0, + "step": 8234 + }, + { + "epoch": 0.6790632472994145, + "grad_norm": 2.151700973510742, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7430235147476196, + "num_tokens": 137773825.0, + "step": 8235 + }, + { + "epoch": 0.679145707924466, + "grad_norm": 2.2040750980377197, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7368361353874207, + "num_tokens": 137797939.0, + "step": 8236 + }, + { + "epoch": 0.6792281685495176, + "grad_norm": 2.0531153678894043, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7603775262832642, + "num_tokens": 137828953.0, + "step": 8237 + }, + { + "epoch": 0.6793106291745692, + "grad_norm": 1.7710644006729126, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.7962480783462524, + "num_tokens": 137864649.0, + "step": 8238 + }, + { + "epoch": 0.6793930897996207, + "grad_norm": 1.899047613143921, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7331720590591431, + "num_tokens": 137896935.0, + "step": 8239 + }, + { + "epoch": 0.6794755504246722, + "grad_norm": 2.458958864212036, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.737824022769928, + "num_tokens": 137921333.0, + "step": 8240 + }, + { + "epoch": 0.6795580110497238, + "grad_norm": 1.8210002183914185, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.761978030204773, + "num_tokens": 137955847.0, + "step": 8241 + }, + { + "epoch": 0.6796404716747753, + "grad_norm": 1.967825174331665, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7627424597740173, + "num_tokens": 137983241.0, + "step": 8242 + }, + { + "epoch": 0.6797229322998268, + "grad_norm": 2.0656826496124268, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7523284554481506, + "num_tokens": 138008883.0, + "step": 8243 + }, + { + "epoch": 0.6798053929248784, + "grad_norm": 1.8916152715682983, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7693520784378052, + "num_tokens": 138038125.0, + "step": 8244 + }, + { + "epoch": 0.6798878535499299, + "grad_norm": 2.0617287158966064, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7228500843048096, + "num_tokens": 138065270.0, + "step": 8245 + }, + { + "epoch": 0.6799703141749814, + "grad_norm": 2.004040241241455, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7561793327331543, + "num_tokens": 138093097.0, + "step": 8246 + }, + { + "epoch": 0.680052774800033, + "grad_norm": 1.797019124031067, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.7391202449798584, + "num_tokens": 138127217.0, + "step": 8247 + }, + { + "epoch": 0.6801352354250845, + "grad_norm": 1.6582322120666504, + "learning_rate": 1e-06, + "loss": 0.6047, + "mean_token_accuracy": 0.7996187210083008, + "num_tokens": 138163664.0, + "step": 8248 + }, + { + "epoch": 0.680217696050136, + "grad_norm": 2.044053792953491, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.754002571105957, + "num_tokens": 138191575.0, + "step": 8249 + }, + { + "epoch": 0.6803001566751876, + "grad_norm": 2.0930211544036865, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7405787706375122, + "num_tokens": 138217799.0, + "step": 8250 + }, + { + "epoch": 0.6803826173002392, + "grad_norm": 1.8566113710403442, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7249401807785034, + "num_tokens": 138253378.0, + "step": 8251 + }, + { + "epoch": 0.6804650779252907, + "grad_norm": 2.16135835647583, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7218712568283081, + "num_tokens": 138279130.0, + "step": 8252 + }, + { + "epoch": 0.6805475385503422, + "grad_norm": 2.0044984817504883, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7664123177528381, + "num_tokens": 138308437.0, + "step": 8253 + }, + { + "epoch": 0.6806299991753938, + "grad_norm": 1.9480746984481812, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7122487425804138, + "num_tokens": 138339866.0, + "step": 8254 + }, + { + "epoch": 0.6807124598004453, + "grad_norm": 1.813564658164978, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7773119211196899, + "num_tokens": 138369487.0, + "step": 8255 + }, + { + "epoch": 0.6807949204254968, + "grad_norm": 1.9588208198547363, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7401254177093506, + "num_tokens": 138399451.0, + "step": 8256 + }, + { + "epoch": 0.6808773810505484, + "grad_norm": 1.9722875356674194, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7754526734352112, + "num_tokens": 138429045.0, + "step": 8257 + }, + { + "epoch": 0.6809598416755999, + "grad_norm": 2.1528236865997314, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7562105655670166, + "num_tokens": 138454308.0, + "step": 8258 + }, + { + "epoch": 0.6810423023006514, + "grad_norm": 2.143817901611328, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7349580526351929, + "num_tokens": 138480400.0, + "step": 8259 + }, + { + "epoch": 0.681124762925703, + "grad_norm": 1.9168052673339844, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7592640519142151, + "num_tokens": 138509527.0, + "step": 8260 + }, + { + "epoch": 0.6812072235507545, + "grad_norm": 2.2762339115142822, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7230196595191956, + "num_tokens": 138534578.0, + "step": 8261 + }, + { + "epoch": 0.681289684175806, + "grad_norm": 2.2467424869537354, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.741631269454956, + "num_tokens": 138559654.0, + "step": 8262 + }, + { + "epoch": 0.6813721448008576, + "grad_norm": 1.7387531995773315, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7433131337165833, + "num_tokens": 138593828.0, + "step": 8263 + }, + { + "epoch": 0.6814546054259091, + "grad_norm": 1.993457555770874, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7296208143234253, + "num_tokens": 138623106.0, + "step": 8264 + }, + { + "epoch": 0.6815370660509606, + "grad_norm": 1.8880375623703003, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7627906799316406, + "num_tokens": 138654232.0, + "step": 8265 + }, + { + "epoch": 0.6816195266760122, + "grad_norm": 1.9151393175125122, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7503238916397095, + "num_tokens": 138685964.0, + "step": 8266 + }, + { + "epoch": 0.6817019873010638, + "grad_norm": 1.9012017250061035, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7383792400360107, + "num_tokens": 138718452.0, + "step": 8267 + }, + { + "epoch": 0.6817844479261153, + "grad_norm": 1.998339056968689, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.739780843257904, + "num_tokens": 138745746.0, + "step": 8268 + }, + { + "epoch": 0.6818669085511668, + "grad_norm": 1.8641035556793213, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7483896017074585, + "num_tokens": 138777246.0, + "step": 8269 + }, + { + "epoch": 0.6819493691762184, + "grad_norm": 1.9550079107284546, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7482324838638306, + "num_tokens": 138807534.0, + "step": 8270 + }, + { + "epoch": 0.6820318298012699, + "grad_norm": 2.173884153366089, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.6954680681228638, + "num_tokens": 138833624.0, + "step": 8271 + }, + { + "epoch": 0.6821142904263214, + "grad_norm": 2.1885569095611572, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7209892272949219, + "num_tokens": 138862052.0, + "step": 8272 + }, + { + "epoch": 0.682196751051373, + "grad_norm": 2.0757360458374023, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7323864698410034, + "num_tokens": 138888697.0, + "step": 8273 + }, + { + "epoch": 0.6822792116764245, + "grad_norm": 1.9597482681274414, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7389754056930542, + "num_tokens": 138917053.0, + "step": 8274 + }, + { + "epoch": 0.682361672301476, + "grad_norm": 1.9516671895980835, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7650256156921387, + "num_tokens": 138945836.0, + "step": 8275 + }, + { + "epoch": 0.6824441329265276, + "grad_norm": 2.181164503097534, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.719386100769043, + "num_tokens": 138973018.0, + "step": 8276 + }, + { + "epoch": 0.6825265935515791, + "grad_norm": 2.103849172592163, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7369987964630127, + "num_tokens": 139000413.0, + "step": 8277 + }, + { + "epoch": 0.6826090541766306, + "grad_norm": 2.0252890586853027, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.719118595123291, + "num_tokens": 139028819.0, + "step": 8278 + }, + { + "epoch": 0.6826915148016822, + "grad_norm": 1.92320716381073, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7343998551368713, + "num_tokens": 139061032.0, + "step": 8279 + }, + { + "epoch": 0.6827739754267337, + "grad_norm": 1.8611997365951538, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7801197171211243, + "num_tokens": 139093289.0, + "step": 8280 + }, + { + "epoch": 0.6828564360517853, + "grad_norm": 1.7536388635635376, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7707712650299072, + "num_tokens": 139126397.0, + "step": 8281 + }, + { + "epoch": 0.6829388966768368, + "grad_norm": 1.906069278717041, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7316755652427673, + "num_tokens": 139156847.0, + "step": 8282 + }, + { + "epoch": 0.6830213573018884, + "grad_norm": 1.8370327949523926, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7706770896911621, + "num_tokens": 139187220.0, + "step": 8283 + }, + { + "epoch": 0.6831038179269399, + "grad_norm": 2.0105886459350586, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7381479740142822, + "num_tokens": 139214921.0, + "step": 8284 + }, + { + "epoch": 0.6831862785519914, + "grad_norm": 2.045332670211792, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7297894358634949, + "num_tokens": 139242635.0, + "step": 8285 + }, + { + "epoch": 0.683268739177043, + "grad_norm": 1.82656991481781, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.7868523597717285, + "num_tokens": 139274996.0, + "step": 8286 + }, + { + "epoch": 0.6833511998020945, + "grad_norm": 1.801379680633545, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7641255259513855, + "num_tokens": 139308881.0, + "step": 8287 + }, + { + "epoch": 0.683433660427146, + "grad_norm": 2.509446382522583, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7578535079956055, + "num_tokens": 139327488.0, + "step": 8288 + }, + { + "epoch": 0.6835161210521976, + "grad_norm": 1.9595407247543335, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7370487451553345, + "num_tokens": 139359028.0, + "step": 8289 + }, + { + "epoch": 0.6835985816772491, + "grad_norm": 1.916771411895752, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7739702463150024, + "num_tokens": 139387998.0, + "step": 8290 + }, + { + "epoch": 0.6836810423023006, + "grad_norm": 1.974684476852417, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7340805530548096, + "num_tokens": 139413835.0, + "step": 8291 + }, + { + "epoch": 0.6837635029273522, + "grad_norm": 1.8252731561660767, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7213084697723389, + "num_tokens": 139447246.0, + "step": 8292 + }, + { + "epoch": 0.6838459635524037, + "grad_norm": 1.9556175470352173, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7115377187728882, + "num_tokens": 139479836.0, + "step": 8293 + }, + { + "epoch": 0.6839284241774553, + "grad_norm": 1.6594772338867188, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7935675978660583, + "num_tokens": 139518669.0, + "step": 8294 + }, + { + "epoch": 0.6840108848025068, + "grad_norm": 1.944485068321228, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7230737209320068, + "num_tokens": 139551233.0, + "step": 8295 + }, + { + "epoch": 0.6840933454275583, + "grad_norm": 1.6966989040374756, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7625483274459839, + "num_tokens": 139588167.0, + "step": 8296 + }, + { + "epoch": 0.6841758060526099, + "grad_norm": 1.9268070459365845, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7233629822731018, + "num_tokens": 139620052.0, + "step": 8297 + }, + { + "epoch": 0.6842582666776614, + "grad_norm": 2.1091675758361816, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7213751077651978, + "num_tokens": 139647257.0, + "step": 8298 + }, + { + "epoch": 0.684340727302713, + "grad_norm": 2.023695468902588, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7164469957351685, + "num_tokens": 139677103.0, + "step": 8299 + }, + { + "epoch": 0.6844231879277645, + "grad_norm": 1.9796435832977295, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7786048054695129, + "num_tokens": 139704523.0, + "step": 8300 + }, + { + "epoch": 0.684505648552816, + "grad_norm": 2.0474488735198975, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7760226726531982, + "num_tokens": 139731017.0, + "step": 8301 + }, + { + "epoch": 0.6845881091778676, + "grad_norm": 1.9133723974227905, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.72890305519104, + "num_tokens": 139761568.0, + "step": 8302 + }, + { + "epoch": 0.6846705698029191, + "grad_norm": 2.001525402069092, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7613388299942017, + "num_tokens": 139787826.0, + "step": 8303 + }, + { + "epoch": 0.6847530304279706, + "grad_norm": 1.7909457683563232, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7562306523323059, + "num_tokens": 139823175.0, + "step": 8304 + }, + { + "epoch": 0.6848354910530222, + "grad_norm": 1.7798316478729248, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7413198351860046, + "num_tokens": 139858329.0, + "step": 8305 + }, + { + "epoch": 0.6849179516780737, + "grad_norm": 1.8775733709335327, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7760435342788696, + "num_tokens": 139887520.0, + "step": 8306 + }, + { + "epoch": 0.6850004123031253, + "grad_norm": 1.9284523725509644, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7537218332290649, + "num_tokens": 139917957.0, + "step": 8307 + }, + { + "epoch": 0.6850828729281768, + "grad_norm": 1.8858122825622559, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7300816774368286, + "num_tokens": 139952760.0, + "step": 8308 + }, + { + "epoch": 0.6851653335532283, + "grad_norm": 2.089667797088623, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7480690479278564, + "num_tokens": 139979057.0, + "step": 8309 + }, + { + "epoch": 0.6852477941782799, + "grad_norm": 1.9054954051971436, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7252036333084106, + "num_tokens": 140010232.0, + "step": 8310 + }, + { + "epoch": 0.6853302548033314, + "grad_norm": 1.7932790517807007, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7329081296920776, + "num_tokens": 140043051.0, + "step": 8311 + }, + { + "epoch": 0.685412715428383, + "grad_norm": 2.1424150466918945, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7599204182624817, + "num_tokens": 140067830.0, + "step": 8312 + }, + { + "epoch": 0.6854951760534345, + "grad_norm": 2.1784708499908447, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7329378128051758, + "num_tokens": 140094292.0, + "step": 8313 + }, + { + "epoch": 0.685577636678486, + "grad_norm": 1.597778558731079, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7864530086517334, + "num_tokens": 140133771.0, + "step": 8314 + }, + { + "epoch": 0.6856600973035376, + "grad_norm": 2.1647610664367676, + "learning_rate": 1e-06, + "loss": 1.0694, + "mean_token_accuracy": 0.6804772019386292, + "num_tokens": 140162721.0, + "step": 8315 + }, + { + "epoch": 0.6857425579285891, + "grad_norm": 1.9929131269454956, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7404763698577881, + "num_tokens": 140195047.0, + "step": 8316 + }, + { + "epoch": 0.6858250185536406, + "grad_norm": 2.0611002445220947, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7747578024864197, + "num_tokens": 140223125.0, + "step": 8317 + }, + { + "epoch": 0.6859074791786922, + "grad_norm": 1.8174595832824707, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7588196992874146, + "num_tokens": 140256725.0, + "step": 8318 + }, + { + "epoch": 0.6859899398037437, + "grad_norm": 2.1939215660095215, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7270689010620117, + "num_tokens": 140282647.0, + "step": 8319 + }, + { + "epoch": 0.6860724004287952, + "grad_norm": 1.8071316480636597, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7793104648590088, + "num_tokens": 140318272.0, + "step": 8320 + }, + { + "epoch": 0.6861548610538468, + "grad_norm": 2.1089823246002197, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7887560725212097, + "num_tokens": 140343721.0, + "step": 8321 + }, + { + "epoch": 0.6862373216788983, + "grad_norm": 2.0183234214782715, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7304688692092896, + "num_tokens": 140370949.0, + "step": 8322 + }, + { + "epoch": 0.6863197823039499, + "grad_norm": 2.081559896469116, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7504677772521973, + "num_tokens": 140397999.0, + "step": 8323 + }, + { + "epoch": 0.6864022429290014, + "grad_norm": 2.0893735885620117, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7452811002731323, + "num_tokens": 140426785.0, + "step": 8324 + }, + { + "epoch": 0.6864847035540529, + "grad_norm": 1.86050546169281, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7378175258636475, + "num_tokens": 140461868.0, + "step": 8325 + }, + { + "epoch": 0.6865671641791045, + "grad_norm": 2.3679275512695312, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7167280912399292, + "num_tokens": 140484093.0, + "step": 8326 + }, + { + "epoch": 0.686649624804156, + "grad_norm": 1.9702346324920654, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7265790104866028, + "num_tokens": 140515067.0, + "step": 8327 + }, + { + "epoch": 0.6867320854292076, + "grad_norm": 1.8906934261322021, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7450594902038574, + "num_tokens": 140544009.0, + "step": 8328 + }, + { + "epoch": 0.6868145460542591, + "grad_norm": 1.8150668144226074, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7066960334777832, + "num_tokens": 140581274.0, + "step": 8329 + }, + { + "epoch": 0.6868970066793106, + "grad_norm": 1.8618794679641724, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7639358043670654, + "num_tokens": 140608352.0, + "step": 8330 + }, + { + "epoch": 0.6869794673043622, + "grad_norm": 2.28851318359375, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7553834915161133, + "num_tokens": 140632960.0, + "step": 8331 + }, + { + "epoch": 0.6870619279294137, + "grad_norm": 1.9456347227096558, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7222371101379395, + "num_tokens": 140662564.0, + "step": 8332 + }, + { + "epoch": 0.6871443885544652, + "grad_norm": 1.9456592798233032, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7422661185264587, + "num_tokens": 140693708.0, + "step": 8333 + }, + { + "epoch": 0.6872268491795168, + "grad_norm": 1.9951367378234863, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7280197143554688, + "num_tokens": 140725370.0, + "step": 8334 + }, + { + "epoch": 0.6873093098045683, + "grad_norm": 1.6421668529510498, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7791638970375061, + "num_tokens": 140764856.0, + "step": 8335 + }, + { + "epoch": 0.6873917704296199, + "grad_norm": 1.7995325326919556, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7433397769927979, + "num_tokens": 140799470.0, + "step": 8336 + }, + { + "epoch": 0.6874742310546714, + "grad_norm": 1.8679659366607666, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7331898808479309, + "num_tokens": 140830100.0, + "step": 8337 + }, + { + "epoch": 0.6875566916797229, + "grad_norm": 1.8924593925476074, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7294385433197021, + "num_tokens": 140862676.0, + "step": 8338 + }, + { + "epoch": 0.6876391523047745, + "grad_norm": 1.7910507917404175, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7166541814804077, + "num_tokens": 140898721.0, + "step": 8339 + }, + { + "epoch": 0.687721612929826, + "grad_norm": 2.2935338020324707, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7185210585594177, + "num_tokens": 140922790.0, + "step": 8340 + }, + { + "epoch": 0.6878040735548775, + "grad_norm": 1.817862868309021, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7529442310333252, + "num_tokens": 140954693.0, + "step": 8341 + }, + { + "epoch": 0.6878865341799291, + "grad_norm": 2.023637533187866, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7280480861663818, + "num_tokens": 140984654.0, + "step": 8342 + }, + { + "epoch": 0.6879689948049806, + "grad_norm": 2.065406322479248, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7535473108291626, + "num_tokens": 141011983.0, + "step": 8343 + }, + { + "epoch": 0.6880514554300322, + "grad_norm": 1.7219970226287842, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7472419738769531, + "num_tokens": 141052633.0, + "step": 8344 + }, + { + "epoch": 0.6881339160550837, + "grad_norm": 1.9118545055389404, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7844641208648682, + "num_tokens": 141084459.0, + "step": 8345 + }, + { + "epoch": 0.6882163766801352, + "grad_norm": 2.0229721069335938, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7476224899291992, + "num_tokens": 141113150.0, + "step": 8346 + }, + { + "epoch": 0.6882988373051868, + "grad_norm": 1.8965803384780884, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.707922101020813, + "num_tokens": 141148348.0, + "step": 8347 + }, + { + "epoch": 0.6883812979302383, + "grad_norm": 1.84409499168396, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7315835952758789, + "num_tokens": 141181922.0, + "step": 8348 + }, + { + "epoch": 0.6884637585552899, + "grad_norm": 2.1244547367095947, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7423461079597473, + "num_tokens": 141210221.0, + "step": 8349 + }, + { + "epoch": 0.6885462191803414, + "grad_norm": 1.9352678060531616, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.746678352355957, + "num_tokens": 141244328.0, + "step": 8350 + }, + { + "epoch": 0.6886286798053929, + "grad_norm": 2.1409924030303955, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7577730417251587, + "num_tokens": 141270335.0, + "step": 8351 + }, + { + "epoch": 0.6887111404304445, + "grad_norm": 1.8998078107833862, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7750256657600403, + "num_tokens": 141299652.0, + "step": 8352 + }, + { + "epoch": 0.688793601055496, + "grad_norm": 2.1134941577911377, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.740435004234314, + "num_tokens": 141326022.0, + "step": 8353 + }, + { + "epoch": 0.6888760616805475, + "grad_norm": 2.043555498123169, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7231558561325073, + "num_tokens": 141353828.0, + "step": 8354 + }, + { + "epoch": 0.6889585223055991, + "grad_norm": 2.3220245838165283, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7848995923995972, + "num_tokens": 141384259.0, + "step": 8355 + }, + { + "epoch": 0.6890409829306506, + "grad_norm": 2.060473918914795, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7391418218612671, + "num_tokens": 141413783.0, + "step": 8356 + }, + { + "epoch": 0.6891234435557021, + "grad_norm": 1.9587451219558716, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7836847901344299, + "num_tokens": 141444774.0, + "step": 8357 + }, + { + "epoch": 0.6892059041807537, + "grad_norm": 2.2425928115844727, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7042361497879028, + "num_tokens": 141470878.0, + "step": 8358 + }, + { + "epoch": 0.6892883648058052, + "grad_norm": 1.821442723274231, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.750247597694397, + "num_tokens": 141503404.0, + "step": 8359 + }, + { + "epoch": 0.6893708254308568, + "grad_norm": 1.9239346981048584, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7645837068557739, + "num_tokens": 141533909.0, + "step": 8360 + }, + { + "epoch": 0.6894532860559083, + "grad_norm": 1.9924312829971313, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.762738823890686, + "num_tokens": 141562013.0, + "step": 8361 + }, + { + "epoch": 0.6895357466809598, + "grad_norm": 2.097604990005493, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7373000383377075, + "num_tokens": 141588866.0, + "step": 8362 + }, + { + "epoch": 0.6896182073060114, + "grad_norm": 1.8569839000701904, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.7857080698013306, + "num_tokens": 141622241.0, + "step": 8363 + }, + { + "epoch": 0.6897006679310629, + "grad_norm": 1.8578922748565674, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7484426498413086, + "num_tokens": 141652943.0, + "step": 8364 + }, + { + "epoch": 0.6897831285561145, + "grad_norm": 1.8839386701583862, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7656865119934082, + "num_tokens": 141685694.0, + "step": 8365 + }, + { + "epoch": 0.689865589181166, + "grad_norm": 2.1160483360290527, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.7876827716827393, + "num_tokens": 141706835.0, + "step": 8366 + }, + { + "epoch": 0.6899480498062175, + "grad_norm": 2.004652976989746, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7599764466285706, + "num_tokens": 141736650.0, + "step": 8367 + }, + { + "epoch": 0.6900305104312691, + "grad_norm": 2.123260498046875, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7505532503128052, + "num_tokens": 141763653.0, + "step": 8368 + }, + { + "epoch": 0.6901129710563206, + "grad_norm": 1.8350719213485718, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7315807938575745, + "num_tokens": 141798925.0, + "step": 8369 + }, + { + "epoch": 0.6901954316813721, + "grad_norm": 2.119044065475464, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7647550106048584, + "num_tokens": 141824212.0, + "step": 8370 + }, + { + "epoch": 0.6902778923064237, + "grad_norm": 1.9598854780197144, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.7470462918281555, + "num_tokens": 141855545.0, + "step": 8371 + }, + { + "epoch": 0.6903603529314752, + "grad_norm": 1.8686169385910034, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7649504542350769, + "num_tokens": 141887164.0, + "step": 8372 + }, + { + "epoch": 0.6904428135565268, + "grad_norm": 2.2996678352355957, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7668924927711487, + "num_tokens": 141909200.0, + "step": 8373 + }, + { + "epoch": 0.6905252741815783, + "grad_norm": 1.9357746839523315, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7451382875442505, + "num_tokens": 141937371.0, + "step": 8374 + }, + { + "epoch": 0.6906077348066298, + "grad_norm": 1.991585612297058, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7467522025108337, + "num_tokens": 141963884.0, + "step": 8375 + }, + { + "epoch": 0.6906901954316814, + "grad_norm": 2.203789710998535, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7487192153930664, + "num_tokens": 141987403.0, + "step": 8376 + }, + { + "epoch": 0.6907726560567329, + "grad_norm": 1.753937005996704, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7440465092658997, + "num_tokens": 142021695.0, + "step": 8377 + }, + { + "epoch": 0.6908551166817845, + "grad_norm": 1.7686940431594849, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.757064700126648, + "num_tokens": 142056745.0, + "step": 8378 + }, + { + "epoch": 0.690937577306836, + "grad_norm": 1.936957597732544, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7424876093864441, + "num_tokens": 142088370.0, + "step": 8379 + }, + { + "epoch": 0.6910200379318875, + "grad_norm": 2.4707822799682617, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7463315725326538, + "num_tokens": 142109202.0, + "step": 8380 + }, + { + "epoch": 0.6911024985569391, + "grad_norm": 1.8410338163375854, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7305478453636169, + "num_tokens": 142145294.0, + "step": 8381 + }, + { + "epoch": 0.6911849591819906, + "grad_norm": 1.8904858827590942, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7510290145874023, + "num_tokens": 142175164.0, + "step": 8382 + }, + { + "epoch": 0.6912674198070421, + "grad_norm": 1.996535062789917, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7305337190628052, + "num_tokens": 142207482.0, + "step": 8383 + }, + { + "epoch": 0.6913498804320937, + "grad_norm": 1.959259271621704, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7463303804397583, + "num_tokens": 142237536.0, + "step": 8384 + }, + { + "epoch": 0.6914323410571452, + "grad_norm": 1.9279216527938843, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.7867697477340698, + "num_tokens": 142267059.0, + "step": 8385 + }, + { + "epoch": 0.6915148016821967, + "grad_norm": 2.13737416267395, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7416017055511475, + "num_tokens": 142293759.0, + "step": 8386 + }, + { + "epoch": 0.6915972623072483, + "grad_norm": 1.8236057758331299, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7678005695343018, + "num_tokens": 142325580.0, + "step": 8387 + }, + { + "epoch": 0.6916797229322998, + "grad_norm": 1.8811172246932983, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7366892695426941, + "num_tokens": 142357387.0, + "step": 8388 + }, + { + "epoch": 0.6917621835573514, + "grad_norm": 1.9248764514923096, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7296143174171448, + "num_tokens": 142388901.0, + "step": 8389 + }, + { + "epoch": 0.6918446441824029, + "grad_norm": 2.1745548248291016, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6940314769744873, + "num_tokens": 142414999.0, + "step": 8390 + }, + { + "epoch": 0.6919271048074545, + "grad_norm": 2.1988441944122314, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.742775559425354, + "num_tokens": 142441994.0, + "step": 8391 + }, + { + "epoch": 0.692009565432506, + "grad_norm": 1.8576682806015015, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.766573965549469, + "num_tokens": 142472417.0, + "step": 8392 + }, + { + "epoch": 0.6920920260575575, + "grad_norm": 1.96564519405365, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7134913206100464, + "num_tokens": 142501797.0, + "step": 8393 + }, + { + "epoch": 0.6921744866826091, + "grad_norm": 1.9094266891479492, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7321125268936157, + "num_tokens": 142531998.0, + "step": 8394 + }, + { + "epoch": 0.6922569473076606, + "grad_norm": 2.328550100326538, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7219495177268982, + "num_tokens": 142553983.0, + "step": 8395 + }, + { + "epoch": 0.6923394079327121, + "grad_norm": 1.8305984735488892, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7702906131744385, + "num_tokens": 142587274.0, + "step": 8396 + }, + { + "epoch": 0.6924218685577637, + "grad_norm": 2.1788105964660645, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7232243418693542, + "num_tokens": 142613156.0, + "step": 8397 + }, + { + "epoch": 0.6925043291828152, + "grad_norm": 1.9163949489593506, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.738328218460083, + "num_tokens": 142642471.0, + "step": 8398 + }, + { + "epoch": 0.6925867898078667, + "grad_norm": 1.7671399116516113, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7169622182846069, + "num_tokens": 142680551.0, + "step": 8399 + }, + { + "epoch": 0.6926692504329183, + "grad_norm": 1.923378348350525, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7069841027259827, + "num_tokens": 142711974.0, + "step": 8400 + }, + { + "epoch": 0.6927517110579698, + "grad_norm": 1.8187254667282104, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7713200449943542, + "num_tokens": 142744652.0, + "step": 8401 + }, + { + "epoch": 0.6928341716830213, + "grad_norm": 2.087043285369873, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7319396734237671, + "num_tokens": 142772259.0, + "step": 8402 + }, + { + "epoch": 0.6929166323080729, + "grad_norm": 1.9838757514953613, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.697097659111023, + "num_tokens": 142803231.0, + "step": 8403 + }, + { + "epoch": 0.6929990929331244, + "grad_norm": 1.8403536081314087, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7617638111114502, + "num_tokens": 142832329.0, + "step": 8404 + }, + { + "epoch": 0.693081553558176, + "grad_norm": 1.9714761972427368, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7102732062339783, + "num_tokens": 142861750.0, + "step": 8405 + }, + { + "epoch": 0.6931640141832275, + "grad_norm": 1.710068702697754, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7611140012741089, + "num_tokens": 142895787.0, + "step": 8406 + }, + { + "epoch": 0.6932464748082791, + "grad_norm": 1.9102221727371216, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7585481405258179, + "num_tokens": 142925740.0, + "step": 8407 + }, + { + "epoch": 0.6933289354333306, + "grad_norm": 1.953620433807373, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7104781270027161, + "num_tokens": 142955393.0, + "step": 8408 + }, + { + "epoch": 0.6934113960583821, + "grad_norm": 2.0538973808288574, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7203693985939026, + "num_tokens": 142986217.0, + "step": 8409 + }, + { + "epoch": 0.6934938566834337, + "grad_norm": 1.686314582824707, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.720025360584259, + "num_tokens": 143023103.0, + "step": 8410 + }, + { + "epoch": 0.6935763173084852, + "grad_norm": 2.4070348739624023, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.72983717918396, + "num_tokens": 143044172.0, + "step": 8411 + }, + { + "epoch": 0.6936587779335367, + "grad_norm": 2.1257762908935547, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7278748750686646, + "num_tokens": 143068976.0, + "step": 8412 + }, + { + "epoch": 0.6937412385585883, + "grad_norm": 1.9337220191955566, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7117754220962524, + "num_tokens": 143102777.0, + "step": 8413 + }, + { + "epoch": 0.6938236991836398, + "grad_norm": 1.9656907320022583, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7494596242904663, + "num_tokens": 143132967.0, + "step": 8414 + }, + { + "epoch": 0.6939061598086913, + "grad_norm": 1.979640007019043, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7049482464790344, + "num_tokens": 143164072.0, + "step": 8415 + }, + { + "epoch": 0.6939886204337429, + "grad_norm": 1.672900676727295, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7494122385978699, + "num_tokens": 143200716.0, + "step": 8416 + }, + { + "epoch": 0.6940710810587944, + "grad_norm": 2.0228378772735596, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7375508546829224, + "num_tokens": 143229797.0, + "step": 8417 + }, + { + "epoch": 0.6941535416838459, + "grad_norm": 2.0175769329071045, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7412772178649902, + "num_tokens": 143256350.0, + "step": 8418 + }, + { + "epoch": 0.6942360023088975, + "grad_norm": 1.7160396575927734, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.8012366890907288, + "num_tokens": 143288712.0, + "step": 8419 + }, + { + "epoch": 0.694318462933949, + "grad_norm": 2.1109447479248047, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7738233804702759, + "num_tokens": 143313190.0, + "step": 8420 + }, + { + "epoch": 0.6944009235590006, + "grad_norm": 1.7842475175857544, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7604761123657227, + "num_tokens": 143348009.0, + "step": 8421 + }, + { + "epoch": 0.6944833841840521, + "grad_norm": 2.19376277923584, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7239369750022888, + "num_tokens": 143372648.0, + "step": 8422 + }, + { + "epoch": 0.6945658448091037, + "grad_norm": 1.8101609945297241, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7553614974021912, + "num_tokens": 143409626.0, + "step": 8423 + }, + { + "epoch": 0.6946483054341552, + "grad_norm": 2.0153918266296387, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.725951611995697, + "num_tokens": 143437112.0, + "step": 8424 + }, + { + "epoch": 0.6947307660592067, + "grad_norm": 1.9974780082702637, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.751064658164978, + "num_tokens": 143468246.0, + "step": 8425 + }, + { + "epoch": 0.6948132266842583, + "grad_norm": 2.032829523086548, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7097682356834412, + "num_tokens": 143498691.0, + "step": 8426 + }, + { + "epoch": 0.6948956873093098, + "grad_norm": 2.210604429244995, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7476699352264404, + "num_tokens": 143523414.0, + "step": 8427 + }, + { + "epoch": 0.6949781479343613, + "grad_norm": 2.1882078647613525, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7340171933174133, + "num_tokens": 143551177.0, + "step": 8428 + }, + { + "epoch": 0.6950606085594129, + "grad_norm": 2.083423614501953, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7506257891654968, + "num_tokens": 143578510.0, + "step": 8429 + }, + { + "epoch": 0.6951430691844644, + "grad_norm": 2.011089563369751, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.7620165944099426, + "num_tokens": 143606105.0, + "step": 8430 + }, + { + "epoch": 0.6952255298095159, + "grad_norm": 1.8885385990142822, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7656891345977783, + "num_tokens": 143635129.0, + "step": 8431 + }, + { + "epoch": 0.6953079904345675, + "grad_norm": 1.9244544506072998, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7450177073478699, + "num_tokens": 143665896.0, + "step": 8432 + }, + { + "epoch": 0.695390451059619, + "grad_norm": 1.7939609289169312, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7410951256752014, + "num_tokens": 143702640.0, + "step": 8433 + }, + { + "epoch": 0.6954729116846706, + "grad_norm": 1.9474433660507202, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7699486613273621, + "num_tokens": 143731460.0, + "step": 8434 + }, + { + "epoch": 0.6955553723097221, + "grad_norm": 2.2068731784820557, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7469750642776489, + "num_tokens": 143754255.0, + "step": 8435 + }, + { + "epoch": 0.6956378329347737, + "grad_norm": 1.9236928224563599, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7577826976776123, + "num_tokens": 143782656.0, + "step": 8436 + }, + { + "epoch": 0.6957202935598252, + "grad_norm": 1.8529936075210571, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7186449766159058, + "num_tokens": 143814555.0, + "step": 8437 + }, + { + "epoch": 0.6958027541848767, + "grad_norm": 2.0070762634277344, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7515262365341187, + "num_tokens": 143845701.0, + "step": 8438 + }, + { + "epoch": 0.6958852148099283, + "grad_norm": 1.9138084650039673, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7483457326889038, + "num_tokens": 143880556.0, + "step": 8439 + }, + { + "epoch": 0.6959676754349798, + "grad_norm": 2.033341884613037, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7275446653366089, + "num_tokens": 143911821.0, + "step": 8440 + }, + { + "epoch": 0.6960501360600313, + "grad_norm": 1.8946363925933838, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7435984015464783, + "num_tokens": 143944025.0, + "step": 8441 + }, + { + "epoch": 0.6961325966850829, + "grad_norm": 2.10664701461792, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7135429382324219, + "num_tokens": 143972442.0, + "step": 8442 + }, + { + "epoch": 0.6962150573101344, + "grad_norm": 1.7165342569351196, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.749144971370697, + "num_tokens": 144008620.0, + "step": 8443 + }, + { + "epoch": 0.6962975179351859, + "grad_norm": 1.8803821802139282, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.7449073195457458, + "num_tokens": 144042643.0, + "step": 8444 + }, + { + "epoch": 0.6963799785602375, + "grad_norm": 1.9410316944122314, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7871566414833069, + "num_tokens": 144072313.0, + "step": 8445 + }, + { + "epoch": 0.696462439185289, + "grad_norm": 1.8979860544204712, + "learning_rate": 1e-06, + "loss": 0.6103, + "mean_token_accuracy": 0.7990126609802246, + "num_tokens": 144100571.0, + "step": 8446 + }, + { + "epoch": 0.6965448998103405, + "grad_norm": 2.2060627937316895, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7263597249984741, + "num_tokens": 144127336.0, + "step": 8447 + }, + { + "epoch": 0.6966273604353921, + "grad_norm": 1.8566561937332153, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7271084785461426, + "num_tokens": 144162446.0, + "step": 8448 + }, + { + "epoch": 0.6967098210604437, + "grad_norm": 1.8459208011627197, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7382130026817322, + "num_tokens": 144195990.0, + "step": 8449 + }, + { + "epoch": 0.6967922816854952, + "grad_norm": 2.089226722717285, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7582737803459167, + "num_tokens": 144223626.0, + "step": 8450 + }, + { + "epoch": 0.6968747423105467, + "grad_norm": 2.3048293590545654, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.743094801902771, + "num_tokens": 144244652.0, + "step": 8451 + }, + { + "epoch": 0.6969572029355983, + "grad_norm": 1.7801564931869507, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.761783242225647, + "num_tokens": 144275427.0, + "step": 8452 + }, + { + "epoch": 0.6970396635606498, + "grad_norm": 1.9039548635482788, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7238767743110657, + "num_tokens": 144307490.0, + "step": 8453 + }, + { + "epoch": 0.6971221241857013, + "grad_norm": 1.9630532264709473, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7614499926567078, + "num_tokens": 144336012.0, + "step": 8454 + }, + { + "epoch": 0.6972045848107529, + "grad_norm": 1.8878308534622192, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7358795404434204, + "num_tokens": 144367950.0, + "step": 8455 + }, + { + "epoch": 0.6972870454358044, + "grad_norm": 2.11889386177063, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7416843175888062, + "num_tokens": 144393212.0, + "step": 8456 + }, + { + "epoch": 0.6973695060608559, + "grad_norm": 1.9287941455841064, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7580749988555908, + "num_tokens": 144422211.0, + "step": 8457 + }, + { + "epoch": 0.6974519666859075, + "grad_norm": 1.8682725429534912, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7547025680541992, + "num_tokens": 144453664.0, + "step": 8458 + }, + { + "epoch": 0.697534427310959, + "grad_norm": 2.072631359100342, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7547129392623901, + "num_tokens": 144479920.0, + "step": 8459 + }, + { + "epoch": 0.6976168879360105, + "grad_norm": 1.8741860389709473, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7479157447814941, + "num_tokens": 144511275.0, + "step": 8460 + }, + { + "epoch": 0.6976993485610621, + "grad_norm": 1.8837555646896362, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7664556503295898, + "num_tokens": 144541818.0, + "step": 8461 + }, + { + "epoch": 0.6977818091861137, + "grad_norm": 1.8620824813842773, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.725004255771637, + "num_tokens": 144576008.0, + "step": 8462 + }, + { + "epoch": 0.6978642698111651, + "grad_norm": 1.8155981302261353, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.7057629823684692, + "num_tokens": 144612810.0, + "step": 8463 + }, + { + "epoch": 0.6979467304362167, + "grad_norm": 1.8415169715881348, + "learning_rate": 1e-06, + "loss": 0.6106, + "mean_token_accuracy": 0.8029031753540039, + "num_tokens": 144642611.0, + "step": 8464 + }, + { + "epoch": 0.6980291910612683, + "grad_norm": 1.9743512868881226, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7268301248550415, + "num_tokens": 144671715.0, + "step": 8465 + }, + { + "epoch": 0.6981116516863198, + "grad_norm": 1.7341526746749878, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.759157121181488, + "num_tokens": 144707671.0, + "step": 8466 + }, + { + "epoch": 0.6981941123113713, + "grad_norm": 1.7357267141342163, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7723461389541626, + "num_tokens": 144743729.0, + "step": 8467 + }, + { + "epoch": 0.6982765729364229, + "grad_norm": 1.7378727197647095, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7637146711349487, + "num_tokens": 144780495.0, + "step": 8468 + }, + { + "epoch": 0.6983590335614744, + "grad_norm": 1.9842026233673096, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7731818556785583, + "num_tokens": 144806417.0, + "step": 8469 + }, + { + "epoch": 0.6984414941865259, + "grad_norm": 1.9282894134521484, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7237153649330139, + "num_tokens": 144838091.0, + "step": 8470 + }, + { + "epoch": 0.6985239548115775, + "grad_norm": 2.007382392883301, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7687723636627197, + "num_tokens": 144867822.0, + "step": 8471 + }, + { + "epoch": 0.698606415436629, + "grad_norm": 1.9703707695007324, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.7351723313331604, + "num_tokens": 144898591.0, + "step": 8472 + }, + { + "epoch": 0.6986888760616805, + "grad_norm": 1.9948889017105103, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7826598882675171, + "num_tokens": 144927586.0, + "step": 8473 + }, + { + "epoch": 0.6987713366867321, + "grad_norm": 2.121361017227173, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.724166750907898, + "num_tokens": 144954447.0, + "step": 8474 + }, + { + "epoch": 0.6988537973117837, + "grad_norm": 1.782447099685669, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7689754962921143, + "num_tokens": 144986492.0, + "step": 8475 + }, + { + "epoch": 0.6989362579368351, + "grad_norm": 1.7974143028259277, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.7964271306991577, + "num_tokens": 145016821.0, + "step": 8476 + }, + { + "epoch": 0.6990187185618867, + "grad_norm": 2.0583834648132324, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.717099666595459, + "num_tokens": 145047151.0, + "step": 8477 + }, + { + "epoch": 0.6991011791869383, + "grad_norm": 2.0321762561798096, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7361468076705933, + "num_tokens": 145075591.0, + "step": 8478 + }, + { + "epoch": 0.6991836398119897, + "grad_norm": 1.8347761631011963, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7298182249069214, + "num_tokens": 145106799.0, + "step": 8479 + }, + { + "epoch": 0.6992661004370413, + "grad_norm": 2.0228724479675293, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7604133486747742, + "num_tokens": 145133755.0, + "step": 8480 + }, + { + "epoch": 0.6993485610620929, + "grad_norm": 1.9007762670516968, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.718461275100708, + "num_tokens": 145165508.0, + "step": 8481 + }, + { + "epoch": 0.6994310216871444, + "grad_norm": 2.004284381866455, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7957521677017212, + "num_tokens": 145194701.0, + "step": 8482 + }, + { + "epoch": 0.6995134823121959, + "grad_norm": 1.9182127714157104, + "learning_rate": 1e-06, + "loss": 0.5724, + "mean_token_accuracy": 0.8071619272232056, + "num_tokens": 145224711.0, + "step": 8483 + }, + { + "epoch": 0.6995959429372475, + "grad_norm": 1.7600349187850952, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7489569783210754, + "num_tokens": 145257325.0, + "step": 8484 + }, + { + "epoch": 0.699678403562299, + "grad_norm": 2.0628645420074463, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7702019214630127, + "num_tokens": 145285336.0, + "step": 8485 + }, + { + "epoch": 0.6997608641873505, + "grad_norm": 1.8171944618225098, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7455638647079468, + "num_tokens": 145316369.0, + "step": 8486 + }, + { + "epoch": 0.6998433248124021, + "grad_norm": 1.733353853225708, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7433214783668518, + "num_tokens": 145354330.0, + "step": 8487 + }, + { + "epoch": 0.6999257854374537, + "grad_norm": 1.9303561449050903, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7233728170394897, + "num_tokens": 145384984.0, + "step": 8488 + }, + { + "epoch": 0.7000082460625051, + "grad_norm": 2.058893918991089, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7558697462081909, + "num_tokens": 145412992.0, + "step": 8489 + }, + { + "epoch": 0.7000907066875567, + "grad_norm": 2.0626749992370605, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7177122831344604, + "num_tokens": 145443272.0, + "step": 8490 + }, + { + "epoch": 0.7001731673126083, + "grad_norm": 2.0469207763671875, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7521735429763794, + "num_tokens": 145469903.0, + "step": 8491 + }, + { + "epoch": 0.7002556279376597, + "grad_norm": 1.929459810256958, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7766865491867065, + "num_tokens": 145497347.0, + "step": 8492 + }, + { + "epoch": 0.7003380885627113, + "grad_norm": 2.1052944660186768, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7500524520874023, + "num_tokens": 145522973.0, + "step": 8493 + }, + { + "epoch": 0.7004205491877629, + "grad_norm": 2.154103994369507, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7296419143676758, + "num_tokens": 145546921.0, + "step": 8494 + }, + { + "epoch": 0.7005030098128143, + "grad_norm": 2.0545120239257812, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7432202696800232, + "num_tokens": 145574997.0, + "step": 8495 + }, + { + "epoch": 0.7005854704378659, + "grad_norm": 2.453500509262085, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7332843542098999, + "num_tokens": 145594759.0, + "step": 8496 + }, + { + "epoch": 0.7006679310629175, + "grad_norm": 1.98404061794281, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7523078322410583, + "num_tokens": 145624577.0, + "step": 8497 + }, + { + "epoch": 0.700750391687969, + "grad_norm": 1.8583946228027344, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.7931469082832336, + "num_tokens": 145657662.0, + "step": 8498 + }, + { + "epoch": 0.7008328523130205, + "grad_norm": 2.2929420471191406, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.7503056526184082, + "num_tokens": 145679526.0, + "step": 8499 + }, + { + "epoch": 0.7009153129380721, + "grad_norm": 2.1275765895843506, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7317790985107422, + "num_tokens": 145707752.0, + "step": 8500 + }, + { + "epoch": 0.7009977735631236, + "grad_norm": 2.173025608062744, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7503875494003296, + "num_tokens": 145733812.0, + "step": 8501 + }, + { + "epoch": 0.7010802341881751, + "grad_norm": 1.820725440979004, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7708975076675415, + "num_tokens": 145770950.0, + "step": 8502 + }, + { + "epoch": 0.7011626948132267, + "grad_norm": 1.9552377462387085, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7501945495605469, + "num_tokens": 145800753.0, + "step": 8503 + }, + { + "epoch": 0.7012451554382783, + "grad_norm": 1.9309991598129272, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7341781258583069, + "num_tokens": 145831473.0, + "step": 8504 + }, + { + "epoch": 0.7013276160633297, + "grad_norm": 2.388741970062256, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.749673068523407, + "num_tokens": 145853923.0, + "step": 8505 + }, + { + "epoch": 0.7014100766883813, + "grad_norm": 1.9751834869384766, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7612436413764954, + "num_tokens": 145883750.0, + "step": 8506 + }, + { + "epoch": 0.7014925373134329, + "grad_norm": 1.7686063051223755, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.736913800239563, + "num_tokens": 145917324.0, + "step": 8507 + }, + { + "epoch": 0.7015749979384843, + "grad_norm": 1.9136683940887451, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7324427366256714, + "num_tokens": 145947731.0, + "step": 8508 + }, + { + "epoch": 0.7016574585635359, + "grad_norm": 1.9004921913146973, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.741581916809082, + "num_tokens": 145980119.0, + "step": 8509 + }, + { + "epoch": 0.7017399191885875, + "grad_norm": 1.9449803829193115, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7678524255752563, + "num_tokens": 146008781.0, + "step": 8510 + }, + { + "epoch": 0.701822379813639, + "grad_norm": 2.089221715927124, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7143267393112183, + "num_tokens": 146037092.0, + "step": 8511 + }, + { + "epoch": 0.7019048404386905, + "grad_norm": 1.913688063621521, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7591729164123535, + "num_tokens": 146068636.0, + "step": 8512 + }, + { + "epoch": 0.7019873010637421, + "grad_norm": 1.9884591102600098, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7246235609054565, + "num_tokens": 146097884.0, + "step": 8513 + }, + { + "epoch": 0.7020697616887936, + "grad_norm": 1.7885349988937378, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7306696772575378, + "num_tokens": 146133948.0, + "step": 8514 + }, + { + "epoch": 0.7021522223138451, + "grad_norm": 1.7428480386734009, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7521793246269226, + "num_tokens": 146169008.0, + "step": 8515 + }, + { + "epoch": 0.7022346829388967, + "grad_norm": 1.9569555521011353, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7533624768257141, + "num_tokens": 146197972.0, + "step": 8516 + }, + { + "epoch": 0.7023171435639483, + "grad_norm": 2.057400703430176, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.76153165102005, + "num_tokens": 146225608.0, + "step": 8517 + }, + { + "epoch": 0.7023996041889997, + "grad_norm": 2.008059501647949, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7452161312103271, + "num_tokens": 146257984.0, + "step": 8518 + }, + { + "epoch": 0.7024820648140513, + "grad_norm": 1.7729876041412354, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7398567795753479, + "num_tokens": 146292260.0, + "step": 8519 + }, + { + "epoch": 0.7025645254391029, + "grad_norm": 1.9099020957946777, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.758409321308136, + "num_tokens": 146322250.0, + "step": 8520 + }, + { + "epoch": 0.7026469860641543, + "grad_norm": 1.7962371110916138, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7463525533676147, + "num_tokens": 146356749.0, + "step": 8521 + }, + { + "epoch": 0.7027294466892059, + "grad_norm": 2.1248881816864014, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7262827157974243, + "num_tokens": 146382308.0, + "step": 8522 + }, + { + "epoch": 0.7028119073142575, + "grad_norm": 1.7984055280685425, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7145839929580688, + "num_tokens": 146420172.0, + "step": 8523 + }, + { + "epoch": 0.7028943679393089, + "grad_norm": 2.1789557933807373, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7739791870117188, + "num_tokens": 146442513.0, + "step": 8524 + }, + { + "epoch": 0.7029768285643605, + "grad_norm": 1.9800623655319214, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7247837781906128, + "num_tokens": 146472940.0, + "step": 8525 + }, + { + "epoch": 0.7030592891894121, + "grad_norm": 2.133662223815918, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7559874057769775, + "num_tokens": 146499004.0, + "step": 8526 + }, + { + "epoch": 0.7031417498144636, + "grad_norm": 1.9538873434066772, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7447109818458557, + "num_tokens": 146528030.0, + "step": 8527 + }, + { + "epoch": 0.7032242104395151, + "grad_norm": 2.0199129581451416, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7570062279701233, + "num_tokens": 146555013.0, + "step": 8528 + }, + { + "epoch": 0.7033066710645667, + "grad_norm": 1.9596024751663208, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7441893815994263, + "num_tokens": 146584721.0, + "step": 8529 + }, + { + "epoch": 0.7033891316896183, + "grad_norm": 1.858548879623413, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7843544483184814, + "num_tokens": 146616911.0, + "step": 8530 + }, + { + "epoch": 0.7034715923146697, + "grad_norm": 1.856477975845337, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7621302604675293, + "num_tokens": 146647497.0, + "step": 8531 + }, + { + "epoch": 0.7035540529397213, + "grad_norm": 1.9389362335205078, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7121843099594116, + "num_tokens": 146676982.0, + "step": 8532 + }, + { + "epoch": 0.7036365135647729, + "grad_norm": 2.078444242477417, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7055254578590393, + "num_tokens": 146705494.0, + "step": 8533 + }, + { + "epoch": 0.7037189741898243, + "grad_norm": 1.8559226989746094, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7842272520065308, + "num_tokens": 146734389.0, + "step": 8534 + }, + { + "epoch": 0.7038014348148759, + "grad_norm": 1.7231122255325317, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.7755104303359985, + "num_tokens": 146770559.0, + "step": 8535 + }, + { + "epoch": 0.7038838954399275, + "grad_norm": 1.861038088798523, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7349939346313477, + "num_tokens": 146806024.0, + "step": 8536 + }, + { + "epoch": 0.7039663560649789, + "grad_norm": 2.025009870529175, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.771891176700592, + "num_tokens": 146833360.0, + "step": 8537 + }, + { + "epoch": 0.7040488166900305, + "grad_norm": 1.7891230583190918, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.7569087743759155, + "num_tokens": 146873268.0, + "step": 8538 + }, + { + "epoch": 0.7041312773150821, + "grad_norm": 1.878542423248291, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7280780076980591, + "num_tokens": 146909401.0, + "step": 8539 + }, + { + "epoch": 0.7042137379401335, + "grad_norm": 2.5605249404907227, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7669929265975952, + "num_tokens": 146929653.0, + "step": 8540 + }, + { + "epoch": 0.7042961985651851, + "grad_norm": 1.8075275421142578, + "learning_rate": 1e-06, + "loss": 0.5787, + "mean_token_accuracy": 0.8143397569656372, + "num_tokens": 146957896.0, + "step": 8541 + }, + { + "epoch": 0.7043786591902367, + "grad_norm": 1.871901512145996, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7312899231910706, + "num_tokens": 146988450.0, + "step": 8542 + }, + { + "epoch": 0.7044611198152883, + "grad_norm": 2.1573615074157715, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7354820966720581, + "num_tokens": 147013790.0, + "step": 8543 + }, + { + "epoch": 0.7045435804403397, + "grad_norm": 2.136329174041748, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7595102190971375, + "num_tokens": 147039751.0, + "step": 8544 + }, + { + "epoch": 0.7046260410653913, + "grad_norm": 2.028278112411499, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7332124710083008, + "num_tokens": 147068045.0, + "step": 8545 + }, + { + "epoch": 0.7047085016904429, + "grad_norm": 1.7581758499145508, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7759320735931396, + "num_tokens": 147101532.0, + "step": 8546 + }, + { + "epoch": 0.7047909623154943, + "grad_norm": 1.983672857284546, + "learning_rate": 1e-06, + "loss": 0.7762, + "mean_token_accuracy": 0.756736695766449, + "num_tokens": 147130177.0, + "step": 8547 + }, + { + "epoch": 0.7048734229405459, + "grad_norm": 1.7611891031265259, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.7897348999977112, + "num_tokens": 147160722.0, + "step": 8548 + }, + { + "epoch": 0.7049558835655975, + "grad_norm": 1.9136322736740112, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7426393032073975, + "num_tokens": 147192160.0, + "step": 8549 + }, + { + "epoch": 0.7050383441906489, + "grad_norm": 1.977889060974121, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7621896266937256, + "num_tokens": 147220406.0, + "step": 8550 + }, + { + "epoch": 0.7051208048157005, + "grad_norm": 2.106837034225464, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.6982967257499695, + "num_tokens": 147247323.0, + "step": 8551 + }, + { + "epoch": 0.7052032654407521, + "grad_norm": 1.929803729057312, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7674432992935181, + "num_tokens": 147275064.0, + "step": 8552 + }, + { + "epoch": 0.7052857260658035, + "grad_norm": 1.8278146982192993, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7219305038452148, + "num_tokens": 147309183.0, + "step": 8553 + }, + { + "epoch": 0.7053681866908551, + "grad_norm": 2.0494370460510254, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7349768280982971, + "num_tokens": 147339685.0, + "step": 8554 + }, + { + "epoch": 0.7054506473159067, + "grad_norm": 2.124906539916992, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7258026599884033, + "num_tokens": 147368131.0, + "step": 8555 + }, + { + "epoch": 0.7055331079409581, + "grad_norm": 1.9611685276031494, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7354483008384705, + "num_tokens": 147399530.0, + "step": 8556 + }, + { + "epoch": 0.7056155685660097, + "grad_norm": 1.9697802066802979, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.755060076713562, + "num_tokens": 147431481.0, + "step": 8557 + }, + { + "epoch": 0.7056980291910613, + "grad_norm": 2.1287479400634766, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7561277151107788, + "num_tokens": 147458446.0, + "step": 8558 + }, + { + "epoch": 0.7057804898161129, + "grad_norm": 1.8809418678283691, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7305361032485962, + "num_tokens": 147492896.0, + "step": 8559 + }, + { + "epoch": 0.7058629504411643, + "grad_norm": 1.9146283864974976, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7638154029846191, + "num_tokens": 147523600.0, + "step": 8560 + }, + { + "epoch": 0.7059454110662159, + "grad_norm": 1.9635730981826782, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7389669418334961, + "num_tokens": 147554353.0, + "step": 8561 + }, + { + "epoch": 0.7060278716912675, + "grad_norm": 2.0103485584259033, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7535065412521362, + "num_tokens": 147583559.0, + "step": 8562 + }, + { + "epoch": 0.7061103323163189, + "grad_norm": 2.0095696449279785, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7442603707313538, + "num_tokens": 147614595.0, + "step": 8563 + }, + { + "epoch": 0.7061927929413705, + "grad_norm": 1.7658765316009521, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7699289321899414, + "num_tokens": 147648032.0, + "step": 8564 + }, + { + "epoch": 0.7062752535664221, + "grad_norm": 2.2109570503234863, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7404909133911133, + "num_tokens": 147671629.0, + "step": 8565 + }, + { + "epoch": 0.7063577141914735, + "grad_norm": 2.0423200130462646, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7232118844985962, + "num_tokens": 147700316.0, + "step": 8566 + }, + { + "epoch": 0.7064401748165251, + "grad_norm": 2.2314629554748535, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7589446902275085, + "num_tokens": 147725159.0, + "step": 8567 + }, + { + "epoch": 0.7065226354415767, + "grad_norm": 1.9944450855255127, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7354601621627808, + "num_tokens": 147757472.0, + "step": 8568 + }, + { + "epoch": 0.7066050960666281, + "grad_norm": 1.8862557411193848, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7224384546279907, + "num_tokens": 147789472.0, + "step": 8569 + }, + { + "epoch": 0.7066875566916797, + "grad_norm": 1.8835976123809814, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7538049817085266, + "num_tokens": 147820882.0, + "step": 8570 + }, + { + "epoch": 0.7067700173167313, + "grad_norm": 1.9762561321258545, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.6864013075828552, + "num_tokens": 147851411.0, + "step": 8571 + }, + { + "epoch": 0.7068524779417829, + "grad_norm": 2.1188676357269287, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7576060891151428, + "num_tokens": 147875372.0, + "step": 8572 + }, + { + "epoch": 0.7069349385668343, + "grad_norm": 1.8733696937561035, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7014431357383728, + "num_tokens": 147907447.0, + "step": 8573 + }, + { + "epoch": 0.7070173991918859, + "grad_norm": 1.885615587234497, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7340233325958252, + "num_tokens": 147938973.0, + "step": 8574 + }, + { + "epoch": 0.7070998598169375, + "grad_norm": 2.068105936050415, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7296836376190186, + "num_tokens": 147965548.0, + "step": 8575 + }, + { + "epoch": 0.7071823204419889, + "grad_norm": 2.003359079360962, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.7531376481056213, + "num_tokens": 147996630.0, + "step": 8576 + }, + { + "epoch": 0.7072647810670405, + "grad_norm": 1.8405349254608154, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.7035583257675171, + "num_tokens": 148031202.0, + "step": 8577 + }, + { + "epoch": 0.7073472416920921, + "grad_norm": 1.9601191282272339, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7776893377304077, + "num_tokens": 148059194.0, + "step": 8578 + }, + { + "epoch": 0.7074297023171435, + "grad_norm": 1.719154715538025, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7720047831535339, + "num_tokens": 148093391.0, + "step": 8579 + }, + { + "epoch": 0.7075121629421951, + "grad_norm": 1.8508788347244263, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7320715188980103, + "num_tokens": 148126711.0, + "step": 8580 + }, + { + "epoch": 0.7075946235672467, + "grad_norm": 2.0432512760162354, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.724690318107605, + "num_tokens": 148154163.0, + "step": 8581 + }, + { + "epoch": 0.7076770841922981, + "grad_norm": 1.9531630277633667, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7398984432220459, + "num_tokens": 148183248.0, + "step": 8582 + }, + { + "epoch": 0.7077595448173497, + "grad_norm": 1.9303158521652222, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7317450046539307, + "num_tokens": 148215033.0, + "step": 8583 + }, + { + "epoch": 0.7078420054424013, + "grad_norm": 2.0721957683563232, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7151147723197937, + "num_tokens": 148243403.0, + "step": 8584 + }, + { + "epoch": 0.7079244660674527, + "grad_norm": 2.043424367904663, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7156659364700317, + "num_tokens": 148271123.0, + "step": 8585 + }, + { + "epoch": 0.7080069266925043, + "grad_norm": 1.7314507961273193, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.7833297848701477, + "num_tokens": 148304288.0, + "step": 8586 + }, + { + "epoch": 0.7080893873175559, + "grad_norm": 1.9882153272628784, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7349692583084106, + "num_tokens": 148334609.0, + "step": 8587 + }, + { + "epoch": 0.7081718479426075, + "grad_norm": 1.9030706882476807, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7402191758155823, + "num_tokens": 148366263.0, + "step": 8588 + }, + { + "epoch": 0.7082543085676589, + "grad_norm": 2.06018328666687, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7769017219543457, + "num_tokens": 148393489.0, + "step": 8589 + }, + { + "epoch": 0.7083367691927105, + "grad_norm": 1.7091021537780762, + "learning_rate": 1e-06, + "loss": 0.5606, + "mean_token_accuracy": 0.8172743916511536, + "num_tokens": 148427249.0, + "step": 8590 + }, + { + "epoch": 0.7084192298177621, + "grad_norm": 1.7459218502044678, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7739176750183105, + "num_tokens": 148462460.0, + "step": 8591 + }, + { + "epoch": 0.7085016904428135, + "grad_norm": 1.8655892610549927, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7557971477508545, + "num_tokens": 148491435.0, + "step": 8592 + }, + { + "epoch": 0.7085841510678651, + "grad_norm": 2.012430191040039, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7264176607131958, + "num_tokens": 148519417.0, + "step": 8593 + }, + { + "epoch": 0.7086666116929167, + "grad_norm": 1.7673777341842651, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.779401421546936, + "num_tokens": 148553319.0, + "step": 8594 + }, + { + "epoch": 0.7087490723179681, + "grad_norm": 1.7758504152297974, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7792967557907104, + "num_tokens": 148589290.0, + "step": 8595 + }, + { + "epoch": 0.7088315329430197, + "grad_norm": 2.1499061584472656, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7621175646781921, + "num_tokens": 148615417.0, + "step": 8596 + }, + { + "epoch": 0.7089139935680713, + "grad_norm": 2.086796998977661, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7650721073150635, + "num_tokens": 148642805.0, + "step": 8597 + }, + { + "epoch": 0.7089964541931227, + "grad_norm": 2.084648609161377, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7233372330665588, + "num_tokens": 148671625.0, + "step": 8598 + }, + { + "epoch": 0.7090789148181743, + "grad_norm": 1.7529810667037964, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7755508422851562, + "num_tokens": 148706908.0, + "step": 8599 + }, + { + "epoch": 0.7091613754432259, + "grad_norm": 1.9726083278656006, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7839627265930176, + "num_tokens": 148736252.0, + "step": 8600 + }, + { + "epoch": 0.7092438360682773, + "grad_norm": 1.9112614393234253, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7313970923423767, + "num_tokens": 148768813.0, + "step": 8601 + }, + { + "epoch": 0.7093262966933289, + "grad_norm": 1.8660386800765991, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.7994210720062256, + "num_tokens": 148798217.0, + "step": 8602 + }, + { + "epoch": 0.7094087573183805, + "grad_norm": 1.834311842918396, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7408992052078247, + "num_tokens": 148831500.0, + "step": 8603 + }, + { + "epoch": 0.7094912179434321, + "grad_norm": 1.9305214881896973, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7523237466812134, + "num_tokens": 148861763.0, + "step": 8604 + }, + { + "epoch": 0.7095736785684835, + "grad_norm": 2.12766695022583, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7364856600761414, + "num_tokens": 148886170.0, + "step": 8605 + }, + { + "epoch": 0.7096561391935351, + "grad_norm": 1.7879546880722046, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7165035009384155, + "num_tokens": 148921528.0, + "step": 8606 + }, + { + "epoch": 0.7097385998185867, + "grad_norm": 2.251020908355713, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7731910943984985, + "num_tokens": 148945541.0, + "step": 8607 + }, + { + "epoch": 0.7098210604436381, + "grad_norm": 2.1807100772857666, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7427341938018799, + "num_tokens": 148969909.0, + "step": 8608 + }, + { + "epoch": 0.7099035210686897, + "grad_norm": 2.3923003673553467, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7365742921829224, + "num_tokens": 148990151.0, + "step": 8609 + }, + { + "epoch": 0.7099859816937413, + "grad_norm": 2.0713374614715576, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7069365978240967, + "num_tokens": 149018595.0, + "step": 8610 + }, + { + "epoch": 0.7100684423187927, + "grad_norm": 2.335557460784912, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.6986212134361267, + "num_tokens": 149042022.0, + "step": 8611 + }, + { + "epoch": 0.7101509029438443, + "grad_norm": 2.0185375213623047, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7495155930519104, + "num_tokens": 149069526.0, + "step": 8612 + }, + { + "epoch": 0.7102333635688959, + "grad_norm": 2.239492893218994, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7558708190917969, + "num_tokens": 149094466.0, + "step": 8613 + }, + { + "epoch": 0.7103158241939473, + "grad_norm": 1.9976630210876465, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7144066691398621, + "num_tokens": 149123066.0, + "step": 8614 + }, + { + "epoch": 0.7103982848189989, + "grad_norm": 1.962231159210205, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7169027328491211, + "num_tokens": 149155346.0, + "step": 8615 + }, + { + "epoch": 0.7104807454440505, + "grad_norm": 2.104099750518799, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7326788902282715, + "num_tokens": 149183269.0, + "step": 8616 + }, + { + "epoch": 0.710563206069102, + "grad_norm": 1.6584320068359375, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.7955665588378906, + "num_tokens": 149219246.0, + "step": 8617 + }, + { + "epoch": 0.7106456666941535, + "grad_norm": 2.0000593662261963, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7360719442367554, + "num_tokens": 149245533.0, + "step": 8618 + }, + { + "epoch": 0.7107281273192051, + "grad_norm": 1.9575661420822144, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7366266250610352, + "num_tokens": 149273487.0, + "step": 8619 + }, + { + "epoch": 0.7108105879442567, + "grad_norm": 1.8133150339126587, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.7963734269142151, + "num_tokens": 149307241.0, + "step": 8620 + }, + { + "epoch": 0.7108930485693081, + "grad_norm": 1.9383407831192017, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7110664248466492, + "num_tokens": 149338846.0, + "step": 8621 + }, + { + "epoch": 0.7109755091943597, + "grad_norm": 2.1724178791046143, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7196160554885864, + "num_tokens": 149366335.0, + "step": 8622 + }, + { + "epoch": 0.7110579698194113, + "grad_norm": 1.9662171602249146, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7704311609268188, + "num_tokens": 149396138.0, + "step": 8623 + }, + { + "epoch": 0.7111404304444627, + "grad_norm": 1.741236925125122, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.7868330478668213, + "num_tokens": 149427985.0, + "step": 8624 + }, + { + "epoch": 0.7112228910695143, + "grad_norm": 1.8584247827529907, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7770989537239075, + "num_tokens": 149459112.0, + "step": 8625 + }, + { + "epoch": 0.7113053516945659, + "grad_norm": 1.6565889120101929, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7826482057571411, + "num_tokens": 149497204.0, + "step": 8626 + }, + { + "epoch": 0.7113878123196173, + "grad_norm": 1.952170729637146, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7357951998710632, + "num_tokens": 149528150.0, + "step": 8627 + }, + { + "epoch": 0.7114702729446689, + "grad_norm": 2.0535101890563965, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7489404082298279, + "num_tokens": 149553820.0, + "step": 8628 + }, + { + "epoch": 0.7115527335697205, + "grad_norm": 1.8474453687667847, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7513983249664307, + "num_tokens": 149585411.0, + "step": 8629 + }, + { + "epoch": 0.711635194194772, + "grad_norm": 1.8616845607757568, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.703836977481842, + "num_tokens": 149619456.0, + "step": 8630 + }, + { + "epoch": 0.7117176548198235, + "grad_norm": 1.7764703035354614, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7650858759880066, + "num_tokens": 149651540.0, + "step": 8631 + }, + { + "epoch": 0.7118001154448751, + "grad_norm": 2.1818714141845703, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7257473468780518, + "num_tokens": 149676667.0, + "step": 8632 + }, + { + "epoch": 0.7118825760699267, + "grad_norm": 1.8616533279418945, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7518620491027832, + "num_tokens": 149708997.0, + "step": 8633 + }, + { + "epoch": 0.7119650366949781, + "grad_norm": 2.1008706092834473, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7373954057693481, + "num_tokens": 149736076.0, + "step": 8634 + }, + { + "epoch": 0.7120474973200297, + "grad_norm": 1.7414134740829468, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7725204229354858, + "num_tokens": 149770615.0, + "step": 8635 + }, + { + "epoch": 0.7121299579450813, + "grad_norm": 1.779023289680481, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7753899693489075, + "num_tokens": 149805711.0, + "step": 8636 + }, + { + "epoch": 0.7122124185701327, + "grad_norm": 1.878438115119934, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7704509496688843, + "num_tokens": 149836525.0, + "step": 8637 + }, + { + "epoch": 0.7122948791951843, + "grad_norm": 1.9071993827819824, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7840670347213745, + "num_tokens": 149867374.0, + "step": 8638 + }, + { + "epoch": 0.7123773398202359, + "grad_norm": 1.8434914350509644, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7639870643615723, + "num_tokens": 149898722.0, + "step": 8639 + }, + { + "epoch": 0.7124598004452873, + "grad_norm": 2.0006072521209717, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7839629650115967, + "num_tokens": 149927231.0, + "step": 8640 + }, + { + "epoch": 0.7125422610703389, + "grad_norm": 2.2940444946289062, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6960394978523254, + "num_tokens": 149953209.0, + "step": 8641 + }, + { + "epoch": 0.7126247216953905, + "grad_norm": 1.6396833658218384, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7573277950286865, + "num_tokens": 149991952.0, + "step": 8642 + }, + { + "epoch": 0.712707182320442, + "grad_norm": 1.7574455738067627, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7948687076568604, + "num_tokens": 150025282.0, + "step": 8643 + }, + { + "epoch": 0.7127896429454935, + "grad_norm": 1.7962501049041748, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.781773567199707, + "num_tokens": 150057726.0, + "step": 8644 + }, + { + "epoch": 0.7128721035705451, + "grad_norm": 1.7726465463638306, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7391932606697083, + "num_tokens": 150092640.0, + "step": 8645 + }, + { + "epoch": 0.7129545641955966, + "grad_norm": 2.3241806030273438, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7374916076660156, + "num_tokens": 150116102.0, + "step": 8646 + }, + { + "epoch": 0.7130370248206481, + "grad_norm": 2.2261905670166016, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.710702657699585, + "num_tokens": 150141252.0, + "step": 8647 + }, + { + "epoch": 0.7131194854456997, + "grad_norm": 2.0489981174468994, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7284189462661743, + "num_tokens": 150168369.0, + "step": 8648 + }, + { + "epoch": 0.7132019460707513, + "grad_norm": 1.9114387035369873, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7336493134498596, + "num_tokens": 150200454.0, + "step": 8649 + }, + { + "epoch": 0.7132844066958027, + "grad_norm": 2.0122010707855225, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.7947080135345459, + "num_tokens": 150224718.0, + "step": 8650 + }, + { + "epoch": 0.7133668673208543, + "grad_norm": 1.7608202695846558, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7091421484947205, + "num_tokens": 150261323.0, + "step": 8651 + }, + { + "epoch": 0.7134493279459059, + "grad_norm": 1.9137917757034302, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7622860670089722, + "num_tokens": 150288700.0, + "step": 8652 + }, + { + "epoch": 0.7135317885709573, + "grad_norm": 1.94706392288208, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7303120493888855, + "num_tokens": 150317037.0, + "step": 8653 + }, + { + "epoch": 0.7136142491960089, + "grad_norm": 2.2647018432617188, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7182953357696533, + "num_tokens": 150340340.0, + "step": 8654 + }, + { + "epoch": 0.7136967098210605, + "grad_norm": 1.7252050638198853, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7724121809005737, + "num_tokens": 150378200.0, + "step": 8655 + }, + { + "epoch": 0.7137791704461119, + "grad_norm": 2.0430986881256104, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7073360681533813, + "num_tokens": 150410081.0, + "step": 8656 + }, + { + "epoch": 0.7138616310711635, + "grad_norm": 1.8211724758148193, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7710360288619995, + "num_tokens": 150441021.0, + "step": 8657 + }, + { + "epoch": 0.7139440916962151, + "grad_norm": 1.7689979076385498, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7834591269493103, + "num_tokens": 150473109.0, + "step": 8658 + }, + { + "epoch": 0.7140265523212665, + "grad_norm": 2.2236154079437256, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7475009560585022, + "num_tokens": 150498070.0, + "step": 8659 + }, + { + "epoch": 0.7141090129463181, + "grad_norm": 1.8367377519607544, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7285275459289551, + "num_tokens": 150531083.0, + "step": 8660 + }, + { + "epoch": 0.7141914735713697, + "grad_norm": 1.7348213195800781, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7703036069869995, + "num_tokens": 150567189.0, + "step": 8661 + }, + { + "epoch": 0.7142739341964212, + "grad_norm": 2.0908215045928955, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7618246078491211, + "num_tokens": 150591195.0, + "step": 8662 + }, + { + "epoch": 0.7143563948214727, + "grad_norm": 1.8842772245407104, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7774139642715454, + "num_tokens": 150620698.0, + "step": 8663 + }, + { + "epoch": 0.7144388554465243, + "grad_norm": 1.9184321165084839, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7744011878967285, + "num_tokens": 150649848.0, + "step": 8664 + }, + { + "epoch": 0.7145213160715759, + "grad_norm": 1.946600079536438, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7567024230957031, + "num_tokens": 150680448.0, + "step": 8665 + }, + { + "epoch": 0.7146037766966273, + "grad_norm": 1.925587773323059, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7410293817520142, + "num_tokens": 150712002.0, + "step": 8666 + }, + { + "epoch": 0.7146862373216789, + "grad_norm": 2.1020302772521973, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7445800304412842, + "num_tokens": 150738541.0, + "step": 8667 + }, + { + "epoch": 0.7147686979467305, + "grad_norm": 2.063171625137329, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6994809508323669, + "num_tokens": 150767499.0, + "step": 8668 + }, + { + "epoch": 0.7148511585717819, + "grad_norm": 2.0799880027770996, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7603131532669067, + "num_tokens": 150794418.0, + "step": 8669 + }, + { + "epoch": 0.7149336191968335, + "grad_norm": 1.983127236366272, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7763588428497314, + "num_tokens": 150825890.0, + "step": 8670 + }, + { + "epoch": 0.7150160798218851, + "grad_norm": 2.097551107406616, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7775100469589233, + "num_tokens": 150849848.0, + "step": 8671 + }, + { + "epoch": 0.7150985404469365, + "grad_norm": 2.1280734539031982, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7527515888214111, + "num_tokens": 150877172.0, + "step": 8672 + }, + { + "epoch": 0.7151810010719881, + "grad_norm": 2.0449271202087402, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7275372743606567, + "num_tokens": 150908522.0, + "step": 8673 + }, + { + "epoch": 0.7152634616970397, + "grad_norm": 1.8369479179382324, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7410305738449097, + "num_tokens": 150943521.0, + "step": 8674 + }, + { + "epoch": 0.7153459223220912, + "grad_norm": 1.6980665922164917, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7428842782974243, + "num_tokens": 150979527.0, + "step": 8675 + }, + { + "epoch": 0.7154283829471427, + "grad_norm": 2.0273022651672363, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7524610161781311, + "num_tokens": 151005277.0, + "step": 8676 + }, + { + "epoch": 0.7155108435721943, + "grad_norm": 1.8579325675964355, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7537312507629395, + "num_tokens": 151037796.0, + "step": 8677 + }, + { + "epoch": 0.7155933041972458, + "grad_norm": 1.9329099655151367, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7709426879882812, + "num_tokens": 151067167.0, + "step": 8678 + }, + { + "epoch": 0.7156757648222973, + "grad_norm": 1.8812230825424194, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7687793970108032, + "num_tokens": 151100197.0, + "step": 8679 + }, + { + "epoch": 0.7157582254473489, + "grad_norm": 2.1518704891204834, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7674254179000854, + "num_tokens": 151123883.0, + "step": 8680 + }, + { + "epoch": 0.7158406860724005, + "grad_norm": 1.8627954721450806, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7580286264419556, + "num_tokens": 151157600.0, + "step": 8681 + }, + { + "epoch": 0.7159231466974519, + "grad_norm": 1.9281901121139526, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7607380151748657, + "num_tokens": 151187569.0, + "step": 8682 + }, + { + "epoch": 0.7160056073225035, + "grad_norm": 1.9075751304626465, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7472236752510071, + "num_tokens": 151217690.0, + "step": 8683 + }, + { + "epoch": 0.7160880679475551, + "grad_norm": 1.972481369972229, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7350068092346191, + "num_tokens": 151248089.0, + "step": 8684 + }, + { + "epoch": 0.7161705285726065, + "grad_norm": 1.9061728715896606, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7194886803627014, + "num_tokens": 151279740.0, + "step": 8685 + }, + { + "epoch": 0.7162529891976581, + "grad_norm": 1.993133783340454, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7551897764205933, + "num_tokens": 151306140.0, + "step": 8686 + }, + { + "epoch": 0.7163354498227097, + "grad_norm": 1.932019591331482, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7456234693527222, + "num_tokens": 151338061.0, + "step": 8687 + }, + { + "epoch": 0.7164179104477612, + "grad_norm": 1.8577772378921509, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.7718274593353271, + "num_tokens": 151370413.0, + "step": 8688 + }, + { + "epoch": 0.7165003710728127, + "grad_norm": 1.7305042743682861, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7210882902145386, + "num_tokens": 151408715.0, + "step": 8689 + }, + { + "epoch": 0.7165828316978643, + "grad_norm": 1.675217628479004, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.7808208465576172, + "num_tokens": 151441374.0, + "step": 8690 + }, + { + "epoch": 0.7166652923229158, + "grad_norm": 2.4183735847473145, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7240151166915894, + "num_tokens": 151461551.0, + "step": 8691 + }, + { + "epoch": 0.7167477529479673, + "grad_norm": 1.989044189453125, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.738359272480011, + "num_tokens": 151489054.0, + "step": 8692 + }, + { + "epoch": 0.7168302135730189, + "grad_norm": 1.8712127208709717, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7761857509613037, + "num_tokens": 151519671.0, + "step": 8693 + }, + { + "epoch": 0.7169126741980705, + "grad_norm": 2.224721908569336, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7278205156326294, + "num_tokens": 151545017.0, + "step": 8694 + }, + { + "epoch": 0.7169951348231219, + "grad_norm": 2.104827404022217, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7461934089660645, + "num_tokens": 151573147.0, + "step": 8695 + }, + { + "epoch": 0.7170775954481735, + "grad_norm": 1.7692641019821167, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7436243295669556, + "num_tokens": 151610239.0, + "step": 8696 + }, + { + "epoch": 0.7171600560732251, + "grad_norm": 1.9846875667572021, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7740155458450317, + "num_tokens": 151640757.0, + "step": 8697 + }, + { + "epoch": 0.7172425166982765, + "grad_norm": 1.8060630559921265, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7684674263000488, + "num_tokens": 151674410.0, + "step": 8698 + }, + { + "epoch": 0.7173249773233281, + "grad_norm": 2.0077764987945557, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7253891229629517, + "num_tokens": 151705556.0, + "step": 8699 + }, + { + "epoch": 0.7174074379483797, + "grad_norm": 1.9659420251846313, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.720822811126709, + "num_tokens": 151738617.0, + "step": 8700 + }, + { + "epoch": 0.7174898985734312, + "grad_norm": 1.7707977294921875, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7647964358329773, + "num_tokens": 151774389.0, + "step": 8701 + }, + { + "epoch": 0.7175723591984827, + "grad_norm": 1.7985857725143433, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7784073352813721, + "num_tokens": 151806784.0, + "step": 8702 + }, + { + "epoch": 0.7176548198235343, + "grad_norm": 1.9856479167938232, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.7875227332115173, + "num_tokens": 151833930.0, + "step": 8703 + }, + { + "epoch": 0.7177372804485858, + "grad_norm": 1.863539695739746, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7602212429046631, + "num_tokens": 151864459.0, + "step": 8704 + }, + { + "epoch": 0.7178197410736373, + "grad_norm": 2.059917449951172, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7536298632621765, + "num_tokens": 151891401.0, + "step": 8705 + }, + { + "epoch": 0.7179022016986889, + "grad_norm": 1.7512321472167969, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7702600955963135, + "num_tokens": 151923221.0, + "step": 8706 + }, + { + "epoch": 0.7179846623237404, + "grad_norm": 1.983546257019043, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7475686073303223, + "num_tokens": 151951919.0, + "step": 8707 + }, + { + "epoch": 0.7180671229487919, + "grad_norm": 1.9612352848052979, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7061048746109009, + "num_tokens": 151983238.0, + "step": 8708 + }, + { + "epoch": 0.7181495835738435, + "grad_norm": 1.9228919744491577, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.7966116666793823, + "num_tokens": 152011803.0, + "step": 8709 + }, + { + "epoch": 0.7182320441988951, + "grad_norm": 1.991383671760559, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7711202502250671, + "num_tokens": 152039206.0, + "step": 8710 + }, + { + "epoch": 0.7183145048239465, + "grad_norm": 1.8875925540924072, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7446737289428711, + "num_tokens": 152071814.0, + "step": 8711 + }, + { + "epoch": 0.7183969654489981, + "grad_norm": 1.9940080642700195, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7492049932479858, + "num_tokens": 152100067.0, + "step": 8712 + }, + { + "epoch": 0.7184794260740497, + "grad_norm": 1.9281673431396484, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7533870935440063, + "num_tokens": 152128938.0, + "step": 8713 + }, + { + "epoch": 0.7185618866991011, + "grad_norm": 1.9922926425933838, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7412874698638916, + "num_tokens": 152157351.0, + "step": 8714 + }, + { + "epoch": 0.7186443473241527, + "grad_norm": 1.571461796760559, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7475453019142151, + "num_tokens": 152203468.0, + "step": 8715 + }, + { + "epoch": 0.7187268079492043, + "grad_norm": 1.9247775077819824, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7578189373016357, + "num_tokens": 152230497.0, + "step": 8716 + }, + { + "epoch": 0.7188092685742558, + "grad_norm": 1.9440233707427979, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7411521077156067, + "num_tokens": 152262593.0, + "step": 8717 + }, + { + "epoch": 0.7188917291993073, + "grad_norm": 1.7994400262832642, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7207760810852051, + "num_tokens": 152300157.0, + "step": 8718 + }, + { + "epoch": 0.7189741898243589, + "grad_norm": 1.9648391008377075, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7830919623374939, + "num_tokens": 152328342.0, + "step": 8719 + }, + { + "epoch": 0.7190566504494104, + "grad_norm": 1.8254029750823975, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7321990728378296, + "num_tokens": 152361094.0, + "step": 8720 + }, + { + "epoch": 0.7191391110744619, + "grad_norm": 2.0096428394317627, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7464509010314941, + "num_tokens": 152389492.0, + "step": 8721 + }, + { + "epoch": 0.7192215716995135, + "grad_norm": 1.8132203817367554, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7536911368370056, + "num_tokens": 152423575.0, + "step": 8722 + }, + { + "epoch": 0.719304032324565, + "grad_norm": 1.8114616870880127, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7830981612205505, + "num_tokens": 152454909.0, + "step": 8723 + }, + { + "epoch": 0.7193864929496165, + "grad_norm": 1.7002463340759277, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.7885906100273132, + "num_tokens": 152487307.0, + "step": 8724 + }, + { + "epoch": 0.7194689535746681, + "grad_norm": 2.0626561641693115, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7669974565505981, + "num_tokens": 152514193.0, + "step": 8725 + }, + { + "epoch": 0.7195514141997197, + "grad_norm": 1.8185425996780396, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.727270245552063, + "num_tokens": 152546922.0, + "step": 8726 + }, + { + "epoch": 0.7196338748247711, + "grad_norm": 1.953173041343689, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7210118174552917, + "num_tokens": 152577245.0, + "step": 8727 + }, + { + "epoch": 0.7197163354498227, + "grad_norm": 1.81888747215271, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7718287706375122, + "num_tokens": 152607179.0, + "step": 8728 + }, + { + "epoch": 0.7197987960748743, + "grad_norm": 1.9237135648727417, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7731785178184509, + "num_tokens": 152638204.0, + "step": 8729 + }, + { + "epoch": 0.7198812566999258, + "grad_norm": 1.7199523448944092, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7850504517555237, + "num_tokens": 152674645.0, + "step": 8730 + }, + { + "epoch": 0.7199637173249773, + "grad_norm": 1.7857606410980225, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.7878518104553223, + "num_tokens": 152703739.0, + "step": 8731 + }, + { + "epoch": 0.7200461779500289, + "grad_norm": 2.226579189300537, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7536481618881226, + "num_tokens": 152728274.0, + "step": 8732 + }, + { + "epoch": 0.7201286385750804, + "grad_norm": 1.6705214977264404, + "learning_rate": 1e-06, + "loss": 0.5027, + "mean_token_accuracy": 0.8123806118965149, + "num_tokens": 152761134.0, + "step": 8733 + }, + { + "epoch": 0.7202110992001319, + "grad_norm": 1.931650161743164, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7568180561065674, + "num_tokens": 152789640.0, + "step": 8734 + }, + { + "epoch": 0.7202935598251835, + "grad_norm": 2.0655672550201416, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7774084210395813, + "num_tokens": 152816194.0, + "step": 8735 + }, + { + "epoch": 0.720376020450235, + "grad_norm": 2.2099337577819824, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7422282695770264, + "num_tokens": 152841638.0, + "step": 8736 + }, + { + "epoch": 0.7204584810752865, + "grad_norm": 1.9760640859603882, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7789828777313232, + "num_tokens": 152869432.0, + "step": 8737 + }, + { + "epoch": 0.7205409417003381, + "grad_norm": 2.1877853870391846, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7268990278244019, + "num_tokens": 152894632.0, + "step": 8738 + }, + { + "epoch": 0.7206234023253896, + "grad_norm": 1.871809482574463, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7431074976921082, + "num_tokens": 152925964.0, + "step": 8739 + }, + { + "epoch": 0.7207058629504411, + "grad_norm": 1.9869569540023804, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7604904174804688, + "num_tokens": 152953474.0, + "step": 8740 + }, + { + "epoch": 0.7207883235754927, + "grad_norm": 1.8292629718780518, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7468432784080505, + "num_tokens": 152984752.0, + "step": 8741 + }, + { + "epoch": 0.7208707842005443, + "grad_norm": 1.8693989515304565, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.7628993988037109, + "num_tokens": 153016681.0, + "step": 8742 + }, + { + "epoch": 0.7209532448255958, + "grad_norm": 1.796884298324585, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7654167413711548, + "num_tokens": 153049284.0, + "step": 8743 + }, + { + "epoch": 0.7210357054506473, + "grad_norm": 1.874024510383606, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7568703293800354, + "num_tokens": 153079624.0, + "step": 8744 + }, + { + "epoch": 0.7211181660756989, + "grad_norm": 2.1614878177642822, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.72453773021698, + "num_tokens": 153103869.0, + "step": 8745 + }, + { + "epoch": 0.7212006267007504, + "grad_norm": 1.7992663383483887, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7434669733047485, + "num_tokens": 153136614.0, + "step": 8746 + }, + { + "epoch": 0.7212830873258019, + "grad_norm": 1.8203102350234985, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7258477210998535, + "num_tokens": 153169351.0, + "step": 8747 + }, + { + "epoch": 0.7213655479508535, + "grad_norm": 1.8386921882629395, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7695522308349609, + "num_tokens": 153199078.0, + "step": 8748 + }, + { + "epoch": 0.721448008575905, + "grad_norm": 2.0542330741882324, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7583834528923035, + "num_tokens": 153225857.0, + "step": 8749 + }, + { + "epoch": 0.7215304692009565, + "grad_norm": 2.1501314640045166, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7389525175094604, + "num_tokens": 153250733.0, + "step": 8750 + }, + { + "epoch": 0.7216129298260081, + "grad_norm": 1.931575894355774, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.7047401666641235, + "num_tokens": 153286350.0, + "step": 8751 + }, + { + "epoch": 0.7216953904510596, + "grad_norm": 1.8290785551071167, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7588703036308289, + "num_tokens": 153319594.0, + "step": 8752 + }, + { + "epoch": 0.7217778510761111, + "grad_norm": 2.0260908603668213, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7380514144897461, + "num_tokens": 153345777.0, + "step": 8753 + }, + { + "epoch": 0.7218603117011627, + "grad_norm": 1.8887248039245605, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7693240642547607, + "num_tokens": 153377392.0, + "step": 8754 + }, + { + "epoch": 0.7219427723262143, + "grad_norm": 1.7313320636749268, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7155318260192871, + "num_tokens": 153414609.0, + "step": 8755 + }, + { + "epoch": 0.7220252329512657, + "grad_norm": 2.007901906967163, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.748224139213562, + "num_tokens": 153444166.0, + "step": 8756 + }, + { + "epoch": 0.7221076935763173, + "grad_norm": 1.9346555471420288, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7318838238716125, + "num_tokens": 153475792.0, + "step": 8757 + }, + { + "epoch": 0.7221901542013689, + "grad_norm": 2.0159010887145996, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7124797105789185, + "num_tokens": 153505204.0, + "step": 8758 + }, + { + "epoch": 0.7222726148264204, + "grad_norm": 2.2464230060577393, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7711282968521118, + "num_tokens": 153529324.0, + "step": 8759 + }, + { + "epoch": 0.7223550754514719, + "grad_norm": 2.0636489391326904, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7601221203804016, + "num_tokens": 153557122.0, + "step": 8760 + }, + { + "epoch": 0.7224375360765235, + "grad_norm": 2.038710355758667, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7491048574447632, + "num_tokens": 153585968.0, + "step": 8761 + }, + { + "epoch": 0.722519996701575, + "grad_norm": 2.0639281272888184, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7159039378166199, + "num_tokens": 153615453.0, + "step": 8762 + }, + { + "epoch": 0.7226024573266265, + "grad_norm": 2.016495943069458, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.7623597383499146, + "num_tokens": 153642945.0, + "step": 8763 + }, + { + "epoch": 0.7226849179516781, + "grad_norm": 2.025723457336426, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.734717607498169, + "num_tokens": 153671499.0, + "step": 8764 + }, + { + "epoch": 0.7227673785767296, + "grad_norm": 1.8243472576141357, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7705045938491821, + "num_tokens": 153705492.0, + "step": 8765 + }, + { + "epoch": 0.7228498392017811, + "grad_norm": 1.7721556425094604, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7337582111358643, + "num_tokens": 153740212.0, + "step": 8766 + }, + { + "epoch": 0.7229322998268327, + "grad_norm": 1.8566819429397583, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7388116121292114, + "num_tokens": 153774019.0, + "step": 8767 + }, + { + "epoch": 0.7230147604518842, + "grad_norm": 1.966141700744629, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7618008255958557, + "num_tokens": 153806205.0, + "step": 8768 + }, + { + "epoch": 0.7230972210769357, + "grad_norm": 2.1398959159851074, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7315958738327026, + "num_tokens": 153834553.0, + "step": 8769 + }, + { + "epoch": 0.7231796817019873, + "grad_norm": 1.913554072380066, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7710734605789185, + "num_tokens": 153869609.0, + "step": 8770 + }, + { + "epoch": 0.7232621423270389, + "grad_norm": 2.3926196098327637, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7244635820388794, + "num_tokens": 153890168.0, + "step": 8771 + }, + { + "epoch": 0.7233446029520904, + "grad_norm": 1.8320116996765137, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7762943506240845, + "num_tokens": 153926635.0, + "step": 8772 + }, + { + "epoch": 0.7234270635771419, + "grad_norm": 2.147468090057373, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6928166151046753, + "num_tokens": 153954882.0, + "step": 8773 + }, + { + "epoch": 0.7235095242021935, + "grad_norm": 1.782284140586853, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7361649870872498, + "num_tokens": 153990711.0, + "step": 8774 + }, + { + "epoch": 0.723591984827245, + "grad_norm": 2.3255999088287354, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7105258107185364, + "num_tokens": 154014244.0, + "step": 8775 + }, + { + "epoch": 0.7236744454522965, + "grad_norm": 2.3174662590026855, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.730436384677887, + "num_tokens": 154037467.0, + "step": 8776 + }, + { + "epoch": 0.7237569060773481, + "grad_norm": 2.02451753616333, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7192232012748718, + "num_tokens": 154068966.0, + "step": 8777 + }, + { + "epoch": 0.7238393667023996, + "grad_norm": 1.9713857173919678, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7436730265617371, + "num_tokens": 154098592.0, + "step": 8778 + }, + { + "epoch": 0.7239218273274511, + "grad_norm": 1.8276313543319702, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7693578004837036, + "num_tokens": 154128544.0, + "step": 8779 + }, + { + "epoch": 0.7240042879525027, + "grad_norm": 1.804630160331726, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.7655714750289917, + "num_tokens": 154158645.0, + "step": 8780 + }, + { + "epoch": 0.7240867485775542, + "grad_norm": 1.8369160890579224, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7447173595428467, + "num_tokens": 154192020.0, + "step": 8781 + }, + { + "epoch": 0.7241692092026057, + "grad_norm": 2.16860294342041, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7400000691413879, + "num_tokens": 154218547.0, + "step": 8782 + }, + { + "epoch": 0.7242516698276573, + "grad_norm": 2.092568874359131, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7149434089660645, + "num_tokens": 154248598.0, + "step": 8783 + }, + { + "epoch": 0.7243341304527088, + "grad_norm": 1.7348307371139526, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7357034683227539, + "num_tokens": 154283043.0, + "step": 8784 + }, + { + "epoch": 0.7244165910777604, + "grad_norm": 2.266423225402832, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.7668724656105042, + "num_tokens": 154306503.0, + "step": 8785 + }, + { + "epoch": 0.7244990517028119, + "grad_norm": 1.8959726095199585, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7276815176010132, + "num_tokens": 154339621.0, + "step": 8786 + }, + { + "epoch": 0.7245815123278635, + "grad_norm": 1.784733772277832, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7584959864616394, + "num_tokens": 154375559.0, + "step": 8787 + }, + { + "epoch": 0.724663972952915, + "grad_norm": 2.094778299331665, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7273944616317749, + "num_tokens": 154403171.0, + "step": 8788 + }, + { + "epoch": 0.7247464335779665, + "grad_norm": 1.8860090970993042, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7563437819480896, + "num_tokens": 154433974.0, + "step": 8789 + }, + { + "epoch": 0.7248288942030181, + "grad_norm": 1.8560330867767334, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7480305433273315, + "num_tokens": 154468778.0, + "step": 8790 + }, + { + "epoch": 0.7249113548280696, + "grad_norm": 2.0007641315460205, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7245444059371948, + "num_tokens": 154498738.0, + "step": 8791 + }, + { + "epoch": 0.7249938154531211, + "grad_norm": 1.9212217330932617, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7424632906913757, + "num_tokens": 154529460.0, + "step": 8792 + }, + { + "epoch": 0.7250762760781727, + "grad_norm": 1.95525324344635, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7961248159408569, + "num_tokens": 154555563.0, + "step": 8793 + }, + { + "epoch": 0.7251587367032242, + "grad_norm": 2.2297730445861816, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7269067764282227, + "num_tokens": 154578741.0, + "step": 8794 + }, + { + "epoch": 0.7252411973282757, + "grad_norm": 1.8001561164855957, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7163093090057373, + "num_tokens": 154613646.0, + "step": 8795 + }, + { + "epoch": 0.7253236579533273, + "grad_norm": 2.0870654582977295, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7568138241767883, + "num_tokens": 154638165.0, + "step": 8796 + }, + { + "epoch": 0.7254061185783788, + "grad_norm": 2.0138726234436035, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7253456115722656, + "num_tokens": 154665518.0, + "step": 8797 + }, + { + "epoch": 0.7254885792034304, + "grad_norm": 1.7447465658187866, + "learning_rate": 1e-06, + "loss": 0.5948, + "mean_token_accuracy": 0.7999480962753296, + "num_tokens": 154696788.0, + "step": 8798 + }, + { + "epoch": 0.7255710398284819, + "grad_norm": 1.8961029052734375, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7599838972091675, + "num_tokens": 154729914.0, + "step": 8799 + }, + { + "epoch": 0.7256535004535334, + "grad_norm": 1.7449694871902466, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7890160083770752, + "num_tokens": 154763292.0, + "step": 8800 + }, + { + "epoch": 0.725735961078585, + "grad_norm": 1.916718602180481, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7475115060806274, + "num_tokens": 154793894.0, + "step": 8801 + }, + { + "epoch": 0.7258184217036365, + "grad_norm": 1.7961115837097168, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7759500741958618, + "num_tokens": 154827488.0, + "step": 8802 + }, + { + "epoch": 0.7259008823286881, + "grad_norm": 2.0072457790374756, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7122194170951843, + "num_tokens": 154858159.0, + "step": 8803 + }, + { + "epoch": 0.7259833429537396, + "grad_norm": 1.932673692703247, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.750929594039917, + "num_tokens": 154890822.0, + "step": 8804 + }, + { + "epoch": 0.7260658035787911, + "grad_norm": 1.8252477645874023, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.7819972038269043, + "num_tokens": 154921742.0, + "step": 8805 + }, + { + "epoch": 0.7261482642038427, + "grad_norm": 2.25089430809021, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7270199060440063, + "num_tokens": 154944234.0, + "step": 8806 + }, + { + "epoch": 0.7262307248288942, + "grad_norm": 2.0479023456573486, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7208343148231506, + "num_tokens": 154971834.0, + "step": 8807 + }, + { + "epoch": 0.7263131854539457, + "grad_norm": 2.053812265396118, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7396407723426819, + "num_tokens": 155000208.0, + "step": 8808 + }, + { + "epoch": 0.7263956460789973, + "grad_norm": 2.1831581592559814, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7322062849998474, + "num_tokens": 155026964.0, + "step": 8809 + }, + { + "epoch": 0.7264781067040488, + "grad_norm": 1.926103115081787, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.768867015838623, + "num_tokens": 155054775.0, + "step": 8810 + }, + { + "epoch": 0.7265605673291003, + "grad_norm": 2.1341803073883057, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7512564063072205, + "num_tokens": 155081145.0, + "step": 8811 + }, + { + "epoch": 0.7266430279541519, + "grad_norm": 1.651204228401184, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.779711127281189, + "num_tokens": 155120378.0, + "step": 8812 + }, + { + "epoch": 0.7267254885792034, + "grad_norm": 1.9779845476150513, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7677870988845825, + "num_tokens": 155148311.0, + "step": 8813 + }, + { + "epoch": 0.726807949204255, + "grad_norm": 2.003858804702759, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.773039698600769, + "num_tokens": 155173772.0, + "step": 8814 + }, + { + "epoch": 0.7268904098293065, + "grad_norm": 2.0220084190368652, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7183274030685425, + "num_tokens": 155203466.0, + "step": 8815 + }, + { + "epoch": 0.726972870454358, + "grad_norm": 1.9688127040863037, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7642857432365417, + "num_tokens": 155229546.0, + "step": 8816 + }, + { + "epoch": 0.7270553310794096, + "grad_norm": 1.95039963722229, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7460131645202637, + "num_tokens": 155258916.0, + "step": 8817 + }, + { + "epoch": 0.7271377917044611, + "grad_norm": 1.9930061101913452, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7324216961860657, + "num_tokens": 155286588.0, + "step": 8818 + }, + { + "epoch": 0.7272202523295127, + "grad_norm": 1.9642435312271118, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7205633521080017, + "num_tokens": 155317581.0, + "step": 8819 + }, + { + "epoch": 0.7273027129545642, + "grad_norm": 1.7210103273391724, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7690601348876953, + "num_tokens": 155352548.0, + "step": 8820 + }, + { + "epoch": 0.7273851735796157, + "grad_norm": 1.7902464866638184, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.767976701259613, + "num_tokens": 155383553.0, + "step": 8821 + }, + { + "epoch": 0.7274676342046673, + "grad_norm": 2.0099244117736816, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7600748538970947, + "num_tokens": 155410231.0, + "step": 8822 + }, + { + "epoch": 0.7275500948297188, + "grad_norm": 2.2304186820983887, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7128233909606934, + "num_tokens": 155434875.0, + "step": 8823 + }, + { + "epoch": 0.7276325554547703, + "grad_norm": 2.181208610534668, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7339329719543457, + "num_tokens": 155464582.0, + "step": 8824 + }, + { + "epoch": 0.7277150160798219, + "grad_norm": 1.6748323440551758, + "learning_rate": 1e-06, + "loss": 0.6098, + "mean_token_accuracy": 0.8068251013755798, + "num_tokens": 155500015.0, + "step": 8825 + }, + { + "epoch": 0.7277974767048734, + "grad_norm": 1.8763504028320312, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7431760430335999, + "num_tokens": 155531504.0, + "step": 8826 + }, + { + "epoch": 0.727879937329925, + "grad_norm": 1.8996996879577637, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.77842116355896, + "num_tokens": 155567160.0, + "step": 8827 + }, + { + "epoch": 0.7279623979549765, + "grad_norm": 2.1353073120117188, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7350972294807434, + "num_tokens": 155591630.0, + "step": 8828 + }, + { + "epoch": 0.728044858580028, + "grad_norm": 2.0610768795013428, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7227778434753418, + "num_tokens": 155618637.0, + "step": 8829 + }, + { + "epoch": 0.7281273192050796, + "grad_norm": 2.2182698249816895, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7469239830970764, + "num_tokens": 155642832.0, + "step": 8830 + }, + { + "epoch": 0.7282097798301311, + "grad_norm": 1.8187329769134521, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.786556601524353, + "num_tokens": 155672434.0, + "step": 8831 + }, + { + "epoch": 0.7282922404551827, + "grad_norm": 2.1534037590026855, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7363599538803101, + "num_tokens": 155698012.0, + "step": 8832 + }, + { + "epoch": 0.7283747010802342, + "grad_norm": 2.0049755573272705, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7233974933624268, + "num_tokens": 155726034.0, + "step": 8833 + }, + { + "epoch": 0.7284571617052857, + "grad_norm": 2.1785480976104736, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7428709268569946, + "num_tokens": 155751261.0, + "step": 8834 + }, + { + "epoch": 0.7285396223303373, + "grad_norm": 2.1539804935455322, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7547838091850281, + "num_tokens": 155775533.0, + "step": 8835 + }, + { + "epoch": 0.7286220829553888, + "grad_norm": 1.8105720281600952, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7526865005493164, + "num_tokens": 155810744.0, + "step": 8836 + }, + { + "epoch": 0.7287045435804403, + "grad_norm": 1.9384888410568237, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7586138844490051, + "num_tokens": 155839928.0, + "step": 8837 + }, + { + "epoch": 0.7287870042054919, + "grad_norm": 1.9350117444992065, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7538325190544128, + "num_tokens": 155869032.0, + "step": 8838 + }, + { + "epoch": 0.7288694648305434, + "grad_norm": 1.8576432466506958, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7874714136123657, + "num_tokens": 155901265.0, + "step": 8839 + }, + { + "epoch": 0.728951925455595, + "grad_norm": 2.0038113594055176, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7527260780334473, + "num_tokens": 155930370.0, + "step": 8840 + }, + { + "epoch": 0.7290343860806465, + "grad_norm": 1.775468349456787, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7533196806907654, + "num_tokens": 155966346.0, + "step": 8841 + }, + { + "epoch": 0.729116846705698, + "grad_norm": 2.0015907287597656, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7555147409439087, + "num_tokens": 155995477.0, + "step": 8842 + }, + { + "epoch": 0.7291993073307496, + "grad_norm": 1.9808210134506226, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7498469352722168, + "num_tokens": 156027755.0, + "step": 8843 + }, + { + "epoch": 0.7292817679558011, + "grad_norm": 1.8504905700683594, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7492707371711731, + "num_tokens": 156058564.0, + "step": 8844 + }, + { + "epoch": 0.7293642285808526, + "grad_norm": 1.945861577987671, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.750119686126709, + "num_tokens": 156092523.0, + "step": 8845 + }, + { + "epoch": 0.7294466892059042, + "grad_norm": 1.8893729448318481, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7418228387832642, + "num_tokens": 156125826.0, + "step": 8846 + }, + { + "epoch": 0.7295291498309557, + "grad_norm": 1.7943274974822998, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7478550672531128, + "num_tokens": 156160967.0, + "step": 8847 + }, + { + "epoch": 0.7296116104560073, + "grad_norm": 1.93734610080719, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7463555335998535, + "num_tokens": 156191605.0, + "step": 8848 + }, + { + "epoch": 0.7296940710810588, + "grad_norm": 2.1580629348754883, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7357337474822998, + "num_tokens": 156216581.0, + "step": 8849 + }, + { + "epoch": 0.7297765317061103, + "grad_norm": 1.8899348974227905, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7462822198867798, + "num_tokens": 156246024.0, + "step": 8850 + }, + { + "epoch": 0.7298589923311619, + "grad_norm": 1.9472886323928833, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7213684916496277, + "num_tokens": 156276553.0, + "step": 8851 + }, + { + "epoch": 0.7299414529562134, + "grad_norm": 1.7729971408843994, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.729931116104126, + "num_tokens": 156310963.0, + "step": 8852 + }, + { + "epoch": 0.730023913581265, + "grad_norm": 2.1006791591644287, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.730446457862854, + "num_tokens": 156337064.0, + "step": 8853 + }, + { + "epoch": 0.7301063742063165, + "grad_norm": 1.7491782903671265, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.7858052253723145, + "num_tokens": 156369106.0, + "step": 8854 + }, + { + "epoch": 0.730188834831368, + "grad_norm": 1.8464555740356445, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7766540050506592, + "num_tokens": 156401980.0, + "step": 8855 + }, + { + "epoch": 0.7302712954564196, + "grad_norm": 1.890541911125183, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7463160753250122, + "num_tokens": 156432999.0, + "step": 8856 + }, + { + "epoch": 0.7303537560814711, + "grad_norm": 1.8820620775222778, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7562609910964966, + "num_tokens": 156463421.0, + "step": 8857 + }, + { + "epoch": 0.7304362167065226, + "grad_norm": 1.7563778162002563, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7593400478363037, + "num_tokens": 156497069.0, + "step": 8858 + }, + { + "epoch": 0.7305186773315742, + "grad_norm": 2.1856560707092285, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7343284487724304, + "num_tokens": 156523862.0, + "step": 8859 + }, + { + "epoch": 0.7306011379566257, + "grad_norm": 1.9215686321258545, + "learning_rate": 1e-06, + "loss": 0.7698, + "mean_token_accuracy": 0.7573521733283997, + "num_tokens": 156551754.0, + "step": 8860 + }, + { + "epoch": 0.7306835985816772, + "grad_norm": 1.8346190452575684, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.7431452870368958, + "num_tokens": 156584170.0, + "step": 8861 + }, + { + "epoch": 0.7307660592067288, + "grad_norm": 1.9331974983215332, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7665202021598816, + "num_tokens": 156613961.0, + "step": 8862 + }, + { + "epoch": 0.7308485198317803, + "grad_norm": 1.9274604320526123, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7732930183410645, + "num_tokens": 156641310.0, + "step": 8863 + }, + { + "epoch": 0.7309309804568319, + "grad_norm": 1.9996116161346436, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7507082223892212, + "num_tokens": 156671667.0, + "step": 8864 + }, + { + "epoch": 0.7310134410818834, + "grad_norm": 1.7443335056304932, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7779169082641602, + "num_tokens": 156706567.0, + "step": 8865 + }, + { + "epoch": 0.731095901706935, + "grad_norm": 2.0916237831115723, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7352844476699829, + "num_tokens": 156735878.0, + "step": 8866 + }, + { + "epoch": 0.7311783623319865, + "grad_norm": 2.5166685581207275, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7394525408744812, + "num_tokens": 156755530.0, + "step": 8867 + }, + { + "epoch": 0.731260822957038, + "grad_norm": 1.9844560623168945, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7169387340545654, + "num_tokens": 156784809.0, + "step": 8868 + }, + { + "epoch": 0.7313432835820896, + "grad_norm": 1.8995862007141113, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.736779510974884, + "num_tokens": 156815665.0, + "step": 8869 + }, + { + "epoch": 0.7314257442071411, + "grad_norm": 1.9182473421096802, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7740644216537476, + "num_tokens": 156844268.0, + "step": 8870 + }, + { + "epoch": 0.7315082048321926, + "grad_norm": 1.8079031705856323, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.749836802482605, + "num_tokens": 156878937.0, + "step": 8871 + }, + { + "epoch": 0.7315906654572442, + "grad_norm": 2.3224170207977295, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7514158487319946, + "num_tokens": 156900018.0, + "step": 8872 + }, + { + "epoch": 0.7316731260822957, + "grad_norm": 1.7656261920928955, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7408030033111572, + "num_tokens": 156933401.0, + "step": 8873 + }, + { + "epoch": 0.7317555867073472, + "grad_norm": 1.9407349824905396, + "learning_rate": 1e-06, + "loss": 0.7637, + "mean_token_accuracy": 0.7611478567123413, + "num_tokens": 156961709.0, + "step": 8874 + }, + { + "epoch": 0.7318380473323988, + "grad_norm": 2.0324008464813232, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.715111255645752, + "num_tokens": 156991633.0, + "step": 8875 + }, + { + "epoch": 0.7319205079574503, + "grad_norm": 1.6724579334259033, + "learning_rate": 1e-06, + "loss": 0.5434, + "mean_token_accuracy": 0.8206621408462524, + "num_tokens": 157023792.0, + "step": 8876 + }, + { + "epoch": 0.7320029685825018, + "grad_norm": 1.9881808757781982, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7507327198982239, + "num_tokens": 157052184.0, + "step": 8877 + }, + { + "epoch": 0.7320854292075534, + "grad_norm": 1.8698533773422241, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7406752109527588, + "num_tokens": 157083969.0, + "step": 8878 + }, + { + "epoch": 0.732167889832605, + "grad_norm": 1.910003662109375, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7823545932769775, + "num_tokens": 157114402.0, + "step": 8879 + }, + { + "epoch": 0.7322503504576565, + "grad_norm": 1.9461100101470947, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7452307939529419, + "num_tokens": 157144571.0, + "step": 8880 + }, + { + "epoch": 0.732332811082708, + "grad_norm": 1.9969335794448853, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7556729316711426, + "num_tokens": 157171613.0, + "step": 8881 + }, + { + "epoch": 0.7324152717077596, + "grad_norm": 1.8873423337936401, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.767333984375, + "num_tokens": 157204984.0, + "step": 8882 + }, + { + "epoch": 0.7324977323328111, + "grad_norm": 2.051281452178955, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7534931898117065, + "num_tokens": 157232565.0, + "step": 8883 + }, + { + "epoch": 0.7325801929578626, + "grad_norm": 1.931978702545166, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.723438560962677, + "num_tokens": 157262930.0, + "step": 8884 + }, + { + "epoch": 0.7326626535829142, + "grad_norm": 1.8860000371932983, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7183976769447327, + "num_tokens": 157294627.0, + "step": 8885 + }, + { + "epoch": 0.7327451142079657, + "grad_norm": 1.8585611581802368, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7586835026741028, + "num_tokens": 157326320.0, + "step": 8886 + }, + { + "epoch": 0.7328275748330172, + "grad_norm": 2.3312549591064453, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7446632385253906, + "num_tokens": 157347721.0, + "step": 8887 + }, + { + "epoch": 0.7329100354580688, + "grad_norm": 2.003441333770752, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7525949478149414, + "num_tokens": 157376509.0, + "step": 8888 + }, + { + "epoch": 0.7329924960831203, + "grad_norm": 2.1763017177581787, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.7537710666656494, + "num_tokens": 157400604.0, + "step": 8889 + }, + { + "epoch": 0.7330749567081718, + "grad_norm": 1.7533503770828247, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7801334857940674, + "num_tokens": 157435583.0, + "step": 8890 + }, + { + "epoch": 0.7331574173332234, + "grad_norm": 1.9466251134872437, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7509288787841797, + "num_tokens": 157467799.0, + "step": 8891 + }, + { + "epoch": 0.7332398779582749, + "grad_norm": 1.8957520723342896, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.757354199886322, + "num_tokens": 157500485.0, + "step": 8892 + }, + { + "epoch": 0.7333223385833265, + "grad_norm": 2.023908853530884, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7671687602996826, + "num_tokens": 157527954.0, + "step": 8893 + }, + { + "epoch": 0.733404799208378, + "grad_norm": 2.0135297775268555, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7529475688934326, + "num_tokens": 157558105.0, + "step": 8894 + }, + { + "epoch": 0.7334872598334296, + "grad_norm": 1.911848783493042, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7610653042793274, + "num_tokens": 157592699.0, + "step": 8895 + }, + { + "epoch": 0.7335697204584811, + "grad_norm": 2.4998300075531006, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7439336776733398, + "num_tokens": 157612714.0, + "step": 8896 + }, + { + "epoch": 0.7336521810835326, + "grad_norm": 2.0682899951934814, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7772019505500793, + "num_tokens": 157639378.0, + "step": 8897 + }, + { + "epoch": 0.7337346417085842, + "grad_norm": 1.5738699436187744, + "learning_rate": 1e-06, + "loss": 0.5451, + "mean_token_accuracy": 0.8210231065750122, + "num_tokens": 157677949.0, + "step": 8898 + }, + { + "epoch": 0.7338171023336357, + "grad_norm": 1.8256852626800537, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7235107421875, + "num_tokens": 157712988.0, + "step": 8899 + }, + { + "epoch": 0.7338995629586872, + "grad_norm": 1.9744764566421509, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7659215927124023, + "num_tokens": 157746524.0, + "step": 8900 + }, + { + "epoch": 0.7339820235837388, + "grad_norm": 1.9926460981369019, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7728512287139893, + "num_tokens": 157774367.0, + "step": 8901 + }, + { + "epoch": 0.7340644842087903, + "grad_norm": 2.0517356395721436, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7767785787582397, + "num_tokens": 157802003.0, + "step": 8902 + }, + { + "epoch": 0.7341469448338418, + "grad_norm": 1.762387752532959, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.7924137711524963, + "num_tokens": 157836129.0, + "step": 8903 + }, + { + "epoch": 0.7342294054588934, + "grad_norm": 1.8963825702667236, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7512369155883789, + "num_tokens": 157866344.0, + "step": 8904 + }, + { + "epoch": 0.7343118660839449, + "grad_norm": 1.8629751205444336, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7537269592285156, + "num_tokens": 157899546.0, + "step": 8905 + }, + { + "epoch": 0.7343943267089964, + "grad_norm": 2.236351251602173, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.708227276802063, + "num_tokens": 157925738.0, + "step": 8906 + }, + { + "epoch": 0.734476787334048, + "grad_norm": 2.322688102722168, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7435188293457031, + "num_tokens": 157948766.0, + "step": 8907 + }, + { + "epoch": 0.7345592479590995, + "grad_norm": 1.976776123046875, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7590317726135254, + "num_tokens": 157979325.0, + "step": 8908 + }, + { + "epoch": 0.7346417085841511, + "grad_norm": 2.2243874073028564, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7248281836509705, + "num_tokens": 158005258.0, + "step": 8909 + }, + { + "epoch": 0.7347241692092026, + "grad_norm": 2.5030980110168457, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7075380086898804, + "num_tokens": 158026244.0, + "step": 8910 + }, + { + "epoch": 0.7348066298342542, + "grad_norm": 2.009422779083252, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7350088357925415, + "num_tokens": 158056109.0, + "step": 8911 + }, + { + "epoch": 0.7348890904593057, + "grad_norm": 1.9176297187805176, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.768877387046814, + "num_tokens": 158087061.0, + "step": 8912 + }, + { + "epoch": 0.7349715510843572, + "grad_norm": 1.8310425281524658, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7530227899551392, + "num_tokens": 158122033.0, + "step": 8913 + }, + { + "epoch": 0.7350540117094088, + "grad_norm": 2.11814022064209, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7251234650611877, + "num_tokens": 158148010.0, + "step": 8914 + }, + { + "epoch": 0.7351364723344603, + "grad_norm": 2.0425429344177246, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7126199007034302, + "num_tokens": 158178071.0, + "step": 8915 + }, + { + "epoch": 0.7352189329595118, + "grad_norm": 1.939417839050293, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.772864580154419, + "num_tokens": 158208924.0, + "step": 8916 + }, + { + "epoch": 0.7353013935845634, + "grad_norm": 2.0803756713867188, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7621023654937744, + "num_tokens": 158234826.0, + "step": 8917 + }, + { + "epoch": 0.7353838542096149, + "grad_norm": 1.9220911264419556, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7625957727432251, + "num_tokens": 158267057.0, + "step": 8918 + }, + { + "epoch": 0.7354663148346664, + "grad_norm": 1.7791926860809326, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7677947282791138, + "num_tokens": 158302540.0, + "step": 8919 + }, + { + "epoch": 0.735548775459718, + "grad_norm": 2.092236042022705, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7368931770324707, + "num_tokens": 158331417.0, + "step": 8920 + }, + { + "epoch": 0.7356312360847695, + "grad_norm": 2.037236452102661, + "learning_rate": 1e-06, + "loss": 0.7656, + "mean_token_accuracy": 0.7463663816452026, + "num_tokens": 158358293.0, + "step": 8921 + }, + { + "epoch": 0.735713696709821, + "grad_norm": 1.6143192052841187, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7807766795158386, + "num_tokens": 158398140.0, + "step": 8922 + }, + { + "epoch": 0.7357961573348726, + "grad_norm": 1.9425177574157715, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7360820770263672, + "num_tokens": 158427701.0, + "step": 8923 + }, + { + "epoch": 0.7358786179599242, + "grad_norm": 1.876175880432129, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7484847903251648, + "num_tokens": 158457688.0, + "step": 8924 + }, + { + "epoch": 0.7359610785849757, + "grad_norm": 2.1483540534973145, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7523351907730103, + "num_tokens": 158481606.0, + "step": 8925 + }, + { + "epoch": 0.7360435392100272, + "grad_norm": 2.3323724269866943, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.759340763092041, + "num_tokens": 158502479.0, + "step": 8926 + }, + { + "epoch": 0.7361259998350788, + "grad_norm": 1.7699865102767944, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7641832828521729, + "num_tokens": 158536074.0, + "step": 8927 + }, + { + "epoch": 0.7362084604601303, + "grad_norm": 1.9063516855239868, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7680400013923645, + "num_tokens": 158567859.0, + "step": 8928 + }, + { + "epoch": 0.7362909210851818, + "grad_norm": 2.0686140060424805, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.7836762070655823, + "num_tokens": 158595135.0, + "step": 8929 + }, + { + "epoch": 0.7363733817102334, + "grad_norm": 1.898685336112976, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7556296586990356, + "num_tokens": 158625991.0, + "step": 8930 + }, + { + "epoch": 0.7364558423352849, + "grad_norm": 2.0920045375823975, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7298593521118164, + "num_tokens": 158652157.0, + "step": 8931 + }, + { + "epoch": 0.7365383029603364, + "grad_norm": 2.004323720932007, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7410304546356201, + "num_tokens": 158681960.0, + "step": 8932 + }, + { + "epoch": 0.736620763585388, + "grad_norm": 1.8679579496383667, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7323918342590332, + "num_tokens": 158713945.0, + "step": 8933 + }, + { + "epoch": 0.7367032242104395, + "grad_norm": 2.1817710399627686, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.746135413646698, + "num_tokens": 158736683.0, + "step": 8934 + }, + { + "epoch": 0.736785684835491, + "grad_norm": 1.890317678451538, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7423223257064819, + "num_tokens": 158766722.0, + "step": 8935 + }, + { + "epoch": 0.7368681454605426, + "grad_norm": 1.7540189027786255, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.738337516784668, + "num_tokens": 158803212.0, + "step": 8936 + }, + { + "epoch": 0.7369506060855942, + "grad_norm": 1.912208914756775, + "learning_rate": 1e-06, + "loss": 0.54, + "mean_token_accuracy": 0.8111205101013184, + "num_tokens": 158829323.0, + "step": 8937 + }, + { + "epoch": 0.7370330667106456, + "grad_norm": 1.918465256690979, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.8011900186538696, + "num_tokens": 158859738.0, + "step": 8938 + }, + { + "epoch": 0.7371155273356972, + "grad_norm": 2.0954294204711914, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.7625296115875244, + "num_tokens": 158885210.0, + "step": 8939 + }, + { + "epoch": 0.7371979879607488, + "grad_norm": 1.8144258260726929, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7322092056274414, + "num_tokens": 158919066.0, + "step": 8940 + }, + { + "epoch": 0.7372804485858003, + "grad_norm": 1.7201476097106934, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7769855856895447, + "num_tokens": 158951625.0, + "step": 8941 + }, + { + "epoch": 0.7373629092108518, + "grad_norm": 1.7582296133041382, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7740408182144165, + "num_tokens": 158984920.0, + "step": 8942 + }, + { + "epoch": 0.7374453698359034, + "grad_norm": 1.886095643043518, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.750036895275116, + "num_tokens": 159017051.0, + "step": 8943 + }, + { + "epoch": 0.7375278304609549, + "grad_norm": 1.816068172454834, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7620161771774292, + "num_tokens": 159049173.0, + "step": 8944 + }, + { + "epoch": 0.7376102910860064, + "grad_norm": 2.007905960083008, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7141516208648682, + "num_tokens": 159079332.0, + "step": 8945 + }, + { + "epoch": 0.737692751711058, + "grad_norm": 2.0039007663726807, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7653011679649353, + "num_tokens": 159111526.0, + "step": 8946 + }, + { + "epoch": 0.7377752123361095, + "grad_norm": 2.084385633468628, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7335506677627563, + "num_tokens": 159137547.0, + "step": 8947 + }, + { + "epoch": 0.737857672961161, + "grad_norm": 1.775144338607788, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.7952448725700378, + "num_tokens": 159170918.0, + "step": 8948 + }, + { + "epoch": 0.7379401335862126, + "grad_norm": 1.9739989042282104, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7708480358123779, + "num_tokens": 159202037.0, + "step": 8949 + }, + { + "epoch": 0.7380225942112641, + "grad_norm": 2.1058127880096436, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.728447675704956, + "num_tokens": 159229178.0, + "step": 8950 + }, + { + "epoch": 0.7381050548363156, + "grad_norm": 2.1877353191375732, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7070000767707825, + "num_tokens": 159256403.0, + "step": 8951 + }, + { + "epoch": 0.7381875154613672, + "grad_norm": 2.109485387802124, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7216050624847412, + "num_tokens": 159282464.0, + "step": 8952 + }, + { + "epoch": 0.7382699760864188, + "grad_norm": 1.9107632637023926, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.7897075414657593, + "num_tokens": 159312088.0, + "step": 8953 + }, + { + "epoch": 0.7383524367114703, + "grad_norm": 1.887188196182251, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7441012859344482, + "num_tokens": 159345076.0, + "step": 8954 + }, + { + "epoch": 0.7384348973365218, + "grad_norm": 1.9917727708816528, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7871052026748657, + "num_tokens": 159372410.0, + "step": 8955 + }, + { + "epoch": 0.7385173579615734, + "grad_norm": 2.1615405082702637, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7727491855621338, + "num_tokens": 159396472.0, + "step": 8956 + }, + { + "epoch": 0.7385998185866249, + "grad_norm": 1.8226566314697266, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7660088539123535, + "num_tokens": 159429741.0, + "step": 8957 + }, + { + "epoch": 0.7386822792116764, + "grad_norm": 1.7637455463409424, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7704802751541138, + "num_tokens": 159465526.0, + "step": 8958 + }, + { + "epoch": 0.738764739836728, + "grad_norm": 1.9153105020523071, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7451916933059692, + "num_tokens": 159496221.0, + "step": 8959 + }, + { + "epoch": 0.7388472004617795, + "grad_norm": 1.9940967559814453, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7635130286216736, + "num_tokens": 159524512.0, + "step": 8960 + }, + { + "epoch": 0.738929661086831, + "grad_norm": 2.026258945465088, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7548767328262329, + "num_tokens": 159554784.0, + "step": 8961 + }, + { + "epoch": 0.7390121217118826, + "grad_norm": 2.0731983184814453, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7218464016914368, + "num_tokens": 159585381.0, + "step": 8962 + }, + { + "epoch": 0.7390945823369341, + "grad_norm": 1.9114806652069092, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7258713245391846, + "num_tokens": 159615686.0, + "step": 8963 + }, + { + "epoch": 0.7391770429619856, + "grad_norm": 1.7985835075378418, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7598127722740173, + "num_tokens": 159651265.0, + "step": 8964 + }, + { + "epoch": 0.7392595035870372, + "grad_norm": 2.0736382007598877, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7348368763923645, + "num_tokens": 159678402.0, + "step": 8965 + }, + { + "epoch": 0.7393419642120888, + "grad_norm": 1.8031256198883057, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7648388147354126, + "num_tokens": 159709907.0, + "step": 8966 + }, + { + "epoch": 0.7394244248371402, + "grad_norm": 2.033019781112671, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7797640562057495, + "num_tokens": 159735844.0, + "step": 8967 + }, + { + "epoch": 0.7395068854621918, + "grad_norm": 1.9874370098114014, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7410191297531128, + "num_tokens": 159764580.0, + "step": 8968 + }, + { + "epoch": 0.7395893460872434, + "grad_norm": 1.8216990232467651, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7735558152198792, + "num_tokens": 159795970.0, + "step": 8969 + }, + { + "epoch": 0.7396718067122949, + "grad_norm": 2.051780939102173, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.770043134689331, + "num_tokens": 159823277.0, + "step": 8970 + }, + { + "epoch": 0.7397542673373464, + "grad_norm": 1.669934868812561, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.749710738658905, + "num_tokens": 159860989.0, + "step": 8971 + }, + { + "epoch": 0.739836727962398, + "grad_norm": 1.9468532800674438, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7477353811264038, + "num_tokens": 159891618.0, + "step": 8972 + }, + { + "epoch": 0.7399191885874495, + "grad_norm": 2.027677536010742, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7499714493751526, + "num_tokens": 159917678.0, + "step": 8973 + }, + { + "epoch": 0.740001649212501, + "grad_norm": 1.566999912261963, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7827049493789673, + "num_tokens": 159957698.0, + "step": 8974 + }, + { + "epoch": 0.7400841098375526, + "grad_norm": 1.875193476676941, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7394524216651917, + "num_tokens": 159991613.0, + "step": 8975 + }, + { + "epoch": 0.7401665704626041, + "grad_norm": 1.8328936100006104, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.786117672920227, + "num_tokens": 160020055.0, + "step": 8976 + }, + { + "epoch": 0.7402490310876556, + "grad_norm": 1.8230754137039185, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7106747627258301, + "num_tokens": 160056389.0, + "step": 8977 + }, + { + "epoch": 0.7403314917127072, + "grad_norm": 1.8000885248184204, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7585978507995605, + "num_tokens": 160090740.0, + "step": 8978 + }, + { + "epoch": 0.7404139523377588, + "grad_norm": 2.250549793243408, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7301676273345947, + "num_tokens": 160115200.0, + "step": 8979 + }, + { + "epoch": 0.7404964129628102, + "grad_norm": 2.148933172225952, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.740114688873291, + "num_tokens": 160139315.0, + "step": 8980 + }, + { + "epoch": 0.7405788735878618, + "grad_norm": 2.1475038528442383, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7679703235626221, + "num_tokens": 160161677.0, + "step": 8981 + }, + { + "epoch": 0.7406613342129134, + "grad_norm": 2.1727192401885986, + "learning_rate": 1e-06, + "loss": 0.7746, + "mean_token_accuracy": 0.7586371898651123, + "num_tokens": 160187660.0, + "step": 8982 + }, + { + "epoch": 0.7407437948379648, + "grad_norm": 1.8414051532745361, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7784981727600098, + "num_tokens": 160217323.0, + "step": 8983 + }, + { + "epoch": 0.7408262554630164, + "grad_norm": 1.888432502746582, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7531688809394836, + "num_tokens": 160249927.0, + "step": 8984 + }, + { + "epoch": 0.740908716088068, + "grad_norm": 1.9776182174682617, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7492941617965698, + "num_tokens": 160280196.0, + "step": 8985 + }, + { + "epoch": 0.7409911767131195, + "grad_norm": 1.6687852144241333, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7830934524536133, + "num_tokens": 160316745.0, + "step": 8986 + }, + { + "epoch": 0.741073637338171, + "grad_norm": 2.002662420272827, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7439982891082764, + "num_tokens": 160343776.0, + "step": 8987 + }, + { + "epoch": 0.7411560979632226, + "grad_norm": 1.5975548028945923, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7835679054260254, + "num_tokens": 160385317.0, + "step": 8988 + }, + { + "epoch": 0.7412385585882741, + "grad_norm": 2.179016590118408, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7333110570907593, + "num_tokens": 160411147.0, + "step": 8989 + }, + { + "epoch": 0.7413210192133256, + "grad_norm": 2.2773451805114746, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7124829292297363, + "num_tokens": 160435757.0, + "step": 8990 + }, + { + "epoch": 0.7414034798383772, + "grad_norm": 1.9360932111740112, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.759340226650238, + "num_tokens": 160467591.0, + "step": 8991 + }, + { + "epoch": 0.7414859404634287, + "grad_norm": 1.8761398792266846, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7636754512786865, + "num_tokens": 160501270.0, + "step": 8992 + }, + { + "epoch": 0.7415684010884802, + "grad_norm": 2.0902721881866455, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6898980140686035, + "num_tokens": 160531031.0, + "step": 8993 + }, + { + "epoch": 0.7416508617135318, + "grad_norm": 2.1741981506347656, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7585748434066772, + "num_tokens": 160556332.0, + "step": 8994 + }, + { + "epoch": 0.7417333223385834, + "grad_norm": 1.9416694641113281, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7540227174758911, + "num_tokens": 160587240.0, + "step": 8995 + }, + { + "epoch": 0.7418157829636348, + "grad_norm": 2.0618972778320312, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7415369749069214, + "num_tokens": 160617897.0, + "step": 8996 + }, + { + "epoch": 0.7418982435886864, + "grad_norm": 1.917732834815979, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7549614906311035, + "num_tokens": 160646191.0, + "step": 8997 + }, + { + "epoch": 0.741980704213738, + "grad_norm": 2.115610122680664, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.757927656173706, + "num_tokens": 160673899.0, + "step": 8998 + }, + { + "epoch": 0.7420631648387894, + "grad_norm": 2.058666944503784, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.7858270406723022, + "num_tokens": 160698838.0, + "step": 8999 + }, + { + "epoch": 0.742145625463841, + "grad_norm": 2.0360424518585205, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7134749889373779, + "num_tokens": 160727294.0, + "step": 9000 + }, + { + "epoch": 0.7422280860888926, + "grad_norm": 2.129053831100464, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7637284994125366, + "num_tokens": 160753269.0, + "step": 9001 + }, + { + "epoch": 0.7423105467139441, + "grad_norm": 1.9386671781539917, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7546439170837402, + "num_tokens": 160782320.0, + "step": 9002 + }, + { + "epoch": 0.7423930073389956, + "grad_norm": 1.8865809440612793, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7836415767669678, + "num_tokens": 160812864.0, + "step": 9003 + }, + { + "epoch": 0.7424754679640472, + "grad_norm": 1.8410576581954956, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7249634265899658, + "num_tokens": 160849703.0, + "step": 9004 + }, + { + "epoch": 0.7425579285890987, + "grad_norm": 2.0280582904815674, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7532353401184082, + "num_tokens": 160876434.0, + "step": 9005 + }, + { + "epoch": 0.7426403892141502, + "grad_norm": 2.033843755722046, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7766886949539185, + "num_tokens": 160905582.0, + "step": 9006 + }, + { + "epoch": 0.7427228498392018, + "grad_norm": 2.1252353191375732, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7341335415840149, + "num_tokens": 160932838.0, + "step": 9007 + }, + { + "epoch": 0.7428053104642534, + "grad_norm": 2.0338873863220215, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7500543594360352, + "num_tokens": 160960046.0, + "step": 9008 + }, + { + "epoch": 0.7428877710893048, + "grad_norm": 2.0825581550598145, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7520319223403931, + "num_tokens": 160990594.0, + "step": 9009 + }, + { + "epoch": 0.7429702317143564, + "grad_norm": 1.8950527906417847, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.742408275604248, + "num_tokens": 161020448.0, + "step": 9010 + }, + { + "epoch": 0.743052692339408, + "grad_norm": 2.0305838584899902, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7398494482040405, + "num_tokens": 161048295.0, + "step": 9011 + }, + { + "epoch": 0.7431351529644594, + "grad_norm": 1.8519772291183472, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7715420722961426, + "num_tokens": 161081449.0, + "step": 9012 + }, + { + "epoch": 0.743217613589511, + "grad_norm": 1.7584452629089355, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7687503099441528, + "num_tokens": 161116460.0, + "step": 9013 + }, + { + "epoch": 0.7433000742145626, + "grad_norm": 1.866018295288086, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7377218008041382, + "num_tokens": 161149604.0, + "step": 9014 + }, + { + "epoch": 0.7433825348396141, + "grad_norm": 1.8948510885238647, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.7835047245025635, + "num_tokens": 161179436.0, + "step": 9015 + }, + { + "epoch": 0.7434649954646656, + "grad_norm": 1.911295771598816, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7525303959846497, + "num_tokens": 161213364.0, + "step": 9016 + }, + { + "epoch": 0.7435474560897172, + "grad_norm": 1.743836760520935, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7664231657981873, + "num_tokens": 161246619.0, + "step": 9017 + }, + { + "epoch": 0.7436299167147687, + "grad_norm": 2.3826184272766113, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7531604170799255, + "num_tokens": 161275604.0, + "step": 9018 + }, + { + "epoch": 0.7437123773398202, + "grad_norm": 1.9891221523284912, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.7853496074676514, + "num_tokens": 161304013.0, + "step": 9019 + }, + { + "epoch": 0.7437948379648718, + "grad_norm": 1.7949498891830444, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7491888999938965, + "num_tokens": 161339557.0, + "step": 9020 + }, + { + "epoch": 0.7438772985899234, + "grad_norm": 1.9659485816955566, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7250214219093323, + "num_tokens": 161368621.0, + "step": 9021 + }, + { + "epoch": 0.7439597592149748, + "grad_norm": 1.5892398357391357, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7794487476348877, + "num_tokens": 161410154.0, + "step": 9022 + }, + { + "epoch": 0.7440422198400264, + "grad_norm": 1.7474489212036133, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.7816789150238037, + "num_tokens": 161442604.0, + "step": 9023 + }, + { + "epoch": 0.744124680465078, + "grad_norm": 2.1084389686584473, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7726832628250122, + "num_tokens": 161469649.0, + "step": 9024 + }, + { + "epoch": 0.7442071410901294, + "grad_norm": 2.0416557788848877, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.749755859375, + "num_tokens": 161497378.0, + "step": 9025 + }, + { + "epoch": 0.744289601715181, + "grad_norm": 2.0977349281311035, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7522311210632324, + "num_tokens": 161523826.0, + "step": 9026 + }, + { + "epoch": 0.7443720623402326, + "grad_norm": 1.789803385734558, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.7926923036575317, + "num_tokens": 161554910.0, + "step": 9027 + }, + { + "epoch": 0.744454522965284, + "grad_norm": 1.9278761148452759, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7460768818855286, + "num_tokens": 161586267.0, + "step": 9028 + }, + { + "epoch": 0.7445369835903356, + "grad_norm": 1.960616111755371, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7550528049468994, + "num_tokens": 161615616.0, + "step": 9029 + }, + { + "epoch": 0.7446194442153872, + "grad_norm": 1.8816901445388794, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7419223189353943, + "num_tokens": 161647389.0, + "step": 9030 + }, + { + "epoch": 0.7447019048404387, + "grad_norm": 1.8465495109558105, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.7902257442474365, + "num_tokens": 161681704.0, + "step": 9031 + }, + { + "epoch": 0.7447843654654902, + "grad_norm": 1.9221785068511963, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7256011962890625, + "num_tokens": 161713284.0, + "step": 9032 + }, + { + "epoch": 0.7448668260905418, + "grad_norm": 1.895801067352295, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7441086769104004, + "num_tokens": 161744675.0, + "step": 9033 + }, + { + "epoch": 0.7449492867155934, + "grad_norm": 1.5755618810653687, + "learning_rate": 1e-06, + "loss": 0.5495, + "mean_token_accuracy": 0.8237391114234924, + "num_tokens": 161783679.0, + "step": 9034 + }, + { + "epoch": 0.7450317473406448, + "grad_norm": 1.8838844299316406, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7803652286529541, + "num_tokens": 161816084.0, + "step": 9035 + }, + { + "epoch": 0.7451142079656964, + "grad_norm": 2.083554267883301, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.7636037468910217, + "num_tokens": 161840592.0, + "step": 9036 + }, + { + "epoch": 0.745196668590748, + "grad_norm": 1.9324332475662231, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7674826383590698, + "num_tokens": 161872925.0, + "step": 9037 + }, + { + "epoch": 0.7452791292157994, + "grad_norm": 1.834650993347168, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7370705604553223, + "num_tokens": 161909467.0, + "step": 9038 + }, + { + "epoch": 0.745361589840851, + "grad_norm": 1.7522855997085571, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7414047718048096, + "num_tokens": 161946978.0, + "step": 9039 + }, + { + "epoch": 0.7454440504659026, + "grad_norm": 1.7225266695022583, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.7927348017692566, + "num_tokens": 161982107.0, + "step": 9040 + }, + { + "epoch": 0.745526511090954, + "grad_norm": 1.9684507846832275, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7591465711593628, + "num_tokens": 162013655.0, + "step": 9041 + }, + { + "epoch": 0.7456089717160056, + "grad_norm": 2.021784782409668, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.76280277967453, + "num_tokens": 162041361.0, + "step": 9042 + }, + { + "epoch": 0.7456914323410572, + "grad_norm": 1.9304696321487427, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7891340255737305, + "num_tokens": 162072480.0, + "step": 9043 + }, + { + "epoch": 0.7457738929661086, + "grad_norm": 1.9115917682647705, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7350181341171265, + "num_tokens": 162102643.0, + "step": 9044 + }, + { + "epoch": 0.7458563535911602, + "grad_norm": 1.902774453163147, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.745276927947998, + "num_tokens": 162131395.0, + "step": 9045 + }, + { + "epoch": 0.7459388142162118, + "grad_norm": 1.877805471420288, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7606037855148315, + "num_tokens": 162164048.0, + "step": 9046 + }, + { + "epoch": 0.7460212748412633, + "grad_norm": 1.815557599067688, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7150514125823975, + "num_tokens": 162198553.0, + "step": 9047 + }, + { + "epoch": 0.7461037354663148, + "grad_norm": 1.7400012016296387, + "learning_rate": 1e-06, + "loss": 0.5755, + "mean_token_accuracy": 0.8006342649459839, + "num_tokens": 162231730.0, + "step": 9048 + }, + { + "epoch": 0.7461861960913664, + "grad_norm": 1.984183430671692, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7247279286384583, + "num_tokens": 162262777.0, + "step": 9049 + }, + { + "epoch": 0.746268656716418, + "grad_norm": 2.2089178562164307, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7097181677818298, + "num_tokens": 162287796.0, + "step": 9050 + }, + { + "epoch": 0.7463511173414694, + "grad_norm": 1.9113337993621826, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7349739074707031, + "num_tokens": 162319079.0, + "step": 9051 + }, + { + "epoch": 0.746433577966521, + "grad_norm": 1.979302167892456, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.71921706199646, + "num_tokens": 162349796.0, + "step": 9052 + }, + { + "epoch": 0.7465160385915726, + "grad_norm": 1.8822720050811768, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7410092949867249, + "num_tokens": 162379382.0, + "step": 9053 + }, + { + "epoch": 0.746598499216624, + "grad_norm": 2.1776280403137207, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7155748605728149, + "num_tokens": 162403897.0, + "step": 9054 + }, + { + "epoch": 0.7466809598416756, + "grad_norm": 1.6998389959335327, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.7873690128326416, + "num_tokens": 162440416.0, + "step": 9055 + }, + { + "epoch": 0.7467634204667272, + "grad_norm": 1.9211541414260864, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7401399612426758, + "num_tokens": 162470902.0, + "step": 9056 + }, + { + "epoch": 0.7468458810917786, + "grad_norm": 1.8050850629806519, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7681494951248169, + "num_tokens": 162503210.0, + "step": 9057 + }, + { + "epoch": 0.7469283417168302, + "grad_norm": 1.990357756614685, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7251318693161011, + "num_tokens": 162533464.0, + "step": 9058 + }, + { + "epoch": 0.7470108023418818, + "grad_norm": 1.7326279878616333, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7466886043548584, + "num_tokens": 162568035.0, + "step": 9059 + }, + { + "epoch": 0.7470932629669332, + "grad_norm": 1.7653172016143799, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7535680532455444, + "num_tokens": 162600867.0, + "step": 9060 + }, + { + "epoch": 0.7471757235919848, + "grad_norm": 1.9057037830352783, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7361096143722534, + "num_tokens": 162631466.0, + "step": 9061 + }, + { + "epoch": 0.7472581842170364, + "grad_norm": 2.0318892002105713, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7550263404846191, + "num_tokens": 162659793.0, + "step": 9062 + }, + { + "epoch": 0.747340644842088, + "grad_norm": 1.9403789043426514, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.7859710454940796, + "num_tokens": 162692414.0, + "step": 9063 + }, + { + "epoch": 0.7474231054671394, + "grad_norm": 2.105091094970703, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7253140807151794, + "num_tokens": 162718501.0, + "step": 9064 + }, + { + "epoch": 0.747505566092191, + "grad_norm": 1.8683388233184814, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.75630784034729, + "num_tokens": 162752953.0, + "step": 9065 + }, + { + "epoch": 0.7475880267172426, + "grad_norm": 1.7924377918243408, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7691252827644348, + "num_tokens": 162786981.0, + "step": 9066 + }, + { + "epoch": 0.747670487342294, + "grad_norm": 1.827752709388733, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7682442665100098, + "num_tokens": 162817263.0, + "step": 9067 + }, + { + "epoch": 0.7477529479673456, + "grad_norm": 1.9106839895248413, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7455137968063354, + "num_tokens": 162846523.0, + "step": 9068 + }, + { + "epoch": 0.7478354085923972, + "grad_norm": 1.9132405519485474, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.768686056137085, + "num_tokens": 162875246.0, + "step": 9069 + }, + { + "epoch": 0.7479178692174486, + "grad_norm": 2.04882550239563, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7616034746170044, + "num_tokens": 162901995.0, + "step": 9070 + }, + { + "epoch": 0.7480003298425002, + "grad_norm": 2.318600654602051, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7581501007080078, + "num_tokens": 162923725.0, + "step": 9071 + }, + { + "epoch": 0.7480827904675518, + "grad_norm": 1.8288217782974243, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7503740787506104, + "num_tokens": 162956759.0, + "step": 9072 + }, + { + "epoch": 0.7481652510926032, + "grad_norm": 1.8635189533233643, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.792084813117981, + "num_tokens": 162988685.0, + "step": 9073 + }, + { + "epoch": 0.7482477117176548, + "grad_norm": 2.119915723800659, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7039811611175537, + "num_tokens": 163017327.0, + "step": 9074 + }, + { + "epoch": 0.7483301723427064, + "grad_norm": 1.7714945077896118, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.770352840423584, + "num_tokens": 163049975.0, + "step": 9075 + }, + { + "epoch": 0.7484126329677578, + "grad_norm": 2.0212554931640625, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7400678396224976, + "num_tokens": 163077868.0, + "step": 9076 + }, + { + "epoch": 0.7484950935928094, + "grad_norm": 1.7940223217010498, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7166143655776978, + "num_tokens": 163113447.0, + "step": 9077 + }, + { + "epoch": 0.748577554217861, + "grad_norm": 1.9531587362289429, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7750580906867981, + "num_tokens": 163141580.0, + "step": 9078 + }, + { + "epoch": 0.7486600148429126, + "grad_norm": 1.5927273035049438, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7964184284210205, + "num_tokens": 163179035.0, + "step": 9079 + }, + { + "epoch": 0.748742475467964, + "grad_norm": 1.777466058731079, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7387485504150391, + "num_tokens": 163216113.0, + "step": 9080 + }, + { + "epoch": 0.7488249360930156, + "grad_norm": 1.9075043201446533, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7778867483139038, + "num_tokens": 163244437.0, + "step": 9081 + }, + { + "epoch": 0.7489073967180672, + "grad_norm": 1.7137726545333862, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.780265212059021, + "num_tokens": 163280056.0, + "step": 9082 + }, + { + "epoch": 0.7489898573431186, + "grad_norm": 1.7377063035964966, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7371392846107483, + "num_tokens": 163317359.0, + "step": 9083 + }, + { + "epoch": 0.7490723179681702, + "grad_norm": 1.817856788635254, + "learning_rate": 1e-06, + "loss": 0.6022, + "mean_token_accuracy": 0.8101858496665955, + "num_tokens": 163345857.0, + "step": 9084 + }, + { + "epoch": 0.7491547785932218, + "grad_norm": 2.1696670055389404, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7538312673568726, + "num_tokens": 163370816.0, + "step": 9085 + }, + { + "epoch": 0.7492372392182732, + "grad_norm": 2.131560802459717, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7598912715911865, + "num_tokens": 163398032.0, + "step": 9086 + }, + { + "epoch": 0.7493196998433248, + "grad_norm": 1.9618560075759888, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7060135006904602, + "num_tokens": 163430222.0, + "step": 9087 + }, + { + "epoch": 0.7494021604683764, + "grad_norm": 2.000002145767212, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7350308895111084, + "num_tokens": 163458854.0, + "step": 9088 + }, + { + "epoch": 0.7494846210934278, + "grad_norm": 1.7571319341659546, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.7852119207382202, + "num_tokens": 163494627.0, + "step": 9089 + }, + { + "epoch": 0.7495670817184794, + "grad_norm": 2.068692445755005, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7415772080421448, + "num_tokens": 163523097.0, + "step": 9090 + }, + { + "epoch": 0.749649542343531, + "grad_norm": 1.887315273284912, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7341034412384033, + "num_tokens": 163554515.0, + "step": 9091 + }, + { + "epoch": 0.7497320029685826, + "grad_norm": 1.9931188821792603, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7568482160568237, + "num_tokens": 163584563.0, + "step": 9092 + }, + { + "epoch": 0.749814463593634, + "grad_norm": 1.6978943347930908, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.7910823822021484, + "num_tokens": 163617320.0, + "step": 9093 + }, + { + "epoch": 0.7498969242186856, + "grad_norm": 1.8008923530578613, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7328324913978577, + "num_tokens": 163653994.0, + "step": 9094 + }, + { + "epoch": 0.7499793848437372, + "grad_norm": 2.0946333408355713, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7405263781547546, + "num_tokens": 163681078.0, + "step": 9095 + }, + { + "epoch": 0.7500618454687886, + "grad_norm": 1.9076881408691406, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.705000638961792, + "num_tokens": 163717124.0, + "step": 9096 + }, + { + "epoch": 0.7501443060938402, + "grad_norm": 1.865962028503418, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7479462027549744, + "num_tokens": 163748470.0, + "step": 9097 + }, + { + "epoch": 0.7502267667188918, + "grad_norm": 1.952995777130127, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7651575803756714, + "num_tokens": 163779892.0, + "step": 9098 + }, + { + "epoch": 0.7503092273439432, + "grad_norm": 1.8458150625228882, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7496166229248047, + "num_tokens": 163810648.0, + "step": 9099 + }, + { + "epoch": 0.7503916879689948, + "grad_norm": 1.9610267877578735, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7760833501815796, + "num_tokens": 163838536.0, + "step": 9100 + }, + { + "epoch": 0.7504741485940464, + "grad_norm": 1.8419373035430908, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7456940412521362, + "num_tokens": 163872816.0, + "step": 9101 + }, + { + "epoch": 0.7505566092190978, + "grad_norm": 1.9053430557250977, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7637330293655396, + "num_tokens": 163901349.0, + "step": 9102 + }, + { + "epoch": 0.7506390698441494, + "grad_norm": 2.0215063095092773, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7333434820175171, + "num_tokens": 163931914.0, + "step": 9103 + }, + { + "epoch": 0.750721530469201, + "grad_norm": 1.856964111328125, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7516387701034546, + "num_tokens": 163962147.0, + "step": 9104 + }, + { + "epoch": 0.7508039910942524, + "grad_norm": 2.0241661071777344, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7089486122131348, + "num_tokens": 163991222.0, + "step": 9105 + }, + { + "epoch": 0.750886451719304, + "grad_norm": 1.9837957620620728, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7473896741867065, + "num_tokens": 164022342.0, + "step": 9106 + }, + { + "epoch": 0.7509689123443556, + "grad_norm": 1.9211345911026, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.748292863368988, + "num_tokens": 164054736.0, + "step": 9107 + }, + { + "epoch": 0.7510513729694072, + "grad_norm": 1.9355199337005615, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7250308394432068, + "num_tokens": 164084943.0, + "step": 9108 + }, + { + "epoch": 0.7511338335944586, + "grad_norm": 1.984993577003479, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7628782391548157, + "num_tokens": 164113482.0, + "step": 9109 + }, + { + "epoch": 0.7512162942195102, + "grad_norm": 1.9855546951293945, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.736635148525238, + "num_tokens": 164142361.0, + "step": 9110 + }, + { + "epoch": 0.7512987548445618, + "grad_norm": 2.268294334411621, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.764938235282898, + "num_tokens": 164166992.0, + "step": 9111 + }, + { + "epoch": 0.7513812154696132, + "grad_norm": 1.87332022190094, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7412081956863403, + "num_tokens": 164200179.0, + "step": 9112 + }, + { + "epoch": 0.7514636760946648, + "grad_norm": 1.9566524028778076, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.740648090839386, + "num_tokens": 164228703.0, + "step": 9113 + }, + { + "epoch": 0.7515461367197164, + "grad_norm": 2.0212080478668213, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7348521947860718, + "num_tokens": 164256750.0, + "step": 9114 + }, + { + "epoch": 0.7516285973447678, + "grad_norm": 2.133821487426758, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7526853084564209, + "num_tokens": 164281623.0, + "step": 9115 + }, + { + "epoch": 0.7517110579698194, + "grad_norm": 2.206432819366455, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7063679099082947, + "num_tokens": 164307482.0, + "step": 9116 + }, + { + "epoch": 0.751793518594871, + "grad_norm": 1.6897708177566528, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7685770988464355, + "num_tokens": 164343712.0, + "step": 9117 + }, + { + "epoch": 0.7518759792199224, + "grad_norm": 1.865107536315918, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7583569288253784, + "num_tokens": 164377813.0, + "step": 9118 + }, + { + "epoch": 0.751958439844974, + "grad_norm": 1.9748543500900269, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7474452257156372, + "num_tokens": 164408155.0, + "step": 9119 + }, + { + "epoch": 0.7520409004700256, + "grad_norm": 1.9563591480255127, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7665493488311768, + "num_tokens": 164437232.0, + "step": 9120 + }, + { + "epoch": 0.752123361095077, + "grad_norm": 2.057119846343994, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7549995183944702, + "num_tokens": 164464207.0, + "step": 9121 + }, + { + "epoch": 0.7522058217201286, + "grad_norm": 1.9088430404663086, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.747235894203186, + "num_tokens": 164497288.0, + "step": 9122 + }, + { + "epoch": 0.7522882823451802, + "grad_norm": 2.118093967437744, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7607372403144836, + "num_tokens": 164522539.0, + "step": 9123 + }, + { + "epoch": 0.7523707429702318, + "grad_norm": 2.0928022861480713, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7488270401954651, + "num_tokens": 164547621.0, + "step": 9124 + }, + { + "epoch": 0.7524532035952832, + "grad_norm": 2.015409469604492, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7373125553131104, + "num_tokens": 164575695.0, + "step": 9125 + }, + { + "epoch": 0.7525356642203348, + "grad_norm": 2.159660816192627, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7345923185348511, + "num_tokens": 164601975.0, + "step": 9126 + }, + { + "epoch": 0.7526181248453864, + "grad_norm": 1.8482600450515747, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7716981172561646, + "num_tokens": 164632875.0, + "step": 9127 + }, + { + "epoch": 0.7527005854704378, + "grad_norm": 2.0081400871276855, + "learning_rate": 1e-06, + "loss": 0.5858, + "mean_token_accuracy": 0.8018843531608582, + "num_tokens": 164662187.0, + "step": 9128 + }, + { + "epoch": 0.7527830460954894, + "grad_norm": 2.075040340423584, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7424154281616211, + "num_tokens": 164690025.0, + "step": 9129 + }, + { + "epoch": 0.752865506720541, + "grad_norm": 2.0309298038482666, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7228766679763794, + "num_tokens": 164720867.0, + "step": 9130 + }, + { + "epoch": 0.7529479673455924, + "grad_norm": 2.1027915477752686, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7553551197052002, + "num_tokens": 164746955.0, + "step": 9131 + }, + { + "epoch": 0.753030427970644, + "grad_norm": 1.7951295375823975, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7765125036239624, + "num_tokens": 164778212.0, + "step": 9132 + }, + { + "epoch": 0.7531128885956956, + "grad_norm": 1.865614891052246, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.759229838848114, + "num_tokens": 164811924.0, + "step": 9133 + }, + { + "epoch": 0.753195349220747, + "grad_norm": 1.9277637004852295, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7395594716072083, + "num_tokens": 164843044.0, + "step": 9134 + }, + { + "epoch": 0.7532778098457986, + "grad_norm": 2.2998690605163574, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.748416543006897, + "num_tokens": 164865553.0, + "step": 9135 + }, + { + "epoch": 0.7533602704708502, + "grad_norm": 1.9855834245681763, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7411884665489197, + "num_tokens": 164894331.0, + "step": 9136 + }, + { + "epoch": 0.7534427310959017, + "grad_norm": 1.9033927917480469, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.7547071576118469, + "num_tokens": 164923161.0, + "step": 9137 + }, + { + "epoch": 0.7535251917209532, + "grad_norm": 1.87874174118042, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7675912380218506, + "num_tokens": 164957006.0, + "step": 9138 + }, + { + "epoch": 0.7536076523460048, + "grad_norm": 2.120985507965088, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7486489415168762, + "num_tokens": 164981798.0, + "step": 9139 + }, + { + "epoch": 0.7536901129710564, + "grad_norm": 1.8608208894729614, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7419723868370056, + "num_tokens": 165014822.0, + "step": 9140 + }, + { + "epoch": 0.7537725735961078, + "grad_norm": 1.9128774404525757, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7000401020050049, + "num_tokens": 165046646.0, + "step": 9141 + }, + { + "epoch": 0.7538550342211594, + "grad_norm": 2.0505459308624268, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7310918569564819, + "num_tokens": 165075458.0, + "step": 9142 + }, + { + "epoch": 0.753937494846211, + "grad_norm": 2.0426793098449707, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7691295742988586, + "num_tokens": 165102304.0, + "step": 9143 + }, + { + "epoch": 0.7540199554712624, + "grad_norm": 2.009341239929199, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.747046947479248, + "num_tokens": 165130414.0, + "step": 9144 + }, + { + "epoch": 0.754102416096314, + "grad_norm": 1.9999629259109497, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7265430688858032, + "num_tokens": 165160457.0, + "step": 9145 + }, + { + "epoch": 0.7541848767213656, + "grad_norm": 1.9032655954360962, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7751715779304504, + "num_tokens": 165187792.0, + "step": 9146 + }, + { + "epoch": 0.754267337346417, + "grad_norm": 2.072031021118164, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7136344313621521, + "num_tokens": 165216085.0, + "step": 9147 + }, + { + "epoch": 0.7543497979714686, + "grad_norm": 1.8901458978652954, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7589986324310303, + "num_tokens": 165248497.0, + "step": 9148 + }, + { + "epoch": 0.7544322585965202, + "grad_norm": 1.9445217847824097, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7928787469863892, + "num_tokens": 165274148.0, + "step": 9149 + }, + { + "epoch": 0.7545147192215717, + "grad_norm": 1.967557668685913, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7675096988677979, + "num_tokens": 165303052.0, + "step": 9150 + }, + { + "epoch": 0.7545971798466232, + "grad_norm": 2.0322465896606445, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7229281067848206, + "num_tokens": 165331987.0, + "step": 9151 + }, + { + "epoch": 0.7546796404716748, + "grad_norm": 2.0101044178009033, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7322597503662109, + "num_tokens": 165359919.0, + "step": 9152 + }, + { + "epoch": 0.7547621010967264, + "grad_norm": 1.951553463935852, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7375993132591248, + "num_tokens": 165390506.0, + "step": 9153 + }, + { + "epoch": 0.7548445617217778, + "grad_norm": 1.8584376573562622, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.7987292408943176, + "num_tokens": 165417679.0, + "step": 9154 + }, + { + "epoch": 0.7549270223468294, + "grad_norm": 1.911300778388977, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7193224430084229, + "num_tokens": 165449371.0, + "step": 9155 + }, + { + "epoch": 0.755009482971881, + "grad_norm": 1.8791521787643433, + "learning_rate": 1e-06, + "loss": 0.734, + "mean_token_accuracy": 0.7701677680015564, + "num_tokens": 165476492.0, + "step": 9156 + }, + { + "epoch": 0.7550919435969324, + "grad_norm": 1.8789395093917847, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7339105606079102, + "num_tokens": 165509273.0, + "step": 9157 + }, + { + "epoch": 0.755174404221984, + "grad_norm": 1.959462285041809, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7741920351982117, + "num_tokens": 165536233.0, + "step": 9158 + }, + { + "epoch": 0.7552568648470356, + "grad_norm": 1.860161304473877, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7789352536201477, + "num_tokens": 165564983.0, + "step": 9159 + }, + { + "epoch": 0.755339325472087, + "grad_norm": 1.9775084257125854, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7247799038887024, + "num_tokens": 165595148.0, + "step": 9160 + }, + { + "epoch": 0.7554217860971386, + "grad_norm": 1.844524621963501, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7552468776702881, + "num_tokens": 165627249.0, + "step": 9161 + }, + { + "epoch": 0.7555042467221902, + "grad_norm": 1.7268427610397339, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7452657222747803, + "num_tokens": 165665304.0, + "step": 9162 + }, + { + "epoch": 0.7555867073472416, + "grad_norm": 2.213486671447754, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7322652339935303, + "num_tokens": 165691272.0, + "step": 9163 + }, + { + "epoch": 0.7556691679722932, + "grad_norm": 1.9736794233322144, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7403178811073303, + "num_tokens": 165719615.0, + "step": 9164 + }, + { + "epoch": 0.7557516285973448, + "grad_norm": 2.1532907485961914, + "learning_rate": 1e-06, + "loss": 1.0795, + "mean_token_accuracy": 0.673827588558197, + "num_tokens": 165748705.0, + "step": 9165 + }, + { + "epoch": 0.7558340892223963, + "grad_norm": 1.968187928199768, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7503306865692139, + "num_tokens": 165776835.0, + "step": 9166 + }, + { + "epoch": 0.7559165498474478, + "grad_norm": 1.8767284154891968, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.7833834886550903, + "num_tokens": 165808534.0, + "step": 9167 + }, + { + "epoch": 0.7559990104724994, + "grad_norm": 1.9737772941589355, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7379589080810547, + "num_tokens": 165841568.0, + "step": 9168 + }, + { + "epoch": 0.756081471097551, + "grad_norm": 2.1119191646575928, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7787721157073975, + "num_tokens": 165868223.0, + "step": 9169 + }, + { + "epoch": 0.7561639317226024, + "grad_norm": 1.8774181604385376, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.779410183429718, + "num_tokens": 165898243.0, + "step": 9170 + }, + { + "epoch": 0.756246392347654, + "grad_norm": 1.6994491815567017, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7449090480804443, + "num_tokens": 165937523.0, + "step": 9171 + }, + { + "epoch": 0.7563288529727056, + "grad_norm": 1.8205313682556152, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7422618865966797, + "num_tokens": 165967581.0, + "step": 9172 + }, + { + "epoch": 0.756411313597757, + "grad_norm": 1.8657878637313843, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7770902514457703, + "num_tokens": 165999761.0, + "step": 9173 + }, + { + "epoch": 0.7564937742228086, + "grad_norm": 1.7916136980056763, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7839425206184387, + "num_tokens": 166030357.0, + "step": 9174 + }, + { + "epoch": 0.7565762348478602, + "grad_norm": 1.8295540809631348, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7349415421485901, + "num_tokens": 166065612.0, + "step": 9175 + }, + { + "epoch": 0.7566586954729116, + "grad_norm": 1.9642221927642822, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7582868933677673, + "num_tokens": 166094508.0, + "step": 9176 + }, + { + "epoch": 0.7567411560979632, + "grad_norm": 1.8841651678085327, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.74847012758255, + "num_tokens": 166125003.0, + "step": 9177 + }, + { + "epoch": 0.7568236167230148, + "grad_norm": 1.7265117168426514, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.744884729385376, + "num_tokens": 166163427.0, + "step": 9178 + }, + { + "epoch": 0.7569060773480663, + "grad_norm": 1.9339369535446167, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7647680044174194, + "num_tokens": 166195018.0, + "step": 9179 + }, + { + "epoch": 0.7569885379731178, + "grad_norm": 1.9994949102401733, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6947281360626221, + "num_tokens": 166225665.0, + "step": 9180 + }, + { + "epoch": 0.7570709985981694, + "grad_norm": 1.8463292121887207, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7714197039604187, + "num_tokens": 166257234.0, + "step": 9181 + }, + { + "epoch": 0.7571534592232209, + "grad_norm": 1.7563661336898804, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7402877807617188, + "num_tokens": 166293510.0, + "step": 9182 + }, + { + "epoch": 0.7572359198482724, + "grad_norm": 1.8184856176376343, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7438774704933167, + "num_tokens": 166324678.0, + "step": 9183 + }, + { + "epoch": 0.757318380473324, + "grad_norm": 1.9536577463150024, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.74233078956604, + "num_tokens": 166353494.0, + "step": 9184 + }, + { + "epoch": 0.7574008410983756, + "grad_norm": 2.0650594234466553, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7373266220092773, + "num_tokens": 166379064.0, + "step": 9185 + }, + { + "epoch": 0.757483301723427, + "grad_norm": 1.8440778255462646, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.726215124130249, + "num_tokens": 166412713.0, + "step": 9186 + }, + { + "epoch": 0.7575657623484786, + "grad_norm": 1.842826247215271, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7329410314559937, + "num_tokens": 166445362.0, + "step": 9187 + }, + { + "epoch": 0.7576482229735302, + "grad_norm": 1.8439708948135376, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.7942721843719482, + "num_tokens": 166474852.0, + "step": 9188 + }, + { + "epoch": 0.7577306835985816, + "grad_norm": 1.8645493984222412, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7166705131530762, + "num_tokens": 166508743.0, + "step": 9189 + }, + { + "epoch": 0.7578131442236332, + "grad_norm": 1.8030469417572021, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.7932404279708862, + "num_tokens": 166541869.0, + "step": 9190 + }, + { + "epoch": 0.7578956048486848, + "grad_norm": 1.842651128768921, + "learning_rate": 1e-06, + "loss": 0.6114, + "mean_token_accuracy": 0.8006925582885742, + "num_tokens": 166570643.0, + "step": 9191 + }, + { + "epoch": 0.7579780654737363, + "grad_norm": 1.803865671157837, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7244625091552734, + "num_tokens": 166604715.0, + "step": 9192 + }, + { + "epoch": 0.7580605260987878, + "grad_norm": 1.921051263809204, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7160369157791138, + "num_tokens": 166636725.0, + "step": 9193 + }, + { + "epoch": 0.7581429867238394, + "grad_norm": 1.8944554328918457, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7798495292663574, + "num_tokens": 166664204.0, + "step": 9194 + }, + { + "epoch": 0.7582254473488909, + "grad_norm": 1.9153896570205688, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7299566864967346, + "num_tokens": 166699635.0, + "step": 9195 + }, + { + "epoch": 0.7583079079739424, + "grad_norm": 1.9046164751052856, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7332713603973389, + "num_tokens": 166730341.0, + "step": 9196 + }, + { + "epoch": 0.758390368598994, + "grad_norm": 2.055438995361328, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7491656541824341, + "num_tokens": 166761703.0, + "step": 9197 + }, + { + "epoch": 0.7584728292240455, + "grad_norm": 1.9621585607528687, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7120882272720337, + "num_tokens": 166792837.0, + "step": 9198 + }, + { + "epoch": 0.758555289849097, + "grad_norm": 1.841557264328003, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7256571054458618, + "num_tokens": 166829154.0, + "step": 9199 + }, + { + "epoch": 0.7586377504741486, + "grad_norm": 1.8928495645523071, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7181200385093689, + "num_tokens": 166862286.0, + "step": 9200 + }, + { + "epoch": 0.7587202110992002, + "grad_norm": 1.8550759553909302, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7525185346603394, + "num_tokens": 166894463.0, + "step": 9201 + }, + { + "epoch": 0.7588026717242516, + "grad_norm": 2.1697988510131836, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7458616495132446, + "num_tokens": 166919326.0, + "step": 9202 + }, + { + "epoch": 0.7588851323493032, + "grad_norm": 1.8821154832839966, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7643852233886719, + "num_tokens": 166949707.0, + "step": 9203 + }, + { + "epoch": 0.7589675929743548, + "grad_norm": 2.0386335849761963, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7070680260658264, + "num_tokens": 166979016.0, + "step": 9204 + }, + { + "epoch": 0.7590500535994062, + "grad_norm": 1.9685134887695312, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7460843324661255, + "num_tokens": 167008044.0, + "step": 9205 + }, + { + "epoch": 0.7591325142244578, + "grad_norm": 1.8445132970809937, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.7975741028785706, + "num_tokens": 167037186.0, + "step": 9206 + }, + { + "epoch": 0.7592149748495094, + "grad_norm": 2.233954429626465, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.745050847530365, + "num_tokens": 167063541.0, + "step": 9207 + }, + { + "epoch": 0.7592974354745609, + "grad_norm": 2.0643386840820312, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7149089574813843, + "num_tokens": 167090273.0, + "step": 9208 + }, + { + "epoch": 0.7593798960996124, + "grad_norm": 1.9639205932617188, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7564541697502136, + "num_tokens": 167119146.0, + "step": 9209 + }, + { + "epoch": 0.759462356724664, + "grad_norm": 2.060723066329956, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7483998537063599, + "num_tokens": 167144092.0, + "step": 9210 + }, + { + "epoch": 0.7595448173497155, + "grad_norm": 2.2118067741394043, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7445863485336304, + "num_tokens": 167169921.0, + "step": 9211 + }, + { + "epoch": 0.759627277974767, + "grad_norm": 2.1664578914642334, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7406395673751831, + "num_tokens": 167196328.0, + "step": 9212 + }, + { + "epoch": 0.7597097385998186, + "grad_norm": 1.783546805381775, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.7863337993621826, + "num_tokens": 167230041.0, + "step": 9213 + }, + { + "epoch": 0.7597921992248702, + "grad_norm": 1.956089735031128, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7858586311340332, + "num_tokens": 167258958.0, + "step": 9214 + }, + { + "epoch": 0.7598746598499216, + "grad_norm": 1.8052644729614258, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.7955063581466675, + "num_tokens": 167291842.0, + "step": 9215 + }, + { + "epoch": 0.7599571204749732, + "grad_norm": 2.2570340633392334, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7139378786087036, + "num_tokens": 167316386.0, + "step": 9216 + }, + { + "epoch": 0.7600395811000248, + "grad_norm": 1.7733157873153687, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.7053369283676147, + "num_tokens": 167355780.0, + "step": 9217 + }, + { + "epoch": 0.7601220417250762, + "grad_norm": 2.019223213195801, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7671540975570679, + "num_tokens": 167381633.0, + "step": 9218 + }, + { + "epoch": 0.7602045023501278, + "grad_norm": 1.8917245864868164, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7816045880317688, + "num_tokens": 167411230.0, + "step": 9219 + }, + { + "epoch": 0.7602869629751794, + "grad_norm": 2.080631971359253, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7508207559585571, + "num_tokens": 167438061.0, + "step": 9220 + }, + { + "epoch": 0.7603694236002309, + "grad_norm": 2.086076259613037, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7474383115768433, + "num_tokens": 167464858.0, + "step": 9221 + }, + { + "epoch": 0.7604518842252824, + "grad_norm": 1.7158087491989136, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7551889419555664, + "num_tokens": 167500416.0, + "step": 9222 + }, + { + "epoch": 0.760534344850334, + "grad_norm": 2.1886839866638184, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.764022707939148, + "num_tokens": 167526160.0, + "step": 9223 + }, + { + "epoch": 0.7606168054753855, + "grad_norm": 1.9785979986190796, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7177977561950684, + "num_tokens": 167555611.0, + "step": 9224 + }, + { + "epoch": 0.760699266100437, + "grad_norm": 1.9960027933120728, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7469749450683594, + "num_tokens": 167585784.0, + "step": 9225 + }, + { + "epoch": 0.7607817267254886, + "grad_norm": 1.9026471376419067, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.705431342124939, + "num_tokens": 167619945.0, + "step": 9226 + }, + { + "epoch": 0.7608641873505401, + "grad_norm": 1.7970608472824097, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.774696409702301, + "num_tokens": 167653748.0, + "step": 9227 + }, + { + "epoch": 0.7609466479755916, + "grad_norm": 1.8473150730133057, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7768283486366272, + "num_tokens": 167684187.0, + "step": 9228 + }, + { + "epoch": 0.7610291086006432, + "grad_norm": 1.9508146047592163, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7588933110237122, + "num_tokens": 167712465.0, + "step": 9229 + }, + { + "epoch": 0.7611115692256948, + "grad_norm": 2.1781980991363525, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7698363065719604, + "num_tokens": 167736578.0, + "step": 9230 + }, + { + "epoch": 0.7611940298507462, + "grad_norm": 1.8709269762039185, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7376353740692139, + "num_tokens": 167766919.0, + "step": 9231 + }, + { + "epoch": 0.7612764904757978, + "grad_norm": 1.9332231283187866, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7309191226959229, + "num_tokens": 167799279.0, + "step": 9232 + }, + { + "epoch": 0.7613589511008494, + "grad_norm": 1.7514036893844604, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7547394037246704, + "num_tokens": 167834101.0, + "step": 9233 + }, + { + "epoch": 0.7614414117259009, + "grad_norm": 1.8281619548797607, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7827200889587402, + "num_tokens": 167863420.0, + "step": 9234 + }, + { + "epoch": 0.7615238723509524, + "grad_norm": 1.775448203086853, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7698296308517456, + "num_tokens": 167896974.0, + "step": 9235 + }, + { + "epoch": 0.761606332976004, + "grad_norm": 1.9934316873550415, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7678183317184448, + "num_tokens": 167924426.0, + "step": 9236 + }, + { + "epoch": 0.7616887936010555, + "grad_norm": 1.9728987216949463, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7050150036811829, + "num_tokens": 167954419.0, + "step": 9237 + }, + { + "epoch": 0.761771254226107, + "grad_norm": 1.817228078842163, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7119150161743164, + "num_tokens": 167988716.0, + "step": 9238 + }, + { + "epoch": 0.7618537148511586, + "grad_norm": 1.9660634994506836, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7539682388305664, + "num_tokens": 168017463.0, + "step": 9239 + }, + { + "epoch": 0.7619361754762101, + "grad_norm": 1.7510370016098022, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7577944993972778, + "num_tokens": 168050968.0, + "step": 9240 + }, + { + "epoch": 0.7620186361012616, + "grad_norm": 2.083056926727295, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7178117036819458, + "num_tokens": 168078623.0, + "step": 9241 + }, + { + "epoch": 0.7621010967263132, + "grad_norm": 1.9003000259399414, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.7941714525222778, + "num_tokens": 168108222.0, + "step": 9242 + }, + { + "epoch": 0.7621835573513647, + "grad_norm": 1.7091137170791626, + "learning_rate": 1e-06, + "loss": 0.5771, + "mean_token_accuracy": 0.8055067658424377, + "num_tokens": 168142560.0, + "step": 9243 + }, + { + "epoch": 0.7622660179764162, + "grad_norm": 2.3078999519348145, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7319986820220947, + "num_tokens": 168166592.0, + "step": 9244 + }, + { + "epoch": 0.7623484786014678, + "grad_norm": 2.119844436645508, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.720771312713623, + "num_tokens": 168194219.0, + "step": 9245 + }, + { + "epoch": 0.7624309392265194, + "grad_norm": 1.8606133460998535, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7265903353691101, + "num_tokens": 168228471.0, + "step": 9246 + }, + { + "epoch": 0.7625133998515708, + "grad_norm": 2.0246710777282715, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7421551942825317, + "num_tokens": 168258957.0, + "step": 9247 + }, + { + "epoch": 0.7625958604766224, + "grad_norm": 2.0995726585388184, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7224903702735901, + "num_tokens": 168286066.0, + "step": 9248 + }, + { + "epoch": 0.762678321101674, + "grad_norm": 1.9324848651885986, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7778993844985962, + "num_tokens": 168314480.0, + "step": 9249 + }, + { + "epoch": 0.7627607817267255, + "grad_norm": 2.0259697437286377, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7626142501831055, + "num_tokens": 168343824.0, + "step": 9250 + }, + { + "epoch": 0.762843242351777, + "grad_norm": 1.7093276977539062, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.752171516418457, + "num_tokens": 168375619.0, + "step": 9251 + }, + { + "epoch": 0.7629257029768286, + "grad_norm": 1.973899483680725, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7443315386772156, + "num_tokens": 168404202.0, + "step": 9252 + }, + { + "epoch": 0.7630081636018801, + "grad_norm": 1.8878428936004639, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7423592209815979, + "num_tokens": 168435116.0, + "step": 9253 + }, + { + "epoch": 0.7630906242269316, + "grad_norm": 1.8577804565429688, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.7631165385246277, + "num_tokens": 168466836.0, + "step": 9254 + }, + { + "epoch": 0.7631730848519832, + "grad_norm": 2.1273226737976074, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7145569324493408, + "num_tokens": 168492759.0, + "step": 9255 + }, + { + "epoch": 0.7632555454770347, + "grad_norm": 1.7974393367767334, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.7635915875434875, + "num_tokens": 168525272.0, + "step": 9256 + }, + { + "epoch": 0.7633380061020862, + "grad_norm": 2.045999765396118, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7464019060134888, + "num_tokens": 168552968.0, + "step": 9257 + }, + { + "epoch": 0.7634204667271378, + "grad_norm": 2.2176859378814697, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7699248790740967, + "num_tokens": 168575698.0, + "step": 9258 + }, + { + "epoch": 0.7635029273521893, + "grad_norm": 1.8176941871643066, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7532293796539307, + "num_tokens": 168608330.0, + "step": 9259 + }, + { + "epoch": 0.7635853879772408, + "grad_norm": 1.7183008193969727, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7455785870552063, + "num_tokens": 168644459.0, + "step": 9260 + }, + { + "epoch": 0.7636678486022924, + "grad_norm": 1.7541344165802002, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7426488399505615, + "num_tokens": 168679779.0, + "step": 9261 + }, + { + "epoch": 0.763750309227344, + "grad_norm": 2.0287768840789795, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7722557783126831, + "num_tokens": 168705490.0, + "step": 9262 + }, + { + "epoch": 0.7638327698523955, + "grad_norm": 1.6822701692581177, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7437175512313843, + "num_tokens": 168744226.0, + "step": 9263 + }, + { + "epoch": 0.763915230477447, + "grad_norm": 1.8179237842559814, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7755147814750671, + "num_tokens": 168774021.0, + "step": 9264 + }, + { + "epoch": 0.7639976911024986, + "grad_norm": 1.8858683109283447, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7229674458503723, + "num_tokens": 168807787.0, + "step": 9265 + }, + { + "epoch": 0.7640801517275501, + "grad_norm": 2.0218911170959473, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7659446001052856, + "num_tokens": 168836765.0, + "step": 9266 + }, + { + "epoch": 0.7641626123526016, + "grad_norm": 1.7600600719451904, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7580132484436035, + "num_tokens": 168872003.0, + "step": 9267 + }, + { + "epoch": 0.7642450729776532, + "grad_norm": 1.894698143005371, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7653056383132935, + "num_tokens": 168905343.0, + "step": 9268 + }, + { + "epoch": 0.7643275336027047, + "grad_norm": 1.7417123317718506, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7315587997436523, + "num_tokens": 168938148.0, + "step": 9269 + }, + { + "epoch": 0.7644099942277562, + "grad_norm": 2.2542037963867188, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7207558155059814, + "num_tokens": 168962739.0, + "step": 9270 + }, + { + "epoch": 0.7644924548528078, + "grad_norm": 2.0071935653686523, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7411882877349854, + "num_tokens": 168991163.0, + "step": 9271 + }, + { + "epoch": 0.7645749154778593, + "grad_norm": 1.9864567518234253, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7338171005249023, + "num_tokens": 169019461.0, + "step": 9272 + }, + { + "epoch": 0.7646573761029108, + "grad_norm": 1.9974980354309082, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7433236837387085, + "num_tokens": 169048192.0, + "step": 9273 + }, + { + "epoch": 0.7647398367279624, + "grad_norm": 1.936221718788147, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7398596405982971, + "num_tokens": 169077845.0, + "step": 9274 + }, + { + "epoch": 0.764822297353014, + "grad_norm": 1.9144338369369507, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7585563063621521, + "num_tokens": 169110978.0, + "step": 9275 + }, + { + "epoch": 0.7649047579780655, + "grad_norm": 1.7527785301208496, + "learning_rate": 1e-06, + "loss": 0.5737, + "mean_token_accuracy": 0.8025491237640381, + "num_tokens": 169144258.0, + "step": 9276 + }, + { + "epoch": 0.764987218603117, + "grad_norm": 1.7477957010269165, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7802876234054565, + "num_tokens": 169177397.0, + "step": 9277 + }, + { + "epoch": 0.7650696792281686, + "grad_norm": 2.0691986083984375, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7060129046440125, + "num_tokens": 169203257.0, + "step": 9278 + }, + { + "epoch": 0.7651521398532201, + "grad_norm": 1.9055322408676147, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.738238513469696, + "num_tokens": 169235059.0, + "step": 9279 + }, + { + "epoch": 0.7652346004782716, + "grad_norm": 1.7599248886108398, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7817057967185974, + "num_tokens": 169268452.0, + "step": 9280 + }, + { + "epoch": 0.7653170611033232, + "grad_norm": 2.0974676609039307, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7264810800552368, + "num_tokens": 169294102.0, + "step": 9281 + }, + { + "epoch": 0.7653995217283747, + "grad_norm": 2.0081613063812256, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7428315877914429, + "num_tokens": 169323060.0, + "step": 9282 + }, + { + "epoch": 0.7654819823534262, + "grad_norm": 1.9732906818389893, + "learning_rate": 1e-06, + "loss": 0.6233, + "mean_token_accuracy": 0.7981505990028381, + "num_tokens": 169349020.0, + "step": 9283 + }, + { + "epoch": 0.7655644429784778, + "grad_norm": 1.750783920288086, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.7964392900466919, + "num_tokens": 169377857.0, + "step": 9284 + }, + { + "epoch": 0.7656469036035293, + "grad_norm": 1.84076988697052, + "learning_rate": 1e-06, + "loss": 0.5971, + "mean_token_accuracy": 0.8038173317909241, + "num_tokens": 169408773.0, + "step": 9285 + }, + { + "epoch": 0.7657293642285808, + "grad_norm": 2.1571755409240723, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7485674023628235, + "num_tokens": 169434802.0, + "step": 9286 + }, + { + "epoch": 0.7658118248536324, + "grad_norm": 2.047511339187622, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7263050079345703, + "num_tokens": 169466323.0, + "step": 9287 + }, + { + "epoch": 0.7658942854786839, + "grad_norm": 2.1179375648498535, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7042105793952942, + "num_tokens": 169494538.0, + "step": 9288 + }, + { + "epoch": 0.7659767461037355, + "grad_norm": 2.090005397796631, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7607554197311401, + "num_tokens": 169521501.0, + "step": 9289 + }, + { + "epoch": 0.766059206728787, + "grad_norm": 1.6474242210388184, + "learning_rate": 1e-06, + "loss": 0.6189, + "mean_token_accuracy": 0.7884094715118408, + "num_tokens": 169558507.0, + "step": 9290 + }, + { + "epoch": 0.7661416673538386, + "grad_norm": 1.6687793731689453, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7441742420196533, + "num_tokens": 169599650.0, + "step": 9291 + }, + { + "epoch": 0.7662241279788901, + "grad_norm": 1.7971643209457397, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7623062133789062, + "num_tokens": 169633313.0, + "step": 9292 + }, + { + "epoch": 0.7663065886039416, + "grad_norm": 2.022949695587158, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7388066053390503, + "num_tokens": 169660085.0, + "step": 9293 + }, + { + "epoch": 0.7663890492289932, + "grad_norm": 1.7826926708221436, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7648141384124756, + "num_tokens": 169694770.0, + "step": 9294 + }, + { + "epoch": 0.7664715098540447, + "grad_norm": 1.9992979764938354, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7633755803108215, + "num_tokens": 169721319.0, + "step": 9295 + }, + { + "epoch": 0.7665539704790962, + "grad_norm": 1.9594340324401855, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7722413539886475, + "num_tokens": 169749469.0, + "step": 9296 + }, + { + "epoch": 0.7666364311041478, + "grad_norm": 2.38289475440979, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7483806610107422, + "num_tokens": 169770159.0, + "step": 9297 + }, + { + "epoch": 0.7667188917291993, + "grad_norm": 1.6924363374710083, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.78470778465271, + "num_tokens": 169809898.0, + "step": 9298 + }, + { + "epoch": 0.7668013523542508, + "grad_norm": 1.8443291187286377, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7427173256874084, + "num_tokens": 169844178.0, + "step": 9299 + }, + { + "epoch": 0.7668838129793024, + "grad_norm": 1.6840382814407349, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7218954563140869, + "num_tokens": 169883743.0, + "step": 9300 + }, + { + "epoch": 0.7669662736043539, + "grad_norm": 2.087550401687622, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7588405609130859, + "num_tokens": 169910275.0, + "step": 9301 + }, + { + "epoch": 0.7670487342294054, + "grad_norm": 2.0061347484588623, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7172772884368896, + "num_tokens": 169939752.0, + "step": 9302 + }, + { + "epoch": 0.767131194854457, + "grad_norm": 1.7116421461105347, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.7617313861846924, + "num_tokens": 169973994.0, + "step": 9303 + }, + { + "epoch": 0.7672136554795085, + "grad_norm": 2.0921459197998047, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7238550186157227, + "num_tokens": 170002094.0, + "step": 9304 + }, + { + "epoch": 0.76729611610456, + "grad_norm": 1.8168543577194214, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7777451872825623, + "num_tokens": 170030811.0, + "step": 9305 + }, + { + "epoch": 0.7673785767296116, + "grad_norm": 2.0237200260162354, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.7455204725265503, + "num_tokens": 170057368.0, + "step": 9306 + }, + { + "epoch": 0.7674610373546632, + "grad_norm": 2.0065715312957764, + "learning_rate": 1e-06, + "loss": 0.7656, + "mean_token_accuracy": 0.759028434753418, + "num_tokens": 170086073.0, + "step": 9307 + }, + { + "epoch": 0.7675434979797147, + "grad_norm": 1.9172941446304321, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7458466291427612, + "num_tokens": 170117197.0, + "step": 9308 + }, + { + "epoch": 0.7676259586047662, + "grad_norm": 2.197932481765747, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7487274408340454, + "num_tokens": 170138494.0, + "step": 9309 + }, + { + "epoch": 0.7677084192298178, + "grad_norm": 1.8252513408660889, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7651760578155518, + "num_tokens": 170171843.0, + "step": 9310 + }, + { + "epoch": 0.7677908798548693, + "grad_norm": 1.916379690170288, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7556474208831787, + "num_tokens": 170202593.0, + "step": 9311 + }, + { + "epoch": 0.7678733404799208, + "grad_norm": 2.0735151767730713, + "learning_rate": 1e-06, + "loss": 0.612, + "mean_token_accuracy": 0.8035273551940918, + "num_tokens": 170225039.0, + "step": 9312 + }, + { + "epoch": 0.7679558011049724, + "grad_norm": 1.9667811393737793, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7235336303710938, + "num_tokens": 170254403.0, + "step": 9313 + }, + { + "epoch": 0.7680382617300239, + "grad_norm": 1.8712478876113892, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7413191795349121, + "num_tokens": 170285909.0, + "step": 9314 + }, + { + "epoch": 0.7681207223550754, + "grad_norm": 1.9985575675964355, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.737247109413147, + "num_tokens": 170313203.0, + "step": 9315 + }, + { + "epoch": 0.768203182980127, + "grad_norm": 1.938163161277771, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7634546756744385, + "num_tokens": 170343607.0, + "step": 9316 + }, + { + "epoch": 0.7682856436051785, + "grad_norm": 1.9534367322921753, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7202579379081726, + "num_tokens": 170373393.0, + "step": 9317 + }, + { + "epoch": 0.76836810423023, + "grad_norm": 1.935418725013733, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7350798845291138, + "num_tokens": 170402901.0, + "step": 9318 + }, + { + "epoch": 0.7684505648552816, + "grad_norm": 2.0280344486236572, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.786960244178772, + "num_tokens": 170430482.0, + "step": 9319 + }, + { + "epoch": 0.7685330254803331, + "grad_norm": 1.9708284139633179, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.730257511138916, + "num_tokens": 170459894.0, + "step": 9320 + }, + { + "epoch": 0.7686154861053847, + "grad_norm": 1.9958890676498413, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7577381134033203, + "num_tokens": 170487731.0, + "step": 9321 + }, + { + "epoch": 0.7686979467304362, + "grad_norm": 1.9774196147918701, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7487955093383789, + "num_tokens": 170517418.0, + "step": 9322 + }, + { + "epoch": 0.7687804073554878, + "grad_norm": 1.8296146392822266, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7596077919006348, + "num_tokens": 170549377.0, + "step": 9323 + }, + { + "epoch": 0.7688628679805393, + "grad_norm": 1.8595409393310547, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7572557926177979, + "num_tokens": 170581981.0, + "step": 9324 + }, + { + "epoch": 0.7689453286055908, + "grad_norm": 1.935479998588562, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7685333490371704, + "num_tokens": 170610969.0, + "step": 9325 + }, + { + "epoch": 0.7690277892306424, + "grad_norm": 1.842729091644287, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7742262482643127, + "num_tokens": 170641862.0, + "step": 9326 + }, + { + "epoch": 0.7691102498556939, + "grad_norm": 1.7444032430648804, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7443685531616211, + "num_tokens": 170676760.0, + "step": 9327 + }, + { + "epoch": 0.7691927104807454, + "grad_norm": 1.8070968389511108, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7544056177139282, + "num_tokens": 170712728.0, + "step": 9328 + }, + { + "epoch": 0.769275171105797, + "grad_norm": 1.8302098512649536, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.793889045715332, + "num_tokens": 170744824.0, + "step": 9329 + }, + { + "epoch": 0.7693576317308485, + "grad_norm": 1.624289870262146, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.762828528881073, + "num_tokens": 170787472.0, + "step": 9330 + }, + { + "epoch": 0.7694400923559, + "grad_norm": 1.822380781173706, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7493246793746948, + "num_tokens": 170819958.0, + "step": 9331 + }, + { + "epoch": 0.7695225529809516, + "grad_norm": 1.937978982925415, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7310633659362793, + "num_tokens": 170851191.0, + "step": 9332 + }, + { + "epoch": 0.7696050136060031, + "grad_norm": 2.272324323654175, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7826639413833618, + "num_tokens": 170874715.0, + "step": 9333 + }, + { + "epoch": 0.7696874742310547, + "grad_norm": 1.949563980102539, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7570669054985046, + "num_tokens": 170903581.0, + "step": 9334 + }, + { + "epoch": 0.7697699348561062, + "grad_norm": 2.0069563388824463, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7354798316955566, + "num_tokens": 170929677.0, + "step": 9335 + }, + { + "epoch": 0.7698523954811577, + "grad_norm": 1.9983195066452026, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7707827091217041, + "num_tokens": 170956017.0, + "step": 9336 + }, + { + "epoch": 0.7699348561062093, + "grad_norm": 2.0756006240844727, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7276119589805603, + "num_tokens": 170983888.0, + "step": 9337 + }, + { + "epoch": 0.7700173167312608, + "grad_norm": 1.840722918510437, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7527416944503784, + "num_tokens": 171017711.0, + "step": 9338 + }, + { + "epoch": 0.7700997773563124, + "grad_norm": 2.081637144088745, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7763948440551758, + "num_tokens": 171043479.0, + "step": 9339 + }, + { + "epoch": 0.7701822379813639, + "grad_norm": 1.8855339288711548, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7522333860397339, + "num_tokens": 171075100.0, + "step": 9340 + }, + { + "epoch": 0.7702646986064154, + "grad_norm": 1.9213999509811401, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.7722814083099365, + "num_tokens": 171103465.0, + "step": 9341 + }, + { + "epoch": 0.770347159231467, + "grad_norm": 2.0730013847351074, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7639617919921875, + "num_tokens": 171130944.0, + "step": 9342 + }, + { + "epoch": 0.7704296198565185, + "grad_norm": 1.9099558591842651, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7587047815322876, + "num_tokens": 171161167.0, + "step": 9343 + }, + { + "epoch": 0.77051208048157, + "grad_norm": 1.8194223642349243, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7874953746795654, + "num_tokens": 171191620.0, + "step": 9344 + }, + { + "epoch": 0.7705945411066216, + "grad_norm": 2.0849828720092773, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7342357635498047, + "num_tokens": 171220361.0, + "step": 9345 + }, + { + "epoch": 0.7706770017316731, + "grad_norm": 1.9330952167510986, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7507621049880981, + "num_tokens": 171248465.0, + "step": 9346 + }, + { + "epoch": 0.7707594623567247, + "grad_norm": 1.978819489479065, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7600853443145752, + "num_tokens": 171276667.0, + "step": 9347 + }, + { + "epoch": 0.7708419229817762, + "grad_norm": 1.8866316080093384, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7576552629470825, + "num_tokens": 171308359.0, + "step": 9348 + }, + { + "epoch": 0.7709243836068277, + "grad_norm": 2.142317056655884, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7539284229278564, + "num_tokens": 171334477.0, + "step": 9349 + }, + { + "epoch": 0.7710068442318793, + "grad_norm": 1.9462006092071533, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7250186800956726, + "num_tokens": 171367431.0, + "step": 9350 + }, + { + "epoch": 0.7710893048569308, + "grad_norm": 1.8001651763916016, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7715253233909607, + "num_tokens": 171399420.0, + "step": 9351 + }, + { + "epoch": 0.7711717654819824, + "grad_norm": 1.7887681722640991, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7788981199264526, + "num_tokens": 171430369.0, + "step": 9352 + }, + { + "epoch": 0.7712542261070339, + "grad_norm": 1.8252544403076172, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7850934863090515, + "num_tokens": 171461473.0, + "step": 9353 + }, + { + "epoch": 0.7713366867320854, + "grad_norm": 1.7887486219406128, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7573749423027039, + "num_tokens": 171492547.0, + "step": 9354 + }, + { + "epoch": 0.771419147357137, + "grad_norm": 1.8254467248916626, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7375931739807129, + "num_tokens": 171526750.0, + "step": 9355 + }, + { + "epoch": 0.7715016079821885, + "grad_norm": 2.066020965576172, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7353227138519287, + "num_tokens": 171556077.0, + "step": 9356 + }, + { + "epoch": 0.77158406860724, + "grad_norm": 1.6846424341201782, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7759498357772827, + "num_tokens": 171591333.0, + "step": 9357 + }, + { + "epoch": 0.7716665292322916, + "grad_norm": 2.1580536365509033, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7565422058105469, + "num_tokens": 171616524.0, + "step": 9358 + }, + { + "epoch": 0.7717489898573431, + "grad_norm": 1.8572285175323486, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.7619089484214783, + "num_tokens": 171647644.0, + "step": 9359 + }, + { + "epoch": 0.7718314504823947, + "grad_norm": 1.902627944946289, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7310636043548584, + "num_tokens": 171677690.0, + "step": 9360 + }, + { + "epoch": 0.7719139111074462, + "grad_norm": 2.0864627361297607, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.731640100479126, + "num_tokens": 171704388.0, + "step": 9361 + }, + { + "epoch": 0.7719963717324977, + "grad_norm": 1.8499159812927246, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7592194676399231, + "num_tokens": 171737003.0, + "step": 9362 + }, + { + "epoch": 0.7720788323575493, + "grad_norm": 1.9008643627166748, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7430737614631653, + "num_tokens": 171768058.0, + "step": 9363 + }, + { + "epoch": 0.7721612929826008, + "grad_norm": 2.0076711177825928, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.735768735408783, + "num_tokens": 171794785.0, + "step": 9364 + }, + { + "epoch": 0.7722437536076523, + "grad_norm": 1.934317708015442, + "learning_rate": 1e-06, + "loss": 0.7808, + "mean_token_accuracy": 0.7510309219360352, + "num_tokens": 171824754.0, + "step": 9365 + }, + { + "epoch": 0.7723262142327039, + "grad_norm": 1.6449835300445557, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7620933651924133, + "num_tokens": 171862523.0, + "step": 9366 + }, + { + "epoch": 0.7724086748577554, + "grad_norm": 1.9876347780227661, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7650549411773682, + "num_tokens": 171891456.0, + "step": 9367 + }, + { + "epoch": 0.772491135482807, + "grad_norm": 1.914758563041687, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.7589657306671143, + "num_tokens": 171919600.0, + "step": 9368 + }, + { + "epoch": 0.7725735961078585, + "grad_norm": 1.4923503398895264, + "learning_rate": 1e-06, + "loss": 0.5436, + "mean_token_accuracy": 0.8237590789794922, + "num_tokens": 171957891.0, + "step": 9369 + }, + { + "epoch": 0.77265605673291, + "grad_norm": 1.7972859144210815, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7137774229049683, + "num_tokens": 171991401.0, + "step": 9370 + }, + { + "epoch": 0.7727385173579616, + "grad_norm": 2.088178873062134, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6977089643478394, + "num_tokens": 172020485.0, + "step": 9371 + }, + { + "epoch": 0.7728209779830131, + "grad_norm": 1.810204029083252, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.74421626329422, + "num_tokens": 172054847.0, + "step": 9372 + }, + { + "epoch": 0.7729034386080647, + "grad_norm": 1.7095637321472168, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7157847285270691, + "num_tokens": 172097810.0, + "step": 9373 + }, + { + "epoch": 0.7729858992331162, + "grad_norm": 2.063727617263794, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7522087693214417, + "num_tokens": 172125032.0, + "step": 9374 + }, + { + "epoch": 0.7730683598581677, + "grad_norm": 1.9798221588134766, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.7941312789916992, + "num_tokens": 172153921.0, + "step": 9375 + }, + { + "epoch": 0.7731508204832193, + "grad_norm": 1.9205743074417114, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7375608682632446, + "num_tokens": 172186718.0, + "step": 9376 + }, + { + "epoch": 0.7732332811082708, + "grad_norm": 1.8475258350372314, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7329269647598267, + "num_tokens": 172218083.0, + "step": 9377 + }, + { + "epoch": 0.7733157417333223, + "grad_norm": 2.1280527114868164, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7445386648178101, + "num_tokens": 172243060.0, + "step": 9378 + }, + { + "epoch": 0.7733982023583739, + "grad_norm": 1.864780306816101, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7413628101348877, + "num_tokens": 172276468.0, + "step": 9379 + }, + { + "epoch": 0.7734806629834254, + "grad_norm": 1.9835492372512817, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7343888282775879, + "num_tokens": 172305364.0, + "step": 9380 + }, + { + "epoch": 0.7735631236084769, + "grad_norm": 2.1746673583984375, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7405588626861572, + "num_tokens": 172330186.0, + "step": 9381 + }, + { + "epoch": 0.7736455842335285, + "grad_norm": 1.973638892173767, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7727413773536682, + "num_tokens": 172359590.0, + "step": 9382 + }, + { + "epoch": 0.77372804485858, + "grad_norm": 2.163987159729004, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.7502603530883789, + "num_tokens": 172382845.0, + "step": 9383 + }, + { + "epoch": 0.7738105054836316, + "grad_norm": 1.896515130996704, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.757677435874939, + "num_tokens": 172413167.0, + "step": 9384 + }, + { + "epoch": 0.7738929661086831, + "grad_norm": 1.8773720264434814, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7493293285369873, + "num_tokens": 172444279.0, + "step": 9385 + }, + { + "epoch": 0.7739754267337347, + "grad_norm": 2.0306897163391113, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7114273309707642, + "num_tokens": 172473879.0, + "step": 9386 + }, + { + "epoch": 0.7740578873587862, + "grad_norm": 2.0508275032043457, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7384719252586365, + "num_tokens": 172502598.0, + "step": 9387 + }, + { + "epoch": 0.7741403479838377, + "grad_norm": 1.863813042640686, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.739922821521759, + "num_tokens": 172532482.0, + "step": 9388 + }, + { + "epoch": 0.7742228086088893, + "grad_norm": 2.1741299629211426, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7370041608810425, + "num_tokens": 172557736.0, + "step": 9389 + }, + { + "epoch": 0.7743052692339408, + "grad_norm": 2.112835645675659, + "learning_rate": 1e-06, + "loss": 0.5779, + "mean_token_accuracy": 0.807379961013794, + "num_tokens": 172582109.0, + "step": 9390 + }, + { + "epoch": 0.7743877298589923, + "grad_norm": 2.007688522338867, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6946787238121033, + "num_tokens": 172613115.0, + "step": 9391 + }, + { + "epoch": 0.7744701904840439, + "grad_norm": 2.0024325847625732, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7157122492790222, + "num_tokens": 172642845.0, + "step": 9392 + }, + { + "epoch": 0.7745526511090954, + "grad_norm": 1.8355989456176758, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7513317465782166, + "num_tokens": 172674286.0, + "step": 9393 + }, + { + "epoch": 0.7746351117341469, + "grad_norm": 1.9135749340057373, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7399826645851135, + "num_tokens": 172704658.0, + "step": 9394 + }, + { + "epoch": 0.7747175723591985, + "grad_norm": 1.687695026397705, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7687985897064209, + "num_tokens": 172739753.0, + "step": 9395 + }, + { + "epoch": 0.77480003298425, + "grad_norm": 1.9835197925567627, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7461841702461243, + "num_tokens": 172768734.0, + "step": 9396 + }, + { + "epoch": 0.7748824936093015, + "grad_norm": 2.212033271789551, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7134224772453308, + "num_tokens": 172793169.0, + "step": 9397 + }, + { + "epoch": 0.7749649542343531, + "grad_norm": 1.8149776458740234, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7818193435668945, + "num_tokens": 172822682.0, + "step": 9398 + }, + { + "epoch": 0.7750474148594046, + "grad_norm": 1.8150453567504883, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.760287880897522, + "num_tokens": 172854994.0, + "step": 9399 + }, + { + "epoch": 0.7751298754844562, + "grad_norm": 1.7451140880584717, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7769147157669067, + "num_tokens": 172889602.0, + "step": 9400 + }, + { + "epoch": 0.7752123361095077, + "grad_norm": 2.0716450214385986, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7635679244995117, + "num_tokens": 172917549.0, + "step": 9401 + }, + { + "epoch": 0.7752947967345593, + "grad_norm": 1.6927369832992554, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7778651714324951, + "num_tokens": 172954296.0, + "step": 9402 + }, + { + "epoch": 0.7753772573596108, + "grad_norm": 2.0518338680267334, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7370566129684448, + "num_tokens": 172981298.0, + "step": 9403 + }, + { + "epoch": 0.7754597179846623, + "grad_norm": 1.6998581886291504, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7592206597328186, + "num_tokens": 173019582.0, + "step": 9404 + }, + { + "epoch": 0.7755421786097139, + "grad_norm": 1.8206186294555664, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7883530855178833, + "num_tokens": 173050174.0, + "step": 9405 + }, + { + "epoch": 0.7756246392347654, + "grad_norm": 1.7694202661514282, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.774387538433075, + "num_tokens": 173082817.0, + "step": 9406 + }, + { + "epoch": 0.7757070998598169, + "grad_norm": 1.9742145538330078, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7540905475616455, + "num_tokens": 173114662.0, + "step": 9407 + }, + { + "epoch": 0.7757895604848685, + "grad_norm": 2.2040960788726807, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7405892014503479, + "num_tokens": 173141604.0, + "step": 9408 + }, + { + "epoch": 0.77587202110992, + "grad_norm": 2.387911081314087, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7331442832946777, + "num_tokens": 173168988.0, + "step": 9409 + }, + { + "epoch": 0.7759544817349715, + "grad_norm": 2.027393341064453, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7270792722702026, + "num_tokens": 173199028.0, + "step": 9410 + }, + { + "epoch": 0.7760369423600231, + "grad_norm": 2.2406132221221924, + "learning_rate": 1e-06, + "loss": 0.786, + "mean_token_accuracy": 0.7504422068595886, + "num_tokens": 173224938.0, + "step": 9411 + }, + { + "epoch": 0.7761194029850746, + "grad_norm": 1.7799030542373657, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7623928785324097, + "num_tokens": 173257680.0, + "step": 9412 + }, + { + "epoch": 0.7762018636101262, + "grad_norm": 2.17040753364563, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7342103123664856, + "num_tokens": 173285408.0, + "step": 9413 + }, + { + "epoch": 0.7762843242351777, + "grad_norm": 1.8929333686828613, + "learning_rate": 1e-06, + "loss": 0.6037, + "mean_token_accuracy": 0.8025383949279785, + "num_tokens": 173315162.0, + "step": 9414 + }, + { + "epoch": 0.7763667848602293, + "grad_norm": 1.9261127710342407, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7406824827194214, + "num_tokens": 173344778.0, + "step": 9415 + }, + { + "epoch": 0.7764492454852808, + "grad_norm": 2.0749542713165283, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7170941829681396, + "num_tokens": 173374051.0, + "step": 9416 + }, + { + "epoch": 0.7765317061103323, + "grad_norm": 2.0619940757751465, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7178289890289307, + "num_tokens": 173407106.0, + "step": 9417 + }, + { + "epoch": 0.7766141667353839, + "grad_norm": 1.790448784828186, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7869954109191895, + "num_tokens": 173439380.0, + "step": 9418 + }, + { + "epoch": 0.7766966273604354, + "grad_norm": 2.114177942276001, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.723589301109314, + "num_tokens": 173465580.0, + "step": 9419 + }, + { + "epoch": 0.7767790879854869, + "grad_norm": 1.982240915298462, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7407634258270264, + "num_tokens": 173495545.0, + "step": 9420 + }, + { + "epoch": 0.7768615486105385, + "grad_norm": 2.1138994693756104, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7775681614875793, + "num_tokens": 173520233.0, + "step": 9421 + }, + { + "epoch": 0.77694400923559, + "grad_norm": 2.09210467338562, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7368699312210083, + "num_tokens": 173545989.0, + "step": 9422 + }, + { + "epoch": 0.7770264698606415, + "grad_norm": 1.8119444847106934, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7923951745033264, + "num_tokens": 173575958.0, + "step": 9423 + }, + { + "epoch": 0.7771089304856931, + "grad_norm": 1.9407521486282349, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7690635323524475, + "num_tokens": 173604128.0, + "step": 9424 + }, + { + "epoch": 0.7771913911107446, + "grad_norm": 1.9558078050613403, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.7580341696739197, + "num_tokens": 173632419.0, + "step": 9425 + }, + { + "epoch": 0.7772738517357961, + "grad_norm": 1.989701509475708, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7329568266868591, + "num_tokens": 173663046.0, + "step": 9426 + }, + { + "epoch": 0.7773563123608477, + "grad_norm": 2.0342743396759033, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7247388362884521, + "num_tokens": 173692103.0, + "step": 9427 + }, + { + "epoch": 0.7774387729858993, + "grad_norm": 1.680428385734558, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7849794626235962, + "num_tokens": 173727533.0, + "step": 9428 + }, + { + "epoch": 0.7775212336109508, + "grad_norm": 1.8508979082107544, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7377243041992188, + "num_tokens": 173759646.0, + "step": 9429 + }, + { + "epoch": 0.7776036942360023, + "grad_norm": 1.7325440645217896, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.751176118850708, + "num_tokens": 173799140.0, + "step": 9430 + }, + { + "epoch": 0.7776861548610539, + "grad_norm": 2.0347445011138916, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7653025388717651, + "num_tokens": 173827562.0, + "step": 9431 + }, + { + "epoch": 0.7777686154861054, + "grad_norm": 1.9944785833358765, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7282648682594299, + "num_tokens": 173856258.0, + "step": 9432 + }, + { + "epoch": 0.7778510761111569, + "grad_norm": 1.9894297122955322, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7281930446624756, + "num_tokens": 173885992.0, + "step": 9433 + }, + { + "epoch": 0.7779335367362085, + "grad_norm": 1.7055795192718506, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7865468263626099, + "num_tokens": 173918431.0, + "step": 9434 + }, + { + "epoch": 0.77801599736126, + "grad_norm": 1.8159478902816772, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7490591406822205, + "num_tokens": 173951666.0, + "step": 9435 + }, + { + "epoch": 0.7780984579863115, + "grad_norm": 1.767507791519165, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7632279396057129, + "num_tokens": 173984148.0, + "step": 9436 + }, + { + "epoch": 0.7781809186113631, + "grad_norm": 2.0996367931365967, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7642063498497009, + "num_tokens": 174007433.0, + "step": 9437 + }, + { + "epoch": 0.7782633792364146, + "grad_norm": 1.9844106435775757, + "learning_rate": 1e-06, + "loss": 0.7852, + "mean_token_accuracy": 0.7489586472511292, + "num_tokens": 174038095.0, + "step": 9438 + }, + { + "epoch": 0.7783458398614661, + "grad_norm": 1.726847529411316, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7311501502990723, + "num_tokens": 174074786.0, + "step": 9439 + }, + { + "epoch": 0.7784283004865177, + "grad_norm": 2.2120277881622314, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7392239570617676, + "num_tokens": 174098093.0, + "step": 9440 + }, + { + "epoch": 0.7785107611115692, + "grad_norm": 1.8456840515136719, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7341053485870361, + "num_tokens": 174129556.0, + "step": 9441 + }, + { + "epoch": 0.7785932217366207, + "grad_norm": 2.1006526947021484, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7118078470230103, + "num_tokens": 174157279.0, + "step": 9442 + }, + { + "epoch": 0.7786756823616723, + "grad_norm": 1.8826113939285278, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7620745897293091, + "num_tokens": 174189004.0, + "step": 9443 + }, + { + "epoch": 0.7787581429867239, + "grad_norm": 1.8403128385543823, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7722489833831787, + "num_tokens": 174222780.0, + "step": 9444 + }, + { + "epoch": 0.7788406036117754, + "grad_norm": 2.0706679821014404, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7357010841369629, + "num_tokens": 174252585.0, + "step": 9445 + }, + { + "epoch": 0.7789230642368269, + "grad_norm": 2.0297696590423584, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7433571815490723, + "num_tokens": 174282507.0, + "step": 9446 + }, + { + "epoch": 0.7790055248618785, + "grad_norm": 2.04245662689209, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7298991084098816, + "num_tokens": 174309789.0, + "step": 9447 + }, + { + "epoch": 0.77908798548693, + "grad_norm": 1.9092410802841187, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7247219085693359, + "num_tokens": 174343073.0, + "step": 9448 + }, + { + "epoch": 0.7791704461119815, + "grad_norm": 1.79450523853302, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.7888954877853394, + "num_tokens": 174377424.0, + "step": 9449 + }, + { + "epoch": 0.7792529067370331, + "grad_norm": 2.184634208679199, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7220755219459534, + "num_tokens": 174401710.0, + "step": 9450 + }, + { + "epoch": 0.7793353673620846, + "grad_norm": 1.9383407831192017, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7456843256950378, + "num_tokens": 174431578.0, + "step": 9451 + }, + { + "epoch": 0.7794178279871361, + "grad_norm": 2.1276745796203613, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7293163537979126, + "num_tokens": 174459743.0, + "step": 9452 + }, + { + "epoch": 0.7795002886121877, + "grad_norm": 1.9751704931259155, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7802255153656006, + "num_tokens": 174486049.0, + "step": 9453 + }, + { + "epoch": 0.7795827492372392, + "grad_norm": 1.7949044704437256, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7780120372772217, + "num_tokens": 174518651.0, + "step": 9454 + }, + { + "epoch": 0.7796652098622907, + "grad_norm": 2.276120662689209, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7483965158462524, + "num_tokens": 174540486.0, + "step": 9455 + }, + { + "epoch": 0.7797476704873423, + "grad_norm": 1.9014636278152466, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7549649477005005, + "num_tokens": 174572979.0, + "step": 9456 + }, + { + "epoch": 0.7798301311123939, + "grad_norm": 2.092698097229004, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7278290390968323, + "num_tokens": 174601310.0, + "step": 9457 + }, + { + "epoch": 0.7799125917374453, + "grad_norm": 1.8602759838104248, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.757890522480011, + "num_tokens": 174631240.0, + "step": 9458 + }, + { + "epoch": 0.7799950523624969, + "grad_norm": 2.1815450191497803, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7002425193786621, + "num_tokens": 174657867.0, + "step": 9459 + }, + { + "epoch": 0.7800775129875485, + "grad_norm": 2.01411509513855, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7568821907043457, + "num_tokens": 174685082.0, + "step": 9460 + }, + { + "epoch": 0.7801599736126, + "grad_norm": 1.8726227283477783, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7813261151313782, + "num_tokens": 174716598.0, + "step": 9461 + }, + { + "epoch": 0.7802424342376515, + "grad_norm": 2.053051471710205, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7061195373535156, + "num_tokens": 174746567.0, + "step": 9462 + }, + { + "epoch": 0.7803248948627031, + "grad_norm": 1.693472981452942, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7398054599761963, + "num_tokens": 174788844.0, + "step": 9463 + }, + { + "epoch": 0.7804073554877546, + "grad_norm": 2.276838779449463, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7314908504486084, + "num_tokens": 174813882.0, + "step": 9464 + }, + { + "epoch": 0.7804898161128061, + "grad_norm": 1.9644265174865723, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.7870258688926697, + "num_tokens": 174844012.0, + "step": 9465 + }, + { + "epoch": 0.7805722767378577, + "grad_norm": 1.767348051071167, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6983421444892883, + "num_tokens": 174882752.0, + "step": 9466 + }, + { + "epoch": 0.7806547373629092, + "grad_norm": 2.1260335445404053, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7478057146072388, + "num_tokens": 174909940.0, + "step": 9467 + }, + { + "epoch": 0.7807371979879607, + "grad_norm": 2.1053426265716553, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7297804355621338, + "num_tokens": 174937219.0, + "step": 9468 + }, + { + "epoch": 0.7808196586130123, + "grad_norm": 2.0261003971099854, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7574805021286011, + "num_tokens": 174966103.0, + "step": 9469 + }, + { + "epoch": 0.7809021192380639, + "grad_norm": 1.9101145267486572, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7447187900543213, + "num_tokens": 174996392.0, + "step": 9470 + }, + { + "epoch": 0.7809845798631153, + "grad_norm": 1.9262807369232178, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7496581077575684, + "num_tokens": 175031860.0, + "step": 9471 + }, + { + "epoch": 0.7810670404881669, + "grad_norm": 1.769757628440857, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7848479747772217, + "num_tokens": 175068665.0, + "step": 9472 + }, + { + "epoch": 0.7811495011132185, + "grad_norm": 1.8150612115859985, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7636042833328247, + "num_tokens": 175099716.0, + "step": 9473 + }, + { + "epoch": 0.78123196173827, + "grad_norm": 2.0029640197753906, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.772413969039917, + "num_tokens": 175128226.0, + "step": 9474 + }, + { + "epoch": 0.7813144223633215, + "grad_norm": 1.8116103410720825, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7118750810623169, + "num_tokens": 175163675.0, + "step": 9475 + }, + { + "epoch": 0.7813968829883731, + "grad_norm": 1.7947014570236206, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7509707808494568, + "num_tokens": 175198041.0, + "step": 9476 + }, + { + "epoch": 0.7814793436134246, + "grad_norm": 2.3727753162384033, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7243901491165161, + "num_tokens": 175219463.0, + "step": 9477 + }, + { + "epoch": 0.7815618042384761, + "grad_norm": 2.1939587593078613, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7440049052238464, + "num_tokens": 175245461.0, + "step": 9478 + }, + { + "epoch": 0.7816442648635277, + "grad_norm": 1.8388571739196777, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7611697912216187, + "num_tokens": 175277677.0, + "step": 9479 + }, + { + "epoch": 0.7817267254885792, + "grad_norm": 2.1160531044006348, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7209404110908508, + "num_tokens": 175303344.0, + "step": 9480 + }, + { + "epoch": 0.7818091861136307, + "grad_norm": 1.8861452341079712, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.796106219291687, + "num_tokens": 175331606.0, + "step": 9481 + }, + { + "epoch": 0.7818916467386823, + "grad_norm": 1.9411814212799072, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7852098941802979, + "num_tokens": 175360308.0, + "step": 9482 + }, + { + "epoch": 0.7819741073637339, + "grad_norm": 1.9827362298965454, + "learning_rate": 1e-06, + "loss": 0.7598, + "mean_token_accuracy": 0.7564388513565063, + "num_tokens": 175388996.0, + "step": 9483 + }, + { + "epoch": 0.7820565679887853, + "grad_norm": 1.8033075332641602, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7281132936477661, + "num_tokens": 175425200.0, + "step": 9484 + }, + { + "epoch": 0.7821390286138369, + "grad_norm": 1.9759308099746704, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7683861255645752, + "num_tokens": 175453568.0, + "step": 9485 + }, + { + "epoch": 0.7822214892388885, + "grad_norm": 1.8972126245498657, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.8008536696434021, + "num_tokens": 175481796.0, + "step": 9486 + }, + { + "epoch": 0.7823039498639399, + "grad_norm": 2.274477005004883, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.758984386920929, + "num_tokens": 175503372.0, + "step": 9487 + }, + { + "epoch": 0.7823864104889915, + "grad_norm": 1.9809160232543945, + "learning_rate": 1e-06, + "loss": 0.7842, + "mean_token_accuracy": 0.7558459043502808, + "num_tokens": 175533794.0, + "step": 9488 + }, + { + "epoch": 0.7824688711140431, + "grad_norm": 2.0939431190490723, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7428664565086365, + "num_tokens": 175561886.0, + "step": 9489 + }, + { + "epoch": 0.7825513317390946, + "grad_norm": 1.6811879873275757, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7351274490356445, + "num_tokens": 175600014.0, + "step": 9490 + }, + { + "epoch": 0.7826337923641461, + "grad_norm": 1.9809856414794922, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7362098693847656, + "num_tokens": 175629857.0, + "step": 9491 + }, + { + "epoch": 0.7827162529891977, + "grad_norm": 2.152204751968384, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7596935033798218, + "num_tokens": 175653937.0, + "step": 9492 + }, + { + "epoch": 0.7827987136142492, + "grad_norm": 2.1155929565429688, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7205200791358948, + "num_tokens": 175679155.0, + "step": 9493 + }, + { + "epoch": 0.7828811742393007, + "grad_norm": 1.9297984838485718, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.751333475112915, + "num_tokens": 175709553.0, + "step": 9494 + }, + { + "epoch": 0.7829636348643523, + "grad_norm": 2.1552200317382812, + "learning_rate": 1e-06, + "loss": 0.6012, + "mean_token_accuracy": 0.799354076385498, + "num_tokens": 175741463.0, + "step": 9495 + }, + { + "epoch": 0.7830460954894038, + "grad_norm": 2.168637275695801, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7774698734283447, + "num_tokens": 175767215.0, + "step": 9496 + }, + { + "epoch": 0.7831285561144553, + "grad_norm": 1.9337584972381592, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7456862926483154, + "num_tokens": 175794657.0, + "step": 9497 + }, + { + "epoch": 0.7832110167395069, + "grad_norm": 1.9194753170013428, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.714430570602417, + "num_tokens": 175826568.0, + "step": 9498 + }, + { + "epoch": 0.7832934773645585, + "grad_norm": 2.2713513374328613, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7412252426147461, + "num_tokens": 175856156.0, + "step": 9499 + }, + { + "epoch": 0.7833759379896099, + "grad_norm": 1.907036304473877, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7552508115768433, + "num_tokens": 175888243.0, + "step": 9500 + }, + { + "epoch": 0.7834583986146615, + "grad_norm": 1.9467135667800903, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7439571619033813, + "num_tokens": 175916811.0, + "step": 9501 + }, + { + "epoch": 0.7835408592397131, + "grad_norm": 2.018305778503418, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7050222158432007, + "num_tokens": 175947608.0, + "step": 9502 + }, + { + "epoch": 0.7836233198647645, + "grad_norm": 2.1220595836639404, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7344595193862915, + "num_tokens": 175973596.0, + "step": 9503 + }, + { + "epoch": 0.7837057804898161, + "grad_norm": 1.782239317893982, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.753929615020752, + "num_tokens": 176009620.0, + "step": 9504 + }, + { + "epoch": 0.7837882411148677, + "grad_norm": 2.112436056137085, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7417266964912415, + "num_tokens": 176038391.0, + "step": 9505 + }, + { + "epoch": 0.7838707017399192, + "grad_norm": 2.1459295749664307, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7221280336380005, + "num_tokens": 176065261.0, + "step": 9506 + }, + { + "epoch": 0.7839531623649707, + "grad_norm": 1.5920765399932861, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7459705471992493, + "num_tokens": 176105197.0, + "step": 9507 + }, + { + "epoch": 0.7840356229900223, + "grad_norm": 1.864490032196045, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7584816813468933, + "num_tokens": 176137113.0, + "step": 9508 + }, + { + "epoch": 0.7841180836150738, + "grad_norm": 1.9502981901168823, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7776532173156738, + "num_tokens": 176164157.0, + "step": 9509 + }, + { + "epoch": 0.7842005442401253, + "grad_norm": 2.1307332515716553, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7335869073867798, + "num_tokens": 176190220.0, + "step": 9510 + }, + { + "epoch": 0.7842830048651769, + "grad_norm": 2.276649236679077, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7346201539039612, + "num_tokens": 176217045.0, + "step": 9511 + }, + { + "epoch": 0.7843654654902285, + "grad_norm": 1.8187716007232666, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7478524446487427, + "num_tokens": 176249276.0, + "step": 9512 + }, + { + "epoch": 0.7844479261152799, + "grad_norm": 1.6699804067611694, + "learning_rate": 1e-06, + "loss": 0.5468, + "mean_token_accuracy": 0.8285700082778931, + "num_tokens": 176281102.0, + "step": 9513 + }, + { + "epoch": 0.7845303867403315, + "grad_norm": 2.5054943561553955, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.700452983379364, + "num_tokens": 176302220.0, + "step": 9514 + }, + { + "epoch": 0.7846128473653831, + "grad_norm": 1.8840548992156982, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7543063759803772, + "num_tokens": 176330924.0, + "step": 9515 + }, + { + "epoch": 0.7846953079904345, + "grad_norm": 1.7977131605148315, + "learning_rate": 1e-06, + "loss": 0.5822, + "mean_token_accuracy": 0.8044781684875488, + "num_tokens": 176360450.0, + "step": 9516 + }, + { + "epoch": 0.7847777686154861, + "grad_norm": 2.138683795928955, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7637156248092651, + "num_tokens": 176384559.0, + "step": 9517 + }, + { + "epoch": 0.7848602292405377, + "grad_norm": 1.828470230102539, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.770649790763855, + "num_tokens": 176416622.0, + "step": 9518 + }, + { + "epoch": 0.7849426898655891, + "grad_norm": 1.9642868041992188, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7684105038642883, + "num_tokens": 176444359.0, + "step": 9519 + }, + { + "epoch": 0.7850251504906407, + "grad_norm": 2.050328493118286, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7517032623291016, + "num_tokens": 176472853.0, + "step": 9520 + }, + { + "epoch": 0.7851076111156923, + "grad_norm": 1.9074803590774536, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7538049817085266, + "num_tokens": 176501010.0, + "step": 9521 + }, + { + "epoch": 0.7851900717407438, + "grad_norm": 2.1219265460968018, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7069225907325745, + "num_tokens": 176530932.0, + "step": 9522 + }, + { + "epoch": 0.7852725323657953, + "grad_norm": 2.600202798843384, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7434394359588623, + "num_tokens": 176548725.0, + "step": 9523 + }, + { + "epoch": 0.7853549929908469, + "grad_norm": 1.8947263956069946, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7349307537078857, + "num_tokens": 176581799.0, + "step": 9524 + }, + { + "epoch": 0.7854374536158985, + "grad_norm": 2.166844129562378, + "learning_rate": 1e-06, + "loss": 0.7767, + "mean_token_accuracy": 0.7492035627365112, + "num_tokens": 176604855.0, + "step": 9525 + }, + { + "epoch": 0.7855199142409499, + "grad_norm": 2.0508506298065186, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7653401494026184, + "num_tokens": 176631039.0, + "step": 9526 + }, + { + "epoch": 0.7856023748660015, + "grad_norm": 1.6363798379898071, + "learning_rate": 1e-06, + "loss": 0.6066, + "mean_token_accuracy": 0.7955904006958008, + "num_tokens": 176666849.0, + "step": 9527 + }, + { + "epoch": 0.7856848354910531, + "grad_norm": 2.115827798843384, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7404781579971313, + "num_tokens": 176692329.0, + "step": 9528 + }, + { + "epoch": 0.7857672961161045, + "grad_norm": 2.0708820819854736, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7624590396881104, + "num_tokens": 176715877.0, + "step": 9529 + }, + { + "epoch": 0.7858497567411561, + "grad_norm": 1.926326036453247, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7475223541259766, + "num_tokens": 176746433.0, + "step": 9530 + }, + { + "epoch": 0.7859322173662077, + "grad_norm": 2.281581401824951, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7347877025604248, + "num_tokens": 176770268.0, + "step": 9531 + }, + { + "epoch": 0.7860146779912591, + "grad_norm": 1.9952919483184814, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7758160829544067, + "num_tokens": 176801524.0, + "step": 9532 + }, + { + "epoch": 0.7860971386163107, + "grad_norm": 1.9557846784591675, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7518362998962402, + "num_tokens": 176832082.0, + "step": 9533 + }, + { + "epoch": 0.7861795992413623, + "grad_norm": 1.9064587354660034, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7686809301376343, + "num_tokens": 176861197.0, + "step": 9534 + }, + { + "epoch": 0.7862620598664138, + "grad_norm": 2.033217191696167, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7683824896812439, + "num_tokens": 176886742.0, + "step": 9535 + }, + { + "epoch": 0.7863445204914653, + "grad_norm": 2.0074706077575684, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.790452778339386, + "num_tokens": 176914316.0, + "step": 9536 + }, + { + "epoch": 0.7864269811165169, + "grad_norm": 1.8691328763961792, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7226991653442383, + "num_tokens": 176949899.0, + "step": 9537 + }, + { + "epoch": 0.7865094417415684, + "grad_norm": 2.023912191390991, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7617077827453613, + "num_tokens": 176978323.0, + "step": 9538 + }, + { + "epoch": 0.7865919023666199, + "grad_norm": 2.043654203414917, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7670947313308716, + "num_tokens": 177003786.0, + "step": 9539 + }, + { + "epoch": 0.7866743629916715, + "grad_norm": 2.116525173187256, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7080521583557129, + "num_tokens": 177031902.0, + "step": 9540 + }, + { + "epoch": 0.7867568236167231, + "grad_norm": 1.916272521018982, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7605741024017334, + "num_tokens": 177063169.0, + "step": 9541 + }, + { + "epoch": 0.7868392842417745, + "grad_norm": 1.9950848817825317, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7440871596336365, + "num_tokens": 177092471.0, + "step": 9542 + }, + { + "epoch": 0.7869217448668261, + "grad_norm": 1.8089244365692139, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7279402017593384, + "num_tokens": 177125486.0, + "step": 9543 + }, + { + "epoch": 0.7870042054918777, + "grad_norm": 2.058511257171631, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7291395664215088, + "num_tokens": 177151313.0, + "step": 9544 + }, + { + "epoch": 0.7870866661169291, + "grad_norm": 2.3681282997131348, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7239177823066711, + "num_tokens": 177172319.0, + "step": 9545 + }, + { + "epoch": 0.7871691267419807, + "grad_norm": 1.8320714235305786, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.773025393486023, + "num_tokens": 177204194.0, + "step": 9546 + }, + { + "epoch": 0.7872515873670323, + "grad_norm": 1.9694600105285645, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7609330415725708, + "num_tokens": 177233209.0, + "step": 9547 + }, + { + "epoch": 0.7873340479920837, + "grad_norm": 1.9488812685012817, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7604444026947021, + "num_tokens": 177261682.0, + "step": 9548 + }, + { + "epoch": 0.7874165086171353, + "grad_norm": 1.945807695388794, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7374507188796997, + "num_tokens": 177289675.0, + "step": 9549 + }, + { + "epoch": 0.7874989692421869, + "grad_norm": 1.636082410812378, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7878322601318359, + "num_tokens": 177328436.0, + "step": 9550 + }, + { + "epoch": 0.7875814298672384, + "grad_norm": 1.9389119148254395, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.763957142829895, + "num_tokens": 177358582.0, + "step": 9551 + }, + { + "epoch": 0.7876638904922899, + "grad_norm": 2.178471088409424, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7771328687667847, + "num_tokens": 177384021.0, + "step": 9552 + }, + { + "epoch": 0.7877463511173415, + "grad_norm": 1.7268503904342651, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7655231356620789, + "num_tokens": 177415917.0, + "step": 9553 + }, + { + "epoch": 0.787828811742393, + "grad_norm": 1.944739580154419, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.7824422717094421, + "num_tokens": 177444420.0, + "step": 9554 + }, + { + "epoch": 0.7879112723674445, + "grad_norm": 2.0627458095550537, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7101329565048218, + "num_tokens": 177477808.0, + "step": 9555 + }, + { + "epoch": 0.7879937329924961, + "grad_norm": 1.9096940755844116, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7123162746429443, + "num_tokens": 177508832.0, + "step": 9556 + }, + { + "epoch": 0.7880761936175477, + "grad_norm": 1.7336223125457764, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7413879036903381, + "num_tokens": 177542380.0, + "step": 9557 + }, + { + "epoch": 0.7881586542425991, + "grad_norm": 1.884813904762268, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7416819334030151, + "num_tokens": 177573501.0, + "step": 9558 + }, + { + "epoch": 0.7882411148676507, + "grad_norm": 2.252704381942749, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7368065118789673, + "num_tokens": 177600900.0, + "step": 9559 + }, + { + "epoch": 0.7883235754927023, + "grad_norm": 1.9320523738861084, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7247825860977173, + "num_tokens": 177633065.0, + "step": 9560 + }, + { + "epoch": 0.7884060361177537, + "grad_norm": 1.8572295904159546, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7588359117507935, + "num_tokens": 177664191.0, + "step": 9561 + }, + { + "epoch": 0.7884884967428053, + "grad_norm": 1.8745485544204712, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7704015970230103, + "num_tokens": 177693828.0, + "step": 9562 + }, + { + "epoch": 0.7885709573678569, + "grad_norm": 1.847677230834961, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7184593677520752, + "num_tokens": 177725688.0, + "step": 9563 + }, + { + "epoch": 0.7886534179929083, + "grad_norm": 1.9914960861206055, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7489039301872253, + "num_tokens": 177753882.0, + "step": 9564 + }, + { + "epoch": 0.7887358786179599, + "grad_norm": 2.0218825340270996, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7236425876617432, + "num_tokens": 177782455.0, + "step": 9565 + }, + { + "epoch": 0.7888183392430115, + "grad_norm": 1.820207953453064, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7623518109321594, + "num_tokens": 177812101.0, + "step": 9566 + }, + { + "epoch": 0.788900799868063, + "grad_norm": 2.038625717163086, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7272464036941528, + "num_tokens": 177840444.0, + "step": 9567 + }, + { + "epoch": 0.7889832604931145, + "grad_norm": 1.700770378112793, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7659093141555786, + "num_tokens": 177872898.0, + "step": 9568 + }, + { + "epoch": 0.7890657211181661, + "grad_norm": 2.1570756435394287, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7491188049316406, + "num_tokens": 177895814.0, + "step": 9569 + }, + { + "epoch": 0.7891481817432177, + "grad_norm": 1.8751246929168701, + "learning_rate": 1e-06, + "loss": 0.5761, + "mean_token_accuracy": 0.8085856437683105, + "num_tokens": 177922095.0, + "step": 9570 + }, + { + "epoch": 0.7892306423682691, + "grad_norm": 1.9697074890136719, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7172097563743591, + "num_tokens": 177952268.0, + "step": 9571 + }, + { + "epoch": 0.7893131029933207, + "grad_norm": 2.258026599884033, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7652640342712402, + "num_tokens": 177975782.0, + "step": 9572 + }, + { + "epoch": 0.7893955636183723, + "grad_norm": 1.8058114051818848, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.7647348642349243, + "num_tokens": 178010566.0, + "step": 9573 + }, + { + "epoch": 0.7894780242434237, + "grad_norm": 1.9551165103912354, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7524006366729736, + "num_tokens": 178042677.0, + "step": 9574 + }, + { + "epoch": 0.7895604848684753, + "grad_norm": 1.8656346797943115, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7424498796463013, + "num_tokens": 178077293.0, + "step": 9575 + }, + { + "epoch": 0.7896429454935269, + "grad_norm": 2.1219897270202637, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7464154958724976, + "num_tokens": 178105400.0, + "step": 9576 + }, + { + "epoch": 0.7897254061185783, + "grad_norm": 1.8288211822509766, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7740075588226318, + "num_tokens": 178138444.0, + "step": 9577 + }, + { + "epoch": 0.7898078667436299, + "grad_norm": 1.7406967878341675, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7386646270751953, + "num_tokens": 178177070.0, + "step": 9578 + }, + { + "epoch": 0.7898903273686815, + "grad_norm": 1.7743500471115112, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.724526047706604, + "num_tokens": 178213627.0, + "step": 9579 + }, + { + "epoch": 0.7899727879937329, + "grad_norm": 1.8137712478637695, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7488986849784851, + "num_tokens": 178249467.0, + "step": 9580 + }, + { + "epoch": 0.7900552486187845, + "grad_norm": 1.7794256210327148, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7241803407669067, + "num_tokens": 178287983.0, + "step": 9581 + }, + { + "epoch": 0.7901377092438361, + "grad_norm": 1.8147616386413574, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7293403744697571, + "num_tokens": 178321504.0, + "step": 9582 + }, + { + "epoch": 0.7902201698688877, + "grad_norm": 1.9389522075653076, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7673304080963135, + "num_tokens": 178349093.0, + "step": 9583 + }, + { + "epoch": 0.7903026304939391, + "grad_norm": 1.9683947563171387, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.742855429649353, + "num_tokens": 178378585.0, + "step": 9584 + }, + { + "epoch": 0.7903850911189907, + "grad_norm": 1.893811821937561, + "learning_rate": 1e-06, + "loss": 0.5642, + "mean_token_accuracy": 0.8089982867240906, + "num_tokens": 178407446.0, + "step": 9585 + }, + { + "epoch": 0.7904675517440423, + "grad_norm": 2.011115789413452, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7501968145370483, + "num_tokens": 178436340.0, + "step": 9586 + }, + { + "epoch": 0.7905500123690937, + "grad_norm": 2.1122288703918457, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7370651960372925, + "num_tokens": 178464957.0, + "step": 9587 + }, + { + "epoch": 0.7906324729941453, + "grad_norm": 1.8658809661865234, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7696084380149841, + "num_tokens": 178499274.0, + "step": 9588 + }, + { + "epoch": 0.7907149336191969, + "grad_norm": 1.8339530229568481, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7840638160705566, + "num_tokens": 178530793.0, + "step": 9589 + }, + { + "epoch": 0.7907973942442483, + "grad_norm": 2.0966975688934326, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.732576847076416, + "num_tokens": 178556186.0, + "step": 9590 + }, + { + "epoch": 0.7908798548692999, + "grad_norm": 2.0803544521331787, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6892023682594299, + "num_tokens": 178585188.0, + "step": 9591 + }, + { + "epoch": 0.7909623154943515, + "grad_norm": 1.9776175022125244, + "learning_rate": 1e-06, + "loss": 0.7642, + "mean_token_accuracy": 0.7599812746047974, + "num_tokens": 178614669.0, + "step": 9592 + }, + { + "epoch": 0.7910447761194029, + "grad_norm": 2.026027202606201, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.731791615486145, + "num_tokens": 178642836.0, + "step": 9593 + }, + { + "epoch": 0.7911272367444545, + "grad_norm": 1.989942193031311, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7244376540184021, + "num_tokens": 178674671.0, + "step": 9594 + }, + { + "epoch": 0.7912096973695061, + "grad_norm": 1.7347749471664429, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7486717700958252, + "num_tokens": 178712299.0, + "step": 9595 + }, + { + "epoch": 0.7912921579945575, + "grad_norm": 2.01420521736145, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7150524854660034, + "num_tokens": 178741967.0, + "step": 9596 + }, + { + "epoch": 0.7913746186196091, + "grad_norm": 2.0257132053375244, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.7938956022262573, + "num_tokens": 178767348.0, + "step": 9597 + }, + { + "epoch": 0.7914570792446607, + "grad_norm": 1.8881055116653442, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7366900444030762, + "num_tokens": 178797751.0, + "step": 9598 + }, + { + "epoch": 0.7915395398697123, + "grad_norm": 1.7510071992874146, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7728711366653442, + "num_tokens": 178831927.0, + "step": 9599 + }, + { + "epoch": 0.7916220004947637, + "grad_norm": 1.972376823425293, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7741040587425232, + "num_tokens": 178860556.0, + "step": 9600 + }, + { + "epoch": 0.7917044611198153, + "grad_norm": 1.8459703922271729, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7498793601989746, + "num_tokens": 178891935.0, + "step": 9601 + }, + { + "epoch": 0.7917869217448669, + "grad_norm": 2.081244707107544, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7177004814147949, + "num_tokens": 178919302.0, + "step": 9602 + }, + { + "epoch": 0.7918693823699183, + "grad_norm": 2.1539924144744873, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.7861742973327637, + "num_tokens": 178945120.0, + "step": 9603 + }, + { + "epoch": 0.7919518429949699, + "grad_norm": 2.1484994888305664, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7195820808410645, + "num_tokens": 178972393.0, + "step": 9604 + }, + { + "epoch": 0.7920343036200215, + "grad_norm": 2.0088694095611572, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7413574457168579, + "num_tokens": 179000393.0, + "step": 9605 + }, + { + "epoch": 0.7921167642450729, + "grad_norm": 1.8942846059799194, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7306278944015503, + "num_tokens": 179030833.0, + "step": 9606 + }, + { + "epoch": 0.7921992248701245, + "grad_norm": 2.545897960662842, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7104278802871704, + "num_tokens": 179050881.0, + "step": 9607 + }, + { + "epoch": 0.7922816854951761, + "grad_norm": 1.7338677644729614, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.7906426787376404, + "num_tokens": 179086638.0, + "step": 9608 + }, + { + "epoch": 0.7923641461202275, + "grad_norm": 1.807305097579956, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.7912322878837585, + "num_tokens": 179118999.0, + "step": 9609 + }, + { + "epoch": 0.7924466067452791, + "grad_norm": 1.7129418849945068, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7567222118377686, + "num_tokens": 179154774.0, + "step": 9610 + }, + { + "epoch": 0.7925290673703307, + "grad_norm": 2.139150619506836, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7543699145317078, + "num_tokens": 179180429.0, + "step": 9611 + }, + { + "epoch": 0.7926115279953823, + "grad_norm": 1.9056652784347534, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7305089235305786, + "num_tokens": 179213104.0, + "step": 9612 + }, + { + "epoch": 0.7926939886204337, + "grad_norm": 1.7814323902130127, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7580066919326782, + "num_tokens": 179247539.0, + "step": 9613 + }, + { + "epoch": 0.7927764492454853, + "grad_norm": 2.2793829441070557, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7188977003097534, + "num_tokens": 179272511.0, + "step": 9614 + }, + { + "epoch": 0.7928589098705369, + "grad_norm": 1.882575511932373, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7813596725463867, + "num_tokens": 179302797.0, + "step": 9615 + }, + { + "epoch": 0.7929413704955883, + "grad_norm": 2.043201208114624, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7673858404159546, + "num_tokens": 179329730.0, + "step": 9616 + }, + { + "epoch": 0.7930238311206399, + "grad_norm": 2.1502668857574463, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7611479759216309, + "num_tokens": 179352699.0, + "step": 9617 + }, + { + "epoch": 0.7931062917456915, + "grad_norm": 2.0396790504455566, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7251691818237305, + "num_tokens": 179383001.0, + "step": 9618 + }, + { + "epoch": 0.7931887523707429, + "grad_norm": 1.7709119319915771, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7705475091934204, + "num_tokens": 179415449.0, + "step": 9619 + }, + { + "epoch": 0.7932712129957945, + "grad_norm": 2.1726725101470947, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7458548545837402, + "num_tokens": 179442008.0, + "step": 9620 + }, + { + "epoch": 0.7933536736208461, + "grad_norm": 1.9951598644256592, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7531630992889404, + "num_tokens": 179469526.0, + "step": 9621 + }, + { + "epoch": 0.7934361342458975, + "grad_norm": 2.1174685955047607, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7234943509101868, + "num_tokens": 179497081.0, + "step": 9622 + }, + { + "epoch": 0.7935185948709491, + "grad_norm": 2.1997554302215576, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7230350971221924, + "num_tokens": 179524521.0, + "step": 9623 + }, + { + "epoch": 0.7936010554960007, + "grad_norm": 1.8279809951782227, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7047699689865112, + "num_tokens": 179563326.0, + "step": 9624 + }, + { + "epoch": 0.7936835161210521, + "grad_norm": 1.8661524057388306, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7761735916137695, + "num_tokens": 179594854.0, + "step": 9625 + }, + { + "epoch": 0.7937659767461037, + "grad_norm": 2.0991463661193848, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7674646377563477, + "num_tokens": 179620707.0, + "step": 9626 + }, + { + "epoch": 0.7938484373711553, + "grad_norm": 1.6318562030792236, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7701418399810791, + "num_tokens": 179659487.0, + "step": 9627 + }, + { + "epoch": 0.7939308979962069, + "grad_norm": 1.7938578128814697, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.7951387166976929, + "num_tokens": 179691123.0, + "step": 9628 + }, + { + "epoch": 0.7940133586212583, + "grad_norm": 2.333315372467041, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7191795706748962, + "num_tokens": 179714782.0, + "step": 9629 + }, + { + "epoch": 0.7940958192463099, + "grad_norm": 1.9667834043502808, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7597417831420898, + "num_tokens": 179741189.0, + "step": 9630 + }, + { + "epoch": 0.7941782798713615, + "grad_norm": 1.918558955192566, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7626448273658752, + "num_tokens": 179771473.0, + "step": 9631 + }, + { + "epoch": 0.7942607404964129, + "grad_norm": 1.9721624851226807, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7420695424079895, + "num_tokens": 179802272.0, + "step": 9632 + }, + { + "epoch": 0.7943432011214645, + "grad_norm": 1.9909303188323975, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.7607361078262329, + "num_tokens": 179831289.0, + "step": 9633 + }, + { + "epoch": 0.7944256617465161, + "grad_norm": 1.8917587995529175, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7404776811599731, + "num_tokens": 179860925.0, + "step": 9634 + }, + { + "epoch": 0.7945081223715675, + "grad_norm": 1.9901095628738403, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7496799230575562, + "num_tokens": 179889910.0, + "step": 9635 + }, + { + "epoch": 0.7945905829966191, + "grad_norm": 1.8647894859313965, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7522318363189697, + "num_tokens": 179921266.0, + "step": 9636 + }, + { + "epoch": 0.7946730436216707, + "grad_norm": 1.7795634269714355, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7792104482650757, + "num_tokens": 179954053.0, + "step": 9637 + }, + { + "epoch": 0.7947555042467221, + "grad_norm": 2.077152967453003, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7521568536758423, + "num_tokens": 179978802.0, + "step": 9638 + }, + { + "epoch": 0.7948379648717737, + "grad_norm": 2.0577213764190674, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7589568495750427, + "num_tokens": 180005533.0, + "step": 9639 + }, + { + "epoch": 0.7949204254968253, + "grad_norm": 2.377817392349243, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7187885046005249, + "num_tokens": 180027864.0, + "step": 9640 + }, + { + "epoch": 0.7950028861218768, + "grad_norm": 1.8915766477584839, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7791547775268555, + "num_tokens": 180057747.0, + "step": 9641 + }, + { + "epoch": 0.7950853467469283, + "grad_norm": 1.983136773109436, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.7950577735900879, + "num_tokens": 180085741.0, + "step": 9642 + }, + { + "epoch": 0.7951678073719799, + "grad_norm": 1.7059327363967896, + "learning_rate": 1e-06, + "loss": 0.5359, + "mean_token_accuracy": 0.8168758153915405, + "num_tokens": 180122449.0, + "step": 9643 + }, + { + "epoch": 0.7952502679970315, + "grad_norm": 1.8052271604537964, + "learning_rate": 1e-06, + "loss": 0.5488, + "mean_token_accuracy": 0.8039870262145996, + "num_tokens": 180153133.0, + "step": 9644 + }, + { + "epoch": 0.7953327286220829, + "grad_norm": 2.063621997833252, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7509835362434387, + "num_tokens": 180183381.0, + "step": 9645 + }, + { + "epoch": 0.7954151892471345, + "grad_norm": 1.7589123249053955, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.7911581993103027, + "num_tokens": 180216052.0, + "step": 9646 + }, + { + "epoch": 0.7954976498721861, + "grad_norm": 1.7317949533462524, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7716500759124756, + "num_tokens": 180247796.0, + "step": 9647 + }, + { + "epoch": 0.7955801104972375, + "grad_norm": 2.164426803588867, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7705051898956299, + "num_tokens": 180271102.0, + "step": 9648 + }, + { + "epoch": 0.7956625711222891, + "grad_norm": 1.7561379671096802, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7470499277114868, + "num_tokens": 180307459.0, + "step": 9649 + }, + { + "epoch": 0.7957450317473407, + "grad_norm": 1.7762774229049683, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.7881374359130859, + "num_tokens": 180339593.0, + "step": 9650 + }, + { + "epoch": 0.7958274923723921, + "grad_norm": 2.0973920822143555, + "learning_rate": 1e-06, + "loss": 1.0302, + "mean_token_accuracy": 0.6839216947555542, + "num_tokens": 180370138.0, + "step": 9651 + }, + { + "epoch": 0.7959099529974437, + "grad_norm": 2.039264678955078, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.73479163646698, + "num_tokens": 180397572.0, + "step": 9652 + }, + { + "epoch": 0.7959924136224953, + "grad_norm": 1.8413938283920288, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.7547101974487305, + "num_tokens": 180426545.0, + "step": 9653 + }, + { + "epoch": 0.7960748742475467, + "grad_norm": 2.0076348781585693, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.740420937538147, + "num_tokens": 180456940.0, + "step": 9654 + }, + { + "epoch": 0.7961573348725983, + "grad_norm": 2.169424533843994, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7546719312667847, + "num_tokens": 180483225.0, + "step": 9655 + }, + { + "epoch": 0.7962397954976499, + "grad_norm": 2.098390817642212, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7281866073608398, + "num_tokens": 180509599.0, + "step": 9656 + }, + { + "epoch": 0.7963222561227014, + "grad_norm": 1.874009370803833, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7202355861663818, + "num_tokens": 180544204.0, + "step": 9657 + }, + { + "epoch": 0.7964047167477529, + "grad_norm": 2.152451515197754, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7572301626205444, + "num_tokens": 180569863.0, + "step": 9658 + }, + { + "epoch": 0.7964871773728045, + "grad_norm": 1.5743014812469482, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.7904495000839233, + "num_tokens": 180609325.0, + "step": 9659 + }, + { + "epoch": 0.7965696379978561, + "grad_norm": 2.040989398956299, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7404707074165344, + "num_tokens": 180637926.0, + "step": 9660 + }, + { + "epoch": 0.7966520986229075, + "grad_norm": 2.1411635875701904, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7436772584915161, + "num_tokens": 180663267.0, + "step": 9661 + }, + { + "epoch": 0.7967345592479591, + "grad_norm": 2.1812310218811035, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7559065818786621, + "num_tokens": 180688874.0, + "step": 9662 + }, + { + "epoch": 0.7968170198730107, + "grad_norm": 1.6191821098327637, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.797386884689331, + "num_tokens": 180725538.0, + "step": 9663 + }, + { + "epoch": 0.7968994804980621, + "grad_norm": 1.9333264827728271, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7428281307220459, + "num_tokens": 180756060.0, + "step": 9664 + }, + { + "epoch": 0.7969819411231137, + "grad_norm": 1.8195892572402954, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7250061631202698, + "num_tokens": 180790137.0, + "step": 9665 + }, + { + "epoch": 0.7970644017481653, + "grad_norm": 2.2094569206237793, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7051115036010742, + "num_tokens": 180814987.0, + "step": 9666 + }, + { + "epoch": 0.7971468623732167, + "grad_norm": 1.8816699981689453, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7073197364807129, + "num_tokens": 180847272.0, + "step": 9667 + }, + { + "epoch": 0.7972293229982683, + "grad_norm": 1.8697720766067505, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.7984602451324463, + "num_tokens": 180877270.0, + "step": 9668 + }, + { + "epoch": 0.7973117836233199, + "grad_norm": 1.7449982166290283, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.7931157946586609, + "num_tokens": 180910298.0, + "step": 9669 + }, + { + "epoch": 0.7973942442483714, + "grad_norm": 1.97696852684021, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7368851900100708, + "num_tokens": 180941231.0, + "step": 9670 + }, + { + "epoch": 0.7974767048734229, + "grad_norm": 1.8027693033218384, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7479342222213745, + "num_tokens": 180975171.0, + "step": 9671 + }, + { + "epoch": 0.7975591654984745, + "grad_norm": 1.9490764141082764, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7779130935668945, + "num_tokens": 181003002.0, + "step": 9672 + }, + { + "epoch": 0.7976416261235261, + "grad_norm": 2.0211033821105957, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7150262594223022, + "num_tokens": 181032354.0, + "step": 9673 + }, + { + "epoch": 0.7977240867485775, + "grad_norm": 1.9694701433181763, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7482227683067322, + "num_tokens": 181060757.0, + "step": 9674 + }, + { + "epoch": 0.7978065473736291, + "grad_norm": 1.9971283674240112, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7438061833381653, + "num_tokens": 181085923.0, + "step": 9675 + }, + { + "epoch": 0.7978890079986807, + "grad_norm": 2.1227519512176514, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7418451309204102, + "num_tokens": 181111459.0, + "step": 9676 + }, + { + "epoch": 0.7979714686237321, + "grad_norm": 1.9893672466278076, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.756822943687439, + "num_tokens": 181138963.0, + "step": 9677 + }, + { + "epoch": 0.7980539292487837, + "grad_norm": 2.003174304962158, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7477323412895203, + "num_tokens": 181167845.0, + "step": 9678 + }, + { + "epoch": 0.7981363898738353, + "grad_norm": 1.9155603647232056, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7449169754981995, + "num_tokens": 181198678.0, + "step": 9679 + }, + { + "epoch": 0.7982188504988867, + "grad_norm": 1.8856289386749268, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7419800758361816, + "num_tokens": 181229089.0, + "step": 9680 + }, + { + "epoch": 0.7983013111239383, + "grad_norm": 1.9092564582824707, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7328077554702759, + "num_tokens": 181259503.0, + "step": 9681 + }, + { + "epoch": 0.7983837717489899, + "grad_norm": 2.276197910308838, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7285399436950684, + "num_tokens": 181280910.0, + "step": 9682 + }, + { + "epoch": 0.7984662323740414, + "grad_norm": 1.7779303789138794, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7612640261650085, + "num_tokens": 181315641.0, + "step": 9683 + }, + { + "epoch": 0.7985486929990929, + "grad_norm": 2.134948492050171, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7384322285652161, + "num_tokens": 181341509.0, + "step": 9684 + }, + { + "epoch": 0.7986311536241445, + "grad_norm": 1.8505489826202393, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.7844148874282837, + "num_tokens": 181373205.0, + "step": 9685 + }, + { + "epoch": 0.798713614249196, + "grad_norm": 2.110401153564453, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.7975220084190369, + "num_tokens": 181396366.0, + "step": 9686 + }, + { + "epoch": 0.7987960748742475, + "grad_norm": 2.1323468685150146, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7202486991882324, + "num_tokens": 181423872.0, + "step": 9687 + }, + { + "epoch": 0.7988785354992991, + "grad_norm": 1.9244306087493896, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7617686986923218, + "num_tokens": 181455514.0, + "step": 9688 + }, + { + "epoch": 0.7989609961243507, + "grad_norm": 1.8158472776412964, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7493122816085815, + "num_tokens": 181489859.0, + "step": 9689 + }, + { + "epoch": 0.7990434567494021, + "grad_norm": 2.033463478088379, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7304995059967041, + "num_tokens": 181520305.0, + "step": 9690 + }, + { + "epoch": 0.7991259173744537, + "grad_norm": 2.0657265186309814, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7632870674133301, + "num_tokens": 181546743.0, + "step": 9691 + }, + { + "epoch": 0.7992083779995053, + "grad_norm": 2.0805790424346924, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.714330792427063, + "num_tokens": 181576845.0, + "step": 9692 + }, + { + "epoch": 0.7992908386245567, + "grad_norm": 1.9263026714324951, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7454363107681274, + "num_tokens": 181606040.0, + "step": 9693 + }, + { + "epoch": 0.7993732992496083, + "grad_norm": 1.8124635219573975, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.7975145578384399, + "num_tokens": 181636689.0, + "step": 9694 + }, + { + "epoch": 0.7994557598746599, + "grad_norm": 2.063113212585449, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7582529783248901, + "num_tokens": 181661945.0, + "step": 9695 + }, + { + "epoch": 0.7995382204997113, + "grad_norm": 2.107062816619873, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7504336833953857, + "num_tokens": 181689070.0, + "step": 9696 + }, + { + "epoch": 0.7996206811247629, + "grad_norm": 1.7372993230819702, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.7953946590423584, + "num_tokens": 181719803.0, + "step": 9697 + }, + { + "epoch": 0.7997031417498145, + "grad_norm": 2.006770372390747, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7127915620803833, + "num_tokens": 181747000.0, + "step": 9698 + }, + { + "epoch": 0.799785602374866, + "grad_norm": 1.9104394912719727, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7647199034690857, + "num_tokens": 181777816.0, + "step": 9699 + }, + { + "epoch": 0.7998680629999175, + "grad_norm": 2.0200376510620117, + "learning_rate": 1e-06, + "loss": 0.6158, + "mean_token_accuracy": 0.7939850687980652, + "num_tokens": 181801872.0, + "step": 9700 + }, + { + "epoch": 0.7999505236249691, + "grad_norm": 2.0043210983276367, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7220456600189209, + "num_tokens": 181831330.0, + "step": 9701 + }, + { + "epoch": 0.8000329842500206, + "grad_norm": 2.025055408477783, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7646459937095642, + "num_tokens": 181858697.0, + "step": 9702 + }, + { + "epoch": 0.8001154448750721, + "grad_norm": 1.80899977684021, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7829007506370544, + "num_tokens": 181889157.0, + "step": 9703 + }, + { + "epoch": 0.8001979055001237, + "grad_norm": 1.615442156791687, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7733317613601685, + "num_tokens": 181925918.0, + "step": 9704 + }, + { + "epoch": 0.8002803661251753, + "grad_norm": 1.8374117612838745, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7690519094467163, + "num_tokens": 181956512.0, + "step": 9705 + }, + { + "epoch": 0.8003628267502267, + "grad_norm": 1.7143555879592896, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7247057557106018, + "num_tokens": 181994743.0, + "step": 9706 + }, + { + "epoch": 0.8004452873752783, + "grad_norm": 1.6874337196350098, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7434506416320801, + "num_tokens": 182034345.0, + "step": 9707 + }, + { + "epoch": 0.8005277480003299, + "grad_norm": 2.2382943630218506, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7430384159088135, + "num_tokens": 182061081.0, + "step": 9708 + }, + { + "epoch": 0.8006102086253813, + "grad_norm": 2.140928268432617, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7152358889579773, + "num_tokens": 182086268.0, + "step": 9709 + }, + { + "epoch": 0.8006926692504329, + "grad_norm": 2.4099323749542236, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7154666185379028, + "num_tokens": 182108141.0, + "step": 9710 + }, + { + "epoch": 0.8007751298754845, + "grad_norm": 1.876783847808838, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7357096672058105, + "num_tokens": 182144306.0, + "step": 9711 + }, + { + "epoch": 0.800857590500536, + "grad_norm": 1.9397183656692505, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7481208443641663, + "num_tokens": 182173310.0, + "step": 9712 + }, + { + "epoch": 0.8009400511255875, + "grad_norm": 2.1803059577941895, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7325477004051208, + "num_tokens": 182199006.0, + "step": 9713 + }, + { + "epoch": 0.8010225117506391, + "grad_norm": 1.983026385307312, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.752826452255249, + "num_tokens": 182226734.0, + "step": 9714 + }, + { + "epoch": 0.8011049723756906, + "grad_norm": 1.961992621421814, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7436915636062622, + "num_tokens": 182258082.0, + "step": 9715 + }, + { + "epoch": 0.8011874330007421, + "grad_norm": 1.6152076721191406, + "learning_rate": 1e-06, + "loss": 0.5171, + "mean_token_accuracy": 0.8279989957809448, + "num_tokens": 182295423.0, + "step": 9716 + }, + { + "epoch": 0.8012698936257937, + "grad_norm": 1.6872326135635376, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.7597013115882874, + "num_tokens": 182330684.0, + "step": 9717 + }, + { + "epoch": 0.8013523542508452, + "grad_norm": 1.8888896703720093, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7554674744606018, + "num_tokens": 182360961.0, + "step": 9718 + }, + { + "epoch": 0.8014348148758967, + "grad_norm": 2.032189130783081, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7821521759033203, + "num_tokens": 182389447.0, + "step": 9719 + }, + { + "epoch": 0.8015172755009483, + "grad_norm": 2.2170755863189697, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7286787033081055, + "num_tokens": 182413861.0, + "step": 9720 + }, + { + "epoch": 0.8015997361259999, + "grad_norm": 2.1762232780456543, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7479324340820312, + "num_tokens": 182440811.0, + "step": 9721 + }, + { + "epoch": 0.8016821967510513, + "grad_norm": 2.0018467903137207, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7388876676559448, + "num_tokens": 182470929.0, + "step": 9722 + }, + { + "epoch": 0.8017646573761029, + "grad_norm": 1.8907356262207031, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.7852225303649902, + "num_tokens": 182499439.0, + "step": 9723 + }, + { + "epoch": 0.8018471180011545, + "grad_norm": 1.7920881509780884, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7322431802749634, + "num_tokens": 182534864.0, + "step": 9724 + }, + { + "epoch": 0.801929578626206, + "grad_norm": 1.6814496517181396, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.761867344379425, + "num_tokens": 182569182.0, + "step": 9725 + }, + { + "epoch": 0.8020120392512575, + "grad_norm": 1.7779386043548584, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7552622556686401, + "num_tokens": 182605110.0, + "step": 9726 + }, + { + "epoch": 0.8020944998763091, + "grad_norm": 1.9120351076126099, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7675700783729553, + "num_tokens": 182635117.0, + "step": 9727 + }, + { + "epoch": 0.8021769605013606, + "grad_norm": 1.7101494073867798, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7633147835731506, + "num_tokens": 182670351.0, + "step": 9728 + }, + { + "epoch": 0.8022594211264121, + "grad_norm": 1.6111605167388916, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.764706552028656, + "num_tokens": 182707325.0, + "step": 9729 + }, + { + "epoch": 0.8023418817514637, + "grad_norm": 1.7347875833511353, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7636992931365967, + "num_tokens": 182742591.0, + "step": 9730 + }, + { + "epoch": 0.8024243423765152, + "grad_norm": 1.8833919763565063, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7471972703933716, + "num_tokens": 182773777.0, + "step": 9731 + }, + { + "epoch": 0.8025068030015667, + "grad_norm": 2.049349308013916, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7076129913330078, + "num_tokens": 182803149.0, + "step": 9732 + }, + { + "epoch": 0.8025892636266183, + "grad_norm": 2.117032766342163, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7461045980453491, + "num_tokens": 182826344.0, + "step": 9733 + }, + { + "epoch": 0.8026717242516699, + "grad_norm": 1.6942973136901855, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.755518913269043, + "num_tokens": 182860436.0, + "step": 9734 + }, + { + "epoch": 0.8027541848767213, + "grad_norm": 2.1257715225219727, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7511305809020996, + "num_tokens": 182886334.0, + "step": 9735 + }, + { + "epoch": 0.8028366455017729, + "grad_norm": 1.9971953630447388, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7572431564331055, + "num_tokens": 182914341.0, + "step": 9736 + }, + { + "epoch": 0.8029191061268245, + "grad_norm": 2.307053565979004, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7548282146453857, + "num_tokens": 182935797.0, + "step": 9737 + }, + { + "epoch": 0.803001566751876, + "grad_norm": 2.016296863555908, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.719185471534729, + "num_tokens": 182969291.0, + "step": 9738 + }, + { + "epoch": 0.8030840273769275, + "grad_norm": 2.11140513420105, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.7990660667419434, + "num_tokens": 182991050.0, + "step": 9739 + }, + { + "epoch": 0.8031664880019791, + "grad_norm": 1.8542888164520264, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7795043587684631, + "num_tokens": 183021549.0, + "step": 9740 + }, + { + "epoch": 0.8032489486270306, + "grad_norm": 1.9580028057098389, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7414671778678894, + "num_tokens": 183050513.0, + "step": 9741 + }, + { + "epoch": 0.8033314092520821, + "grad_norm": 1.9473494291305542, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7310538291931152, + "num_tokens": 183081331.0, + "step": 9742 + }, + { + "epoch": 0.8034138698771337, + "grad_norm": 2.1516101360321045, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.703644871711731, + "num_tokens": 183111929.0, + "step": 9743 + }, + { + "epoch": 0.8034963305021852, + "grad_norm": 1.9722063541412354, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7462493777275085, + "num_tokens": 183138948.0, + "step": 9744 + }, + { + "epoch": 0.8035787911272367, + "grad_norm": 1.940069317817688, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.721728503704071, + "num_tokens": 183169758.0, + "step": 9745 + }, + { + "epoch": 0.8036612517522883, + "grad_norm": 1.8169828653335571, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7163482904434204, + "num_tokens": 183206021.0, + "step": 9746 + }, + { + "epoch": 0.8037437123773398, + "grad_norm": 2.198232889175415, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7322341799736023, + "num_tokens": 183230251.0, + "step": 9747 + }, + { + "epoch": 0.8038261730023913, + "grad_norm": 2.152409315109253, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7260884046554565, + "num_tokens": 183255792.0, + "step": 9748 + }, + { + "epoch": 0.8039086336274429, + "grad_norm": 1.7123326063156128, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7750742435455322, + "num_tokens": 183290991.0, + "step": 9749 + }, + { + "epoch": 0.8039910942524945, + "grad_norm": 1.8646972179412842, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7437530755996704, + "num_tokens": 183322692.0, + "step": 9750 + }, + { + "epoch": 0.804073554877546, + "grad_norm": 1.9849752187728882, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7612598538398743, + "num_tokens": 183351145.0, + "step": 9751 + }, + { + "epoch": 0.8041560155025975, + "grad_norm": 2.024824380874634, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7519789934158325, + "num_tokens": 183378277.0, + "step": 9752 + }, + { + "epoch": 0.8042384761276491, + "grad_norm": 1.9692137241363525, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7724965810775757, + "num_tokens": 183404678.0, + "step": 9753 + }, + { + "epoch": 0.8043209367527006, + "grad_norm": 1.9224528074264526, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7600085139274597, + "num_tokens": 183436032.0, + "step": 9754 + }, + { + "epoch": 0.8044033973777521, + "grad_norm": 1.760770320892334, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7754019498825073, + "num_tokens": 183468546.0, + "step": 9755 + }, + { + "epoch": 0.8044858580028037, + "grad_norm": 1.8204433917999268, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7477592825889587, + "num_tokens": 183506249.0, + "step": 9756 + }, + { + "epoch": 0.8045683186278552, + "grad_norm": 2.0216383934020996, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7345645427703857, + "num_tokens": 183536849.0, + "step": 9757 + }, + { + "epoch": 0.8046507792529067, + "grad_norm": 1.9073188304901123, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7539110779762268, + "num_tokens": 183568710.0, + "step": 9758 + }, + { + "epoch": 0.8047332398779583, + "grad_norm": 1.8998931646347046, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7682048678398132, + "num_tokens": 183598408.0, + "step": 9759 + }, + { + "epoch": 0.8048157005030098, + "grad_norm": 1.9684628248214722, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.766847550868988, + "num_tokens": 183624626.0, + "step": 9760 + }, + { + "epoch": 0.8048981611280613, + "grad_norm": 2.011892557144165, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7296062707901001, + "num_tokens": 183652041.0, + "step": 9761 + }, + { + "epoch": 0.8049806217531129, + "grad_norm": 1.860494613647461, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7180583477020264, + "num_tokens": 183686204.0, + "step": 9762 + }, + { + "epoch": 0.8050630823781644, + "grad_norm": 1.8285633325576782, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7840059995651245, + "num_tokens": 183719904.0, + "step": 9763 + }, + { + "epoch": 0.805145543003216, + "grad_norm": 1.8351212739944458, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7691894173622131, + "num_tokens": 183750616.0, + "step": 9764 + }, + { + "epoch": 0.8052280036282675, + "grad_norm": 1.6775702238082886, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7170206308364868, + "num_tokens": 183793390.0, + "step": 9765 + }, + { + "epoch": 0.8053104642533191, + "grad_norm": 2.022735357284546, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7484179735183716, + "num_tokens": 183819555.0, + "step": 9766 + }, + { + "epoch": 0.8053929248783706, + "grad_norm": 1.6823294162750244, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7423375248908997, + "num_tokens": 183857685.0, + "step": 9767 + }, + { + "epoch": 0.8054753855034221, + "grad_norm": 1.9959795475006104, + "learning_rate": 1e-06, + "loss": 0.7821, + "mean_token_accuracy": 0.7492034435272217, + "num_tokens": 183884109.0, + "step": 9768 + }, + { + "epoch": 0.8055578461284737, + "grad_norm": 2.0975406169891357, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7350775599479675, + "num_tokens": 183911597.0, + "step": 9769 + }, + { + "epoch": 0.8056403067535252, + "grad_norm": 1.6765965223312378, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7441238164901733, + "num_tokens": 183950543.0, + "step": 9770 + }, + { + "epoch": 0.8057227673785767, + "grad_norm": 1.8841558694839478, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.7526299953460693, + "num_tokens": 183979537.0, + "step": 9771 + }, + { + "epoch": 0.8058052280036283, + "grad_norm": 1.9299970865249634, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7337291240692139, + "num_tokens": 184009046.0, + "step": 9772 + }, + { + "epoch": 0.8058876886286798, + "grad_norm": 1.8786852359771729, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7419957518577576, + "num_tokens": 184041974.0, + "step": 9773 + }, + { + "epoch": 0.8059701492537313, + "grad_norm": 1.8607847690582275, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7797167301177979, + "num_tokens": 184071141.0, + "step": 9774 + }, + { + "epoch": 0.8060526098787829, + "grad_norm": 2.0788824558258057, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7254421710968018, + "num_tokens": 184099377.0, + "step": 9775 + }, + { + "epoch": 0.8061350705038344, + "grad_norm": 1.9170727729797363, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7485888004302979, + "num_tokens": 184129384.0, + "step": 9776 + }, + { + "epoch": 0.806217531128886, + "grad_norm": 2.0690715312957764, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7337416410446167, + "num_tokens": 184159302.0, + "step": 9777 + }, + { + "epoch": 0.8062999917539375, + "grad_norm": 2.114048957824707, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7525471448898315, + "num_tokens": 184188540.0, + "step": 9778 + }, + { + "epoch": 0.806382452378989, + "grad_norm": 1.943865180015564, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7622964978218079, + "num_tokens": 184217195.0, + "step": 9779 + }, + { + "epoch": 0.8064649130040406, + "grad_norm": 1.9648209810256958, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7273733615875244, + "num_tokens": 184247396.0, + "step": 9780 + }, + { + "epoch": 0.8065473736290921, + "grad_norm": 1.8272427320480347, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7356979846954346, + "num_tokens": 184279797.0, + "step": 9781 + }, + { + "epoch": 0.8066298342541437, + "grad_norm": 2.025334119796753, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7715209722518921, + "num_tokens": 184304551.0, + "step": 9782 + }, + { + "epoch": 0.8067122948791952, + "grad_norm": 1.9123823642730713, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7759066820144653, + "num_tokens": 184334796.0, + "step": 9783 + }, + { + "epoch": 0.8067947555042467, + "grad_norm": 2.105379819869995, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7271677255630493, + "num_tokens": 184360073.0, + "step": 9784 + }, + { + "epoch": 0.8068772161292983, + "grad_norm": 1.9828712940216064, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7717853784561157, + "num_tokens": 184390330.0, + "step": 9785 + }, + { + "epoch": 0.8069596767543498, + "grad_norm": 1.9527597427368164, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7416123151779175, + "num_tokens": 184421150.0, + "step": 9786 + }, + { + "epoch": 0.8070421373794013, + "grad_norm": 1.9944161176681519, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.719550609588623, + "num_tokens": 184451908.0, + "step": 9787 + }, + { + "epoch": 0.8071245980044529, + "grad_norm": 2.2368268966674805, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7005506157875061, + "num_tokens": 184476052.0, + "step": 9788 + }, + { + "epoch": 0.8072070586295044, + "grad_norm": 1.9605458974838257, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7699301838874817, + "num_tokens": 184504024.0, + "step": 9789 + }, + { + "epoch": 0.8072895192545559, + "grad_norm": 1.9357234239578247, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7673238515853882, + "num_tokens": 184537519.0, + "step": 9790 + }, + { + "epoch": 0.8073719798796075, + "grad_norm": 1.9973816871643066, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7331224083900452, + "num_tokens": 184568192.0, + "step": 9791 + }, + { + "epoch": 0.807454440504659, + "grad_norm": 1.9066563844680786, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7798146605491638, + "num_tokens": 184597914.0, + "step": 9792 + }, + { + "epoch": 0.8075369011297105, + "grad_norm": 2.1118662357330322, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7547955513000488, + "num_tokens": 184622948.0, + "step": 9793 + }, + { + "epoch": 0.8076193617547621, + "grad_norm": 1.725404977798462, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7525685429573059, + "num_tokens": 184661696.0, + "step": 9794 + }, + { + "epoch": 0.8077018223798137, + "grad_norm": 2.177736282348633, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7293827533721924, + "num_tokens": 184687523.0, + "step": 9795 + }, + { + "epoch": 0.8077842830048652, + "grad_norm": 2.0156800746917725, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7617126703262329, + "num_tokens": 184714955.0, + "step": 9796 + }, + { + "epoch": 0.8078667436299167, + "grad_norm": 1.860169529914856, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.7993801236152649, + "num_tokens": 184746908.0, + "step": 9797 + }, + { + "epoch": 0.8079492042549683, + "grad_norm": 1.7748359441757202, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7830643653869629, + "num_tokens": 184779942.0, + "step": 9798 + }, + { + "epoch": 0.8080316648800198, + "grad_norm": 1.9122635126113892, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7482602596282959, + "num_tokens": 184807912.0, + "step": 9799 + }, + { + "epoch": 0.8081141255050713, + "grad_norm": 1.976055383682251, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7659685015678406, + "num_tokens": 184834751.0, + "step": 9800 + }, + { + "epoch": 0.8081965861301229, + "grad_norm": 1.8687200546264648, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7480391263961792, + "num_tokens": 184867118.0, + "step": 9801 + }, + { + "epoch": 0.8082790467551744, + "grad_norm": 2.0765347480773926, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7286813259124756, + "num_tokens": 184894800.0, + "step": 9802 + }, + { + "epoch": 0.8083615073802259, + "grad_norm": 1.9108250141143799, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7437577247619629, + "num_tokens": 184924382.0, + "step": 9803 + }, + { + "epoch": 0.8084439680052775, + "grad_norm": 1.860430121421814, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.7969754934310913, + "num_tokens": 184953459.0, + "step": 9804 + }, + { + "epoch": 0.808526428630329, + "grad_norm": 1.8341678380966187, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7337628602981567, + "num_tokens": 184984738.0, + "step": 9805 + }, + { + "epoch": 0.8086088892553805, + "grad_norm": 1.9589054584503174, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7432368993759155, + "num_tokens": 185014367.0, + "step": 9806 + }, + { + "epoch": 0.8086913498804321, + "grad_norm": 1.8328677415847778, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7433550357818604, + "num_tokens": 185048359.0, + "step": 9807 + }, + { + "epoch": 0.8087738105054836, + "grad_norm": 1.893494963645935, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7566587328910828, + "num_tokens": 185080262.0, + "step": 9808 + }, + { + "epoch": 0.8088562711305352, + "grad_norm": 1.8374253511428833, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7347143292427063, + "num_tokens": 185116311.0, + "step": 9809 + }, + { + "epoch": 0.8089387317555867, + "grad_norm": 1.9935886859893799, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.751205563545227, + "num_tokens": 185143870.0, + "step": 9810 + }, + { + "epoch": 0.8090211923806383, + "grad_norm": 1.72432279586792, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7645803093910217, + "num_tokens": 185183010.0, + "step": 9811 + }, + { + "epoch": 0.8091036530056898, + "grad_norm": 2.1008191108703613, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7350445985794067, + "num_tokens": 185208930.0, + "step": 9812 + }, + { + "epoch": 0.8091861136307413, + "grad_norm": 1.9773242473602295, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7666778564453125, + "num_tokens": 185237348.0, + "step": 9813 + }, + { + "epoch": 0.8092685742557929, + "grad_norm": 2.070951461791992, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7638048529624939, + "num_tokens": 185266212.0, + "step": 9814 + }, + { + "epoch": 0.8093510348808444, + "grad_norm": 1.7913942337036133, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7525524497032166, + "num_tokens": 185301120.0, + "step": 9815 + }, + { + "epoch": 0.8094334955058959, + "grad_norm": 2.0296761989593506, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7562741041183472, + "num_tokens": 185328950.0, + "step": 9816 + }, + { + "epoch": 0.8095159561309475, + "grad_norm": 1.9775549173355103, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7212511301040649, + "num_tokens": 185360749.0, + "step": 9817 + }, + { + "epoch": 0.809598416755999, + "grad_norm": 1.986501693725586, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7235418558120728, + "num_tokens": 185391101.0, + "step": 9818 + }, + { + "epoch": 0.8096808773810505, + "grad_norm": 1.8814882040023804, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7913769483566284, + "num_tokens": 185421389.0, + "step": 9819 + }, + { + "epoch": 0.8097633380061021, + "grad_norm": 1.9319745302200317, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7278786897659302, + "num_tokens": 185452001.0, + "step": 9820 + }, + { + "epoch": 0.8098457986311536, + "grad_norm": 1.9563825130462646, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7664299607276917, + "num_tokens": 185480008.0, + "step": 9821 + }, + { + "epoch": 0.8099282592562052, + "grad_norm": 1.9278682470321655, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7421574592590332, + "num_tokens": 185509778.0, + "step": 9822 + }, + { + "epoch": 0.8100107198812567, + "grad_norm": 1.848526954650879, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7645761966705322, + "num_tokens": 185541727.0, + "step": 9823 + }, + { + "epoch": 0.8100931805063082, + "grad_norm": 1.7690573930740356, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7376377582550049, + "num_tokens": 185574225.0, + "step": 9824 + }, + { + "epoch": 0.8101756411313598, + "grad_norm": 1.8233050107955933, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.7926989197731018, + "num_tokens": 185604496.0, + "step": 9825 + }, + { + "epoch": 0.8102581017564113, + "grad_norm": 1.9346168041229248, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7188907265663147, + "num_tokens": 185635416.0, + "step": 9826 + }, + { + "epoch": 0.8103405623814629, + "grad_norm": 1.9885331392288208, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7770335674285889, + "num_tokens": 185663698.0, + "step": 9827 + }, + { + "epoch": 0.8104230230065144, + "grad_norm": 1.9331703186035156, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7255380153656006, + "num_tokens": 185693547.0, + "step": 9828 + }, + { + "epoch": 0.8105054836315659, + "grad_norm": 1.96898353099823, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7285263538360596, + "num_tokens": 185723297.0, + "step": 9829 + }, + { + "epoch": 0.8105879442566175, + "grad_norm": 2.106562376022339, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7217270135879517, + "num_tokens": 185748960.0, + "step": 9830 + }, + { + "epoch": 0.810670404881669, + "grad_norm": 1.8706287145614624, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7673940658569336, + "num_tokens": 185782369.0, + "step": 9831 + }, + { + "epoch": 0.8107528655067205, + "grad_norm": 2.1575565338134766, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7215088605880737, + "num_tokens": 185809050.0, + "step": 9832 + }, + { + "epoch": 0.8108353261317721, + "grad_norm": 1.8590004444122314, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7791743278503418, + "num_tokens": 185838450.0, + "step": 9833 + }, + { + "epoch": 0.8109177867568236, + "grad_norm": 2.0554206371307373, + "learning_rate": 1e-06, + "loss": 0.5635, + "mean_token_accuracy": 0.7973945736885071, + "num_tokens": 185862351.0, + "step": 9834 + }, + { + "epoch": 0.8110002473818752, + "grad_norm": 1.7430583238601685, + "learning_rate": 1e-06, + "loss": 0.5452, + "mean_token_accuracy": 0.8187167644500732, + "num_tokens": 185894645.0, + "step": 9835 + }, + { + "epoch": 0.8110827080069267, + "grad_norm": 2.083574056625366, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7057361602783203, + "num_tokens": 185925597.0, + "step": 9836 + }, + { + "epoch": 0.8111651686319782, + "grad_norm": 1.7568559646606445, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7539781332015991, + "num_tokens": 185963357.0, + "step": 9837 + }, + { + "epoch": 0.8112476292570298, + "grad_norm": 2.0262625217437744, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7245883345603943, + "num_tokens": 185989384.0, + "step": 9838 + }, + { + "epoch": 0.8113300898820813, + "grad_norm": 1.8227366209030151, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7725119590759277, + "num_tokens": 186021752.0, + "step": 9839 + }, + { + "epoch": 0.8114125505071328, + "grad_norm": 1.9579887390136719, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7395064234733582, + "num_tokens": 186053830.0, + "step": 9840 + }, + { + "epoch": 0.8114950111321844, + "grad_norm": 2.0633749961853027, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7506181597709656, + "num_tokens": 186078106.0, + "step": 9841 + }, + { + "epoch": 0.8115774717572359, + "grad_norm": 1.8287229537963867, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7779353857040405, + "num_tokens": 186110270.0, + "step": 9842 + }, + { + "epoch": 0.8116599323822875, + "grad_norm": 1.9941694736480713, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7703518867492676, + "num_tokens": 186137705.0, + "step": 9843 + }, + { + "epoch": 0.811742393007339, + "grad_norm": 1.9256283044815063, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7599161267280579, + "num_tokens": 186166117.0, + "step": 9844 + }, + { + "epoch": 0.8118248536323905, + "grad_norm": 2.249051332473755, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7201787829399109, + "num_tokens": 186191618.0, + "step": 9845 + }, + { + "epoch": 0.8119073142574421, + "grad_norm": 2.0173988342285156, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7473118305206299, + "num_tokens": 186219680.0, + "step": 9846 + }, + { + "epoch": 0.8119897748824936, + "grad_norm": 1.9116888046264648, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7707586288452148, + "num_tokens": 186251420.0, + "step": 9847 + }, + { + "epoch": 0.8120722355075451, + "grad_norm": 2.136590003967285, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.744516134262085, + "num_tokens": 186276901.0, + "step": 9848 + }, + { + "epoch": 0.8121546961325967, + "grad_norm": 1.9831995964050293, + "learning_rate": 1e-06, + "loss": 0.6047, + "mean_token_accuracy": 0.7955597639083862, + "num_tokens": 186304920.0, + "step": 9849 + }, + { + "epoch": 0.8122371567576482, + "grad_norm": 1.7464138269424438, + "learning_rate": 1e-06, + "loss": 0.5527, + "mean_token_accuracy": 0.8197844624519348, + "num_tokens": 186337240.0, + "step": 9850 + }, + { + "epoch": 0.8123196173826998, + "grad_norm": 2.273296356201172, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7391538619995117, + "num_tokens": 186361580.0, + "step": 9851 + }, + { + "epoch": 0.8124020780077513, + "grad_norm": 1.7368438243865967, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.745602011680603, + "num_tokens": 186396535.0, + "step": 9852 + }, + { + "epoch": 0.8124845386328028, + "grad_norm": 1.9228700399398804, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7778239250183105, + "num_tokens": 186425989.0, + "step": 9853 + }, + { + "epoch": 0.8125669992578544, + "grad_norm": 1.771616816520691, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7721397876739502, + "num_tokens": 186459487.0, + "step": 9854 + }, + { + "epoch": 0.8126494598829059, + "grad_norm": 2.010154962539673, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7282624244689941, + "num_tokens": 186488387.0, + "step": 9855 + }, + { + "epoch": 0.8127319205079574, + "grad_norm": 2.0966858863830566, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7514846324920654, + "num_tokens": 186515468.0, + "step": 9856 + }, + { + "epoch": 0.812814381133009, + "grad_norm": 1.7951029539108276, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.736458420753479, + "num_tokens": 186550059.0, + "step": 9857 + }, + { + "epoch": 0.8128968417580605, + "grad_norm": 2.111684799194336, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7254678606987, + "num_tokens": 186576680.0, + "step": 9858 + }, + { + "epoch": 0.8129793023831121, + "grad_norm": 1.9752531051635742, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7686451077461243, + "num_tokens": 186604640.0, + "step": 9859 + }, + { + "epoch": 0.8130617630081636, + "grad_norm": 1.8532459735870361, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7411407232284546, + "num_tokens": 186635309.0, + "step": 9860 + }, + { + "epoch": 0.8131442236332151, + "grad_norm": 2.0604710578918457, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7631356716156006, + "num_tokens": 186662425.0, + "step": 9861 + }, + { + "epoch": 0.8132266842582667, + "grad_norm": 1.919647216796875, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7557046413421631, + "num_tokens": 186689928.0, + "step": 9862 + }, + { + "epoch": 0.8133091448833182, + "grad_norm": 1.8509283065795898, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7550637722015381, + "num_tokens": 186722167.0, + "step": 9863 + }, + { + "epoch": 0.8133916055083698, + "grad_norm": 1.8816747665405273, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7278530597686768, + "num_tokens": 186755618.0, + "step": 9864 + }, + { + "epoch": 0.8134740661334213, + "grad_norm": 1.9426735639572144, + "learning_rate": 1e-06, + "loss": 0.771, + "mean_token_accuracy": 0.7440229654312134, + "num_tokens": 186783767.0, + "step": 9865 + }, + { + "epoch": 0.8135565267584728, + "grad_norm": 1.8649253845214844, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7368829250335693, + "num_tokens": 186816087.0, + "step": 9866 + }, + { + "epoch": 0.8136389873835244, + "grad_norm": 1.7921295166015625, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.723792552947998, + "num_tokens": 186851133.0, + "step": 9867 + }, + { + "epoch": 0.8137214480085759, + "grad_norm": 1.865280270576477, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7570748329162598, + "num_tokens": 186881313.0, + "step": 9868 + }, + { + "epoch": 0.8138039086336274, + "grad_norm": 2.163501262664795, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7222414016723633, + "num_tokens": 186904157.0, + "step": 9869 + }, + { + "epoch": 0.813886369258679, + "grad_norm": 1.8145968914031982, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7911220788955688, + "num_tokens": 186933758.0, + "step": 9870 + }, + { + "epoch": 0.8139688298837305, + "grad_norm": 1.858710765838623, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7680344581604004, + "num_tokens": 186963416.0, + "step": 9871 + }, + { + "epoch": 0.8140512905087821, + "grad_norm": 1.7821635007858276, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7810397148132324, + "num_tokens": 186993578.0, + "step": 9872 + }, + { + "epoch": 0.8141337511338336, + "grad_norm": 2.0695436000823975, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7280890941619873, + "num_tokens": 187019864.0, + "step": 9873 + }, + { + "epoch": 0.8142162117588851, + "grad_norm": 2.1173248291015625, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7290027141571045, + "num_tokens": 187047723.0, + "step": 9874 + }, + { + "epoch": 0.8142986723839367, + "grad_norm": 2.160538673400879, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7699587345123291, + "num_tokens": 187071165.0, + "step": 9875 + }, + { + "epoch": 0.8143811330089882, + "grad_norm": 1.732667326927185, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7323988676071167, + "num_tokens": 187107767.0, + "step": 9876 + }, + { + "epoch": 0.8144635936340398, + "grad_norm": 1.8615796566009521, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.75232994556427, + "num_tokens": 187141697.0, + "step": 9877 + }, + { + "epoch": 0.8145460542590913, + "grad_norm": 2.038480281829834, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7576594352722168, + "num_tokens": 187168962.0, + "step": 9878 + }, + { + "epoch": 0.8146285148841428, + "grad_norm": 1.8932347297668457, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.7967565655708313, + "num_tokens": 187198720.0, + "step": 9879 + }, + { + "epoch": 0.8147109755091944, + "grad_norm": 2.200817346572876, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7293692827224731, + "num_tokens": 187224021.0, + "step": 9880 + }, + { + "epoch": 0.8147934361342459, + "grad_norm": 1.763405680656433, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7544305324554443, + "num_tokens": 187258039.0, + "step": 9881 + }, + { + "epoch": 0.8148758967592974, + "grad_norm": 1.9028983116149902, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7510204911231995, + "num_tokens": 187290489.0, + "step": 9882 + }, + { + "epoch": 0.814958357384349, + "grad_norm": 1.9524288177490234, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7774466276168823, + "num_tokens": 187319082.0, + "step": 9883 + }, + { + "epoch": 0.8150408180094005, + "grad_norm": 2.0767953395843506, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.718525767326355, + "num_tokens": 187345203.0, + "step": 9884 + }, + { + "epoch": 0.815123278634452, + "grad_norm": 1.9714103937149048, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7874274253845215, + "num_tokens": 187370427.0, + "step": 9885 + }, + { + "epoch": 0.8152057392595036, + "grad_norm": 1.7282829284667969, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7554816007614136, + "num_tokens": 187405458.0, + "step": 9886 + }, + { + "epoch": 0.8152881998845551, + "grad_norm": 2.167314052581787, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7519101500511169, + "num_tokens": 187429802.0, + "step": 9887 + }, + { + "epoch": 0.8153706605096067, + "grad_norm": 1.6776701211929321, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.7885962128639221, + "num_tokens": 187462371.0, + "step": 9888 + }, + { + "epoch": 0.8154531211346582, + "grad_norm": 1.923188328742981, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7769773006439209, + "num_tokens": 187491500.0, + "step": 9889 + }, + { + "epoch": 0.8155355817597097, + "grad_norm": 2.1885979175567627, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7364143133163452, + "num_tokens": 187518752.0, + "step": 9890 + }, + { + "epoch": 0.8156180423847613, + "grad_norm": 1.8064451217651367, + "learning_rate": 1e-06, + "loss": 0.7584, + "mean_token_accuracy": 0.7609497904777527, + "num_tokens": 187554535.0, + "step": 9891 + }, + { + "epoch": 0.8157005030098128, + "grad_norm": 1.79527747631073, + "learning_rate": 1e-06, + "loss": 0.7621, + "mean_token_accuracy": 0.7639176249504089, + "num_tokens": 187589563.0, + "step": 9892 + }, + { + "epoch": 0.8157829636348644, + "grad_norm": 2.020087957382202, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7310472726821899, + "num_tokens": 187617949.0, + "step": 9893 + }, + { + "epoch": 0.8158654242599159, + "grad_norm": 2.468999147415161, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7523075938224792, + "num_tokens": 187638522.0, + "step": 9894 + }, + { + "epoch": 0.8159478848849674, + "grad_norm": 2.1365933418273926, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.758549690246582, + "num_tokens": 187663391.0, + "step": 9895 + }, + { + "epoch": 0.816030345510019, + "grad_norm": 1.9245131015777588, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7732491493225098, + "num_tokens": 187691966.0, + "step": 9896 + }, + { + "epoch": 0.8161128061350705, + "grad_norm": 1.9304267168045044, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7594584822654724, + "num_tokens": 187721665.0, + "step": 9897 + }, + { + "epoch": 0.816195266760122, + "grad_norm": 1.6832982301712036, + "learning_rate": 1e-06, + "loss": 0.5499, + "mean_token_accuracy": 0.8111920356750488, + "num_tokens": 187753586.0, + "step": 9898 + }, + { + "epoch": 0.8162777273851736, + "grad_norm": 1.76267671585083, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7375893592834473, + "num_tokens": 187787748.0, + "step": 9899 + }, + { + "epoch": 0.8163601880102251, + "grad_norm": 1.7603126764297485, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7445694804191589, + "num_tokens": 187821608.0, + "step": 9900 + }, + { + "epoch": 0.8164426486352766, + "grad_norm": 2.039125442504883, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7762040495872498, + "num_tokens": 187851353.0, + "step": 9901 + }, + { + "epoch": 0.8165251092603282, + "grad_norm": 2.076432704925537, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7287975549697876, + "num_tokens": 187882145.0, + "step": 9902 + }, + { + "epoch": 0.8166075698853797, + "grad_norm": 2.3029592037200928, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7356521487236023, + "num_tokens": 187905531.0, + "step": 9903 + }, + { + "epoch": 0.8166900305104313, + "grad_norm": 2.38997483253479, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7162830233573914, + "num_tokens": 187927656.0, + "step": 9904 + }, + { + "epoch": 0.8167724911354828, + "grad_norm": 1.7169287204742432, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.7874587774276733, + "num_tokens": 187960627.0, + "step": 9905 + }, + { + "epoch": 0.8168549517605344, + "grad_norm": 2.0739779472351074, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.724894106388092, + "num_tokens": 187988894.0, + "step": 9906 + }, + { + "epoch": 0.8169374123855859, + "grad_norm": 1.9511798620224, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.7524471879005432, + "num_tokens": 188021909.0, + "step": 9907 + }, + { + "epoch": 0.8170198730106374, + "grad_norm": 2.021852493286133, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7414147853851318, + "num_tokens": 188050450.0, + "step": 9908 + }, + { + "epoch": 0.817102333635689, + "grad_norm": 1.8540918827056885, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7390600442886353, + "num_tokens": 188085284.0, + "step": 9909 + }, + { + "epoch": 0.8171847942607405, + "grad_norm": 2.3872134685516357, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7367619276046753, + "num_tokens": 188106830.0, + "step": 9910 + }, + { + "epoch": 0.817267254885792, + "grad_norm": 1.9034277200698853, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7338207364082336, + "num_tokens": 188134979.0, + "step": 9911 + }, + { + "epoch": 0.8173497155108436, + "grad_norm": 2.740511417388916, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7590603232383728, + "num_tokens": 188163192.0, + "step": 9912 + }, + { + "epoch": 0.8174321761358951, + "grad_norm": 1.9621329307556152, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7687993049621582, + "num_tokens": 188191084.0, + "step": 9913 + }, + { + "epoch": 0.8175146367609466, + "grad_norm": 1.800611138343811, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.799547016620636, + "num_tokens": 188223590.0, + "step": 9914 + }, + { + "epoch": 0.8175970973859982, + "grad_norm": 1.945802092552185, + "learning_rate": 1e-06, + "loss": 0.7724, + "mean_token_accuracy": 0.7617790699005127, + "num_tokens": 188252502.0, + "step": 9915 + }, + { + "epoch": 0.8176795580110497, + "grad_norm": 1.7543227672576904, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7345746755599976, + "num_tokens": 188288181.0, + "step": 9916 + }, + { + "epoch": 0.8177620186361012, + "grad_norm": 1.8805272579193115, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.7851519584655762, + "num_tokens": 188316128.0, + "step": 9917 + }, + { + "epoch": 0.8178444792611528, + "grad_norm": 1.8390405178070068, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7680255174636841, + "num_tokens": 188345925.0, + "step": 9918 + }, + { + "epoch": 0.8179269398862044, + "grad_norm": 1.9295870065689087, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7324988842010498, + "num_tokens": 188376972.0, + "step": 9919 + }, + { + "epoch": 0.8180094005112559, + "grad_norm": 1.5866481065750122, + "learning_rate": 1e-06, + "loss": 0.5705, + "mean_token_accuracy": 0.8133357763290405, + "num_tokens": 188417521.0, + "step": 9920 + }, + { + "epoch": 0.8180918611363074, + "grad_norm": 1.8122196197509766, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7560784220695496, + "num_tokens": 188449577.0, + "step": 9921 + }, + { + "epoch": 0.818174321761359, + "grad_norm": 2.1398367881774902, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.73807692527771, + "num_tokens": 188474671.0, + "step": 9922 + }, + { + "epoch": 0.8182567823864105, + "grad_norm": 2.154413938522339, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.6984697580337524, + "num_tokens": 188503016.0, + "step": 9923 + }, + { + "epoch": 0.818339243011462, + "grad_norm": 2.0272321701049805, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7331686019897461, + "num_tokens": 188531368.0, + "step": 9924 + }, + { + "epoch": 0.8184217036365136, + "grad_norm": 2.0958521366119385, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7219579219818115, + "num_tokens": 188557062.0, + "step": 9925 + }, + { + "epoch": 0.8185041642615651, + "grad_norm": 1.7218844890594482, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.77364182472229, + "num_tokens": 188591670.0, + "step": 9926 + }, + { + "epoch": 0.8185866248866166, + "grad_norm": 1.795964241027832, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7634462714195251, + "num_tokens": 188625381.0, + "step": 9927 + }, + { + "epoch": 0.8186690855116682, + "grad_norm": 1.85146963596344, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7300782203674316, + "num_tokens": 188663070.0, + "step": 9928 + }, + { + "epoch": 0.8187515461367197, + "grad_norm": 1.7600054740905762, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.763006329536438, + "num_tokens": 188698337.0, + "step": 9929 + }, + { + "epoch": 0.8188340067617712, + "grad_norm": 2.293837308883667, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7552843689918518, + "num_tokens": 188720508.0, + "step": 9930 + }, + { + "epoch": 0.8189164673868228, + "grad_norm": 2.1137402057647705, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.7576784491539001, + "num_tokens": 188747132.0, + "step": 9931 + }, + { + "epoch": 0.8189989280118743, + "grad_norm": 1.8968676328659058, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7287456393241882, + "num_tokens": 188779924.0, + "step": 9932 + }, + { + "epoch": 0.8190813886369259, + "grad_norm": 2.0602190494537354, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7655266523361206, + "num_tokens": 188806390.0, + "step": 9933 + }, + { + "epoch": 0.8191638492619774, + "grad_norm": 1.7774330377578735, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7646735906600952, + "num_tokens": 188840488.0, + "step": 9934 + }, + { + "epoch": 0.819246309887029, + "grad_norm": 1.665216088294983, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.7955522537231445, + "num_tokens": 188875785.0, + "step": 9935 + }, + { + "epoch": 0.8193287705120805, + "grad_norm": 1.9419625997543335, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7174022197723389, + "num_tokens": 188905866.0, + "step": 9936 + }, + { + "epoch": 0.819411231137132, + "grad_norm": 1.9821547269821167, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7240546345710754, + "num_tokens": 188935659.0, + "step": 9937 + }, + { + "epoch": 0.8194936917621836, + "grad_norm": 1.8404604196548462, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.766589879989624, + "num_tokens": 188964718.0, + "step": 9938 + }, + { + "epoch": 0.8195761523872351, + "grad_norm": 2.0837242603302, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.7387106418609619, + "num_tokens": 188990502.0, + "step": 9939 + }, + { + "epoch": 0.8196586130122866, + "grad_norm": 1.8933131694793701, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.733136773109436, + "num_tokens": 189021268.0, + "step": 9940 + }, + { + "epoch": 0.8197410736373382, + "grad_norm": 2.0353262424468994, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7092415690422058, + "num_tokens": 189050026.0, + "step": 9941 + }, + { + "epoch": 0.8198235342623897, + "grad_norm": 1.6147867441177368, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7769197225570679, + "num_tokens": 189085942.0, + "step": 9942 + }, + { + "epoch": 0.8199059948874412, + "grad_norm": 2.013282060623169, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7150371670722961, + "num_tokens": 189117835.0, + "step": 9943 + }, + { + "epoch": 0.8199884555124928, + "grad_norm": 2.046361207962036, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7346436977386475, + "num_tokens": 189145455.0, + "step": 9944 + }, + { + "epoch": 0.8200709161375443, + "grad_norm": 2.069115161895752, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.748110294342041, + "num_tokens": 189172142.0, + "step": 9945 + }, + { + "epoch": 0.8201533767625958, + "grad_norm": 1.9187500476837158, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7565301656723022, + "num_tokens": 189201948.0, + "step": 9946 + }, + { + "epoch": 0.8202358373876474, + "grad_norm": 2.236316442489624, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7234192490577698, + "num_tokens": 189224391.0, + "step": 9947 + }, + { + "epoch": 0.820318298012699, + "grad_norm": 1.95846426486969, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7753764390945435, + "num_tokens": 189258221.0, + "step": 9948 + }, + { + "epoch": 0.8204007586377505, + "grad_norm": 1.7866312265396118, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7197529673576355, + "num_tokens": 189297968.0, + "step": 9949 + }, + { + "epoch": 0.820483219262802, + "grad_norm": 2.105196952819824, + "learning_rate": 1e-06, + "loss": 0.7712, + "mean_token_accuracy": 0.7562534213066101, + "num_tokens": 189323529.0, + "step": 9950 + }, + { + "epoch": 0.8205656798878536, + "grad_norm": 1.6936060190200806, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7675662040710449, + "num_tokens": 189359712.0, + "step": 9951 + }, + { + "epoch": 0.8206481405129051, + "grad_norm": 1.9708043336868286, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7418636083602905, + "num_tokens": 189388995.0, + "step": 9952 + }, + { + "epoch": 0.8207306011379566, + "grad_norm": 2.021555185317993, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7106689214706421, + "num_tokens": 189419463.0, + "step": 9953 + }, + { + "epoch": 0.8208130617630082, + "grad_norm": 1.936152458190918, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.7788141965866089, + "num_tokens": 189447421.0, + "step": 9954 + }, + { + "epoch": 0.8208955223880597, + "grad_norm": 2.3548357486724854, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7583860158920288, + "num_tokens": 189468165.0, + "step": 9955 + }, + { + "epoch": 0.8209779830131112, + "grad_norm": 1.861580729484558, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7536427974700928, + "num_tokens": 189501964.0, + "step": 9956 + }, + { + "epoch": 0.8210604436381628, + "grad_norm": 1.8361743688583374, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7075284719467163, + "num_tokens": 189538846.0, + "step": 9957 + }, + { + "epoch": 0.8211429042632143, + "grad_norm": 1.9474645853042603, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7522647380828857, + "num_tokens": 189571638.0, + "step": 9958 + }, + { + "epoch": 0.8212253648882658, + "grad_norm": 1.9176266193389893, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7560490965843201, + "num_tokens": 189599614.0, + "step": 9959 + }, + { + "epoch": 0.8213078255133174, + "grad_norm": 1.8397891521453857, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.762096107006073, + "num_tokens": 189630314.0, + "step": 9960 + }, + { + "epoch": 0.821390286138369, + "grad_norm": 1.8349802494049072, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.7877237796783447, + "num_tokens": 189659266.0, + "step": 9961 + }, + { + "epoch": 0.8214727467634204, + "grad_norm": 1.9546561241149902, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7824618816375732, + "num_tokens": 189687589.0, + "step": 9962 + }, + { + "epoch": 0.821555207388472, + "grad_norm": 2.113374948501587, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7630327939987183, + "num_tokens": 189712159.0, + "step": 9963 + }, + { + "epoch": 0.8216376680135236, + "grad_norm": 1.986696481704712, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7820570468902588, + "num_tokens": 189739312.0, + "step": 9964 + }, + { + "epoch": 0.8217201286385751, + "grad_norm": 1.826738953590393, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7610514760017395, + "num_tokens": 189772579.0, + "step": 9965 + }, + { + "epoch": 0.8218025892636266, + "grad_norm": 1.919584035873413, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7621246576309204, + "num_tokens": 189800760.0, + "step": 9966 + }, + { + "epoch": 0.8218850498886782, + "grad_norm": 1.9326848983764648, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7423489093780518, + "num_tokens": 189830617.0, + "step": 9967 + }, + { + "epoch": 0.8219675105137297, + "grad_norm": 1.8747806549072266, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7564270496368408, + "num_tokens": 189861547.0, + "step": 9968 + }, + { + "epoch": 0.8220499711387812, + "grad_norm": 2.076936960220337, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.770976722240448, + "num_tokens": 189887198.0, + "step": 9969 + }, + { + "epoch": 0.8221324317638328, + "grad_norm": 1.733260989189148, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7706665992736816, + "num_tokens": 189922011.0, + "step": 9970 + }, + { + "epoch": 0.8222148923888843, + "grad_norm": 2.037792444229126, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7630800008773804, + "num_tokens": 189949332.0, + "step": 9971 + }, + { + "epoch": 0.8222973530139358, + "grad_norm": 2.068118095397949, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7396190762519836, + "num_tokens": 189974768.0, + "step": 9972 + }, + { + "epoch": 0.8223798136389874, + "grad_norm": 1.9931581020355225, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7577096223831177, + "num_tokens": 190004060.0, + "step": 9973 + }, + { + "epoch": 0.822462274264039, + "grad_norm": 1.8028641939163208, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.6989122629165649, + "num_tokens": 190040379.0, + "step": 9974 + }, + { + "epoch": 0.8225447348890904, + "grad_norm": 1.8675267696380615, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7719230651855469, + "num_tokens": 190072926.0, + "step": 9975 + }, + { + "epoch": 0.822627195514142, + "grad_norm": 1.98756742477417, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7509269714355469, + "num_tokens": 190101302.0, + "step": 9976 + }, + { + "epoch": 0.8227096561391936, + "grad_norm": 1.728308081626892, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.774019718170166, + "num_tokens": 190134690.0, + "step": 9977 + }, + { + "epoch": 0.822792116764245, + "grad_norm": 1.9228770732879639, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.7736954092979431, + "num_tokens": 190162761.0, + "step": 9978 + }, + { + "epoch": 0.8228745773892966, + "grad_norm": 1.9354292154312134, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7359675168991089, + "num_tokens": 190192473.0, + "step": 9979 + }, + { + "epoch": 0.8229570380143482, + "grad_norm": 1.8382152318954468, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7248032093048096, + "num_tokens": 190227072.0, + "step": 9980 + }, + { + "epoch": 0.8230394986393997, + "grad_norm": 1.9791721105575562, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.7806707620620728, + "num_tokens": 190253930.0, + "step": 9981 + }, + { + "epoch": 0.8231219592644512, + "grad_norm": 1.9863201379776, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7147640585899353, + "num_tokens": 190282696.0, + "step": 9982 + }, + { + "epoch": 0.8232044198895028, + "grad_norm": 1.7184158563613892, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.80840003490448, + "num_tokens": 190318484.0, + "step": 9983 + }, + { + "epoch": 0.8232868805145543, + "grad_norm": 1.896880865097046, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7570710182189941, + "num_tokens": 190347866.0, + "step": 9984 + }, + { + "epoch": 0.8233693411396058, + "grad_norm": 1.8933508396148682, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7191504240036011, + "num_tokens": 190378400.0, + "step": 9985 + }, + { + "epoch": 0.8234518017646574, + "grad_norm": 2.2778944969177246, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7474093437194824, + "num_tokens": 190401030.0, + "step": 9986 + }, + { + "epoch": 0.823534262389709, + "grad_norm": 1.9589208364486694, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.755183219909668, + "num_tokens": 190429480.0, + "step": 9987 + }, + { + "epoch": 0.8236167230147604, + "grad_norm": 1.7083412408828735, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7801013588905334, + "num_tokens": 190465932.0, + "step": 9988 + }, + { + "epoch": 0.823699183639812, + "grad_norm": 1.9013789892196655, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.7801636457443237, + "num_tokens": 190493793.0, + "step": 9989 + }, + { + "epoch": 0.8237816442648636, + "grad_norm": 1.6886353492736816, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7422776222229004, + "num_tokens": 190530771.0, + "step": 9990 + }, + { + "epoch": 0.823864104889915, + "grad_norm": 2.219299793243408, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7648016214370728, + "num_tokens": 190552914.0, + "step": 9991 + }, + { + "epoch": 0.8239465655149666, + "grad_norm": 1.7912371158599854, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.751724898815155, + "num_tokens": 190590026.0, + "step": 9992 + }, + { + "epoch": 0.8240290261400182, + "grad_norm": 2.1692652702331543, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.76448655128479, + "num_tokens": 190614787.0, + "step": 9993 + }, + { + "epoch": 0.8241114867650697, + "grad_norm": 1.7574228048324585, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7779617309570312, + "num_tokens": 190645777.0, + "step": 9994 + }, + { + "epoch": 0.8241939473901212, + "grad_norm": 2.0158097743988037, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7316572666168213, + "num_tokens": 190674673.0, + "step": 9995 + }, + { + "epoch": 0.8242764080151728, + "grad_norm": 2.019846200942993, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7401882410049438, + "num_tokens": 190704229.0, + "step": 9996 + }, + { + "epoch": 0.8243588686402243, + "grad_norm": 1.8034465312957764, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.773677408695221, + "num_tokens": 190739822.0, + "step": 9997 + }, + { + "epoch": 0.8244413292652758, + "grad_norm": 2.2853851318359375, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.741043746471405, + "num_tokens": 190765183.0, + "step": 9998 + }, + { + "epoch": 0.8245237898903274, + "grad_norm": 1.9859936237335205, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7530363202095032, + "num_tokens": 190795023.0, + "step": 9999 + }, + { + "epoch": 0.824606250515379, + "grad_norm": 1.8458096981048584, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.776448667049408, + "num_tokens": 190825357.0, + "step": 10000 + }, + { + "epoch": 0.8246887111404304, + "grad_norm": 2.3617727756500244, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7698979377746582, + "num_tokens": 190849528.0, + "step": 10001 + }, + { + "epoch": 0.824771171765482, + "grad_norm": 1.8113865852355957, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7631877660751343, + "num_tokens": 190882494.0, + "step": 10002 + }, + { + "epoch": 0.8248536323905336, + "grad_norm": 1.6739373207092285, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7805166244506836, + "num_tokens": 190918553.0, + "step": 10003 + }, + { + "epoch": 0.824936093015585, + "grad_norm": 2.0494558811187744, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.7487964034080505, + "num_tokens": 190944643.0, + "step": 10004 + }, + { + "epoch": 0.8250185536406366, + "grad_norm": 2.025742292404175, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7248562574386597, + "num_tokens": 190973295.0, + "step": 10005 + }, + { + "epoch": 0.8251010142656882, + "grad_norm": 1.8544977903366089, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7498171925544739, + "num_tokens": 191002402.0, + "step": 10006 + }, + { + "epoch": 0.8251834748907396, + "grad_norm": 2.1429569721221924, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7526330947875977, + "num_tokens": 191029851.0, + "step": 10007 + }, + { + "epoch": 0.8252659355157912, + "grad_norm": 1.8461211919784546, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.750485360622406, + "num_tokens": 191062500.0, + "step": 10008 + }, + { + "epoch": 0.8253483961408428, + "grad_norm": 2.2848658561706543, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7265128493309021, + "num_tokens": 191087847.0, + "step": 10009 + }, + { + "epoch": 0.8254308567658943, + "grad_norm": 2.1645538806915283, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7473611831665039, + "num_tokens": 191113277.0, + "step": 10010 + }, + { + "epoch": 0.8255133173909458, + "grad_norm": 2.516059398651123, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7012444734573364, + "num_tokens": 191134195.0, + "step": 10011 + }, + { + "epoch": 0.8255957780159974, + "grad_norm": 1.847664475440979, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7536145448684692, + "num_tokens": 191166487.0, + "step": 10012 + }, + { + "epoch": 0.825678238641049, + "grad_norm": 1.9249885082244873, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7373438477516174, + "num_tokens": 191199290.0, + "step": 10013 + }, + { + "epoch": 0.8257606992661004, + "grad_norm": 1.7487260103225708, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7712494730949402, + "num_tokens": 191235917.0, + "step": 10014 + }, + { + "epoch": 0.825843159891152, + "grad_norm": 1.710317611694336, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7781025171279907, + "num_tokens": 191269881.0, + "step": 10015 + }, + { + "epoch": 0.8259256205162036, + "grad_norm": 1.8401682376861572, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7698664665222168, + "num_tokens": 191302874.0, + "step": 10016 + }, + { + "epoch": 0.826008081141255, + "grad_norm": 2.1041154861450195, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7371078729629517, + "num_tokens": 191330838.0, + "step": 10017 + }, + { + "epoch": 0.8260905417663066, + "grad_norm": 1.9866687059402466, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7493231892585754, + "num_tokens": 191359918.0, + "step": 10018 + }, + { + "epoch": 0.8261730023913582, + "grad_norm": 1.8485974073410034, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7156652212142944, + "num_tokens": 191394992.0, + "step": 10019 + }, + { + "epoch": 0.8262554630164096, + "grad_norm": 1.961662769317627, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7716462016105652, + "num_tokens": 191424501.0, + "step": 10020 + }, + { + "epoch": 0.8263379236414612, + "grad_norm": 1.8572313785552979, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7287753820419312, + "num_tokens": 191457241.0, + "step": 10021 + }, + { + "epoch": 0.8264203842665128, + "grad_norm": 1.9700305461883545, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7324203252792358, + "num_tokens": 191485983.0, + "step": 10022 + }, + { + "epoch": 0.8265028448915642, + "grad_norm": 2.0057270526885986, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.740047037601471, + "num_tokens": 191515618.0, + "step": 10023 + }, + { + "epoch": 0.8265853055166158, + "grad_norm": 1.8833801746368408, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.751011848449707, + "num_tokens": 191549492.0, + "step": 10024 + }, + { + "epoch": 0.8266677661416674, + "grad_norm": 1.889615774154663, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.7608277797698975, + "num_tokens": 191582940.0, + "step": 10025 + }, + { + "epoch": 0.8267502267667189, + "grad_norm": 1.8983330726623535, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7299313545227051, + "num_tokens": 191614317.0, + "step": 10026 + }, + { + "epoch": 0.8268326873917704, + "grad_norm": 1.9750323295593262, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7316626310348511, + "num_tokens": 191641768.0, + "step": 10027 + }, + { + "epoch": 0.826915148016822, + "grad_norm": 2.124962329864502, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7274227738380432, + "num_tokens": 191670147.0, + "step": 10028 + }, + { + "epoch": 0.8269976086418735, + "grad_norm": 1.9958080053329468, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7316636443138123, + "num_tokens": 191700122.0, + "step": 10029 + }, + { + "epoch": 0.827080069266925, + "grad_norm": 1.8695406913757324, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7593817710876465, + "num_tokens": 191734017.0, + "step": 10030 + }, + { + "epoch": 0.8271625298919766, + "grad_norm": 1.716240644454956, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.7985986471176147, + "num_tokens": 191771379.0, + "step": 10031 + }, + { + "epoch": 0.8272449905170282, + "grad_norm": 1.7673512697219849, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7614411115646362, + "num_tokens": 191803937.0, + "step": 10032 + }, + { + "epoch": 0.8273274511420796, + "grad_norm": 2.124781608581543, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7287561893463135, + "num_tokens": 191832388.0, + "step": 10033 + }, + { + "epoch": 0.8274099117671312, + "grad_norm": 1.908668041229248, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7772660255432129, + "num_tokens": 191859914.0, + "step": 10034 + }, + { + "epoch": 0.8274923723921828, + "grad_norm": 1.8051350116729736, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7453892827033997, + "num_tokens": 191894278.0, + "step": 10035 + }, + { + "epoch": 0.8275748330172342, + "grad_norm": 1.9047328233718872, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7803922891616821, + "num_tokens": 191923232.0, + "step": 10036 + }, + { + "epoch": 0.8276572936422858, + "grad_norm": 2.0146310329437256, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.728277325630188, + "num_tokens": 191949749.0, + "step": 10037 + }, + { + "epoch": 0.8277397542673374, + "grad_norm": 1.9195384979248047, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7740918397903442, + "num_tokens": 191976117.0, + "step": 10038 + }, + { + "epoch": 0.8278222148923888, + "grad_norm": 1.9596707820892334, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7336557507514954, + "num_tokens": 192007544.0, + "step": 10039 + }, + { + "epoch": 0.8279046755174404, + "grad_norm": 1.8000868558883667, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7732164263725281, + "num_tokens": 192038740.0, + "step": 10040 + }, + { + "epoch": 0.827987136142492, + "grad_norm": 1.7950258255004883, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.7772232294082642, + "num_tokens": 192072630.0, + "step": 10041 + }, + { + "epoch": 0.8280695967675435, + "grad_norm": 1.9984678030014038, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.765819787979126, + "num_tokens": 192105466.0, + "step": 10042 + }, + { + "epoch": 0.828152057392595, + "grad_norm": 2.181783437728882, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6954277753829956, + "num_tokens": 192132358.0, + "step": 10043 + }, + { + "epoch": 0.8282345180176466, + "grad_norm": 1.9593225717544556, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7143193483352661, + "num_tokens": 192164746.0, + "step": 10044 + }, + { + "epoch": 0.8283169786426982, + "grad_norm": 1.991473913192749, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7177646160125732, + "num_tokens": 192192992.0, + "step": 10045 + }, + { + "epoch": 0.8283994392677496, + "grad_norm": 2.258983612060547, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7536400556564331, + "num_tokens": 192216949.0, + "step": 10046 + }, + { + "epoch": 0.8284818998928012, + "grad_norm": 2.0257015228271484, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7331600785255432, + "num_tokens": 192242353.0, + "step": 10047 + }, + { + "epoch": 0.8285643605178528, + "grad_norm": 1.8663461208343506, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7762396335601807, + "num_tokens": 192270542.0, + "step": 10048 + }, + { + "epoch": 0.8286468211429042, + "grad_norm": 2.112104654312134, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.709430992603302, + "num_tokens": 192297097.0, + "step": 10049 + }, + { + "epoch": 0.8287292817679558, + "grad_norm": 2.2030162811279297, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.7855716943740845, + "num_tokens": 192318176.0, + "step": 10050 + }, + { + "epoch": 0.8288117423930074, + "grad_norm": 1.8996859788894653, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7529096007347107, + "num_tokens": 192348080.0, + "step": 10051 + }, + { + "epoch": 0.8288942030180588, + "grad_norm": 1.9063166379928589, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.748306155204773, + "num_tokens": 192377083.0, + "step": 10052 + }, + { + "epoch": 0.8289766636431104, + "grad_norm": 1.7833895683288574, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7620145678520203, + "num_tokens": 192411131.0, + "step": 10053 + }, + { + "epoch": 0.829059124268162, + "grad_norm": 2.0746960639953613, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7208640575408936, + "num_tokens": 192440952.0, + "step": 10054 + }, + { + "epoch": 0.8291415848932135, + "grad_norm": 2.1250650882720947, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7397563457489014, + "num_tokens": 192467343.0, + "step": 10055 + }, + { + "epoch": 0.829224045518265, + "grad_norm": 2.044498920440674, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7673490047454834, + "num_tokens": 192492423.0, + "step": 10056 + }, + { + "epoch": 0.8293065061433166, + "grad_norm": 1.9744263887405396, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7699426412582397, + "num_tokens": 192519541.0, + "step": 10057 + }, + { + "epoch": 0.8293889667683682, + "grad_norm": 2.071012258529663, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7485880255699158, + "num_tokens": 192545191.0, + "step": 10058 + }, + { + "epoch": 0.8294714273934196, + "grad_norm": 1.8395346403121948, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7639346718788147, + "num_tokens": 192579337.0, + "step": 10059 + }, + { + "epoch": 0.8295538880184712, + "grad_norm": 1.817939043045044, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7072620391845703, + "num_tokens": 192616552.0, + "step": 10060 + }, + { + "epoch": 0.8296363486435228, + "grad_norm": 1.7910985946655273, + "learning_rate": 1e-06, + "loss": 0.5361, + "mean_token_accuracy": 0.8317800164222717, + "num_tokens": 192647272.0, + "step": 10061 + }, + { + "epoch": 0.8297188092685742, + "grad_norm": 1.9190069437026978, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7763752937316895, + "num_tokens": 192675685.0, + "step": 10062 + }, + { + "epoch": 0.8298012698936258, + "grad_norm": 1.898887276649475, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7318047881126404, + "num_tokens": 192706086.0, + "step": 10063 + }, + { + "epoch": 0.8298837305186774, + "grad_norm": 1.876725673675537, + "learning_rate": 1e-06, + "loss": 0.5763, + "mean_token_accuracy": 0.8059781789779663, + "num_tokens": 192736367.0, + "step": 10064 + }, + { + "epoch": 0.8299661911437288, + "grad_norm": 2.0460410118103027, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7577176690101624, + "num_tokens": 192764661.0, + "step": 10065 + }, + { + "epoch": 0.8300486517687804, + "grad_norm": 1.9853143692016602, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7636558413505554, + "num_tokens": 192794312.0, + "step": 10066 + }, + { + "epoch": 0.830131112393832, + "grad_norm": 2.1086325645446777, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7311117053031921, + "num_tokens": 192822126.0, + "step": 10067 + }, + { + "epoch": 0.8302135730188834, + "grad_norm": 2.213881015777588, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7384542226791382, + "num_tokens": 192847188.0, + "step": 10068 + }, + { + "epoch": 0.830296033643935, + "grad_norm": 1.9617265462875366, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7191052436828613, + "num_tokens": 192877402.0, + "step": 10069 + }, + { + "epoch": 0.8303784942689866, + "grad_norm": 1.8628307580947876, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7417519688606262, + "num_tokens": 192907302.0, + "step": 10070 + }, + { + "epoch": 0.8304609548940382, + "grad_norm": 2.0004801750183105, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7433452606201172, + "num_tokens": 192934388.0, + "step": 10071 + }, + { + "epoch": 0.8305434155190896, + "grad_norm": 1.9130651950836182, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.7576009035110474, + "num_tokens": 192962441.0, + "step": 10072 + }, + { + "epoch": 0.8306258761441412, + "grad_norm": 2.065502643585205, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7260563373565674, + "num_tokens": 192989349.0, + "step": 10073 + }, + { + "epoch": 0.8307083367691928, + "grad_norm": 1.823988437652588, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7721829414367676, + "num_tokens": 193021665.0, + "step": 10074 + }, + { + "epoch": 0.8307907973942442, + "grad_norm": 1.7337932586669922, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.8000588417053223, + "num_tokens": 193055560.0, + "step": 10075 + }, + { + "epoch": 0.8308732580192958, + "grad_norm": 1.9365334510803223, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7608888149261475, + "num_tokens": 193084594.0, + "step": 10076 + }, + { + "epoch": 0.8309557186443474, + "grad_norm": 2.305880308151245, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7346761226654053, + "num_tokens": 193106836.0, + "step": 10077 + }, + { + "epoch": 0.8310381792693988, + "grad_norm": 1.6543712615966797, + "learning_rate": 1e-06, + "loss": 0.5883, + "mean_token_accuracy": 0.7959712147712708, + "num_tokens": 193142637.0, + "step": 10078 + }, + { + "epoch": 0.8311206398944504, + "grad_norm": 1.8733587265014648, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7481603026390076, + "num_tokens": 193175218.0, + "step": 10079 + }, + { + "epoch": 0.831203100519502, + "grad_norm": 1.9730165004730225, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7486017942428589, + "num_tokens": 193207873.0, + "step": 10080 + }, + { + "epoch": 0.8312855611445534, + "grad_norm": 1.6994646787643433, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.7897061109542847, + "num_tokens": 193239971.0, + "step": 10081 + }, + { + "epoch": 0.831368021769605, + "grad_norm": 1.902105689048767, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.747931182384491, + "num_tokens": 193269523.0, + "step": 10082 + }, + { + "epoch": 0.8314504823946566, + "grad_norm": 1.704073190689087, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7619892954826355, + "num_tokens": 193306488.0, + "step": 10083 + }, + { + "epoch": 0.831532943019708, + "grad_norm": 1.7536545991897583, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7447569966316223, + "num_tokens": 193343509.0, + "step": 10084 + }, + { + "epoch": 0.8316154036447596, + "grad_norm": 1.8877875804901123, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7453176379203796, + "num_tokens": 193374233.0, + "step": 10085 + }, + { + "epoch": 0.8316978642698112, + "grad_norm": 2.1033875942230225, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7288520336151123, + "num_tokens": 193402147.0, + "step": 10086 + }, + { + "epoch": 0.8317803248948628, + "grad_norm": 1.8829737901687622, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7603201866149902, + "num_tokens": 193429762.0, + "step": 10087 + }, + { + "epoch": 0.8318627855199142, + "grad_norm": 1.9668391942977905, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.741414487361908, + "num_tokens": 193462884.0, + "step": 10088 + }, + { + "epoch": 0.8319452461449658, + "grad_norm": 2.230699300765991, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7492198944091797, + "num_tokens": 193486347.0, + "step": 10089 + }, + { + "epoch": 0.8320277067700174, + "grad_norm": 1.7070947885513306, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7645382881164551, + "num_tokens": 193520654.0, + "step": 10090 + }, + { + "epoch": 0.8321101673950688, + "grad_norm": 2.0518932342529297, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.750490128993988, + "num_tokens": 193547387.0, + "step": 10091 + }, + { + "epoch": 0.8321926280201204, + "grad_norm": 1.7557381391525269, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7375632524490356, + "num_tokens": 193585935.0, + "step": 10092 + }, + { + "epoch": 0.832275088645172, + "grad_norm": 1.7470695972442627, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.7973045110702515, + "num_tokens": 193618979.0, + "step": 10093 + }, + { + "epoch": 0.8323575492702234, + "grad_norm": 1.8788632154464722, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7667861580848694, + "num_tokens": 193649313.0, + "step": 10094 + }, + { + "epoch": 0.832440009895275, + "grad_norm": 1.958191990852356, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7302914261817932, + "num_tokens": 193679384.0, + "step": 10095 + }, + { + "epoch": 0.8325224705203266, + "grad_norm": 1.6450146436691284, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.742005467414856, + "num_tokens": 193718288.0, + "step": 10096 + }, + { + "epoch": 0.832604931145378, + "grad_norm": 2.0350022315979004, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.739795446395874, + "num_tokens": 193745640.0, + "step": 10097 + }, + { + "epoch": 0.8326873917704296, + "grad_norm": 1.9686996936798096, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7502219676971436, + "num_tokens": 193773795.0, + "step": 10098 + }, + { + "epoch": 0.8327698523954812, + "grad_norm": 1.8830288648605347, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.74391770362854, + "num_tokens": 193803480.0, + "step": 10099 + }, + { + "epoch": 0.8328523130205326, + "grad_norm": 2.2288146018981934, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7280986309051514, + "num_tokens": 193827650.0, + "step": 10100 + }, + { + "epoch": 0.8329347736455842, + "grad_norm": 1.7555885314941406, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7757683992385864, + "num_tokens": 193861850.0, + "step": 10101 + }, + { + "epoch": 0.8330172342706358, + "grad_norm": 1.686779499053955, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7510175704956055, + "num_tokens": 193898320.0, + "step": 10102 + }, + { + "epoch": 0.8330996948956874, + "grad_norm": 2.361372709274292, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7124567627906799, + "num_tokens": 193921721.0, + "step": 10103 + }, + { + "epoch": 0.8331821555207388, + "grad_norm": 1.7291250228881836, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.794475793838501, + "num_tokens": 193956782.0, + "step": 10104 + }, + { + "epoch": 0.8332646161457904, + "grad_norm": 1.73421311378479, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7946464419364929, + "num_tokens": 193989266.0, + "step": 10105 + }, + { + "epoch": 0.833347076770842, + "grad_norm": 2.1210100650787354, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7345362901687622, + "num_tokens": 194017990.0, + "step": 10106 + }, + { + "epoch": 0.8334295373958934, + "grad_norm": 1.838594913482666, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7447469234466553, + "num_tokens": 194051499.0, + "step": 10107 + }, + { + "epoch": 0.833511998020945, + "grad_norm": 2.0052382946014404, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7264630198478699, + "num_tokens": 194080036.0, + "step": 10108 + }, + { + "epoch": 0.8335944586459966, + "grad_norm": 2.239774703979492, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7256483435630798, + "num_tokens": 194104451.0, + "step": 10109 + }, + { + "epoch": 0.833676919271048, + "grad_norm": 1.8698580265045166, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7707337141036987, + "num_tokens": 194136851.0, + "step": 10110 + }, + { + "epoch": 0.8337593798960996, + "grad_norm": 2.4279918670654297, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7249294519424438, + "num_tokens": 194157347.0, + "step": 10111 + }, + { + "epoch": 0.8338418405211512, + "grad_norm": 1.8462145328521729, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7861522436141968, + "num_tokens": 194187253.0, + "step": 10112 + }, + { + "epoch": 0.8339243011462026, + "grad_norm": 2.137930154800415, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7189036011695862, + "num_tokens": 194214544.0, + "step": 10113 + }, + { + "epoch": 0.8340067617712542, + "grad_norm": 1.7283841371536255, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7427236437797546, + "num_tokens": 194249739.0, + "step": 10114 + }, + { + "epoch": 0.8340892223963058, + "grad_norm": 2.041442632675171, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7630019187927246, + "num_tokens": 194276245.0, + "step": 10115 + }, + { + "epoch": 0.8341716830213572, + "grad_norm": 2.104095697402954, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7610121965408325, + "num_tokens": 194303902.0, + "step": 10116 + }, + { + "epoch": 0.8342541436464088, + "grad_norm": 1.8831654787063599, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7399157881736755, + "num_tokens": 194335131.0, + "step": 10117 + }, + { + "epoch": 0.8343366042714604, + "grad_norm": 1.811776876449585, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.7904462814331055, + "num_tokens": 194367141.0, + "step": 10118 + }, + { + "epoch": 0.834419064896512, + "grad_norm": 2.0230791568756104, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7666774988174438, + "num_tokens": 194393831.0, + "step": 10119 + }, + { + "epoch": 0.8345015255215634, + "grad_norm": 2.128812074661255, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7625255584716797, + "num_tokens": 194419816.0, + "step": 10120 + }, + { + "epoch": 0.834583986146615, + "grad_norm": 2.311239242553711, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7530312538146973, + "num_tokens": 194440362.0, + "step": 10121 + }, + { + "epoch": 0.8346664467716666, + "grad_norm": 1.6858426332473755, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7593070268630981, + "num_tokens": 194479836.0, + "step": 10122 + }, + { + "epoch": 0.834748907396718, + "grad_norm": 2.0011684894561768, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.731408417224884, + "num_tokens": 194507646.0, + "step": 10123 + }, + { + "epoch": 0.8348313680217696, + "grad_norm": 1.9662219285964966, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7295317649841309, + "num_tokens": 194537377.0, + "step": 10124 + }, + { + "epoch": 0.8349138286468212, + "grad_norm": 1.9351059198379517, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7613528966903687, + "num_tokens": 194566809.0, + "step": 10125 + }, + { + "epoch": 0.8349962892718726, + "grad_norm": 1.7722846269607544, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.7977920770645142, + "num_tokens": 194599869.0, + "step": 10126 + }, + { + "epoch": 0.8350787498969242, + "grad_norm": 2.1238412857055664, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.7612552642822266, + "num_tokens": 194625911.0, + "step": 10127 + }, + { + "epoch": 0.8351612105219758, + "grad_norm": 2.0522940158843994, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7041399478912354, + "num_tokens": 194655203.0, + "step": 10128 + }, + { + "epoch": 0.8352436711470272, + "grad_norm": 1.6733797788619995, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7617877721786499, + "num_tokens": 194693272.0, + "step": 10129 + }, + { + "epoch": 0.8353261317720788, + "grad_norm": 1.954379916191101, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.715005099773407, + "num_tokens": 194723679.0, + "step": 10130 + }, + { + "epoch": 0.8354085923971304, + "grad_norm": 1.904678463935852, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7507502436637878, + "num_tokens": 194753846.0, + "step": 10131 + }, + { + "epoch": 0.835491053022182, + "grad_norm": 1.908461093902588, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7383584976196289, + "num_tokens": 194785697.0, + "step": 10132 + }, + { + "epoch": 0.8355735136472334, + "grad_norm": 1.8168230056762695, + "learning_rate": 1e-06, + "loss": 0.5916, + "mean_token_accuracy": 0.808832049369812, + "num_tokens": 194815624.0, + "step": 10133 + }, + { + "epoch": 0.835655974272285, + "grad_norm": 1.945160984992981, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7441642880439758, + "num_tokens": 194843531.0, + "step": 10134 + }, + { + "epoch": 0.8357384348973366, + "grad_norm": 1.8197563886642456, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7722464799880981, + "num_tokens": 194872999.0, + "step": 10135 + }, + { + "epoch": 0.835820895522388, + "grad_norm": 1.7749601602554321, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7571491599082947, + "num_tokens": 194908585.0, + "step": 10136 + }, + { + "epoch": 0.8359033561474396, + "grad_norm": 2.0675313472747803, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7756497263908386, + "num_tokens": 194934720.0, + "step": 10137 + }, + { + "epoch": 0.8359858167724912, + "grad_norm": 1.8969612121582031, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7666096091270447, + "num_tokens": 194965602.0, + "step": 10138 + }, + { + "epoch": 0.8360682773975426, + "grad_norm": 1.8477468490600586, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7551147937774658, + "num_tokens": 194996730.0, + "step": 10139 + }, + { + "epoch": 0.8361507380225942, + "grad_norm": 1.7931528091430664, + "learning_rate": 1e-06, + "loss": 0.5769, + "mean_token_accuracy": 0.8075950145721436, + "num_tokens": 195027521.0, + "step": 10140 + }, + { + "epoch": 0.8362331986476458, + "grad_norm": 1.7464452981948853, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7853325605392456, + "num_tokens": 195059882.0, + "step": 10141 + }, + { + "epoch": 0.8363156592726972, + "grad_norm": 1.8756377696990967, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7752867937088013, + "num_tokens": 195091334.0, + "step": 10142 + }, + { + "epoch": 0.8363981198977488, + "grad_norm": 2.152554988861084, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7104464173316956, + "num_tokens": 195117167.0, + "step": 10143 + }, + { + "epoch": 0.8364805805228004, + "grad_norm": 2.016840696334839, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7143118381500244, + "num_tokens": 195147280.0, + "step": 10144 + }, + { + "epoch": 0.8365630411478518, + "grad_norm": 1.8466486930847168, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7718302011489868, + "num_tokens": 195177608.0, + "step": 10145 + }, + { + "epoch": 0.8366455017729034, + "grad_norm": 2.026367664337158, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7959441542625427, + "num_tokens": 195205714.0, + "step": 10146 + }, + { + "epoch": 0.836727962397955, + "grad_norm": 1.7071504592895508, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7398803234100342, + "num_tokens": 195240523.0, + "step": 10147 + }, + { + "epoch": 0.8368104230230066, + "grad_norm": 1.8503535985946655, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7443495988845825, + "num_tokens": 195273788.0, + "step": 10148 + }, + { + "epoch": 0.836892883648058, + "grad_norm": 1.9975543022155762, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.731249213218689, + "num_tokens": 195302072.0, + "step": 10149 + }, + { + "epoch": 0.8369753442731096, + "grad_norm": 1.9834264516830444, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7436578273773193, + "num_tokens": 195330567.0, + "step": 10150 + }, + { + "epoch": 0.8370578048981612, + "grad_norm": 1.8350818157196045, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7391225099563599, + "num_tokens": 195364988.0, + "step": 10151 + }, + { + "epoch": 0.8371402655232126, + "grad_norm": 1.6824983358383179, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7652420997619629, + "num_tokens": 195404137.0, + "step": 10152 + }, + { + "epoch": 0.8372227261482642, + "grad_norm": 1.9896222352981567, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.741378664970398, + "num_tokens": 195435795.0, + "step": 10153 + }, + { + "epoch": 0.8373051867733158, + "grad_norm": 1.7710202932357788, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7637649774551392, + "num_tokens": 195467419.0, + "step": 10154 + }, + { + "epoch": 0.8373876473983672, + "grad_norm": 2.0075647830963135, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6988909840583801, + "num_tokens": 195498251.0, + "step": 10155 + }, + { + "epoch": 0.8374701080234188, + "grad_norm": 2.0880377292633057, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7279915809631348, + "num_tokens": 195524901.0, + "step": 10156 + }, + { + "epoch": 0.8375525686484704, + "grad_norm": 2.0755867958068848, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7358239889144897, + "num_tokens": 195552329.0, + "step": 10157 + }, + { + "epoch": 0.8376350292735218, + "grad_norm": 1.8571065664291382, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7568290829658508, + "num_tokens": 195584126.0, + "step": 10158 + }, + { + "epoch": 0.8377174898985734, + "grad_norm": 1.9156469106674194, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7308707237243652, + "num_tokens": 195614259.0, + "step": 10159 + }, + { + "epoch": 0.837799950523625, + "grad_norm": 1.8336377143859863, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7742865085601807, + "num_tokens": 195642672.0, + "step": 10160 + }, + { + "epoch": 0.8378824111486765, + "grad_norm": 2.0417580604553223, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.735831618309021, + "num_tokens": 195671061.0, + "step": 10161 + }, + { + "epoch": 0.837964871773728, + "grad_norm": 1.9893393516540527, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7308376431465149, + "num_tokens": 195701195.0, + "step": 10162 + }, + { + "epoch": 0.8380473323987796, + "grad_norm": 1.751739740371704, + "learning_rate": 1e-06, + "loss": 0.5903, + "mean_token_accuracy": 0.7985633015632629, + "num_tokens": 195731879.0, + "step": 10163 + }, + { + "epoch": 0.8381297930238312, + "grad_norm": 1.7316516637802124, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7425356507301331, + "num_tokens": 195765449.0, + "step": 10164 + }, + { + "epoch": 0.8382122536488826, + "grad_norm": 1.90834379196167, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7694828510284424, + "num_tokens": 195796079.0, + "step": 10165 + }, + { + "epoch": 0.8382947142739342, + "grad_norm": 1.938401699066162, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7317490577697754, + "num_tokens": 195830411.0, + "step": 10166 + }, + { + "epoch": 0.8383771748989858, + "grad_norm": 1.769209384918213, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.747634768486023, + "num_tokens": 195866092.0, + "step": 10167 + }, + { + "epoch": 0.8384596355240372, + "grad_norm": 2.0873847007751465, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7419706583023071, + "num_tokens": 195893251.0, + "step": 10168 + }, + { + "epoch": 0.8385420961490888, + "grad_norm": 1.9511131048202515, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7222892045974731, + "num_tokens": 195926215.0, + "step": 10169 + }, + { + "epoch": 0.8386245567741404, + "grad_norm": 1.8043864965438843, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7492555379867554, + "num_tokens": 195959210.0, + "step": 10170 + }, + { + "epoch": 0.8387070173991918, + "grad_norm": 1.9224733114242554, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.7758055329322815, + "num_tokens": 195988892.0, + "step": 10171 + }, + { + "epoch": 0.8387894780242434, + "grad_norm": 1.763458013534546, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7486831545829773, + "num_tokens": 196025105.0, + "step": 10172 + }, + { + "epoch": 0.838871938649295, + "grad_norm": 1.9737557172775269, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7282554507255554, + "num_tokens": 196053943.0, + "step": 10173 + }, + { + "epoch": 0.8389543992743465, + "grad_norm": 1.8256633281707764, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7268946170806885, + "num_tokens": 196090026.0, + "step": 10174 + }, + { + "epoch": 0.839036859899398, + "grad_norm": 2.1504948139190674, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7214168310165405, + "num_tokens": 196114786.0, + "step": 10175 + }, + { + "epoch": 0.8391193205244496, + "grad_norm": 2.144202470779419, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7715664505958557, + "num_tokens": 196140305.0, + "step": 10176 + }, + { + "epoch": 0.8392017811495011, + "grad_norm": 1.9571666717529297, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7338810563087463, + "num_tokens": 196171282.0, + "step": 10177 + }, + { + "epoch": 0.8392842417745526, + "grad_norm": 1.9044747352600098, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7183623909950256, + "num_tokens": 196203530.0, + "step": 10178 + }, + { + "epoch": 0.8393667023996042, + "grad_norm": 2.1475777626037598, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.704430341720581, + "num_tokens": 196230839.0, + "step": 10179 + }, + { + "epoch": 0.8394491630246558, + "grad_norm": 2.040454864501953, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.7854090332984924, + "num_tokens": 196259278.0, + "step": 10180 + }, + { + "epoch": 0.8395316236497072, + "grad_norm": 2.0520291328430176, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.733035683631897, + "num_tokens": 196288661.0, + "step": 10181 + }, + { + "epoch": 0.8396140842747588, + "grad_norm": 2.2969181537628174, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7312028408050537, + "num_tokens": 196310818.0, + "step": 10182 + }, + { + "epoch": 0.8396965448998104, + "grad_norm": 1.9225847721099854, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7713032960891724, + "num_tokens": 196339049.0, + "step": 10183 + }, + { + "epoch": 0.8397790055248618, + "grad_norm": 2.1774744987487793, + "learning_rate": 1e-06, + "loss": 0.6189, + "mean_token_accuracy": 0.7964844703674316, + "num_tokens": 196364266.0, + "step": 10184 + }, + { + "epoch": 0.8398614661499134, + "grad_norm": 1.9126490354537964, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7569247484207153, + "num_tokens": 196394955.0, + "step": 10185 + }, + { + "epoch": 0.839943926774965, + "grad_norm": 1.8562233448028564, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.759438693523407, + "num_tokens": 196426473.0, + "step": 10186 + }, + { + "epoch": 0.8400263874000164, + "grad_norm": 1.920649766921997, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7488269805908203, + "num_tokens": 196456949.0, + "step": 10187 + }, + { + "epoch": 0.840108848025068, + "grad_norm": 1.882954478263855, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.7505096197128296, + "num_tokens": 196485997.0, + "step": 10188 + }, + { + "epoch": 0.8401913086501196, + "grad_norm": 1.7332994937896729, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7650813460350037, + "num_tokens": 196520933.0, + "step": 10189 + }, + { + "epoch": 0.8402737692751711, + "grad_norm": 1.9101122617721558, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7796293497085571, + "num_tokens": 196554225.0, + "step": 10190 + }, + { + "epoch": 0.8403562299002226, + "grad_norm": 2.0612752437591553, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7599876523017883, + "num_tokens": 196580909.0, + "step": 10191 + }, + { + "epoch": 0.8404386905252742, + "grad_norm": 1.980150818824768, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.7799241542816162, + "num_tokens": 196609376.0, + "step": 10192 + }, + { + "epoch": 0.8405211511503258, + "grad_norm": 1.6808876991271973, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7397210001945496, + "num_tokens": 196646595.0, + "step": 10193 + }, + { + "epoch": 0.8406036117753772, + "grad_norm": 1.807989478111267, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7859387397766113, + "num_tokens": 196674769.0, + "step": 10194 + }, + { + "epoch": 0.8406860724004288, + "grad_norm": 1.8819048404693604, + "learning_rate": 1e-06, + "loss": 0.5737, + "mean_token_accuracy": 0.8144693374633789, + "num_tokens": 196701937.0, + "step": 10195 + }, + { + "epoch": 0.8407685330254804, + "grad_norm": 1.7231146097183228, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7742687463760376, + "num_tokens": 196737267.0, + "step": 10196 + }, + { + "epoch": 0.8408509936505318, + "grad_norm": 1.8170552253723145, + "learning_rate": 1e-06, + "loss": 0.6079, + "mean_token_accuracy": 0.8018937110900879, + "num_tokens": 196766276.0, + "step": 10197 + }, + { + "epoch": 0.8409334542755834, + "grad_norm": 2.250908374786377, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7336136102676392, + "num_tokens": 196792331.0, + "step": 10198 + }, + { + "epoch": 0.841015914900635, + "grad_norm": 1.7725862264633179, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7147924304008484, + "num_tokens": 196828045.0, + "step": 10199 + }, + { + "epoch": 0.8410983755256864, + "grad_norm": 1.9043924808502197, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7634016871452332, + "num_tokens": 196856651.0, + "step": 10200 + }, + { + "epoch": 0.841180836150738, + "grad_norm": 1.854259967803955, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7205572724342346, + "num_tokens": 196890179.0, + "step": 10201 + }, + { + "epoch": 0.8412632967757896, + "grad_norm": 2.168466329574585, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7326668500900269, + "num_tokens": 196917610.0, + "step": 10202 + }, + { + "epoch": 0.841345757400841, + "grad_norm": 1.9889264106750488, + "learning_rate": 1e-06, + "loss": 0.5987, + "mean_token_accuracy": 0.8031328320503235, + "num_tokens": 196944572.0, + "step": 10203 + }, + { + "epoch": 0.8414282180258926, + "grad_norm": 2.1265757083892822, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.753950834274292, + "num_tokens": 196969158.0, + "step": 10204 + }, + { + "epoch": 0.8415106786509442, + "grad_norm": 1.8561559915542603, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7710621356964111, + "num_tokens": 196998430.0, + "step": 10205 + }, + { + "epoch": 0.8415931392759957, + "grad_norm": 1.9914849996566772, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7085518836975098, + "num_tokens": 197026782.0, + "step": 10206 + }, + { + "epoch": 0.8416755999010472, + "grad_norm": 1.9204885959625244, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7552794218063354, + "num_tokens": 197055866.0, + "step": 10207 + }, + { + "epoch": 0.8417580605260988, + "grad_norm": 2.0295372009277344, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7218769192695618, + "num_tokens": 197083191.0, + "step": 10208 + }, + { + "epoch": 0.8418405211511504, + "grad_norm": 1.7483952045440674, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.7663367986679077, + "num_tokens": 197120161.0, + "step": 10209 + }, + { + "epoch": 0.8419229817762018, + "grad_norm": 1.794825553894043, + "learning_rate": 1e-06, + "loss": 0.6056, + "mean_token_accuracy": 0.8013280630111694, + "num_tokens": 197150518.0, + "step": 10210 + }, + { + "epoch": 0.8420054424012534, + "grad_norm": 2.2778687477111816, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7182848453521729, + "num_tokens": 197177082.0, + "step": 10211 + }, + { + "epoch": 0.842087903026305, + "grad_norm": 2.0356831550598145, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7108166217803955, + "num_tokens": 197210277.0, + "step": 10212 + }, + { + "epoch": 0.8421703636513564, + "grad_norm": 1.8782323598861694, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7685613632202148, + "num_tokens": 197241342.0, + "step": 10213 + }, + { + "epoch": 0.842252824276408, + "grad_norm": 1.9574923515319824, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7208336591720581, + "num_tokens": 197271338.0, + "step": 10214 + }, + { + "epoch": 0.8423352849014596, + "grad_norm": 2.161543369293213, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7712579965591431, + "num_tokens": 197292273.0, + "step": 10215 + }, + { + "epoch": 0.842417745526511, + "grad_norm": 1.7745623588562012, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.7918896675109863, + "num_tokens": 197324600.0, + "step": 10216 + }, + { + "epoch": 0.8425002061515626, + "grad_norm": 1.9969478845596313, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.758380651473999, + "num_tokens": 197351111.0, + "step": 10217 + }, + { + "epoch": 0.8425826667766142, + "grad_norm": 1.9227229356765747, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.78688645362854, + "num_tokens": 197381365.0, + "step": 10218 + }, + { + "epoch": 0.8426651274016657, + "grad_norm": 2.067666530609131, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7644456624984741, + "num_tokens": 197408176.0, + "step": 10219 + }, + { + "epoch": 0.8427475880267172, + "grad_norm": 1.8267662525177002, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7202221155166626, + "num_tokens": 197441600.0, + "step": 10220 + }, + { + "epoch": 0.8428300486517688, + "grad_norm": 1.7156226634979248, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7458911538124084, + "num_tokens": 197477243.0, + "step": 10221 + }, + { + "epoch": 0.8429125092768203, + "grad_norm": 1.750902533531189, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.758198618888855, + "num_tokens": 197513506.0, + "step": 10222 + }, + { + "epoch": 0.8429949699018718, + "grad_norm": 2.2178597450256348, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7202305793762207, + "num_tokens": 197538930.0, + "step": 10223 + }, + { + "epoch": 0.8430774305269234, + "grad_norm": 1.888181209564209, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7572389841079712, + "num_tokens": 197569070.0, + "step": 10224 + }, + { + "epoch": 0.843159891151975, + "grad_norm": 2.043698310852051, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7094792127609253, + "num_tokens": 197597955.0, + "step": 10225 + }, + { + "epoch": 0.8432423517770264, + "grad_norm": 1.8385770320892334, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7719001173973083, + "num_tokens": 197629582.0, + "step": 10226 + }, + { + "epoch": 0.843324812402078, + "grad_norm": 2.141993999481201, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7327604293823242, + "num_tokens": 197656189.0, + "step": 10227 + }, + { + "epoch": 0.8434072730271296, + "grad_norm": 2.046855926513672, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7560552358627319, + "num_tokens": 197681121.0, + "step": 10228 + }, + { + "epoch": 0.843489733652181, + "grad_norm": 1.896613359451294, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7514402270317078, + "num_tokens": 197713101.0, + "step": 10229 + }, + { + "epoch": 0.8435721942772326, + "grad_norm": 1.9867373704910278, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7731587290763855, + "num_tokens": 197740276.0, + "step": 10230 + }, + { + "epoch": 0.8436546549022842, + "grad_norm": 1.9592429399490356, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7057266235351562, + "num_tokens": 197770562.0, + "step": 10231 + }, + { + "epoch": 0.8437371155273357, + "grad_norm": 1.8919060230255127, + "learning_rate": 1e-06, + "loss": 0.6038, + "mean_token_accuracy": 0.7973705530166626, + "num_tokens": 197799568.0, + "step": 10232 + }, + { + "epoch": 0.8438195761523872, + "grad_norm": 2.2528226375579834, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7609396576881409, + "num_tokens": 197824451.0, + "step": 10233 + }, + { + "epoch": 0.8439020367774388, + "grad_norm": 1.954380989074707, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7665733098983765, + "num_tokens": 197854832.0, + "step": 10234 + }, + { + "epoch": 0.8439844974024903, + "grad_norm": 2.0011656284332275, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7187551856040955, + "num_tokens": 197883411.0, + "step": 10235 + }, + { + "epoch": 0.8440669580275418, + "grad_norm": 2.083632707595825, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.743992805480957, + "num_tokens": 197909684.0, + "step": 10236 + }, + { + "epoch": 0.8441494186525934, + "grad_norm": 1.981667160987854, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7321648597717285, + "num_tokens": 197940579.0, + "step": 10237 + }, + { + "epoch": 0.8442318792776449, + "grad_norm": 1.8053911924362183, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.7770321369171143, + "num_tokens": 197972331.0, + "step": 10238 + }, + { + "epoch": 0.8443143399026964, + "grad_norm": 1.647039532661438, + "learning_rate": 1e-06, + "loss": 0.575, + "mean_token_accuracy": 0.8089646100997925, + "num_tokens": 198008189.0, + "step": 10239 + }, + { + "epoch": 0.844396800527748, + "grad_norm": 2.1405279636383057, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7043887972831726, + "num_tokens": 198034100.0, + "step": 10240 + }, + { + "epoch": 0.8444792611527996, + "grad_norm": 1.7610970735549927, + "learning_rate": 1e-06, + "loss": 0.6195, + "mean_token_accuracy": 0.7985392212867737, + "num_tokens": 198066483.0, + "step": 10241 + }, + { + "epoch": 0.844561721777851, + "grad_norm": 1.9609094858169556, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7553378343582153, + "num_tokens": 198094706.0, + "step": 10242 + }, + { + "epoch": 0.8446441824029026, + "grad_norm": 1.7837637662887573, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7829422950744629, + "num_tokens": 198128372.0, + "step": 10243 + }, + { + "epoch": 0.8447266430279542, + "grad_norm": 1.8012592792510986, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7437649369239807, + "num_tokens": 198160719.0, + "step": 10244 + }, + { + "epoch": 0.8448091036530057, + "grad_norm": 1.7956736087799072, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7458688020706177, + "num_tokens": 198194591.0, + "step": 10245 + }, + { + "epoch": 0.8448915642780572, + "grad_norm": 2.1156184673309326, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7676854729652405, + "num_tokens": 198218267.0, + "step": 10246 + }, + { + "epoch": 0.8449740249031088, + "grad_norm": 2.220811128616333, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7552740573883057, + "num_tokens": 198242922.0, + "step": 10247 + }, + { + "epoch": 0.8450564855281603, + "grad_norm": 1.9447548389434814, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7117335796356201, + "num_tokens": 198274443.0, + "step": 10248 + }, + { + "epoch": 0.8451389461532118, + "grad_norm": 1.955580234527588, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7308650016784668, + "num_tokens": 198302825.0, + "step": 10249 + }, + { + "epoch": 0.8452214067782634, + "grad_norm": 1.9111803770065308, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7503923773765564, + "num_tokens": 198333754.0, + "step": 10250 + }, + { + "epoch": 0.8453038674033149, + "grad_norm": 2.0710670948028564, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7274566292762756, + "num_tokens": 198362578.0, + "step": 10251 + }, + { + "epoch": 0.8453863280283664, + "grad_norm": 2.0416882038116455, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7630501985549927, + "num_tokens": 198388423.0, + "step": 10252 + }, + { + "epoch": 0.845468788653418, + "grad_norm": 1.8780879974365234, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7441047430038452, + "num_tokens": 198420729.0, + "step": 10253 + }, + { + "epoch": 0.8455512492784696, + "grad_norm": 1.8963514566421509, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7844458818435669, + "num_tokens": 198450664.0, + "step": 10254 + }, + { + "epoch": 0.845633709903521, + "grad_norm": 2.0970382690429688, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7497764825820923, + "num_tokens": 198479029.0, + "step": 10255 + }, + { + "epoch": 0.8457161705285726, + "grad_norm": 1.7147386074066162, + "learning_rate": 1e-06, + "loss": 0.6051, + "mean_token_accuracy": 0.791496992111206, + "num_tokens": 198511124.0, + "step": 10256 + }, + { + "epoch": 0.8457986311536242, + "grad_norm": 1.9407591819763184, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.7869638800621033, + "num_tokens": 198540412.0, + "step": 10257 + }, + { + "epoch": 0.8458810917786757, + "grad_norm": 2.015803575515747, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.77287757396698, + "num_tokens": 198565540.0, + "step": 10258 + }, + { + "epoch": 0.8459635524037272, + "grad_norm": 1.9009674787521362, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7757423520088196, + "num_tokens": 198594590.0, + "step": 10259 + }, + { + "epoch": 0.8460460130287788, + "grad_norm": 1.8742735385894775, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7597676515579224, + "num_tokens": 198626995.0, + "step": 10260 + }, + { + "epoch": 0.8461284736538303, + "grad_norm": 1.7689133882522583, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7500057220458984, + "num_tokens": 198660185.0, + "step": 10261 + }, + { + "epoch": 0.8462109342788818, + "grad_norm": 1.9181755781173706, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7483211755752563, + "num_tokens": 198693351.0, + "step": 10262 + }, + { + "epoch": 0.8462933949039334, + "grad_norm": 1.928985595703125, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7810014486312866, + "num_tokens": 198724439.0, + "step": 10263 + }, + { + "epoch": 0.8463758555289849, + "grad_norm": 1.7708063125610352, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7574635744094849, + "num_tokens": 198756795.0, + "step": 10264 + }, + { + "epoch": 0.8464583161540364, + "grad_norm": 2.2949228286743164, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7161507606506348, + "num_tokens": 198782797.0, + "step": 10265 + }, + { + "epoch": 0.846540776779088, + "grad_norm": 1.9241328239440918, + "learning_rate": 1e-06, + "loss": 0.7926, + "mean_token_accuracy": 0.7621387839317322, + "num_tokens": 198813064.0, + "step": 10266 + }, + { + "epoch": 0.8466232374041395, + "grad_norm": 2.1760213375091553, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7700611352920532, + "num_tokens": 198836748.0, + "step": 10267 + }, + { + "epoch": 0.846705698029191, + "grad_norm": 1.9970481395721436, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7464337348937988, + "num_tokens": 198864950.0, + "step": 10268 + }, + { + "epoch": 0.8467881586542426, + "grad_norm": 1.8510974645614624, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7445123195648193, + "num_tokens": 198897131.0, + "step": 10269 + }, + { + "epoch": 0.8468706192792942, + "grad_norm": 1.7627451419830322, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7438612580299377, + "num_tokens": 198930694.0, + "step": 10270 + }, + { + "epoch": 0.8469530799043457, + "grad_norm": 1.7591986656188965, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7520077228546143, + "num_tokens": 198964826.0, + "step": 10271 + }, + { + "epoch": 0.8470355405293972, + "grad_norm": 1.965021014213562, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7392025589942932, + "num_tokens": 198994610.0, + "step": 10272 + }, + { + "epoch": 0.8471180011544488, + "grad_norm": 2.062577724456787, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7592971324920654, + "num_tokens": 199022571.0, + "step": 10273 + }, + { + "epoch": 0.8472004617795003, + "grad_norm": 2.290034055709839, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7406427264213562, + "num_tokens": 199045865.0, + "step": 10274 + }, + { + "epoch": 0.8472829224045518, + "grad_norm": 2.030022621154785, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7629813551902771, + "num_tokens": 199074233.0, + "step": 10275 + }, + { + "epoch": 0.8473653830296034, + "grad_norm": 2.081002950668335, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7505096793174744, + "num_tokens": 199098429.0, + "step": 10276 + }, + { + "epoch": 0.8474478436546549, + "grad_norm": 2.111375331878662, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7438749670982361, + "num_tokens": 199124585.0, + "step": 10277 + }, + { + "epoch": 0.8475303042797064, + "grad_norm": 1.8931804895401, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7463767528533936, + "num_tokens": 199159938.0, + "step": 10278 + }, + { + "epoch": 0.847612764904758, + "grad_norm": 1.9813909530639648, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.718356192111969, + "num_tokens": 199193286.0, + "step": 10279 + }, + { + "epoch": 0.8476952255298095, + "grad_norm": 1.8347975015640259, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7317435145378113, + "num_tokens": 199224651.0, + "step": 10280 + }, + { + "epoch": 0.847777686154861, + "grad_norm": 2.3023009300231934, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.735271692276001, + "num_tokens": 199248971.0, + "step": 10281 + }, + { + "epoch": 0.8478601467799126, + "grad_norm": 1.6664584875106812, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8003628253936768, + "num_tokens": 199283260.0, + "step": 10282 + }, + { + "epoch": 0.8479426074049641, + "grad_norm": 1.9178909063339233, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.762278139591217, + "num_tokens": 199314004.0, + "step": 10283 + }, + { + "epoch": 0.8480250680300156, + "grad_norm": 1.8047163486480713, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7696186304092407, + "num_tokens": 199343317.0, + "step": 10284 + }, + { + "epoch": 0.8481075286550672, + "grad_norm": 2.1553635597229004, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7361493110656738, + "num_tokens": 199368640.0, + "step": 10285 + }, + { + "epoch": 0.8481899892801188, + "grad_norm": 1.7538847923278809, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7256044149398804, + "num_tokens": 199406499.0, + "step": 10286 + }, + { + "epoch": 0.8482724499051703, + "grad_norm": 1.948610782623291, + "learning_rate": 1e-06, + "loss": 0.7767, + "mean_token_accuracy": 0.7468064427375793, + "num_tokens": 199435486.0, + "step": 10287 + }, + { + "epoch": 0.8483549105302218, + "grad_norm": 1.9582021236419678, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7664800882339478, + "num_tokens": 199465319.0, + "step": 10288 + }, + { + "epoch": 0.8484373711552734, + "grad_norm": 1.8490046262741089, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7787860631942749, + "num_tokens": 199499031.0, + "step": 10289 + }, + { + "epoch": 0.8485198317803249, + "grad_norm": 1.8643897771835327, + "learning_rate": 1e-06, + "loss": 0.5761, + "mean_token_accuracy": 0.8017073273658752, + "num_tokens": 199528860.0, + "step": 10290 + }, + { + "epoch": 0.8486022924053764, + "grad_norm": 2.051867961883545, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7099977731704712, + "num_tokens": 199556516.0, + "step": 10291 + }, + { + "epoch": 0.848684753030428, + "grad_norm": 1.9495738744735718, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7303443551063538, + "num_tokens": 199586266.0, + "step": 10292 + }, + { + "epoch": 0.8487672136554795, + "grad_norm": 1.9386582374572754, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7600470781326294, + "num_tokens": 199616751.0, + "step": 10293 + }, + { + "epoch": 0.848849674280531, + "grad_norm": 2.065190315246582, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7273369431495667, + "num_tokens": 199646188.0, + "step": 10294 + }, + { + "epoch": 0.8489321349055826, + "grad_norm": 1.773760199546814, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7610405683517456, + "num_tokens": 199678648.0, + "step": 10295 + }, + { + "epoch": 0.8490145955306341, + "grad_norm": 2.0067341327667236, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7623255252838135, + "num_tokens": 199706351.0, + "step": 10296 + }, + { + "epoch": 0.8490970561556856, + "grad_norm": 1.7388641834259033, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7622166872024536, + "num_tokens": 199739982.0, + "step": 10297 + }, + { + "epoch": 0.8491795167807372, + "grad_norm": 1.9396204948425293, + "learning_rate": 1e-06, + "loss": 0.6436, + "mean_token_accuracy": 0.7971686124801636, + "num_tokens": 199766763.0, + "step": 10298 + }, + { + "epoch": 0.8492619774057887, + "grad_norm": 2.010056495666504, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7302842736244202, + "num_tokens": 199793818.0, + "step": 10299 + }, + { + "epoch": 0.8493444380308403, + "grad_norm": 1.6891400814056396, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.7870649099349976, + "num_tokens": 199826218.0, + "step": 10300 + }, + { + "epoch": 0.8494268986558918, + "grad_norm": 1.9952906370162964, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7468510270118713, + "num_tokens": 199854701.0, + "step": 10301 + }, + { + "epoch": 0.8495093592809434, + "grad_norm": 1.7978876829147339, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7840790152549744, + "num_tokens": 199889243.0, + "step": 10302 + }, + { + "epoch": 0.8495918199059949, + "grad_norm": 1.8062092065811157, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7338833808898926, + "num_tokens": 199924840.0, + "step": 10303 + }, + { + "epoch": 0.8496742805310464, + "grad_norm": 1.8837785720825195, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.785457968711853, + "num_tokens": 199952488.0, + "step": 10304 + }, + { + "epoch": 0.849756741156098, + "grad_norm": 1.9579566717147827, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7366875410079956, + "num_tokens": 199981249.0, + "step": 10305 + }, + { + "epoch": 0.8498392017811495, + "grad_norm": 2.2991011142730713, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7234896421432495, + "num_tokens": 200005158.0, + "step": 10306 + }, + { + "epoch": 0.849921662406201, + "grad_norm": 2.051593065261841, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7422192096710205, + "num_tokens": 200032617.0, + "step": 10307 + }, + { + "epoch": 0.8500041230312526, + "grad_norm": 2.050292730331421, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7609173059463501, + "num_tokens": 200058148.0, + "step": 10308 + }, + { + "epoch": 0.8500865836563041, + "grad_norm": 1.772102952003479, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7818264365196228, + "num_tokens": 200090814.0, + "step": 10309 + }, + { + "epoch": 0.8501690442813556, + "grad_norm": 1.858298897743225, + "learning_rate": 1e-06, + "loss": 1.0604, + "mean_token_accuracy": 0.6779413819313049, + "num_tokens": 200126241.0, + "step": 10310 + }, + { + "epoch": 0.8502515049064072, + "grad_norm": 1.8876866102218628, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7355619668960571, + "num_tokens": 200157770.0, + "step": 10311 + }, + { + "epoch": 0.8503339655314587, + "grad_norm": 1.8277182579040527, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7482736706733704, + "num_tokens": 200191643.0, + "step": 10312 + }, + { + "epoch": 0.8504164261565103, + "grad_norm": 2.0128705501556396, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7724833488464355, + "num_tokens": 200217324.0, + "step": 10313 + }, + { + "epoch": 0.8504988867815618, + "grad_norm": 1.7762283086776733, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7734008431434631, + "num_tokens": 200251773.0, + "step": 10314 + }, + { + "epoch": 0.8505813474066134, + "grad_norm": 2.099133014678955, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7403678297996521, + "num_tokens": 200279088.0, + "step": 10315 + }, + { + "epoch": 0.8506638080316649, + "grad_norm": 2.0492959022521973, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7309820652008057, + "num_tokens": 200305615.0, + "step": 10316 + }, + { + "epoch": 0.8507462686567164, + "grad_norm": 1.8749902248382568, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7574345469474792, + "num_tokens": 200336539.0, + "step": 10317 + }, + { + "epoch": 0.850828729281768, + "grad_norm": 2.008687734603882, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7397469282150269, + "num_tokens": 200368039.0, + "step": 10318 + }, + { + "epoch": 0.8509111899068195, + "grad_norm": 1.8599677085876465, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7171841859817505, + "num_tokens": 200403438.0, + "step": 10319 + }, + { + "epoch": 0.850993650531871, + "grad_norm": 2.1772384643554688, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7717301249504089, + "num_tokens": 200429450.0, + "step": 10320 + }, + { + "epoch": 0.8510761111569226, + "grad_norm": 2.0046188831329346, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7635694742202759, + "num_tokens": 200458237.0, + "step": 10321 + }, + { + "epoch": 0.8511585717819741, + "grad_norm": 1.8889294862747192, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7584642171859741, + "num_tokens": 200486959.0, + "step": 10322 + }, + { + "epoch": 0.8512410324070256, + "grad_norm": 1.8711098432540894, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7742345333099365, + "num_tokens": 200516965.0, + "step": 10323 + }, + { + "epoch": 0.8513234930320772, + "grad_norm": 1.9605176448822021, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.726311981678009, + "num_tokens": 200546409.0, + "step": 10324 + }, + { + "epoch": 0.8514059536571287, + "grad_norm": 1.8928929567337036, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7071825265884399, + "num_tokens": 200578047.0, + "step": 10325 + }, + { + "epoch": 0.8514884142821803, + "grad_norm": 2.045081853866577, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7045501470565796, + "num_tokens": 200608062.0, + "step": 10326 + }, + { + "epoch": 0.8515708749072318, + "grad_norm": 1.7629454135894775, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7587355375289917, + "num_tokens": 200641796.0, + "step": 10327 + }, + { + "epoch": 0.8516533355322833, + "grad_norm": 1.8094241619110107, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7277107834815979, + "num_tokens": 200677613.0, + "step": 10328 + }, + { + "epoch": 0.8517357961573349, + "grad_norm": 1.8754479885101318, + "learning_rate": 1e-06, + "loss": 0.5993, + "mean_token_accuracy": 0.7957479953765869, + "num_tokens": 200703344.0, + "step": 10329 + }, + { + "epoch": 0.8518182567823864, + "grad_norm": 1.9773048162460327, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7555516958236694, + "num_tokens": 200732416.0, + "step": 10330 + }, + { + "epoch": 0.851900717407438, + "grad_norm": 1.7555489540100098, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7170190811157227, + "num_tokens": 200769814.0, + "step": 10331 + }, + { + "epoch": 0.8519831780324895, + "grad_norm": 2.211958408355713, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7119098901748657, + "num_tokens": 200793260.0, + "step": 10332 + }, + { + "epoch": 0.852065638657541, + "grad_norm": 1.7973153591156006, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7618889808654785, + "num_tokens": 200827184.0, + "step": 10333 + }, + { + "epoch": 0.8521480992825926, + "grad_norm": 2.238471508026123, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7417389750480652, + "num_tokens": 200852134.0, + "step": 10334 + }, + { + "epoch": 0.8522305599076441, + "grad_norm": 2.014169216156006, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7548953890800476, + "num_tokens": 200881263.0, + "step": 10335 + }, + { + "epoch": 0.8523130205326956, + "grad_norm": 1.773654818534851, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7324532270431519, + "num_tokens": 200916260.0, + "step": 10336 + }, + { + "epoch": 0.8523954811577472, + "grad_norm": 1.8149595260620117, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7536990642547607, + "num_tokens": 200949560.0, + "step": 10337 + }, + { + "epoch": 0.8524779417827987, + "grad_norm": 1.8020939826965332, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7452179193496704, + "num_tokens": 200986284.0, + "step": 10338 + }, + { + "epoch": 0.8525604024078502, + "grad_norm": 2.234285831451416, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7394648194313049, + "num_tokens": 201009711.0, + "step": 10339 + }, + { + "epoch": 0.8526428630329018, + "grad_norm": 2.009369134902954, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7818195819854736, + "num_tokens": 201037935.0, + "step": 10340 + }, + { + "epoch": 0.8527253236579533, + "grad_norm": 2.11590838432312, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7202405333518982, + "num_tokens": 201064013.0, + "step": 10341 + }, + { + "epoch": 0.8528077842830049, + "grad_norm": 1.910803198814392, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.773934543132782, + "num_tokens": 201091938.0, + "step": 10342 + }, + { + "epoch": 0.8528902449080564, + "grad_norm": 1.927456259727478, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7816309928894043, + "num_tokens": 201123282.0, + "step": 10343 + }, + { + "epoch": 0.8529727055331079, + "grad_norm": 2.1387453079223633, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.76997971534729, + "num_tokens": 201146225.0, + "step": 10344 + }, + { + "epoch": 0.8530551661581595, + "grad_norm": 1.8826284408569336, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.754900336265564, + "num_tokens": 201177959.0, + "step": 10345 + }, + { + "epoch": 0.853137626783211, + "grad_norm": 1.8288800716400146, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7086122632026672, + "num_tokens": 201211132.0, + "step": 10346 + }, + { + "epoch": 0.8532200874082626, + "grad_norm": 1.803097128868103, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7383374571800232, + "num_tokens": 201246670.0, + "step": 10347 + }, + { + "epoch": 0.8533025480333141, + "grad_norm": 1.9975894689559937, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7346154451370239, + "num_tokens": 201274370.0, + "step": 10348 + }, + { + "epoch": 0.8533850086583656, + "grad_norm": 1.7793108224868774, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7660836577415466, + "num_tokens": 201310564.0, + "step": 10349 + }, + { + "epoch": 0.8534674692834172, + "grad_norm": 1.8107255697250366, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7810757160186768, + "num_tokens": 201341418.0, + "step": 10350 + }, + { + "epoch": 0.8535499299084687, + "grad_norm": 1.8724130392074585, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7576079964637756, + "num_tokens": 201371228.0, + "step": 10351 + }, + { + "epoch": 0.8536323905335202, + "grad_norm": 2.0740084648132324, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7513457536697388, + "num_tokens": 201400533.0, + "step": 10352 + }, + { + "epoch": 0.8537148511585718, + "grad_norm": 2.148505926132202, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.732460618019104, + "num_tokens": 201425286.0, + "step": 10353 + }, + { + "epoch": 0.8537973117836233, + "grad_norm": 2.185011148452759, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7432155609130859, + "num_tokens": 201451281.0, + "step": 10354 + }, + { + "epoch": 0.8538797724086749, + "grad_norm": 1.9262691736221313, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7428069114685059, + "num_tokens": 201479938.0, + "step": 10355 + }, + { + "epoch": 0.8539622330337264, + "grad_norm": 1.771100401878357, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.7946345806121826, + "num_tokens": 201512695.0, + "step": 10356 + }, + { + "epoch": 0.8540446936587779, + "grad_norm": 1.7113909721374512, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7907313704490662, + "num_tokens": 201547122.0, + "step": 10357 + }, + { + "epoch": 0.8541271542838295, + "grad_norm": 1.8108946084976196, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7420160174369812, + "num_tokens": 201578681.0, + "step": 10358 + }, + { + "epoch": 0.854209614908881, + "grad_norm": 1.8757117986679077, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7533777952194214, + "num_tokens": 201607053.0, + "step": 10359 + }, + { + "epoch": 0.8542920755339325, + "grad_norm": 2.1535494327545166, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7612250447273254, + "num_tokens": 201634426.0, + "step": 10360 + }, + { + "epoch": 0.8543745361589841, + "grad_norm": 2.0222644805908203, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7657995223999023, + "num_tokens": 201661004.0, + "step": 10361 + }, + { + "epoch": 0.8544569967840356, + "grad_norm": 1.9979037046432495, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7714871168136597, + "num_tokens": 201688863.0, + "step": 10362 + }, + { + "epoch": 0.8545394574090872, + "grad_norm": 1.6951199769973755, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.7972185015678406, + "num_tokens": 201724994.0, + "step": 10363 + }, + { + "epoch": 0.8546219180341387, + "grad_norm": 1.7048553228378296, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.7890341281890869, + "num_tokens": 201757566.0, + "step": 10364 + }, + { + "epoch": 0.8547043786591902, + "grad_norm": 1.7532451152801514, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7660509347915649, + "num_tokens": 201790597.0, + "step": 10365 + }, + { + "epoch": 0.8547868392842418, + "grad_norm": 1.874399185180664, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.7858644127845764, + "num_tokens": 201819087.0, + "step": 10366 + }, + { + "epoch": 0.8548692999092933, + "grad_norm": 2.1141605377197266, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7566251754760742, + "num_tokens": 201844693.0, + "step": 10367 + }, + { + "epoch": 0.8549517605343449, + "grad_norm": 1.8448922634124756, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.74039626121521, + "num_tokens": 201875949.0, + "step": 10368 + }, + { + "epoch": 0.8550342211593964, + "grad_norm": 2.203115463256836, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7381234169006348, + "num_tokens": 201900750.0, + "step": 10369 + }, + { + "epoch": 0.8551166817844479, + "grad_norm": 1.9372947216033936, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.751888632774353, + "num_tokens": 201930824.0, + "step": 10370 + }, + { + "epoch": 0.8551991424094995, + "grad_norm": 1.901764988899231, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7265340089797974, + "num_tokens": 201964480.0, + "step": 10371 + }, + { + "epoch": 0.855281603034551, + "grad_norm": 2.1573355197906494, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7179145812988281, + "num_tokens": 201989502.0, + "step": 10372 + }, + { + "epoch": 0.8553640636596025, + "grad_norm": 1.9365328550338745, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7389777302742004, + "num_tokens": 202019929.0, + "step": 10373 + }, + { + "epoch": 0.8554465242846541, + "grad_norm": 1.8538811206817627, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7390352487564087, + "num_tokens": 202049639.0, + "step": 10374 + }, + { + "epoch": 0.8555289849097056, + "grad_norm": 1.9770238399505615, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7343714833259583, + "num_tokens": 202078945.0, + "step": 10375 + }, + { + "epoch": 0.8556114455347571, + "grad_norm": 2.0206334590911865, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7261266708374023, + "num_tokens": 202107405.0, + "step": 10376 + }, + { + "epoch": 0.8556939061598087, + "grad_norm": 1.7849195003509521, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7799729704856873, + "num_tokens": 202141703.0, + "step": 10377 + }, + { + "epoch": 0.8557763667848602, + "grad_norm": 1.8091853857040405, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.775452733039856, + "num_tokens": 202175191.0, + "step": 10378 + }, + { + "epoch": 0.8558588274099118, + "grad_norm": 2.0984442234039307, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7393772602081299, + "num_tokens": 202202758.0, + "step": 10379 + }, + { + "epoch": 0.8559412880349633, + "grad_norm": 2.460836887359619, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7324519753456116, + "num_tokens": 202224030.0, + "step": 10380 + }, + { + "epoch": 0.8560237486600148, + "grad_norm": 1.7531577348709106, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.7461116313934326, + "num_tokens": 202261268.0, + "step": 10381 + }, + { + "epoch": 0.8561062092850664, + "grad_norm": 1.8594621419906616, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7404686212539673, + "num_tokens": 202298625.0, + "step": 10382 + }, + { + "epoch": 0.8561886699101179, + "grad_norm": 2.0325541496276855, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7310307621955872, + "num_tokens": 202330293.0, + "step": 10383 + }, + { + "epoch": 0.8562711305351695, + "grad_norm": 1.785079836845398, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7449405193328857, + "num_tokens": 202364493.0, + "step": 10384 + }, + { + "epoch": 0.856353591160221, + "grad_norm": 2.206468343734741, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6886099576950073, + "num_tokens": 202390489.0, + "step": 10385 + }, + { + "epoch": 0.8564360517852725, + "grad_norm": 2.266045331954956, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7577453851699829, + "num_tokens": 202410481.0, + "step": 10386 + }, + { + "epoch": 0.8565185124103241, + "grad_norm": 2.1106832027435303, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.763332724571228, + "num_tokens": 202437732.0, + "step": 10387 + }, + { + "epoch": 0.8566009730353756, + "grad_norm": 1.9558017253875732, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7525122761726379, + "num_tokens": 202469942.0, + "step": 10388 + }, + { + "epoch": 0.8566834336604271, + "grad_norm": 1.8370312452316284, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7769364714622498, + "num_tokens": 202501351.0, + "step": 10389 + }, + { + "epoch": 0.8567658942854787, + "grad_norm": 1.9318093061447144, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.7894158959388733, + "num_tokens": 202530816.0, + "step": 10390 + }, + { + "epoch": 0.8568483549105302, + "grad_norm": 1.7728385925292969, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7288588285446167, + "num_tokens": 202567792.0, + "step": 10391 + }, + { + "epoch": 0.8569308155355818, + "grad_norm": 1.831595778465271, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7665849924087524, + "num_tokens": 202601571.0, + "step": 10392 + }, + { + "epoch": 0.8570132761606333, + "grad_norm": 2.186160087585449, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7534871697425842, + "num_tokens": 202626651.0, + "step": 10393 + }, + { + "epoch": 0.8570957367856848, + "grad_norm": 2.046705722808838, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7604562044143677, + "num_tokens": 202657076.0, + "step": 10394 + }, + { + "epoch": 0.8571781974107364, + "grad_norm": 1.8947197198867798, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7350317239761353, + "num_tokens": 202688453.0, + "step": 10395 + }, + { + "epoch": 0.8572606580357879, + "grad_norm": 2.3799333572387695, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7595270872116089, + "num_tokens": 202711851.0, + "step": 10396 + }, + { + "epoch": 0.8573431186608395, + "grad_norm": 2.034559965133667, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7580684423446655, + "num_tokens": 202738186.0, + "step": 10397 + }, + { + "epoch": 0.857425579285891, + "grad_norm": 1.9128385782241821, + "learning_rate": 1e-06, + "loss": 0.7642, + "mean_token_accuracy": 0.7601561546325684, + "num_tokens": 202767436.0, + "step": 10398 + }, + { + "epoch": 0.8575080399109425, + "grad_norm": 1.814538836479187, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.8019552230834961, + "num_tokens": 202799235.0, + "step": 10399 + }, + { + "epoch": 0.8575905005359941, + "grad_norm": 2.0102267265319824, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.7959561944007874, + "num_tokens": 202825299.0, + "step": 10400 + }, + { + "epoch": 0.8576729611610456, + "grad_norm": 1.7856857776641846, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7469332218170166, + "num_tokens": 202859959.0, + "step": 10401 + }, + { + "epoch": 0.8577554217860971, + "grad_norm": 1.9124939441680908, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7128899097442627, + "num_tokens": 202891370.0, + "step": 10402 + }, + { + "epoch": 0.8578378824111487, + "grad_norm": 2.164050817489624, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7205121517181396, + "num_tokens": 202917276.0, + "step": 10403 + }, + { + "epoch": 0.8579203430362002, + "grad_norm": 1.9285414218902588, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7538536787033081, + "num_tokens": 202948100.0, + "step": 10404 + }, + { + "epoch": 0.8580028036612517, + "grad_norm": 2.0579183101654053, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7284446358680725, + "num_tokens": 202975440.0, + "step": 10405 + }, + { + "epoch": 0.8580852642863033, + "grad_norm": 1.8217968940734863, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7695518732070923, + "num_tokens": 203008772.0, + "step": 10406 + }, + { + "epoch": 0.8581677249113548, + "grad_norm": 1.9697967767715454, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.7945055961608887, + "num_tokens": 203034517.0, + "step": 10407 + }, + { + "epoch": 0.8582501855364064, + "grad_norm": 1.815719485282898, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7430442571640015, + "num_tokens": 203069098.0, + "step": 10408 + }, + { + "epoch": 0.8583326461614579, + "grad_norm": 2.1054704189300537, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7244963645935059, + "num_tokens": 203096186.0, + "step": 10409 + }, + { + "epoch": 0.8584151067865095, + "grad_norm": 2.055373191833496, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7537646293640137, + "num_tokens": 203121270.0, + "step": 10410 + }, + { + "epoch": 0.858497567411561, + "grad_norm": 2.0201401710510254, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7017619013786316, + "num_tokens": 203150262.0, + "step": 10411 + }, + { + "epoch": 0.8585800280366125, + "grad_norm": 2.1437695026397705, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.727810800075531, + "num_tokens": 203176958.0, + "step": 10412 + }, + { + "epoch": 0.8586624886616641, + "grad_norm": 2.0433590412139893, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7759461402893066, + "num_tokens": 203204097.0, + "step": 10413 + }, + { + "epoch": 0.8587449492867156, + "grad_norm": 1.8582367897033691, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7483866810798645, + "num_tokens": 203237782.0, + "step": 10414 + }, + { + "epoch": 0.8588274099117671, + "grad_norm": 2.015303611755371, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7648472785949707, + "num_tokens": 203263032.0, + "step": 10415 + }, + { + "epoch": 0.8589098705368187, + "grad_norm": 1.6849792003631592, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7777082920074463, + "num_tokens": 203299380.0, + "step": 10416 + }, + { + "epoch": 0.8589923311618702, + "grad_norm": 2.0759944915771484, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7398930191993713, + "num_tokens": 203323792.0, + "step": 10417 + }, + { + "epoch": 0.8590747917869217, + "grad_norm": 1.877301812171936, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7585576772689819, + "num_tokens": 203353686.0, + "step": 10418 + }, + { + "epoch": 0.8591572524119733, + "grad_norm": 2.1085097789764404, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7461452484130859, + "num_tokens": 203379352.0, + "step": 10419 + }, + { + "epoch": 0.8592397130370248, + "grad_norm": 1.7750407457351685, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7841591835021973, + "num_tokens": 203410042.0, + "step": 10420 + }, + { + "epoch": 0.8593221736620763, + "grad_norm": 2.20167875289917, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7408944368362427, + "num_tokens": 203434979.0, + "step": 10421 + }, + { + "epoch": 0.8594046342871279, + "grad_norm": 1.7564505338668823, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7541725635528564, + "num_tokens": 203469758.0, + "step": 10422 + }, + { + "epoch": 0.8594870949121795, + "grad_norm": 1.9794502258300781, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7429376840591431, + "num_tokens": 203500303.0, + "step": 10423 + }, + { + "epoch": 0.859569555537231, + "grad_norm": 2.396049976348877, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.726641058921814, + "num_tokens": 203521481.0, + "step": 10424 + }, + { + "epoch": 0.8596520161622825, + "grad_norm": 1.9408468008041382, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7296596169471741, + "num_tokens": 203551304.0, + "step": 10425 + }, + { + "epoch": 0.8597344767873341, + "grad_norm": 1.920552134513855, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7376145124435425, + "num_tokens": 203583479.0, + "step": 10426 + }, + { + "epoch": 0.8598169374123856, + "grad_norm": 1.8057036399841309, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.76426100730896, + "num_tokens": 203615982.0, + "step": 10427 + }, + { + "epoch": 0.8598993980374371, + "grad_norm": 1.8838987350463867, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7529142498970032, + "num_tokens": 203644248.0, + "step": 10428 + }, + { + "epoch": 0.8599818586624887, + "grad_norm": 1.8053799867630005, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.8034812808036804, + "num_tokens": 203674464.0, + "step": 10429 + }, + { + "epoch": 0.8600643192875402, + "grad_norm": 2.1321682929992676, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7191901206970215, + "num_tokens": 203700471.0, + "step": 10430 + }, + { + "epoch": 0.8601467799125917, + "grad_norm": 1.994422197341919, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7062454223632812, + "num_tokens": 203731461.0, + "step": 10431 + }, + { + "epoch": 0.8602292405376433, + "grad_norm": 1.863913893699646, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7748270630836487, + "num_tokens": 203761242.0, + "step": 10432 + }, + { + "epoch": 0.8603117011626948, + "grad_norm": 1.767627477645874, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7761520147323608, + "num_tokens": 203793546.0, + "step": 10433 + }, + { + "epoch": 0.8603941617877463, + "grad_norm": 2.158766031265259, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.725866973400116, + "num_tokens": 203819304.0, + "step": 10434 + }, + { + "epoch": 0.8604766224127979, + "grad_norm": 1.8209820985794067, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7609109878540039, + "num_tokens": 203852745.0, + "step": 10435 + }, + { + "epoch": 0.8605590830378494, + "grad_norm": 1.9695006608963013, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7886789441108704, + "num_tokens": 203881442.0, + "step": 10436 + }, + { + "epoch": 0.8606415436629009, + "grad_norm": 2.1138713359832764, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7655459642410278, + "num_tokens": 203905516.0, + "step": 10437 + }, + { + "epoch": 0.8607240042879525, + "grad_norm": 2.0160982608795166, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.75294029712677, + "num_tokens": 203936884.0, + "step": 10438 + }, + { + "epoch": 0.860806464913004, + "grad_norm": 1.7604602575302124, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7800787091255188, + "num_tokens": 203967337.0, + "step": 10439 + }, + { + "epoch": 0.8608889255380556, + "grad_norm": 1.9136286973953247, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7661378383636475, + "num_tokens": 204000901.0, + "step": 10440 + }, + { + "epoch": 0.8609713861631071, + "grad_norm": 2.0381736755371094, + "learning_rate": 1e-06, + "loss": 0.7723, + "mean_token_accuracy": 0.7509660720825195, + "num_tokens": 204027303.0, + "step": 10441 + }, + { + "epoch": 0.8610538467881587, + "grad_norm": 1.8198180198669434, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7260586619377136, + "num_tokens": 204062188.0, + "step": 10442 + }, + { + "epoch": 0.8611363074132102, + "grad_norm": 1.8859394788742065, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7690661549568176, + "num_tokens": 204089843.0, + "step": 10443 + }, + { + "epoch": 0.8612187680382617, + "grad_norm": 2.036414623260498, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.763572633266449, + "num_tokens": 204117552.0, + "step": 10444 + }, + { + "epoch": 0.8613012286633133, + "grad_norm": 2.109145402908325, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7558577060699463, + "num_tokens": 204141653.0, + "step": 10445 + }, + { + "epoch": 0.8613836892883648, + "grad_norm": 1.8144478797912598, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7621544599533081, + "num_tokens": 204176717.0, + "step": 10446 + }, + { + "epoch": 0.8614661499134163, + "grad_norm": 2.038942575454712, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7344554662704468, + "num_tokens": 204205279.0, + "step": 10447 + }, + { + "epoch": 0.8615486105384679, + "grad_norm": 2.087069272994995, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7665309906005859, + "num_tokens": 204232709.0, + "step": 10448 + }, + { + "epoch": 0.8616310711635194, + "grad_norm": 1.9762569665908813, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7715980410575867, + "num_tokens": 204261413.0, + "step": 10449 + }, + { + "epoch": 0.8617135317885709, + "grad_norm": 1.7944039106369019, + "learning_rate": 1e-06, + "loss": 0.7617, + "mean_token_accuracy": 0.7608190774917603, + "num_tokens": 204294849.0, + "step": 10450 + }, + { + "epoch": 0.8617959924136225, + "grad_norm": 1.8572651147842407, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7430969476699829, + "num_tokens": 204328323.0, + "step": 10451 + }, + { + "epoch": 0.861878453038674, + "grad_norm": 2.349618911743164, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7160747051239014, + "num_tokens": 204350829.0, + "step": 10452 + }, + { + "epoch": 0.8619609136637256, + "grad_norm": 2.013681411743164, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7057459354400635, + "num_tokens": 204380192.0, + "step": 10453 + }, + { + "epoch": 0.8620433742887771, + "grad_norm": 1.7865846157073975, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7577643394470215, + "num_tokens": 204411959.0, + "step": 10454 + }, + { + "epoch": 0.8621258349138287, + "grad_norm": 2.08121919631958, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7461453676223755, + "num_tokens": 204440800.0, + "step": 10455 + }, + { + "epoch": 0.8622082955388802, + "grad_norm": 1.9729706048965454, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7436760663986206, + "num_tokens": 204469886.0, + "step": 10456 + }, + { + "epoch": 0.8622907561639317, + "grad_norm": 2.090463161468506, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7496570348739624, + "num_tokens": 204500183.0, + "step": 10457 + }, + { + "epoch": 0.8623732167889833, + "grad_norm": 1.9267966747283936, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7403209805488586, + "num_tokens": 204531310.0, + "step": 10458 + }, + { + "epoch": 0.8624556774140348, + "grad_norm": 2.022210121154785, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7315704822540283, + "num_tokens": 204561064.0, + "step": 10459 + }, + { + "epoch": 0.8625381380390863, + "grad_norm": 1.9242368936538696, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7822234630584717, + "num_tokens": 204587752.0, + "step": 10460 + }, + { + "epoch": 0.8626205986641379, + "grad_norm": 1.889824628829956, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.735130250453949, + "num_tokens": 204623699.0, + "step": 10461 + }, + { + "epoch": 0.8627030592891894, + "grad_norm": 1.8138997554779053, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7295002341270447, + "num_tokens": 204657485.0, + "step": 10462 + }, + { + "epoch": 0.8627855199142409, + "grad_norm": 1.9113637208938599, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.765293300151825, + "num_tokens": 204687899.0, + "step": 10463 + }, + { + "epoch": 0.8628679805392925, + "grad_norm": 2.017235279083252, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7708935737609863, + "num_tokens": 204715665.0, + "step": 10464 + }, + { + "epoch": 0.862950441164344, + "grad_norm": 1.8422034978866577, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7746418714523315, + "num_tokens": 204746289.0, + "step": 10465 + }, + { + "epoch": 0.8630329017893955, + "grad_norm": 2.0575644969940186, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7261942625045776, + "num_tokens": 204772989.0, + "step": 10466 + }, + { + "epoch": 0.8631153624144471, + "grad_norm": 2.071549415588379, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7564666271209717, + "num_tokens": 204799705.0, + "step": 10467 + }, + { + "epoch": 0.8631978230394987, + "grad_norm": 2.016995668411255, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7509657144546509, + "num_tokens": 204826096.0, + "step": 10468 + }, + { + "epoch": 0.8632802836645502, + "grad_norm": 2.15401291847229, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.726294219493866, + "num_tokens": 204852550.0, + "step": 10469 + }, + { + "epoch": 0.8633627442896017, + "grad_norm": 1.947890043258667, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7464056015014648, + "num_tokens": 204883355.0, + "step": 10470 + }, + { + "epoch": 0.8634452049146533, + "grad_norm": 2.003300189971924, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7122995853424072, + "num_tokens": 204914266.0, + "step": 10471 + }, + { + "epoch": 0.8635276655397048, + "grad_norm": 2.141245126724243, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7624322175979614, + "num_tokens": 204939371.0, + "step": 10472 + }, + { + "epoch": 0.8636101261647563, + "grad_norm": 2.067603588104248, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7221089601516724, + "num_tokens": 204968051.0, + "step": 10473 + }, + { + "epoch": 0.8636925867898079, + "grad_norm": 1.9289295673370361, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7429324388504028, + "num_tokens": 204998943.0, + "step": 10474 + }, + { + "epoch": 0.8637750474148594, + "grad_norm": 2.075115442276001, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7361049652099609, + "num_tokens": 205025426.0, + "step": 10475 + }, + { + "epoch": 0.8638575080399109, + "grad_norm": 1.685671329498291, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7719871997833252, + "num_tokens": 205062992.0, + "step": 10476 + }, + { + "epoch": 0.8639399686649625, + "grad_norm": 1.971401333808899, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7383126020431519, + "num_tokens": 205090400.0, + "step": 10477 + }, + { + "epoch": 0.864022429290014, + "grad_norm": 1.9853737354278564, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.736762285232544, + "num_tokens": 205116155.0, + "step": 10478 + }, + { + "epoch": 0.8641048899150655, + "grad_norm": 2.111449956893921, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.753435492515564, + "num_tokens": 205141925.0, + "step": 10479 + }, + { + "epoch": 0.8641873505401171, + "grad_norm": 2.0069687366485596, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7566965818405151, + "num_tokens": 205170490.0, + "step": 10480 + }, + { + "epoch": 0.8642698111651687, + "grad_norm": 2.0154178142547607, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7598915100097656, + "num_tokens": 205200851.0, + "step": 10481 + }, + { + "epoch": 0.8643522717902201, + "grad_norm": 1.7487164735794067, + "learning_rate": 1e-06, + "loss": 0.6192, + "mean_token_accuracy": 0.7966781854629517, + "num_tokens": 205234424.0, + "step": 10482 + }, + { + "epoch": 0.8644347324152717, + "grad_norm": 2.1288328170776367, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7574760317802429, + "num_tokens": 205258658.0, + "step": 10483 + }, + { + "epoch": 0.8645171930403233, + "grad_norm": 1.8906792402267456, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7730327844619751, + "num_tokens": 205289050.0, + "step": 10484 + }, + { + "epoch": 0.8645996536653748, + "grad_norm": 1.926275610923767, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7192087769508362, + "num_tokens": 205321666.0, + "step": 10485 + }, + { + "epoch": 0.8646821142904263, + "grad_norm": 2.2442195415496826, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7240615487098694, + "num_tokens": 205346262.0, + "step": 10486 + }, + { + "epoch": 0.8647645749154779, + "grad_norm": 1.8495550155639648, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7444362044334412, + "num_tokens": 205381380.0, + "step": 10487 + }, + { + "epoch": 0.8648470355405294, + "grad_norm": 1.7898632287979126, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.7959696054458618, + "num_tokens": 205414170.0, + "step": 10488 + }, + { + "epoch": 0.8649294961655809, + "grad_norm": 2.0203094482421875, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7606521844863892, + "num_tokens": 205442304.0, + "step": 10489 + }, + { + "epoch": 0.8650119567906325, + "grad_norm": 2.1258938312530518, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7483460307121277, + "num_tokens": 205467680.0, + "step": 10490 + }, + { + "epoch": 0.865094417415684, + "grad_norm": 2.0835158824920654, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7185157537460327, + "num_tokens": 205493902.0, + "step": 10491 + }, + { + "epoch": 0.8651768780407355, + "grad_norm": 1.9606237411499023, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7280359268188477, + "num_tokens": 205523750.0, + "step": 10492 + }, + { + "epoch": 0.8652593386657871, + "grad_norm": 1.8895063400268555, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7070571184158325, + "num_tokens": 205557804.0, + "step": 10493 + }, + { + "epoch": 0.8653417992908387, + "grad_norm": 1.8243801593780518, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7646975517272949, + "num_tokens": 205590155.0, + "step": 10494 + }, + { + "epoch": 0.8654242599158901, + "grad_norm": 1.779916763305664, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7870159149169922, + "num_tokens": 205625323.0, + "step": 10495 + }, + { + "epoch": 0.8655067205409417, + "grad_norm": 1.9083234071731567, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7676767110824585, + "num_tokens": 205656323.0, + "step": 10496 + }, + { + "epoch": 0.8655891811659933, + "grad_norm": 1.8533942699432373, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7309473752975464, + "num_tokens": 205689892.0, + "step": 10497 + }, + { + "epoch": 0.8656716417910447, + "grad_norm": 1.9282654523849487, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.732180655002594, + "num_tokens": 205721011.0, + "step": 10498 + }, + { + "epoch": 0.8657541024160963, + "grad_norm": 1.5485559701919556, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7799985408782959, + "num_tokens": 205760165.0, + "step": 10499 + }, + { + "epoch": 0.8658365630411479, + "grad_norm": 1.8535939455032349, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7434849143028259, + "num_tokens": 205791277.0, + "step": 10500 + }, + { + "epoch": 0.8659190236661994, + "grad_norm": 1.9055672883987427, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7839854955673218, + "num_tokens": 205819802.0, + "step": 10501 + }, + { + "epoch": 0.8660014842912509, + "grad_norm": 2.0414254665374756, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7476434707641602, + "num_tokens": 205848495.0, + "step": 10502 + }, + { + "epoch": 0.8660839449163025, + "grad_norm": 1.6255682706832886, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7603442072868347, + "num_tokens": 205886833.0, + "step": 10503 + }, + { + "epoch": 0.866166405541354, + "grad_norm": 1.8959230184555054, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.6980328559875488, + "num_tokens": 205919613.0, + "step": 10504 + }, + { + "epoch": 0.8662488661664055, + "grad_norm": 2.1355698108673096, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7190412282943726, + "num_tokens": 205947072.0, + "step": 10505 + }, + { + "epoch": 0.8663313267914571, + "grad_norm": 1.7596409320831299, + "learning_rate": 1e-06, + "loss": 0.6027, + "mean_token_accuracy": 0.798384428024292, + "num_tokens": 205984341.0, + "step": 10506 + }, + { + "epoch": 0.8664137874165087, + "grad_norm": 1.8110613822937012, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7386918067932129, + "num_tokens": 206018429.0, + "step": 10507 + }, + { + "epoch": 0.8664962480415601, + "grad_norm": 1.7600057125091553, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7214257717132568, + "num_tokens": 206053890.0, + "step": 10508 + }, + { + "epoch": 0.8665787086666117, + "grad_norm": 1.8635339736938477, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7715511322021484, + "num_tokens": 206086429.0, + "step": 10509 + }, + { + "epoch": 0.8666611692916633, + "grad_norm": 2.1248035430908203, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7749769687652588, + "num_tokens": 206109581.0, + "step": 10510 + }, + { + "epoch": 0.8667436299167147, + "grad_norm": 1.93972909450531, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7714591026306152, + "num_tokens": 206136708.0, + "step": 10511 + }, + { + "epoch": 0.8668260905417663, + "grad_norm": 1.9228771924972534, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7716864347457886, + "num_tokens": 206165971.0, + "step": 10512 + }, + { + "epoch": 0.8669085511668179, + "grad_norm": 2.036377191543579, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7151668071746826, + "num_tokens": 206195299.0, + "step": 10513 + }, + { + "epoch": 0.8669910117918694, + "grad_norm": 2.0693435668945312, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7242863178253174, + "num_tokens": 206226151.0, + "step": 10514 + }, + { + "epoch": 0.8670734724169209, + "grad_norm": 1.8446332216262817, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7507901787757874, + "num_tokens": 206256379.0, + "step": 10515 + }, + { + "epoch": 0.8671559330419725, + "grad_norm": 1.9622299671173096, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7545773983001709, + "num_tokens": 206286129.0, + "step": 10516 + }, + { + "epoch": 0.867238393667024, + "grad_norm": 1.9370547533035278, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7448352575302124, + "num_tokens": 206315039.0, + "step": 10517 + }, + { + "epoch": 0.8673208542920755, + "grad_norm": 1.925313115119934, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.759700357913971, + "num_tokens": 206342845.0, + "step": 10518 + }, + { + "epoch": 0.8674033149171271, + "grad_norm": 2.012519121170044, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7569754719734192, + "num_tokens": 206370544.0, + "step": 10519 + }, + { + "epoch": 0.8674857755421786, + "grad_norm": 1.8123281002044678, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.761022686958313, + "num_tokens": 206404182.0, + "step": 10520 + }, + { + "epoch": 0.8675682361672301, + "grad_norm": 2.0903077125549316, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7412570714950562, + "num_tokens": 206430096.0, + "step": 10521 + }, + { + "epoch": 0.8676506967922817, + "grad_norm": 1.8925282955169678, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7576336860656738, + "num_tokens": 206459043.0, + "step": 10522 + }, + { + "epoch": 0.8677331574173333, + "grad_norm": 1.8753280639648438, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7534952759742737, + "num_tokens": 206490208.0, + "step": 10523 + }, + { + "epoch": 0.8678156180423847, + "grad_norm": 1.8590110540390015, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7623310089111328, + "num_tokens": 206520776.0, + "step": 10524 + }, + { + "epoch": 0.8678980786674363, + "grad_norm": 1.8561149835586548, + "learning_rate": 1e-06, + "loss": 0.7808, + "mean_token_accuracy": 0.7504727244377136, + "num_tokens": 206551720.0, + "step": 10525 + }, + { + "epoch": 0.8679805392924879, + "grad_norm": 1.8896026611328125, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7708489298820496, + "num_tokens": 206582950.0, + "step": 10526 + }, + { + "epoch": 0.8680629999175393, + "grad_norm": 2.0870909690856934, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7613719701766968, + "num_tokens": 206608209.0, + "step": 10527 + }, + { + "epoch": 0.8681454605425909, + "grad_norm": 1.7874507904052734, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7423664331436157, + "num_tokens": 206642103.0, + "step": 10528 + }, + { + "epoch": 0.8682279211676425, + "grad_norm": 1.879988670349121, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7474766969680786, + "num_tokens": 206674415.0, + "step": 10529 + }, + { + "epoch": 0.868310381792694, + "grad_norm": 1.5691496133804321, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7706495523452759, + "num_tokens": 206717596.0, + "step": 10530 + }, + { + "epoch": 0.8683928424177455, + "grad_norm": 1.7689756155014038, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7714453935623169, + "num_tokens": 206748410.0, + "step": 10531 + }, + { + "epoch": 0.8684753030427971, + "grad_norm": 1.860636830329895, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7367692589759827, + "num_tokens": 206780095.0, + "step": 10532 + }, + { + "epoch": 0.8685577636678486, + "grad_norm": 1.8349188566207886, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7224903106689453, + "num_tokens": 206813941.0, + "step": 10533 + }, + { + "epoch": 0.8686402242929001, + "grad_norm": 1.8861982822418213, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7541080713272095, + "num_tokens": 206842868.0, + "step": 10534 + }, + { + "epoch": 0.8687226849179517, + "grad_norm": 1.9297754764556885, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.7643204927444458, + "num_tokens": 206873860.0, + "step": 10535 + }, + { + "epoch": 0.8688051455430033, + "grad_norm": 2.0793650150299072, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7405773997306824, + "num_tokens": 206902708.0, + "step": 10536 + }, + { + "epoch": 0.8688876061680547, + "grad_norm": 1.7916960716247559, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7755091190338135, + "num_tokens": 206937593.0, + "step": 10537 + }, + { + "epoch": 0.8689700667931063, + "grad_norm": 1.9767273664474487, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7520098090171814, + "num_tokens": 206966333.0, + "step": 10538 + }, + { + "epoch": 0.8690525274181579, + "grad_norm": 1.9480699300765991, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7581453323364258, + "num_tokens": 206994190.0, + "step": 10539 + }, + { + "epoch": 0.8691349880432093, + "grad_norm": 1.9651075601577759, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7819329500198364, + "num_tokens": 207021173.0, + "step": 10540 + }, + { + "epoch": 0.8692174486682609, + "grad_norm": 1.9878898859024048, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7586624622344971, + "num_tokens": 207053738.0, + "step": 10541 + }, + { + "epoch": 0.8692999092933125, + "grad_norm": 2.234602451324463, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7724817991256714, + "num_tokens": 207076211.0, + "step": 10542 + }, + { + "epoch": 0.8693823699183639, + "grad_norm": 2.107370376586914, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7729170322418213, + "num_tokens": 207100077.0, + "step": 10543 + }, + { + "epoch": 0.8694648305434155, + "grad_norm": 1.7852579355239868, + "learning_rate": 1e-06, + "loss": 0.7543, + "mean_token_accuracy": 0.7558958530426025, + "num_tokens": 207134544.0, + "step": 10544 + }, + { + "epoch": 0.8695472911684671, + "grad_norm": 2.0074803829193115, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7553409337997437, + "num_tokens": 207164389.0, + "step": 10545 + }, + { + "epoch": 0.8696297517935186, + "grad_norm": 2.2688395977020264, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7208974361419678, + "num_tokens": 207187330.0, + "step": 10546 + }, + { + "epoch": 0.8697122124185701, + "grad_norm": 2.045523166656494, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.721474289894104, + "num_tokens": 207215788.0, + "step": 10547 + }, + { + "epoch": 0.8697946730436217, + "grad_norm": 2.1298696994781494, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7634097337722778, + "num_tokens": 207239351.0, + "step": 10548 + }, + { + "epoch": 0.8698771336686733, + "grad_norm": 1.9746181964874268, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7528785467147827, + "num_tokens": 207269036.0, + "step": 10549 + }, + { + "epoch": 0.8699595942937247, + "grad_norm": 2.2334840297698975, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7651886343955994, + "num_tokens": 207303054.0, + "step": 10550 + }, + { + "epoch": 0.8700420549187763, + "grad_norm": 1.9453309774398804, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7210961580276489, + "num_tokens": 207332690.0, + "step": 10551 + }, + { + "epoch": 0.8701245155438279, + "grad_norm": 2.0088438987731934, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7824053764343262, + "num_tokens": 207359303.0, + "step": 10552 + }, + { + "epoch": 0.8702069761688793, + "grad_norm": 1.8502095937728882, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7362481951713562, + "num_tokens": 207391041.0, + "step": 10553 + }, + { + "epoch": 0.8702894367939309, + "grad_norm": 1.7453705072402954, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7684922218322754, + "num_tokens": 207428005.0, + "step": 10554 + }, + { + "epoch": 0.8703718974189825, + "grad_norm": 1.985129714012146, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7464699149131775, + "num_tokens": 207462333.0, + "step": 10555 + }, + { + "epoch": 0.8704543580440339, + "grad_norm": 1.7284244298934937, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.7843700647354126, + "num_tokens": 207495193.0, + "step": 10556 + }, + { + "epoch": 0.8705368186690855, + "grad_norm": 2.0034255981445312, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7257276177406311, + "num_tokens": 207526242.0, + "step": 10557 + }, + { + "epoch": 0.8706192792941371, + "grad_norm": 1.9419288635253906, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.773246705532074, + "num_tokens": 207556276.0, + "step": 10558 + }, + { + "epoch": 0.8707017399191885, + "grad_norm": 2.2079169750213623, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7212857007980347, + "num_tokens": 207585842.0, + "step": 10559 + }, + { + "epoch": 0.8707842005442401, + "grad_norm": 1.835051417350769, + "learning_rate": 1e-06, + "loss": 0.5512, + "mean_token_accuracy": 0.8112647533416748, + "num_tokens": 207616780.0, + "step": 10560 + }, + { + "epoch": 0.8708666611692917, + "grad_norm": 1.6254589557647705, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7874525785446167, + "num_tokens": 207653825.0, + "step": 10561 + }, + { + "epoch": 0.8709491217943433, + "grad_norm": 2.320916175842285, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7418748140335083, + "num_tokens": 207677586.0, + "step": 10562 + }, + { + "epoch": 0.8710315824193947, + "grad_norm": 1.8663866519927979, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7597360610961914, + "num_tokens": 207708640.0, + "step": 10563 + }, + { + "epoch": 0.8711140430444463, + "grad_norm": 1.9801528453826904, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7286019325256348, + "num_tokens": 207737283.0, + "step": 10564 + }, + { + "epoch": 0.8711965036694979, + "grad_norm": 1.709240198135376, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7905502915382385, + "num_tokens": 207775101.0, + "step": 10565 + }, + { + "epoch": 0.8712789642945493, + "grad_norm": 2.0212740898132324, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7366072535514832, + "num_tokens": 207805096.0, + "step": 10566 + }, + { + "epoch": 0.8713614249196009, + "grad_norm": 1.9279848337173462, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7746469974517822, + "num_tokens": 207834905.0, + "step": 10567 + }, + { + "epoch": 0.8714438855446525, + "grad_norm": 2.1149115562438965, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7660570740699768, + "num_tokens": 207863509.0, + "step": 10568 + }, + { + "epoch": 0.8715263461697039, + "grad_norm": 1.854864239692688, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7698245048522949, + "num_tokens": 207898638.0, + "step": 10569 + }, + { + "epoch": 0.8716088067947555, + "grad_norm": 1.9912282228469849, + "learning_rate": 1e-06, + "loss": 0.5099, + "mean_token_accuracy": 0.8281779885292053, + "num_tokens": 207924384.0, + "step": 10570 + }, + { + "epoch": 0.8716912674198071, + "grad_norm": 2.0866267681121826, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7439062595367432, + "num_tokens": 207950834.0, + "step": 10571 + }, + { + "epoch": 0.8717737280448585, + "grad_norm": 2.016300678253174, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7485790252685547, + "num_tokens": 207982534.0, + "step": 10572 + }, + { + "epoch": 0.8718561886699101, + "grad_norm": 1.8056203126907349, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7314208745956421, + "num_tokens": 208018184.0, + "step": 10573 + }, + { + "epoch": 0.8719386492949617, + "grad_norm": 1.980725646018982, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7486560940742493, + "num_tokens": 208047454.0, + "step": 10574 + }, + { + "epoch": 0.8720211099200132, + "grad_norm": 1.9985119104385376, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7094964981079102, + "num_tokens": 208078376.0, + "step": 10575 + }, + { + "epoch": 0.8721035705450647, + "grad_norm": 1.7181264162063599, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7745639681816101, + "num_tokens": 208113021.0, + "step": 10576 + }, + { + "epoch": 0.8721860311701163, + "grad_norm": 2.187114953994751, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7346205711364746, + "num_tokens": 208139400.0, + "step": 10577 + }, + { + "epoch": 0.8722684917951679, + "grad_norm": 1.713254690170288, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7445953488349915, + "num_tokens": 208175667.0, + "step": 10578 + }, + { + "epoch": 0.8723509524202193, + "grad_norm": 1.6611496210098267, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7685390710830688, + "num_tokens": 208211107.0, + "step": 10579 + }, + { + "epoch": 0.8724334130452709, + "grad_norm": 2.0014801025390625, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7602300643920898, + "num_tokens": 208239438.0, + "step": 10580 + }, + { + "epoch": 0.8725158736703225, + "grad_norm": 1.5201821327209473, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7717297077178955, + "num_tokens": 208282716.0, + "step": 10581 + }, + { + "epoch": 0.8725983342953739, + "grad_norm": 1.820363998413086, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7272703051567078, + "num_tokens": 208316758.0, + "step": 10582 + }, + { + "epoch": 0.8726807949204255, + "grad_norm": 1.8519370555877686, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7562146186828613, + "num_tokens": 208350153.0, + "step": 10583 + }, + { + "epoch": 0.8727632555454771, + "grad_norm": 1.9303605556488037, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7178382873535156, + "num_tokens": 208381333.0, + "step": 10584 + }, + { + "epoch": 0.8728457161705285, + "grad_norm": 1.7889677286148071, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7378629446029663, + "num_tokens": 208414613.0, + "step": 10585 + }, + { + "epoch": 0.8729281767955801, + "grad_norm": 2.0307869911193848, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7141818404197693, + "num_tokens": 208444477.0, + "step": 10586 + }, + { + "epoch": 0.8730106374206317, + "grad_norm": 1.771377682685852, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7531486749649048, + "num_tokens": 208477231.0, + "step": 10587 + }, + { + "epoch": 0.8730930980456831, + "grad_norm": 1.6691704988479614, + "learning_rate": 1e-06, + "loss": 0.6049, + "mean_token_accuracy": 0.7954946160316467, + "num_tokens": 208511794.0, + "step": 10588 + }, + { + "epoch": 0.8731755586707347, + "grad_norm": 1.9435338973999023, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7290775775909424, + "num_tokens": 208542704.0, + "step": 10589 + }, + { + "epoch": 0.8732580192957863, + "grad_norm": 1.9682320356369019, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7359550595283508, + "num_tokens": 208570189.0, + "step": 10590 + }, + { + "epoch": 0.8733404799208379, + "grad_norm": 1.817025899887085, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7339357137680054, + "num_tokens": 208603134.0, + "step": 10591 + }, + { + "epoch": 0.8734229405458893, + "grad_norm": 2.1658833026885986, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7490258812904358, + "num_tokens": 208633777.0, + "step": 10592 + }, + { + "epoch": 0.8735054011709409, + "grad_norm": 1.7984771728515625, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7458150386810303, + "num_tokens": 208668939.0, + "step": 10593 + }, + { + "epoch": 0.8735878617959925, + "grad_norm": 2.4689736366271973, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7323063015937805, + "num_tokens": 208687702.0, + "step": 10594 + }, + { + "epoch": 0.8736703224210439, + "grad_norm": 2.002804756164551, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7158875465393066, + "num_tokens": 208715608.0, + "step": 10595 + }, + { + "epoch": 0.8737527830460955, + "grad_norm": 1.9364445209503174, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7319674491882324, + "num_tokens": 208743879.0, + "step": 10596 + }, + { + "epoch": 0.8738352436711471, + "grad_norm": 1.7319345474243164, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7805615663528442, + "num_tokens": 208773369.0, + "step": 10597 + }, + { + "epoch": 0.8739177042961985, + "grad_norm": 2.0399436950683594, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7653029561042786, + "num_tokens": 208799163.0, + "step": 10598 + }, + { + "epoch": 0.8740001649212501, + "grad_norm": 2.07151198387146, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7428122162818909, + "num_tokens": 208826240.0, + "step": 10599 + }, + { + "epoch": 0.8740826255463017, + "grad_norm": 2.3157172203063965, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7604554891586304, + "num_tokens": 208847121.0, + "step": 10600 + }, + { + "epoch": 0.8741650861713531, + "grad_norm": 1.9007771015167236, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7208710312843323, + "num_tokens": 208877589.0, + "step": 10601 + }, + { + "epoch": 0.8742475467964047, + "grad_norm": 1.9164196252822876, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7372919917106628, + "num_tokens": 208906763.0, + "step": 10602 + }, + { + "epoch": 0.8743300074214563, + "grad_norm": 1.7919034957885742, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7764155268669128, + "num_tokens": 208940696.0, + "step": 10603 + }, + { + "epoch": 0.8744124680465077, + "grad_norm": 1.8673721551895142, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7515516877174377, + "num_tokens": 208972602.0, + "step": 10604 + }, + { + "epoch": 0.8744949286715593, + "grad_norm": 1.672611951828003, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7410805225372314, + "num_tokens": 209010671.0, + "step": 10605 + }, + { + "epoch": 0.8745773892966109, + "grad_norm": 2.0876684188842773, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7304971218109131, + "num_tokens": 209037424.0, + "step": 10606 + }, + { + "epoch": 0.8746598499216625, + "grad_norm": 1.9172881841659546, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7571462392807007, + "num_tokens": 209065251.0, + "step": 10607 + }, + { + "epoch": 0.8747423105467139, + "grad_norm": 2.1133639812469482, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6930153369903564, + "num_tokens": 209093395.0, + "step": 10608 + }, + { + "epoch": 0.8748247711717655, + "grad_norm": 1.7057230472564697, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7775909900665283, + "num_tokens": 209125778.0, + "step": 10609 + }, + { + "epoch": 0.8749072317968171, + "grad_norm": 1.8873333930969238, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7177994251251221, + "num_tokens": 209156877.0, + "step": 10610 + }, + { + "epoch": 0.8749896924218685, + "grad_norm": 1.8995617628097534, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7398676872253418, + "num_tokens": 209188280.0, + "step": 10611 + }, + { + "epoch": 0.8750721530469201, + "grad_norm": 1.8342955112457275, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7680411338806152, + "num_tokens": 209220640.0, + "step": 10612 + }, + { + "epoch": 0.8751546136719717, + "grad_norm": 1.7584679126739502, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.7911249995231628, + "num_tokens": 209253470.0, + "step": 10613 + }, + { + "epoch": 0.8752370742970231, + "grad_norm": 1.896882176399231, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.757597804069519, + "num_tokens": 209287486.0, + "step": 10614 + }, + { + "epoch": 0.8753195349220747, + "grad_norm": 1.994866967201233, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7057973742485046, + "num_tokens": 209315481.0, + "step": 10615 + }, + { + "epoch": 0.8754019955471263, + "grad_norm": 1.9579542875289917, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7077285051345825, + "num_tokens": 209346567.0, + "step": 10616 + }, + { + "epoch": 0.8754844561721777, + "grad_norm": 2.270432949066162, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7366803884506226, + "num_tokens": 209371767.0, + "step": 10617 + }, + { + "epoch": 0.8755669167972293, + "grad_norm": 2.0266401767730713, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7446187734603882, + "num_tokens": 209398980.0, + "step": 10618 + }, + { + "epoch": 0.8756493774222809, + "grad_norm": 2.1795644760131836, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7233235239982605, + "num_tokens": 209424412.0, + "step": 10619 + }, + { + "epoch": 0.8757318380473323, + "grad_norm": 1.8545403480529785, + "learning_rate": 1e-06, + "loss": 0.593, + "mean_token_accuracy": 0.7909988164901733, + "num_tokens": 209451870.0, + "step": 10620 + }, + { + "epoch": 0.8758142986723839, + "grad_norm": 1.768679141998291, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7079806327819824, + "num_tokens": 209492928.0, + "step": 10621 + }, + { + "epoch": 0.8758967592974355, + "grad_norm": 2.1589951515197754, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7599747180938721, + "num_tokens": 209518509.0, + "step": 10622 + }, + { + "epoch": 0.8759792199224871, + "grad_norm": 1.8447898626327515, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7365689277648926, + "num_tokens": 209551808.0, + "step": 10623 + }, + { + "epoch": 0.8760616805475385, + "grad_norm": 2.0613949298858643, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7254786491394043, + "num_tokens": 209578594.0, + "step": 10624 + }, + { + "epoch": 0.8761441411725901, + "grad_norm": 1.8744990825653076, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7453815937042236, + "num_tokens": 209610089.0, + "step": 10625 + }, + { + "epoch": 0.8762266017976417, + "grad_norm": 1.6702338457107544, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.763886570930481, + "num_tokens": 209646944.0, + "step": 10626 + }, + { + "epoch": 0.8763090624226931, + "grad_norm": 1.8323438167572021, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.765544056892395, + "num_tokens": 209679045.0, + "step": 10627 + }, + { + "epoch": 0.8763915230477447, + "grad_norm": 1.915937066078186, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7385767102241516, + "num_tokens": 209707625.0, + "step": 10628 + }, + { + "epoch": 0.8764739836727963, + "grad_norm": 2.198070526123047, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7288087606430054, + "num_tokens": 209734486.0, + "step": 10629 + }, + { + "epoch": 0.8765564442978477, + "grad_norm": 1.8470066785812378, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7658466100692749, + "num_tokens": 209765249.0, + "step": 10630 + }, + { + "epoch": 0.8766389049228993, + "grad_norm": 1.9275774955749512, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7726840972900391, + "num_tokens": 209794005.0, + "step": 10631 + }, + { + "epoch": 0.8767213655479509, + "grad_norm": 1.944821834564209, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7076512575149536, + "num_tokens": 209826329.0, + "step": 10632 + }, + { + "epoch": 0.8768038261730023, + "grad_norm": 1.7565373182296753, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7484104633331299, + "num_tokens": 209864770.0, + "step": 10633 + }, + { + "epoch": 0.8768862867980539, + "grad_norm": 1.9114899635314941, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7392609119415283, + "num_tokens": 209895966.0, + "step": 10634 + }, + { + "epoch": 0.8769687474231055, + "grad_norm": 2.1828532218933105, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7562689781188965, + "num_tokens": 209921246.0, + "step": 10635 + }, + { + "epoch": 0.8770512080481571, + "grad_norm": 1.7585070133209229, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7685921788215637, + "num_tokens": 209956621.0, + "step": 10636 + }, + { + "epoch": 0.8771336686732085, + "grad_norm": 1.9249693155288696, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7673864364624023, + "num_tokens": 209985519.0, + "step": 10637 + }, + { + "epoch": 0.8772161292982601, + "grad_norm": 1.6769808530807495, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.767756462097168, + "num_tokens": 210019012.0, + "step": 10638 + }, + { + "epoch": 0.8772985899233117, + "grad_norm": 1.785918951034546, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7657557725906372, + "num_tokens": 210052517.0, + "step": 10639 + }, + { + "epoch": 0.8773810505483631, + "grad_norm": 2.1952669620513916, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7107383608818054, + "num_tokens": 210077104.0, + "step": 10640 + }, + { + "epoch": 0.8774635111734147, + "grad_norm": 2.0198733806610107, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7461078763008118, + "num_tokens": 210104823.0, + "step": 10641 + }, + { + "epoch": 0.8775459717984663, + "grad_norm": 2.3703644275665283, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7200261354446411, + "num_tokens": 210127911.0, + "step": 10642 + }, + { + "epoch": 0.8776284324235177, + "grad_norm": 1.9596375226974487, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7603094577789307, + "num_tokens": 210156426.0, + "step": 10643 + }, + { + "epoch": 0.8777108930485693, + "grad_norm": 2.008112668991089, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7391707897186279, + "num_tokens": 210184753.0, + "step": 10644 + }, + { + "epoch": 0.8777933536736209, + "grad_norm": 2.174043893814087, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7368467450141907, + "num_tokens": 210210548.0, + "step": 10645 + }, + { + "epoch": 0.8778758142986723, + "grad_norm": 2.0933468341827393, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7213660478591919, + "num_tokens": 210239717.0, + "step": 10646 + }, + { + "epoch": 0.8779582749237239, + "grad_norm": 2.3146920204162598, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7067106366157532, + "num_tokens": 210264088.0, + "step": 10647 + }, + { + "epoch": 0.8780407355487755, + "grad_norm": 1.8323280811309814, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7726701498031616, + "num_tokens": 210293575.0, + "step": 10648 + }, + { + "epoch": 0.878123196173827, + "grad_norm": 2.1252248287200928, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7344874739646912, + "num_tokens": 210320076.0, + "step": 10649 + }, + { + "epoch": 0.8782056567988785, + "grad_norm": 1.8805580139160156, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7306673526763916, + "num_tokens": 210353343.0, + "step": 10650 + }, + { + "epoch": 0.8782881174239301, + "grad_norm": 1.7839106321334839, + "learning_rate": 1e-06, + "loss": 0.5437, + "mean_token_accuracy": 0.8216089010238647, + "num_tokens": 210383767.0, + "step": 10651 + }, + { + "epoch": 0.8783705780489817, + "grad_norm": 1.7777587175369263, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7711995244026184, + "num_tokens": 210413564.0, + "step": 10652 + }, + { + "epoch": 0.8784530386740331, + "grad_norm": 1.9020715951919556, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7715951800346375, + "num_tokens": 210443891.0, + "step": 10653 + }, + { + "epoch": 0.8785354992990847, + "grad_norm": 1.9519712924957275, + "learning_rate": 1e-06, + "loss": 0.8014, + "mean_token_accuracy": 0.7435123324394226, + "num_tokens": 210476504.0, + "step": 10654 + }, + { + "epoch": 0.8786179599241363, + "grad_norm": 2.0328617095947266, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7019222378730774, + "num_tokens": 210506026.0, + "step": 10655 + }, + { + "epoch": 0.8787004205491877, + "grad_norm": 1.9052343368530273, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7726107239723206, + "num_tokens": 210539208.0, + "step": 10656 + }, + { + "epoch": 0.8787828811742393, + "grad_norm": 1.676681399345398, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7707405090332031, + "num_tokens": 210573758.0, + "step": 10657 + }, + { + "epoch": 0.8788653417992909, + "grad_norm": 1.9619989395141602, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7569438219070435, + "num_tokens": 210602484.0, + "step": 10658 + }, + { + "epoch": 0.8789478024243423, + "grad_norm": 2.2747833728790283, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7655813694000244, + "num_tokens": 210623773.0, + "step": 10659 + }, + { + "epoch": 0.8790302630493939, + "grad_norm": 1.862557291984558, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7364637851715088, + "num_tokens": 210654383.0, + "step": 10660 + }, + { + "epoch": 0.8791127236744455, + "grad_norm": 1.793546438217163, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7678316235542297, + "num_tokens": 210686487.0, + "step": 10661 + }, + { + "epoch": 0.879195184299497, + "grad_norm": 1.8316951990127563, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7534741163253784, + "num_tokens": 210720863.0, + "step": 10662 + }, + { + "epoch": 0.8792776449245485, + "grad_norm": 1.8755466938018799, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7476343512535095, + "num_tokens": 210753037.0, + "step": 10663 + }, + { + "epoch": 0.8793601055496001, + "grad_norm": 1.964293122291565, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7416247129440308, + "num_tokens": 210784788.0, + "step": 10664 + }, + { + "epoch": 0.8794425661746516, + "grad_norm": 1.7510559558868408, + "learning_rate": 1e-06, + "loss": 0.5386, + "mean_token_accuracy": 0.8173186779022217, + "num_tokens": 210815195.0, + "step": 10665 + }, + { + "epoch": 0.8795250267997031, + "grad_norm": 1.8239561319351196, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7606078386306763, + "num_tokens": 210844352.0, + "step": 10666 + }, + { + "epoch": 0.8796074874247547, + "grad_norm": 1.976230263710022, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7719743251800537, + "num_tokens": 210872238.0, + "step": 10667 + }, + { + "epoch": 0.8796899480498063, + "grad_norm": 2.0787339210510254, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7490171194076538, + "num_tokens": 210898526.0, + "step": 10668 + }, + { + "epoch": 0.8797724086748577, + "grad_norm": 1.7863901853561401, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.7855585813522339, + "num_tokens": 210928501.0, + "step": 10669 + }, + { + "epoch": 0.8798548692999093, + "grad_norm": 2.075371265411377, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.723892331123352, + "num_tokens": 210954419.0, + "step": 10670 + }, + { + "epoch": 0.8799373299249609, + "grad_norm": 1.648038625717163, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7814432978630066, + "num_tokens": 210995648.0, + "step": 10671 + }, + { + "epoch": 0.8800197905500123, + "grad_norm": 1.5641971826553345, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.785108208656311, + "num_tokens": 211035155.0, + "step": 10672 + }, + { + "epoch": 0.8801022511750639, + "grad_norm": 1.9869967699050903, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7840132713317871, + "num_tokens": 211062161.0, + "step": 10673 + }, + { + "epoch": 0.8801847118001155, + "grad_norm": 1.8099792003631592, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7433854937553406, + "num_tokens": 211093671.0, + "step": 10674 + }, + { + "epoch": 0.8802671724251669, + "grad_norm": 2.0557897090911865, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7241683602333069, + "num_tokens": 211121771.0, + "step": 10675 + }, + { + "epoch": 0.8803496330502185, + "grad_norm": 1.8402206897735596, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7239566445350647, + "num_tokens": 211154776.0, + "step": 10676 + }, + { + "epoch": 0.8804320936752701, + "grad_norm": 1.8015260696411133, + "learning_rate": 1e-06, + "loss": 0.6251, + "mean_token_accuracy": 0.7942810654640198, + "num_tokens": 211185478.0, + "step": 10677 + }, + { + "epoch": 0.8805145543003216, + "grad_norm": 1.8398040533065796, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7880532741546631, + "num_tokens": 211215217.0, + "step": 10678 + }, + { + "epoch": 0.8805970149253731, + "grad_norm": 2.0423388481140137, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7563991546630859, + "num_tokens": 211243577.0, + "step": 10679 + }, + { + "epoch": 0.8806794755504247, + "grad_norm": 2.0837504863739014, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7752564549446106, + "num_tokens": 211269596.0, + "step": 10680 + }, + { + "epoch": 0.8807619361754762, + "grad_norm": 2.1226041316986084, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7486087083816528, + "num_tokens": 211299092.0, + "step": 10681 + }, + { + "epoch": 0.8808443968005277, + "grad_norm": 1.902868628501892, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7313281297683716, + "num_tokens": 211332200.0, + "step": 10682 + }, + { + "epoch": 0.8809268574255793, + "grad_norm": 2.0197248458862305, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7351847887039185, + "num_tokens": 211360778.0, + "step": 10683 + }, + { + "epoch": 0.8810093180506309, + "grad_norm": 1.6585253477096558, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7866697311401367, + "num_tokens": 211398743.0, + "step": 10684 + }, + { + "epoch": 0.8810917786756823, + "grad_norm": 2.4078660011291504, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.721740186214447, + "num_tokens": 211419368.0, + "step": 10685 + }, + { + "epoch": 0.8811742393007339, + "grad_norm": 1.7688699960708618, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7188172340393066, + "num_tokens": 211457398.0, + "step": 10686 + }, + { + "epoch": 0.8812566999257855, + "grad_norm": 1.7575838565826416, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7695576548576355, + "num_tokens": 211489299.0, + "step": 10687 + }, + { + "epoch": 0.8813391605508369, + "grad_norm": 2.268632650375366, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.729529082775116, + "num_tokens": 211513303.0, + "step": 10688 + }, + { + "epoch": 0.8814216211758885, + "grad_norm": 2.197774887084961, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7338641881942749, + "num_tokens": 211537152.0, + "step": 10689 + }, + { + "epoch": 0.8815040818009401, + "grad_norm": 1.7833454608917236, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7876228094100952, + "num_tokens": 211570675.0, + "step": 10690 + }, + { + "epoch": 0.8815865424259915, + "grad_norm": 1.9910826683044434, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7685769200325012, + "num_tokens": 211596509.0, + "step": 10691 + }, + { + "epoch": 0.8816690030510431, + "grad_norm": 1.8783758878707886, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7691022157669067, + "num_tokens": 211625033.0, + "step": 10692 + }, + { + "epoch": 0.8817514636760947, + "grad_norm": 1.9843865633010864, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7452806234359741, + "num_tokens": 211653003.0, + "step": 10693 + }, + { + "epoch": 0.8818339243011462, + "grad_norm": 1.7312064170837402, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.743452250957489, + "num_tokens": 211692114.0, + "step": 10694 + }, + { + "epoch": 0.8819163849261977, + "grad_norm": 1.7050292491912842, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7679775357246399, + "num_tokens": 211725248.0, + "step": 10695 + }, + { + "epoch": 0.8819988455512493, + "grad_norm": 1.9998033046722412, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7641385793685913, + "num_tokens": 211752535.0, + "step": 10696 + }, + { + "epoch": 0.8820813061763008, + "grad_norm": 2.057582139968872, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.733688473701477, + "num_tokens": 211779617.0, + "step": 10697 + }, + { + "epoch": 0.8821637668013523, + "grad_norm": 1.9594117403030396, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7413357496261597, + "num_tokens": 211807541.0, + "step": 10698 + }, + { + "epoch": 0.8822462274264039, + "grad_norm": 1.9754172563552856, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7523322105407715, + "num_tokens": 211835470.0, + "step": 10699 + }, + { + "epoch": 0.8823286880514555, + "grad_norm": 1.86762535572052, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7496433854103088, + "num_tokens": 211869030.0, + "step": 10700 + }, + { + "epoch": 0.8824111486765069, + "grad_norm": 2.0451362133026123, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.735040545463562, + "num_tokens": 211896660.0, + "step": 10701 + }, + { + "epoch": 0.8824936093015585, + "grad_norm": 1.9209874868392944, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7238866090774536, + "num_tokens": 211926947.0, + "step": 10702 + }, + { + "epoch": 0.8825760699266101, + "grad_norm": 2.031761884689331, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7187234163284302, + "num_tokens": 211956948.0, + "step": 10703 + }, + { + "epoch": 0.8826585305516615, + "grad_norm": 1.9304304122924805, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.7004094123840332, + "num_tokens": 211990163.0, + "step": 10704 + }, + { + "epoch": 0.8827409911767131, + "grad_norm": 1.8990464210510254, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7502856254577637, + "num_tokens": 212023451.0, + "step": 10705 + }, + { + "epoch": 0.8828234518017647, + "grad_norm": 1.9473495483398438, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7514548897743225, + "num_tokens": 212053097.0, + "step": 10706 + }, + { + "epoch": 0.8829059124268162, + "grad_norm": 2.092456817626953, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.763443112373352, + "num_tokens": 212081446.0, + "step": 10707 + }, + { + "epoch": 0.8829883730518677, + "grad_norm": 1.7219974994659424, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.783562421798706, + "num_tokens": 212116173.0, + "step": 10708 + }, + { + "epoch": 0.8830708336769193, + "grad_norm": 1.9411835670471191, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7261642217636108, + "num_tokens": 212148205.0, + "step": 10709 + }, + { + "epoch": 0.8831532943019708, + "grad_norm": 1.9076999425888062, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6985710859298706, + "num_tokens": 212180071.0, + "step": 10710 + }, + { + "epoch": 0.8832357549270223, + "grad_norm": 1.793395757675171, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7652533054351807, + "num_tokens": 212210739.0, + "step": 10711 + }, + { + "epoch": 0.8833182155520739, + "grad_norm": 2.103311777114868, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7528092861175537, + "num_tokens": 212235191.0, + "step": 10712 + }, + { + "epoch": 0.8834006761771255, + "grad_norm": 2.004903554916382, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7459524273872375, + "num_tokens": 212262922.0, + "step": 10713 + }, + { + "epoch": 0.8834831368021769, + "grad_norm": 1.8453922271728516, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7735086679458618, + "num_tokens": 212294011.0, + "step": 10714 + }, + { + "epoch": 0.8835655974272285, + "grad_norm": 1.9851319789886475, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7647250294685364, + "num_tokens": 212320820.0, + "step": 10715 + }, + { + "epoch": 0.8836480580522801, + "grad_norm": 1.642167329788208, + "learning_rate": 1e-06, + "loss": 0.5881, + "mean_token_accuracy": 0.8004493713378906, + "num_tokens": 212355433.0, + "step": 10716 + }, + { + "epoch": 0.8837305186773315, + "grad_norm": 1.7988916635513306, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7193652987480164, + "num_tokens": 212391280.0, + "step": 10717 + }, + { + "epoch": 0.8838129793023831, + "grad_norm": 1.8711003065109253, + "learning_rate": 1e-06, + "loss": 0.7656, + "mean_token_accuracy": 0.7646214365959167, + "num_tokens": 212423446.0, + "step": 10718 + }, + { + "epoch": 0.8838954399274347, + "grad_norm": 2.3512794971466064, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7111798524856567, + "num_tokens": 212444701.0, + "step": 10719 + }, + { + "epoch": 0.8839779005524862, + "grad_norm": 1.8664143085479736, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7395122647285461, + "num_tokens": 212474993.0, + "step": 10720 + }, + { + "epoch": 0.8840603611775377, + "grad_norm": 2.1097781658172607, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7044773101806641, + "num_tokens": 212501578.0, + "step": 10721 + }, + { + "epoch": 0.8841428218025893, + "grad_norm": 1.944577932357788, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7527796030044556, + "num_tokens": 212529521.0, + "step": 10722 + }, + { + "epoch": 0.8842252824276408, + "grad_norm": 1.9919614791870117, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7395235300064087, + "num_tokens": 212558298.0, + "step": 10723 + }, + { + "epoch": 0.8843077430526923, + "grad_norm": 1.9631521701812744, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7478653192520142, + "num_tokens": 212585584.0, + "step": 10724 + }, + { + "epoch": 0.8843902036777439, + "grad_norm": 1.9709113836288452, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7328854203224182, + "num_tokens": 212615296.0, + "step": 10725 + }, + { + "epoch": 0.8844726643027954, + "grad_norm": 1.7881598472595215, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7749794125556946, + "num_tokens": 212650673.0, + "step": 10726 + }, + { + "epoch": 0.8845551249278469, + "grad_norm": 1.8626806735992432, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.745890736579895, + "num_tokens": 212682848.0, + "step": 10727 + }, + { + "epoch": 0.8846375855528985, + "grad_norm": 1.943521499633789, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7281945943832397, + "num_tokens": 212713756.0, + "step": 10728 + }, + { + "epoch": 0.8847200461779501, + "grad_norm": 1.8908681869506836, + "learning_rate": 1e-06, + "loss": 0.7883, + "mean_token_accuracy": 0.7502048015594482, + "num_tokens": 212744068.0, + "step": 10729 + }, + { + "epoch": 0.8848025068030015, + "grad_norm": 2.147390842437744, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7500689029693604, + "num_tokens": 212771216.0, + "step": 10730 + }, + { + "epoch": 0.8848849674280531, + "grad_norm": 1.7168316841125488, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7690014243125916, + "num_tokens": 212805330.0, + "step": 10731 + }, + { + "epoch": 0.8849674280531047, + "grad_norm": 2.012566328048706, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7850648760795593, + "num_tokens": 212832916.0, + "step": 10732 + }, + { + "epoch": 0.8850498886781561, + "grad_norm": 1.8178421258926392, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7355985641479492, + "num_tokens": 212868229.0, + "step": 10733 + }, + { + "epoch": 0.8851323493032077, + "grad_norm": 1.909281849861145, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7174373269081116, + "num_tokens": 212899763.0, + "step": 10734 + }, + { + "epoch": 0.8852148099282593, + "grad_norm": 1.984819769859314, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7358888387680054, + "num_tokens": 212929174.0, + "step": 10735 + }, + { + "epoch": 0.8852972705533108, + "grad_norm": 2.1552977561950684, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.770041286945343, + "num_tokens": 212952670.0, + "step": 10736 + }, + { + "epoch": 0.8853797311783623, + "grad_norm": 1.9466527700424194, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7676928043365479, + "num_tokens": 212980790.0, + "step": 10737 + }, + { + "epoch": 0.8854621918034139, + "grad_norm": 1.8264864683151245, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.781758189201355, + "num_tokens": 213010395.0, + "step": 10738 + }, + { + "epoch": 0.8855446524284654, + "grad_norm": 1.8995834589004517, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.7810420989990234, + "num_tokens": 213039436.0, + "step": 10739 + }, + { + "epoch": 0.8856271130535169, + "grad_norm": 2.1552488803863525, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7230965495109558, + "num_tokens": 213065286.0, + "step": 10740 + }, + { + "epoch": 0.8857095736785685, + "grad_norm": 1.7577767372131348, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7227748036384583, + "num_tokens": 213105682.0, + "step": 10741 + }, + { + "epoch": 0.88579203430362, + "grad_norm": 2.161475658416748, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7245190143585205, + "num_tokens": 213131029.0, + "step": 10742 + }, + { + "epoch": 0.8858744949286715, + "grad_norm": 1.9638315439224243, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.7930850982666016, + "num_tokens": 213160639.0, + "step": 10743 + }, + { + "epoch": 0.8859569555537231, + "grad_norm": 1.9321542978286743, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.753232479095459, + "num_tokens": 213188827.0, + "step": 10744 + }, + { + "epoch": 0.8860394161787747, + "grad_norm": 1.6702207326889038, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.7867084741592407, + "num_tokens": 213226266.0, + "step": 10745 + }, + { + "epoch": 0.8861218768038261, + "grad_norm": 1.8509835004806519, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.749721884727478, + "num_tokens": 213257426.0, + "step": 10746 + }, + { + "epoch": 0.8862043374288777, + "grad_norm": 1.813348650932312, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.7902203798294067, + "num_tokens": 213293255.0, + "step": 10747 + }, + { + "epoch": 0.8862867980539293, + "grad_norm": 1.8511992692947388, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7619616985321045, + "num_tokens": 213326194.0, + "step": 10748 + }, + { + "epoch": 0.8863692586789808, + "grad_norm": 1.9771660566329956, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7363423705101013, + "num_tokens": 213354379.0, + "step": 10749 + }, + { + "epoch": 0.8864517193040323, + "grad_norm": 1.7187026739120483, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.7905400991439819, + "num_tokens": 213391412.0, + "step": 10750 + }, + { + "epoch": 0.8865341799290839, + "grad_norm": 2.0092732906341553, + "learning_rate": 1e-06, + "loss": 0.7808, + "mean_token_accuracy": 0.7563766241073608, + "num_tokens": 213422800.0, + "step": 10751 + }, + { + "epoch": 0.8866166405541354, + "grad_norm": 1.8367031812667847, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7319422364234924, + "num_tokens": 213456253.0, + "step": 10752 + }, + { + "epoch": 0.8866991011791869, + "grad_norm": 2.122697114944458, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7396460771560669, + "num_tokens": 213480174.0, + "step": 10753 + }, + { + "epoch": 0.8867815618042385, + "grad_norm": 1.9406726360321045, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7393463850021362, + "num_tokens": 213510165.0, + "step": 10754 + }, + { + "epoch": 0.88686402242929, + "grad_norm": 1.6924054622650146, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7631784677505493, + "num_tokens": 213545605.0, + "step": 10755 + }, + { + "epoch": 0.8869464830543415, + "grad_norm": 1.743064045906067, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7628788948059082, + "num_tokens": 213581794.0, + "step": 10756 + }, + { + "epoch": 0.8870289436793931, + "grad_norm": 1.858041763305664, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.7477715611457825, + "num_tokens": 213612630.0, + "step": 10757 + }, + { + "epoch": 0.8871114043044446, + "grad_norm": 1.8958579301834106, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7673916220664978, + "num_tokens": 213645129.0, + "step": 10758 + }, + { + "epoch": 0.8871938649294961, + "grad_norm": 2.0721449851989746, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7754877805709839, + "num_tokens": 213673482.0, + "step": 10759 + }, + { + "epoch": 0.8872763255545477, + "grad_norm": 2.1650230884552, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7288378477096558, + "num_tokens": 213699631.0, + "step": 10760 + }, + { + "epoch": 0.8873587861795993, + "grad_norm": 1.7748597860336304, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7704691290855408, + "num_tokens": 213730100.0, + "step": 10761 + }, + { + "epoch": 0.8874412468046508, + "grad_norm": 2.2516403198242188, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7438761591911316, + "num_tokens": 213752236.0, + "step": 10762 + }, + { + "epoch": 0.8875237074297023, + "grad_norm": 1.839532732963562, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7646467685699463, + "num_tokens": 213783489.0, + "step": 10763 + }, + { + "epoch": 0.8876061680547539, + "grad_norm": 1.7955671548843384, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7342849969863892, + "num_tokens": 213818093.0, + "step": 10764 + }, + { + "epoch": 0.8876886286798054, + "grad_norm": 1.8708343505859375, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7767161130905151, + "num_tokens": 213848101.0, + "step": 10765 + }, + { + "epoch": 0.8877710893048569, + "grad_norm": 1.9908137321472168, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7586885690689087, + "num_tokens": 213879040.0, + "step": 10766 + }, + { + "epoch": 0.8878535499299085, + "grad_norm": 1.773404836654663, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7610238194465637, + "num_tokens": 213912671.0, + "step": 10767 + }, + { + "epoch": 0.88793601055496, + "grad_norm": 1.887935996055603, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7054214477539062, + "num_tokens": 213946539.0, + "step": 10768 + }, + { + "epoch": 0.8880184711800115, + "grad_norm": 2.0582966804504395, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7081985473632812, + "num_tokens": 213977584.0, + "step": 10769 + }, + { + "epoch": 0.8881009318050631, + "grad_norm": 1.6674799919128418, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7649341821670532, + "num_tokens": 214014661.0, + "step": 10770 + }, + { + "epoch": 0.8881833924301146, + "grad_norm": 2.2840354442596436, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7367320656776428, + "num_tokens": 214039993.0, + "step": 10771 + }, + { + "epoch": 0.8882658530551661, + "grad_norm": 1.8802121877670288, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7753621339797974, + "num_tokens": 214067913.0, + "step": 10772 + }, + { + "epoch": 0.8883483136802177, + "grad_norm": 2.1076180934906006, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7444157600402832, + "num_tokens": 214092301.0, + "step": 10773 + }, + { + "epoch": 0.8884307743052693, + "grad_norm": 1.8593642711639404, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7416603565216064, + "num_tokens": 214126333.0, + "step": 10774 + }, + { + "epoch": 0.8885132349303207, + "grad_norm": 2.0127432346343994, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7443054914474487, + "num_tokens": 214156622.0, + "step": 10775 + }, + { + "epoch": 0.8885956955553723, + "grad_norm": 1.9809733629226685, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.741657555103302, + "num_tokens": 214186166.0, + "step": 10776 + }, + { + "epoch": 0.8886781561804239, + "grad_norm": 1.729695200920105, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7544943690299988, + "num_tokens": 214221238.0, + "step": 10777 + }, + { + "epoch": 0.8887606168054754, + "grad_norm": 1.834851622581482, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7877292633056641, + "num_tokens": 214251497.0, + "step": 10778 + }, + { + "epoch": 0.8888430774305269, + "grad_norm": 1.9551160335540771, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7286083698272705, + "num_tokens": 214282284.0, + "step": 10779 + }, + { + "epoch": 0.8889255380555785, + "grad_norm": 1.8313418626785278, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7245239019393921, + "num_tokens": 214314635.0, + "step": 10780 + }, + { + "epoch": 0.88900799868063, + "grad_norm": 1.8354352712631226, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.731881856918335, + "num_tokens": 214345295.0, + "step": 10781 + }, + { + "epoch": 0.8890904593056815, + "grad_norm": 2.218299627304077, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7803695201873779, + "num_tokens": 214367875.0, + "step": 10782 + }, + { + "epoch": 0.8891729199307331, + "grad_norm": 1.6693520545959473, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.756812572479248, + "num_tokens": 214405290.0, + "step": 10783 + }, + { + "epoch": 0.8892553805557846, + "grad_norm": 2.030510187149048, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7214446663856506, + "num_tokens": 214434991.0, + "step": 10784 + }, + { + "epoch": 0.8893378411808361, + "grad_norm": 1.9074214696884155, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7429122924804688, + "num_tokens": 214466254.0, + "step": 10785 + }, + { + "epoch": 0.8894203018058877, + "grad_norm": 1.8807411193847656, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7435605525970459, + "num_tokens": 214496975.0, + "step": 10786 + }, + { + "epoch": 0.8895027624309392, + "grad_norm": 1.9884073734283447, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.753422200679779, + "num_tokens": 214525986.0, + "step": 10787 + }, + { + "epoch": 0.8895852230559907, + "grad_norm": 2.187223196029663, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7593685388565063, + "num_tokens": 214549056.0, + "step": 10788 + }, + { + "epoch": 0.8896676836810423, + "grad_norm": 1.8539341688156128, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7637103796005249, + "num_tokens": 214584532.0, + "step": 10789 + }, + { + "epoch": 0.8897501443060939, + "grad_norm": 1.9372093677520752, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7585151195526123, + "num_tokens": 214612372.0, + "step": 10790 + }, + { + "epoch": 0.8898326049311454, + "grad_norm": 2.243765115737915, + "learning_rate": 1e-06, + "loss": 0.7768, + "mean_token_accuracy": 0.7407520413398743, + "num_tokens": 214635474.0, + "step": 10791 + }, + { + "epoch": 0.8899150655561969, + "grad_norm": 1.9872140884399414, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7427706718444824, + "num_tokens": 214663886.0, + "step": 10792 + }, + { + "epoch": 0.8899975261812485, + "grad_norm": 1.902208924293518, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.7818571329116821, + "num_tokens": 214697076.0, + "step": 10793 + }, + { + "epoch": 0.8900799868063, + "grad_norm": 2.133600950241089, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.7484604716300964, + "num_tokens": 214722117.0, + "step": 10794 + }, + { + "epoch": 0.8901624474313515, + "grad_norm": 1.9421210289001465, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7579426765441895, + "num_tokens": 214751533.0, + "step": 10795 + }, + { + "epoch": 0.8902449080564031, + "grad_norm": 1.928783893585205, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.7956573963165283, + "num_tokens": 214778258.0, + "step": 10796 + }, + { + "epoch": 0.8903273686814546, + "grad_norm": 1.8853322267532349, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7651081085205078, + "num_tokens": 214810749.0, + "step": 10797 + }, + { + "epoch": 0.8904098293065061, + "grad_norm": 2.0113470554351807, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7540537118911743, + "num_tokens": 214838931.0, + "step": 10798 + }, + { + "epoch": 0.8904922899315577, + "grad_norm": 1.7742295265197754, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7586795091629028, + "num_tokens": 214871940.0, + "step": 10799 + }, + { + "epoch": 0.8905747505566092, + "grad_norm": 1.830554723739624, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.7915709614753723, + "num_tokens": 214903104.0, + "step": 10800 + }, + { + "epoch": 0.8906572111816607, + "grad_norm": 2.235316753387451, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7176638245582581, + "num_tokens": 214928295.0, + "step": 10801 + }, + { + "epoch": 0.8907396718067123, + "grad_norm": 1.9197295904159546, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7610257267951965, + "num_tokens": 214958812.0, + "step": 10802 + }, + { + "epoch": 0.8908221324317638, + "grad_norm": 1.887567162513733, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7448989748954773, + "num_tokens": 214987988.0, + "step": 10803 + }, + { + "epoch": 0.8909045930568154, + "grad_norm": 2.2328779697418213, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7351186275482178, + "num_tokens": 215012584.0, + "step": 10804 + }, + { + "epoch": 0.8909870536818669, + "grad_norm": 1.8211742639541626, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7633641362190247, + "num_tokens": 215043689.0, + "step": 10805 + }, + { + "epoch": 0.8910695143069185, + "grad_norm": 1.8969603776931763, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7657026052474976, + "num_tokens": 215071604.0, + "step": 10806 + }, + { + "epoch": 0.89115197493197, + "grad_norm": 1.9466089010238647, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7509104609489441, + "num_tokens": 215099462.0, + "step": 10807 + }, + { + "epoch": 0.8912344355570215, + "grad_norm": 2.0391905307769775, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7734372615814209, + "num_tokens": 215124234.0, + "step": 10808 + }, + { + "epoch": 0.8913168961820731, + "grad_norm": 1.939946174621582, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7510350942611694, + "num_tokens": 215153957.0, + "step": 10809 + }, + { + "epoch": 0.8913993568071246, + "grad_norm": 2.228093147277832, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7609536647796631, + "num_tokens": 215177118.0, + "step": 10810 + }, + { + "epoch": 0.8914818174321761, + "grad_norm": 2.3402771949768066, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7285231351852417, + "num_tokens": 215200086.0, + "step": 10811 + }, + { + "epoch": 0.8915642780572277, + "grad_norm": 1.9112157821655273, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.754874587059021, + "num_tokens": 215230582.0, + "step": 10812 + }, + { + "epoch": 0.8916467386822792, + "grad_norm": 2.1825308799743652, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7277816534042358, + "num_tokens": 215257287.0, + "step": 10813 + }, + { + "epoch": 0.8917291993073307, + "grad_norm": 2.134636402130127, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7490630745887756, + "num_tokens": 215281032.0, + "step": 10814 + }, + { + "epoch": 0.8918116599323823, + "grad_norm": 2.1173031330108643, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7323037385940552, + "num_tokens": 215307990.0, + "step": 10815 + }, + { + "epoch": 0.8918941205574338, + "grad_norm": 1.834282398223877, + "learning_rate": 1e-06, + "loss": 0.6011, + "mean_token_accuracy": 0.8029347062110901, + "num_tokens": 215339815.0, + "step": 10816 + }, + { + "epoch": 0.8919765811824854, + "grad_norm": 1.7406566143035889, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7830052375793457, + "num_tokens": 215376733.0, + "step": 10817 + }, + { + "epoch": 0.8920590418075369, + "grad_norm": 1.8118445873260498, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7297542691230774, + "num_tokens": 215410272.0, + "step": 10818 + }, + { + "epoch": 0.8921415024325884, + "grad_norm": 2.0532009601593018, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7448468208312988, + "num_tokens": 215438904.0, + "step": 10819 + }, + { + "epoch": 0.89222396305764, + "grad_norm": 1.8547741174697876, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7788196802139282, + "num_tokens": 215471333.0, + "step": 10820 + }, + { + "epoch": 0.8923064236826915, + "grad_norm": 2.372143030166626, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7247311472892761, + "num_tokens": 215494360.0, + "step": 10821 + }, + { + "epoch": 0.8923888843077431, + "grad_norm": 1.8524751663208008, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7764439582824707, + "num_tokens": 215526934.0, + "step": 10822 + }, + { + "epoch": 0.8924713449327946, + "grad_norm": 1.7706656455993652, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7790709137916565, + "num_tokens": 215559070.0, + "step": 10823 + }, + { + "epoch": 0.8925538055578461, + "grad_norm": 1.7229540348052979, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7361387014389038, + "num_tokens": 215598720.0, + "step": 10824 + }, + { + "epoch": 0.8926362661828977, + "grad_norm": 1.8165854215621948, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7609230279922485, + "num_tokens": 215632338.0, + "step": 10825 + }, + { + "epoch": 0.8927187268079492, + "grad_norm": 1.9322882890701294, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7235940098762512, + "num_tokens": 215662382.0, + "step": 10826 + }, + { + "epoch": 0.8928011874330007, + "grad_norm": 1.8462934494018555, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.7998801469802856, + "num_tokens": 215698200.0, + "step": 10827 + }, + { + "epoch": 0.8928836480580523, + "grad_norm": 2.0449256896972656, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7535855770111084, + "num_tokens": 215725807.0, + "step": 10828 + }, + { + "epoch": 0.8929661086831038, + "grad_norm": 1.8287543058395386, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7767132520675659, + "num_tokens": 215758305.0, + "step": 10829 + }, + { + "epoch": 0.8930485693081553, + "grad_norm": 2.019477367401123, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.714623212814331, + "num_tokens": 215788594.0, + "step": 10830 + }, + { + "epoch": 0.8931310299332069, + "grad_norm": 1.9125984907150269, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7563128471374512, + "num_tokens": 215818404.0, + "step": 10831 + }, + { + "epoch": 0.8932134905582584, + "grad_norm": 1.8391075134277344, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7649163603782654, + "num_tokens": 215848307.0, + "step": 10832 + }, + { + "epoch": 0.89329595118331, + "grad_norm": 1.957434892654419, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7447041869163513, + "num_tokens": 215877458.0, + "step": 10833 + }, + { + "epoch": 0.8933784118083615, + "grad_norm": 2.297806978225708, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7132259607315063, + "num_tokens": 215902694.0, + "step": 10834 + }, + { + "epoch": 0.8934608724334131, + "grad_norm": 1.8686151504516602, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7720061540603638, + "num_tokens": 215933702.0, + "step": 10835 + }, + { + "epoch": 0.8935433330584646, + "grad_norm": 2.165820598602295, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7274336218833923, + "num_tokens": 215958470.0, + "step": 10836 + }, + { + "epoch": 0.8936257936835161, + "grad_norm": 2.018354892730713, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7279166579246521, + "num_tokens": 215989412.0, + "step": 10837 + }, + { + "epoch": 0.8937082543085677, + "grad_norm": 1.7589755058288574, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.7824302911758423, + "num_tokens": 216022521.0, + "step": 10838 + }, + { + "epoch": 0.8937907149336192, + "grad_norm": 1.9251309633255005, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.744520902633667, + "num_tokens": 216053737.0, + "step": 10839 + }, + { + "epoch": 0.8938731755586707, + "grad_norm": 2.0608630180358887, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7511951923370361, + "num_tokens": 216079489.0, + "step": 10840 + }, + { + "epoch": 0.8939556361837223, + "grad_norm": 2.008490562438965, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7103276252746582, + "num_tokens": 216110834.0, + "step": 10841 + }, + { + "epoch": 0.8940380968087738, + "grad_norm": 1.8605343103408813, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7609140872955322, + "num_tokens": 216144353.0, + "step": 10842 + }, + { + "epoch": 0.8941205574338253, + "grad_norm": 2.1192617416381836, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7530810832977295, + "num_tokens": 216171534.0, + "step": 10843 + }, + { + "epoch": 0.8942030180588769, + "grad_norm": 1.9071459770202637, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7688324451446533, + "num_tokens": 216201312.0, + "step": 10844 + }, + { + "epoch": 0.8942854786839284, + "grad_norm": 1.895714282989502, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7867612838745117, + "num_tokens": 216229701.0, + "step": 10845 + }, + { + "epoch": 0.89436793930898, + "grad_norm": 1.9136569499969482, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7215139865875244, + "num_tokens": 216261248.0, + "step": 10846 + }, + { + "epoch": 0.8944503999340315, + "grad_norm": 1.8632233142852783, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7700996398925781, + "num_tokens": 216293565.0, + "step": 10847 + }, + { + "epoch": 0.894532860559083, + "grad_norm": 2.312870502471924, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7234190702438354, + "num_tokens": 216321505.0, + "step": 10848 + }, + { + "epoch": 0.8946153211841346, + "grad_norm": 1.908669352531433, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7515153884887695, + "num_tokens": 216351314.0, + "step": 10849 + }, + { + "epoch": 0.8946977818091861, + "grad_norm": 1.9030057191848755, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7510248422622681, + "num_tokens": 216381769.0, + "step": 10850 + }, + { + "epoch": 0.8947802424342377, + "grad_norm": 1.9833444356918335, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.700305700302124, + "num_tokens": 216411635.0, + "step": 10851 + }, + { + "epoch": 0.8948627030592892, + "grad_norm": 1.9001976251602173, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7831900119781494, + "num_tokens": 216440424.0, + "step": 10852 + }, + { + "epoch": 0.8949451636843407, + "grad_norm": 1.8335360288619995, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7670193910598755, + "num_tokens": 216471638.0, + "step": 10853 + }, + { + "epoch": 0.8950276243093923, + "grad_norm": 1.9597581624984741, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7149924039840698, + "num_tokens": 216501050.0, + "step": 10854 + }, + { + "epoch": 0.8951100849344438, + "grad_norm": 2.0615804195404053, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7308646440505981, + "num_tokens": 216530851.0, + "step": 10855 + }, + { + "epoch": 0.8951925455594953, + "grad_norm": 2.1220226287841797, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7703824043273926, + "num_tokens": 216555289.0, + "step": 10856 + }, + { + "epoch": 0.8952750061845469, + "grad_norm": 1.8632837533950806, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7478560209274292, + "num_tokens": 216587292.0, + "step": 10857 + }, + { + "epoch": 0.8953574668095984, + "grad_norm": 1.9018949270248413, + "learning_rate": 1e-06, + "loss": 0.5733, + "mean_token_accuracy": 0.8065972924232483, + "num_tokens": 216617749.0, + "step": 10858 + }, + { + "epoch": 0.89543992743465, + "grad_norm": 2.2789316177368164, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7378286123275757, + "num_tokens": 216641701.0, + "step": 10859 + }, + { + "epoch": 0.8955223880597015, + "grad_norm": 1.885461688041687, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.761677622795105, + "num_tokens": 216671602.0, + "step": 10860 + }, + { + "epoch": 0.895604848684753, + "grad_norm": 2.058946132659912, + "learning_rate": 1e-06, + "loss": 0.7738, + "mean_token_accuracy": 0.7565009593963623, + "num_tokens": 216699154.0, + "step": 10861 + }, + { + "epoch": 0.8956873093098046, + "grad_norm": 1.752548336982727, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.764230489730835, + "num_tokens": 216736155.0, + "step": 10862 + }, + { + "epoch": 0.8957697699348561, + "grad_norm": 1.9020334482192993, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7303687334060669, + "num_tokens": 216769664.0, + "step": 10863 + }, + { + "epoch": 0.8958522305599076, + "grad_norm": 2.0784084796905518, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7777252197265625, + "num_tokens": 216796899.0, + "step": 10864 + }, + { + "epoch": 0.8959346911849592, + "grad_norm": 1.8175818920135498, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7267944812774658, + "num_tokens": 216827812.0, + "step": 10865 + }, + { + "epoch": 0.8960171518100107, + "grad_norm": 1.973321557044983, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7781553268432617, + "num_tokens": 216854699.0, + "step": 10866 + }, + { + "epoch": 0.8960996124350623, + "grad_norm": 1.9379589557647705, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7536686658859253, + "num_tokens": 216882264.0, + "step": 10867 + }, + { + "epoch": 0.8961820730601138, + "grad_norm": 1.712281346321106, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.8005900382995605, + "num_tokens": 216915142.0, + "step": 10868 + }, + { + "epoch": 0.8962645336851653, + "grad_norm": 1.9260058403015137, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7647262215614319, + "num_tokens": 216945452.0, + "step": 10869 + }, + { + "epoch": 0.8963469943102169, + "grad_norm": 2.0054166316986084, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7408896088600159, + "num_tokens": 216972409.0, + "step": 10870 + }, + { + "epoch": 0.8964294549352684, + "grad_norm": 1.7746868133544922, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.780056357383728, + "num_tokens": 217002490.0, + "step": 10871 + }, + { + "epoch": 0.89651191556032, + "grad_norm": 2.0495946407318115, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7349381446838379, + "num_tokens": 217032430.0, + "step": 10872 + }, + { + "epoch": 0.8965943761853715, + "grad_norm": 1.7761985063552856, + "learning_rate": 1e-06, + "loss": 0.7746, + "mean_token_accuracy": 0.7517292499542236, + "num_tokens": 217068483.0, + "step": 10873 + }, + { + "epoch": 0.896676836810423, + "grad_norm": 1.9250210523605347, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7603703737258911, + "num_tokens": 217097975.0, + "step": 10874 + }, + { + "epoch": 0.8967592974354746, + "grad_norm": 1.7907705307006836, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7650290727615356, + "num_tokens": 217132071.0, + "step": 10875 + }, + { + "epoch": 0.8968417580605261, + "grad_norm": 1.5941200256347656, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7736762762069702, + "num_tokens": 217173207.0, + "step": 10876 + }, + { + "epoch": 0.8969242186855776, + "grad_norm": 1.7800493240356445, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7643486261367798, + "num_tokens": 217207898.0, + "step": 10877 + }, + { + "epoch": 0.8970066793106292, + "grad_norm": 1.7601850032806396, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7521716952323914, + "num_tokens": 217242773.0, + "step": 10878 + }, + { + "epoch": 0.8970891399356807, + "grad_norm": 2.326824188232422, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7221587300300598, + "num_tokens": 217265982.0, + "step": 10879 + }, + { + "epoch": 0.8971716005607322, + "grad_norm": 1.8115323781967163, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7570526003837585, + "num_tokens": 217303758.0, + "step": 10880 + }, + { + "epoch": 0.8972540611857838, + "grad_norm": 1.9895163774490356, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7402362823486328, + "num_tokens": 217332093.0, + "step": 10881 + }, + { + "epoch": 0.8973365218108353, + "grad_norm": 2.1072492599487305, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7321463823318481, + "num_tokens": 217358647.0, + "step": 10882 + }, + { + "epoch": 0.8974189824358869, + "grad_norm": 1.6913648843765259, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7687869668006897, + "num_tokens": 217393832.0, + "step": 10883 + }, + { + "epoch": 0.8975014430609384, + "grad_norm": 1.9334187507629395, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7318557500839233, + "num_tokens": 217424585.0, + "step": 10884 + }, + { + "epoch": 0.89758390368599, + "grad_norm": 1.803848385810852, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7478063106536865, + "num_tokens": 217455921.0, + "step": 10885 + }, + { + "epoch": 0.8976663643110415, + "grad_norm": 1.7341643571853638, + "learning_rate": 1e-06, + "loss": 0.6025, + "mean_token_accuracy": 0.8035340309143066, + "num_tokens": 217486538.0, + "step": 10886 + }, + { + "epoch": 0.897748824936093, + "grad_norm": 1.8384734392166138, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.7799042463302612, + "num_tokens": 217519183.0, + "step": 10887 + }, + { + "epoch": 0.8978312855611446, + "grad_norm": 1.8279789686203003, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7661114931106567, + "num_tokens": 217552179.0, + "step": 10888 + }, + { + "epoch": 0.8979137461861961, + "grad_norm": 1.849531650543213, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.7909334897994995, + "num_tokens": 217580946.0, + "step": 10889 + }, + { + "epoch": 0.8979962068112476, + "grad_norm": 1.8740320205688477, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7737175226211548, + "num_tokens": 217610095.0, + "step": 10890 + }, + { + "epoch": 0.8980786674362992, + "grad_norm": 2.147826910018921, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7514528036117554, + "num_tokens": 217635710.0, + "step": 10891 + }, + { + "epoch": 0.8981611280613507, + "grad_norm": 2.0195255279541016, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7277052402496338, + "num_tokens": 217663838.0, + "step": 10892 + }, + { + "epoch": 0.8982435886864022, + "grad_norm": 1.864830732345581, + "learning_rate": 1e-06, + "loss": 0.7702, + "mean_token_accuracy": 0.7414243221282959, + "num_tokens": 217694257.0, + "step": 10893 + }, + { + "epoch": 0.8983260493114538, + "grad_norm": 1.8797595500946045, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7193245887756348, + "num_tokens": 217724551.0, + "step": 10894 + }, + { + "epoch": 0.8984085099365053, + "grad_norm": 1.745664358139038, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7701305150985718, + "num_tokens": 217759293.0, + "step": 10895 + }, + { + "epoch": 0.8984909705615569, + "grad_norm": 1.9839818477630615, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7280324697494507, + "num_tokens": 217788664.0, + "step": 10896 + }, + { + "epoch": 0.8985734311866084, + "grad_norm": 1.9704248905181885, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7227814197540283, + "num_tokens": 217817925.0, + "step": 10897 + }, + { + "epoch": 0.89865589181166, + "grad_norm": 2.21431565284729, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7512556314468384, + "num_tokens": 217843531.0, + "step": 10898 + }, + { + "epoch": 0.8987383524367115, + "grad_norm": 2.0333991050720215, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7610058784484863, + "num_tokens": 217869493.0, + "step": 10899 + }, + { + "epoch": 0.898820813061763, + "grad_norm": 1.8258005380630493, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.754203200340271, + "num_tokens": 217900958.0, + "step": 10900 + }, + { + "epoch": 0.8989032736868146, + "grad_norm": 2.139840602874756, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7531725168228149, + "num_tokens": 217924917.0, + "step": 10901 + }, + { + "epoch": 0.8989857343118661, + "grad_norm": 2.6355369091033936, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7427719235420227, + "num_tokens": 217945790.0, + "step": 10902 + }, + { + "epoch": 0.8990681949369176, + "grad_norm": 1.8174444437026978, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7660235166549683, + "num_tokens": 217979229.0, + "step": 10903 + }, + { + "epoch": 0.8991506555619692, + "grad_norm": 2.2551615238189697, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7566115260124207, + "num_tokens": 218009779.0, + "step": 10904 + }, + { + "epoch": 0.8992331161870207, + "grad_norm": 2.100339651107788, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7248990535736084, + "num_tokens": 218036759.0, + "step": 10905 + }, + { + "epoch": 0.8993155768120722, + "grad_norm": 1.9607826471328735, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7851547002792358, + "num_tokens": 218063627.0, + "step": 10906 + }, + { + "epoch": 0.8993980374371238, + "grad_norm": 2.0919058322906494, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7819889783859253, + "num_tokens": 218087112.0, + "step": 10907 + }, + { + "epoch": 0.8994804980621753, + "grad_norm": 1.828705906867981, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7503955364227295, + "num_tokens": 218117418.0, + "step": 10908 + }, + { + "epoch": 0.8995629586872268, + "grad_norm": 2.005559206008911, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7366676330566406, + "num_tokens": 218147417.0, + "step": 10909 + }, + { + "epoch": 0.8996454193122784, + "grad_norm": 1.9825059175491333, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7449654936790466, + "num_tokens": 218176216.0, + "step": 10910 + }, + { + "epoch": 0.8997278799373299, + "grad_norm": 1.7343270778656006, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7747336626052856, + "num_tokens": 218210990.0, + "step": 10911 + }, + { + "epoch": 0.8998103405623815, + "grad_norm": 2.2773430347442627, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7619411945343018, + "num_tokens": 218233219.0, + "step": 10912 + }, + { + "epoch": 0.899892801187433, + "grad_norm": 1.9645054340362549, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7406045198440552, + "num_tokens": 218264732.0, + "step": 10913 + }, + { + "epoch": 0.8999752618124846, + "grad_norm": 2.1507012844085693, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7721196413040161, + "num_tokens": 218289971.0, + "step": 10914 + }, + { + "epoch": 0.9000577224375361, + "grad_norm": 1.99701726436615, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7505964040756226, + "num_tokens": 218320014.0, + "step": 10915 + }, + { + "epoch": 0.9001401830625876, + "grad_norm": 1.9128364324569702, + "learning_rate": 1e-06, + "loss": 0.7584, + "mean_token_accuracy": 0.7536612153053284, + "num_tokens": 218350534.0, + "step": 10916 + }, + { + "epoch": 0.9002226436876392, + "grad_norm": 1.7923954725265503, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7147353887557983, + "num_tokens": 218388062.0, + "step": 10917 + }, + { + "epoch": 0.9003051043126907, + "grad_norm": 2.3358163833618164, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7319570779800415, + "num_tokens": 218410321.0, + "step": 10918 + }, + { + "epoch": 0.9003875649377422, + "grad_norm": 1.8843578100204468, + "learning_rate": 1e-06, + "loss": 0.765, + "mean_token_accuracy": 0.7631827592849731, + "num_tokens": 218441039.0, + "step": 10919 + }, + { + "epoch": 0.9004700255627938, + "grad_norm": 1.9720244407653809, + "learning_rate": 1e-06, + "loss": 0.5869, + "mean_token_accuracy": 0.8093125820159912, + "num_tokens": 218468843.0, + "step": 10920 + }, + { + "epoch": 0.9005524861878453, + "grad_norm": 2.017287254333496, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7631164789199829, + "num_tokens": 218493905.0, + "step": 10921 + }, + { + "epoch": 0.9006349468128968, + "grad_norm": 1.773423433303833, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7407642006874084, + "num_tokens": 218527808.0, + "step": 10922 + }, + { + "epoch": 0.9007174074379484, + "grad_norm": 1.9713902473449707, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7420699596405029, + "num_tokens": 218555745.0, + "step": 10923 + }, + { + "epoch": 0.9007998680629999, + "grad_norm": 1.7135086059570312, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7484070062637329, + "num_tokens": 218593105.0, + "step": 10924 + }, + { + "epoch": 0.9008823286880514, + "grad_norm": 2.126711368560791, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7392836213111877, + "num_tokens": 218620902.0, + "step": 10925 + }, + { + "epoch": 0.900964789313103, + "grad_norm": 1.8271528482437134, + "learning_rate": 1e-06, + "loss": 0.6096, + "mean_token_accuracy": 0.7971468567848206, + "num_tokens": 218649136.0, + "step": 10926 + }, + { + "epoch": 0.9010472499381545, + "grad_norm": 1.9738003015518188, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7366378903388977, + "num_tokens": 218676540.0, + "step": 10927 + }, + { + "epoch": 0.9011297105632061, + "grad_norm": 1.9901049137115479, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7603472471237183, + "num_tokens": 218703588.0, + "step": 10928 + }, + { + "epoch": 0.9012121711882576, + "grad_norm": 2.1514387130737305, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7408699989318848, + "num_tokens": 218727910.0, + "step": 10929 + }, + { + "epoch": 0.9012946318133092, + "grad_norm": 1.8159739971160889, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6933101415634155, + "num_tokens": 218766080.0, + "step": 10930 + }, + { + "epoch": 0.9013770924383607, + "grad_norm": 2.02608323097229, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7640860676765442, + "num_tokens": 218791885.0, + "step": 10931 + }, + { + "epoch": 0.9014595530634122, + "grad_norm": 2.225048065185547, + "learning_rate": 1e-06, + "loss": 1.0648, + "mean_token_accuracy": 0.6809547543525696, + "num_tokens": 218818262.0, + "step": 10932 + }, + { + "epoch": 0.9015420136884638, + "grad_norm": 1.990959644317627, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7193448543548584, + "num_tokens": 218846612.0, + "step": 10933 + }, + { + "epoch": 0.9016244743135153, + "grad_norm": 1.6534820795059204, + "learning_rate": 1e-06, + "loss": 0.6192, + "mean_token_accuracy": 0.7944631576538086, + "num_tokens": 218881466.0, + "step": 10934 + }, + { + "epoch": 0.9017069349385668, + "grad_norm": 1.8451420068740845, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7414397597312927, + "num_tokens": 218913614.0, + "step": 10935 + }, + { + "epoch": 0.9017893955636184, + "grad_norm": 1.8712457418441772, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7564406394958496, + "num_tokens": 218945590.0, + "step": 10936 + }, + { + "epoch": 0.9018718561886699, + "grad_norm": 1.6942377090454102, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7672076225280762, + "num_tokens": 218978610.0, + "step": 10937 + }, + { + "epoch": 0.9019543168137214, + "grad_norm": 2.2617452144622803, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.726718544960022, + "num_tokens": 219003135.0, + "step": 10938 + }, + { + "epoch": 0.902036777438773, + "grad_norm": 2.137603759765625, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7626201510429382, + "num_tokens": 219026979.0, + "step": 10939 + }, + { + "epoch": 0.9021192380638245, + "grad_norm": 2.0328972339630127, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7547653913497925, + "num_tokens": 219054044.0, + "step": 10940 + }, + { + "epoch": 0.902201698688876, + "grad_norm": 1.9252880811691284, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7471469640731812, + "num_tokens": 219084469.0, + "step": 10941 + }, + { + "epoch": 0.9022841593139276, + "grad_norm": 1.8537458181381226, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7218576669692993, + "num_tokens": 219117837.0, + "step": 10942 + }, + { + "epoch": 0.9023666199389792, + "grad_norm": 1.8308676481246948, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7917808890342712, + "num_tokens": 219149941.0, + "step": 10943 + }, + { + "epoch": 0.9024490805640307, + "grad_norm": 1.957897663116455, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7808512449264526, + "num_tokens": 219176317.0, + "step": 10944 + }, + { + "epoch": 0.9025315411890822, + "grad_norm": 2.0713605880737305, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7059910297393799, + "num_tokens": 219202658.0, + "step": 10945 + }, + { + "epoch": 0.9026140018141338, + "grad_norm": 2.1437692642211914, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7539991140365601, + "num_tokens": 219227672.0, + "step": 10946 + }, + { + "epoch": 0.9026964624391853, + "grad_norm": 2.039597272872925, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.8024775981903076, + "num_tokens": 219253170.0, + "step": 10947 + }, + { + "epoch": 0.9027789230642368, + "grad_norm": 2.0211830139160156, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7475537061691284, + "num_tokens": 219280690.0, + "step": 10948 + }, + { + "epoch": 0.9028613836892884, + "grad_norm": 1.8698633909225464, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7629799842834473, + "num_tokens": 219312072.0, + "step": 10949 + }, + { + "epoch": 0.9029438443143399, + "grad_norm": 2.31567645072937, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7505391836166382, + "num_tokens": 219336149.0, + "step": 10950 + }, + { + "epoch": 0.9030263049393914, + "grad_norm": 1.9222266674041748, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.7845975160598755, + "num_tokens": 219362546.0, + "step": 10951 + }, + { + "epoch": 0.903108765564443, + "grad_norm": 1.775974988937378, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7495912909507751, + "num_tokens": 219400881.0, + "step": 10952 + }, + { + "epoch": 0.9031912261894945, + "grad_norm": 1.9383912086486816, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7585526704788208, + "num_tokens": 219431045.0, + "step": 10953 + }, + { + "epoch": 0.903273686814546, + "grad_norm": 2.0280392169952393, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7461444139480591, + "num_tokens": 219459897.0, + "step": 10954 + }, + { + "epoch": 0.9033561474395976, + "grad_norm": 1.8231287002563477, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7870557308197021, + "num_tokens": 219495704.0, + "step": 10955 + }, + { + "epoch": 0.9034386080646492, + "grad_norm": 1.8593695163726807, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7560915350914001, + "num_tokens": 219527311.0, + "step": 10956 + }, + { + "epoch": 0.9035210686897006, + "grad_norm": 1.7755470275878906, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7548564672470093, + "num_tokens": 219562589.0, + "step": 10957 + }, + { + "epoch": 0.9036035293147522, + "grad_norm": 1.800072193145752, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7786468863487244, + "num_tokens": 219594414.0, + "step": 10958 + }, + { + "epoch": 0.9036859899398038, + "grad_norm": 1.7449257373809814, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.7959260940551758, + "num_tokens": 219627883.0, + "step": 10959 + }, + { + "epoch": 0.9037684505648553, + "grad_norm": 2.190225839614868, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7240256667137146, + "num_tokens": 219651905.0, + "step": 10960 + }, + { + "epoch": 0.9038509111899068, + "grad_norm": 1.6825177669525146, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7706311941146851, + "num_tokens": 219689095.0, + "step": 10961 + }, + { + "epoch": 0.9039333718149584, + "grad_norm": 1.9413033723831177, + "learning_rate": 1e-06, + "loss": 0.5903, + "mean_token_accuracy": 0.7999272346496582, + "num_tokens": 219716531.0, + "step": 10962 + }, + { + "epoch": 0.9040158324400099, + "grad_norm": 2.108663320541382, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6988505125045776, + "num_tokens": 219745069.0, + "step": 10963 + }, + { + "epoch": 0.9040982930650614, + "grad_norm": 1.8691866397857666, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7710838913917542, + "num_tokens": 219773398.0, + "step": 10964 + }, + { + "epoch": 0.904180753690113, + "grad_norm": 2.340310573577881, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7253062725067139, + "num_tokens": 219797025.0, + "step": 10965 + }, + { + "epoch": 0.9042632143151645, + "grad_norm": 2.055807113647461, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7456727027893066, + "num_tokens": 219825422.0, + "step": 10966 + }, + { + "epoch": 0.904345674940216, + "grad_norm": 1.9964083433151245, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7465101480484009, + "num_tokens": 219855106.0, + "step": 10967 + }, + { + "epoch": 0.9044281355652676, + "grad_norm": 1.9809528589248657, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7707434892654419, + "num_tokens": 219882018.0, + "step": 10968 + }, + { + "epoch": 0.9045105961903191, + "grad_norm": 2.012479543685913, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.739376962184906, + "num_tokens": 219912339.0, + "step": 10969 + }, + { + "epoch": 0.9045930568153706, + "grad_norm": 1.7550041675567627, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.7911750674247742, + "num_tokens": 219945107.0, + "step": 10970 + }, + { + "epoch": 0.9046755174404222, + "grad_norm": 2.1414847373962402, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.736518144607544, + "num_tokens": 219971330.0, + "step": 10971 + }, + { + "epoch": 0.9047579780654738, + "grad_norm": 2.0331339836120605, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7660121917724609, + "num_tokens": 219998115.0, + "step": 10972 + }, + { + "epoch": 0.9048404386905253, + "grad_norm": 1.7940685749053955, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7622660398483276, + "num_tokens": 220030996.0, + "step": 10973 + }, + { + "epoch": 0.9049228993155768, + "grad_norm": 1.8907676935195923, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7698672413825989, + "num_tokens": 220059575.0, + "step": 10974 + }, + { + "epoch": 0.9050053599406284, + "grad_norm": 2.0793886184692383, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7616055011749268, + "num_tokens": 220083708.0, + "step": 10975 + }, + { + "epoch": 0.9050878205656799, + "grad_norm": 1.9740924835205078, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7590447664260864, + "num_tokens": 220110352.0, + "step": 10976 + }, + { + "epoch": 0.9051702811907314, + "grad_norm": 1.8942605257034302, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.711665153503418, + "num_tokens": 220142931.0, + "step": 10977 + }, + { + "epoch": 0.905252741815783, + "grad_norm": 2.074284553527832, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7731777429580688, + "num_tokens": 220169387.0, + "step": 10978 + }, + { + "epoch": 0.9053352024408345, + "grad_norm": 1.8707084655761719, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7542086243629456, + "num_tokens": 220202797.0, + "step": 10979 + }, + { + "epoch": 0.905417663065886, + "grad_norm": 1.8448677062988281, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7408015131950378, + "num_tokens": 220234857.0, + "step": 10980 + }, + { + "epoch": 0.9055001236909376, + "grad_norm": 1.9191550016403198, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7310627698898315, + "num_tokens": 220265175.0, + "step": 10981 + }, + { + "epoch": 0.9055825843159891, + "grad_norm": 2.020158529281616, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7356464266777039, + "num_tokens": 220293170.0, + "step": 10982 + }, + { + "epoch": 0.9056650449410406, + "grad_norm": 1.9606748819351196, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.7919896841049194, + "num_tokens": 220320405.0, + "step": 10983 + }, + { + "epoch": 0.9057475055660922, + "grad_norm": 1.8869593143463135, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7438004016876221, + "num_tokens": 220348294.0, + "step": 10984 + }, + { + "epoch": 0.9058299661911438, + "grad_norm": 1.7940962314605713, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7854374647140503, + "num_tokens": 220381034.0, + "step": 10985 + }, + { + "epoch": 0.9059124268161952, + "grad_norm": 1.8825578689575195, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7656375765800476, + "num_tokens": 220410693.0, + "step": 10986 + }, + { + "epoch": 0.9059948874412468, + "grad_norm": 1.7195556163787842, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7789188027381897, + "num_tokens": 220447481.0, + "step": 10987 + }, + { + "epoch": 0.9060773480662984, + "grad_norm": 2.05336856842041, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7335139513015747, + "num_tokens": 220474181.0, + "step": 10988 + }, + { + "epoch": 0.9061598086913499, + "grad_norm": 1.931876301765442, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7683656215667725, + "num_tokens": 220502923.0, + "step": 10989 + }, + { + "epoch": 0.9062422693164014, + "grad_norm": 1.8028749227523804, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7534287571907043, + "num_tokens": 220536711.0, + "step": 10990 + }, + { + "epoch": 0.906324729941453, + "grad_norm": 2.157169818878174, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7480984926223755, + "num_tokens": 220563237.0, + "step": 10991 + }, + { + "epoch": 0.9064071905665045, + "grad_norm": 1.6403837203979492, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7450710535049438, + "num_tokens": 220600028.0, + "step": 10992 + }, + { + "epoch": 0.906489651191556, + "grad_norm": 1.9090003967285156, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.74550461769104, + "num_tokens": 220631382.0, + "step": 10993 + }, + { + "epoch": 0.9065721118166076, + "grad_norm": 2.038625478744507, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7471088171005249, + "num_tokens": 220659092.0, + "step": 10994 + }, + { + "epoch": 0.9066545724416591, + "grad_norm": 1.944282054901123, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7513746619224548, + "num_tokens": 220692468.0, + "step": 10995 + }, + { + "epoch": 0.9067370330667106, + "grad_norm": 1.7456364631652832, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.7842265367507935, + "num_tokens": 220724776.0, + "step": 10996 + }, + { + "epoch": 0.9068194936917622, + "grad_norm": 1.9650439023971558, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.771642804145813, + "num_tokens": 220755174.0, + "step": 10997 + }, + { + "epoch": 0.9069019543168138, + "grad_norm": 1.8764588832855225, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7605149745941162, + "num_tokens": 220786685.0, + "step": 10998 + }, + { + "epoch": 0.9069844149418652, + "grad_norm": 2.3083813190460205, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7320161461830139, + "num_tokens": 220808544.0, + "step": 10999 + }, + { + "epoch": 0.9070668755669168, + "grad_norm": 1.9691413640975952, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.725282609462738, + "num_tokens": 220839730.0, + "step": 11000 + }, + { + "epoch": 0.9071493361919684, + "grad_norm": 1.9372540712356567, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7669885158538818, + "num_tokens": 220867432.0, + "step": 11001 + }, + { + "epoch": 0.9072317968170198, + "grad_norm": 1.7651827335357666, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7671607136726379, + "num_tokens": 220899552.0, + "step": 11002 + }, + { + "epoch": 0.9073142574420714, + "grad_norm": 2.102573871612549, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.755200207233429, + "num_tokens": 220926560.0, + "step": 11003 + }, + { + "epoch": 0.907396718067123, + "grad_norm": 2.3236453533172607, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7352960109710693, + "num_tokens": 220950427.0, + "step": 11004 + }, + { + "epoch": 0.9074791786921745, + "grad_norm": 1.958789587020874, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7813886404037476, + "num_tokens": 220977755.0, + "step": 11005 + }, + { + "epoch": 0.907561639317226, + "grad_norm": 1.9186594486236572, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7597342729568481, + "num_tokens": 221006664.0, + "step": 11006 + }, + { + "epoch": 0.9076440999422776, + "grad_norm": 1.9980740547180176, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.7007777094841003, + "num_tokens": 221036207.0, + "step": 11007 + }, + { + "epoch": 0.9077265605673291, + "grad_norm": 2.513498544692993, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7319344878196716, + "num_tokens": 221056191.0, + "step": 11008 + }, + { + "epoch": 0.9078090211923806, + "grad_norm": 1.9434877634048462, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7469605207443237, + "num_tokens": 221087537.0, + "step": 11009 + }, + { + "epoch": 0.9078914818174322, + "grad_norm": 1.9093234539031982, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7488486766815186, + "num_tokens": 221118680.0, + "step": 11010 + }, + { + "epoch": 0.9079739424424838, + "grad_norm": 1.9591255187988281, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6892539262771606, + "num_tokens": 221150075.0, + "step": 11011 + }, + { + "epoch": 0.9080564030675352, + "grad_norm": 1.6776036024093628, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7762027382850647, + "num_tokens": 221188236.0, + "step": 11012 + }, + { + "epoch": 0.9081388636925868, + "grad_norm": 2.0284080505371094, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7613207697868347, + "num_tokens": 221215367.0, + "step": 11013 + }, + { + "epoch": 0.9082213243176384, + "grad_norm": 1.7759724855422974, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.764041006565094, + "num_tokens": 221249595.0, + "step": 11014 + }, + { + "epoch": 0.9083037849426898, + "grad_norm": 1.982828140258789, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.724165678024292, + "num_tokens": 221278625.0, + "step": 11015 + }, + { + "epoch": 0.9083862455677414, + "grad_norm": 1.7914966344833374, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7275478839874268, + "num_tokens": 221312491.0, + "step": 11016 + }, + { + "epoch": 0.908468706192793, + "grad_norm": 1.9294638633728027, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7408835887908936, + "num_tokens": 221345123.0, + "step": 11017 + }, + { + "epoch": 0.9085511668178444, + "grad_norm": 1.8169469833374023, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7319198250770569, + "num_tokens": 221380469.0, + "step": 11018 + }, + { + "epoch": 0.908633627442896, + "grad_norm": 2.034223794937134, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7834138870239258, + "num_tokens": 221406972.0, + "step": 11019 + }, + { + "epoch": 0.9087160880679476, + "grad_norm": 1.9701298475265503, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7679970264434814, + "num_tokens": 221436121.0, + "step": 11020 + }, + { + "epoch": 0.9087985486929991, + "grad_norm": 2.10616135597229, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7371541261672974, + "num_tokens": 221463639.0, + "step": 11021 + }, + { + "epoch": 0.9088810093180506, + "grad_norm": 2.099792957305908, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7236262559890747, + "num_tokens": 221492539.0, + "step": 11022 + }, + { + "epoch": 0.9089634699431022, + "grad_norm": 2.174844264984131, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.746606707572937, + "num_tokens": 221517639.0, + "step": 11023 + }, + { + "epoch": 0.9090459305681537, + "grad_norm": 1.6787923574447632, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7858034372329712, + "num_tokens": 221555831.0, + "step": 11024 + }, + { + "epoch": 0.9091283911932052, + "grad_norm": 1.9612996578216553, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7661930322647095, + "num_tokens": 221584454.0, + "step": 11025 + }, + { + "epoch": 0.9092108518182568, + "grad_norm": 2.1669669151306152, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7433743476867676, + "num_tokens": 221609366.0, + "step": 11026 + }, + { + "epoch": 0.9092933124433084, + "grad_norm": 1.852588415145874, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7542940974235535, + "num_tokens": 221642417.0, + "step": 11027 + }, + { + "epoch": 0.9093757730683598, + "grad_norm": 2.114819049835205, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.688579797744751, + "num_tokens": 221673971.0, + "step": 11028 + }, + { + "epoch": 0.9094582336934114, + "grad_norm": 1.9164172410964966, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7483659982681274, + "num_tokens": 221703848.0, + "step": 11029 + }, + { + "epoch": 0.909540694318463, + "grad_norm": 1.991563320159912, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.7506691217422485, + "num_tokens": 221730952.0, + "step": 11030 + }, + { + "epoch": 0.9096231549435144, + "grad_norm": 1.9799083471298218, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7467294931411743, + "num_tokens": 221757013.0, + "step": 11031 + }, + { + "epoch": 0.909705615568566, + "grad_norm": 2.2462854385375977, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7503935694694519, + "num_tokens": 221780975.0, + "step": 11032 + }, + { + "epoch": 0.9097880761936176, + "grad_norm": 1.938713788986206, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7831478714942932, + "num_tokens": 221809684.0, + "step": 11033 + }, + { + "epoch": 0.9098705368186691, + "grad_norm": 2.1384332180023193, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.712052583694458, + "num_tokens": 221835176.0, + "step": 11034 + }, + { + "epoch": 0.9099529974437206, + "grad_norm": 1.8832296133041382, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7560117244720459, + "num_tokens": 221866804.0, + "step": 11035 + }, + { + "epoch": 0.9100354580687722, + "grad_norm": 1.884098768234253, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7190642356872559, + "num_tokens": 221897406.0, + "step": 11036 + }, + { + "epoch": 0.9101179186938237, + "grad_norm": 1.9043995141983032, + "learning_rate": 1e-06, + "loss": 0.7774, + "mean_token_accuracy": 0.7527620792388916, + "num_tokens": 221927263.0, + "step": 11037 + }, + { + "epoch": 0.9102003793188752, + "grad_norm": 2.0162553787231445, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.77053302526474, + "num_tokens": 221952697.0, + "step": 11038 + }, + { + "epoch": 0.9102828399439268, + "grad_norm": 1.9392133951187134, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7409243583679199, + "num_tokens": 221983572.0, + "step": 11039 + }, + { + "epoch": 0.9103653005689784, + "grad_norm": 1.6224006414413452, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7801584601402283, + "num_tokens": 222019295.0, + "step": 11040 + }, + { + "epoch": 0.9104477611940298, + "grad_norm": 1.9715020656585693, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.7907307147979736, + "num_tokens": 222046201.0, + "step": 11041 + }, + { + "epoch": 0.9105302218190814, + "grad_norm": 1.8772584199905396, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.770658552646637, + "num_tokens": 222077303.0, + "step": 11042 + }, + { + "epoch": 0.910612682444133, + "grad_norm": 1.9760432243347168, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7515565752983093, + "num_tokens": 222108480.0, + "step": 11043 + }, + { + "epoch": 0.9106951430691844, + "grad_norm": 1.9522110223770142, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7569401264190674, + "num_tokens": 222138967.0, + "step": 11044 + }, + { + "epoch": 0.910777603694236, + "grad_norm": 1.8777037858963013, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7608263492584229, + "num_tokens": 222168944.0, + "step": 11045 + }, + { + "epoch": 0.9108600643192876, + "grad_norm": 2.1947367191314697, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7185999751091003, + "num_tokens": 222193522.0, + "step": 11046 + }, + { + "epoch": 0.910942524944339, + "grad_norm": 1.9371472597122192, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7485253214836121, + "num_tokens": 222223344.0, + "step": 11047 + }, + { + "epoch": 0.9110249855693906, + "grad_norm": 2.0847091674804688, + "learning_rate": 1e-06, + "loss": 0.7883, + "mean_token_accuracy": 0.7507102489471436, + "num_tokens": 222248716.0, + "step": 11048 + }, + { + "epoch": 0.9111074461944422, + "grad_norm": 1.8038827180862427, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7393338680267334, + "num_tokens": 222282865.0, + "step": 11049 + }, + { + "epoch": 0.9111899068194937, + "grad_norm": 1.8943661451339722, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7747454643249512, + "num_tokens": 222311347.0, + "step": 11050 + }, + { + "epoch": 0.9112723674445452, + "grad_norm": 1.6376477479934692, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.7858893871307373, + "num_tokens": 222347923.0, + "step": 11051 + }, + { + "epoch": 0.9113548280695968, + "grad_norm": 2.1403942108154297, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7138112187385559, + "num_tokens": 222374114.0, + "step": 11052 + }, + { + "epoch": 0.9114372886946484, + "grad_norm": 1.773640513420105, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7684709429740906, + "num_tokens": 222412112.0, + "step": 11053 + }, + { + "epoch": 0.9115197493196998, + "grad_norm": 1.6914551258087158, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7391148805618286, + "num_tokens": 222448965.0, + "step": 11054 + }, + { + "epoch": 0.9116022099447514, + "grad_norm": 1.961247205734253, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7283177375793457, + "num_tokens": 222477725.0, + "step": 11055 + }, + { + "epoch": 0.911684670569803, + "grad_norm": 1.9118382930755615, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7656298875808716, + "num_tokens": 222508131.0, + "step": 11056 + }, + { + "epoch": 0.9117671311948544, + "grad_norm": 1.7835447788238525, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7267599105834961, + "num_tokens": 222546023.0, + "step": 11057 + }, + { + "epoch": 0.911849591819906, + "grad_norm": 2.0471513271331787, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7574746608734131, + "num_tokens": 222573314.0, + "step": 11058 + }, + { + "epoch": 0.9119320524449576, + "grad_norm": 2.0783395767211914, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7375288605690002, + "num_tokens": 222602407.0, + "step": 11059 + }, + { + "epoch": 0.912014513070009, + "grad_norm": 1.7242459058761597, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.756404459476471, + "num_tokens": 222638653.0, + "step": 11060 + }, + { + "epoch": 0.9120969736950606, + "grad_norm": 1.7454302310943604, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.7881789207458496, + "num_tokens": 222670031.0, + "step": 11061 + }, + { + "epoch": 0.9121794343201122, + "grad_norm": 1.9120968580245972, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7227692008018494, + "num_tokens": 222703289.0, + "step": 11062 + }, + { + "epoch": 0.9122618949451636, + "grad_norm": 2.0784246921539307, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7358771562576294, + "num_tokens": 222731889.0, + "step": 11063 + }, + { + "epoch": 0.9123443555702152, + "grad_norm": 1.7810676097869873, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7467048764228821, + "num_tokens": 222764812.0, + "step": 11064 + }, + { + "epoch": 0.9124268161952668, + "grad_norm": 1.961181640625, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7549575567245483, + "num_tokens": 222793263.0, + "step": 11065 + }, + { + "epoch": 0.9125092768203183, + "grad_norm": 2.2387771606445312, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.740723729133606, + "num_tokens": 222815863.0, + "step": 11066 + }, + { + "epoch": 0.9125917374453698, + "grad_norm": 1.9282118082046509, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7570307850837708, + "num_tokens": 222846104.0, + "step": 11067 + }, + { + "epoch": 0.9126741980704214, + "grad_norm": 1.9659427404403687, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7592995166778564, + "num_tokens": 222873927.0, + "step": 11068 + }, + { + "epoch": 0.912756658695473, + "grad_norm": 1.628791332244873, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7467784881591797, + "num_tokens": 222912568.0, + "step": 11069 + }, + { + "epoch": 0.9128391193205244, + "grad_norm": 1.8959531784057617, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7581708431243896, + "num_tokens": 222942289.0, + "step": 11070 + }, + { + "epoch": 0.912921579945576, + "grad_norm": 2.0762412548065186, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7254606485366821, + "num_tokens": 222968374.0, + "step": 11071 + }, + { + "epoch": 0.9130040405706276, + "grad_norm": 1.9753613471984863, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7256151437759399, + "num_tokens": 222997306.0, + "step": 11072 + }, + { + "epoch": 0.913086501195679, + "grad_norm": 1.905377745628357, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7330477237701416, + "num_tokens": 223029587.0, + "step": 11073 + }, + { + "epoch": 0.9131689618207306, + "grad_norm": 2.149460554122925, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7296867370605469, + "num_tokens": 223054552.0, + "step": 11074 + }, + { + "epoch": 0.9132514224457822, + "grad_norm": 2.2072856426239014, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7491580247879028, + "num_tokens": 223081242.0, + "step": 11075 + }, + { + "epoch": 0.9133338830708336, + "grad_norm": 1.725441813468933, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7725249528884888, + "num_tokens": 223117122.0, + "step": 11076 + }, + { + "epoch": 0.9134163436958852, + "grad_norm": 1.987913727760315, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7574038505554199, + "num_tokens": 223148464.0, + "step": 11077 + }, + { + "epoch": 0.9134988043209368, + "grad_norm": 1.8792777061462402, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7809827923774719, + "num_tokens": 223181480.0, + "step": 11078 + }, + { + "epoch": 0.9135812649459882, + "grad_norm": 2.135561466217041, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7417067885398865, + "num_tokens": 223207980.0, + "step": 11079 + }, + { + "epoch": 0.9136637255710398, + "grad_norm": 2.142333745956421, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7341489791870117, + "num_tokens": 223234233.0, + "step": 11080 + }, + { + "epoch": 0.9137461861960914, + "grad_norm": 1.8459593057632446, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7469051480293274, + "num_tokens": 223266655.0, + "step": 11081 + }, + { + "epoch": 0.913828646821143, + "grad_norm": 1.7378493547439575, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.7919683456420898, + "num_tokens": 223301660.0, + "step": 11082 + }, + { + "epoch": 0.9139111074461944, + "grad_norm": 1.5824775695800781, + "learning_rate": 1e-06, + "loss": 0.5624, + "mean_token_accuracy": 0.8174039721488953, + "num_tokens": 223337382.0, + "step": 11083 + }, + { + "epoch": 0.913993568071246, + "grad_norm": 1.8576610088348389, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7584928870201111, + "num_tokens": 223368237.0, + "step": 11084 + }, + { + "epoch": 0.9140760286962976, + "grad_norm": 2.0186069011688232, + "learning_rate": 1e-06, + "loss": 0.5795, + "mean_token_accuracy": 0.8024598956108093, + "num_tokens": 223394063.0, + "step": 11085 + }, + { + "epoch": 0.914158489321349, + "grad_norm": 2.1243515014648438, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7368403673171997, + "num_tokens": 223419051.0, + "step": 11086 + }, + { + "epoch": 0.9142409499464006, + "grad_norm": 1.9695366621017456, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7331292629241943, + "num_tokens": 223449119.0, + "step": 11087 + }, + { + "epoch": 0.9143234105714522, + "grad_norm": 2.009960889816284, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7600346207618713, + "num_tokens": 223474834.0, + "step": 11088 + }, + { + "epoch": 0.9144058711965036, + "grad_norm": 2.085577964782715, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7699958086013794, + "num_tokens": 223499395.0, + "step": 11089 + }, + { + "epoch": 0.9144883318215552, + "grad_norm": 1.8303946256637573, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7627556324005127, + "num_tokens": 223532290.0, + "step": 11090 + }, + { + "epoch": 0.9145707924466068, + "grad_norm": 2.097166061401367, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7343425750732422, + "num_tokens": 223561243.0, + "step": 11091 + }, + { + "epoch": 0.9146532530716582, + "grad_norm": 1.7242753505706787, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7790446877479553, + "num_tokens": 223597325.0, + "step": 11092 + }, + { + "epoch": 0.9147357136967098, + "grad_norm": 1.6574254035949707, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7944413423538208, + "num_tokens": 223633387.0, + "step": 11093 + }, + { + "epoch": 0.9148181743217614, + "grad_norm": 1.840144157409668, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7387100458145142, + "num_tokens": 223669037.0, + "step": 11094 + }, + { + "epoch": 0.914900634946813, + "grad_norm": 1.987540364265442, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7627261877059937, + "num_tokens": 223700025.0, + "step": 11095 + }, + { + "epoch": 0.9149830955718644, + "grad_norm": 1.7167258262634277, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7668022513389587, + "num_tokens": 223737545.0, + "step": 11096 + }, + { + "epoch": 0.915065556196916, + "grad_norm": 1.8628097772598267, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7396177649497986, + "num_tokens": 223770236.0, + "step": 11097 + }, + { + "epoch": 0.9151480168219676, + "grad_norm": 1.9356776475906372, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7489162683486938, + "num_tokens": 223797281.0, + "step": 11098 + }, + { + "epoch": 0.915230477447019, + "grad_norm": 2.191052198410034, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7445641160011292, + "num_tokens": 223821252.0, + "step": 11099 + }, + { + "epoch": 0.9153129380720706, + "grad_norm": 1.7903426885604858, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7494734525680542, + "num_tokens": 223854097.0, + "step": 11100 + }, + { + "epoch": 0.9153953986971222, + "grad_norm": 2.0603482723236084, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7578641176223755, + "num_tokens": 223880897.0, + "step": 11101 + }, + { + "epoch": 0.9154778593221736, + "grad_norm": 1.8328925371170044, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7763356566429138, + "num_tokens": 223910614.0, + "step": 11102 + }, + { + "epoch": 0.9155603199472252, + "grad_norm": 1.8336663246154785, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7504684925079346, + "num_tokens": 223945315.0, + "step": 11103 + }, + { + "epoch": 0.9156427805722768, + "grad_norm": 2.215479612350464, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7542122006416321, + "num_tokens": 223972015.0, + "step": 11104 + }, + { + "epoch": 0.9157252411973282, + "grad_norm": 1.7820144891738892, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7400321960449219, + "num_tokens": 224006329.0, + "step": 11105 + }, + { + "epoch": 0.9158077018223798, + "grad_norm": 2.0619280338287354, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7517062425613403, + "num_tokens": 224032416.0, + "step": 11106 + }, + { + "epoch": 0.9158901624474314, + "grad_norm": 1.8948909044265747, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7590171694755554, + "num_tokens": 224064348.0, + "step": 11107 + }, + { + "epoch": 0.9159726230724828, + "grad_norm": 2.169762134552002, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7210118770599365, + "num_tokens": 224091788.0, + "step": 11108 + }, + { + "epoch": 0.9160550836975344, + "grad_norm": 2.076935052871704, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7197878956794739, + "num_tokens": 224119078.0, + "step": 11109 + }, + { + "epoch": 0.916137544322586, + "grad_norm": 1.8490878343582153, + "learning_rate": 1e-06, + "loss": 0.5782, + "mean_token_accuracy": 0.8037846088409424, + "num_tokens": 224150500.0, + "step": 11110 + }, + { + "epoch": 0.9162200049476376, + "grad_norm": 2.0147650241851807, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7390667200088501, + "num_tokens": 224179025.0, + "step": 11111 + }, + { + "epoch": 0.916302465572689, + "grad_norm": 1.9504985809326172, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7252813577651978, + "num_tokens": 224206718.0, + "step": 11112 + }, + { + "epoch": 0.9163849261977406, + "grad_norm": 1.9203295707702637, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7615083456039429, + "num_tokens": 224236085.0, + "step": 11113 + }, + { + "epoch": 0.9164673868227922, + "grad_norm": 1.808163046836853, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7749056220054626, + "num_tokens": 224265605.0, + "step": 11114 + }, + { + "epoch": 0.9165498474478436, + "grad_norm": 2.2064707279205322, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7695490717887878, + "num_tokens": 224288442.0, + "step": 11115 + }, + { + "epoch": 0.9166323080728952, + "grad_norm": 1.8975805044174194, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7505499720573425, + "num_tokens": 224319793.0, + "step": 11116 + }, + { + "epoch": 0.9167147686979468, + "grad_norm": 2.3069076538085938, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7454044818878174, + "num_tokens": 224343877.0, + "step": 11117 + }, + { + "epoch": 0.9167972293229982, + "grad_norm": 2.1723365783691406, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7254854440689087, + "num_tokens": 224369762.0, + "step": 11118 + }, + { + "epoch": 0.9168796899480498, + "grad_norm": 1.8550318479537964, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7391249537467957, + "num_tokens": 224401135.0, + "step": 11119 + }, + { + "epoch": 0.9169621505731014, + "grad_norm": 2.0243098735809326, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.7862400412559509, + "num_tokens": 224426888.0, + "step": 11120 + }, + { + "epoch": 0.9170446111981528, + "grad_norm": 1.880224585533142, + "learning_rate": 1e-06, + "loss": 0.7926, + "mean_token_accuracy": 0.748298168182373, + "num_tokens": 224457503.0, + "step": 11121 + }, + { + "epoch": 0.9171270718232044, + "grad_norm": 1.8881584405899048, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7355530261993408, + "num_tokens": 224487810.0, + "step": 11122 + }, + { + "epoch": 0.917209532448256, + "grad_norm": 2.1786839962005615, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7600122094154358, + "num_tokens": 224513047.0, + "step": 11123 + }, + { + "epoch": 0.9172919930733074, + "grad_norm": 1.6717545986175537, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.7822189927101135, + "num_tokens": 224548683.0, + "step": 11124 + }, + { + "epoch": 0.917374453698359, + "grad_norm": 1.8393827676773071, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7758479118347168, + "num_tokens": 224584439.0, + "step": 11125 + }, + { + "epoch": 0.9174569143234106, + "grad_norm": 2.019740343093872, + "learning_rate": 1e-06, + "loss": 0.5999, + "mean_token_accuracy": 0.792759120464325, + "num_tokens": 224608934.0, + "step": 11126 + }, + { + "epoch": 0.9175393749484622, + "grad_norm": 2.20084810256958, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7070597410202026, + "num_tokens": 224632896.0, + "step": 11127 + }, + { + "epoch": 0.9176218355735136, + "grad_norm": 1.711154818534851, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7764564752578735, + "num_tokens": 224668538.0, + "step": 11128 + }, + { + "epoch": 0.9177042961985652, + "grad_norm": 2.0579307079315186, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7716401219367981, + "num_tokens": 224694203.0, + "step": 11129 + }, + { + "epoch": 0.9177867568236168, + "grad_norm": 1.998964548110962, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7327142953872681, + "num_tokens": 224724114.0, + "step": 11130 + }, + { + "epoch": 0.9178692174486682, + "grad_norm": 1.8975590467453003, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7830748558044434, + "num_tokens": 224754492.0, + "step": 11131 + }, + { + "epoch": 0.9179516780737198, + "grad_norm": 1.9422533512115479, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.737269401550293, + "num_tokens": 224783793.0, + "step": 11132 + }, + { + "epoch": 0.9180341386987714, + "grad_norm": 1.9622830152511597, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7528396844863892, + "num_tokens": 224814828.0, + "step": 11133 + }, + { + "epoch": 0.9181165993238228, + "grad_norm": 2.3090548515319824, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7381975054740906, + "num_tokens": 224836990.0, + "step": 11134 + }, + { + "epoch": 0.9181990599488744, + "grad_norm": 1.8725193738937378, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7770374417304993, + "num_tokens": 224868645.0, + "step": 11135 + }, + { + "epoch": 0.918281520573926, + "grad_norm": 2.0318427085876465, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7438912987709045, + "num_tokens": 224897239.0, + "step": 11136 + }, + { + "epoch": 0.9183639811989774, + "grad_norm": 2.010608434677124, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7419272065162659, + "num_tokens": 224928921.0, + "step": 11137 + }, + { + "epoch": 0.918446441824029, + "grad_norm": 1.904904842376709, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7407741546630859, + "num_tokens": 224959699.0, + "step": 11138 + }, + { + "epoch": 0.9185289024490806, + "grad_norm": 2.0677709579467773, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7120634913444519, + "num_tokens": 224994563.0, + "step": 11139 + }, + { + "epoch": 0.918611363074132, + "grad_norm": 1.886349081993103, + "learning_rate": 1e-06, + "loss": 0.7877, + "mean_token_accuracy": 0.7641775608062744, + "num_tokens": 225024806.0, + "step": 11140 + }, + { + "epoch": 0.9186938236991836, + "grad_norm": 1.9140335321426392, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7566391229629517, + "num_tokens": 225053200.0, + "step": 11141 + }, + { + "epoch": 0.9187762843242352, + "grad_norm": 1.9030858278274536, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7593094110488892, + "num_tokens": 225085281.0, + "step": 11142 + }, + { + "epoch": 0.9188587449492868, + "grad_norm": 1.735337495803833, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.741997241973877, + "num_tokens": 225122061.0, + "step": 11143 + }, + { + "epoch": 0.9189412055743382, + "grad_norm": 1.939875602722168, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7397080659866333, + "num_tokens": 225150006.0, + "step": 11144 + }, + { + "epoch": 0.9190236661993898, + "grad_norm": 1.9145325422286987, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7297054529190063, + "num_tokens": 225181095.0, + "step": 11145 + }, + { + "epoch": 0.9191061268244414, + "grad_norm": 2.0161101818084717, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7680709362030029, + "num_tokens": 225208248.0, + "step": 11146 + }, + { + "epoch": 0.9191885874494928, + "grad_norm": 1.954527497291565, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7313933968544006, + "num_tokens": 225238320.0, + "step": 11147 + }, + { + "epoch": 0.9192710480745444, + "grad_norm": 2.0673508644104004, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7295800447463989, + "num_tokens": 225264583.0, + "step": 11148 + }, + { + "epoch": 0.919353508699596, + "grad_norm": 1.7877141237258911, + "learning_rate": 1e-06, + "loss": 0.6094, + "mean_token_accuracy": 0.7967556715011597, + "num_tokens": 225295551.0, + "step": 11149 + }, + { + "epoch": 0.9194359693246474, + "grad_norm": 1.8524134159088135, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7564669847488403, + "num_tokens": 225331111.0, + "step": 11150 + }, + { + "epoch": 0.919518429949699, + "grad_norm": 1.7276408672332764, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7433250546455383, + "num_tokens": 225365762.0, + "step": 11151 + }, + { + "epoch": 0.9196008905747506, + "grad_norm": 1.8426181077957153, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7537309527397156, + "num_tokens": 225397898.0, + "step": 11152 + }, + { + "epoch": 0.919683351199802, + "grad_norm": 1.777331829071045, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7575560808181763, + "num_tokens": 225429794.0, + "step": 11153 + }, + { + "epoch": 0.9197658118248536, + "grad_norm": 1.9379585981369019, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.772650420665741, + "num_tokens": 225458265.0, + "step": 11154 + }, + { + "epoch": 0.9198482724499052, + "grad_norm": 1.7513593435287476, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.726538360118866, + "num_tokens": 225495537.0, + "step": 11155 + }, + { + "epoch": 0.9199307330749568, + "grad_norm": 1.8401210308074951, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.746139645576477, + "num_tokens": 225530515.0, + "step": 11156 + }, + { + "epoch": 0.9200131937000082, + "grad_norm": 2.1174252033233643, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7247195839881897, + "num_tokens": 225556042.0, + "step": 11157 + }, + { + "epoch": 0.9200956543250598, + "grad_norm": 1.893829345703125, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7395948171615601, + "num_tokens": 225584859.0, + "step": 11158 + }, + { + "epoch": 0.9201781149501114, + "grad_norm": 1.921858787536621, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7331417798995972, + "num_tokens": 225615600.0, + "step": 11159 + }, + { + "epoch": 0.9202605755751628, + "grad_norm": 1.8340510129928589, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.75449538230896, + "num_tokens": 225646640.0, + "step": 11160 + }, + { + "epoch": 0.9203430362002144, + "grad_norm": 1.9468327760696411, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7593790888786316, + "num_tokens": 225677615.0, + "step": 11161 + }, + { + "epoch": 0.920425496825266, + "grad_norm": 2.008596658706665, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7321986556053162, + "num_tokens": 225704920.0, + "step": 11162 + }, + { + "epoch": 0.9205079574503174, + "grad_norm": 1.7785779237747192, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7776216268539429, + "num_tokens": 225739829.0, + "step": 11163 + }, + { + "epoch": 0.920590418075369, + "grad_norm": 1.8590089082717896, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7775921821594238, + "num_tokens": 225768615.0, + "step": 11164 + }, + { + "epoch": 0.9206728787004206, + "grad_norm": 2.0380144119262695, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7739025354385376, + "num_tokens": 225794921.0, + "step": 11165 + }, + { + "epoch": 0.920755339325472, + "grad_norm": 1.9206198453903198, + "learning_rate": 1e-06, + "loss": 0.6098, + "mean_token_accuracy": 0.7919897437095642, + "num_tokens": 225824647.0, + "step": 11166 + }, + { + "epoch": 0.9208377999505236, + "grad_norm": 1.7952171564102173, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7668740749359131, + "num_tokens": 225859584.0, + "step": 11167 + }, + { + "epoch": 0.9209202605755752, + "grad_norm": 1.9897127151489258, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7461336255073547, + "num_tokens": 225889547.0, + "step": 11168 + }, + { + "epoch": 0.9210027212006267, + "grad_norm": 1.8829758167266846, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.732419490814209, + "num_tokens": 225921683.0, + "step": 11169 + }, + { + "epoch": 0.9210851818256782, + "grad_norm": 2.043307065963745, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7120374441146851, + "num_tokens": 225951071.0, + "step": 11170 + }, + { + "epoch": 0.9211676424507298, + "grad_norm": 1.7367435693740845, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.758068323135376, + "num_tokens": 225987042.0, + "step": 11171 + }, + { + "epoch": 0.9212501030757814, + "grad_norm": 1.9441930055618286, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7619963884353638, + "num_tokens": 226017395.0, + "step": 11172 + }, + { + "epoch": 0.9213325637008328, + "grad_norm": 2.0882859230041504, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7351888418197632, + "num_tokens": 226046307.0, + "step": 11173 + }, + { + "epoch": 0.9214150243258844, + "grad_norm": 1.7586331367492676, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7804750204086304, + "num_tokens": 226080090.0, + "step": 11174 + }, + { + "epoch": 0.921497484950936, + "grad_norm": 1.8795504570007324, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.7716566324234009, + "num_tokens": 226110190.0, + "step": 11175 + }, + { + "epoch": 0.9215799455759874, + "grad_norm": 2.0442733764648438, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7731971144676208, + "num_tokens": 226138658.0, + "step": 11176 + }, + { + "epoch": 0.921662406201039, + "grad_norm": 1.7485874891281128, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7707622051239014, + "num_tokens": 226171965.0, + "step": 11177 + }, + { + "epoch": 0.9217448668260906, + "grad_norm": 2.097468614578247, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7297215461730957, + "num_tokens": 226202030.0, + "step": 11178 + }, + { + "epoch": 0.921827327451142, + "grad_norm": 1.9081212282180786, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.768219530582428, + "num_tokens": 226230637.0, + "step": 11179 + }, + { + "epoch": 0.9219097880761936, + "grad_norm": 1.8302820920944214, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7827214002609253, + "num_tokens": 226261806.0, + "step": 11180 + }, + { + "epoch": 0.9219922487012452, + "grad_norm": 1.7307884693145752, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7926503419876099, + "num_tokens": 226296208.0, + "step": 11181 + }, + { + "epoch": 0.9220747093262966, + "grad_norm": 1.7734735012054443, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.792149007320404, + "num_tokens": 226327519.0, + "step": 11182 + }, + { + "epoch": 0.9221571699513482, + "grad_norm": 2.0027546882629395, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7184098958969116, + "num_tokens": 226357486.0, + "step": 11183 + }, + { + "epoch": 0.9222396305763998, + "grad_norm": 2.1569626331329346, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7484263181686401, + "num_tokens": 226382478.0, + "step": 11184 + }, + { + "epoch": 0.9223220912014513, + "grad_norm": 1.9634349346160889, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7622647881507874, + "num_tokens": 226408546.0, + "step": 11185 + }, + { + "epoch": 0.9224045518265028, + "grad_norm": 1.9704748392105103, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7397555708885193, + "num_tokens": 226436858.0, + "step": 11186 + }, + { + "epoch": 0.9224870124515544, + "grad_norm": 2.082611083984375, + "learning_rate": 1e-06, + "loss": 0.765, + "mean_token_accuracy": 0.7489602565765381, + "num_tokens": 226462916.0, + "step": 11187 + }, + { + "epoch": 0.922569473076606, + "grad_norm": 2.0125133991241455, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7479113340377808, + "num_tokens": 226490276.0, + "step": 11188 + }, + { + "epoch": 0.9226519337016574, + "grad_norm": 1.9108556509017944, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.7611796855926514, + "num_tokens": 226519453.0, + "step": 11189 + }, + { + "epoch": 0.922734394326709, + "grad_norm": 1.8383464813232422, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.7649033069610596, + "num_tokens": 226550722.0, + "step": 11190 + }, + { + "epoch": 0.9228168549517606, + "grad_norm": 1.9432307481765747, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7414642572402954, + "num_tokens": 226579188.0, + "step": 11191 + }, + { + "epoch": 0.922899315576812, + "grad_norm": 1.7411260604858398, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7711024284362793, + "num_tokens": 226611933.0, + "step": 11192 + }, + { + "epoch": 0.9229817762018636, + "grad_norm": 2.0507726669311523, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7121044397354126, + "num_tokens": 226641697.0, + "step": 11193 + }, + { + "epoch": 0.9230642368269152, + "grad_norm": 1.894387125968933, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7600454688072205, + "num_tokens": 226672026.0, + "step": 11194 + }, + { + "epoch": 0.9231466974519666, + "grad_norm": 1.6825517416000366, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7546856999397278, + "num_tokens": 226708275.0, + "step": 11195 + }, + { + "epoch": 0.9232291580770182, + "grad_norm": 1.994971752166748, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7131775617599487, + "num_tokens": 226736816.0, + "step": 11196 + }, + { + "epoch": 0.9233116187020698, + "grad_norm": 1.8953773975372314, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.780887246131897, + "num_tokens": 226764685.0, + "step": 11197 + }, + { + "epoch": 0.9233940793271213, + "grad_norm": 1.9965704679489136, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7674505710601807, + "num_tokens": 226792825.0, + "step": 11198 + }, + { + "epoch": 0.9234765399521728, + "grad_norm": 1.8612109422683716, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7582398056983948, + "num_tokens": 226824160.0, + "step": 11199 + }, + { + "epoch": 0.9235590005772244, + "grad_norm": 1.8037052154541016, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7352581024169922, + "num_tokens": 226857338.0, + "step": 11200 + }, + { + "epoch": 0.9236414612022759, + "grad_norm": 2.0272951126098633, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7322728633880615, + "num_tokens": 226883876.0, + "step": 11201 + }, + { + "epoch": 0.9237239218273274, + "grad_norm": 1.9491294622421265, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7709758281707764, + "num_tokens": 226913912.0, + "step": 11202 + }, + { + "epoch": 0.923806382452379, + "grad_norm": 1.8147577047348022, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7398847937583923, + "num_tokens": 226947893.0, + "step": 11203 + }, + { + "epoch": 0.9238888430774306, + "grad_norm": 1.910335898399353, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7497382760047913, + "num_tokens": 226976720.0, + "step": 11204 + }, + { + "epoch": 0.923971303702482, + "grad_norm": 1.6557577848434448, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.791947603225708, + "num_tokens": 227015197.0, + "step": 11205 + }, + { + "epoch": 0.9240537643275336, + "grad_norm": 1.9733415842056274, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7063112258911133, + "num_tokens": 227046524.0, + "step": 11206 + }, + { + "epoch": 0.9241362249525852, + "grad_norm": 1.7402409315109253, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7822076082229614, + "num_tokens": 227082842.0, + "step": 11207 + }, + { + "epoch": 0.9242186855776366, + "grad_norm": 2.0486106872558594, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7122572660446167, + "num_tokens": 227113426.0, + "step": 11208 + }, + { + "epoch": 0.9243011462026882, + "grad_norm": 1.9201608896255493, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7371881008148193, + "num_tokens": 227142510.0, + "step": 11209 + }, + { + "epoch": 0.9243836068277398, + "grad_norm": 1.8206417560577393, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.7866497039794922, + "num_tokens": 227172197.0, + "step": 11210 + }, + { + "epoch": 0.9244660674527913, + "grad_norm": 2.051151752471924, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7405324578285217, + "num_tokens": 227200844.0, + "step": 11211 + }, + { + "epoch": 0.9245485280778428, + "grad_norm": 1.6873104572296143, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.751541793346405, + "num_tokens": 227239271.0, + "step": 11212 + }, + { + "epoch": 0.9246309887028944, + "grad_norm": 1.8824639320373535, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7647444009780884, + "num_tokens": 227269317.0, + "step": 11213 + }, + { + "epoch": 0.9247134493279459, + "grad_norm": 1.8157340288162231, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7255829572677612, + "num_tokens": 227304451.0, + "step": 11214 + }, + { + "epoch": 0.9247959099529974, + "grad_norm": 1.740599513053894, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.752754807472229, + "num_tokens": 227340988.0, + "step": 11215 + }, + { + "epoch": 0.924878370578049, + "grad_norm": 1.8583060503005981, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7420037984848022, + "num_tokens": 227372693.0, + "step": 11216 + }, + { + "epoch": 0.9249608312031005, + "grad_norm": 2.1277027130126953, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.7851845026016235, + "num_tokens": 227396888.0, + "step": 11217 + }, + { + "epoch": 0.925043291828152, + "grad_norm": 1.7671352624893188, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7380712628364563, + "num_tokens": 227429451.0, + "step": 11218 + }, + { + "epoch": 0.9251257524532036, + "grad_norm": 1.7702257633209229, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7953468561172485, + "num_tokens": 227462296.0, + "step": 11219 + }, + { + "epoch": 0.9252082130782552, + "grad_norm": 2.000102996826172, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.7695946097373962, + "num_tokens": 227490931.0, + "step": 11220 + }, + { + "epoch": 0.9252906737033066, + "grad_norm": 1.7556062936782837, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7674278616905212, + "num_tokens": 227524190.0, + "step": 11221 + }, + { + "epoch": 0.9253731343283582, + "grad_norm": 2.0874669551849365, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7696850299835205, + "num_tokens": 227548888.0, + "step": 11222 + }, + { + "epoch": 0.9254555949534098, + "grad_norm": 2.02020525932312, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7566115856170654, + "num_tokens": 227578543.0, + "step": 11223 + }, + { + "epoch": 0.9255380555784612, + "grad_norm": 2.0253021717071533, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7506265640258789, + "num_tokens": 227607394.0, + "step": 11224 + }, + { + "epoch": 0.9256205162035128, + "grad_norm": 1.960836410522461, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7333375215530396, + "num_tokens": 227638123.0, + "step": 11225 + }, + { + "epoch": 0.9257029768285644, + "grad_norm": 1.8595460653305054, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7211185693740845, + "num_tokens": 227669239.0, + "step": 11226 + }, + { + "epoch": 0.9257854374536159, + "grad_norm": 2.1421732902526855, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7433724403381348, + "num_tokens": 227697373.0, + "step": 11227 + }, + { + "epoch": 0.9258678980786674, + "grad_norm": 2.0151009559631348, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7551209330558777, + "num_tokens": 227725875.0, + "step": 11228 + }, + { + "epoch": 0.925950358703719, + "grad_norm": 2.210118055343628, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7245186567306519, + "num_tokens": 227749571.0, + "step": 11229 + }, + { + "epoch": 0.9260328193287705, + "grad_norm": 1.7030084133148193, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7403357028961182, + "num_tokens": 227786005.0, + "step": 11230 + }, + { + "epoch": 0.926115279953822, + "grad_norm": 1.9962959289550781, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7388044595718384, + "num_tokens": 227814529.0, + "step": 11231 + }, + { + "epoch": 0.9261977405788736, + "grad_norm": 1.8142540454864502, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7759918570518494, + "num_tokens": 227846629.0, + "step": 11232 + }, + { + "epoch": 0.9262802012039252, + "grad_norm": 2.049713611602783, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7825253009796143, + "num_tokens": 227873278.0, + "step": 11233 + }, + { + "epoch": 0.9263626618289766, + "grad_norm": 2.0697989463806152, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7786203622817993, + "num_tokens": 227902983.0, + "step": 11234 + }, + { + "epoch": 0.9264451224540282, + "grad_norm": 1.8748235702514648, + "learning_rate": 1e-06, + "loss": 0.7688, + "mean_token_accuracy": 0.7604447603225708, + "num_tokens": 227932874.0, + "step": 11235 + }, + { + "epoch": 0.9265275830790798, + "grad_norm": 2.0999577045440674, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7250890731811523, + "num_tokens": 227958453.0, + "step": 11236 + }, + { + "epoch": 0.9266100437041312, + "grad_norm": 2.0620791912078857, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7311707139015198, + "num_tokens": 227985959.0, + "step": 11237 + }, + { + "epoch": 0.9266925043291828, + "grad_norm": 1.947906494140625, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7476441860198975, + "num_tokens": 228014498.0, + "step": 11238 + }, + { + "epoch": 0.9267749649542344, + "grad_norm": 1.959558367729187, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7339113354682922, + "num_tokens": 228046127.0, + "step": 11239 + }, + { + "epoch": 0.9268574255792859, + "grad_norm": 1.9323186874389648, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.7963365316390991, + "num_tokens": 228073623.0, + "step": 11240 + }, + { + "epoch": 0.9269398862043374, + "grad_norm": 2.012214422225952, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.7322527170181274, + "num_tokens": 228100019.0, + "step": 11241 + }, + { + "epoch": 0.927022346829389, + "grad_norm": 1.7800877094268799, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7588735818862915, + "num_tokens": 228135447.0, + "step": 11242 + }, + { + "epoch": 0.9271048074544405, + "grad_norm": 1.8170205354690552, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7798095345497131, + "num_tokens": 228168771.0, + "step": 11243 + }, + { + "epoch": 0.927187268079492, + "grad_norm": 1.9613797664642334, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7441264390945435, + "num_tokens": 228197233.0, + "step": 11244 + }, + { + "epoch": 0.9272697287045436, + "grad_norm": 1.9514260292053223, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7464450597763062, + "num_tokens": 228228315.0, + "step": 11245 + }, + { + "epoch": 0.9273521893295951, + "grad_norm": 1.9994566440582275, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7332972288131714, + "num_tokens": 228256386.0, + "step": 11246 + }, + { + "epoch": 0.9274346499546466, + "grad_norm": 1.8081263303756714, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6948031783103943, + "num_tokens": 228293292.0, + "step": 11247 + }, + { + "epoch": 0.9275171105796982, + "grad_norm": 1.7810230255126953, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.748550534248352, + "num_tokens": 228325541.0, + "step": 11248 + }, + { + "epoch": 0.9275995712047498, + "grad_norm": 1.7496819496154785, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7408894300460815, + "num_tokens": 228361233.0, + "step": 11249 + }, + { + "epoch": 0.9276820318298012, + "grad_norm": 2.1272246837615967, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7557445764541626, + "num_tokens": 228388220.0, + "step": 11250 + }, + { + "epoch": 0.9277644924548528, + "grad_norm": 1.8477544784545898, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.744202196598053, + "num_tokens": 228421218.0, + "step": 11251 + }, + { + "epoch": 0.9278469530799044, + "grad_norm": 1.8432905673980713, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7872795462608337, + "num_tokens": 228452391.0, + "step": 11252 + }, + { + "epoch": 0.9279294137049559, + "grad_norm": 1.7581944465637207, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7619527578353882, + "num_tokens": 228483847.0, + "step": 11253 + }, + { + "epoch": 0.9280118743300074, + "grad_norm": 1.6995370388031006, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7632569670677185, + "num_tokens": 228519942.0, + "step": 11254 + }, + { + "epoch": 0.928094334955059, + "grad_norm": 1.8202769756317139, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7799007892608643, + "num_tokens": 228553550.0, + "step": 11255 + }, + { + "epoch": 0.9281767955801105, + "grad_norm": 1.8643200397491455, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7450796961784363, + "num_tokens": 228583691.0, + "step": 11256 + }, + { + "epoch": 0.928259256205162, + "grad_norm": 1.7878568172454834, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7395808696746826, + "num_tokens": 228618940.0, + "step": 11257 + }, + { + "epoch": 0.9283417168302136, + "grad_norm": 1.8577988147735596, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7426782250404358, + "num_tokens": 228649654.0, + "step": 11258 + }, + { + "epoch": 0.9284241774552651, + "grad_norm": 2.025076389312744, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.726230263710022, + "num_tokens": 228681424.0, + "step": 11259 + }, + { + "epoch": 0.9285066380803166, + "grad_norm": 1.9563318490982056, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7690340876579285, + "num_tokens": 228710993.0, + "step": 11260 + }, + { + "epoch": 0.9285890987053682, + "grad_norm": 2.0918045043945312, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7585145235061646, + "num_tokens": 228737885.0, + "step": 11261 + }, + { + "epoch": 0.9286715593304197, + "grad_norm": 2.0280349254608154, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7463308572769165, + "num_tokens": 228767428.0, + "step": 11262 + }, + { + "epoch": 0.9287540199554712, + "grad_norm": 1.9788563251495361, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7503153681755066, + "num_tokens": 228797497.0, + "step": 11263 + }, + { + "epoch": 0.9288364805805228, + "grad_norm": 2.009364604949951, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7149540781974792, + "num_tokens": 228829712.0, + "step": 11264 + }, + { + "epoch": 0.9289189412055744, + "grad_norm": 1.9437426328659058, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.7850812077522278, + "num_tokens": 228859924.0, + "step": 11265 + }, + { + "epoch": 0.9290014018306259, + "grad_norm": 2.299103260040283, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.743301272392273, + "num_tokens": 228880913.0, + "step": 11266 + }, + { + "epoch": 0.9290838624556774, + "grad_norm": 1.9758517742156982, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6933162808418274, + "num_tokens": 228913510.0, + "step": 11267 + }, + { + "epoch": 0.929166323080729, + "grad_norm": 1.7708793878555298, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7552844285964966, + "num_tokens": 228949957.0, + "step": 11268 + }, + { + "epoch": 0.9292487837057805, + "grad_norm": 2.1919944286346436, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.749807596206665, + "num_tokens": 228973431.0, + "step": 11269 + }, + { + "epoch": 0.929331244330832, + "grad_norm": 2.3107166290283203, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7436234951019287, + "num_tokens": 228995432.0, + "step": 11270 + }, + { + "epoch": 0.9294137049558836, + "grad_norm": 1.9430488348007202, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7161179184913635, + "num_tokens": 229026257.0, + "step": 11271 + }, + { + "epoch": 0.9294961655809351, + "grad_norm": 1.8807387351989746, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7383762001991272, + "num_tokens": 229055647.0, + "step": 11272 + }, + { + "epoch": 0.9295786262059866, + "grad_norm": 1.8247278928756714, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.7946965098381042, + "num_tokens": 229088833.0, + "step": 11273 + }, + { + "epoch": 0.9296610868310382, + "grad_norm": 2.3195254802703857, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7173042297363281, + "num_tokens": 229111593.0, + "step": 11274 + }, + { + "epoch": 0.9297435474560897, + "grad_norm": 1.9590215682983398, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7194186449050903, + "num_tokens": 229143697.0, + "step": 11275 + }, + { + "epoch": 0.9298260080811412, + "grad_norm": 1.8168165683746338, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7519088387489319, + "num_tokens": 229179216.0, + "step": 11276 + }, + { + "epoch": 0.9299084687061928, + "grad_norm": 2.0606470108032227, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7190483808517456, + "num_tokens": 229207607.0, + "step": 11277 + }, + { + "epoch": 0.9299909293312443, + "grad_norm": 1.8508429527282715, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7885870337486267, + "num_tokens": 229238549.0, + "step": 11278 + }, + { + "epoch": 0.9300733899562958, + "grad_norm": 2.16463565826416, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.7894697785377502, + "num_tokens": 229262246.0, + "step": 11279 + }, + { + "epoch": 0.9301558505813474, + "grad_norm": 2.0089433193206787, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7317218780517578, + "num_tokens": 229289110.0, + "step": 11280 + }, + { + "epoch": 0.930238311206399, + "grad_norm": 2.258579730987549, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7513735294342041, + "num_tokens": 229310194.0, + "step": 11281 + }, + { + "epoch": 0.9303207718314505, + "grad_norm": 1.7869980335235596, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7370513677597046, + "num_tokens": 229345208.0, + "step": 11282 + }, + { + "epoch": 0.930403232456502, + "grad_norm": 1.9124583005905151, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7421437501907349, + "num_tokens": 229380240.0, + "step": 11283 + }, + { + "epoch": 0.9304856930815536, + "grad_norm": 1.8719326257705688, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7346208095550537, + "num_tokens": 229411774.0, + "step": 11284 + }, + { + "epoch": 0.9305681537066051, + "grad_norm": 1.9535150527954102, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7530190348625183, + "num_tokens": 229439417.0, + "step": 11285 + }, + { + "epoch": 0.9306506143316566, + "grad_norm": 1.684149146080017, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.7941758632659912, + "num_tokens": 229477316.0, + "step": 11286 + }, + { + "epoch": 0.9307330749567082, + "grad_norm": 1.8844149112701416, + "learning_rate": 1e-06, + "loss": 0.594, + "mean_token_accuracy": 0.7757512927055359, + "num_tokens": 229505652.0, + "step": 11287 + }, + { + "epoch": 0.9308155355817597, + "grad_norm": 1.8977679014205933, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7351771593093872, + "num_tokens": 229536871.0, + "step": 11288 + }, + { + "epoch": 0.9308979962068112, + "grad_norm": 1.9418836832046509, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.758499026298523, + "num_tokens": 229566725.0, + "step": 11289 + }, + { + "epoch": 0.9309804568318628, + "grad_norm": 2.2199604511260986, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7437105178833008, + "num_tokens": 229589533.0, + "step": 11290 + }, + { + "epoch": 0.9310629174569143, + "grad_norm": 2.1597628593444824, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.729662299156189, + "num_tokens": 229615032.0, + "step": 11291 + }, + { + "epoch": 0.9311453780819658, + "grad_norm": 2.123584508895874, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7224215269088745, + "num_tokens": 229641071.0, + "step": 11292 + }, + { + "epoch": 0.9312278387070174, + "grad_norm": 1.9319521188735962, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7706146240234375, + "num_tokens": 229669318.0, + "step": 11293 + }, + { + "epoch": 0.931310299332069, + "grad_norm": 1.8111655712127686, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7547348737716675, + "num_tokens": 229701645.0, + "step": 11294 + }, + { + "epoch": 0.9313927599571205, + "grad_norm": 1.8766824007034302, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7407065629959106, + "num_tokens": 229731501.0, + "step": 11295 + }, + { + "epoch": 0.931475220582172, + "grad_norm": 1.898493766784668, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7459891438484192, + "num_tokens": 229760483.0, + "step": 11296 + }, + { + "epoch": 0.9315576812072236, + "grad_norm": 2.018333673477173, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.772291898727417, + "num_tokens": 229789240.0, + "step": 11297 + }, + { + "epoch": 0.9316401418322751, + "grad_norm": 1.893126368522644, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7588175535202026, + "num_tokens": 229819922.0, + "step": 11298 + }, + { + "epoch": 0.9317226024573266, + "grad_norm": 2.0451650619506836, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.740929365158081, + "num_tokens": 229846885.0, + "step": 11299 + }, + { + "epoch": 0.9318050630823782, + "grad_norm": 2.0806987285614014, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7344741821289062, + "num_tokens": 229872080.0, + "step": 11300 + }, + { + "epoch": 0.9318875237074297, + "grad_norm": 2.2932541370391846, + "learning_rate": 1e-06, + "loss": 0.7619, + "mean_token_accuracy": 0.7622862458229065, + "num_tokens": 229894689.0, + "step": 11301 + }, + { + "epoch": 0.9319699843324812, + "grad_norm": 1.9980902671813965, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7413903474807739, + "num_tokens": 229924729.0, + "step": 11302 + }, + { + "epoch": 0.9320524449575328, + "grad_norm": 1.887345314025879, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.7808392643928528, + "num_tokens": 229957405.0, + "step": 11303 + }, + { + "epoch": 0.9321349055825843, + "grad_norm": 2.119422197341919, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7569403648376465, + "num_tokens": 229983847.0, + "step": 11304 + }, + { + "epoch": 0.9322173662076358, + "grad_norm": 1.930336833000183, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.7799086570739746, + "num_tokens": 230013072.0, + "step": 11305 + }, + { + "epoch": 0.9322998268326874, + "grad_norm": 1.976129174232483, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7646210789680481, + "num_tokens": 230044562.0, + "step": 11306 + }, + { + "epoch": 0.9323822874577389, + "grad_norm": 1.9682552814483643, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.7558201551437378, + "num_tokens": 230077432.0, + "step": 11307 + }, + { + "epoch": 0.9324647480827905, + "grad_norm": 1.9067132472991943, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.7857397198677063, + "num_tokens": 230108403.0, + "step": 11308 + }, + { + "epoch": 0.932547208707842, + "grad_norm": 1.8820149898529053, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7344816327095032, + "num_tokens": 230140333.0, + "step": 11309 + }, + { + "epoch": 0.9326296693328936, + "grad_norm": 1.9052085876464844, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7446193695068359, + "num_tokens": 230169744.0, + "step": 11310 + }, + { + "epoch": 0.9327121299579451, + "grad_norm": 1.8002607822418213, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7701349258422852, + "num_tokens": 230202498.0, + "step": 11311 + }, + { + "epoch": 0.9327945905829966, + "grad_norm": 1.7290090322494507, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7580832839012146, + "num_tokens": 230235557.0, + "step": 11312 + }, + { + "epoch": 0.9328770512080482, + "grad_norm": 1.897074818611145, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7666598558425903, + "num_tokens": 230264875.0, + "step": 11313 + }, + { + "epoch": 0.9329595118330997, + "grad_norm": 2.0241916179656982, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6901806592941284, + "num_tokens": 230293143.0, + "step": 11314 + }, + { + "epoch": 0.9330419724581512, + "grad_norm": 1.7501646280288696, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7668063640594482, + "num_tokens": 230331779.0, + "step": 11315 + }, + { + "epoch": 0.9331244330832028, + "grad_norm": 1.9989701509475708, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7661420702934265, + "num_tokens": 230359172.0, + "step": 11316 + }, + { + "epoch": 0.9332068937082543, + "grad_norm": 1.9109411239624023, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7805516123771667, + "num_tokens": 230385846.0, + "step": 11317 + }, + { + "epoch": 0.9332893543333058, + "grad_norm": 2.106581211090088, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7412866950035095, + "num_tokens": 230413696.0, + "step": 11318 + }, + { + "epoch": 0.9333718149583574, + "grad_norm": 2.142885446548462, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7251598834991455, + "num_tokens": 230440590.0, + "step": 11319 + }, + { + "epoch": 0.9334542755834089, + "grad_norm": 1.8353869915008545, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7463710904121399, + "num_tokens": 230471361.0, + "step": 11320 + }, + { + "epoch": 0.9335367362084604, + "grad_norm": 1.8312811851501465, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7423756122589111, + "num_tokens": 230504943.0, + "step": 11321 + }, + { + "epoch": 0.933619196833512, + "grad_norm": 2.1375575065612793, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7142356634140015, + "num_tokens": 230530756.0, + "step": 11322 + }, + { + "epoch": 0.9337016574585635, + "grad_norm": 2.028689384460449, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7705997824668884, + "num_tokens": 230558812.0, + "step": 11323 + }, + { + "epoch": 0.9337841180836151, + "grad_norm": 1.6486411094665527, + "learning_rate": 1e-06, + "loss": 0.5879, + "mean_token_accuracy": 0.7985121011734009, + "num_tokens": 230595438.0, + "step": 11324 + }, + { + "epoch": 0.9338665787086666, + "grad_norm": 2.042024612426758, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7424759864807129, + "num_tokens": 230622671.0, + "step": 11325 + }, + { + "epoch": 0.9339490393337182, + "grad_norm": 2.0951173305511475, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7548287510871887, + "num_tokens": 230650091.0, + "step": 11326 + }, + { + "epoch": 0.9340314999587697, + "grad_norm": 1.9994986057281494, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7242546677589417, + "num_tokens": 230678479.0, + "step": 11327 + }, + { + "epoch": 0.9341139605838212, + "grad_norm": 1.6535096168518066, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7526007294654846, + "num_tokens": 230717010.0, + "step": 11328 + }, + { + "epoch": 0.9341964212088728, + "grad_norm": 1.8994749784469604, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7321733236312866, + "num_tokens": 230747508.0, + "step": 11329 + }, + { + "epoch": 0.9342788818339243, + "grad_norm": 1.7308320999145508, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7440545558929443, + "num_tokens": 230780964.0, + "step": 11330 + }, + { + "epoch": 0.9343613424589758, + "grad_norm": 1.959901213645935, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7700503468513489, + "num_tokens": 230810136.0, + "step": 11331 + }, + { + "epoch": 0.9344438030840274, + "grad_norm": 1.9692816734313965, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7234188318252563, + "num_tokens": 230841519.0, + "step": 11332 + }, + { + "epoch": 0.9345262637090789, + "grad_norm": 1.77646803855896, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.757756233215332, + "num_tokens": 230874987.0, + "step": 11333 + }, + { + "epoch": 0.9346087243341304, + "grad_norm": 2.1051907539367676, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7430812120437622, + "num_tokens": 230901189.0, + "step": 11334 + }, + { + "epoch": 0.934691184959182, + "grad_norm": 2.118635654449463, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7487224340438843, + "num_tokens": 230926423.0, + "step": 11335 + }, + { + "epoch": 0.9347736455842335, + "grad_norm": 1.9329707622528076, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7712714672088623, + "num_tokens": 230954859.0, + "step": 11336 + }, + { + "epoch": 0.934856106209285, + "grad_norm": 1.6729238033294678, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.780343770980835, + "num_tokens": 230990299.0, + "step": 11337 + }, + { + "epoch": 0.9349385668343366, + "grad_norm": 2.0261590480804443, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7552447319030762, + "num_tokens": 231021817.0, + "step": 11338 + }, + { + "epoch": 0.9350210274593881, + "grad_norm": 1.908540964126587, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7641699314117432, + "num_tokens": 231050074.0, + "step": 11339 + }, + { + "epoch": 0.9351034880844397, + "grad_norm": 1.733760952949524, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.7941411733627319, + "num_tokens": 231085325.0, + "step": 11340 + }, + { + "epoch": 0.9351859487094912, + "grad_norm": 2.0005900859832764, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.7596503496170044, + "num_tokens": 231114926.0, + "step": 11341 + }, + { + "epoch": 0.9352684093345428, + "grad_norm": 2.025743007659912, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7259791493415833, + "num_tokens": 231146515.0, + "step": 11342 + }, + { + "epoch": 0.9353508699595943, + "grad_norm": 2.274770975112915, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7633316516876221, + "num_tokens": 231167777.0, + "step": 11343 + }, + { + "epoch": 0.9354333305846458, + "grad_norm": 1.8321127891540527, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7186573147773743, + "num_tokens": 231200591.0, + "step": 11344 + }, + { + "epoch": 0.9355157912096974, + "grad_norm": 2.1172714233398438, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.745088517665863, + "num_tokens": 231227392.0, + "step": 11345 + }, + { + "epoch": 0.9355982518347489, + "grad_norm": 1.6427457332611084, + "learning_rate": 1e-06, + "loss": 0.517, + "mean_token_accuracy": 0.8204982280731201, + "num_tokens": 231261877.0, + "step": 11346 + }, + { + "epoch": 0.9356807124598004, + "grad_norm": 2.325542688369751, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7125715017318726, + "num_tokens": 231287118.0, + "step": 11347 + }, + { + "epoch": 0.935763173084852, + "grad_norm": 1.9845105409622192, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7310976982116699, + "num_tokens": 231315305.0, + "step": 11348 + }, + { + "epoch": 0.9358456337099035, + "grad_norm": 2.2123143672943115, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7437834739685059, + "num_tokens": 231340668.0, + "step": 11349 + }, + { + "epoch": 0.935928094334955, + "grad_norm": 1.95681631565094, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.742292046546936, + "num_tokens": 231370962.0, + "step": 11350 + }, + { + "epoch": 0.9360105549600066, + "grad_norm": 1.818444848060608, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7695462107658386, + "num_tokens": 231399697.0, + "step": 11351 + }, + { + "epoch": 0.9360930155850581, + "grad_norm": 1.95918869972229, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7206445336341858, + "num_tokens": 231429451.0, + "step": 11352 + }, + { + "epoch": 0.9361754762101097, + "grad_norm": 2.0077412128448486, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7478092908859253, + "num_tokens": 231457585.0, + "step": 11353 + }, + { + "epoch": 0.9362579368351612, + "grad_norm": 2.1757099628448486, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.738086462020874, + "num_tokens": 231483835.0, + "step": 11354 + }, + { + "epoch": 0.9363403974602128, + "grad_norm": 1.6754519939422607, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7700414061546326, + "num_tokens": 231519555.0, + "step": 11355 + }, + { + "epoch": 0.9364228580852643, + "grad_norm": 1.8510156869888306, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7358959913253784, + "num_tokens": 231552274.0, + "step": 11356 + }, + { + "epoch": 0.9365053187103158, + "grad_norm": 1.9110651016235352, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7620877623558044, + "num_tokens": 231582050.0, + "step": 11357 + }, + { + "epoch": 0.9365877793353674, + "grad_norm": 1.6685421466827393, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7403129935264587, + "num_tokens": 231621929.0, + "step": 11358 + }, + { + "epoch": 0.9366702399604189, + "grad_norm": 1.9235858917236328, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7218332886695862, + "num_tokens": 231654233.0, + "step": 11359 + }, + { + "epoch": 0.9367527005854704, + "grad_norm": 2.2124693393707275, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7170591354370117, + "num_tokens": 231676557.0, + "step": 11360 + }, + { + "epoch": 0.936835161210522, + "grad_norm": 2.2691314220428467, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7187099456787109, + "num_tokens": 231701985.0, + "step": 11361 + }, + { + "epoch": 0.9369176218355735, + "grad_norm": 2.107172966003418, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7340076565742493, + "num_tokens": 231727617.0, + "step": 11362 + }, + { + "epoch": 0.937000082460625, + "grad_norm": 1.7966318130493164, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7636420726776123, + "num_tokens": 231763193.0, + "step": 11363 + }, + { + "epoch": 0.9370825430856766, + "grad_norm": 1.910047173500061, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7627806067466736, + "num_tokens": 231792697.0, + "step": 11364 + }, + { + "epoch": 0.9371650037107281, + "grad_norm": 2.072161912918091, + "learning_rate": 1e-06, + "loss": 0.7808, + "mean_token_accuracy": 0.7476629018783569, + "num_tokens": 231819875.0, + "step": 11365 + }, + { + "epoch": 0.9372474643357797, + "grad_norm": 1.8016959428787231, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7614566087722778, + "num_tokens": 231855764.0, + "step": 11366 + }, + { + "epoch": 0.9373299249608312, + "grad_norm": 1.8496657609939575, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7907775640487671, + "num_tokens": 231883953.0, + "step": 11367 + }, + { + "epoch": 0.9374123855858827, + "grad_norm": 2.351036787033081, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7891750335693359, + "num_tokens": 231906397.0, + "step": 11368 + }, + { + "epoch": 0.9374948462109343, + "grad_norm": 1.8322335481643677, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.744972825050354, + "num_tokens": 231941087.0, + "step": 11369 + }, + { + "epoch": 0.9375773068359858, + "grad_norm": 2.0909340381622314, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7284569144248962, + "num_tokens": 231966863.0, + "step": 11370 + }, + { + "epoch": 0.9376597674610374, + "grad_norm": 1.8352516889572144, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7644802331924438, + "num_tokens": 231997776.0, + "step": 11371 + }, + { + "epoch": 0.9377422280860889, + "grad_norm": 2.1456363201141357, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7558183670043945, + "num_tokens": 232022065.0, + "step": 11372 + }, + { + "epoch": 0.9378246887111404, + "grad_norm": 1.8952550888061523, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7783951759338379, + "num_tokens": 232052866.0, + "step": 11373 + }, + { + "epoch": 0.937907149336192, + "grad_norm": 1.901287317276001, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7331674098968506, + "num_tokens": 232085200.0, + "step": 11374 + }, + { + "epoch": 0.9379896099612435, + "grad_norm": 1.8940421342849731, + "learning_rate": 1e-06, + "loss": 0.7926, + "mean_token_accuracy": 0.7562431693077087, + "num_tokens": 232115449.0, + "step": 11375 + }, + { + "epoch": 0.938072070586295, + "grad_norm": 2.326460123062134, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7083292007446289, + "num_tokens": 232139805.0, + "step": 11376 + }, + { + "epoch": 0.9381545312113466, + "grad_norm": 2.10659122467041, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7289483547210693, + "num_tokens": 232166863.0, + "step": 11377 + }, + { + "epoch": 0.9382369918363981, + "grad_norm": 1.941669225692749, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.7820632457733154, + "num_tokens": 232194878.0, + "step": 11378 + }, + { + "epoch": 0.9383194524614497, + "grad_norm": 2.213247776031494, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7223735451698303, + "num_tokens": 232220899.0, + "step": 11379 + }, + { + "epoch": 0.9384019130865012, + "grad_norm": 1.8759970664978027, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7702882885932922, + "num_tokens": 232250252.0, + "step": 11380 + }, + { + "epoch": 0.9384843737115527, + "grad_norm": 1.772324800491333, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7762603759765625, + "num_tokens": 232281157.0, + "step": 11381 + }, + { + "epoch": 0.9385668343366043, + "grad_norm": 1.931607723236084, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7790402173995972, + "num_tokens": 232309954.0, + "step": 11382 + }, + { + "epoch": 0.9386492949616558, + "grad_norm": 1.8515757322311401, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7644533514976501, + "num_tokens": 232342616.0, + "step": 11383 + }, + { + "epoch": 0.9387317555867073, + "grad_norm": 1.851493239402771, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7453263998031616, + "num_tokens": 232375411.0, + "step": 11384 + }, + { + "epoch": 0.9388142162117589, + "grad_norm": 1.718320608139038, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7673130035400391, + "num_tokens": 232409627.0, + "step": 11385 + }, + { + "epoch": 0.9388966768368104, + "grad_norm": 1.9167709350585938, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7698585987091064, + "num_tokens": 232440267.0, + "step": 11386 + }, + { + "epoch": 0.938979137461862, + "grad_norm": 1.843074083328247, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7730221748352051, + "num_tokens": 232468630.0, + "step": 11387 + }, + { + "epoch": 0.9390615980869135, + "grad_norm": 2.133965492248535, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.760608434677124, + "num_tokens": 232494829.0, + "step": 11388 + }, + { + "epoch": 0.939144058711965, + "grad_norm": 1.9718382358551025, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.756835401058197, + "num_tokens": 232523195.0, + "step": 11389 + }, + { + "epoch": 0.9392265193370166, + "grad_norm": 1.998361587524414, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7434577941894531, + "num_tokens": 232552274.0, + "step": 11390 + }, + { + "epoch": 0.9393089799620681, + "grad_norm": 1.9250119924545288, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7760109901428223, + "num_tokens": 232582303.0, + "step": 11391 + }, + { + "epoch": 0.9393914405871197, + "grad_norm": 1.867612361907959, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7732385396957397, + "num_tokens": 232611780.0, + "step": 11392 + }, + { + "epoch": 0.9394739012121712, + "grad_norm": 1.8353585004806519, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7519416809082031, + "num_tokens": 232644319.0, + "step": 11393 + }, + { + "epoch": 0.9395563618372227, + "grad_norm": 1.7725220918655396, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7703563570976257, + "num_tokens": 232679045.0, + "step": 11394 + }, + { + "epoch": 0.9396388224622743, + "grad_norm": 2.01654052734375, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7624571323394775, + "num_tokens": 232706676.0, + "step": 11395 + }, + { + "epoch": 0.9397212830873258, + "grad_norm": 1.9736460447311401, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7480023503303528, + "num_tokens": 232734978.0, + "step": 11396 + }, + { + "epoch": 0.9398037437123773, + "grad_norm": 2.0043785572052, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.770430862903595, + "num_tokens": 232763014.0, + "step": 11397 + }, + { + "epoch": 0.9398862043374289, + "grad_norm": 1.662376046180725, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.7937712669372559, + "num_tokens": 232799658.0, + "step": 11398 + }, + { + "epoch": 0.9399686649624804, + "grad_norm": 1.9396371841430664, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7533186674118042, + "num_tokens": 232830700.0, + "step": 11399 + }, + { + "epoch": 0.9400511255875319, + "grad_norm": 1.9243453741073608, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7771446108818054, + "num_tokens": 232858271.0, + "step": 11400 + }, + { + "epoch": 0.9401335862125835, + "grad_norm": 2.1455862522125244, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.7793818712234497, + "num_tokens": 232881707.0, + "step": 11401 + }, + { + "epoch": 0.940216046837635, + "grad_norm": 1.884690761566162, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7564770579338074, + "num_tokens": 232915588.0, + "step": 11402 + }, + { + "epoch": 0.9402985074626866, + "grad_norm": 2.009497880935669, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7407290935516357, + "num_tokens": 232942360.0, + "step": 11403 + }, + { + "epoch": 0.9403809680877381, + "grad_norm": 2.055309295654297, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.7889984846115112, + "num_tokens": 232969557.0, + "step": 11404 + }, + { + "epoch": 0.9404634287127897, + "grad_norm": 1.9985464811325073, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7574145793914795, + "num_tokens": 233000004.0, + "step": 11405 + }, + { + "epoch": 0.9405458893378412, + "grad_norm": 1.9854393005371094, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7856670618057251, + "num_tokens": 233027567.0, + "step": 11406 + }, + { + "epoch": 0.9406283499628927, + "grad_norm": 2.1419684886932373, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7861984968185425, + "num_tokens": 233052041.0, + "step": 11407 + }, + { + "epoch": 0.9407108105879443, + "grad_norm": 1.9681925773620605, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7645748853683472, + "num_tokens": 233080732.0, + "step": 11408 + }, + { + "epoch": 0.9407932712129958, + "grad_norm": 1.8361879587173462, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7582866549491882, + "num_tokens": 233116146.0, + "step": 11409 + }, + { + "epoch": 0.9408757318380473, + "grad_norm": 2.079625368118286, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.745589017868042, + "num_tokens": 233141715.0, + "step": 11410 + }, + { + "epoch": 0.9409581924630989, + "grad_norm": 1.9293557405471802, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7223786115646362, + "num_tokens": 233173414.0, + "step": 11411 + }, + { + "epoch": 0.9410406530881504, + "grad_norm": 1.8338770866394043, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7463786005973816, + "num_tokens": 233208084.0, + "step": 11412 + }, + { + "epoch": 0.9411231137132019, + "grad_norm": 1.8512139320373535, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.730625569820404, + "num_tokens": 233242447.0, + "step": 11413 + }, + { + "epoch": 0.9412055743382535, + "grad_norm": 1.9802470207214355, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7247049808502197, + "num_tokens": 233274950.0, + "step": 11414 + }, + { + "epoch": 0.941288034963305, + "grad_norm": 2.2965948581695557, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7374742031097412, + "num_tokens": 233298218.0, + "step": 11415 + }, + { + "epoch": 0.9413704955883566, + "grad_norm": 1.8511507511138916, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7556169033050537, + "num_tokens": 233328445.0, + "step": 11416 + }, + { + "epoch": 0.9414529562134081, + "grad_norm": 1.9024521112442017, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7703219056129456, + "num_tokens": 233360467.0, + "step": 11417 + }, + { + "epoch": 0.9415354168384596, + "grad_norm": 2.0797550678253174, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7726957201957703, + "num_tokens": 233386884.0, + "step": 11418 + }, + { + "epoch": 0.9416178774635112, + "grad_norm": 1.7263665199279785, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7625495195388794, + "num_tokens": 233420951.0, + "step": 11419 + }, + { + "epoch": 0.9417003380885627, + "grad_norm": 1.6613965034484863, + "learning_rate": 1e-06, + "loss": 0.6214, + "mean_token_accuracy": 0.8043673038482666, + "num_tokens": 233453466.0, + "step": 11420 + }, + { + "epoch": 0.9417827987136143, + "grad_norm": 1.6088536977767944, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7971658706665039, + "num_tokens": 233490569.0, + "step": 11421 + }, + { + "epoch": 0.9418652593386658, + "grad_norm": 1.9976506233215332, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7468377351760864, + "num_tokens": 233523112.0, + "step": 11422 + }, + { + "epoch": 0.9419477199637173, + "grad_norm": 1.9729663133621216, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7418504953384399, + "num_tokens": 233551962.0, + "step": 11423 + }, + { + "epoch": 0.9420301805887689, + "grad_norm": 1.9461479187011719, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7417245507240295, + "num_tokens": 233580532.0, + "step": 11424 + }, + { + "epoch": 0.9421126412138204, + "grad_norm": 1.9014793634414673, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7516252398490906, + "num_tokens": 233611707.0, + "step": 11425 + }, + { + "epoch": 0.9421951018388719, + "grad_norm": 1.8468250036239624, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7030206918716431, + "num_tokens": 233645709.0, + "step": 11426 + }, + { + "epoch": 0.9422775624639235, + "grad_norm": 1.8248564004898071, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.78932124376297, + "num_tokens": 233675989.0, + "step": 11427 + }, + { + "epoch": 0.942360023088975, + "grad_norm": 1.8061760663986206, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7796358466148376, + "num_tokens": 233708128.0, + "step": 11428 + }, + { + "epoch": 0.9424424837140265, + "grad_norm": 1.9033325910568237, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7716045379638672, + "num_tokens": 233735047.0, + "step": 11429 + }, + { + "epoch": 0.9425249443390781, + "grad_norm": 1.8919296264648438, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7669396996498108, + "num_tokens": 233765597.0, + "step": 11430 + }, + { + "epoch": 0.9426074049641296, + "grad_norm": 1.8708773851394653, + "learning_rate": 1e-06, + "loss": 0.6034, + "mean_token_accuracy": 0.7974650859832764, + "num_tokens": 233793993.0, + "step": 11431 + }, + { + "epoch": 0.9426898655891812, + "grad_norm": 1.8646310567855835, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7825639247894287, + "num_tokens": 233824282.0, + "step": 11432 + }, + { + "epoch": 0.9427723262142327, + "grad_norm": 2.0487818717956543, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7264140844345093, + "num_tokens": 233853853.0, + "step": 11433 + }, + { + "epoch": 0.9428547868392843, + "grad_norm": 2.1095688343048096, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7897154688835144, + "num_tokens": 233882972.0, + "step": 11434 + }, + { + "epoch": 0.9429372474643358, + "grad_norm": 1.8628425598144531, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7450395822525024, + "num_tokens": 233918278.0, + "step": 11435 + }, + { + "epoch": 0.9430197080893873, + "grad_norm": 1.717355728149414, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7319982051849365, + "num_tokens": 233954100.0, + "step": 11436 + }, + { + "epoch": 0.9431021687144389, + "grad_norm": 2.1973166465759277, + "learning_rate": 1e-06, + "loss": 0.7706, + "mean_token_accuracy": 0.7643565535545349, + "num_tokens": 233975044.0, + "step": 11437 + }, + { + "epoch": 0.9431846293394904, + "grad_norm": 2.0017621517181396, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7381829023361206, + "num_tokens": 234006964.0, + "step": 11438 + }, + { + "epoch": 0.9432670899645419, + "grad_norm": 1.8818308115005493, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7550862431526184, + "num_tokens": 234041215.0, + "step": 11439 + }, + { + "epoch": 0.9433495505895935, + "grad_norm": 2.3395702838897705, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7526376843452454, + "num_tokens": 234064575.0, + "step": 11440 + }, + { + "epoch": 0.943432011214645, + "grad_norm": 1.6475633382797241, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7496518492698669, + "num_tokens": 234102646.0, + "step": 11441 + }, + { + "epoch": 0.9435144718396965, + "grad_norm": 1.9612191915512085, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7141345143318176, + "num_tokens": 234132833.0, + "step": 11442 + }, + { + "epoch": 0.9435969324647481, + "grad_norm": 1.7973963022232056, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7212690711021423, + "num_tokens": 234167960.0, + "step": 11443 + }, + { + "epoch": 0.9436793930897996, + "grad_norm": 2.061837673187256, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7610692381858826, + "num_tokens": 234192538.0, + "step": 11444 + }, + { + "epoch": 0.9437618537148511, + "grad_norm": 1.8733059167861938, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7218111753463745, + "num_tokens": 234222341.0, + "step": 11445 + }, + { + "epoch": 0.9438443143399027, + "grad_norm": 1.7840672731399536, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7459766864776611, + "num_tokens": 234258074.0, + "step": 11446 + }, + { + "epoch": 0.9439267749649543, + "grad_norm": 1.7762748003005981, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7620903253555298, + "num_tokens": 234290391.0, + "step": 11447 + }, + { + "epoch": 0.9440092355900058, + "grad_norm": 1.993098497390747, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7276496887207031, + "num_tokens": 234321246.0, + "step": 11448 + }, + { + "epoch": 0.9440916962150573, + "grad_norm": 2.0988357067108154, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.768491268157959, + "num_tokens": 234345544.0, + "step": 11449 + }, + { + "epoch": 0.9441741568401089, + "grad_norm": 1.7909924983978271, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7337490320205688, + "num_tokens": 234381233.0, + "step": 11450 + }, + { + "epoch": 0.9442566174651604, + "grad_norm": 1.8215619325637817, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.7823435664176941, + "num_tokens": 234416622.0, + "step": 11451 + }, + { + "epoch": 0.9443390780902119, + "grad_norm": 2.1890649795532227, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.699465811252594, + "num_tokens": 234442417.0, + "step": 11452 + }, + { + "epoch": 0.9444215387152635, + "grad_norm": 2.0384438037872314, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7683664560317993, + "num_tokens": 234469686.0, + "step": 11453 + }, + { + "epoch": 0.944503999340315, + "grad_norm": 2.0526468753814697, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7143503427505493, + "num_tokens": 234499122.0, + "step": 11454 + }, + { + "epoch": 0.9445864599653665, + "grad_norm": 1.5715292692184448, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7631043195724487, + "num_tokens": 234542106.0, + "step": 11455 + }, + { + "epoch": 0.9446689205904181, + "grad_norm": 1.8664332628250122, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7424815893173218, + "num_tokens": 234572891.0, + "step": 11456 + }, + { + "epoch": 0.9447513812154696, + "grad_norm": 1.8905991315841675, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7153689861297607, + "num_tokens": 234604419.0, + "step": 11457 + }, + { + "epoch": 0.9448338418405211, + "grad_norm": 1.8642414808273315, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7596812844276428, + "num_tokens": 234632299.0, + "step": 11458 + }, + { + "epoch": 0.9449163024655727, + "grad_norm": 2.047137498855591, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7353115081787109, + "num_tokens": 234661777.0, + "step": 11459 + }, + { + "epoch": 0.9449987630906242, + "grad_norm": 2.3379366397857666, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7333685755729675, + "num_tokens": 234686635.0, + "step": 11460 + }, + { + "epoch": 0.9450812237156757, + "grad_norm": 2.008183479309082, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7634630799293518, + "num_tokens": 234714832.0, + "step": 11461 + }, + { + "epoch": 0.9451636843407273, + "grad_norm": 1.7484309673309326, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7686233520507812, + "num_tokens": 234749743.0, + "step": 11462 + }, + { + "epoch": 0.9452461449657789, + "grad_norm": 2.0808334350585938, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7094389200210571, + "num_tokens": 234777512.0, + "step": 11463 + }, + { + "epoch": 0.9453286055908304, + "grad_norm": 1.8599708080291748, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.7953517436981201, + "num_tokens": 234805974.0, + "step": 11464 + }, + { + "epoch": 0.9454110662158819, + "grad_norm": 1.888319492340088, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7711703777313232, + "num_tokens": 234836351.0, + "step": 11465 + }, + { + "epoch": 0.9454935268409335, + "grad_norm": 1.9255136251449585, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7423150539398193, + "num_tokens": 234868789.0, + "step": 11466 + }, + { + "epoch": 0.945575987465985, + "grad_norm": 1.8475475311279297, + "learning_rate": 1e-06, + "loss": 0.57, + "mean_token_accuracy": 0.8143883943557739, + "num_tokens": 234897085.0, + "step": 11467 + }, + { + "epoch": 0.9456584480910365, + "grad_norm": 2.3092427253723145, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7185817956924438, + "num_tokens": 234923049.0, + "step": 11468 + }, + { + "epoch": 0.9457409087160881, + "grad_norm": 1.8275610208511353, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7686029076576233, + "num_tokens": 234954619.0, + "step": 11469 + }, + { + "epoch": 0.9458233693411396, + "grad_norm": 1.8606914281845093, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7340811491012573, + "num_tokens": 234987737.0, + "step": 11470 + }, + { + "epoch": 0.9459058299661911, + "grad_norm": 1.9198962450027466, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7222285270690918, + "num_tokens": 235022482.0, + "step": 11471 + }, + { + "epoch": 0.9459882905912427, + "grad_norm": 1.9494432210922241, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7443128824234009, + "num_tokens": 235053710.0, + "step": 11472 + }, + { + "epoch": 0.9460707512162942, + "grad_norm": 1.8682811260223389, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.720030665397644, + "num_tokens": 235087053.0, + "step": 11473 + }, + { + "epoch": 0.9461532118413457, + "grad_norm": 1.9954047203063965, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7430652976036072, + "num_tokens": 235120476.0, + "step": 11474 + }, + { + "epoch": 0.9462356724663973, + "grad_norm": 2.1525256633758545, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7574089765548706, + "num_tokens": 235145661.0, + "step": 11475 + }, + { + "epoch": 0.9463181330914489, + "grad_norm": 2.1031014919281006, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7469009160995483, + "num_tokens": 235174158.0, + "step": 11476 + }, + { + "epoch": 0.9464005937165003, + "grad_norm": 2.3462231159210205, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7577428817749023, + "num_tokens": 235194178.0, + "step": 11477 + }, + { + "epoch": 0.9464830543415519, + "grad_norm": 2.0609118938446045, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7686092853546143, + "num_tokens": 235222698.0, + "step": 11478 + }, + { + "epoch": 0.9465655149666035, + "grad_norm": 1.627995252609253, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7542514801025391, + "num_tokens": 235258660.0, + "step": 11479 + }, + { + "epoch": 0.946647975591655, + "grad_norm": 2.336082935333252, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.742688775062561, + "num_tokens": 235281513.0, + "step": 11480 + }, + { + "epoch": 0.9467304362167065, + "grad_norm": 2.06266450881958, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7308995723724365, + "num_tokens": 235309182.0, + "step": 11481 + }, + { + "epoch": 0.9468128968417581, + "grad_norm": 1.9844253063201904, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7694961428642273, + "num_tokens": 235335477.0, + "step": 11482 + }, + { + "epoch": 0.9468953574668096, + "grad_norm": 1.9651440382003784, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7808563113212585, + "num_tokens": 235362970.0, + "step": 11483 + }, + { + "epoch": 0.9469778180918611, + "grad_norm": 1.8143041133880615, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.7895093560218811, + "num_tokens": 235395099.0, + "step": 11484 + }, + { + "epoch": 0.9470602787169127, + "grad_norm": 1.9353913068771362, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7337915301322937, + "num_tokens": 235424731.0, + "step": 11485 + }, + { + "epoch": 0.9471427393419642, + "grad_norm": 1.6810675859451294, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7553848028182983, + "num_tokens": 235460841.0, + "step": 11486 + }, + { + "epoch": 0.9472251999670157, + "grad_norm": 2.1441595554351807, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7333025336265564, + "num_tokens": 235484897.0, + "step": 11487 + }, + { + "epoch": 0.9473076605920673, + "grad_norm": 2.0086116790771484, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7193639278411865, + "num_tokens": 235512729.0, + "step": 11488 + }, + { + "epoch": 0.9473901212171189, + "grad_norm": 2.0758635997772217, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7614567279815674, + "num_tokens": 235541153.0, + "step": 11489 + }, + { + "epoch": 0.9474725818421703, + "grad_norm": 2.0794084072113037, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7153152823448181, + "num_tokens": 235568547.0, + "step": 11490 + }, + { + "epoch": 0.9475550424672219, + "grad_norm": 2.2582178115844727, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7654696106910706, + "num_tokens": 235590134.0, + "step": 11491 + }, + { + "epoch": 0.9476375030922735, + "grad_norm": 2.3524625301361084, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7518664598464966, + "num_tokens": 235614179.0, + "step": 11492 + }, + { + "epoch": 0.947719963717325, + "grad_norm": 1.9342154264450073, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7701661586761475, + "num_tokens": 235642833.0, + "step": 11493 + }, + { + "epoch": 0.9478024243423765, + "grad_norm": 1.8872936964035034, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7579586505889893, + "num_tokens": 235672484.0, + "step": 11494 + }, + { + "epoch": 0.9478848849674281, + "grad_norm": 1.8414729833602905, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7740515470504761, + "num_tokens": 235703108.0, + "step": 11495 + }, + { + "epoch": 0.9479673455924796, + "grad_norm": 1.7851775884628296, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7605091333389282, + "num_tokens": 235740507.0, + "step": 11496 + }, + { + "epoch": 0.9480498062175311, + "grad_norm": 1.9028630256652832, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7419041395187378, + "num_tokens": 235769689.0, + "step": 11497 + }, + { + "epoch": 0.9481322668425827, + "grad_norm": 2.0060372352600098, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7367055416107178, + "num_tokens": 235797905.0, + "step": 11498 + }, + { + "epoch": 0.9482147274676342, + "grad_norm": 1.9454137086868286, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7679558992385864, + "num_tokens": 235824616.0, + "step": 11499 + }, + { + "epoch": 0.9482971880926857, + "grad_norm": 1.9082242250442505, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7282633781433105, + "num_tokens": 235856421.0, + "step": 11500 + }, + { + "epoch": 0.9483796487177373, + "grad_norm": 2.0810019969940186, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7277570962905884, + "num_tokens": 235882855.0, + "step": 11501 + }, + { + "epoch": 0.9484621093427889, + "grad_norm": 2.2001028060913086, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.7417701482772827, + "num_tokens": 235905689.0, + "step": 11502 + }, + { + "epoch": 0.9485445699678403, + "grad_norm": 1.8124033212661743, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.768471360206604, + "num_tokens": 235939504.0, + "step": 11503 + }, + { + "epoch": 0.9486270305928919, + "grad_norm": 2.231065511703491, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7440938949584961, + "num_tokens": 235964326.0, + "step": 11504 + }, + { + "epoch": 0.9487094912179435, + "grad_norm": 2.197993516921997, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7253673076629639, + "num_tokens": 235988919.0, + "step": 11505 + }, + { + "epoch": 0.9487919518429949, + "grad_norm": 1.7643145322799683, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7444949746131897, + "num_tokens": 236023115.0, + "step": 11506 + }, + { + "epoch": 0.9488744124680465, + "grad_norm": 1.9906938076019287, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.744917631149292, + "num_tokens": 236052375.0, + "step": 11507 + }, + { + "epoch": 0.9489568730930981, + "grad_norm": 2.227668285369873, + "learning_rate": 1e-06, + "loss": 0.786, + "mean_token_accuracy": 0.7413241267204285, + "num_tokens": 236077661.0, + "step": 11508 + }, + { + "epoch": 0.9490393337181496, + "grad_norm": 2.0721607208251953, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.739102840423584, + "num_tokens": 236103212.0, + "step": 11509 + }, + { + "epoch": 0.9491217943432011, + "grad_norm": 1.9519000053405762, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7468081116676331, + "num_tokens": 236134994.0, + "step": 11510 + }, + { + "epoch": 0.9492042549682527, + "grad_norm": 2.148817539215088, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7147774696350098, + "num_tokens": 236162145.0, + "step": 11511 + }, + { + "epoch": 0.9492867155933042, + "grad_norm": 1.857323408126831, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7727503776550293, + "num_tokens": 236191556.0, + "step": 11512 + }, + { + "epoch": 0.9493691762183557, + "grad_norm": 1.9411213397979736, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7753902673721313, + "num_tokens": 236219678.0, + "step": 11513 + }, + { + "epoch": 0.9494516368434073, + "grad_norm": 1.7182847261428833, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7662355303764343, + "num_tokens": 236253311.0, + "step": 11514 + }, + { + "epoch": 0.9495340974684588, + "grad_norm": 1.933101773262024, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7298349142074585, + "num_tokens": 236282886.0, + "step": 11515 + }, + { + "epoch": 0.9496165580935103, + "grad_norm": 1.977683424949646, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7813695669174194, + "num_tokens": 236309235.0, + "step": 11516 + }, + { + "epoch": 0.9496990187185619, + "grad_norm": 1.840309977531433, + "learning_rate": 1e-06, + "loss": 0.771, + "mean_token_accuracy": 0.761050283908844, + "num_tokens": 236346174.0, + "step": 11517 + }, + { + "epoch": 0.9497814793436135, + "grad_norm": 2.4973292350769043, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7066333889961243, + "num_tokens": 236366705.0, + "step": 11518 + }, + { + "epoch": 0.9498639399686649, + "grad_norm": 1.8360222578048706, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7519258260726929, + "num_tokens": 236398406.0, + "step": 11519 + }, + { + "epoch": 0.9499464005937165, + "grad_norm": 1.8904573917388916, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7671804428100586, + "num_tokens": 236429226.0, + "step": 11520 + }, + { + "epoch": 0.9500288612187681, + "grad_norm": 2.207817792892456, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7479000091552734, + "num_tokens": 236454188.0, + "step": 11521 + }, + { + "epoch": 0.9501113218438195, + "grad_norm": 1.7949718236923218, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7718613743782043, + "num_tokens": 236486280.0, + "step": 11522 + }, + { + "epoch": 0.9501937824688711, + "grad_norm": 1.7215532064437866, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.741028368473053, + "num_tokens": 236524267.0, + "step": 11523 + }, + { + "epoch": 0.9502762430939227, + "grad_norm": 1.7936948537826538, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.7939527034759521, + "num_tokens": 236555668.0, + "step": 11524 + }, + { + "epoch": 0.9503587037189742, + "grad_norm": 2.287612199783325, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7237437963485718, + "num_tokens": 236579889.0, + "step": 11525 + }, + { + "epoch": 0.9504411643440257, + "grad_norm": 1.9409396648406982, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7773440480232239, + "num_tokens": 236606532.0, + "step": 11526 + }, + { + "epoch": 0.9505236249690773, + "grad_norm": 1.888783574104309, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.763526439666748, + "num_tokens": 236639190.0, + "step": 11527 + }, + { + "epoch": 0.9506060855941288, + "grad_norm": 2.2578301429748535, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7352813482284546, + "num_tokens": 236666220.0, + "step": 11528 + }, + { + "epoch": 0.9506885462191803, + "grad_norm": 1.8683967590332031, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.7841140031814575, + "num_tokens": 236694388.0, + "step": 11529 + }, + { + "epoch": 0.9507710068442319, + "grad_norm": 1.9759591817855835, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7836071252822876, + "num_tokens": 236719202.0, + "step": 11530 + }, + { + "epoch": 0.9508534674692835, + "grad_norm": 1.74403977394104, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7405831217765808, + "num_tokens": 236754166.0, + "step": 11531 + }, + { + "epoch": 0.9509359280943349, + "grad_norm": 1.7203630208969116, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7708015441894531, + "num_tokens": 236786161.0, + "step": 11532 + }, + { + "epoch": 0.9510183887193865, + "grad_norm": 1.8827747106552124, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7515211701393127, + "num_tokens": 236815331.0, + "step": 11533 + }, + { + "epoch": 0.9511008493444381, + "grad_norm": 2.096194267272949, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7384140491485596, + "num_tokens": 236843110.0, + "step": 11534 + }, + { + "epoch": 0.9511833099694895, + "grad_norm": 1.9717961549758911, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7354952096939087, + "num_tokens": 236873943.0, + "step": 11535 + }, + { + "epoch": 0.9512657705945411, + "grad_norm": 1.8513751029968262, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6935725808143616, + "num_tokens": 236910451.0, + "step": 11536 + }, + { + "epoch": 0.9513482312195927, + "grad_norm": 1.8043360710144043, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.77665114402771, + "num_tokens": 236943235.0, + "step": 11537 + }, + { + "epoch": 0.9514306918446441, + "grad_norm": 2.000596523284912, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7295326590538025, + "num_tokens": 236971882.0, + "step": 11538 + }, + { + "epoch": 0.9515131524696957, + "grad_norm": 2.1932878494262695, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7670725584030151, + "num_tokens": 236997331.0, + "step": 11539 + }, + { + "epoch": 0.9515956130947473, + "grad_norm": 2.1106746196746826, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7511100769042969, + "num_tokens": 237021937.0, + "step": 11540 + }, + { + "epoch": 0.9516780737197988, + "grad_norm": 1.903049349784851, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7129005789756775, + "num_tokens": 237053656.0, + "step": 11541 + }, + { + "epoch": 0.9517605343448503, + "grad_norm": 1.8672865629196167, + "learning_rate": 1e-06, + "loss": 0.5469, + "mean_token_accuracy": 0.8225328922271729, + "num_tokens": 237080920.0, + "step": 11542 + }, + { + "epoch": 0.9518429949699019, + "grad_norm": 1.816400170326233, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7720879912376404, + "num_tokens": 237114646.0, + "step": 11543 + }, + { + "epoch": 0.9519254555949535, + "grad_norm": 1.9123051166534424, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7748940587043762, + "num_tokens": 237143266.0, + "step": 11544 + }, + { + "epoch": 0.9520079162200049, + "grad_norm": 1.7841066122055054, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7661153674125671, + "num_tokens": 237175925.0, + "step": 11545 + }, + { + "epoch": 0.9520903768450565, + "grad_norm": 1.7718346118927002, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.764548659324646, + "num_tokens": 237209590.0, + "step": 11546 + }, + { + "epoch": 0.9521728374701081, + "grad_norm": 1.9430149793624878, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7308193445205688, + "num_tokens": 237241617.0, + "step": 11547 + }, + { + "epoch": 0.9522552980951595, + "grad_norm": 1.9317762851715088, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7528602480888367, + "num_tokens": 237271940.0, + "step": 11548 + }, + { + "epoch": 0.9523377587202111, + "grad_norm": 1.8083077669143677, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7760539650917053, + "num_tokens": 237306241.0, + "step": 11549 + }, + { + "epoch": 0.9524202193452627, + "grad_norm": 2.0635712146759033, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7679595351219177, + "num_tokens": 237330836.0, + "step": 11550 + }, + { + "epoch": 0.9525026799703141, + "grad_norm": 1.939563274383545, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7506414651870728, + "num_tokens": 237362726.0, + "step": 11551 + }, + { + "epoch": 0.9525851405953657, + "grad_norm": 2.097463607788086, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7357476949691772, + "num_tokens": 237393257.0, + "step": 11552 + }, + { + "epoch": 0.9526676012204173, + "grad_norm": 2.359405517578125, + "learning_rate": 1e-06, + "loss": 0.7799, + "mean_token_accuracy": 0.755059003829956, + "num_tokens": 237413644.0, + "step": 11553 + }, + { + "epoch": 0.9527500618454688, + "grad_norm": 1.9950382709503174, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7481083869934082, + "num_tokens": 237441736.0, + "step": 11554 + }, + { + "epoch": 0.9528325224705203, + "grad_norm": 1.6680243015289307, + "learning_rate": 1e-06, + "loss": 0.573, + "mean_token_accuracy": 0.8118559122085571, + "num_tokens": 237480444.0, + "step": 11555 + }, + { + "epoch": 0.9529149830955719, + "grad_norm": 1.8982040882110596, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7330006957054138, + "num_tokens": 237513229.0, + "step": 11556 + }, + { + "epoch": 0.9529974437206234, + "grad_norm": 2.1950087547302246, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.730704665184021, + "num_tokens": 237538628.0, + "step": 11557 + }, + { + "epoch": 0.9530799043456749, + "grad_norm": 1.9473096132278442, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7667971253395081, + "num_tokens": 237568322.0, + "step": 11558 + }, + { + "epoch": 0.9531623649707265, + "grad_norm": 2.184899091720581, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7591143846511841, + "num_tokens": 237593221.0, + "step": 11559 + }, + { + "epoch": 0.9532448255957781, + "grad_norm": 1.8750526905059814, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7612866163253784, + "num_tokens": 237623906.0, + "step": 11560 + }, + { + "epoch": 0.9533272862208295, + "grad_norm": 2.0251383781433105, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7467590570449829, + "num_tokens": 237653012.0, + "step": 11561 + }, + { + "epoch": 0.9534097468458811, + "grad_norm": 1.9935256242752075, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.782712996006012, + "num_tokens": 237680853.0, + "step": 11562 + }, + { + "epoch": 0.9534922074709327, + "grad_norm": 1.8060258626937866, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.78733229637146, + "num_tokens": 237714281.0, + "step": 11563 + }, + { + "epoch": 0.9535746680959841, + "grad_norm": 1.8504717350006104, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7710492610931396, + "num_tokens": 237743863.0, + "step": 11564 + }, + { + "epoch": 0.9536571287210357, + "grad_norm": 1.7681437730789185, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.749360203742981, + "num_tokens": 237780639.0, + "step": 11565 + }, + { + "epoch": 0.9537395893460873, + "grad_norm": 1.8221983909606934, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.789466381072998, + "num_tokens": 237811417.0, + "step": 11566 + }, + { + "epoch": 0.9538220499711387, + "grad_norm": 2.035611867904663, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.759611964225769, + "num_tokens": 237839877.0, + "step": 11567 + }, + { + "epoch": 0.9539045105961903, + "grad_norm": 2.007688522338867, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7272205948829651, + "num_tokens": 237869419.0, + "step": 11568 + }, + { + "epoch": 0.9539869712212419, + "grad_norm": 1.839294195175171, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7655216455459595, + "num_tokens": 237905338.0, + "step": 11569 + }, + { + "epoch": 0.9540694318462934, + "grad_norm": 1.7687737941741943, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7552852630615234, + "num_tokens": 237940407.0, + "step": 11570 + }, + { + "epoch": 0.9541518924713449, + "grad_norm": 2.011451244354248, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7473008632659912, + "num_tokens": 237967947.0, + "step": 11571 + }, + { + "epoch": 0.9542343530963965, + "grad_norm": 1.9796977043151855, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7483661770820618, + "num_tokens": 237997452.0, + "step": 11572 + }, + { + "epoch": 0.954316813721448, + "grad_norm": 1.9208813905715942, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.7550294399261475, + "num_tokens": 238027824.0, + "step": 11573 + }, + { + "epoch": 0.9543992743464995, + "grad_norm": 1.909510850906372, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7844161987304688, + "num_tokens": 238057340.0, + "step": 11574 + }, + { + "epoch": 0.9544817349715511, + "grad_norm": 1.7250653505325317, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7385997772216797, + "num_tokens": 238093883.0, + "step": 11575 + }, + { + "epoch": 0.9545641955966027, + "grad_norm": 2.013946056365967, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7247692346572876, + "num_tokens": 238121786.0, + "step": 11576 + }, + { + "epoch": 0.9546466562216541, + "grad_norm": 1.9595752954483032, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.771274745464325, + "num_tokens": 238151430.0, + "step": 11577 + }, + { + "epoch": 0.9547291168467057, + "grad_norm": 1.9645335674285889, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7449042201042175, + "num_tokens": 238180406.0, + "step": 11578 + }, + { + "epoch": 0.9548115774717573, + "grad_norm": 2.2078349590301514, + "learning_rate": 1e-06, + "loss": 0.7883, + "mean_token_accuracy": 0.7573193311691284, + "num_tokens": 238204246.0, + "step": 11579 + }, + { + "epoch": 0.9548940380968087, + "grad_norm": 1.7933297157287598, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.7888655066490173, + "num_tokens": 238235264.0, + "step": 11580 + }, + { + "epoch": 0.9549764987218603, + "grad_norm": 1.841805338859558, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7203969955444336, + "num_tokens": 238268042.0, + "step": 11581 + }, + { + "epoch": 0.9550589593469119, + "grad_norm": 1.9742072820663452, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7329550385475159, + "num_tokens": 238296726.0, + "step": 11582 + }, + { + "epoch": 0.9551414199719633, + "grad_norm": 2.696657657623291, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.7952394485473633, + "num_tokens": 238322674.0, + "step": 11583 + }, + { + "epoch": 0.9552238805970149, + "grad_norm": 2.0350425243377686, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7377549409866333, + "num_tokens": 238352741.0, + "step": 11584 + }, + { + "epoch": 0.9553063412220665, + "grad_norm": 2.0760979652404785, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7492920160293579, + "num_tokens": 238381161.0, + "step": 11585 + }, + { + "epoch": 0.955388801847118, + "grad_norm": 1.9961644411087036, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7357916831970215, + "num_tokens": 238413731.0, + "step": 11586 + }, + { + "epoch": 0.9554712624721695, + "grad_norm": 2.3875296115875244, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7050398588180542, + "num_tokens": 238434461.0, + "step": 11587 + }, + { + "epoch": 0.9555537230972211, + "grad_norm": 1.915619134902954, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7646366357803345, + "num_tokens": 238462863.0, + "step": 11588 + }, + { + "epoch": 0.9556361837222727, + "grad_norm": 2.18864369392395, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7682172060012817, + "num_tokens": 238485382.0, + "step": 11589 + }, + { + "epoch": 0.9557186443473241, + "grad_norm": 1.964892864227295, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7229965925216675, + "num_tokens": 238513344.0, + "step": 11590 + }, + { + "epoch": 0.9558011049723757, + "grad_norm": 1.9030539989471436, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7715214490890503, + "num_tokens": 238540550.0, + "step": 11591 + }, + { + "epoch": 0.9558835655974273, + "grad_norm": 2.0803942680358887, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7715461254119873, + "num_tokens": 238569544.0, + "step": 11592 + }, + { + "epoch": 0.9559660262224787, + "grad_norm": 2.1031949520111084, + "learning_rate": 1e-06, + "loss": 0.7852, + "mean_token_accuracy": 0.7461265325546265, + "num_tokens": 238594558.0, + "step": 11593 + }, + { + "epoch": 0.9560484868475303, + "grad_norm": 1.9358251094818115, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7162075042724609, + "num_tokens": 238627588.0, + "step": 11594 + }, + { + "epoch": 0.9561309474725819, + "grad_norm": 1.9340749979019165, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7799088358879089, + "num_tokens": 238657476.0, + "step": 11595 + }, + { + "epoch": 0.9562134080976333, + "grad_norm": 1.8485159873962402, + "learning_rate": 1e-06, + "loss": 0.6, + "mean_token_accuracy": 0.7939204573631287, + "num_tokens": 238686165.0, + "step": 11596 + }, + { + "epoch": 0.9562958687226849, + "grad_norm": 2.0278563499450684, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7332068085670471, + "num_tokens": 238716744.0, + "step": 11597 + }, + { + "epoch": 0.9563783293477365, + "grad_norm": 1.7728074789047241, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7757047414779663, + "num_tokens": 238752329.0, + "step": 11598 + }, + { + "epoch": 0.9564607899727879, + "grad_norm": 1.987346887588501, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7356704473495483, + "num_tokens": 238780873.0, + "step": 11599 + }, + { + "epoch": 0.9565432505978395, + "grad_norm": 1.6615736484527588, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7790092825889587, + "num_tokens": 238815316.0, + "step": 11600 + }, + { + "epoch": 0.9566257112228911, + "grad_norm": 1.9062130451202393, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7709103226661682, + "num_tokens": 238846574.0, + "step": 11601 + }, + { + "epoch": 0.9567081718479427, + "grad_norm": 2.6306302547454834, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7191177606582642, + "num_tokens": 238865265.0, + "step": 11602 + }, + { + "epoch": 0.9567906324729941, + "grad_norm": 1.8718522787094116, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7535750865936279, + "num_tokens": 238901408.0, + "step": 11603 + }, + { + "epoch": 0.9568730930980457, + "grad_norm": 1.9534063339233398, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7503832578659058, + "num_tokens": 238930561.0, + "step": 11604 + }, + { + "epoch": 0.9569555537230973, + "grad_norm": 1.8840901851654053, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.755524754524231, + "num_tokens": 238961349.0, + "step": 11605 + }, + { + "epoch": 0.9570380143481487, + "grad_norm": 1.9994087219238281, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.735745370388031, + "num_tokens": 238987426.0, + "step": 11606 + }, + { + "epoch": 0.9571204749732003, + "grad_norm": 2.2977936267852783, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7314325571060181, + "num_tokens": 239011619.0, + "step": 11607 + }, + { + "epoch": 0.9572029355982519, + "grad_norm": 2.032351016998291, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7476295232772827, + "num_tokens": 239039618.0, + "step": 11608 + }, + { + "epoch": 0.9572853962233033, + "grad_norm": 1.9190858602523804, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7717364430427551, + "num_tokens": 239069136.0, + "step": 11609 + }, + { + "epoch": 0.9573678568483549, + "grad_norm": 1.9158005714416504, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7573009133338928, + "num_tokens": 239099404.0, + "step": 11610 + }, + { + "epoch": 0.9574503174734065, + "grad_norm": 1.916157841682434, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7584373950958252, + "num_tokens": 239131555.0, + "step": 11611 + }, + { + "epoch": 0.9575327780984579, + "grad_norm": 1.7718158960342407, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7394452095031738, + "num_tokens": 239166981.0, + "step": 11612 + }, + { + "epoch": 0.9576152387235095, + "grad_norm": 2.0510942935943604, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7566121816635132, + "num_tokens": 239196979.0, + "step": 11613 + }, + { + "epoch": 0.9576976993485611, + "grad_norm": 2.079874038696289, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.709036111831665, + "num_tokens": 239223774.0, + "step": 11614 + }, + { + "epoch": 0.9577801599736127, + "grad_norm": 1.9584368467330933, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.736078143119812, + "num_tokens": 239255398.0, + "step": 11615 + }, + { + "epoch": 0.9578626205986641, + "grad_norm": 1.8199392557144165, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7862006425857544, + "num_tokens": 239288828.0, + "step": 11616 + }, + { + "epoch": 0.9579450812237157, + "grad_norm": 1.7295246124267578, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7349601984024048, + "num_tokens": 239326213.0, + "step": 11617 + }, + { + "epoch": 0.9580275418487673, + "grad_norm": 1.8784704208374023, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7729079723358154, + "num_tokens": 239358187.0, + "step": 11618 + }, + { + "epoch": 0.9581100024738187, + "grad_norm": 1.9019217491149902, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7634966969490051, + "num_tokens": 239387706.0, + "step": 11619 + }, + { + "epoch": 0.9581924630988703, + "grad_norm": 1.8395917415618896, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7533541917800903, + "num_tokens": 239419888.0, + "step": 11620 + }, + { + "epoch": 0.9582749237239219, + "grad_norm": 1.8562126159667969, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7406175136566162, + "num_tokens": 239451792.0, + "step": 11621 + }, + { + "epoch": 0.9583573843489733, + "grad_norm": 2.003563165664673, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.770043134689331, + "num_tokens": 239478442.0, + "step": 11622 + }, + { + "epoch": 0.9584398449740249, + "grad_norm": 2.0016369819641113, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.704237163066864, + "num_tokens": 239508793.0, + "step": 11623 + }, + { + "epoch": 0.9585223055990765, + "grad_norm": 1.9713010787963867, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7783116102218628, + "num_tokens": 239533135.0, + "step": 11624 + }, + { + "epoch": 0.9586047662241279, + "grad_norm": 1.630976676940918, + "learning_rate": 1e-06, + "loss": 0.5285, + "mean_token_accuracy": 0.820990264415741, + "num_tokens": 239569390.0, + "step": 11625 + }, + { + "epoch": 0.9586872268491795, + "grad_norm": 1.8610508441925049, + "learning_rate": 1e-06, + "loss": 0.7688, + "mean_token_accuracy": 0.7532390356063843, + "num_tokens": 239601581.0, + "step": 11626 + }, + { + "epoch": 0.9587696874742311, + "grad_norm": 2.0148346424102783, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7349774837493896, + "num_tokens": 239628571.0, + "step": 11627 + }, + { + "epoch": 0.9588521480992825, + "grad_norm": 2.1964023113250732, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7397456169128418, + "num_tokens": 239655778.0, + "step": 11628 + }, + { + "epoch": 0.9589346087243341, + "grad_norm": 2.0150039196014404, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7382380962371826, + "num_tokens": 239684538.0, + "step": 11629 + }, + { + "epoch": 0.9590170693493857, + "grad_norm": 2.0074942111968994, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7318411469459534, + "num_tokens": 239712385.0, + "step": 11630 + }, + { + "epoch": 0.9590995299744373, + "grad_norm": 1.907607913017273, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7734309434890747, + "num_tokens": 239739289.0, + "step": 11631 + }, + { + "epoch": 0.9591819905994887, + "grad_norm": 1.9143768548965454, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.7867960929870605, + "num_tokens": 239772069.0, + "step": 11632 + }, + { + "epoch": 0.9592644512245403, + "grad_norm": 1.8607969284057617, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7499622106552124, + "num_tokens": 239806088.0, + "step": 11633 + }, + { + "epoch": 0.9593469118495919, + "grad_norm": 1.8643492460250854, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7680953741073608, + "num_tokens": 239840176.0, + "step": 11634 + }, + { + "epoch": 0.9594293724746433, + "grad_norm": 1.9008151292800903, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7581515312194824, + "num_tokens": 239868858.0, + "step": 11635 + }, + { + "epoch": 0.9595118330996949, + "grad_norm": 1.707919716835022, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7682477235794067, + "num_tokens": 239902520.0, + "step": 11636 + }, + { + "epoch": 0.9595942937247465, + "grad_norm": 1.864327311515808, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7694920301437378, + "num_tokens": 239934196.0, + "step": 11637 + }, + { + "epoch": 0.9596767543497979, + "grad_norm": 1.7882411479949951, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.7468883991241455, + "num_tokens": 239966049.0, + "step": 11638 + }, + { + "epoch": 0.9597592149748495, + "grad_norm": 2.052212715148926, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7443682551383972, + "num_tokens": 239994531.0, + "step": 11639 + }, + { + "epoch": 0.9598416755999011, + "grad_norm": 1.9809973239898682, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7421436309814453, + "num_tokens": 240024649.0, + "step": 11640 + }, + { + "epoch": 0.9599241362249525, + "grad_norm": 2.110978364944458, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7513163089752197, + "num_tokens": 240050431.0, + "step": 11641 + }, + { + "epoch": 0.9600065968500041, + "grad_norm": 1.9096932411193848, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7238574028015137, + "num_tokens": 240083774.0, + "step": 11642 + }, + { + "epoch": 0.9600890574750557, + "grad_norm": 2.010650396347046, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7185477614402771, + "num_tokens": 240113096.0, + "step": 11643 + }, + { + "epoch": 0.9601715181001071, + "grad_norm": 1.9024937152862549, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7228338122367859, + "num_tokens": 240145097.0, + "step": 11644 + }, + { + "epoch": 0.9602539787251587, + "grad_norm": 1.9662225246429443, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7254797220230103, + "num_tokens": 240174572.0, + "step": 11645 + }, + { + "epoch": 0.9603364393502103, + "grad_norm": 1.9578582048416138, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7660578489303589, + "num_tokens": 240202168.0, + "step": 11646 + }, + { + "epoch": 0.9604188999752619, + "grad_norm": 1.8971848487854004, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7704013586044312, + "num_tokens": 240234711.0, + "step": 11647 + }, + { + "epoch": 0.9605013606003133, + "grad_norm": 2.0149989128112793, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7498816847801208, + "num_tokens": 240262895.0, + "step": 11648 + }, + { + "epoch": 0.9605838212253649, + "grad_norm": 1.8687913417816162, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7680151462554932, + "num_tokens": 240292242.0, + "step": 11649 + }, + { + "epoch": 0.9606662818504165, + "grad_norm": 1.7601325511932373, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7786400318145752, + "num_tokens": 240324656.0, + "step": 11650 + }, + { + "epoch": 0.9607487424754679, + "grad_norm": 1.795242190361023, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.7958747744560242, + "num_tokens": 240359288.0, + "step": 11651 + }, + { + "epoch": 0.9608312031005195, + "grad_norm": 1.8635019063949585, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7725564241409302, + "num_tokens": 240389032.0, + "step": 11652 + }, + { + "epoch": 0.9609136637255711, + "grad_norm": 1.8573753833770752, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7453245520591736, + "num_tokens": 240419369.0, + "step": 11653 + }, + { + "epoch": 0.9609961243506225, + "grad_norm": 1.9413630962371826, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7311000823974609, + "num_tokens": 240449933.0, + "step": 11654 + }, + { + "epoch": 0.9610785849756741, + "grad_norm": 1.805291771888733, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7674430012702942, + "num_tokens": 240483316.0, + "step": 11655 + }, + { + "epoch": 0.9611610456007257, + "grad_norm": 2.197014093399048, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7326542139053345, + "num_tokens": 240506779.0, + "step": 11656 + }, + { + "epoch": 0.9612435062257771, + "grad_norm": 2.0440962314605713, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7183966636657715, + "num_tokens": 240534910.0, + "step": 11657 + }, + { + "epoch": 0.9613259668508287, + "grad_norm": 1.8220072984695435, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.773734450340271, + "num_tokens": 240565698.0, + "step": 11658 + }, + { + "epoch": 0.9614084274758803, + "grad_norm": 1.5441360473632812, + "learning_rate": 1e-06, + "loss": 0.4652, + "mean_token_accuracy": 0.8284724950790405, + "num_tokens": 240600704.0, + "step": 11659 + }, + { + "epoch": 0.9614908881009318, + "grad_norm": 2.262378454208374, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7275176048278809, + "num_tokens": 240624706.0, + "step": 11660 + }, + { + "epoch": 0.9615733487259833, + "grad_norm": 2.04695463180542, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7592782974243164, + "num_tokens": 240652631.0, + "step": 11661 + }, + { + "epoch": 0.9616558093510349, + "grad_norm": 2.1515843868255615, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7498694062232971, + "num_tokens": 240677786.0, + "step": 11662 + }, + { + "epoch": 0.9617382699760865, + "grad_norm": 2.112614870071411, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7194947004318237, + "num_tokens": 240705190.0, + "step": 11663 + }, + { + "epoch": 0.9618207306011379, + "grad_norm": 1.866228699684143, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7638946771621704, + "num_tokens": 240738385.0, + "step": 11664 + }, + { + "epoch": 0.9619031912261895, + "grad_norm": 1.9105137586593628, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7238802313804626, + "num_tokens": 240769824.0, + "step": 11665 + }, + { + "epoch": 0.9619856518512411, + "grad_norm": 2.1011643409729004, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7157832384109497, + "num_tokens": 240799110.0, + "step": 11666 + }, + { + "epoch": 0.9620681124762925, + "grad_norm": 1.9461418390274048, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7738394737243652, + "num_tokens": 240830446.0, + "step": 11667 + }, + { + "epoch": 0.9621505731013441, + "grad_norm": 1.8843244314193726, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7753719091415405, + "num_tokens": 240859400.0, + "step": 11668 + }, + { + "epoch": 0.9622330337263957, + "grad_norm": 2.277958631515503, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7410280704498291, + "num_tokens": 240883326.0, + "step": 11669 + }, + { + "epoch": 0.9623154943514471, + "grad_norm": 1.9983336925506592, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.7544664144515991, + "num_tokens": 240910913.0, + "step": 11670 + }, + { + "epoch": 0.9623979549764987, + "grad_norm": 2.125704765319824, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7483940720558167, + "num_tokens": 240935314.0, + "step": 11671 + }, + { + "epoch": 0.9624804156015503, + "grad_norm": 2.0232818126678467, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7716006636619568, + "num_tokens": 240963176.0, + "step": 11672 + }, + { + "epoch": 0.9625628762266017, + "grad_norm": 1.7156280279159546, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.7832115888595581, + "num_tokens": 240994315.0, + "step": 11673 + }, + { + "epoch": 0.9626453368516533, + "grad_norm": 2.0263946056365967, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7717044353485107, + "num_tokens": 241020965.0, + "step": 11674 + }, + { + "epoch": 0.9627277974767049, + "grad_norm": 1.9132760763168335, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.7689948678016663, + "num_tokens": 241050678.0, + "step": 11675 + }, + { + "epoch": 0.9628102581017565, + "grad_norm": 1.7364591360092163, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7528504729270935, + "num_tokens": 241084180.0, + "step": 11676 + }, + { + "epoch": 0.9628927187268079, + "grad_norm": 1.6506320238113403, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7719999551773071, + "num_tokens": 241122127.0, + "step": 11677 + }, + { + "epoch": 0.9629751793518595, + "grad_norm": 1.8979156017303467, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7769830822944641, + "num_tokens": 241150346.0, + "step": 11678 + }, + { + "epoch": 0.9630576399769111, + "grad_norm": 1.8037954568862915, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7572096586227417, + "num_tokens": 241185164.0, + "step": 11679 + }, + { + "epoch": 0.9631401006019625, + "grad_norm": 1.7976473569869995, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7572343945503235, + "num_tokens": 241219853.0, + "step": 11680 + }, + { + "epoch": 0.9632225612270141, + "grad_norm": 1.9046558141708374, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7659503817558289, + "num_tokens": 241252464.0, + "step": 11681 + }, + { + "epoch": 0.9633050218520657, + "grad_norm": 1.8608722686767578, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7475708723068237, + "num_tokens": 241283224.0, + "step": 11682 + }, + { + "epoch": 0.9633874824771171, + "grad_norm": 1.6935877799987793, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7421392798423767, + "num_tokens": 241318335.0, + "step": 11683 + }, + { + "epoch": 0.9634699431021687, + "grad_norm": 1.6806732416152954, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7366178035736084, + "num_tokens": 241353175.0, + "step": 11684 + }, + { + "epoch": 0.9635524037272203, + "grad_norm": 2.005486249923706, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7271867990493774, + "num_tokens": 241381195.0, + "step": 11685 + }, + { + "epoch": 0.9636348643522717, + "grad_norm": 1.868273138999939, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7413946390151978, + "num_tokens": 241414817.0, + "step": 11686 + }, + { + "epoch": 0.9637173249773233, + "grad_norm": 1.9326320886611938, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7406421899795532, + "num_tokens": 241448083.0, + "step": 11687 + }, + { + "epoch": 0.9637997856023749, + "grad_norm": 2.1298274993896484, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7563648223876953, + "num_tokens": 241474577.0, + "step": 11688 + }, + { + "epoch": 0.9638822462274264, + "grad_norm": 1.8338176012039185, + "learning_rate": 1e-06, + "loss": 0.5812, + "mean_token_accuracy": 0.8084635734558105, + "num_tokens": 241503418.0, + "step": 11689 + }, + { + "epoch": 0.9639647068524779, + "grad_norm": 1.9799805879592896, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7528764605522156, + "num_tokens": 241529996.0, + "step": 11690 + }, + { + "epoch": 0.9640471674775295, + "grad_norm": 1.9080229997634888, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7308002710342407, + "num_tokens": 241561258.0, + "step": 11691 + }, + { + "epoch": 0.9641296281025811, + "grad_norm": 1.859902262687683, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.730926513671875, + "num_tokens": 241595616.0, + "step": 11692 + }, + { + "epoch": 0.9642120887276325, + "grad_norm": 1.838201642036438, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.747434139251709, + "num_tokens": 241631377.0, + "step": 11693 + }, + { + "epoch": 0.9642945493526841, + "grad_norm": 2.0987422466278076, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7704216241836548, + "num_tokens": 241654064.0, + "step": 11694 + }, + { + "epoch": 0.9643770099777357, + "grad_norm": 2.1093387603759766, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.770677924156189, + "num_tokens": 241679590.0, + "step": 11695 + }, + { + "epoch": 0.9644594706027871, + "grad_norm": 1.7957216501235962, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7714390754699707, + "num_tokens": 241712912.0, + "step": 11696 + }, + { + "epoch": 0.9645419312278387, + "grad_norm": 2.054461717605591, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7558801770210266, + "num_tokens": 241742445.0, + "step": 11697 + }, + { + "epoch": 0.9646243918528903, + "grad_norm": 2.0498416423797607, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7168397903442383, + "num_tokens": 241771151.0, + "step": 11698 + }, + { + "epoch": 0.9647068524779417, + "grad_norm": 1.947296142578125, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7609801292419434, + "num_tokens": 241801266.0, + "step": 11699 + }, + { + "epoch": 0.9647893131029933, + "grad_norm": 1.7609632015228271, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.757560133934021, + "num_tokens": 241833974.0, + "step": 11700 + }, + { + "epoch": 0.9648717737280449, + "grad_norm": 2.0396947860717773, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7242127060890198, + "num_tokens": 241863040.0, + "step": 11701 + }, + { + "epoch": 0.9649542343530964, + "grad_norm": 1.9865174293518066, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7267123460769653, + "num_tokens": 241892002.0, + "step": 11702 + }, + { + "epoch": 0.9650366949781479, + "grad_norm": 1.7825126647949219, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7678712606430054, + "num_tokens": 241929918.0, + "step": 11703 + }, + { + "epoch": 0.9651191556031995, + "grad_norm": 1.9385639429092407, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7491436004638672, + "num_tokens": 241961462.0, + "step": 11704 + }, + { + "epoch": 0.965201616228251, + "grad_norm": 1.7305150032043457, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7730225920677185, + "num_tokens": 241994227.0, + "step": 11705 + }, + { + "epoch": 0.9652840768533025, + "grad_norm": 2.0248525142669678, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7341467142105103, + "num_tokens": 242022574.0, + "step": 11706 + }, + { + "epoch": 0.9653665374783541, + "grad_norm": 2.007237434387207, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.74955815076828, + "num_tokens": 242051838.0, + "step": 11707 + }, + { + "epoch": 0.9654489981034057, + "grad_norm": 1.7633546590805054, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7744854092597961, + "num_tokens": 242087616.0, + "step": 11708 + }, + { + "epoch": 0.9655314587284571, + "grad_norm": 1.7209330797195435, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7788550853729248, + "num_tokens": 242120318.0, + "step": 11709 + }, + { + "epoch": 0.9656139193535087, + "grad_norm": 1.8011000156402588, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7492512464523315, + "num_tokens": 242152869.0, + "step": 11710 + }, + { + "epoch": 0.9656963799785603, + "grad_norm": 1.8224788904190063, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7728358507156372, + "num_tokens": 242188140.0, + "step": 11711 + }, + { + "epoch": 0.9657788406036117, + "grad_norm": 1.7132437229156494, + "learning_rate": 1e-06, + "loss": 0.5598, + "mean_token_accuracy": 0.8150391578674316, + "num_tokens": 242222529.0, + "step": 11712 + }, + { + "epoch": 0.9658613012286633, + "grad_norm": 1.8528271913528442, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.752210795879364, + "num_tokens": 242254114.0, + "step": 11713 + }, + { + "epoch": 0.9659437618537149, + "grad_norm": 1.6551519632339478, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7870745658874512, + "num_tokens": 242291664.0, + "step": 11714 + }, + { + "epoch": 0.9660262224787663, + "grad_norm": 1.9046391248703003, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7641749382019043, + "num_tokens": 242320963.0, + "step": 11715 + }, + { + "epoch": 0.9661086831038179, + "grad_norm": 2.0239474773406982, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6992282271385193, + "num_tokens": 242351122.0, + "step": 11716 + }, + { + "epoch": 0.9661911437288695, + "grad_norm": 1.8880008459091187, + "learning_rate": 1e-06, + "loss": 0.6046, + "mean_token_accuracy": 0.7921802997589111, + "num_tokens": 242379914.0, + "step": 11717 + }, + { + "epoch": 0.966273604353921, + "grad_norm": 1.825742244720459, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7603720426559448, + "num_tokens": 242412371.0, + "step": 11718 + }, + { + "epoch": 0.9663560649789725, + "grad_norm": 1.8122791051864624, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7393261790275574, + "num_tokens": 242444117.0, + "step": 11719 + }, + { + "epoch": 0.9664385256040241, + "grad_norm": 1.8704969882965088, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7574459314346313, + "num_tokens": 242476377.0, + "step": 11720 + }, + { + "epoch": 0.9665209862290756, + "grad_norm": 1.9033546447753906, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7496988773345947, + "num_tokens": 242512034.0, + "step": 11721 + }, + { + "epoch": 0.9666034468541271, + "grad_norm": 1.794734001159668, + "learning_rate": 1e-06, + "loss": 0.7498, + "mean_token_accuracy": 0.7610765695571899, + "num_tokens": 242545466.0, + "step": 11722 + }, + { + "epoch": 0.9666859074791787, + "grad_norm": 2.085371971130371, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7207894325256348, + "num_tokens": 242571884.0, + "step": 11723 + }, + { + "epoch": 0.9667683681042303, + "grad_norm": 2.2031049728393555, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.786770224571228, + "num_tokens": 242597751.0, + "step": 11724 + }, + { + "epoch": 0.9668508287292817, + "grad_norm": 1.7688300609588623, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7615395188331604, + "num_tokens": 242632362.0, + "step": 11725 + }, + { + "epoch": 0.9669332893543333, + "grad_norm": 1.8831157684326172, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7067135572433472, + "num_tokens": 242664460.0, + "step": 11726 + }, + { + "epoch": 0.9670157499793849, + "grad_norm": 2.0034239292144775, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7190842628479004, + "num_tokens": 242692653.0, + "step": 11727 + }, + { + "epoch": 0.9670982106044363, + "grad_norm": 1.9669700860977173, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7273896932601929, + "num_tokens": 242719365.0, + "step": 11728 + }, + { + "epoch": 0.9671806712294879, + "grad_norm": 1.8164336681365967, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.7898446321487427, + "num_tokens": 242749720.0, + "step": 11729 + }, + { + "epoch": 0.9672631318545395, + "grad_norm": 2.296046733856201, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7277283072471619, + "num_tokens": 242772368.0, + "step": 11730 + }, + { + "epoch": 0.967345592479591, + "grad_norm": 1.900639295578003, + "learning_rate": 1e-06, + "loss": 0.6239, + "mean_token_accuracy": 0.7904617786407471, + "num_tokens": 242803153.0, + "step": 11731 + }, + { + "epoch": 0.9674280531046425, + "grad_norm": 1.9690319299697876, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7257575988769531, + "num_tokens": 242830912.0, + "step": 11732 + }, + { + "epoch": 0.9675105137296941, + "grad_norm": 1.8607548475265503, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7486726641654968, + "num_tokens": 242861609.0, + "step": 11733 + }, + { + "epoch": 0.9675929743547456, + "grad_norm": 1.7704778909683228, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7765275239944458, + "num_tokens": 242894033.0, + "step": 11734 + }, + { + "epoch": 0.9676754349797971, + "grad_norm": 1.7058773040771484, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7870341539382935, + "num_tokens": 242929404.0, + "step": 11735 + }, + { + "epoch": 0.9677578956048487, + "grad_norm": 1.9949151277542114, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7499305009841919, + "num_tokens": 242957968.0, + "step": 11736 + }, + { + "epoch": 0.9678403562299002, + "grad_norm": 2.1010823249816895, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7284278869628906, + "num_tokens": 242988145.0, + "step": 11737 + }, + { + "epoch": 0.9679228168549517, + "grad_norm": 1.867409586906433, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7560296654701233, + "num_tokens": 243021490.0, + "step": 11738 + }, + { + "epoch": 0.9680052774800033, + "grad_norm": 1.647129774093628, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7771348357200623, + "num_tokens": 243059172.0, + "step": 11739 + }, + { + "epoch": 0.9680877381050549, + "grad_norm": 1.9373210668563843, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7508443593978882, + "num_tokens": 243089198.0, + "step": 11740 + }, + { + "epoch": 0.9681701987301063, + "grad_norm": 2.0358786582946777, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7511016130447388, + "num_tokens": 243116292.0, + "step": 11741 + }, + { + "epoch": 0.9682526593551579, + "grad_norm": 2.027228593826294, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7502769231796265, + "num_tokens": 243143012.0, + "step": 11742 + }, + { + "epoch": 0.9683351199802095, + "grad_norm": 2.0163862705230713, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7645779848098755, + "num_tokens": 243169611.0, + "step": 11743 + }, + { + "epoch": 0.968417580605261, + "grad_norm": 1.9139840602874756, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7423930764198303, + "num_tokens": 243200832.0, + "step": 11744 + }, + { + "epoch": 0.9685000412303125, + "grad_norm": 1.676306962966919, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.7902243137359619, + "num_tokens": 243235262.0, + "step": 11745 + }, + { + "epoch": 0.9685825018553641, + "grad_norm": 2.2028558254241943, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7448418140411377, + "num_tokens": 243257897.0, + "step": 11746 + }, + { + "epoch": 0.9686649624804156, + "grad_norm": 1.8790456056594849, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7733930349349976, + "num_tokens": 243288467.0, + "step": 11747 + }, + { + "epoch": 0.9687474231054671, + "grad_norm": 1.8113189935684204, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7410449981689453, + "num_tokens": 243323267.0, + "step": 11748 + }, + { + "epoch": 0.9688298837305187, + "grad_norm": 1.8231085538864136, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7913379669189453, + "num_tokens": 243354846.0, + "step": 11749 + }, + { + "epoch": 0.9689123443555702, + "grad_norm": 1.9056531190872192, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7062469720840454, + "num_tokens": 243386780.0, + "step": 11750 + }, + { + "epoch": 0.9689948049806217, + "grad_norm": 1.8899890184402466, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7875394225120544, + "num_tokens": 243414126.0, + "step": 11751 + }, + { + "epoch": 0.9690772656056733, + "grad_norm": 2.441800355911255, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7238917946815491, + "num_tokens": 243437624.0, + "step": 11752 + }, + { + "epoch": 0.9691597262307249, + "grad_norm": 1.947268009185791, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.747516393661499, + "num_tokens": 243465882.0, + "step": 11753 + }, + { + "epoch": 0.9692421868557763, + "grad_norm": 1.8390578031539917, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7487364411354065, + "num_tokens": 243499959.0, + "step": 11754 + }, + { + "epoch": 0.9693246474808279, + "grad_norm": 1.8767818212509155, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7390457391738892, + "num_tokens": 243532022.0, + "step": 11755 + }, + { + "epoch": 0.9694071081058795, + "grad_norm": 2.0352797508239746, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7070048451423645, + "num_tokens": 243559426.0, + "step": 11756 + }, + { + "epoch": 0.969489568730931, + "grad_norm": 2.051647424697876, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7432857751846313, + "num_tokens": 243584793.0, + "step": 11757 + }, + { + "epoch": 0.9695720293559825, + "grad_norm": 1.8435758352279663, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7484028339385986, + "num_tokens": 243616050.0, + "step": 11758 + }, + { + "epoch": 0.9696544899810341, + "grad_norm": 1.7694180011749268, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7254058718681335, + "num_tokens": 243651502.0, + "step": 11759 + }, + { + "epoch": 0.9697369506060856, + "grad_norm": 1.8720836639404297, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.7618204355239868, + "num_tokens": 243683772.0, + "step": 11760 + }, + { + "epoch": 0.9698194112311371, + "grad_norm": 2.164112091064453, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.718498706817627, + "num_tokens": 243710393.0, + "step": 11761 + }, + { + "epoch": 0.9699018718561887, + "grad_norm": 2.1126205921173096, + "learning_rate": 1e-06, + "loss": 0.7746, + "mean_token_accuracy": 0.742639422416687, + "num_tokens": 243735769.0, + "step": 11762 + }, + { + "epoch": 0.9699843324812402, + "grad_norm": 1.9684245586395264, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7659406661987305, + "num_tokens": 243762837.0, + "step": 11763 + }, + { + "epoch": 0.9700667931062917, + "grad_norm": 2.058401346206665, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7305656671524048, + "num_tokens": 243789088.0, + "step": 11764 + }, + { + "epoch": 0.9701492537313433, + "grad_norm": 2.0285561084747314, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7804243564605713, + "num_tokens": 243814474.0, + "step": 11765 + }, + { + "epoch": 0.9702317143563948, + "grad_norm": 1.8063864707946777, + "learning_rate": 1e-06, + "loss": 0.5768, + "mean_token_accuracy": 0.8052884936332703, + "num_tokens": 243844736.0, + "step": 11766 + }, + { + "epoch": 0.9703141749814463, + "grad_norm": 1.7866092920303345, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7736276984214783, + "num_tokens": 243877977.0, + "step": 11767 + }, + { + "epoch": 0.9703966356064979, + "grad_norm": 1.9458357095718384, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7373684644699097, + "num_tokens": 243907556.0, + "step": 11768 + }, + { + "epoch": 0.9704790962315495, + "grad_norm": 2.0461390018463135, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6938298940658569, + "num_tokens": 243936680.0, + "step": 11769 + }, + { + "epoch": 0.970561556856601, + "grad_norm": 1.8831205368041992, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7364621162414551, + "num_tokens": 243968633.0, + "step": 11770 + }, + { + "epoch": 0.9706440174816525, + "grad_norm": 1.8106584548950195, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7320143580436707, + "num_tokens": 244006402.0, + "step": 11771 + }, + { + "epoch": 0.9707264781067041, + "grad_norm": 1.996406078338623, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7518455982208252, + "num_tokens": 244036135.0, + "step": 11772 + }, + { + "epoch": 0.9708089387317556, + "grad_norm": 1.8406308889389038, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.7839610576629639, + "num_tokens": 244064626.0, + "step": 11773 + }, + { + "epoch": 0.9708913993568071, + "grad_norm": 1.865172266960144, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7643629312515259, + "num_tokens": 244097112.0, + "step": 11774 + }, + { + "epoch": 0.9709738599818587, + "grad_norm": 1.9499170780181885, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7690355777740479, + "num_tokens": 244128996.0, + "step": 11775 + }, + { + "epoch": 0.9710563206069102, + "grad_norm": 1.8083667755126953, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7858189344406128, + "num_tokens": 244160819.0, + "step": 11776 + }, + { + "epoch": 0.9711387812319617, + "grad_norm": 1.9118410348892212, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7257324457168579, + "num_tokens": 244193167.0, + "step": 11777 + }, + { + "epoch": 0.9712212418570133, + "grad_norm": 1.9350433349609375, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7124039530754089, + "num_tokens": 244222853.0, + "step": 11778 + }, + { + "epoch": 0.9713037024820648, + "grad_norm": 1.9541447162628174, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.8042476177215576, + "num_tokens": 244247903.0, + "step": 11779 + }, + { + "epoch": 0.9713861631071163, + "grad_norm": 1.8557294607162476, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7483962774276733, + "num_tokens": 244281699.0, + "step": 11780 + }, + { + "epoch": 0.9714686237321679, + "grad_norm": 2.0092577934265137, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7654827833175659, + "num_tokens": 244308608.0, + "step": 11781 + }, + { + "epoch": 0.9715510843572194, + "grad_norm": 2.1210789680480957, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7643234729766846, + "num_tokens": 244335922.0, + "step": 11782 + }, + { + "epoch": 0.971633544982271, + "grad_norm": 1.7292428016662598, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7655700445175171, + "num_tokens": 244372722.0, + "step": 11783 + }, + { + "epoch": 0.9717160056073225, + "grad_norm": 1.8664506673812866, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7141098976135254, + "num_tokens": 244404785.0, + "step": 11784 + }, + { + "epoch": 0.9717984662323741, + "grad_norm": 1.8610299825668335, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7527352571487427, + "num_tokens": 244434185.0, + "step": 11785 + }, + { + "epoch": 0.9718809268574256, + "grad_norm": 1.9453068971633911, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.7658507823944092, + "num_tokens": 244463188.0, + "step": 11786 + }, + { + "epoch": 0.9719633874824771, + "grad_norm": 1.94106125831604, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7654938101768494, + "num_tokens": 244494270.0, + "step": 11787 + }, + { + "epoch": 0.9720458481075287, + "grad_norm": 2.052541732788086, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7283753752708435, + "num_tokens": 244523598.0, + "step": 11788 + }, + { + "epoch": 0.9721283087325802, + "grad_norm": 1.9813321828842163, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7334601283073425, + "num_tokens": 244554224.0, + "step": 11789 + }, + { + "epoch": 0.9722107693576317, + "grad_norm": 1.8445509672164917, + "learning_rate": 1e-06, + "loss": 0.5772, + "mean_token_accuracy": 0.8081713914871216, + "num_tokens": 244581274.0, + "step": 11790 + }, + { + "epoch": 0.9722932299826833, + "grad_norm": 1.9310343265533447, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7303287386894226, + "num_tokens": 244609458.0, + "step": 11791 + }, + { + "epoch": 0.9723756906077348, + "grad_norm": 1.9401618242263794, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7161561846733093, + "num_tokens": 244639369.0, + "step": 11792 + }, + { + "epoch": 0.9724581512327863, + "grad_norm": 1.7891007661819458, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7822036743164062, + "num_tokens": 244674659.0, + "step": 11793 + }, + { + "epoch": 0.9725406118578379, + "grad_norm": 1.9597179889678955, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.8010502457618713, + "num_tokens": 244702504.0, + "step": 11794 + }, + { + "epoch": 0.9726230724828894, + "grad_norm": 1.9700398445129395, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7435380816459656, + "num_tokens": 244731879.0, + "step": 11795 + }, + { + "epoch": 0.972705533107941, + "grad_norm": 2.0881805419921875, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.7901140451431274, + "num_tokens": 244759608.0, + "step": 11796 + }, + { + "epoch": 0.9727879937329925, + "grad_norm": 1.8891807794570923, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7733599543571472, + "num_tokens": 244791641.0, + "step": 11797 + }, + { + "epoch": 0.972870454358044, + "grad_norm": 2.0058131217956543, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7097722291946411, + "num_tokens": 244823013.0, + "step": 11798 + }, + { + "epoch": 0.9729529149830956, + "grad_norm": 1.9569642543792725, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.740074098110199, + "num_tokens": 244851756.0, + "step": 11799 + }, + { + "epoch": 0.9730353756081471, + "grad_norm": 1.8798846006393433, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7776230573654175, + "num_tokens": 244882043.0, + "step": 11800 + }, + { + "epoch": 0.9731178362331987, + "grad_norm": 2.0456671714782715, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.7829700708389282, + "num_tokens": 244910009.0, + "step": 11801 + }, + { + "epoch": 0.9732002968582502, + "grad_norm": 1.9211422204971313, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7616334557533264, + "num_tokens": 244938106.0, + "step": 11802 + }, + { + "epoch": 0.9732827574833017, + "grad_norm": 1.761510968208313, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7401933073997498, + "num_tokens": 244972575.0, + "step": 11803 + }, + { + "epoch": 0.9733652181083533, + "grad_norm": 2.002762794494629, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7634522318840027, + "num_tokens": 245002316.0, + "step": 11804 + }, + { + "epoch": 0.9734476787334048, + "grad_norm": 1.7291961908340454, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.6986597180366516, + "num_tokens": 245040452.0, + "step": 11805 + }, + { + "epoch": 0.9735301393584563, + "grad_norm": 1.8799570798873901, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7555623650550842, + "num_tokens": 245069772.0, + "step": 11806 + }, + { + "epoch": 0.9736125999835079, + "grad_norm": 1.689270257949829, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7370306253433228, + "num_tokens": 245107458.0, + "step": 11807 + }, + { + "epoch": 0.9736950606085594, + "grad_norm": 2.202242612838745, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7380207180976868, + "num_tokens": 245131511.0, + "step": 11808 + }, + { + "epoch": 0.9737775212336109, + "grad_norm": 1.9938701391220093, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7739854454994202, + "num_tokens": 245158810.0, + "step": 11809 + }, + { + "epoch": 0.9738599818586625, + "grad_norm": 2.125213623046875, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7691545486450195, + "num_tokens": 245183066.0, + "step": 11810 + }, + { + "epoch": 0.973942442483714, + "grad_norm": 1.6946189403533936, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7753406763076782, + "num_tokens": 245221601.0, + "step": 11811 + }, + { + "epoch": 0.9740249031087655, + "grad_norm": 1.8223800659179688, + "learning_rate": 1e-06, + "loss": 0.5514, + "mean_token_accuracy": 0.813499927520752, + "num_tokens": 245249361.0, + "step": 11812 + }, + { + "epoch": 0.9741073637338171, + "grad_norm": 1.832908034324646, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7600356936454773, + "num_tokens": 245282195.0, + "step": 11813 + }, + { + "epoch": 0.9741898243588687, + "grad_norm": 2.0340702533721924, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7557438611984253, + "num_tokens": 245310297.0, + "step": 11814 + }, + { + "epoch": 0.9742722849839202, + "grad_norm": 1.8523832559585571, + "learning_rate": 1e-06, + "loss": 0.7767, + "mean_token_accuracy": 0.7448457479476929, + "num_tokens": 245342357.0, + "step": 11815 + }, + { + "epoch": 0.9743547456089717, + "grad_norm": 1.8208184242248535, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7579890489578247, + "num_tokens": 245372506.0, + "step": 11816 + }, + { + "epoch": 0.9744372062340233, + "grad_norm": 1.965691328048706, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.739500105381012, + "num_tokens": 245402529.0, + "step": 11817 + }, + { + "epoch": 0.9745196668590748, + "grad_norm": 1.8321950435638428, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7389319539070129, + "num_tokens": 245433894.0, + "step": 11818 + }, + { + "epoch": 0.9746021274841263, + "grad_norm": 1.9750124216079712, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7434661388397217, + "num_tokens": 245462465.0, + "step": 11819 + }, + { + "epoch": 0.9746845881091779, + "grad_norm": 1.8882116079330444, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7710851430892944, + "num_tokens": 245488157.0, + "step": 11820 + }, + { + "epoch": 0.9747670487342294, + "grad_norm": 1.5591511726379395, + "learning_rate": 1e-06, + "loss": 0.5387, + "mean_token_accuracy": 0.8236343264579773, + "num_tokens": 245523698.0, + "step": 11821 + }, + { + "epoch": 0.9748495093592809, + "grad_norm": 1.7762174606323242, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.788301944732666, + "num_tokens": 245556452.0, + "step": 11822 + }, + { + "epoch": 0.9749319699843325, + "grad_norm": 2.2562546730041504, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7469024658203125, + "num_tokens": 245578526.0, + "step": 11823 + }, + { + "epoch": 0.975014430609384, + "grad_norm": 1.9833987951278687, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7727477550506592, + "num_tokens": 245605427.0, + "step": 11824 + }, + { + "epoch": 0.9750968912344355, + "grad_norm": 1.9307926893234253, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7161861658096313, + "num_tokens": 245641638.0, + "step": 11825 + }, + { + "epoch": 0.9751793518594871, + "grad_norm": 1.908341407775879, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.7530317306518555, + "num_tokens": 245671242.0, + "step": 11826 + }, + { + "epoch": 0.9752618124845386, + "grad_norm": 2.1995277404785156, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7361416816711426, + "num_tokens": 245696083.0, + "step": 11827 + }, + { + "epoch": 0.9753442731095902, + "grad_norm": 1.9015579223632812, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7499516010284424, + "num_tokens": 245730257.0, + "step": 11828 + }, + { + "epoch": 0.9754267337346417, + "grad_norm": 2.01068377494812, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7508500814437866, + "num_tokens": 245759547.0, + "step": 11829 + }, + { + "epoch": 0.9755091943596933, + "grad_norm": 1.6395589113235474, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7421209812164307, + "num_tokens": 245800020.0, + "step": 11830 + }, + { + "epoch": 0.9755916549847448, + "grad_norm": 1.9359508752822876, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7673710584640503, + "num_tokens": 245827910.0, + "step": 11831 + }, + { + "epoch": 0.9756741156097963, + "grad_norm": 2.024766683578491, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7420487999916077, + "num_tokens": 245854561.0, + "step": 11832 + }, + { + "epoch": 0.9757565762348479, + "grad_norm": 1.9953770637512207, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7378466129302979, + "num_tokens": 245883840.0, + "step": 11833 + }, + { + "epoch": 0.9758390368598994, + "grad_norm": 2.158073663711548, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7245612740516663, + "num_tokens": 245911261.0, + "step": 11834 + }, + { + "epoch": 0.9759214974849509, + "grad_norm": 2.021361827850342, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7305305600166321, + "num_tokens": 245940393.0, + "step": 11835 + }, + { + "epoch": 0.9760039581100025, + "grad_norm": 2.0022125244140625, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.744342029094696, + "num_tokens": 245968280.0, + "step": 11836 + }, + { + "epoch": 0.976086418735054, + "grad_norm": 1.9661859273910522, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.741635262966156, + "num_tokens": 245997378.0, + "step": 11837 + }, + { + "epoch": 0.9761688793601055, + "grad_norm": 1.9927809238433838, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7566925287246704, + "num_tokens": 246025288.0, + "step": 11838 + }, + { + "epoch": 0.9762513399851571, + "grad_norm": 2.2885799407958984, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7424019575119019, + "num_tokens": 246048616.0, + "step": 11839 + }, + { + "epoch": 0.9763338006102086, + "grad_norm": 2.1199533939361572, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7122154831886292, + "num_tokens": 246075819.0, + "step": 11840 + }, + { + "epoch": 0.9764162612352602, + "grad_norm": 2.0540337562561035, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7216373682022095, + "num_tokens": 246103352.0, + "step": 11841 + }, + { + "epoch": 0.9764987218603117, + "grad_norm": 1.8920786380767822, + "learning_rate": 1e-06, + "loss": 0.7723, + "mean_token_accuracy": 0.7612248659133911, + "num_tokens": 246132968.0, + "step": 11842 + }, + { + "epoch": 0.9765811824853632, + "grad_norm": 1.971782922744751, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7756475806236267, + "num_tokens": 246163417.0, + "step": 11843 + }, + { + "epoch": 0.9766636431104148, + "grad_norm": 1.7819675207138062, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.7693555355072021, + "num_tokens": 246197981.0, + "step": 11844 + }, + { + "epoch": 0.9767461037354663, + "grad_norm": 2.0478053092956543, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.7844671010971069, + "num_tokens": 246224297.0, + "step": 11845 + }, + { + "epoch": 0.9768285643605179, + "grad_norm": 1.9083818197250366, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.753775417804718, + "num_tokens": 246254474.0, + "step": 11846 + }, + { + "epoch": 0.9769110249855694, + "grad_norm": 2.0368382930755615, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7818921208381653, + "num_tokens": 246282890.0, + "step": 11847 + }, + { + "epoch": 0.9769934856106209, + "grad_norm": 1.8191906213760376, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7682396173477173, + "num_tokens": 246314435.0, + "step": 11848 + }, + { + "epoch": 0.9770759462356725, + "grad_norm": 1.9689230918884277, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7536746263504028, + "num_tokens": 246345867.0, + "step": 11849 + }, + { + "epoch": 0.977158406860724, + "grad_norm": 2.2737834453582764, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7332442402839661, + "num_tokens": 246369527.0, + "step": 11850 + }, + { + "epoch": 0.9772408674857755, + "grad_norm": 2.251039743423462, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.716184139251709, + "num_tokens": 246393164.0, + "step": 11851 + }, + { + "epoch": 0.9773233281108271, + "grad_norm": 2.1690311431884766, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7375165224075317, + "num_tokens": 246418695.0, + "step": 11852 + }, + { + "epoch": 0.9774057887358786, + "grad_norm": 1.8182759284973145, + "learning_rate": 1e-06, + "loss": 0.5955, + "mean_token_accuracy": 0.8038306832313538, + "num_tokens": 246448483.0, + "step": 11853 + }, + { + "epoch": 0.9774882493609302, + "grad_norm": 1.795818567276001, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.8005598783493042, + "num_tokens": 246478358.0, + "step": 11854 + }, + { + "epoch": 0.9775707099859817, + "grad_norm": 2.045370578765869, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7761446237564087, + "num_tokens": 246504333.0, + "step": 11855 + }, + { + "epoch": 0.9776531706110332, + "grad_norm": 1.8476402759552002, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7463738918304443, + "num_tokens": 246535321.0, + "step": 11856 + }, + { + "epoch": 0.9777356312360848, + "grad_norm": 1.8973580598831177, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7670304775238037, + "num_tokens": 246564997.0, + "step": 11857 + }, + { + "epoch": 0.9778180918611363, + "grad_norm": 2.1527583599090576, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7332111597061157, + "num_tokens": 246591693.0, + "step": 11858 + }, + { + "epoch": 0.9779005524861878, + "grad_norm": 1.900678038597107, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7562004923820496, + "num_tokens": 246620311.0, + "step": 11859 + }, + { + "epoch": 0.9779830131112394, + "grad_norm": 1.6940702199935913, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7478545904159546, + "num_tokens": 246654836.0, + "step": 11860 + }, + { + "epoch": 0.9780654737362909, + "grad_norm": 1.9597275257110596, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.754024863243103, + "num_tokens": 246682360.0, + "step": 11861 + }, + { + "epoch": 0.9781479343613425, + "grad_norm": 1.8970141410827637, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.767713189125061, + "num_tokens": 246715729.0, + "step": 11862 + }, + { + "epoch": 0.978230394986394, + "grad_norm": 2.2259020805358887, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7611401081085205, + "num_tokens": 246743289.0, + "step": 11863 + }, + { + "epoch": 0.9783128556114455, + "grad_norm": 1.8869056701660156, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7843265533447266, + "num_tokens": 246770692.0, + "step": 11864 + }, + { + "epoch": 0.9783953162364971, + "grad_norm": 1.9043196439743042, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.724403977394104, + "num_tokens": 246802943.0, + "step": 11865 + }, + { + "epoch": 0.9784777768615486, + "grad_norm": 2.055701971054077, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7709521055221558, + "num_tokens": 246829978.0, + "step": 11866 + }, + { + "epoch": 0.9785602374866001, + "grad_norm": 2.2292277812957764, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.719516396522522, + "num_tokens": 246856646.0, + "step": 11867 + }, + { + "epoch": 0.9786426981116517, + "grad_norm": 2.01693058013916, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.7598147392272949, + "num_tokens": 246884361.0, + "step": 11868 + }, + { + "epoch": 0.9787251587367032, + "grad_norm": 1.7523680925369263, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.7943853139877319, + "num_tokens": 246916955.0, + "step": 11869 + }, + { + "epoch": 0.9788076193617548, + "grad_norm": 2.153252124786377, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7210507392883301, + "num_tokens": 246943808.0, + "step": 11870 + }, + { + "epoch": 0.9788900799868063, + "grad_norm": 1.867107629776001, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7329195737838745, + "num_tokens": 246974377.0, + "step": 11871 + }, + { + "epoch": 0.9789725406118578, + "grad_norm": 1.755164384841919, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7851212620735168, + "num_tokens": 247009636.0, + "step": 11872 + }, + { + "epoch": 0.9790550012369094, + "grad_norm": 2.0510873794555664, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7228664755821228, + "num_tokens": 247037053.0, + "step": 11873 + }, + { + "epoch": 0.9791374618619609, + "grad_norm": 2.069567918777466, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7460865378379822, + "num_tokens": 247062698.0, + "step": 11874 + }, + { + "epoch": 0.9792199224870125, + "grad_norm": 1.9319565296173096, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7572661638259888, + "num_tokens": 247091723.0, + "step": 11875 + }, + { + "epoch": 0.979302383112064, + "grad_norm": 1.994568943977356, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7415483593940735, + "num_tokens": 247120453.0, + "step": 11876 + }, + { + "epoch": 0.9793848437371155, + "grad_norm": 1.8425137996673584, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7628185153007507, + "num_tokens": 247150783.0, + "step": 11877 + }, + { + "epoch": 0.9794673043621671, + "grad_norm": 1.9234727621078491, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7814818024635315, + "num_tokens": 247178226.0, + "step": 11878 + }, + { + "epoch": 0.9795497649872186, + "grad_norm": 1.7575017213821411, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7662700414657593, + "num_tokens": 247212125.0, + "step": 11879 + }, + { + "epoch": 0.9796322256122701, + "grad_norm": 2.0789377689361572, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7721792459487915, + "num_tokens": 247237079.0, + "step": 11880 + }, + { + "epoch": 0.9797146862373217, + "grad_norm": 1.7890574932098389, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7751708030700684, + "num_tokens": 247269172.0, + "step": 11881 + }, + { + "epoch": 0.9797971468623732, + "grad_norm": 1.8555821180343628, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7194758653640747, + "num_tokens": 247304640.0, + "step": 11882 + }, + { + "epoch": 0.9798796074874248, + "grad_norm": 2.1037404537200928, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7312327027320862, + "num_tokens": 247331201.0, + "step": 11883 + }, + { + "epoch": 0.9799620681124763, + "grad_norm": 1.9286764860153198, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7262513637542725, + "num_tokens": 247362702.0, + "step": 11884 + }, + { + "epoch": 0.9800445287375278, + "grad_norm": 1.7825723886489868, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7319722175598145, + "num_tokens": 247398104.0, + "step": 11885 + }, + { + "epoch": 0.9801269893625794, + "grad_norm": 1.7150564193725586, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7518091797828674, + "num_tokens": 247434902.0, + "step": 11886 + }, + { + "epoch": 0.9802094499876309, + "grad_norm": 2.1576766967773438, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.7892500162124634, + "num_tokens": 247462682.0, + "step": 11887 + }, + { + "epoch": 0.9802919106126824, + "grad_norm": 2.325746774673462, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7249846458435059, + "num_tokens": 247484316.0, + "step": 11888 + }, + { + "epoch": 0.980374371237734, + "grad_norm": 1.6982892751693726, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7553199529647827, + "num_tokens": 247519622.0, + "step": 11889 + }, + { + "epoch": 0.9804568318627855, + "grad_norm": 1.959899663925171, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7217357158660889, + "num_tokens": 247549667.0, + "step": 11890 + }, + { + "epoch": 0.9805392924878371, + "grad_norm": 1.7701956033706665, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7574671506881714, + "num_tokens": 247585113.0, + "step": 11891 + }, + { + "epoch": 0.9806217531128886, + "grad_norm": 1.9193713665008545, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7395761013031006, + "num_tokens": 247617371.0, + "step": 11892 + }, + { + "epoch": 0.9807042137379401, + "grad_norm": 2.059616804122925, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.753627359867096, + "num_tokens": 247643055.0, + "step": 11893 + }, + { + "epoch": 0.9807866743629917, + "grad_norm": 1.832082986831665, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7256819009780884, + "num_tokens": 247675893.0, + "step": 11894 + }, + { + "epoch": 0.9808691349880432, + "grad_norm": 1.8901243209838867, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7424924373626709, + "num_tokens": 247707773.0, + "step": 11895 + }, + { + "epoch": 0.9809515956130948, + "grad_norm": 2.110827922821045, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7396836280822754, + "num_tokens": 247734653.0, + "step": 11896 + }, + { + "epoch": 0.9810340562381463, + "grad_norm": 1.7840029001235962, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7715733051300049, + "num_tokens": 247767595.0, + "step": 11897 + }, + { + "epoch": 0.9811165168631978, + "grad_norm": 1.9331914186477661, + "learning_rate": 1e-06, + "loss": 0.5976, + "mean_token_accuracy": 0.8025438189506531, + "num_tokens": 247794870.0, + "step": 11898 + }, + { + "epoch": 0.9811989774882494, + "grad_norm": 2.0286026000976562, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.718381404876709, + "num_tokens": 247825506.0, + "step": 11899 + }, + { + "epoch": 0.9812814381133009, + "grad_norm": 2.283026695251465, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7343786954879761, + "num_tokens": 247847772.0, + "step": 11900 + }, + { + "epoch": 0.9813638987383524, + "grad_norm": 1.9693347215652466, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7571161985397339, + "num_tokens": 247875912.0, + "step": 11901 + }, + { + "epoch": 0.981446359363404, + "grad_norm": 1.8555666208267212, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7554954290390015, + "num_tokens": 247908007.0, + "step": 11902 + }, + { + "epoch": 0.9815288199884555, + "grad_norm": 1.7221699953079224, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7869953513145447, + "num_tokens": 247943468.0, + "step": 11903 + }, + { + "epoch": 0.981611280613507, + "grad_norm": 1.9700536727905273, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7219230532646179, + "num_tokens": 247973460.0, + "step": 11904 + }, + { + "epoch": 0.9816937412385586, + "grad_norm": 1.8960986137390137, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7284377813339233, + "num_tokens": 248005571.0, + "step": 11905 + }, + { + "epoch": 0.9817762018636101, + "grad_norm": 1.8691766262054443, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7530929446220398, + "num_tokens": 248036157.0, + "step": 11906 + }, + { + "epoch": 0.9818586624886617, + "grad_norm": 1.5989618301391602, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.7917210459709167, + "num_tokens": 248073575.0, + "step": 11907 + }, + { + "epoch": 0.9819411231137132, + "grad_norm": 1.939622163772583, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.7651668190956116, + "num_tokens": 248102484.0, + "step": 11908 + }, + { + "epoch": 0.9820235837387647, + "grad_norm": 1.9839838743209839, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7473584413528442, + "num_tokens": 248133524.0, + "step": 11909 + }, + { + "epoch": 0.9821060443638163, + "grad_norm": 2.057295322418213, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7780483961105347, + "num_tokens": 248159882.0, + "step": 11910 + }, + { + "epoch": 0.9821885049888678, + "grad_norm": 2.070714235305786, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7208799123764038, + "num_tokens": 248188196.0, + "step": 11911 + }, + { + "epoch": 0.9822709656139194, + "grad_norm": 2.1263816356658936, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7280871272087097, + "num_tokens": 248213030.0, + "step": 11912 + }, + { + "epoch": 0.9823534262389709, + "grad_norm": 1.9158117771148682, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7232593894004822, + "num_tokens": 248245184.0, + "step": 11913 + }, + { + "epoch": 0.9824358868640224, + "grad_norm": 1.9249885082244873, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7532851696014404, + "num_tokens": 248276974.0, + "step": 11914 + }, + { + "epoch": 0.982518347489074, + "grad_norm": 2.1460156440734863, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7537941932678223, + "num_tokens": 248302718.0, + "step": 11915 + }, + { + "epoch": 0.9826008081141255, + "grad_norm": 2.1871190071105957, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7697054743766785, + "num_tokens": 248326921.0, + "step": 11916 + }, + { + "epoch": 0.982683268739177, + "grad_norm": 1.8293168544769287, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.773773729801178, + "num_tokens": 248356983.0, + "step": 11917 + }, + { + "epoch": 0.9827657293642286, + "grad_norm": 2.231154680252075, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7316055297851562, + "num_tokens": 248382089.0, + "step": 11918 + }, + { + "epoch": 0.9828481899892801, + "grad_norm": 2.117780923843384, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7775539755821228, + "num_tokens": 248406419.0, + "step": 11919 + }, + { + "epoch": 0.9829306506143316, + "grad_norm": 1.8292677402496338, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7590789794921875, + "num_tokens": 248439034.0, + "step": 11920 + }, + { + "epoch": 0.9830131112393832, + "grad_norm": 2.1648736000061035, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7459225058555603, + "num_tokens": 248465836.0, + "step": 11921 + }, + { + "epoch": 0.9830955718644347, + "grad_norm": 1.898341178894043, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7852261662483215, + "num_tokens": 248497417.0, + "step": 11922 + }, + { + "epoch": 0.9831780324894863, + "grad_norm": 1.9872063398361206, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7742434740066528, + "num_tokens": 248524530.0, + "step": 11923 + }, + { + "epoch": 0.9832604931145378, + "grad_norm": 1.8493573665618896, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7633365392684937, + "num_tokens": 248552591.0, + "step": 11924 + }, + { + "epoch": 0.9833429537395894, + "grad_norm": 2.0371363162994385, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7609180212020874, + "num_tokens": 248579913.0, + "step": 11925 + }, + { + "epoch": 0.9834254143646409, + "grad_norm": 2.1749260425567627, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7363153696060181, + "num_tokens": 248605871.0, + "step": 11926 + }, + { + "epoch": 0.9835078749896924, + "grad_norm": 2.0795650482177734, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7764019966125488, + "num_tokens": 248630817.0, + "step": 11927 + }, + { + "epoch": 0.983590335614744, + "grad_norm": 1.869041085243225, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7309991717338562, + "num_tokens": 248663033.0, + "step": 11928 + }, + { + "epoch": 0.9836727962397955, + "grad_norm": 2.0170743465423584, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7773547172546387, + "num_tokens": 248689639.0, + "step": 11929 + }, + { + "epoch": 0.983755256864847, + "grad_norm": 1.8883851766586304, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.7525455951690674, + "num_tokens": 248719354.0, + "step": 11930 + }, + { + "epoch": 0.9838377174898986, + "grad_norm": 2.3010334968566895, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7695632576942444, + "num_tokens": 248740537.0, + "step": 11931 + }, + { + "epoch": 0.9839201781149501, + "grad_norm": 2.3719723224639893, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.7514472007751465, + "num_tokens": 248761028.0, + "step": 11932 + }, + { + "epoch": 0.9840026387400016, + "grad_norm": 1.9894074201583862, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7374653816223145, + "num_tokens": 248789936.0, + "step": 11933 + }, + { + "epoch": 0.9840850993650532, + "grad_norm": 1.8454018831253052, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7437241077423096, + "num_tokens": 248820694.0, + "step": 11934 + }, + { + "epoch": 0.9841675599901047, + "grad_norm": 1.9055135250091553, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7245081663131714, + "num_tokens": 248854628.0, + "step": 11935 + }, + { + "epoch": 0.9842500206151563, + "grad_norm": 1.8258826732635498, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7650485038757324, + "num_tokens": 248884063.0, + "step": 11936 + }, + { + "epoch": 0.9843324812402078, + "grad_norm": 1.6926826238632202, + "learning_rate": 1e-06, + "loss": 0.5698, + "mean_token_accuracy": 0.8114743232727051, + "num_tokens": 248915762.0, + "step": 11937 + }, + { + "epoch": 0.9844149418652594, + "grad_norm": 1.5112433433532715, + "learning_rate": 1e-06, + "loss": 0.6189, + "mean_token_accuracy": 0.8019846081733704, + "num_tokens": 248956405.0, + "step": 11938 + }, + { + "epoch": 0.9844974024903109, + "grad_norm": 1.9225455522537231, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7690058946609497, + "num_tokens": 248988406.0, + "step": 11939 + }, + { + "epoch": 0.9845798631153624, + "grad_norm": 2.056804895401001, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7572263479232788, + "num_tokens": 249020113.0, + "step": 11940 + }, + { + "epoch": 0.984662323740414, + "grad_norm": 1.943381905555725, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7254279851913452, + "num_tokens": 249050704.0, + "step": 11941 + }, + { + "epoch": 0.9847447843654655, + "grad_norm": 2.07468581199646, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7669773101806641, + "num_tokens": 249074451.0, + "step": 11942 + }, + { + "epoch": 0.984827244990517, + "grad_norm": 2.0264010429382324, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7614456415176392, + "num_tokens": 249103786.0, + "step": 11943 + }, + { + "epoch": 0.9849097056155686, + "grad_norm": 1.9548840522766113, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7357760071754456, + "num_tokens": 249135287.0, + "step": 11944 + }, + { + "epoch": 0.9849921662406201, + "grad_norm": 2.0814931392669678, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7521886825561523, + "num_tokens": 249164211.0, + "step": 11945 + }, + { + "epoch": 0.9850746268656716, + "grad_norm": 1.9606164693832397, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7474827170372009, + "num_tokens": 249194656.0, + "step": 11946 + }, + { + "epoch": 0.9851570874907232, + "grad_norm": 2.0421981811523438, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7460200786590576, + "num_tokens": 249224357.0, + "step": 11947 + }, + { + "epoch": 0.9852395481157747, + "grad_norm": 2.058680534362793, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7477849721908569, + "num_tokens": 249249154.0, + "step": 11948 + }, + { + "epoch": 0.9853220087408262, + "grad_norm": 2.1049752235412598, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7251157760620117, + "num_tokens": 249275700.0, + "step": 11949 + }, + { + "epoch": 0.9854044693658778, + "grad_norm": 1.9154093265533447, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7701196670532227, + "num_tokens": 249306379.0, + "step": 11950 + }, + { + "epoch": 0.9854869299909294, + "grad_norm": 1.840446949005127, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.729242205619812, + "num_tokens": 249339814.0, + "step": 11951 + }, + { + "epoch": 0.9855693906159809, + "grad_norm": 1.924756646156311, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7264785766601562, + "num_tokens": 249371882.0, + "step": 11952 + }, + { + "epoch": 0.9856518512410324, + "grad_norm": 1.9850168228149414, + "learning_rate": 1e-06, + "loss": 1.061, + "mean_token_accuracy": 0.6863118410110474, + "num_tokens": 249402716.0, + "step": 11953 + }, + { + "epoch": 0.985734311866084, + "grad_norm": 1.9372056722640991, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7750827074050903, + "num_tokens": 249433530.0, + "step": 11954 + }, + { + "epoch": 0.9858167724911355, + "grad_norm": 1.919823169708252, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7403998374938965, + "num_tokens": 249464722.0, + "step": 11955 + }, + { + "epoch": 0.985899233116187, + "grad_norm": 1.8379364013671875, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.727190375328064, + "num_tokens": 249499366.0, + "step": 11956 + }, + { + "epoch": 0.9859816937412386, + "grad_norm": 1.8869097232818604, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.7871589064598083, + "num_tokens": 249528350.0, + "step": 11957 + }, + { + "epoch": 0.9860641543662901, + "grad_norm": 1.9685274362564087, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7657926082611084, + "num_tokens": 249556870.0, + "step": 11958 + }, + { + "epoch": 0.9861466149913416, + "grad_norm": 1.8937335014343262, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.7829594612121582, + "num_tokens": 249587882.0, + "step": 11959 + }, + { + "epoch": 0.9862290756163932, + "grad_norm": 1.6938748359680176, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7858973741531372, + "num_tokens": 249622360.0, + "step": 11960 + }, + { + "epoch": 0.9863115362414447, + "grad_norm": 2.002919912338257, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7600307464599609, + "num_tokens": 249651023.0, + "step": 11961 + }, + { + "epoch": 0.9863939968664962, + "grad_norm": 2.0725741386413574, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7792007923126221, + "num_tokens": 249675425.0, + "step": 11962 + }, + { + "epoch": 0.9864764574915478, + "grad_norm": 2.226267099380493, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7403897047042847, + "num_tokens": 249700246.0, + "step": 11963 + }, + { + "epoch": 0.9865589181165993, + "grad_norm": 1.78556227684021, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7683676481246948, + "num_tokens": 249733699.0, + "step": 11964 + }, + { + "epoch": 0.9866413787416508, + "grad_norm": 2.5105960369110107, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7300012111663818, + "num_tokens": 249753876.0, + "step": 11965 + }, + { + "epoch": 0.9867238393667024, + "grad_norm": 1.8834456205368042, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7377601861953735, + "num_tokens": 249786081.0, + "step": 11966 + }, + { + "epoch": 0.986806299991754, + "grad_norm": 1.780144453048706, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7765172719955444, + "num_tokens": 249820270.0, + "step": 11967 + }, + { + "epoch": 0.9868887606168055, + "grad_norm": 2.0090439319610596, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.719611406326294, + "num_tokens": 249849293.0, + "step": 11968 + }, + { + "epoch": 0.986971221241857, + "grad_norm": 1.8725416660308838, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7485013008117676, + "num_tokens": 249883202.0, + "step": 11969 + }, + { + "epoch": 0.9870536818669086, + "grad_norm": 1.8064584732055664, + "learning_rate": 1e-06, + "loss": 0.7598, + "mean_token_accuracy": 0.761420726776123, + "num_tokens": 249914332.0, + "step": 11970 + }, + { + "epoch": 0.9871361424919601, + "grad_norm": 1.8803536891937256, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7770330905914307, + "num_tokens": 249945131.0, + "step": 11971 + }, + { + "epoch": 0.9872186031170116, + "grad_norm": 1.9388988018035889, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7536174058914185, + "num_tokens": 249979742.0, + "step": 11972 + }, + { + "epoch": 0.9873010637420632, + "grad_norm": 1.868360161781311, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7573437690734863, + "num_tokens": 250009948.0, + "step": 11973 + }, + { + "epoch": 0.9873835243671147, + "grad_norm": 1.874868631362915, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7324914336204529, + "num_tokens": 250042412.0, + "step": 11974 + }, + { + "epoch": 0.9874659849921662, + "grad_norm": 1.9048583507537842, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.7683526277542114, + "num_tokens": 250070811.0, + "step": 11975 + }, + { + "epoch": 0.9875484456172178, + "grad_norm": 1.9392181634902954, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7602695226669312, + "num_tokens": 250098819.0, + "step": 11976 + }, + { + "epoch": 0.9876309062422693, + "grad_norm": 1.8217740058898926, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7593139410018921, + "num_tokens": 250136184.0, + "step": 11977 + }, + { + "epoch": 0.9877133668673208, + "grad_norm": 1.7352352142333984, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7741644382476807, + "num_tokens": 250174351.0, + "step": 11978 + }, + { + "epoch": 0.9877958274923724, + "grad_norm": 1.8447184562683105, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7300626039505005, + "num_tokens": 250207552.0, + "step": 11979 + }, + { + "epoch": 0.987878288117424, + "grad_norm": 1.788292646408081, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.742361843585968, + "num_tokens": 250240223.0, + "step": 11980 + }, + { + "epoch": 0.9879607487424754, + "grad_norm": 1.751354455947876, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.8018636107444763, + "num_tokens": 250271013.0, + "step": 11981 + }, + { + "epoch": 0.988043209367527, + "grad_norm": 1.9375863075256348, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.766287088394165, + "num_tokens": 250301533.0, + "step": 11982 + }, + { + "epoch": 0.9881256699925786, + "grad_norm": 1.8917784690856934, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7850828766822815, + "num_tokens": 250329440.0, + "step": 11983 + }, + { + "epoch": 0.9882081306176301, + "grad_norm": 1.7441399097442627, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.7863376140594482, + "num_tokens": 250360652.0, + "step": 11984 + }, + { + "epoch": 0.9882905912426816, + "grad_norm": 1.968909502029419, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7732444405555725, + "num_tokens": 250389786.0, + "step": 11985 + }, + { + "epoch": 0.9883730518677332, + "grad_norm": 2.0435354709625244, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7171218395233154, + "num_tokens": 250418571.0, + "step": 11986 + }, + { + "epoch": 0.9884555124927847, + "grad_norm": 1.707934856414795, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.8024452924728394, + "num_tokens": 250451274.0, + "step": 11987 + }, + { + "epoch": 0.9885379731178362, + "grad_norm": 1.9638465642929077, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7692080736160278, + "num_tokens": 250480886.0, + "step": 11988 + }, + { + "epoch": 0.9886204337428878, + "grad_norm": 1.7578181028366089, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.777855396270752, + "num_tokens": 250514306.0, + "step": 11989 + }, + { + "epoch": 0.9887028943679393, + "grad_norm": 1.7524219751358032, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7682599425315857, + "num_tokens": 250549451.0, + "step": 11990 + }, + { + "epoch": 0.9887853549929908, + "grad_norm": 1.6859122514724731, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.7912885546684265, + "num_tokens": 250587301.0, + "step": 11991 + }, + { + "epoch": 0.9888678156180424, + "grad_norm": 1.9639488458633423, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7634454965591431, + "num_tokens": 250617513.0, + "step": 11992 + }, + { + "epoch": 0.988950276243094, + "grad_norm": 2.0705955028533936, + "learning_rate": 1e-06, + "loss": 0.5934, + "mean_token_accuracy": 0.799022912979126, + "num_tokens": 250641961.0, + "step": 11993 + }, + { + "epoch": 0.9890327368681454, + "grad_norm": 1.933171033859253, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.7933536171913147, + "num_tokens": 250670465.0, + "step": 11994 + }, + { + "epoch": 0.989115197493197, + "grad_norm": 1.9307140111923218, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7197433710098267, + "num_tokens": 250700355.0, + "step": 11995 + }, + { + "epoch": 0.9891976581182486, + "grad_norm": 1.7730884552001953, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.7834223508834839, + "num_tokens": 250731666.0, + "step": 11996 + }, + { + "epoch": 0.9892801187433, + "grad_norm": 2.363659143447876, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7319862842559814, + "num_tokens": 250752452.0, + "step": 11997 + }, + { + "epoch": 0.9893625793683516, + "grad_norm": 1.5939452648162842, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7539629936218262, + "num_tokens": 250794310.0, + "step": 11998 + }, + { + "epoch": 0.9894450399934032, + "grad_norm": 1.61501145362854, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7690168619155884, + "num_tokens": 250833164.0, + "step": 11999 + }, + { + "epoch": 0.9895275006184547, + "grad_norm": 2.0918052196502686, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.750077486038208, + "num_tokens": 250860444.0, + "step": 12000 + }, + { + "epoch": 0.9896099612435062, + "grad_norm": 1.776859998703003, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7479861974716187, + "num_tokens": 250893913.0, + "step": 12001 + }, + { + "epoch": 0.9896924218685578, + "grad_norm": 2.1026368141174316, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7431411743164062, + "num_tokens": 250918914.0, + "step": 12002 + }, + { + "epoch": 0.9897748824936093, + "grad_norm": 1.6336101293563843, + "learning_rate": 1e-06, + "loss": 0.6128, + "mean_token_accuracy": 0.7974491119384766, + "num_tokens": 250952840.0, + "step": 12003 + }, + { + "epoch": 0.9898573431186608, + "grad_norm": 2.020602226257324, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7550647258758545, + "num_tokens": 250978839.0, + "step": 12004 + }, + { + "epoch": 0.9899398037437124, + "grad_norm": 1.859273910522461, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.75148606300354, + "num_tokens": 251009848.0, + "step": 12005 + }, + { + "epoch": 0.990022264368764, + "grad_norm": 1.7380435466766357, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.763170599937439, + "num_tokens": 251046238.0, + "step": 12006 + }, + { + "epoch": 0.9901047249938154, + "grad_norm": 2.1324093341827393, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.751907229423523, + "num_tokens": 251072645.0, + "step": 12007 + }, + { + "epoch": 0.990187185618867, + "grad_norm": 1.878023386001587, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7794431447982788, + "num_tokens": 251105671.0, + "step": 12008 + }, + { + "epoch": 0.9902696462439186, + "grad_norm": 1.8966312408447266, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.755429744720459, + "num_tokens": 251138203.0, + "step": 12009 + }, + { + "epoch": 0.99035210686897, + "grad_norm": 1.6718331575393677, + "learning_rate": 1e-06, + "loss": 0.583, + "mean_token_accuracy": 0.7984960079193115, + "num_tokens": 251172636.0, + "step": 12010 + }, + { + "epoch": 0.9904345674940216, + "grad_norm": 2.1359572410583496, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7418856620788574, + "num_tokens": 251200083.0, + "step": 12011 + }, + { + "epoch": 0.9905170281190732, + "grad_norm": 1.987635850906372, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.768437922000885, + "num_tokens": 251229810.0, + "step": 12012 + }, + { + "epoch": 0.9905994887441247, + "grad_norm": 2.1061179637908936, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7476566433906555, + "num_tokens": 251253828.0, + "step": 12013 + }, + { + "epoch": 0.9906819493691762, + "grad_norm": 1.7787528038024902, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7686535120010376, + "num_tokens": 251287890.0, + "step": 12014 + }, + { + "epoch": 0.9907644099942278, + "grad_norm": 1.783341646194458, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7316564917564392, + "num_tokens": 251323854.0, + "step": 12015 + }, + { + "epoch": 0.9908468706192793, + "grad_norm": 1.7887170314788818, + "learning_rate": 1e-06, + "loss": 0.5712, + "mean_token_accuracy": 0.8103238344192505, + "num_tokens": 251354144.0, + "step": 12016 + }, + { + "epoch": 0.9909293312443308, + "grad_norm": 2.016843795776367, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7516568899154663, + "num_tokens": 251383735.0, + "step": 12017 + }, + { + "epoch": 0.9910117918693824, + "grad_norm": 1.7812329530715942, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7496090531349182, + "num_tokens": 251418011.0, + "step": 12018 + }, + { + "epoch": 0.991094252494434, + "grad_norm": 1.859479546546936, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.774520754814148, + "num_tokens": 251447553.0, + "step": 12019 + }, + { + "epoch": 0.9911767131194854, + "grad_norm": 2.192993402481079, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7620282173156738, + "num_tokens": 251471493.0, + "step": 12020 + }, + { + "epoch": 0.991259173744537, + "grad_norm": 2.1717233657836914, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7315421104431152, + "num_tokens": 251495248.0, + "step": 12021 + }, + { + "epoch": 0.9913416343695886, + "grad_norm": 2.008248805999756, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7297115325927734, + "num_tokens": 251527067.0, + "step": 12022 + }, + { + "epoch": 0.99142409499464, + "grad_norm": 2.045968770980835, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.76054447889328, + "num_tokens": 251552443.0, + "step": 12023 + }, + { + "epoch": 0.9915065556196916, + "grad_norm": 1.9495000839233398, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7503187656402588, + "num_tokens": 251584112.0, + "step": 12024 + }, + { + "epoch": 0.9915890162447432, + "grad_norm": 1.9141865968704224, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7756538391113281, + "num_tokens": 251613271.0, + "step": 12025 + }, + { + "epoch": 0.9916714768697946, + "grad_norm": 1.9821085929870605, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7264841794967651, + "num_tokens": 251642550.0, + "step": 12026 + }, + { + "epoch": 0.9917539374948462, + "grad_norm": 1.8602519035339355, + "learning_rate": 1e-06, + "loss": 0.6011, + "mean_token_accuracy": 0.800955593585968, + "num_tokens": 251672559.0, + "step": 12027 + }, + { + "epoch": 0.9918363981198978, + "grad_norm": 1.9651895761489868, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7789345979690552, + "num_tokens": 251700997.0, + "step": 12028 + }, + { + "epoch": 0.9919188587449493, + "grad_norm": 1.8992149829864502, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7501971125602722, + "num_tokens": 251732656.0, + "step": 12029 + }, + { + "epoch": 0.9920013193700008, + "grad_norm": 2.2125625610351562, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.737906277179718, + "num_tokens": 251757600.0, + "step": 12030 + }, + { + "epoch": 0.9920837799950524, + "grad_norm": 1.9343597888946533, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7603673338890076, + "num_tokens": 251787396.0, + "step": 12031 + }, + { + "epoch": 0.992166240620104, + "grad_norm": 2.0661282539367676, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7467576265335083, + "num_tokens": 251812868.0, + "step": 12032 + }, + { + "epoch": 0.9922487012451554, + "grad_norm": 1.640203833580017, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7624093294143677, + "num_tokens": 251847996.0, + "step": 12033 + }, + { + "epoch": 0.992331161870207, + "grad_norm": 1.9657331705093384, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7389813661575317, + "num_tokens": 251879725.0, + "step": 12034 + }, + { + "epoch": 0.9924136224952586, + "grad_norm": 2.2019431591033936, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.721850574016571, + "num_tokens": 251906324.0, + "step": 12035 + }, + { + "epoch": 0.99249608312031, + "grad_norm": 2.1218395233154297, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7602127194404602, + "num_tokens": 251933648.0, + "step": 12036 + }, + { + "epoch": 0.9925785437453616, + "grad_norm": 1.888411045074463, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7697739601135254, + "num_tokens": 251964465.0, + "step": 12037 + }, + { + "epoch": 0.9926610043704132, + "grad_norm": 1.9350608587265015, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7692955732345581, + "num_tokens": 251997382.0, + "step": 12038 + }, + { + "epoch": 0.9927434649954646, + "grad_norm": 2.0436298847198486, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7383946776390076, + "num_tokens": 252024026.0, + "step": 12039 + }, + { + "epoch": 0.9928259256205162, + "grad_norm": 1.88393235206604, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.746540904045105, + "num_tokens": 252054438.0, + "step": 12040 + }, + { + "epoch": 0.9929083862455678, + "grad_norm": 1.8028937578201294, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7797428369522095, + "num_tokens": 252088072.0, + "step": 12041 + }, + { + "epoch": 0.9929908468706192, + "grad_norm": 1.835628867149353, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7702434062957764, + "num_tokens": 252118790.0, + "step": 12042 + }, + { + "epoch": 0.9930733074956708, + "grad_norm": 1.9952359199523926, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.7916804552078247, + "num_tokens": 252143696.0, + "step": 12043 + }, + { + "epoch": 0.9931557681207224, + "grad_norm": 2.035433053970337, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.734811007976532, + "num_tokens": 252173323.0, + "step": 12044 + }, + { + "epoch": 0.9932382287457739, + "grad_norm": 1.9770127534866333, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7374627590179443, + "num_tokens": 252200610.0, + "step": 12045 + }, + { + "epoch": 0.9933206893708254, + "grad_norm": 1.8714590072631836, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7680133581161499, + "num_tokens": 252231480.0, + "step": 12046 + }, + { + "epoch": 0.993403149995877, + "grad_norm": 2.1511659622192383, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7568498253822327, + "num_tokens": 252254872.0, + "step": 12047 + }, + { + "epoch": 0.9934856106209285, + "grad_norm": 1.9236962795257568, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7400512099266052, + "num_tokens": 252289336.0, + "step": 12048 + }, + { + "epoch": 0.99356807124598, + "grad_norm": 1.8878754377365112, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7666621208190918, + "num_tokens": 252318223.0, + "step": 12049 + }, + { + "epoch": 0.9936505318710316, + "grad_norm": 1.9036452770233154, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7483471632003784, + "num_tokens": 252351376.0, + "step": 12050 + }, + { + "epoch": 0.9937329924960832, + "grad_norm": 1.876563310623169, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7451735734939575, + "num_tokens": 252383561.0, + "step": 12051 + }, + { + "epoch": 0.9938154531211346, + "grad_norm": 1.7348400354385376, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7870197296142578, + "num_tokens": 252418599.0, + "step": 12052 + }, + { + "epoch": 0.9938979137461862, + "grad_norm": 1.9565426111221313, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7510027289390564, + "num_tokens": 252447181.0, + "step": 12053 + }, + { + "epoch": 0.9939803743712378, + "grad_norm": 1.834418535232544, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7304104566574097, + "num_tokens": 252479483.0, + "step": 12054 + }, + { + "epoch": 0.9940628349962892, + "grad_norm": 1.814710259437561, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7657395005226135, + "num_tokens": 252513127.0, + "step": 12055 + }, + { + "epoch": 0.9941452956213408, + "grad_norm": 1.8646999597549438, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7689695358276367, + "num_tokens": 252546585.0, + "step": 12056 + }, + { + "epoch": 0.9942277562463924, + "grad_norm": 1.9489587545394897, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.732315719127655, + "num_tokens": 252575783.0, + "step": 12057 + }, + { + "epoch": 0.9943102168714438, + "grad_norm": 2.146514415740967, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7329388856887817, + "num_tokens": 252607142.0, + "step": 12058 + }, + { + "epoch": 0.9943926774964954, + "grad_norm": 2.486090898513794, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7538765668869019, + "num_tokens": 252627311.0, + "step": 12059 + }, + { + "epoch": 0.994475138121547, + "grad_norm": 1.7363852262496948, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7318427562713623, + "num_tokens": 252662990.0, + "step": 12060 + }, + { + "epoch": 0.9945575987465985, + "grad_norm": 1.7368565797805786, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7795270681381226, + "num_tokens": 252698299.0, + "step": 12061 + }, + { + "epoch": 0.99464005937165, + "grad_norm": 1.8590850830078125, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7746590375900269, + "num_tokens": 252729198.0, + "step": 12062 + }, + { + "epoch": 0.9947225199967016, + "grad_norm": 1.8530733585357666, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7234328985214233, + "num_tokens": 252764732.0, + "step": 12063 + }, + { + "epoch": 0.9948049806217532, + "grad_norm": 1.8084241151809692, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7132319211959839, + "num_tokens": 252798772.0, + "step": 12064 + }, + { + "epoch": 0.9948874412468046, + "grad_norm": 2.0913140773773193, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7214558720588684, + "num_tokens": 252828290.0, + "step": 12065 + }, + { + "epoch": 0.9949699018718562, + "grad_norm": 1.8902302980422974, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.7445535659790039, + "num_tokens": 252860170.0, + "step": 12066 + }, + { + "epoch": 0.9950523624969078, + "grad_norm": 1.7762047052383423, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.7984631657600403, + "num_tokens": 252890563.0, + "step": 12067 + }, + { + "epoch": 0.9951348231219592, + "grad_norm": 1.6148688793182373, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.7959656715393066, + "num_tokens": 252927104.0, + "step": 12068 + }, + { + "epoch": 0.9952172837470108, + "grad_norm": 1.8669564723968506, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7317871451377869, + "num_tokens": 252961220.0, + "step": 12069 + }, + { + "epoch": 0.9952997443720624, + "grad_norm": 1.8675806522369385, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7570713758468628, + "num_tokens": 252993399.0, + "step": 12070 + }, + { + "epoch": 0.9953822049971138, + "grad_norm": 1.913940191268921, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7720718383789062, + "num_tokens": 253023915.0, + "step": 12071 + }, + { + "epoch": 0.9954646656221654, + "grad_norm": 2.331332206726074, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7403749227523804, + "num_tokens": 253045357.0, + "step": 12072 + }, + { + "epoch": 0.995547126247217, + "grad_norm": 2.2129831314086914, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7600318789482117, + "num_tokens": 253068247.0, + "step": 12073 + }, + { + "epoch": 0.9956295868722685, + "grad_norm": 1.8399324417114258, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.7840796709060669, + "num_tokens": 253099613.0, + "step": 12074 + }, + { + "epoch": 0.99571204749732, + "grad_norm": 2.147351026535034, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7318547964096069, + "num_tokens": 253125566.0, + "step": 12075 + }, + { + "epoch": 0.9957945081223716, + "grad_norm": 1.9649266004562378, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7862756252288818, + "num_tokens": 253157590.0, + "step": 12076 + }, + { + "epoch": 0.9958769687474232, + "grad_norm": 2.1905908584594727, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7758604884147644, + "num_tokens": 253179844.0, + "step": 12077 + }, + { + "epoch": 0.9959594293724746, + "grad_norm": 1.99252188205719, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7399423122406006, + "num_tokens": 253207751.0, + "step": 12078 + }, + { + "epoch": 0.9960418899975262, + "grad_norm": 1.749751329421997, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7634009122848511, + "num_tokens": 253244339.0, + "step": 12079 + }, + { + "epoch": 0.9961243506225778, + "grad_norm": 1.789841890335083, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7409886121749878, + "num_tokens": 253283428.0, + "step": 12080 + }, + { + "epoch": 0.9962068112476292, + "grad_norm": 1.5984832048416138, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7692335844039917, + "num_tokens": 253320323.0, + "step": 12081 + }, + { + "epoch": 0.9962892718726808, + "grad_norm": 1.8132432699203491, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7371062636375427, + "num_tokens": 253353576.0, + "step": 12082 + }, + { + "epoch": 0.9963717324977324, + "grad_norm": 1.7692959308624268, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7270777225494385, + "num_tokens": 253390055.0, + "step": 12083 + }, + { + "epoch": 0.9964541931227838, + "grad_norm": 2.0707383155822754, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7418178915977478, + "num_tokens": 253415759.0, + "step": 12084 + }, + { + "epoch": 0.9965366537478354, + "grad_norm": 2.2831032276153564, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7414008378982544, + "num_tokens": 253438529.0, + "step": 12085 + }, + { + "epoch": 0.996619114372887, + "grad_norm": 1.9234062433242798, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.735740065574646, + "num_tokens": 253469233.0, + "step": 12086 + }, + { + "epoch": 0.9967015749979384, + "grad_norm": 2.03328013420105, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7568680644035339, + "num_tokens": 253496841.0, + "step": 12087 + }, + { + "epoch": 0.99678403562299, + "grad_norm": 2.160367727279663, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7339973449707031, + "num_tokens": 253522466.0, + "step": 12088 + }, + { + "epoch": 0.9968664962480416, + "grad_norm": 2.125704288482666, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7344515919685364, + "num_tokens": 253548306.0, + "step": 12089 + }, + { + "epoch": 0.9969489568730932, + "grad_norm": 1.9562548398971558, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7051366567611694, + "num_tokens": 253579001.0, + "step": 12090 + }, + { + "epoch": 0.9970314174981446, + "grad_norm": 2.035538911819458, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7670713067054749, + "num_tokens": 253606407.0, + "step": 12091 + }, + { + "epoch": 0.9971138781231962, + "grad_norm": 2.000718593597412, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7807654738426208, + "num_tokens": 253632639.0, + "step": 12092 + }, + { + "epoch": 0.9971963387482478, + "grad_norm": 2.020655632019043, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7671908140182495, + "num_tokens": 253657448.0, + "step": 12093 + }, + { + "epoch": 0.9972787993732992, + "grad_norm": 1.9757578372955322, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.72196364402771, + "num_tokens": 253687038.0, + "step": 12094 + }, + { + "epoch": 0.9973612599983508, + "grad_norm": 1.7967051267623901, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7413694858551025, + "num_tokens": 253721558.0, + "step": 12095 + }, + { + "epoch": 0.9974437206234024, + "grad_norm": 1.9090338945388794, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7631727457046509, + "num_tokens": 253752723.0, + "step": 12096 + }, + { + "epoch": 0.9975261812484538, + "grad_norm": 1.7297992706298828, + "learning_rate": 1e-06, + "loss": 0.7999, + "mean_token_accuracy": 0.7531770467758179, + "num_tokens": 253789573.0, + "step": 12097 + }, + { + "epoch": 0.9976086418735054, + "grad_norm": 1.884379506111145, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7576441168785095, + "num_tokens": 253818614.0, + "step": 12098 + }, + { + "epoch": 0.997691102498557, + "grad_norm": 2.4344215393066406, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7051151394844055, + "num_tokens": 253840341.0, + "step": 12099 + }, + { + "epoch": 0.9977735631236084, + "grad_norm": 2.1020584106445312, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7407166361808777, + "num_tokens": 253868125.0, + "step": 12100 + }, + { + "epoch": 0.99785602374866, + "grad_norm": 2.2007176876068115, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7227041125297546, + "num_tokens": 253893134.0, + "step": 12101 + }, + { + "epoch": 0.9979384843737116, + "grad_norm": 2.18878436088562, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7567672729492188, + "num_tokens": 253916444.0, + "step": 12102 + }, + { + "epoch": 0.998020944998763, + "grad_norm": 1.9749737977981567, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7562149167060852, + "num_tokens": 253946644.0, + "step": 12103 + }, + { + "epoch": 0.9981034056238146, + "grad_norm": 1.798069953918457, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7648942470550537, + "num_tokens": 253982372.0, + "step": 12104 + }, + { + "epoch": 0.9981858662488662, + "grad_norm": 2.035630464553833, + "learning_rate": 1e-06, + "loss": 0.6193, + "mean_token_accuracy": 0.8040591478347778, + "num_tokens": 254011469.0, + "step": 12105 + }, + { + "epoch": 0.9982683268739178, + "grad_norm": 1.923529863357544, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7566578388214111, + "num_tokens": 254040537.0, + "step": 12106 + }, + { + "epoch": 0.9983507874989692, + "grad_norm": 2.019932508468628, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7317953705787659, + "num_tokens": 254070053.0, + "step": 12107 + }, + { + "epoch": 0.9984332481240208, + "grad_norm": 2.1677939891815186, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7237358689308167, + "num_tokens": 254094630.0, + "step": 12108 + }, + { + "epoch": 0.9985157087490724, + "grad_norm": 1.7609713077545166, + "learning_rate": 1e-06, + "loss": 0.5932, + "mean_token_accuracy": 0.808790922164917, + "num_tokens": 254127791.0, + "step": 12109 + }, + { + "epoch": 0.9985981693741238, + "grad_norm": 1.9552242755889893, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7415282726287842, + "num_tokens": 254157038.0, + "step": 12110 + }, + { + "epoch": 0.9986806299991754, + "grad_norm": 1.9990317821502686, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7570413947105408, + "num_tokens": 254186283.0, + "step": 12111 + }, + { + "epoch": 0.998763090624227, + "grad_norm": 1.9713314771652222, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7535037398338318, + "num_tokens": 254217442.0, + "step": 12112 + }, + { + "epoch": 0.9988455512492784, + "grad_norm": 1.8024756908416748, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.7958534359931946, + "num_tokens": 254249326.0, + "step": 12113 + }, + { + "epoch": 0.99892801187433, + "grad_norm": 2.0154623985290527, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7698363661766052, + "num_tokens": 254275123.0, + "step": 12114 + }, + { + "epoch": 0.9990104724993816, + "grad_norm": 1.7643392086029053, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7525507807731628, + "num_tokens": 254307977.0, + "step": 12115 + }, + { + "epoch": 0.999092933124433, + "grad_norm": 1.89352285861969, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7687094211578369, + "num_tokens": 254339335.0, + "step": 12116 + }, + { + "epoch": 0.9991753937494846, + "grad_norm": 1.8812075853347778, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7490612268447876, + "num_tokens": 254368331.0, + "step": 12117 + }, + { + "epoch": 0.9992578543745362, + "grad_norm": 1.8990435600280762, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7752288579940796, + "num_tokens": 254397868.0, + "step": 12118 + }, + { + "epoch": 0.9993403149995876, + "grad_norm": 1.8360852003097534, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7592473030090332, + "num_tokens": 254428837.0, + "step": 12119 + }, + { + "epoch": 0.9994227756246392, + "grad_norm": 1.9270656108856201, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7483611106872559, + "num_tokens": 254459695.0, + "step": 12120 + }, + { + "epoch": 0.9995052362496908, + "grad_norm": 1.7292848825454712, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7576123476028442, + "num_tokens": 254492638.0, + "step": 12121 + }, + { + "epoch": 0.9995876968747424, + "grad_norm": 1.9741538763046265, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.728018045425415, + "num_tokens": 254520690.0, + "step": 12122 + }, + { + "epoch": 0.9996701574997938, + "grad_norm": 1.746362328529358, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7850841879844666, + "num_tokens": 254557928.0, + "step": 12123 + }, + { + "epoch": 0.9997526181248454, + "grad_norm": 2.015625, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.752417266368866, + "num_tokens": 254585565.0, + "step": 12124 + }, + { + "epoch": 0.999835078749897, + "grad_norm": 2.052448034286499, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7315057516098022, + "num_tokens": 254609873.0, + "step": 12125 + }, + { + "epoch": 0.9999175393749484, + "grad_norm": 1.944898247718811, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7622863054275513, + "num_tokens": 254637119.0, + "step": 12126 + }, + { + "epoch": 1.0, + "grad_norm": 1.7232701778411865, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7793911695480347, + "num_tokens": 254670786.0, + "step": 12127 + }, + { + "epoch": 1.0000824606250516, + "grad_norm": 1.7651561498641968, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7298891544342041, + "num_tokens": 254702264.0, + "step": 12128 + }, + { + "epoch": 1.0001649212501031, + "grad_norm": 1.8673337697982788, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7637153267860413, + "num_tokens": 254732667.0, + "step": 12129 + }, + { + "epoch": 1.0002473818751547, + "grad_norm": 2.00140643119812, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7354286909103394, + "num_tokens": 254758382.0, + "step": 12130 + }, + { + "epoch": 1.000329842500206, + "grad_norm": 1.8434362411499023, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.7530584931373596, + "num_tokens": 254789540.0, + "step": 12131 + }, + { + "epoch": 1.0004123031252576, + "grad_norm": 1.906174898147583, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.736565351486206, + "num_tokens": 254819049.0, + "step": 12132 + }, + { + "epoch": 1.0004947637503092, + "grad_norm": 1.9803136587142944, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7791396379470825, + "num_tokens": 254847976.0, + "step": 12133 + }, + { + "epoch": 1.0005772243753608, + "grad_norm": 1.823257327079773, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.7919334769248962, + "num_tokens": 254876050.0, + "step": 12134 + }, + { + "epoch": 1.0006596850004124, + "grad_norm": 1.833191156387329, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.7950617671012878, + "num_tokens": 254904621.0, + "step": 12135 + }, + { + "epoch": 1.000742145625464, + "grad_norm": 2.193664073944092, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7355830669403076, + "num_tokens": 254929216.0, + "step": 12136 + }, + { + "epoch": 1.0008246062505153, + "grad_norm": 2.1899328231811523, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7475104928016663, + "num_tokens": 254954145.0, + "step": 12137 + }, + { + "epoch": 1.0009070668755669, + "grad_norm": 1.7944132089614868, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7444482445716858, + "num_tokens": 254986263.0, + "step": 12138 + }, + { + "epoch": 1.0009895275006184, + "grad_norm": 1.874447226524353, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7624351382255554, + "num_tokens": 255021057.0, + "step": 12139 + }, + { + "epoch": 1.00107198812567, + "grad_norm": 1.977653980255127, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7235623598098755, + "num_tokens": 255053388.0, + "step": 12140 + }, + { + "epoch": 1.0011544487507216, + "grad_norm": 1.9454275369644165, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.7905242443084717, + "num_tokens": 255082842.0, + "step": 12141 + }, + { + "epoch": 1.0012369093757731, + "grad_norm": 1.598478078842163, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.8007704615592957, + "num_tokens": 255120888.0, + "step": 12142 + }, + { + "epoch": 1.0013193700008247, + "grad_norm": 2.0564796924591064, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7155100107192993, + "num_tokens": 255151204.0, + "step": 12143 + }, + { + "epoch": 1.001401830625876, + "grad_norm": 1.8760064840316772, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7571029663085938, + "num_tokens": 255184789.0, + "step": 12144 + }, + { + "epoch": 1.0014842912509276, + "grad_norm": 1.8731147050857544, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7691383957862854, + "num_tokens": 255217835.0, + "step": 12145 + }, + { + "epoch": 1.0015667518759792, + "grad_norm": 1.9259804487228394, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7432906627655029, + "num_tokens": 255248515.0, + "step": 12146 + }, + { + "epoch": 1.0016492125010308, + "grad_norm": 2.051147222518921, + "learning_rate": 1e-06, + "loss": 0.7887, + "mean_token_accuracy": 0.7534918189048767, + "num_tokens": 255276058.0, + "step": 12147 + }, + { + "epoch": 1.0017316731260824, + "grad_norm": 1.834775686264038, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7298001646995544, + "num_tokens": 255311770.0, + "step": 12148 + }, + { + "epoch": 1.001814133751134, + "grad_norm": 2.093337059020996, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7648656368255615, + "num_tokens": 255339284.0, + "step": 12149 + }, + { + "epoch": 1.0018965943761853, + "grad_norm": 2.044067621231079, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7482895255088806, + "num_tokens": 255367699.0, + "step": 12150 + }, + { + "epoch": 1.0019790550012369, + "grad_norm": 1.823346495628357, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7848096489906311, + "num_tokens": 255403838.0, + "step": 12151 + }, + { + "epoch": 1.0020615156262884, + "grad_norm": 2.0218355655670166, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7460687160491943, + "num_tokens": 255432092.0, + "step": 12152 + }, + { + "epoch": 1.00214397625134, + "grad_norm": 2.149427890777588, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.7916018962860107, + "num_tokens": 255456350.0, + "step": 12153 + }, + { + "epoch": 1.0022264368763916, + "grad_norm": 2.1683826446533203, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7420065402984619, + "num_tokens": 255482919.0, + "step": 12154 + }, + { + "epoch": 1.0023088975014431, + "grad_norm": 2.103548049926758, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.7919456958770752, + "num_tokens": 255506721.0, + "step": 12155 + }, + { + "epoch": 1.0023913581264945, + "grad_norm": 2.0161192417144775, + "learning_rate": 1e-06, + "loss": 0.603, + "mean_token_accuracy": 0.7959868907928467, + "num_tokens": 255537492.0, + "step": 12156 + }, + { + "epoch": 1.002473818751546, + "grad_norm": 1.6707828044891357, + "learning_rate": 1e-06, + "loss": 0.4894, + "mean_token_accuracy": 0.8314401507377625, + "num_tokens": 255576558.0, + "step": 12157 + }, + { + "epoch": 1.0025562793765976, + "grad_norm": 2.181377410888672, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7161568999290466, + "num_tokens": 255602705.0, + "step": 12158 + }, + { + "epoch": 1.0026387400016492, + "grad_norm": 1.8579323291778564, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7483018636703491, + "num_tokens": 255636985.0, + "step": 12159 + }, + { + "epoch": 1.0027212006267008, + "grad_norm": 2.096406936645508, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7484806776046753, + "num_tokens": 255664050.0, + "step": 12160 + }, + { + "epoch": 1.0028036612517524, + "grad_norm": 1.9818120002746582, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7456302642822266, + "num_tokens": 255690774.0, + "step": 12161 + }, + { + "epoch": 1.002886121876804, + "grad_norm": 1.9909179210662842, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.761263906955719, + "num_tokens": 255718223.0, + "step": 12162 + }, + { + "epoch": 1.0029685825018553, + "grad_norm": 2.2305779457092285, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7458246946334839, + "num_tokens": 255741628.0, + "step": 12163 + }, + { + "epoch": 1.0030510431269068, + "grad_norm": 2.1044435501098633, + "learning_rate": 1e-06, + "loss": 0.569, + "mean_token_accuracy": 0.805447518825531, + "num_tokens": 255767667.0, + "step": 12164 + }, + { + "epoch": 1.0031335037519584, + "grad_norm": 1.912609577178955, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.7859281301498413, + "num_tokens": 255797314.0, + "step": 12165 + }, + { + "epoch": 1.00321596437701, + "grad_norm": 2.0263376235961914, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7731548547744751, + "num_tokens": 255825472.0, + "step": 12166 + }, + { + "epoch": 1.0032984250020616, + "grad_norm": 1.8207505941390991, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7149828672409058, + "num_tokens": 255863256.0, + "step": 12167 + }, + { + "epoch": 1.0033808856271131, + "grad_norm": 1.899923324584961, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7180399298667908, + "num_tokens": 255896034.0, + "step": 12168 + }, + { + "epoch": 1.0034633462521645, + "grad_norm": 1.984399676322937, + "learning_rate": 1e-06, + "loss": 0.5349, + "mean_token_accuracy": 0.8150001764297485, + "num_tokens": 255923112.0, + "step": 12169 + }, + { + "epoch": 1.003545806877216, + "grad_norm": 1.9650092124938965, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7690420150756836, + "num_tokens": 255954595.0, + "step": 12170 + }, + { + "epoch": 1.0036282675022676, + "grad_norm": 1.922327995300293, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7392383813858032, + "num_tokens": 255988917.0, + "step": 12171 + }, + { + "epoch": 1.0037107281273192, + "grad_norm": 1.8585624694824219, + "learning_rate": 1e-06, + "loss": 0.5806, + "mean_token_accuracy": 0.8078997135162354, + "num_tokens": 256020825.0, + "step": 12172 + }, + { + "epoch": 1.0037931887523708, + "grad_norm": 1.9761475324630737, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.737410306930542, + "num_tokens": 256049565.0, + "step": 12173 + }, + { + "epoch": 1.0038756493774224, + "grad_norm": 1.9870837926864624, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7538328170776367, + "num_tokens": 256078924.0, + "step": 12174 + }, + { + "epoch": 1.003958110002474, + "grad_norm": 1.9656322002410889, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7339982986450195, + "num_tokens": 256109817.0, + "step": 12175 + }, + { + "epoch": 1.0040405706275253, + "grad_norm": 1.9461580514907837, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.741031289100647, + "num_tokens": 256142564.0, + "step": 12176 + }, + { + "epoch": 1.0041230312525768, + "grad_norm": 1.8150486946105957, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7451685667037964, + "num_tokens": 256177349.0, + "step": 12177 + }, + { + "epoch": 1.0042054918776284, + "grad_norm": 2.226550817489624, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7155035734176636, + "num_tokens": 256203508.0, + "step": 12178 + }, + { + "epoch": 1.00428795250268, + "grad_norm": 2.053597927093506, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7337014675140381, + "num_tokens": 256233706.0, + "step": 12179 + }, + { + "epoch": 1.0043704131277316, + "grad_norm": 1.8984484672546387, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.7989441752433777, + "num_tokens": 256267269.0, + "step": 12180 + }, + { + "epoch": 1.0044528737527831, + "grad_norm": 2.015937328338623, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7504746913909912, + "num_tokens": 256295618.0, + "step": 12181 + }, + { + "epoch": 1.0045353343778345, + "grad_norm": 1.8101617097854614, + "learning_rate": 1e-06, + "loss": 0.608, + "mean_token_accuracy": 0.7983928918838501, + "num_tokens": 256332372.0, + "step": 12182 + }, + { + "epoch": 1.004617795002886, + "grad_norm": 2.1483449935913086, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7379122972488403, + "num_tokens": 256358945.0, + "step": 12183 + }, + { + "epoch": 1.0047002556279376, + "grad_norm": 1.983518123626709, + "learning_rate": 1e-06, + "loss": 0.5935, + "mean_token_accuracy": 0.808281421661377, + "num_tokens": 256387368.0, + "step": 12184 + }, + { + "epoch": 1.0047827162529892, + "grad_norm": 1.9636390209197998, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7535291910171509, + "num_tokens": 256418037.0, + "step": 12185 + }, + { + "epoch": 1.0048651768780408, + "grad_norm": 1.7666854858398438, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7725450396537781, + "num_tokens": 256453376.0, + "step": 12186 + }, + { + "epoch": 1.0049476375030924, + "grad_norm": 1.9586280584335327, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7645467519760132, + "num_tokens": 256481321.0, + "step": 12187 + }, + { + "epoch": 1.005030098128144, + "grad_norm": 1.893069863319397, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7252546548843384, + "num_tokens": 256515027.0, + "step": 12188 + }, + { + "epoch": 1.0051125587531953, + "grad_norm": 1.870793342590332, + "learning_rate": 1e-06, + "loss": 0.5916, + "mean_token_accuracy": 0.7866823077201843, + "num_tokens": 256542309.0, + "step": 12189 + }, + { + "epoch": 1.0051950193782468, + "grad_norm": 1.8736919164657593, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7579822540283203, + "num_tokens": 256576054.0, + "step": 12190 + }, + { + "epoch": 1.0052774800032984, + "grad_norm": 1.9961992502212524, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.7635494470596313, + "num_tokens": 256605070.0, + "step": 12191 + }, + { + "epoch": 1.00535994062835, + "grad_norm": 1.9132368564605713, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7743295431137085, + "num_tokens": 256637597.0, + "step": 12192 + }, + { + "epoch": 1.0054424012534016, + "grad_norm": 2.241311550140381, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7585867047309875, + "num_tokens": 256660127.0, + "step": 12193 + }, + { + "epoch": 1.0055248618784531, + "grad_norm": 1.9489192962646484, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.7776474952697754, + "num_tokens": 256691192.0, + "step": 12194 + }, + { + "epoch": 1.0056073225035045, + "grad_norm": 1.6737990379333496, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.8019715547561646, + "num_tokens": 256727954.0, + "step": 12195 + }, + { + "epoch": 1.005689783128556, + "grad_norm": 2.1037402153015137, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7466470003128052, + "num_tokens": 256755542.0, + "step": 12196 + }, + { + "epoch": 1.0057722437536076, + "grad_norm": 1.8501503467559814, + "learning_rate": 1e-06, + "loss": 0.5648, + "mean_token_accuracy": 0.8170339465141296, + "num_tokens": 256785709.0, + "step": 12197 + }, + { + "epoch": 1.0058547043786592, + "grad_norm": 2.2918381690979004, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7296533584594727, + "num_tokens": 256809122.0, + "step": 12198 + }, + { + "epoch": 1.0059371650037108, + "grad_norm": 2.1101443767547607, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.7838200330734253, + "num_tokens": 256834265.0, + "step": 12199 + }, + { + "epoch": 1.0060196256287623, + "grad_norm": 1.8654664754867554, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.770446240901947, + "num_tokens": 256867771.0, + "step": 12200 + }, + { + "epoch": 1.0061020862538137, + "grad_norm": 1.8104742765426636, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7668852210044861, + "num_tokens": 256903951.0, + "step": 12201 + }, + { + "epoch": 1.0061845468788653, + "grad_norm": 2.0284931659698486, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.749802827835083, + "num_tokens": 256934807.0, + "step": 12202 + }, + { + "epoch": 1.0062670075039168, + "grad_norm": 1.9180896282196045, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7644352912902832, + "num_tokens": 256965856.0, + "step": 12203 + }, + { + "epoch": 1.0063494681289684, + "grad_norm": 1.7767081260681152, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7655717134475708, + "num_tokens": 257000040.0, + "step": 12204 + }, + { + "epoch": 1.00643192875402, + "grad_norm": 2.0700883865356445, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7709805369377136, + "num_tokens": 257026224.0, + "step": 12205 + }, + { + "epoch": 1.0065143893790716, + "grad_norm": 1.9451199769973755, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7881835103034973, + "num_tokens": 257054307.0, + "step": 12206 + }, + { + "epoch": 1.0065968500041231, + "grad_norm": 2.0425031185150146, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7189199924468994, + "num_tokens": 257084296.0, + "step": 12207 + }, + { + "epoch": 1.0066793106291745, + "grad_norm": 2.0471739768981934, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7703302502632141, + "num_tokens": 257112689.0, + "step": 12208 + }, + { + "epoch": 1.006761771254226, + "grad_norm": 1.8361239433288574, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7525538206100464, + "num_tokens": 257145095.0, + "step": 12209 + }, + { + "epoch": 1.0068442318792776, + "grad_norm": 1.8143117427825928, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7781476378440857, + "num_tokens": 257178649.0, + "step": 12210 + }, + { + "epoch": 1.0069266925043292, + "grad_norm": 1.9808028936386108, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7436752319335938, + "num_tokens": 257208366.0, + "step": 12211 + }, + { + "epoch": 1.0070091531293808, + "grad_norm": 1.9790159463882446, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7274304628372192, + "num_tokens": 257240601.0, + "step": 12212 + }, + { + "epoch": 1.0070916137544323, + "grad_norm": 2.1970789432525635, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7722375392913818, + "num_tokens": 257266452.0, + "step": 12213 + }, + { + "epoch": 1.0071740743794837, + "grad_norm": 1.8871241807937622, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7679208517074585, + "num_tokens": 257298519.0, + "step": 12214 + }, + { + "epoch": 1.0072565350045353, + "grad_norm": 2.2837235927581787, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7418773174285889, + "num_tokens": 257322382.0, + "step": 12215 + }, + { + "epoch": 1.0073389956295868, + "grad_norm": 2.026228666305542, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7812663316726685, + "num_tokens": 257351445.0, + "step": 12216 + }, + { + "epoch": 1.0074214562546384, + "grad_norm": 2.306389331817627, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7350189089775085, + "num_tokens": 257375881.0, + "step": 12217 + }, + { + "epoch": 1.00750391687969, + "grad_norm": 2.0711472034454346, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7729872465133667, + "num_tokens": 257401766.0, + "step": 12218 + }, + { + "epoch": 1.0075863775047416, + "grad_norm": 2.2141268253326416, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7426849603652954, + "num_tokens": 257425692.0, + "step": 12219 + }, + { + "epoch": 1.0076688381297931, + "grad_norm": 1.9745957851409912, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.729767382144928, + "num_tokens": 257455158.0, + "step": 12220 + }, + { + "epoch": 1.0077512987548445, + "grad_norm": 1.827231526374817, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.7663778066635132, + "num_tokens": 257488748.0, + "step": 12221 + }, + { + "epoch": 1.007833759379896, + "grad_norm": 1.8257368803024292, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7928667664527893, + "num_tokens": 257525655.0, + "step": 12222 + }, + { + "epoch": 1.0079162200049476, + "grad_norm": 2.223032236099243, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.7955907583236694, + "num_tokens": 257550079.0, + "step": 12223 + }, + { + "epoch": 1.0079986806299992, + "grad_norm": 2.13273549079895, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7721152305603027, + "num_tokens": 257578986.0, + "step": 12224 + }, + { + "epoch": 1.0080811412550508, + "grad_norm": 1.9094595909118652, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7457471489906311, + "num_tokens": 257610619.0, + "step": 12225 + }, + { + "epoch": 1.0081636018801023, + "grad_norm": 1.8460015058517456, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7656458616256714, + "num_tokens": 257644497.0, + "step": 12226 + }, + { + "epoch": 1.0082460625051537, + "grad_norm": 1.9441758394241333, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7713339328765869, + "num_tokens": 257674408.0, + "step": 12227 + }, + { + "epoch": 1.0083285231302053, + "grad_norm": 1.7977118492126465, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7327799797058105, + "num_tokens": 257709053.0, + "step": 12228 + }, + { + "epoch": 1.0084109837552568, + "grad_norm": 2.2078497409820557, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7367159128189087, + "num_tokens": 257731946.0, + "step": 12229 + }, + { + "epoch": 1.0084934443803084, + "grad_norm": 1.9871535301208496, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.7982754707336426, + "num_tokens": 257760716.0, + "step": 12230 + }, + { + "epoch": 1.00857590500536, + "grad_norm": 1.9350838661193848, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.7895205020904541, + "num_tokens": 257789676.0, + "step": 12231 + }, + { + "epoch": 1.0086583656304116, + "grad_norm": 2.0188000202178955, + "learning_rate": 1e-06, + "loss": 0.6128, + "mean_token_accuracy": 0.7828609347343445, + "num_tokens": 257816188.0, + "step": 12232 + }, + { + "epoch": 1.008740826255463, + "grad_norm": 1.8630030155181885, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.7891307473182678, + "num_tokens": 257849303.0, + "step": 12233 + }, + { + "epoch": 1.0088232868805145, + "grad_norm": 2.000826597213745, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7787524461746216, + "num_tokens": 257879339.0, + "step": 12234 + }, + { + "epoch": 1.008905747505566, + "grad_norm": 1.9098032712936401, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7313174605369568, + "num_tokens": 257910843.0, + "step": 12235 + }, + { + "epoch": 1.0089882081306176, + "grad_norm": 2.0739858150482178, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7436951398849487, + "num_tokens": 257940132.0, + "step": 12236 + }, + { + "epoch": 1.0090706687556692, + "grad_norm": 2.0089120864868164, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7696120142936707, + "num_tokens": 257966955.0, + "step": 12237 + }, + { + "epoch": 1.0091531293807208, + "grad_norm": 2.1242024898529053, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7716557383537292, + "num_tokens": 257997056.0, + "step": 12238 + }, + { + "epoch": 1.0092355900057723, + "grad_norm": 1.8621906042099, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7632014751434326, + "num_tokens": 258030944.0, + "step": 12239 + }, + { + "epoch": 1.0093180506308237, + "grad_norm": 2.1385316848754883, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7759064435958862, + "num_tokens": 258055322.0, + "step": 12240 + }, + { + "epoch": 1.0094005112558753, + "grad_norm": 2.118441343307495, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7451777458190918, + "num_tokens": 258080752.0, + "step": 12241 + }, + { + "epoch": 1.0094829718809268, + "grad_norm": 2.0111446380615234, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7480626702308655, + "num_tokens": 258107318.0, + "step": 12242 + }, + { + "epoch": 1.0095654325059784, + "grad_norm": 2.1118013858795166, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7809524536132812, + "num_tokens": 258133959.0, + "step": 12243 + }, + { + "epoch": 1.00964789313103, + "grad_norm": 1.7808146476745605, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7705399990081787, + "num_tokens": 258166293.0, + "step": 12244 + }, + { + "epoch": 1.0097303537560816, + "grad_norm": 1.8108848333358765, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7679109573364258, + "num_tokens": 258199163.0, + "step": 12245 + }, + { + "epoch": 1.009812814381133, + "grad_norm": 2.009087324142456, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7463212609291077, + "num_tokens": 258228608.0, + "step": 12246 + }, + { + "epoch": 1.0098952750061845, + "grad_norm": 2.1490771770477295, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.7773120999336243, + "num_tokens": 258253453.0, + "step": 12247 + }, + { + "epoch": 1.009977735631236, + "grad_norm": 1.9641191959381104, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7425069808959961, + "num_tokens": 258286012.0, + "step": 12248 + }, + { + "epoch": 1.0100601962562876, + "grad_norm": 1.8742059469223022, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7528123259544373, + "num_tokens": 258318375.0, + "step": 12249 + }, + { + "epoch": 1.0101426568813392, + "grad_norm": 2.1337778568267822, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7388808727264404, + "num_tokens": 258346246.0, + "step": 12250 + }, + { + "epoch": 1.0102251175063908, + "grad_norm": 1.9687714576721191, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7205892205238342, + "num_tokens": 258379561.0, + "step": 12251 + }, + { + "epoch": 1.0103075781314423, + "grad_norm": 1.9736976623535156, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.7644643783569336, + "num_tokens": 258409371.0, + "step": 12252 + }, + { + "epoch": 1.0103900387564937, + "grad_norm": 1.8949213027954102, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.767177164554596, + "num_tokens": 258440866.0, + "step": 12253 + }, + { + "epoch": 1.0104724993815453, + "grad_norm": 1.96869957447052, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7544590830802917, + "num_tokens": 258470422.0, + "step": 12254 + }, + { + "epoch": 1.0105549600065968, + "grad_norm": 2.0769286155700684, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7830468416213989, + "num_tokens": 258496767.0, + "step": 12255 + }, + { + "epoch": 1.0106374206316484, + "grad_norm": 1.9703588485717773, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.7862303853034973, + "num_tokens": 258526130.0, + "step": 12256 + }, + { + "epoch": 1.0107198812567, + "grad_norm": 1.9629374742507935, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.757286548614502, + "num_tokens": 258558929.0, + "step": 12257 + }, + { + "epoch": 1.0108023418817516, + "grad_norm": 2.106705665588379, + "learning_rate": 1e-06, + "loss": 0.7483, + "mean_token_accuracy": 0.7537619471549988, + "num_tokens": 258588342.0, + "step": 12258 + }, + { + "epoch": 1.010884802506803, + "grad_norm": 2.035037040710449, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7690048217773438, + "num_tokens": 258616973.0, + "step": 12259 + }, + { + "epoch": 1.0109672631318545, + "grad_norm": 2.0197577476501465, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7173838019371033, + "num_tokens": 258645837.0, + "step": 12260 + }, + { + "epoch": 1.011049723756906, + "grad_norm": 1.8162580728530884, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.771112859249115, + "num_tokens": 258679146.0, + "step": 12261 + }, + { + "epoch": 1.0111321843819576, + "grad_norm": 1.7066478729248047, + "learning_rate": 1e-06, + "loss": 0.5873, + "mean_token_accuracy": 0.8100731372833252, + "num_tokens": 258712574.0, + "step": 12262 + }, + { + "epoch": 1.0112146450070092, + "grad_norm": 1.998997449874878, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7765268087387085, + "num_tokens": 258738609.0, + "step": 12263 + }, + { + "epoch": 1.0112971056320608, + "grad_norm": 2.1638128757476807, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7462289333343506, + "num_tokens": 258765089.0, + "step": 12264 + }, + { + "epoch": 1.0113795662571123, + "grad_norm": 2.1367509365081787, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7729078531265259, + "num_tokens": 258787975.0, + "step": 12265 + }, + { + "epoch": 1.0114620268821637, + "grad_norm": 1.9139490127563477, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.754645049571991, + "num_tokens": 258818542.0, + "step": 12266 + }, + { + "epoch": 1.0115444875072153, + "grad_norm": 1.9324531555175781, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7725019454956055, + "num_tokens": 258848050.0, + "step": 12267 + }, + { + "epoch": 1.0116269481322668, + "grad_norm": 1.5766873359680176, + "learning_rate": 1e-06, + "loss": 0.5581, + "mean_token_accuracy": 0.8176493644714355, + "num_tokens": 258887073.0, + "step": 12268 + }, + { + "epoch": 1.0117094087573184, + "grad_norm": 1.983925700187683, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7866706848144531, + "num_tokens": 258913838.0, + "step": 12269 + }, + { + "epoch": 1.01179186938237, + "grad_norm": 1.881753921508789, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.7959836721420288, + "num_tokens": 258942592.0, + "step": 12270 + }, + { + "epoch": 1.0118743300074216, + "grad_norm": 1.8062479496002197, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7744148969650269, + "num_tokens": 258972312.0, + "step": 12271 + }, + { + "epoch": 1.011956790632473, + "grad_norm": 1.8390989303588867, + "learning_rate": 1e-06, + "loss": 0.6219, + "mean_token_accuracy": 0.7942522168159485, + "num_tokens": 259004210.0, + "step": 12272 + }, + { + "epoch": 1.0120392512575245, + "grad_norm": 1.7697994709014893, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7689883708953857, + "num_tokens": 259043204.0, + "step": 12273 + }, + { + "epoch": 1.012121711882576, + "grad_norm": 1.807965874671936, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7797229290008545, + "num_tokens": 259079961.0, + "step": 12274 + }, + { + "epoch": 1.0122041725076276, + "grad_norm": 2.214325189590454, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.775034487247467, + "num_tokens": 259104435.0, + "step": 12275 + }, + { + "epoch": 1.0122866331326792, + "grad_norm": 1.9947483539581299, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7361735105514526, + "num_tokens": 259133497.0, + "step": 12276 + }, + { + "epoch": 1.0123690937577308, + "grad_norm": 2.211437702178955, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7259773015975952, + "num_tokens": 259160533.0, + "step": 12277 + }, + { + "epoch": 1.0124515543827821, + "grad_norm": 1.9215006828308105, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.760176956653595, + "num_tokens": 259190409.0, + "step": 12278 + }, + { + "epoch": 1.0125340150078337, + "grad_norm": 2.122668981552124, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7620794773101807, + "num_tokens": 259217114.0, + "step": 12279 + }, + { + "epoch": 1.0126164756328853, + "grad_norm": 1.7217278480529785, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7777549624443054, + "num_tokens": 259254015.0, + "step": 12280 + }, + { + "epoch": 1.0126989362579368, + "grad_norm": 1.8586338758468628, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7534056305885315, + "num_tokens": 259285571.0, + "step": 12281 + }, + { + "epoch": 1.0127813968829884, + "grad_norm": 2.200136184692383, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7739427089691162, + "num_tokens": 259308208.0, + "step": 12282 + }, + { + "epoch": 1.01286385750804, + "grad_norm": 1.7971128225326538, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7391688823699951, + "num_tokens": 259345846.0, + "step": 12283 + }, + { + "epoch": 1.0129463181330915, + "grad_norm": 1.7507305145263672, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.7874436974525452, + "num_tokens": 259379495.0, + "step": 12284 + }, + { + "epoch": 1.013028778758143, + "grad_norm": 1.910636305809021, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7621829509735107, + "num_tokens": 259411833.0, + "step": 12285 + }, + { + "epoch": 1.0131112393831945, + "grad_norm": 1.974506139755249, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.7895407676696777, + "num_tokens": 259442747.0, + "step": 12286 + }, + { + "epoch": 1.013193700008246, + "grad_norm": 2.0404927730560303, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7552869915962219, + "num_tokens": 259473009.0, + "step": 12287 + }, + { + "epoch": 1.0132761606332976, + "grad_norm": 2.430722713470459, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7289372086524963, + "num_tokens": 259495134.0, + "step": 12288 + }, + { + "epoch": 1.0133586212583492, + "grad_norm": 1.9080928564071655, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7728804349899292, + "num_tokens": 259526256.0, + "step": 12289 + }, + { + "epoch": 1.0134410818834008, + "grad_norm": 1.9946329593658447, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7738701105117798, + "num_tokens": 259556028.0, + "step": 12290 + }, + { + "epoch": 1.0135235425084521, + "grad_norm": 1.887902021408081, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7674401998519897, + "num_tokens": 259587747.0, + "step": 12291 + }, + { + "epoch": 1.0136060031335037, + "grad_norm": 2.0094480514526367, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7752575278282166, + "num_tokens": 259616600.0, + "step": 12292 + }, + { + "epoch": 1.0136884637585553, + "grad_norm": 2.050180435180664, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7378194332122803, + "num_tokens": 259643648.0, + "step": 12293 + }, + { + "epoch": 1.0137709243836068, + "grad_norm": 1.708353877067566, + "learning_rate": 1e-06, + "loss": 0.6077, + "mean_token_accuracy": 0.8009750843048096, + "num_tokens": 259678897.0, + "step": 12294 + }, + { + "epoch": 1.0138533850086584, + "grad_norm": 1.8460614681243896, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.780541718006134, + "num_tokens": 259712310.0, + "step": 12295 + }, + { + "epoch": 1.01393584563371, + "grad_norm": 2.02406907081604, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7852927446365356, + "num_tokens": 259741089.0, + "step": 12296 + }, + { + "epoch": 1.0140183062587615, + "grad_norm": 1.7811568975448608, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.753298282623291, + "num_tokens": 259776428.0, + "step": 12297 + }, + { + "epoch": 1.014100766883813, + "grad_norm": 2.2101457118988037, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7702771425247192, + "num_tokens": 259802014.0, + "step": 12298 + }, + { + "epoch": 1.0141832275088645, + "grad_norm": 2.0421395301818848, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7811110019683838, + "num_tokens": 259829493.0, + "step": 12299 + }, + { + "epoch": 1.014265688133916, + "grad_norm": 2.0753068923950195, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7612231969833374, + "num_tokens": 259856354.0, + "step": 12300 + }, + { + "epoch": 1.0143481487589676, + "grad_norm": 1.9147281646728516, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7803471684455872, + "num_tokens": 259889961.0, + "step": 12301 + }, + { + "epoch": 1.0144306093840192, + "grad_norm": 2.452388286590576, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7276024222373962, + "num_tokens": 259912506.0, + "step": 12302 + }, + { + "epoch": 1.0145130700090708, + "grad_norm": 1.9909778833389282, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7107803821563721, + "num_tokens": 259946098.0, + "step": 12303 + }, + { + "epoch": 1.0145955306341221, + "grad_norm": 2.1222870349884033, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.788330614566803, + "num_tokens": 259973898.0, + "step": 12304 + }, + { + "epoch": 1.0146779912591737, + "grad_norm": 1.9077695608139038, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7823859453201294, + "num_tokens": 260006518.0, + "step": 12305 + }, + { + "epoch": 1.0147604518842253, + "grad_norm": 2.252929210662842, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7339805364608765, + "num_tokens": 260031745.0, + "step": 12306 + }, + { + "epoch": 1.0148429125092768, + "grad_norm": 2.2207682132720947, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.73971027135849, + "num_tokens": 260057748.0, + "step": 12307 + }, + { + "epoch": 1.0149253731343284, + "grad_norm": 1.8865736722946167, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7566844820976257, + "num_tokens": 260092094.0, + "step": 12308 + }, + { + "epoch": 1.01500783375938, + "grad_norm": 1.9692624807357788, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7514380812644958, + "num_tokens": 260122054.0, + "step": 12309 + }, + { + "epoch": 1.0150902943844313, + "grad_norm": 1.8188108205795288, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.747206449508667, + "num_tokens": 260153591.0, + "step": 12310 + }, + { + "epoch": 1.015172755009483, + "grad_norm": 2.1232919692993164, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7616409063339233, + "num_tokens": 260180623.0, + "step": 12311 + }, + { + "epoch": 1.0152552156345345, + "grad_norm": 2.140631914138794, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.8024117946624756, + "num_tokens": 260205459.0, + "step": 12312 + }, + { + "epoch": 1.015337676259586, + "grad_norm": 2.0190985202789307, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7381559610366821, + "num_tokens": 260234577.0, + "step": 12313 + }, + { + "epoch": 1.0154201368846376, + "grad_norm": 2.1721341609954834, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7481912970542908, + "num_tokens": 260261672.0, + "step": 12314 + }, + { + "epoch": 1.0155025975096892, + "grad_norm": 1.7891461849212646, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7590000033378601, + "num_tokens": 260298599.0, + "step": 12315 + }, + { + "epoch": 1.0155850581347408, + "grad_norm": 2.1406936645507812, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7552419304847717, + "num_tokens": 260324983.0, + "step": 12316 + }, + { + "epoch": 1.015667518759792, + "grad_norm": 2.0007028579711914, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7797041535377502, + "num_tokens": 260353396.0, + "step": 12317 + }, + { + "epoch": 1.0157499793848437, + "grad_norm": 1.7933238744735718, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7812716960906982, + "num_tokens": 260388416.0, + "step": 12318 + }, + { + "epoch": 1.0158324400098953, + "grad_norm": 1.7649420499801636, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7759937644004822, + "num_tokens": 260421152.0, + "step": 12319 + }, + { + "epoch": 1.0159149006349468, + "grad_norm": 2.058938503265381, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7421384453773499, + "num_tokens": 260452103.0, + "step": 12320 + }, + { + "epoch": 1.0159973612599984, + "grad_norm": 1.8061327934265137, + "learning_rate": 1e-06, + "loss": 0.734, + "mean_token_accuracy": 0.7637633085250854, + "num_tokens": 260488378.0, + "step": 12321 + }, + { + "epoch": 1.01607982188505, + "grad_norm": 1.8154023885726929, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7650126814842224, + "num_tokens": 260523342.0, + "step": 12322 + }, + { + "epoch": 1.0161622825101013, + "grad_norm": 2.0848500728607178, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7890475392341614, + "num_tokens": 260554286.0, + "step": 12323 + }, + { + "epoch": 1.016244743135153, + "grad_norm": 2.1651394367218018, + "learning_rate": 1e-06, + "loss": 0.604, + "mean_token_accuracy": 0.7976868152618408, + "num_tokens": 260583142.0, + "step": 12324 + }, + { + "epoch": 1.0163272037602045, + "grad_norm": 2.312424659729004, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7671162486076355, + "num_tokens": 260607364.0, + "step": 12325 + }, + { + "epoch": 1.016409664385256, + "grad_norm": 1.8976777791976929, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7620984315872192, + "num_tokens": 260636423.0, + "step": 12326 + }, + { + "epoch": 1.0164921250103076, + "grad_norm": 1.9379329681396484, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7226510047912598, + "num_tokens": 260669778.0, + "step": 12327 + }, + { + "epoch": 1.0165745856353592, + "grad_norm": 1.5862711668014526, + "learning_rate": 1e-06, + "loss": 0.5966, + "mean_token_accuracy": 0.8104075193405151, + "num_tokens": 260709036.0, + "step": 12328 + }, + { + "epoch": 1.0166570462604108, + "grad_norm": 1.8233742713928223, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7809057831764221, + "num_tokens": 260745799.0, + "step": 12329 + }, + { + "epoch": 1.016739506885462, + "grad_norm": 2.169055461883545, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.766360878944397, + "num_tokens": 260771187.0, + "step": 12330 + }, + { + "epoch": 1.0168219675105137, + "grad_norm": 1.9892288446426392, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7636687755584717, + "num_tokens": 260800660.0, + "step": 12331 + }, + { + "epoch": 1.0169044281355653, + "grad_norm": 2.173408031463623, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7606543302536011, + "num_tokens": 260825144.0, + "step": 12332 + }, + { + "epoch": 1.0169868887606168, + "grad_norm": 1.8888102769851685, + "learning_rate": 1e-06, + "loss": 0.594, + "mean_token_accuracy": 0.8046628832817078, + "num_tokens": 260855782.0, + "step": 12333 + }, + { + "epoch": 1.0170693493856684, + "grad_norm": 2.17521595954895, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.748862624168396, + "num_tokens": 260881760.0, + "step": 12334 + }, + { + "epoch": 1.01715181001072, + "grad_norm": 2.441115140914917, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.746660590171814, + "num_tokens": 260902226.0, + "step": 12335 + }, + { + "epoch": 1.0172342706357713, + "grad_norm": 1.9283541440963745, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7526836395263672, + "num_tokens": 260934371.0, + "step": 12336 + }, + { + "epoch": 1.017316731260823, + "grad_norm": 1.9785840511322021, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.767452597618103, + "num_tokens": 260963956.0, + "step": 12337 + }, + { + "epoch": 1.0173991918858745, + "grad_norm": 2.327892303466797, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7306476831436157, + "num_tokens": 260989661.0, + "step": 12338 + }, + { + "epoch": 1.017481652510926, + "grad_norm": 1.795301079750061, + "learning_rate": 1e-06, + "loss": 0.4918, + "mean_token_accuracy": 0.8117414712905884, + "num_tokens": 261020084.0, + "step": 12339 + }, + { + "epoch": 1.0175641131359776, + "grad_norm": 1.9116547107696533, + "learning_rate": 1e-06, + "loss": 0.5669, + "mean_token_accuracy": 0.8120908737182617, + "num_tokens": 261050079.0, + "step": 12340 + }, + { + "epoch": 1.0176465737610292, + "grad_norm": 1.9701393842697144, + "learning_rate": 1e-06, + "loss": 0.6252, + "mean_token_accuracy": 0.7933660745620728, + "num_tokens": 261079634.0, + "step": 12341 + }, + { + "epoch": 1.0177290343860808, + "grad_norm": 1.7556867599487305, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7478785514831543, + "num_tokens": 261114363.0, + "step": 12342 + }, + { + "epoch": 1.017811495011132, + "grad_norm": 2.090254068374634, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.770443320274353, + "num_tokens": 261143635.0, + "step": 12343 + }, + { + "epoch": 1.0178939556361837, + "grad_norm": 1.9684284925460815, + "learning_rate": 1e-06, + "loss": 0.6046, + "mean_token_accuracy": 0.7950173616409302, + "num_tokens": 261173822.0, + "step": 12344 + }, + { + "epoch": 1.0179764162612353, + "grad_norm": 2.0715203285217285, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7820050716400146, + "num_tokens": 261204260.0, + "step": 12345 + }, + { + "epoch": 1.0180588768862868, + "grad_norm": 1.9039751291275024, + "learning_rate": 1e-06, + "loss": 0.5749, + "mean_token_accuracy": 0.8050113916397095, + "num_tokens": 261233598.0, + "step": 12346 + }, + { + "epoch": 1.0181413375113384, + "grad_norm": 2.2389309406280518, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7208800315856934, + "num_tokens": 261262848.0, + "step": 12347 + }, + { + "epoch": 1.01822379813639, + "grad_norm": 2.1173667907714844, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7434003353118896, + "num_tokens": 261294520.0, + "step": 12348 + }, + { + "epoch": 1.0183062587614413, + "grad_norm": 1.7835153341293335, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.8043071031570435, + "num_tokens": 261329627.0, + "step": 12349 + }, + { + "epoch": 1.018388719386493, + "grad_norm": 2.052056312561035, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7550616264343262, + "num_tokens": 261360013.0, + "step": 12350 + }, + { + "epoch": 1.0184711800115445, + "grad_norm": 1.9550355672836304, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7423707246780396, + "num_tokens": 261390165.0, + "step": 12351 + }, + { + "epoch": 1.018553640636596, + "grad_norm": 2.200728178024292, + "learning_rate": 1e-06, + "loss": 0.5917, + "mean_token_accuracy": 0.805507481098175, + "num_tokens": 261412608.0, + "step": 12352 + }, + { + "epoch": 1.0186361012616476, + "grad_norm": 2.307997465133667, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7434067726135254, + "num_tokens": 261440411.0, + "step": 12353 + }, + { + "epoch": 1.0187185618866992, + "grad_norm": 2.280164957046509, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7672100067138672, + "num_tokens": 261462653.0, + "step": 12354 + }, + { + "epoch": 1.0188010225117505, + "grad_norm": 2.1911582946777344, + "learning_rate": 1e-06, + "loss": 0.784, + "mean_token_accuracy": 0.7449041604995728, + "num_tokens": 261489208.0, + "step": 12355 + }, + { + "epoch": 1.018883483136802, + "grad_norm": 1.8894785642623901, + "learning_rate": 1e-06, + "loss": 0.6115, + "mean_token_accuracy": 0.7955700159072876, + "num_tokens": 261523721.0, + "step": 12356 + }, + { + "epoch": 1.0189659437618537, + "grad_norm": 2.062920093536377, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7618594765663147, + "num_tokens": 261550905.0, + "step": 12357 + }, + { + "epoch": 1.0190484043869052, + "grad_norm": 2.1411092281341553, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.7706502676010132, + "num_tokens": 261575799.0, + "step": 12358 + }, + { + "epoch": 1.0191308650119568, + "grad_norm": 1.9448460340499878, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7802956700325012, + "num_tokens": 261606643.0, + "step": 12359 + }, + { + "epoch": 1.0192133256370084, + "grad_norm": 1.6783103942871094, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.7913810014724731, + "num_tokens": 261644199.0, + "step": 12360 + }, + { + "epoch": 1.01929578626206, + "grad_norm": 2.193150520324707, + "learning_rate": 1e-06, + "loss": 0.6205, + "mean_token_accuracy": 0.7898430824279785, + "num_tokens": 261667484.0, + "step": 12361 + }, + { + "epoch": 1.0193782468871113, + "grad_norm": 2.2337710857391357, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.785539984703064, + "num_tokens": 261692594.0, + "step": 12362 + }, + { + "epoch": 1.019460707512163, + "grad_norm": 1.820020318031311, + "learning_rate": 1e-06, + "loss": 0.5908, + "mean_token_accuracy": 0.8154247403144836, + "num_tokens": 261724594.0, + "step": 12363 + }, + { + "epoch": 1.0195431681372145, + "grad_norm": 1.9825836420059204, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7439761161804199, + "num_tokens": 261757145.0, + "step": 12364 + }, + { + "epoch": 1.019625628762266, + "grad_norm": 2.125659704208374, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7695395350456238, + "num_tokens": 261783959.0, + "step": 12365 + }, + { + "epoch": 1.0197080893873176, + "grad_norm": 1.9481741189956665, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7791824340820312, + "num_tokens": 261814089.0, + "step": 12366 + }, + { + "epoch": 1.0197905500123692, + "grad_norm": 2.251372814178467, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7467899322509766, + "num_tokens": 261841512.0, + "step": 12367 + }, + { + "epoch": 1.0198730106374205, + "grad_norm": 1.9349216222763062, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7559508681297302, + "num_tokens": 261868594.0, + "step": 12368 + }, + { + "epoch": 1.019955471262472, + "grad_norm": 1.9842623472213745, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.763464093208313, + "num_tokens": 261897842.0, + "step": 12369 + }, + { + "epoch": 1.0200379318875237, + "grad_norm": 1.8764066696166992, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7630382776260376, + "num_tokens": 261933633.0, + "step": 12370 + }, + { + "epoch": 1.0201203925125752, + "grad_norm": 1.9939749240875244, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7702378034591675, + "num_tokens": 261963299.0, + "step": 12371 + }, + { + "epoch": 1.0202028531376268, + "grad_norm": 2.066070079803467, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7592077851295471, + "num_tokens": 261993663.0, + "step": 12372 + }, + { + "epoch": 1.0202853137626784, + "grad_norm": 1.9007281064987183, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.7560827732086182, + "num_tokens": 262027729.0, + "step": 12373 + }, + { + "epoch": 1.02036777438773, + "grad_norm": 2.0242464542388916, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.7869462370872498, + "num_tokens": 262053724.0, + "step": 12374 + }, + { + "epoch": 1.0204502350127813, + "grad_norm": 2.0515711307525635, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7890763282775879, + "num_tokens": 262082434.0, + "step": 12375 + }, + { + "epoch": 1.0205326956378329, + "grad_norm": 1.6883577108383179, + "learning_rate": 1e-06, + "loss": 0.5131, + "mean_token_accuracy": 0.8283752202987671, + "num_tokens": 262116374.0, + "step": 12376 + }, + { + "epoch": 1.0206151562628845, + "grad_norm": 2.2826240062713623, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7344030141830444, + "num_tokens": 262140203.0, + "step": 12377 + }, + { + "epoch": 1.020697616887936, + "grad_norm": 1.708383321762085, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7536983489990234, + "num_tokens": 262175082.0, + "step": 12378 + }, + { + "epoch": 1.0207800775129876, + "grad_norm": 1.905909776687622, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7707221508026123, + "num_tokens": 262208085.0, + "step": 12379 + }, + { + "epoch": 1.0208625381380392, + "grad_norm": 2.1810708045959473, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7210836410522461, + "num_tokens": 262233892.0, + "step": 12380 + }, + { + "epoch": 1.0209449987630905, + "grad_norm": 1.8792352676391602, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7658344507217407, + "num_tokens": 262264281.0, + "step": 12381 + }, + { + "epoch": 1.021027459388142, + "grad_norm": 2.002293825149536, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7753753662109375, + "num_tokens": 262294463.0, + "step": 12382 + }, + { + "epoch": 1.0211099200131937, + "grad_norm": 1.9125707149505615, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7558122277259827, + "num_tokens": 262327122.0, + "step": 12383 + }, + { + "epoch": 1.0211923806382452, + "grad_norm": 2.066161632537842, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7651764154434204, + "num_tokens": 262356224.0, + "step": 12384 + }, + { + "epoch": 1.0212748412632968, + "grad_norm": 1.7774299383163452, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7373994588851929, + "num_tokens": 262395280.0, + "step": 12385 + }, + { + "epoch": 1.0213573018883484, + "grad_norm": 2.1440999507904053, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7324990034103394, + "num_tokens": 262424256.0, + "step": 12386 + }, + { + "epoch": 1.0214397625134, + "grad_norm": 2.273085832595825, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7729306221008301, + "num_tokens": 262447454.0, + "step": 12387 + }, + { + "epoch": 1.0215222231384513, + "grad_norm": 2.0963430404663086, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7635520696640015, + "num_tokens": 262475989.0, + "step": 12388 + }, + { + "epoch": 1.0216046837635029, + "grad_norm": 1.8733950853347778, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.7572516202926636, + "num_tokens": 262508090.0, + "step": 12389 + }, + { + "epoch": 1.0216871443885545, + "grad_norm": 2.583613634109497, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7238034009933472, + "num_tokens": 262529729.0, + "step": 12390 + }, + { + "epoch": 1.021769605013606, + "grad_norm": 1.9548635482788086, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7677960395812988, + "num_tokens": 262559285.0, + "step": 12391 + }, + { + "epoch": 1.0218520656386576, + "grad_norm": 1.8254637718200684, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7657544016838074, + "num_tokens": 262592865.0, + "step": 12392 + }, + { + "epoch": 1.0219345262637092, + "grad_norm": 2.014784336090088, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7760798335075378, + "num_tokens": 262621525.0, + "step": 12393 + }, + { + "epoch": 1.0220169868887605, + "grad_norm": 1.9383524656295776, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7544528245925903, + "num_tokens": 262657111.0, + "step": 12394 + }, + { + "epoch": 1.022099447513812, + "grad_norm": 1.9247349500656128, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7305715680122375, + "num_tokens": 262689481.0, + "step": 12395 + }, + { + "epoch": 1.0221819081388637, + "grad_norm": 1.9837018251419067, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7673243284225464, + "num_tokens": 262723905.0, + "step": 12396 + }, + { + "epoch": 1.0222643687639152, + "grad_norm": 1.838236689567566, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7685052752494812, + "num_tokens": 262757338.0, + "step": 12397 + }, + { + "epoch": 1.0223468293889668, + "grad_norm": 1.8273711204528809, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7717336416244507, + "num_tokens": 262789221.0, + "step": 12398 + }, + { + "epoch": 1.0224292900140184, + "grad_norm": 2.07060170173645, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7416322827339172, + "num_tokens": 262816535.0, + "step": 12399 + }, + { + "epoch": 1.0225117506390697, + "grad_norm": 1.6560447216033936, + "learning_rate": 1e-06, + "loss": 0.5838, + "mean_token_accuracy": 0.8068398833274841, + "num_tokens": 262853276.0, + "step": 12400 + }, + { + "epoch": 1.0225942112641213, + "grad_norm": 2.2700653076171875, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7509689331054688, + "num_tokens": 262878882.0, + "step": 12401 + }, + { + "epoch": 1.0226766718891729, + "grad_norm": 1.9251657724380493, + "learning_rate": 1e-06, + "loss": 0.5321, + "mean_token_accuracy": 0.8247437477111816, + "num_tokens": 262910682.0, + "step": 12402 + }, + { + "epoch": 1.0227591325142245, + "grad_norm": 1.627408504486084, + "learning_rate": 1e-06, + "loss": 0.5747, + "mean_token_accuracy": 0.812682032585144, + "num_tokens": 262946838.0, + "step": 12403 + }, + { + "epoch": 1.022841593139276, + "grad_norm": 1.8175352811813354, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7738175988197327, + "num_tokens": 262986058.0, + "step": 12404 + }, + { + "epoch": 1.0229240537643276, + "grad_norm": 1.744654893875122, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.7902536392211914, + "num_tokens": 263020759.0, + "step": 12405 + }, + { + "epoch": 1.0230065143893792, + "grad_norm": 2.0240156650543213, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7610976099967957, + "num_tokens": 263047587.0, + "step": 12406 + }, + { + "epoch": 1.0230889750144305, + "grad_norm": 2.843973398208618, + "learning_rate": 1e-06, + "loss": 0.5404, + "mean_token_accuracy": 0.8102338314056396, + "num_tokens": 263069019.0, + "step": 12407 + }, + { + "epoch": 1.023171435639482, + "grad_norm": 2.0682358741760254, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7788412570953369, + "num_tokens": 263096952.0, + "step": 12408 + }, + { + "epoch": 1.0232538962645337, + "grad_norm": 1.6945221424102783, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7849042415618896, + "num_tokens": 263132713.0, + "step": 12409 + }, + { + "epoch": 1.0233363568895852, + "grad_norm": 1.6872419118881226, + "learning_rate": 1e-06, + "loss": 0.5705, + "mean_token_accuracy": 0.8089132308959961, + "num_tokens": 263170568.0, + "step": 12410 + }, + { + "epoch": 1.0234188175146368, + "grad_norm": 1.726569414138794, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7466331720352173, + "num_tokens": 263208109.0, + "step": 12411 + }, + { + "epoch": 1.0235012781396884, + "grad_norm": 1.9757920503616333, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.744529128074646, + "num_tokens": 263239174.0, + "step": 12412 + }, + { + "epoch": 1.0235837387647397, + "grad_norm": 1.8917145729064941, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7647223472595215, + "num_tokens": 263271362.0, + "step": 12413 + }, + { + "epoch": 1.0236661993897913, + "grad_norm": 1.9214990139007568, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7397564649581909, + "num_tokens": 263301751.0, + "step": 12414 + }, + { + "epoch": 1.0237486600148429, + "grad_norm": 1.8015940189361572, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7426525354385376, + "num_tokens": 263337173.0, + "step": 12415 + }, + { + "epoch": 1.0238311206398945, + "grad_norm": 2.1034700870513916, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7303563356399536, + "num_tokens": 263365130.0, + "step": 12416 + }, + { + "epoch": 1.023913581264946, + "grad_norm": 2.0714969635009766, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.7798466682434082, + "num_tokens": 263393360.0, + "step": 12417 + }, + { + "epoch": 1.0239960418899976, + "grad_norm": 2.18855619430542, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7687837481498718, + "num_tokens": 263419142.0, + "step": 12418 + }, + { + "epoch": 1.0240785025150492, + "grad_norm": 1.9464306831359863, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.7883034348487854, + "num_tokens": 263450482.0, + "step": 12419 + }, + { + "epoch": 1.0241609631401005, + "grad_norm": 1.990751028060913, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7775499820709229, + "num_tokens": 263481682.0, + "step": 12420 + }, + { + "epoch": 1.024243423765152, + "grad_norm": 1.9148123264312744, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7620340585708618, + "num_tokens": 263513165.0, + "step": 12421 + }, + { + "epoch": 1.0243258843902037, + "grad_norm": 1.958864450454712, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.774104654788971, + "num_tokens": 263543104.0, + "step": 12422 + }, + { + "epoch": 1.0244083450152552, + "grad_norm": 2.1413791179656982, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7577463984489441, + "num_tokens": 263571161.0, + "step": 12423 + }, + { + "epoch": 1.0244908056403068, + "grad_norm": 1.967506766319275, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7756032943725586, + "num_tokens": 263604093.0, + "step": 12424 + }, + { + "epoch": 1.0245732662653584, + "grad_norm": 2.0160598754882812, + "learning_rate": 1e-06, + "loss": 0.6166, + "mean_token_accuracy": 0.789269208908081, + "num_tokens": 263631469.0, + "step": 12425 + }, + { + "epoch": 1.0246557268904097, + "grad_norm": 2.04323148727417, + "learning_rate": 1e-06, + "loss": 0.5145, + "mean_token_accuracy": 0.8300507068634033, + "num_tokens": 263656619.0, + "step": 12426 + }, + { + "epoch": 1.0247381875154613, + "grad_norm": 1.7974908351898193, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7790805101394653, + "num_tokens": 263688088.0, + "step": 12427 + }, + { + "epoch": 1.0248206481405129, + "grad_norm": 2.1711578369140625, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7187578082084656, + "num_tokens": 263716994.0, + "step": 12428 + }, + { + "epoch": 1.0249031087655645, + "grad_norm": 1.9899924993515015, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.7926533818244934, + "num_tokens": 263746637.0, + "step": 12429 + }, + { + "epoch": 1.024985569390616, + "grad_norm": 1.8655990362167358, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7355862259864807, + "num_tokens": 263782744.0, + "step": 12430 + }, + { + "epoch": 1.0250680300156676, + "grad_norm": 2.0308640003204346, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7623163461685181, + "num_tokens": 263811050.0, + "step": 12431 + }, + { + "epoch": 1.025150490640719, + "grad_norm": 2.0456838607788086, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7596330642700195, + "num_tokens": 263840832.0, + "step": 12432 + }, + { + "epoch": 1.0252329512657705, + "grad_norm": 1.73726224899292, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7679986953735352, + "num_tokens": 263876490.0, + "step": 12433 + }, + { + "epoch": 1.025315411890822, + "grad_norm": 1.9168075323104858, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.7831177115440369, + "num_tokens": 263908787.0, + "step": 12434 + }, + { + "epoch": 1.0253978725158737, + "grad_norm": 2.1082890033721924, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7751429080963135, + "num_tokens": 263934909.0, + "step": 12435 + }, + { + "epoch": 1.0254803331409252, + "grad_norm": 2.041020631790161, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7331327199935913, + "num_tokens": 263963109.0, + "step": 12436 + }, + { + "epoch": 1.0255627937659768, + "grad_norm": 2.2272701263427734, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7677050232887268, + "num_tokens": 263989925.0, + "step": 12437 + }, + { + "epoch": 1.0256452543910284, + "grad_norm": 2.261323928833008, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7377719283103943, + "num_tokens": 264014300.0, + "step": 12438 + }, + { + "epoch": 1.0257277150160797, + "grad_norm": 1.9583094120025635, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7396566867828369, + "num_tokens": 264046932.0, + "step": 12439 + }, + { + "epoch": 1.0258101756411313, + "grad_norm": 2.3556010723114014, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7272983193397522, + "num_tokens": 264071481.0, + "step": 12440 + }, + { + "epoch": 1.0258926362661829, + "grad_norm": 2.159554958343506, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7763102054595947, + "num_tokens": 264098544.0, + "step": 12441 + }, + { + "epoch": 1.0259750968912345, + "grad_norm": 1.9471244812011719, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7805881500244141, + "num_tokens": 264128820.0, + "step": 12442 + }, + { + "epoch": 1.026057557516286, + "grad_norm": 2.2058682441711426, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7194841504096985, + "num_tokens": 264156567.0, + "step": 12443 + }, + { + "epoch": 1.0261400181413376, + "grad_norm": 2.436145067214966, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7467629909515381, + "num_tokens": 264180924.0, + "step": 12444 + }, + { + "epoch": 1.026222478766389, + "grad_norm": 1.805086374282837, + "learning_rate": 1e-06, + "loss": 0.5781, + "mean_token_accuracy": 0.8173591494560242, + "num_tokens": 264213155.0, + "step": 12445 + }, + { + "epoch": 1.0263049393914405, + "grad_norm": 1.9081637859344482, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7690193057060242, + "num_tokens": 264245697.0, + "step": 12446 + }, + { + "epoch": 1.026387400016492, + "grad_norm": 2.2065682411193848, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.746770977973938, + "num_tokens": 264269875.0, + "step": 12447 + }, + { + "epoch": 1.0264698606415437, + "grad_norm": 2.4686851501464844, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.7002760171890259, + "num_tokens": 264292344.0, + "step": 12448 + }, + { + "epoch": 1.0265523212665952, + "grad_norm": 1.93061363697052, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.737941563129425, + "num_tokens": 264324018.0, + "step": 12449 + }, + { + "epoch": 1.0266347818916468, + "grad_norm": 1.9932340383529663, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7746385931968689, + "num_tokens": 264353084.0, + "step": 12450 + }, + { + "epoch": 1.0267172425166984, + "grad_norm": 1.8368480205535889, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7687466740608215, + "num_tokens": 264387781.0, + "step": 12451 + }, + { + "epoch": 1.0267997031417497, + "grad_norm": 1.956601619720459, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7747184634208679, + "num_tokens": 264416684.0, + "step": 12452 + }, + { + "epoch": 1.0268821637668013, + "grad_norm": 2.2338809967041016, + "learning_rate": 1e-06, + "loss": 0.8234, + "mean_token_accuracy": 0.735592782497406, + "num_tokens": 264443912.0, + "step": 12453 + }, + { + "epoch": 1.0269646243918529, + "grad_norm": 2.107511043548584, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7407780885696411, + "num_tokens": 264471555.0, + "step": 12454 + }, + { + "epoch": 1.0270470850169044, + "grad_norm": 2.174069404602051, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7467194199562073, + "num_tokens": 264496346.0, + "step": 12455 + }, + { + "epoch": 1.027129545641956, + "grad_norm": 1.830958366394043, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.7810990810394287, + "num_tokens": 264528514.0, + "step": 12456 + }, + { + "epoch": 1.0272120062670076, + "grad_norm": 1.9340362548828125, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7421168684959412, + "num_tokens": 264560528.0, + "step": 12457 + }, + { + "epoch": 1.027294466892059, + "grad_norm": 1.8924534320831299, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.741818368434906, + "num_tokens": 264594232.0, + "step": 12458 + }, + { + "epoch": 1.0273769275171105, + "grad_norm": 1.9311329126358032, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7412244081497192, + "num_tokens": 264626518.0, + "step": 12459 + }, + { + "epoch": 1.027459388142162, + "grad_norm": 1.7812551259994507, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7749505639076233, + "num_tokens": 264661777.0, + "step": 12460 + }, + { + "epoch": 1.0275418487672137, + "grad_norm": 1.7668081521987915, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.752839207649231, + "num_tokens": 264701407.0, + "step": 12461 + }, + { + "epoch": 1.0276243093922652, + "grad_norm": 2.4004626274108887, + "learning_rate": 1e-06, + "loss": 0.7483, + "mean_token_accuracy": 0.7656487226486206, + "num_tokens": 264722463.0, + "step": 12462 + }, + { + "epoch": 1.0277067700173168, + "grad_norm": 1.9147491455078125, + "learning_rate": 1e-06, + "loss": 0.7563, + "mean_token_accuracy": 0.7557626962661743, + "num_tokens": 264754922.0, + "step": 12463 + }, + { + "epoch": 1.0277892306423682, + "grad_norm": 2.2069735527038574, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7412449717521667, + "num_tokens": 264782026.0, + "step": 12464 + }, + { + "epoch": 1.0278716912674197, + "grad_norm": 2.0539283752441406, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.7498993873596191, + "num_tokens": 264811641.0, + "step": 12465 + }, + { + "epoch": 1.0279541518924713, + "grad_norm": 1.7973848581314087, + "learning_rate": 1e-06, + "loss": 0.5781, + "mean_token_accuracy": 0.8036550283432007, + "num_tokens": 264848839.0, + "step": 12466 + }, + { + "epoch": 1.0280366125175229, + "grad_norm": 1.880887746810913, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7490662932395935, + "num_tokens": 264882339.0, + "step": 12467 + }, + { + "epoch": 1.0281190731425744, + "grad_norm": 2.3032901287078857, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7447142601013184, + "num_tokens": 264905636.0, + "step": 12468 + }, + { + "epoch": 1.028201533767626, + "grad_norm": 2.183778762817383, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7128087282180786, + "num_tokens": 264933037.0, + "step": 12469 + }, + { + "epoch": 1.0282839943926776, + "grad_norm": 2.1807870864868164, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7692888379096985, + "num_tokens": 264958119.0, + "step": 12470 + }, + { + "epoch": 1.028366455017729, + "grad_norm": 1.9260295629501343, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7811766266822815, + "num_tokens": 264991211.0, + "step": 12471 + }, + { + "epoch": 1.0284489156427805, + "grad_norm": 2.0711915493011475, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7442348003387451, + "num_tokens": 265020697.0, + "step": 12472 + }, + { + "epoch": 1.028531376267832, + "grad_norm": 2.03676700592041, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7494745254516602, + "num_tokens": 265052125.0, + "step": 12473 + }, + { + "epoch": 1.0286138368928837, + "grad_norm": 2.1561572551727295, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7310197353363037, + "num_tokens": 265079116.0, + "step": 12474 + }, + { + "epoch": 1.0286962975179352, + "grad_norm": 2.3740878105163574, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7387934923171997, + "num_tokens": 265103129.0, + "step": 12475 + }, + { + "epoch": 1.0287787581429868, + "grad_norm": 2.1663460731506348, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7599337697029114, + "num_tokens": 265128040.0, + "step": 12476 + }, + { + "epoch": 1.0288612187680382, + "grad_norm": 2.0903408527374268, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7640043497085571, + "num_tokens": 265154613.0, + "step": 12477 + }, + { + "epoch": 1.0289436793930897, + "grad_norm": 2.0446434020996094, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7430693507194519, + "num_tokens": 265184583.0, + "step": 12478 + }, + { + "epoch": 1.0290261400181413, + "grad_norm": 1.894161343574524, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7728897333145142, + "num_tokens": 265217847.0, + "step": 12479 + }, + { + "epoch": 1.0291086006431929, + "grad_norm": 1.9221270084381104, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.778172492980957, + "num_tokens": 265248530.0, + "step": 12480 + }, + { + "epoch": 1.0291910612682444, + "grad_norm": 2.0174431800842285, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7367210388183594, + "num_tokens": 265277792.0, + "step": 12481 + }, + { + "epoch": 1.029273521893296, + "grad_norm": 2.1781163215637207, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7525871992111206, + "num_tokens": 265304675.0, + "step": 12482 + }, + { + "epoch": 1.0293559825183476, + "grad_norm": 2.2016749382019043, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7384214401245117, + "num_tokens": 265329464.0, + "step": 12483 + }, + { + "epoch": 1.029438443143399, + "grad_norm": 2.153085231781006, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.755297064781189, + "num_tokens": 265358579.0, + "step": 12484 + }, + { + "epoch": 1.0295209037684505, + "grad_norm": 1.9752824306488037, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7565579414367676, + "num_tokens": 265388023.0, + "step": 12485 + }, + { + "epoch": 1.029603364393502, + "grad_norm": 2.3690786361694336, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7431073784828186, + "num_tokens": 265413196.0, + "step": 12486 + }, + { + "epoch": 1.0296858250185537, + "grad_norm": 1.9395124912261963, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.743806779384613, + "num_tokens": 265444531.0, + "step": 12487 + }, + { + "epoch": 1.0297682856436052, + "grad_norm": 1.8611218929290771, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7527675032615662, + "num_tokens": 265475747.0, + "step": 12488 + }, + { + "epoch": 1.0298507462686568, + "grad_norm": 2.0730702877044678, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7384607791900635, + "num_tokens": 265505282.0, + "step": 12489 + }, + { + "epoch": 1.0299332068937082, + "grad_norm": 2.0491600036621094, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7604321241378784, + "num_tokens": 265537337.0, + "step": 12490 + }, + { + "epoch": 1.0300156675187597, + "grad_norm": 2.054316759109497, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.77561354637146, + "num_tokens": 265563245.0, + "step": 12491 + }, + { + "epoch": 1.0300981281438113, + "grad_norm": 1.8339601755142212, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7503625154495239, + "num_tokens": 265598623.0, + "step": 12492 + }, + { + "epoch": 1.0301805887688629, + "grad_norm": 2.056112766265869, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7588438391685486, + "num_tokens": 265628516.0, + "step": 12493 + }, + { + "epoch": 1.0302630493939144, + "grad_norm": 1.9772428274154663, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7364192605018616, + "num_tokens": 265661234.0, + "step": 12494 + }, + { + "epoch": 1.030345510018966, + "grad_norm": 2.0315654277801514, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7207402586936951, + "num_tokens": 265694721.0, + "step": 12495 + }, + { + "epoch": 1.0304279706440176, + "grad_norm": 2.1154050827026367, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.742754340171814, + "num_tokens": 265722782.0, + "step": 12496 + }, + { + "epoch": 1.030510431269069, + "grad_norm": 1.9830447435379028, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.78257155418396, + "num_tokens": 265752474.0, + "step": 12497 + }, + { + "epoch": 1.0305928918941205, + "grad_norm": 1.9218577146530151, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7455257773399353, + "num_tokens": 265785559.0, + "step": 12498 + }, + { + "epoch": 1.030675352519172, + "grad_norm": 1.7912321090698242, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7890573740005493, + "num_tokens": 265820425.0, + "step": 12499 + }, + { + "epoch": 1.0307578131442237, + "grad_norm": 2.058164119720459, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.771851122379303, + "num_tokens": 265848179.0, + "step": 12500 + }, + { + "epoch": 1.0308402737692752, + "grad_norm": 1.8923946619033813, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.7907317876815796, + "num_tokens": 265880507.0, + "step": 12501 + }, + { + "epoch": 1.0309227343943268, + "grad_norm": 1.9266799688339233, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7235747575759888, + "num_tokens": 265913591.0, + "step": 12502 + }, + { + "epoch": 1.0310051950193782, + "grad_norm": 2.0924341678619385, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7560826539993286, + "num_tokens": 265943386.0, + "step": 12503 + }, + { + "epoch": 1.0310876556444297, + "grad_norm": 1.8634616136550903, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7817136645317078, + "num_tokens": 265975972.0, + "step": 12504 + }, + { + "epoch": 1.0311701162694813, + "grad_norm": 2.294858932495117, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7506933212280273, + "num_tokens": 265999730.0, + "step": 12505 + }, + { + "epoch": 1.0312525768945329, + "grad_norm": 1.884527325630188, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.765387773513794, + "num_tokens": 266033452.0, + "step": 12506 + }, + { + "epoch": 1.0313350375195844, + "grad_norm": 2.296034574508667, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7548913955688477, + "num_tokens": 266059396.0, + "step": 12507 + }, + { + "epoch": 1.031417498144636, + "grad_norm": 1.9907727241516113, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7649613618850708, + "num_tokens": 266087034.0, + "step": 12508 + }, + { + "epoch": 1.0314999587696874, + "grad_norm": 1.9639194011688232, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7308247685432434, + "num_tokens": 266119107.0, + "step": 12509 + }, + { + "epoch": 1.031582419394739, + "grad_norm": 2.130542039871216, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7435792088508606, + "num_tokens": 266148336.0, + "step": 12510 + }, + { + "epoch": 1.0316648800197905, + "grad_norm": 1.5398249626159668, + "learning_rate": 1e-06, + "loss": 0.4881, + "mean_token_accuracy": 0.8307061195373535, + "num_tokens": 266183069.0, + "step": 12511 + }, + { + "epoch": 1.031747340644842, + "grad_norm": 1.908073902130127, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7285938858985901, + "num_tokens": 266215097.0, + "step": 12512 + }, + { + "epoch": 1.0318298012698937, + "grad_norm": 1.851480484008789, + "learning_rate": 1e-06, + "loss": 0.7303, + "mean_token_accuracy": 0.7673658728599548, + "num_tokens": 266249670.0, + "step": 12513 + }, + { + "epoch": 1.0319122618949452, + "grad_norm": 2.0797977447509766, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7610734701156616, + "num_tokens": 266277522.0, + "step": 12514 + }, + { + "epoch": 1.0319947225199968, + "grad_norm": 2.1378347873687744, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7445931434631348, + "num_tokens": 266304924.0, + "step": 12515 + }, + { + "epoch": 1.0320771831450481, + "grad_norm": 2.1403558254241943, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7623164653778076, + "num_tokens": 266331774.0, + "step": 12516 + }, + { + "epoch": 1.0321596437700997, + "grad_norm": 1.833418369293213, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7818032503128052, + "num_tokens": 266365958.0, + "step": 12517 + }, + { + "epoch": 1.0322421043951513, + "grad_norm": 1.976318597793579, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7341123819351196, + "num_tokens": 266394887.0, + "step": 12518 + }, + { + "epoch": 1.0323245650202029, + "grad_norm": 1.9724937677383423, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7316293120384216, + "num_tokens": 266426935.0, + "step": 12519 + }, + { + "epoch": 1.0324070256452544, + "grad_norm": 1.9975506067276, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.729712963104248, + "num_tokens": 266457008.0, + "step": 12520 + }, + { + "epoch": 1.032489486270306, + "grad_norm": 2.153348684310913, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.732813835144043, + "num_tokens": 266485693.0, + "step": 12521 + }, + { + "epoch": 1.0325719468953574, + "grad_norm": 2.1965558528900146, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7363588809967041, + "num_tokens": 266513271.0, + "step": 12522 + }, + { + "epoch": 1.032654407520409, + "grad_norm": 2.0181422233581543, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7529939413070679, + "num_tokens": 266542684.0, + "step": 12523 + }, + { + "epoch": 1.0327368681454605, + "grad_norm": 1.8859388828277588, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7308663129806519, + "num_tokens": 266577017.0, + "step": 12524 + }, + { + "epoch": 1.032819328770512, + "grad_norm": 2.132175922393799, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7379878759384155, + "num_tokens": 266603990.0, + "step": 12525 + }, + { + "epoch": 1.0329017893955637, + "grad_norm": 1.803462266921997, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7637016773223877, + "num_tokens": 266637342.0, + "step": 12526 + }, + { + "epoch": 1.0329842500206152, + "grad_norm": 2.079678773880005, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.757831871509552, + "num_tokens": 266667306.0, + "step": 12527 + }, + { + "epoch": 1.0330667106456668, + "grad_norm": 1.8904510736465454, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.727906346321106, + "num_tokens": 266700658.0, + "step": 12528 + }, + { + "epoch": 1.0331491712707181, + "grad_norm": 1.9040377140045166, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7624539136886597, + "num_tokens": 266732752.0, + "step": 12529 + }, + { + "epoch": 1.0332316318957697, + "grad_norm": 1.8223350048065186, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7708351016044617, + "num_tokens": 266765168.0, + "step": 12530 + }, + { + "epoch": 1.0333140925208213, + "grad_norm": 2.154431104660034, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.76991206407547, + "num_tokens": 266792531.0, + "step": 12531 + }, + { + "epoch": 1.0333965531458729, + "grad_norm": 1.8411732912063599, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.7662521600723267, + "num_tokens": 266824091.0, + "step": 12532 + }, + { + "epoch": 1.0334790137709244, + "grad_norm": 2.008000135421753, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.728480339050293, + "num_tokens": 266857026.0, + "step": 12533 + }, + { + "epoch": 1.033561474395976, + "grad_norm": 1.9922927618026733, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7637026906013489, + "num_tokens": 266885725.0, + "step": 12534 + }, + { + "epoch": 1.0336439350210274, + "grad_norm": 1.8848336935043335, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7870597243309021, + "num_tokens": 266917984.0, + "step": 12535 + }, + { + "epoch": 1.033726395646079, + "grad_norm": 1.9167507886886597, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.7578939199447632, + "num_tokens": 266949945.0, + "step": 12536 + }, + { + "epoch": 1.0338088562711305, + "grad_norm": 1.9678338766098022, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7669700384140015, + "num_tokens": 266982037.0, + "step": 12537 + }, + { + "epoch": 1.033891316896182, + "grad_norm": 2.2437615394592285, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7501202821731567, + "num_tokens": 267008078.0, + "step": 12538 + }, + { + "epoch": 1.0339737775212337, + "grad_norm": 2.189809799194336, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7497233152389526, + "num_tokens": 267033737.0, + "step": 12539 + }, + { + "epoch": 1.0340562381462852, + "grad_norm": 1.8758121728897095, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7705250978469849, + "num_tokens": 267063908.0, + "step": 12540 + }, + { + "epoch": 1.0341386987713368, + "grad_norm": 1.8437477350234985, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7885158061981201, + "num_tokens": 267095750.0, + "step": 12541 + }, + { + "epoch": 1.0342211593963881, + "grad_norm": 1.6955509185791016, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.7672545909881592, + "num_tokens": 267133905.0, + "step": 12542 + }, + { + "epoch": 1.0343036200214397, + "grad_norm": 2.2858712673187256, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7500431537628174, + "num_tokens": 267157872.0, + "step": 12543 + }, + { + "epoch": 1.0343860806464913, + "grad_norm": 1.8425121307373047, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7497262358665466, + "num_tokens": 267193463.0, + "step": 12544 + }, + { + "epoch": 1.0344685412715429, + "grad_norm": 2.087916612625122, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7615883350372314, + "num_tokens": 267219650.0, + "step": 12545 + }, + { + "epoch": 1.0345510018965944, + "grad_norm": 1.9183361530303955, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.7782086730003357, + "num_tokens": 267252354.0, + "step": 12546 + }, + { + "epoch": 1.034633462521646, + "grad_norm": 1.9904729127883911, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.70488041639328, + "num_tokens": 267288034.0, + "step": 12547 + }, + { + "epoch": 1.0347159231466974, + "grad_norm": 1.9516551494598389, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7658093571662903, + "num_tokens": 267319266.0, + "step": 12548 + }, + { + "epoch": 1.034798383771749, + "grad_norm": 2.040907144546509, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7197548151016235, + "num_tokens": 267348659.0, + "step": 12549 + }, + { + "epoch": 1.0348808443968005, + "grad_norm": 2.144353151321411, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7680981159210205, + "num_tokens": 267376460.0, + "step": 12550 + }, + { + "epoch": 1.034963305021852, + "grad_norm": 1.8266346454620361, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.7847639322280884, + "num_tokens": 267411552.0, + "step": 12551 + }, + { + "epoch": 1.0350457656469036, + "grad_norm": 2.1516292095184326, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7307072877883911, + "num_tokens": 267439165.0, + "step": 12552 + }, + { + "epoch": 1.0351282262719552, + "grad_norm": 2.3115956783294678, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7313352823257446, + "num_tokens": 267464187.0, + "step": 12553 + }, + { + "epoch": 1.0352106868970066, + "grad_norm": 2.1734893321990967, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7265618443489075, + "num_tokens": 267489555.0, + "step": 12554 + }, + { + "epoch": 1.0352931475220581, + "grad_norm": 1.9070988893508911, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.7901627421379089, + "num_tokens": 267520401.0, + "step": 12555 + }, + { + "epoch": 1.0353756081471097, + "grad_norm": 2.057262897491455, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.732977032661438, + "num_tokens": 267548947.0, + "step": 12556 + }, + { + "epoch": 1.0354580687721613, + "grad_norm": 2.073140859603882, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7673600912094116, + "num_tokens": 267577222.0, + "step": 12557 + }, + { + "epoch": 1.0355405293972129, + "grad_norm": 1.9703874588012695, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7799152135848999, + "num_tokens": 267608333.0, + "step": 12558 + }, + { + "epoch": 1.0356229900222644, + "grad_norm": 1.894591212272644, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7322715520858765, + "num_tokens": 267640064.0, + "step": 12559 + }, + { + "epoch": 1.035705450647316, + "grad_norm": 2.1501786708831787, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7561294436454773, + "num_tokens": 267664731.0, + "step": 12560 + }, + { + "epoch": 1.0357879112723674, + "grad_norm": 1.9096678495407104, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7492856979370117, + "num_tokens": 267698888.0, + "step": 12561 + }, + { + "epoch": 1.035870371897419, + "grad_norm": 1.8471689224243164, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7675572633743286, + "num_tokens": 267731889.0, + "step": 12562 + }, + { + "epoch": 1.0359528325224705, + "grad_norm": 2.029287099838257, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7822020053863525, + "num_tokens": 267760901.0, + "step": 12563 + }, + { + "epoch": 1.036035293147522, + "grad_norm": 2.135075330734253, + "learning_rate": 1e-06, + "loss": 0.7483, + "mean_token_accuracy": 0.7617048621177673, + "num_tokens": 267788551.0, + "step": 12564 + }, + { + "epoch": 1.0361177537725736, + "grad_norm": 2.2474241256713867, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7504258155822754, + "num_tokens": 267813051.0, + "step": 12565 + }, + { + "epoch": 1.0362002143976252, + "grad_norm": 2.021575689315796, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7590600252151489, + "num_tokens": 267841668.0, + "step": 12566 + }, + { + "epoch": 1.0362826750226766, + "grad_norm": 1.952453374862671, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7083761692047119, + "num_tokens": 267873349.0, + "step": 12567 + }, + { + "epoch": 1.0363651356477281, + "grad_norm": 1.9113932847976685, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7478024363517761, + "num_tokens": 267902399.0, + "step": 12568 + }, + { + "epoch": 1.0364475962727797, + "grad_norm": 1.8039805889129639, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.7879229187965393, + "num_tokens": 267935935.0, + "step": 12569 + }, + { + "epoch": 1.0365300568978313, + "grad_norm": 1.7585327625274658, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7733867764472961, + "num_tokens": 267972420.0, + "step": 12570 + }, + { + "epoch": 1.0366125175228829, + "grad_norm": 2.1275789737701416, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.7452998161315918, + "num_tokens": 267998909.0, + "step": 12571 + }, + { + "epoch": 1.0366949781479344, + "grad_norm": 1.8625737428665161, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.7897289991378784, + "num_tokens": 268030875.0, + "step": 12572 + }, + { + "epoch": 1.036777438772986, + "grad_norm": 1.9005340337753296, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7350651621818542, + "num_tokens": 268063386.0, + "step": 12573 + }, + { + "epoch": 1.0368598993980374, + "grad_norm": 1.8609509468078613, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.7738369703292847, + "num_tokens": 268093890.0, + "step": 12574 + }, + { + "epoch": 1.036942360023089, + "grad_norm": 2.0644242763519287, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7357157468795776, + "num_tokens": 268120450.0, + "step": 12575 + }, + { + "epoch": 1.0370248206481405, + "grad_norm": 2.2259016036987305, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7255688905715942, + "num_tokens": 268147866.0, + "step": 12576 + }, + { + "epoch": 1.037107281273192, + "grad_norm": 1.8141884803771973, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7733951807022095, + "num_tokens": 268180758.0, + "step": 12577 + }, + { + "epoch": 1.0371897418982436, + "grad_norm": 1.9264293909072876, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7666373252868652, + "num_tokens": 268211694.0, + "step": 12578 + }, + { + "epoch": 1.0372722025232952, + "grad_norm": 2.1239967346191406, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.741280198097229, + "num_tokens": 268239147.0, + "step": 12579 + }, + { + "epoch": 1.0373546631483466, + "grad_norm": 1.966090440750122, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7417387962341309, + "num_tokens": 268270213.0, + "step": 12580 + }, + { + "epoch": 1.0374371237733981, + "grad_norm": 1.8533512353897095, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7775459885597229, + "num_tokens": 268301565.0, + "step": 12581 + }, + { + "epoch": 1.0375195843984497, + "grad_norm": 2.1658847332000732, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7561154365539551, + "num_tokens": 268326914.0, + "step": 12582 + }, + { + "epoch": 1.0376020450235013, + "grad_norm": 1.9306044578552246, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.7834481596946716, + "num_tokens": 268357972.0, + "step": 12583 + }, + { + "epoch": 1.0376845056485529, + "grad_norm": 1.9628514051437378, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7436567544937134, + "num_tokens": 268389596.0, + "step": 12584 + }, + { + "epoch": 1.0377669662736044, + "grad_norm": 1.8434934616088867, + "learning_rate": 1e-06, + "loss": 0.6023, + "mean_token_accuracy": 0.8017874956130981, + "num_tokens": 268420959.0, + "step": 12585 + }, + { + "epoch": 1.037849426898656, + "grad_norm": 1.8953510522842407, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7502756118774414, + "num_tokens": 268451418.0, + "step": 12586 + }, + { + "epoch": 1.0379318875237074, + "grad_norm": 1.8966271877288818, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.7545597553253174, + "num_tokens": 268485258.0, + "step": 12587 + }, + { + "epoch": 1.038014348148759, + "grad_norm": 2.026858329772949, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7490785717964172, + "num_tokens": 268515565.0, + "step": 12588 + }, + { + "epoch": 1.0380968087738105, + "grad_norm": 1.8559826612472534, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7727830410003662, + "num_tokens": 268555047.0, + "step": 12589 + }, + { + "epoch": 1.038179269398862, + "grad_norm": 2.114462375640869, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7660201787948608, + "num_tokens": 268583958.0, + "step": 12590 + }, + { + "epoch": 1.0382617300239136, + "grad_norm": 2.028524398803711, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7382258176803589, + "num_tokens": 268613036.0, + "step": 12591 + }, + { + "epoch": 1.0383441906489652, + "grad_norm": 1.969692349433899, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7532581090927124, + "num_tokens": 268644524.0, + "step": 12592 + }, + { + "epoch": 1.0384266512740166, + "grad_norm": 2.0446908473968506, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7464419603347778, + "num_tokens": 268671938.0, + "step": 12593 + }, + { + "epoch": 1.0385091118990681, + "grad_norm": 1.9215731620788574, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.7878626585006714, + "num_tokens": 268701521.0, + "step": 12594 + }, + { + "epoch": 1.0385915725241197, + "grad_norm": 2.123201370239258, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7530317902565002, + "num_tokens": 268731113.0, + "step": 12595 + }, + { + "epoch": 1.0386740331491713, + "grad_norm": 2.167459011077881, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7442125678062439, + "num_tokens": 268757529.0, + "step": 12596 + }, + { + "epoch": 1.0387564937742229, + "grad_norm": 1.7128424644470215, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.7931540012359619, + "num_tokens": 268798617.0, + "step": 12597 + }, + { + "epoch": 1.0388389543992744, + "grad_norm": 2.0377116203308105, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.723578929901123, + "num_tokens": 268828677.0, + "step": 12598 + }, + { + "epoch": 1.0389214150243258, + "grad_norm": 2.2143197059631348, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7262111306190491, + "num_tokens": 268854926.0, + "step": 12599 + }, + { + "epoch": 1.0390038756493774, + "grad_norm": 2.346498727798462, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7405089139938354, + "num_tokens": 268877937.0, + "step": 12600 + }, + { + "epoch": 1.039086336274429, + "grad_norm": 2.0205113887786865, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7325141429901123, + "num_tokens": 268904492.0, + "step": 12601 + }, + { + "epoch": 1.0391687968994805, + "grad_norm": 2.0497868061065674, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7481446862220764, + "num_tokens": 268932230.0, + "step": 12602 + }, + { + "epoch": 1.039251257524532, + "grad_norm": 1.9837298393249512, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.7552344799041748, + "num_tokens": 268963658.0, + "step": 12603 + }, + { + "epoch": 1.0393337181495836, + "grad_norm": 1.8319406509399414, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7638546228408813, + "num_tokens": 268997421.0, + "step": 12604 + }, + { + "epoch": 1.0394161787746352, + "grad_norm": 2.0201871395111084, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7823587656021118, + "num_tokens": 269027216.0, + "step": 12605 + }, + { + "epoch": 1.0394986393996866, + "grad_norm": 1.7024333477020264, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.7925444841384888, + "num_tokens": 269063200.0, + "step": 12606 + }, + { + "epoch": 1.0395811000247381, + "grad_norm": 1.7893561124801636, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.7852834463119507, + "num_tokens": 269097630.0, + "step": 12607 + }, + { + "epoch": 1.0396635606497897, + "grad_norm": 2.1725308895111084, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7423731088638306, + "num_tokens": 269122859.0, + "step": 12608 + }, + { + "epoch": 1.0397460212748413, + "grad_norm": 1.7688415050506592, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.7975748777389526, + "num_tokens": 269157587.0, + "step": 12609 + }, + { + "epoch": 1.0398284818998929, + "grad_norm": 2.0814623832702637, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7815210819244385, + "num_tokens": 269184550.0, + "step": 12610 + }, + { + "epoch": 1.0399109425249444, + "grad_norm": 2.086200714111328, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7820112705230713, + "num_tokens": 269211498.0, + "step": 12611 + }, + { + "epoch": 1.0399934031499958, + "grad_norm": 1.9691723585128784, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7883065938949585, + "num_tokens": 269242106.0, + "step": 12612 + }, + { + "epoch": 1.0400758637750473, + "grad_norm": 1.9296270608901978, + "learning_rate": 1e-06, + "loss": 0.5556, + "mean_token_accuracy": 0.8163989782333374, + "num_tokens": 269273878.0, + "step": 12613 + }, + { + "epoch": 1.040158324400099, + "grad_norm": 2.1434130668640137, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7149859666824341, + "num_tokens": 269304136.0, + "step": 12614 + }, + { + "epoch": 1.0402407850251505, + "grad_norm": 2.1386783123016357, + "learning_rate": 1e-06, + "loss": 0.5406, + "mean_token_accuracy": 0.8180172443389893, + "num_tokens": 269327268.0, + "step": 12615 + }, + { + "epoch": 1.040323245650202, + "grad_norm": 1.9819856882095337, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.776726484298706, + "num_tokens": 269355756.0, + "step": 12616 + }, + { + "epoch": 1.0404057062752536, + "grad_norm": 1.9648354053497314, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7475770711898804, + "num_tokens": 269391264.0, + "step": 12617 + }, + { + "epoch": 1.0404881669003052, + "grad_norm": 1.8752031326293945, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7470592260360718, + "num_tokens": 269426964.0, + "step": 12618 + }, + { + "epoch": 1.0405706275253566, + "grad_norm": 1.6641887426376343, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7752788662910461, + "num_tokens": 269468407.0, + "step": 12619 + }, + { + "epoch": 1.0406530881504081, + "grad_norm": 1.8206019401550293, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7711625099182129, + "num_tokens": 269502056.0, + "step": 12620 + }, + { + "epoch": 1.0407355487754597, + "grad_norm": 1.9406081438064575, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7498950958251953, + "num_tokens": 269537065.0, + "step": 12621 + }, + { + "epoch": 1.0408180094005113, + "grad_norm": 1.9836312532424927, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7490151524543762, + "num_tokens": 269566696.0, + "step": 12622 + }, + { + "epoch": 1.0409004700255629, + "grad_norm": 1.7694834470748901, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7782028913497925, + "num_tokens": 269603056.0, + "step": 12623 + }, + { + "epoch": 1.0409829306506144, + "grad_norm": 2.467195749282837, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7639589309692383, + "num_tokens": 269625245.0, + "step": 12624 + }, + { + "epoch": 1.0410653912756658, + "grad_norm": 1.8972997665405273, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7361956238746643, + "num_tokens": 269656304.0, + "step": 12625 + }, + { + "epoch": 1.0411478519007173, + "grad_norm": 2.236938238143921, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7496480345726013, + "num_tokens": 269681216.0, + "step": 12626 + }, + { + "epoch": 1.041230312525769, + "grad_norm": 1.9116276502609253, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.745847761631012, + "num_tokens": 269714150.0, + "step": 12627 + }, + { + "epoch": 1.0413127731508205, + "grad_norm": 2.0231990814208984, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.7940114140510559, + "num_tokens": 269743313.0, + "step": 12628 + }, + { + "epoch": 1.041395233775872, + "grad_norm": 2.321009874343872, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7480292320251465, + "num_tokens": 269767261.0, + "step": 12629 + }, + { + "epoch": 1.0414776944009236, + "grad_norm": 1.900816559791565, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7427093982696533, + "num_tokens": 269799806.0, + "step": 12630 + }, + { + "epoch": 1.0415601550259752, + "grad_norm": 1.890350580215454, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7434237003326416, + "num_tokens": 269833296.0, + "step": 12631 + }, + { + "epoch": 1.0416426156510266, + "grad_norm": 1.9633560180664062, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7798784971237183, + "num_tokens": 269862938.0, + "step": 12632 + }, + { + "epoch": 1.0417250762760781, + "grad_norm": 1.9884477853775024, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7778490781784058, + "num_tokens": 269892842.0, + "step": 12633 + }, + { + "epoch": 1.0418075369011297, + "grad_norm": 2.2240467071533203, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7467105388641357, + "num_tokens": 269916113.0, + "step": 12634 + }, + { + "epoch": 1.0418899975261813, + "grad_norm": 2.026890754699707, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7638698816299438, + "num_tokens": 269947415.0, + "step": 12635 + }, + { + "epoch": 1.0419724581512328, + "grad_norm": 1.9580531120300293, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7836397886276245, + "num_tokens": 269980614.0, + "step": 12636 + }, + { + "epoch": 1.0420549187762844, + "grad_norm": 2.4008595943450928, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7411861419677734, + "num_tokens": 270002657.0, + "step": 12637 + }, + { + "epoch": 1.0421373794013358, + "grad_norm": 2.0452237129211426, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.751654863357544, + "num_tokens": 270032543.0, + "step": 12638 + }, + { + "epoch": 1.0422198400263873, + "grad_norm": 1.8991975784301758, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.7905799150466919, + "num_tokens": 270062926.0, + "step": 12639 + }, + { + "epoch": 1.042302300651439, + "grad_norm": 2.0783655643463135, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7470934987068176, + "num_tokens": 270093598.0, + "step": 12640 + }, + { + "epoch": 1.0423847612764905, + "grad_norm": 1.9875599145889282, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7769648432731628, + "num_tokens": 270124176.0, + "step": 12641 + }, + { + "epoch": 1.042467221901542, + "grad_norm": 1.9299311637878418, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7747616767883301, + "num_tokens": 270156824.0, + "step": 12642 + }, + { + "epoch": 1.0425496825265936, + "grad_norm": 2.1092214584350586, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7701120376586914, + "num_tokens": 270182190.0, + "step": 12643 + }, + { + "epoch": 1.042632143151645, + "grad_norm": 1.9292645454406738, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7621721029281616, + "num_tokens": 270215476.0, + "step": 12644 + }, + { + "epoch": 1.0427146037766966, + "grad_norm": 2.0634689331054688, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7856465578079224, + "num_tokens": 270242198.0, + "step": 12645 + }, + { + "epoch": 1.0427970644017481, + "grad_norm": 2.279632568359375, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7269433736801147, + "num_tokens": 270266502.0, + "step": 12646 + }, + { + "epoch": 1.0428795250267997, + "grad_norm": 1.7268739938735962, + "learning_rate": 1e-06, + "loss": 0.7621, + "mean_token_accuracy": 0.7638740539550781, + "num_tokens": 270305926.0, + "step": 12647 + }, + { + "epoch": 1.0429619856518513, + "grad_norm": 1.9101390838623047, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7253087162971497, + "num_tokens": 270339689.0, + "step": 12648 + }, + { + "epoch": 1.0430444462769028, + "grad_norm": 2.2131407260894775, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7503293752670288, + "num_tokens": 270364742.0, + "step": 12649 + }, + { + "epoch": 1.0431269069019544, + "grad_norm": 1.9036039113998413, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7724621295928955, + "num_tokens": 270397895.0, + "step": 12650 + }, + { + "epoch": 1.0432093675270058, + "grad_norm": 2.1094970703125, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7600213289260864, + "num_tokens": 270426057.0, + "step": 12651 + }, + { + "epoch": 1.0432918281520573, + "grad_norm": 2.044506072998047, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.7565667629241943, + "num_tokens": 270455226.0, + "step": 12652 + }, + { + "epoch": 1.043374288777109, + "grad_norm": 1.8765394687652588, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7543878555297852, + "num_tokens": 270486601.0, + "step": 12653 + }, + { + "epoch": 1.0434567494021605, + "grad_norm": 1.8193660974502563, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.7802468538284302, + "num_tokens": 270522379.0, + "step": 12654 + }, + { + "epoch": 1.043539210027212, + "grad_norm": 2.46081280708313, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7449657917022705, + "num_tokens": 270545108.0, + "step": 12655 + }, + { + "epoch": 1.0436216706522636, + "grad_norm": 1.792718529701233, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7650740742683411, + "num_tokens": 270581994.0, + "step": 12656 + }, + { + "epoch": 1.043704131277315, + "grad_norm": 1.769608497619629, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7318268418312073, + "num_tokens": 270619671.0, + "step": 12657 + }, + { + "epoch": 1.0437865919023666, + "grad_norm": 2.317125082015991, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7593515515327454, + "num_tokens": 270645476.0, + "step": 12658 + }, + { + "epoch": 1.0438690525274181, + "grad_norm": 2.0493438243865967, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.7940765023231506, + "num_tokens": 270672343.0, + "step": 12659 + }, + { + "epoch": 1.0439515131524697, + "grad_norm": 1.9770244359970093, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.7909742593765259, + "num_tokens": 270700798.0, + "step": 12660 + }, + { + "epoch": 1.0440339737775213, + "grad_norm": 2.4669389724731445, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7547147274017334, + "num_tokens": 270732576.0, + "step": 12661 + }, + { + "epoch": 1.0441164344025728, + "grad_norm": 2.2454802989959717, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7193921208381653, + "num_tokens": 270759072.0, + "step": 12662 + }, + { + "epoch": 1.0441988950276242, + "grad_norm": 1.926713228225708, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.777793288230896, + "num_tokens": 270792886.0, + "step": 12663 + }, + { + "epoch": 1.0442813556526758, + "grad_norm": 1.918574333190918, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7732682824134827, + "num_tokens": 270822115.0, + "step": 12664 + }, + { + "epoch": 1.0443638162777273, + "grad_norm": 2.161494493484497, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7220825552940369, + "num_tokens": 270848136.0, + "step": 12665 + }, + { + "epoch": 1.044446276902779, + "grad_norm": 1.6901330947875977, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7807130813598633, + "num_tokens": 270886992.0, + "step": 12666 + }, + { + "epoch": 1.0445287375278305, + "grad_norm": 1.9659451246261597, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.7921934127807617, + "num_tokens": 270916070.0, + "step": 12667 + }, + { + "epoch": 1.044611198152882, + "grad_norm": 1.9090588092803955, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.7849389314651489, + "num_tokens": 270948453.0, + "step": 12668 + }, + { + "epoch": 1.0446936587779336, + "grad_norm": 2.0370771884918213, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.792648196220398, + "num_tokens": 270979917.0, + "step": 12669 + }, + { + "epoch": 1.044776119402985, + "grad_norm": 2.246123790740967, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7913883924484253, + "num_tokens": 271004426.0, + "step": 12670 + }, + { + "epoch": 1.0448585800280366, + "grad_norm": 1.7915098667144775, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7766710519790649, + "num_tokens": 271039268.0, + "step": 12671 + }, + { + "epoch": 1.0449410406530881, + "grad_norm": 1.8357590436935425, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.743470311164856, + "num_tokens": 271074913.0, + "step": 12672 + }, + { + "epoch": 1.0450235012781397, + "grad_norm": 1.8204565048217773, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.738759458065033, + "num_tokens": 271107910.0, + "step": 12673 + }, + { + "epoch": 1.0451059619031913, + "grad_norm": 1.8580037355422974, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7907181978225708, + "num_tokens": 271142124.0, + "step": 12674 + }, + { + "epoch": 1.0451884225282428, + "grad_norm": 2.266674757003784, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7166808843612671, + "num_tokens": 271165477.0, + "step": 12675 + }, + { + "epoch": 1.0452708831532942, + "grad_norm": 2.085552215576172, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7703878283500671, + "num_tokens": 271193422.0, + "step": 12676 + }, + { + "epoch": 1.0453533437783458, + "grad_norm": 1.9445143938064575, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.7953796982765198, + "num_tokens": 271221213.0, + "step": 12677 + }, + { + "epoch": 1.0454358044033973, + "grad_norm": 2.2002031803131104, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7761743068695068, + "num_tokens": 271248310.0, + "step": 12678 + }, + { + "epoch": 1.045518265028449, + "grad_norm": 1.7854653596878052, + "learning_rate": 1e-06, + "loss": 0.584, + "mean_token_accuracy": 0.8104144334793091, + "num_tokens": 271281933.0, + "step": 12679 + }, + { + "epoch": 1.0456007256535005, + "grad_norm": 1.935498833656311, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.7824228405952454, + "num_tokens": 271311753.0, + "step": 12680 + }, + { + "epoch": 1.045683186278552, + "grad_norm": 2.1291613578796387, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7693890929222107, + "num_tokens": 271338895.0, + "step": 12681 + }, + { + "epoch": 1.0457656469036036, + "grad_norm": 2.095794677734375, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7809257507324219, + "num_tokens": 271365928.0, + "step": 12682 + }, + { + "epoch": 1.045848107528655, + "grad_norm": 1.8757891654968262, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7761662602424622, + "num_tokens": 271398887.0, + "step": 12683 + }, + { + "epoch": 1.0459305681537066, + "grad_norm": 1.9094253778457642, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.8010505437850952, + "num_tokens": 271429450.0, + "step": 12684 + }, + { + "epoch": 1.0460130287787581, + "grad_norm": 2.0487465858459473, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7266758680343628, + "num_tokens": 271458570.0, + "step": 12685 + }, + { + "epoch": 1.0460954894038097, + "grad_norm": 2.1625936031341553, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7455418705940247, + "num_tokens": 271486113.0, + "step": 12686 + }, + { + "epoch": 1.0461779500288613, + "grad_norm": 1.9692078828811646, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7524786591529846, + "num_tokens": 271515534.0, + "step": 12687 + }, + { + "epoch": 1.0462604106539128, + "grad_norm": 2.0531833171844482, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7484794855117798, + "num_tokens": 271546764.0, + "step": 12688 + }, + { + "epoch": 1.0463428712789642, + "grad_norm": 2.142033576965332, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7711173295974731, + "num_tokens": 271575276.0, + "step": 12689 + }, + { + "epoch": 1.0464253319040158, + "grad_norm": 2.0505928993225098, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.7920001745223999, + "num_tokens": 271603756.0, + "step": 12690 + }, + { + "epoch": 1.0465077925290673, + "grad_norm": 2.08923077583313, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.7495073080062866, + "num_tokens": 271631378.0, + "step": 12691 + }, + { + "epoch": 1.046590253154119, + "grad_norm": 1.7662051916122437, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.8047957420349121, + "num_tokens": 271666640.0, + "step": 12692 + }, + { + "epoch": 1.0466727137791705, + "grad_norm": 1.669041395187378, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.7820671796798706, + "num_tokens": 271703237.0, + "step": 12693 + }, + { + "epoch": 1.046755174404222, + "grad_norm": 1.7117846012115479, + "learning_rate": 1e-06, + "loss": 0.6093, + "mean_token_accuracy": 0.8092275857925415, + "num_tokens": 271740212.0, + "step": 12694 + }, + { + "epoch": 1.0468376350292736, + "grad_norm": 1.8244738578796387, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7684352397918701, + "num_tokens": 271773227.0, + "step": 12695 + }, + { + "epoch": 1.046920095654325, + "grad_norm": 1.7048285007476807, + "learning_rate": 1e-06, + "loss": 0.5972, + "mean_token_accuracy": 0.8040953874588013, + "num_tokens": 271810226.0, + "step": 12696 + }, + { + "epoch": 1.0470025562793766, + "grad_norm": 2.239109754562378, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7344871163368225, + "num_tokens": 271836933.0, + "step": 12697 + }, + { + "epoch": 1.0470850169044281, + "grad_norm": 1.818928837776184, + "learning_rate": 1e-06, + "loss": 0.4925, + "mean_token_accuracy": 0.8284643888473511, + "num_tokens": 271866995.0, + "step": 12698 + }, + { + "epoch": 1.0471674775294797, + "grad_norm": 1.9022806882858276, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7495080232620239, + "num_tokens": 271899193.0, + "step": 12699 + }, + { + "epoch": 1.0472499381545313, + "grad_norm": 1.7739439010620117, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.7981947660446167, + "num_tokens": 271933951.0, + "step": 12700 + }, + { + "epoch": 1.0473323987795828, + "grad_norm": 2.070868492126465, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7535978555679321, + "num_tokens": 271963874.0, + "step": 12701 + }, + { + "epoch": 1.0474148594046342, + "grad_norm": 1.9646350145339966, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7290859222412109, + "num_tokens": 271995652.0, + "step": 12702 + }, + { + "epoch": 1.0474973200296858, + "grad_norm": 1.7023484706878662, + "learning_rate": 1e-06, + "loss": 0.5933, + "mean_token_accuracy": 0.807402491569519, + "num_tokens": 272031820.0, + "step": 12703 + }, + { + "epoch": 1.0475797806547373, + "grad_norm": 1.9948962926864624, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7334446310997009, + "num_tokens": 272060712.0, + "step": 12704 + }, + { + "epoch": 1.047662241279789, + "grad_norm": 2.2348546981811523, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7138172388076782, + "num_tokens": 272088951.0, + "step": 12705 + }, + { + "epoch": 1.0477447019048405, + "grad_norm": 2.07487154006958, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7392358183860779, + "num_tokens": 272117123.0, + "step": 12706 + }, + { + "epoch": 1.047827162529892, + "grad_norm": 2.112927198410034, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7711619138717651, + "num_tokens": 272142597.0, + "step": 12707 + }, + { + "epoch": 1.0479096231549434, + "grad_norm": 2.1110193729400635, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7697399854660034, + "num_tokens": 272169975.0, + "step": 12708 + }, + { + "epoch": 1.047992083779995, + "grad_norm": 1.9010337591171265, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.7894269227981567, + "num_tokens": 272199829.0, + "step": 12709 + }, + { + "epoch": 1.0480745444050465, + "grad_norm": 1.8832828998565674, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7637659311294556, + "num_tokens": 272231662.0, + "step": 12710 + }, + { + "epoch": 1.0481570050300981, + "grad_norm": 2.1147782802581787, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7164773941040039, + "num_tokens": 272262896.0, + "step": 12711 + }, + { + "epoch": 1.0482394656551497, + "grad_norm": 1.9746179580688477, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.735882043838501, + "num_tokens": 272296803.0, + "step": 12712 + }, + { + "epoch": 1.0483219262802013, + "grad_norm": 2.152587652206421, + "learning_rate": 1e-06, + "loss": 0.5757, + "mean_token_accuracy": 0.807399332523346, + "num_tokens": 272325448.0, + "step": 12713 + }, + { + "epoch": 1.0484043869052528, + "grad_norm": 1.7649725675582886, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.7815342545509338, + "num_tokens": 272359750.0, + "step": 12714 + }, + { + "epoch": 1.0484868475303042, + "grad_norm": 2.0041041374206543, + "learning_rate": 1e-06, + "loss": 0.7673, + "mean_token_accuracy": 0.7629091739654541, + "num_tokens": 272386437.0, + "step": 12715 + }, + { + "epoch": 1.0485693081553558, + "grad_norm": 1.9257878065109253, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.7813582420349121, + "num_tokens": 272417583.0, + "step": 12716 + }, + { + "epoch": 1.0486517687804073, + "grad_norm": 1.9840716123580933, + "learning_rate": 1e-06, + "loss": 0.553, + "mean_token_accuracy": 0.817076563835144, + "num_tokens": 272447677.0, + "step": 12717 + }, + { + "epoch": 1.048734229405459, + "grad_norm": 1.9560461044311523, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7549611926078796, + "num_tokens": 272481061.0, + "step": 12718 + }, + { + "epoch": 1.0488166900305105, + "grad_norm": 1.7164736986160278, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7539044618606567, + "num_tokens": 272521689.0, + "step": 12719 + }, + { + "epoch": 1.048899150655562, + "grad_norm": 1.7787911891937256, + "learning_rate": 1e-06, + "loss": 0.6014, + "mean_token_accuracy": 0.7942434549331665, + "num_tokens": 272554885.0, + "step": 12720 + }, + { + "epoch": 1.0489816112806134, + "grad_norm": 2.034055471420288, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7115069627761841, + "num_tokens": 272585306.0, + "step": 12721 + }, + { + "epoch": 1.049064071905665, + "grad_norm": 1.907213568687439, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7309660911560059, + "num_tokens": 272622988.0, + "step": 12722 + }, + { + "epoch": 1.0491465325307165, + "grad_norm": 2.028853416442871, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7751814126968384, + "num_tokens": 272651623.0, + "step": 12723 + }, + { + "epoch": 1.0492289931557681, + "grad_norm": 1.6459916830062866, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.7941515445709229, + "num_tokens": 272688500.0, + "step": 12724 + }, + { + "epoch": 1.0493114537808197, + "grad_norm": 2.2087435722351074, + "learning_rate": 1e-06, + "loss": 0.7673, + "mean_token_accuracy": 0.7509167790412903, + "num_tokens": 272712883.0, + "step": 12725 + }, + { + "epoch": 1.0493939144058713, + "grad_norm": 1.8228771686553955, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.7799884080886841, + "num_tokens": 272745513.0, + "step": 12726 + }, + { + "epoch": 1.0494763750309228, + "grad_norm": 2.1379382610321045, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7502174377441406, + "num_tokens": 272774215.0, + "step": 12727 + }, + { + "epoch": 1.0495588356559742, + "grad_norm": 2.280334711074829, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7759146690368652, + "num_tokens": 272797311.0, + "step": 12728 + }, + { + "epoch": 1.0496412962810258, + "grad_norm": 1.9257168769836426, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.776294469833374, + "num_tokens": 272826571.0, + "step": 12729 + }, + { + "epoch": 1.0497237569060773, + "grad_norm": 1.7265037298202515, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.770472526550293, + "num_tokens": 272865879.0, + "step": 12730 + }, + { + "epoch": 1.049806217531129, + "grad_norm": 1.938833236694336, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7561689615249634, + "num_tokens": 272898298.0, + "step": 12731 + }, + { + "epoch": 1.0498886781561805, + "grad_norm": 2.214648485183716, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7148037552833557, + "num_tokens": 272926318.0, + "step": 12732 + }, + { + "epoch": 1.049971138781232, + "grad_norm": 1.852007508277893, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7656503319740295, + "num_tokens": 272960154.0, + "step": 12733 + }, + { + "epoch": 1.0500535994062834, + "grad_norm": 1.9270458221435547, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7579610347747803, + "num_tokens": 272993699.0, + "step": 12734 + }, + { + "epoch": 1.050136060031335, + "grad_norm": 2.0697572231292725, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.740950345993042, + "num_tokens": 273022146.0, + "step": 12735 + }, + { + "epoch": 1.0502185206563865, + "grad_norm": 2.28251051902771, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.755890965461731, + "num_tokens": 273046351.0, + "step": 12736 + }, + { + "epoch": 1.0503009812814381, + "grad_norm": 2.0084023475646973, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7510988712310791, + "num_tokens": 273077336.0, + "step": 12737 + }, + { + "epoch": 1.0503834419064897, + "grad_norm": 2.133951187133789, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7676249146461487, + "num_tokens": 273102307.0, + "step": 12738 + }, + { + "epoch": 1.0504659025315413, + "grad_norm": 1.9047019481658936, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7531371116638184, + "num_tokens": 273132899.0, + "step": 12739 + }, + { + "epoch": 1.0505483631565928, + "grad_norm": 2.2474255561828613, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7545918822288513, + "num_tokens": 273158032.0, + "step": 12740 + }, + { + "epoch": 1.0506308237816442, + "grad_norm": 2.1200056076049805, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7644875049591064, + "num_tokens": 273184759.0, + "step": 12741 + }, + { + "epoch": 1.0507132844066958, + "grad_norm": 2.1026272773742676, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7727937698364258, + "num_tokens": 273209249.0, + "step": 12742 + }, + { + "epoch": 1.0507957450317473, + "grad_norm": 1.9421783685684204, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7673254013061523, + "num_tokens": 273239684.0, + "step": 12743 + }, + { + "epoch": 1.050878205656799, + "grad_norm": 1.9620647430419922, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7784706354141235, + "num_tokens": 273269818.0, + "step": 12744 + }, + { + "epoch": 1.0509606662818505, + "grad_norm": 1.9153176546096802, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7321003675460815, + "num_tokens": 273303726.0, + "step": 12745 + }, + { + "epoch": 1.051043126906902, + "grad_norm": 1.7140140533447266, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7639005184173584, + "num_tokens": 273343564.0, + "step": 12746 + }, + { + "epoch": 1.0511255875319534, + "grad_norm": 1.9091600179672241, + "learning_rate": 1e-06, + "loss": 0.7543, + "mean_token_accuracy": 0.7538810968399048, + "num_tokens": 273379176.0, + "step": 12747 + }, + { + "epoch": 1.051208048157005, + "grad_norm": 2.175715446472168, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.7958986163139343, + "num_tokens": 273405038.0, + "step": 12748 + }, + { + "epoch": 1.0512905087820565, + "grad_norm": 2.075298547744751, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.7879986763000488, + "num_tokens": 273431563.0, + "step": 12749 + }, + { + "epoch": 1.0513729694071081, + "grad_norm": 1.8351476192474365, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7656989097595215, + "num_tokens": 273464801.0, + "step": 12750 + }, + { + "epoch": 1.0514554300321597, + "grad_norm": 2.0268099308013916, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7456928491592407, + "num_tokens": 273494313.0, + "step": 12751 + }, + { + "epoch": 1.0515378906572113, + "grad_norm": 1.8346965312957764, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7780134081840515, + "num_tokens": 273528200.0, + "step": 12752 + }, + { + "epoch": 1.0516203512822626, + "grad_norm": 2.1930551528930664, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7319464087486267, + "num_tokens": 273554105.0, + "step": 12753 + }, + { + "epoch": 1.0517028119073142, + "grad_norm": 2.2540688514709473, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7836173176765442, + "num_tokens": 273578564.0, + "step": 12754 + }, + { + "epoch": 1.0517852725323658, + "grad_norm": 1.9507968425750732, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.7825522422790527, + "num_tokens": 273607989.0, + "step": 12755 + }, + { + "epoch": 1.0518677331574173, + "grad_norm": 1.9792120456695557, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7382116913795471, + "num_tokens": 273637612.0, + "step": 12756 + }, + { + "epoch": 1.051950193782469, + "grad_norm": 1.9974545240402222, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7254718542098999, + "num_tokens": 273667197.0, + "step": 12757 + }, + { + "epoch": 1.0520326544075205, + "grad_norm": 1.9712271690368652, + "learning_rate": 1e-06, + "loss": 0.5718, + "mean_token_accuracy": 0.807483434677124, + "num_tokens": 273697053.0, + "step": 12758 + }, + { + "epoch": 1.052115115032572, + "grad_norm": 2.022030830383301, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7681288719177246, + "num_tokens": 273729242.0, + "step": 12759 + }, + { + "epoch": 1.0521975756576234, + "grad_norm": 1.984807014465332, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7398568391799927, + "num_tokens": 273760452.0, + "step": 12760 + }, + { + "epoch": 1.052280036282675, + "grad_norm": 2.0452558994293213, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7607488632202148, + "num_tokens": 273791091.0, + "step": 12761 + }, + { + "epoch": 1.0523624969077265, + "grad_norm": 1.9881826639175415, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7602324485778809, + "num_tokens": 273822785.0, + "step": 12762 + }, + { + "epoch": 1.0524449575327781, + "grad_norm": 1.9838253259658813, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7815834283828735, + "num_tokens": 273849781.0, + "step": 12763 + }, + { + "epoch": 1.0525274181578297, + "grad_norm": 1.7146950960159302, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.7955203056335449, + "num_tokens": 273885491.0, + "step": 12764 + }, + { + "epoch": 1.0526098787828813, + "grad_norm": 2.2830283641815186, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.749636709690094, + "num_tokens": 273909201.0, + "step": 12765 + }, + { + "epoch": 1.0526923394079326, + "grad_norm": 2.0972235202789307, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7759872674942017, + "num_tokens": 273936049.0, + "step": 12766 + }, + { + "epoch": 1.0527748000329842, + "grad_norm": 2.124561309814453, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7745903730392456, + "num_tokens": 273964215.0, + "step": 12767 + }, + { + "epoch": 1.0528572606580358, + "grad_norm": 2.2291042804718018, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7393972873687744, + "num_tokens": 273989234.0, + "step": 12768 + }, + { + "epoch": 1.0529397212830873, + "grad_norm": 2.0267670154571533, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7572550773620605, + "num_tokens": 274017754.0, + "step": 12769 + }, + { + "epoch": 1.053022181908139, + "grad_norm": 2.2301790714263916, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7349190711975098, + "num_tokens": 274044561.0, + "step": 12770 + }, + { + "epoch": 1.0531046425331905, + "grad_norm": 1.979494333267212, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7657420635223389, + "num_tokens": 274078034.0, + "step": 12771 + }, + { + "epoch": 1.053187103158242, + "grad_norm": 1.9667075872421265, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7621918320655823, + "num_tokens": 274108972.0, + "step": 12772 + }, + { + "epoch": 1.0532695637832934, + "grad_norm": 1.9417980909347534, + "learning_rate": 1e-06, + "loss": 0.5715, + "mean_token_accuracy": 0.8092290163040161, + "num_tokens": 274136692.0, + "step": 12773 + }, + { + "epoch": 1.053352024408345, + "grad_norm": 2.023017168045044, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7409307956695557, + "num_tokens": 274167485.0, + "step": 12774 + }, + { + "epoch": 1.0534344850333965, + "grad_norm": 2.0144412517547607, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7472609877586365, + "num_tokens": 274194593.0, + "step": 12775 + }, + { + "epoch": 1.0535169456584481, + "grad_norm": 1.8299530744552612, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7468517422676086, + "num_tokens": 274229431.0, + "step": 12776 + }, + { + "epoch": 1.0535994062834997, + "grad_norm": 1.831778883934021, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.750205397605896, + "num_tokens": 274263525.0, + "step": 12777 + }, + { + "epoch": 1.0536818669085513, + "grad_norm": 2.0892627239227295, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7242708206176758, + "num_tokens": 274292889.0, + "step": 12778 + }, + { + "epoch": 1.0537643275336026, + "grad_norm": 1.90255606174469, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7483336925506592, + "num_tokens": 274327047.0, + "step": 12779 + }, + { + "epoch": 1.0538467881586542, + "grad_norm": 1.9100309610366821, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7489776611328125, + "num_tokens": 274359951.0, + "step": 12780 + }, + { + "epoch": 1.0539292487837058, + "grad_norm": 1.8453749418258667, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.7934634685516357, + "num_tokens": 274389420.0, + "step": 12781 + }, + { + "epoch": 1.0540117094087573, + "grad_norm": 1.8638720512390137, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7558153867721558, + "num_tokens": 274424087.0, + "step": 12782 + }, + { + "epoch": 1.054094170033809, + "grad_norm": 1.9970569610595703, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.744784951210022, + "num_tokens": 274456112.0, + "step": 12783 + }, + { + "epoch": 1.0541766306588605, + "grad_norm": 1.9075454473495483, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7547680735588074, + "num_tokens": 274485741.0, + "step": 12784 + }, + { + "epoch": 1.054259091283912, + "grad_norm": 2.276050329208374, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7379456162452698, + "num_tokens": 274510559.0, + "step": 12785 + }, + { + "epoch": 1.0543415519089634, + "grad_norm": 1.9716473817825317, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7584383487701416, + "num_tokens": 274540481.0, + "step": 12786 + }, + { + "epoch": 1.054424012534015, + "grad_norm": 1.9744411706924438, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7684575319290161, + "num_tokens": 274569396.0, + "step": 12787 + }, + { + "epoch": 1.0545064731590665, + "grad_norm": 2.08638596534729, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7481567859649658, + "num_tokens": 274597054.0, + "step": 12788 + }, + { + "epoch": 1.054588933784118, + "grad_norm": 1.955287218093872, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7633728981018066, + "num_tokens": 274628080.0, + "step": 12789 + }, + { + "epoch": 1.0546713944091697, + "grad_norm": 2.0459396839141846, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7594721913337708, + "num_tokens": 274656435.0, + "step": 12790 + }, + { + "epoch": 1.0547538550342213, + "grad_norm": 2.084719657897949, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7681687474250793, + "num_tokens": 274683782.0, + "step": 12791 + }, + { + "epoch": 1.0548363156592726, + "grad_norm": 1.875161051750183, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.7568609714508057, + "num_tokens": 274717659.0, + "step": 12792 + }, + { + "epoch": 1.0549187762843242, + "grad_norm": 2.102680206298828, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7499686479568481, + "num_tokens": 274745293.0, + "step": 12793 + }, + { + "epoch": 1.0550012369093758, + "grad_norm": 1.810132384300232, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7888320088386536, + "num_tokens": 274783529.0, + "step": 12794 + }, + { + "epoch": 1.0550836975344273, + "grad_norm": 2.1007091999053955, + "learning_rate": 1e-06, + "loss": 0.7722, + "mean_token_accuracy": 0.7534120082855225, + "num_tokens": 274812560.0, + "step": 12795 + }, + { + "epoch": 1.055166158159479, + "grad_norm": 1.9772135019302368, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7853736877441406, + "num_tokens": 274842325.0, + "step": 12796 + }, + { + "epoch": 1.0552486187845305, + "grad_norm": 2.1398630142211914, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7661244869232178, + "num_tokens": 274869608.0, + "step": 12797 + }, + { + "epoch": 1.0553310794095818, + "grad_norm": 1.9417283535003662, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7308074235916138, + "num_tokens": 274902063.0, + "step": 12798 + }, + { + "epoch": 1.0554135400346334, + "grad_norm": 1.9654760360717773, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.7810056805610657, + "num_tokens": 274932244.0, + "step": 12799 + }, + { + "epoch": 1.055496000659685, + "grad_norm": 2.0859179496765137, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7356880903244019, + "num_tokens": 274962633.0, + "step": 12800 + }, + { + "epoch": 1.0555784612847365, + "grad_norm": 2.12168550491333, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7548974752426147, + "num_tokens": 274990589.0, + "step": 12801 + }, + { + "epoch": 1.055660921909788, + "grad_norm": 1.9728949069976807, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.7957383990287781, + "num_tokens": 275017899.0, + "step": 12802 + }, + { + "epoch": 1.0557433825348397, + "grad_norm": 1.8959866762161255, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7593244314193726, + "num_tokens": 275051034.0, + "step": 12803 + }, + { + "epoch": 1.0558258431598913, + "grad_norm": 2.0399763584136963, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7233635783195496, + "num_tokens": 275083614.0, + "step": 12804 + }, + { + "epoch": 1.0559083037849426, + "grad_norm": 2.0657031536102295, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.722864031791687, + "num_tokens": 275112046.0, + "step": 12805 + }, + { + "epoch": 1.0559907644099942, + "grad_norm": 1.9617342948913574, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7831094264984131, + "num_tokens": 275145645.0, + "step": 12806 + }, + { + "epoch": 1.0560732250350457, + "grad_norm": 2.1731717586517334, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7487596273422241, + "num_tokens": 275173291.0, + "step": 12807 + }, + { + "epoch": 1.0561556856600973, + "grad_norm": 1.877347707748413, + "learning_rate": 1e-06, + "loss": 0.5151, + "mean_token_accuracy": 0.8226510882377625, + "num_tokens": 275204185.0, + "step": 12808 + }, + { + "epoch": 1.056238146285149, + "grad_norm": 1.9257912635803223, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7895935773849487, + "num_tokens": 275234062.0, + "step": 12809 + }, + { + "epoch": 1.0563206069102005, + "grad_norm": 1.7338470220565796, + "learning_rate": 1e-06, + "loss": 0.6112, + "mean_token_accuracy": 0.7936362028121948, + "num_tokens": 275267424.0, + "step": 12810 + }, + { + "epoch": 1.0564030675352518, + "grad_norm": 1.8124555349349976, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7702370882034302, + "num_tokens": 275302726.0, + "step": 12811 + }, + { + "epoch": 1.0564855281603034, + "grad_norm": 1.813324213027954, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7517274022102356, + "num_tokens": 275336083.0, + "step": 12812 + }, + { + "epoch": 1.056567988785355, + "grad_norm": 1.666707992553711, + "learning_rate": 1e-06, + "loss": 0.5826, + "mean_token_accuracy": 0.8001694083213806, + "num_tokens": 275373354.0, + "step": 12813 + }, + { + "epoch": 1.0566504494104065, + "grad_norm": 1.9960167407989502, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7735707759857178, + "num_tokens": 275406804.0, + "step": 12814 + }, + { + "epoch": 1.056732910035458, + "grad_norm": 2.1008410453796387, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.796671450138092, + "num_tokens": 275433992.0, + "step": 12815 + }, + { + "epoch": 1.0568153706605097, + "grad_norm": 1.9008092880249023, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7750453352928162, + "num_tokens": 275466283.0, + "step": 12816 + }, + { + "epoch": 1.0568978312855613, + "grad_norm": 2.061997652053833, + "learning_rate": 1e-06, + "loss": 0.7712, + "mean_token_accuracy": 0.756205141544342, + "num_tokens": 275493237.0, + "step": 12817 + }, + { + "epoch": 1.0569802919106126, + "grad_norm": 1.9354948997497559, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7549700736999512, + "num_tokens": 275520792.0, + "step": 12818 + }, + { + "epoch": 1.0570627525356642, + "grad_norm": 1.6592158079147339, + "learning_rate": 1e-06, + "loss": 0.5058, + "mean_token_accuracy": 0.8323872089385986, + "num_tokens": 275557927.0, + "step": 12819 + }, + { + "epoch": 1.0571452131607157, + "grad_norm": 1.6920775175094604, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7720451354980469, + "num_tokens": 275597928.0, + "step": 12820 + }, + { + "epoch": 1.0572276737857673, + "grad_norm": 2.131124258041382, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7224478721618652, + "num_tokens": 275628348.0, + "step": 12821 + }, + { + "epoch": 1.057310134410819, + "grad_norm": 1.865451693534851, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7700827121734619, + "num_tokens": 275662139.0, + "step": 12822 + }, + { + "epoch": 1.0573925950358705, + "grad_norm": 2.0035910606384277, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7316902875900269, + "num_tokens": 275691520.0, + "step": 12823 + }, + { + "epoch": 1.0574750556609218, + "grad_norm": 2.117185354232788, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7251913547515869, + "num_tokens": 275720556.0, + "step": 12824 + }, + { + "epoch": 1.0575575162859734, + "grad_norm": 2.110997438430786, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7692278623580933, + "num_tokens": 275746682.0, + "step": 12825 + }, + { + "epoch": 1.057639976911025, + "grad_norm": 2.1010308265686035, + "learning_rate": 1e-06, + "loss": 0.5616, + "mean_token_accuracy": 0.8161147236824036, + "num_tokens": 275770517.0, + "step": 12826 + }, + { + "epoch": 1.0577224375360765, + "grad_norm": 2.0174453258514404, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7210928201675415, + "num_tokens": 275799453.0, + "step": 12827 + }, + { + "epoch": 1.057804898161128, + "grad_norm": 1.8868571519851685, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7617995738983154, + "num_tokens": 275833718.0, + "step": 12828 + }, + { + "epoch": 1.0578873587861797, + "grad_norm": 2.0768768787384033, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7179503440856934, + "num_tokens": 275862711.0, + "step": 12829 + }, + { + "epoch": 1.0579698194112312, + "grad_norm": 2.10795259475708, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.744775116443634, + "num_tokens": 275888269.0, + "step": 12830 + }, + { + "epoch": 1.0580522800362826, + "grad_norm": 2.02081036567688, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7626432180404663, + "num_tokens": 275918843.0, + "step": 12831 + }, + { + "epoch": 1.0581347406613342, + "grad_norm": 2.0331079959869385, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7521518468856812, + "num_tokens": 275946753.0, + "step": 12832 + }, + { + "epoch": 1.0582172012863857, + "grad_norm": 2.1971373558044434, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7565442323684692, + "num_tokens": 275973154.0, + "step": 12833 + }, + { + "epoch": 1.0582996619114373, + "grad_norm": 1.9276831150054932, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7760705947875977, + "num_tokens": 276005843.0, + "step": 12834 + }, + { + "epoch": 1.058382122536489, + "grad_norm": 2.4617538452148438, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7345964908599854, + "num_tokens": 276029439.0, + "step": 12835 + }, + { + "epoch": 1.0584645831615405, + "grad_norm": 2.1770002841949463, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7757283449172974, + "num_tokens": 276053584.0, + "step": 12836 + }, + { + "epoch": 1.0585470437865918, + "grad_norm": 2.062283754348755, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7592905759811401, + "num_tokens": 276084918.0, + "step": 12837 + }, + { + "epoch": 1.0586295044116434, + "grad_norm": 1.8826847076416016, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7683438658714294, + "num_tokens": 276116440.0, + "step": 12838 + }, + { + "epoch": 1.058711965036695, + "grad_norm": 2.2372963428497314, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7509055137634277, + "num_tokens": 276142580.0, + "step": 12839 + }, + { + "epoch": 1.0587944256617465, + "grad_norm": 1.9834173917770386, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7797070741653442, + "num_tokens": 276171966.0, + "step": 12840 + }, + { + "epoch": 1.058876886286798, + "grad_norm": 2.1797358989715576, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7458031177520752, + "num_tokens": 276198082.0, + "step": 12841 + }, + { + "epoch": 1.0589593469118497, + "grad_norm": 2.025249719619751, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.726769745349884, + "num_tokens": 276227093.0, + "step": 12842 + }, + { + "epoch": 1.059041807536901, + "grad_norm": 1.8552274703979492, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7569290399551392, + "num_tokens": 276262057.0, + "step": 12843 + }, + { + "epoch": 1.0591242681619526, + "grad_norm": 1.8990695476531982, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7772991061210632, + "num_tokens": 276293005.0, + "step": 12844 + }, + { + "epoch": 1.0592067287870042, + "grad_norm": 2.0682740211486816, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.764737606048584, + "num_tokens": 276321011.0, + "step": 12845 + }, + { + "epoch": 1.0592891894120557, + "grad_norm": 2.1388609409332275, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7608190774917603, + "num_tokens": 276346181.0, + "step": 12846 + }, + { + "epoch": 1.0593716500371073, + "grad_norm": 1.9172617197036743, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.8021707534790039, + "num_tokens": 276376093.0, + "step": 12847 + }, + { + "epoch": 1.0594541106621589, + "grad_norm": 2.0933868885040283, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7470657825469971, + "num_tokens": 276404000.0, + "step": 12848 + }, + { + "epoch": 1.0595365712872105, + "grad_norm": 1.982985496520996, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.790400505065918, + "num_tokens": 276432350.0, + "step": 12849 + }, + { + "epoch": 1.0596190319122618, + "grad_norm": 2.344822406768799, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.7928269505500793, + "num_tokens": 276455150.0, + "step": 12850 + }, + { + "epoch": 1.0597014925373134, + "grad_norm": 2.0886571407318115, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7587396502494812, + "num_tokens": 276483520.0, + "step": 12851 + }, + { + "epoch": 1.059783953162365, + "grad_norm": 1.9680817127227783, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7641750574111938, + "num_tokens": 276514531.0, + "step": 12852 + }, + { + "epoch": 1.0598664137874165, + "grad_norm": 1.8665176630020142, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7772237062454224, + "num_tokens": 276548667.0, + "step": 12853 + }, + { + "epoch": 1.059948874412468, + "grad_norm": 1.6941642761230469, + "learning_rate": 1e-06, + "loss": 0.5942, + "mean_token_accuracy": 0.800485372543335, + "num_tokens": 276583485.0, + "step": 12854 + }, + { + "epoch": 1.0600313350375197, + "grad_norm": 1.780929446220398, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7790271639823914, + "num_tokens": 276621011.0, + "step": 12855 + }, + { + "epoch": 1.060113795662571, + "grad_norm": 2.013274669647217, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.8031491637229919, + "num_tokens": 276648039.0, + "step": 12856 + }, + { + "epoch": 1.0601962562876226, + "grad_norm": 1.822461485862732, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7811535596847534, + "num_tokens": 276684313.0, + "step": 12857 + }, + { + "epoch": 1.0602787169126742, + "grad_norm": 2.1512725353240967, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7611714601516724, + "num_tokens": 276711184.0, + "step": 12858 + }, + { + "epoch": 1.0603611775377257, + "grad_norm": 2.3541431427001953, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7608275413513184, + "num_tokens": 276732533.0, + "step": 12859 + }, + { + "epoch": 1.0604436381627773, + "grad_norm": 3.709554433822632, + "learning_rate": 1e-06, + "loss": 0.5335, + "mean_token_accuracy": 0.8111404180526733, + "num_tokens": 276761619.0, + "step": 12860 + }, + { + "epoch": 1.0605260987878289, + "grad_norm": 2.243839740753174, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7459480166435242, + "num_tokens": 276788568.0, + "step": 12861 + }, + { + "epoch": 1.0606085594128802, + "grad_norm": 2.1513185501098633, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7708970904350281, + "num_tokens": 276813042.0, + "step": 12862 + }, + { + "epoch": 1.0606910200379318, + "grad_norm": 2.244908571243286, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7303230166435242, + "num_tokens": 276837302.0, + "step": 12863 + }, + { + "epoch": 1.0607734806629834, + "grad_norm": 1.883514404296875, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7532258033752441, + "num_tokens": 276869101.0, + "step": 12864 + }, + { + "epoch": 1.060855941288035, + "grad_norm": 1.8335566520690918, + "learning_rate": 1e-06, + "loss": 0.5044, + "mean_token_accuracy": 0.8316078782081604, + "num_tokens": 276899721.0, + "step": 12865 + }, + { + "epoch": 1.0609384019130865, + "grad_norm": 1.992107629776001, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.7889286279678345, + "num_tokens": 276926465.0, + "step": 12866 + }, + { + "epoch": 1.061020862538138, + "grad_norm": 1.8329516649246216, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7791691422462463, + "num_tokens": 276957934.0, + "step": 12867 + }, + { + "epoch": 1.0611033231631897, + "grad_norm": 1.8956722021102905, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7567731142044067, + "num_tokens": 276992177.0, + "step": 12868 + }, + { + "epoch": 1.061185783788241, + "grad_norm": 2.0887343883514404, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7647876739501953, + "num_tokens": 277020904.0, + "step": 12869 + }, + { + "epoch": 1.0612682444132926, + "grad_norm": 1.824193000793457, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7555819749832153, + "num_tokens": 277057521.0, + "step": 12870 + }, + { + "epoch": 1.0613507050383442, + "grad_norm": 2.023296594619751, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.762129545211792, + "num_tokens": 277088377.0, + "step": 12871 + }, + { + "epoch": 1.0614331656633957, + "grad_norm": 1.9663769006729126, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.7811156511306763, + "num_tokens": 277121317.0, + "step": 12872 + }, + { + "epoch": 1.0615156262884473, + "grad_norm": 2.2045693397521973, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7277590036392212, + "num_tokens": 277149845.0, + "step": 12873 + }, + { + "epoch": 1.0615980869134989, + "grad_norm": 2.0190227031707764, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.7974256277084351, + "num_tokens": 277176085.0, + "step": 12874 + }, + { + "epoch": 1.0616805475385505, + "grad_norm": 1.8506989479064941, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.749700665473938, + "num_tokens": 277206714.0, + "step": 12875 + }, + { + "epoch": 1.0617630081636018, + "grad_norm": 2.475942850112915, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7402291893959045, + "num_tokens": 277228773.0, + "step": 12876 + }, + { + "epoch": 1.0618454687886534, + "grad_norm": 2.447222948074341, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7665022611618042, + "num_tokens": 277249088.0, + "step": 12877 + }, + { + "epoch": 1.061927929413705, + "grad_norm": 2.1148593425750732, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7463141679763794, + "num_tokens": 277277696.0, + "step": 12878 + }, + { + "epoch": 1.0620103900387565, + "grad_norm": 1.983670711517334, + "learning_rate": 1e-06, + "loss": 0.7585, + "mean_token_accuracy": 0.7534670233726501, + "num_tokens": 277305486.0, + "step": 12879 + }, + { + "epoch": 1.062092850663808, + "grad_norm": 2.0866756439208984, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7440313696861267, + "num_tokens": 277331165.0, + "step": 12880 + }, + { + "epoch": 1.0621753112888597, + "grad_norm": 1.9000592231750488, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.7851812839508057, + "num_tokens": 277365474.0, + "step": 12881 + }, + { + "epoch": 1.062257771913911, + "grad_norm": 1.9298619031906128, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7680591344833374, + "num_tokens": 277400142.0, + "step": 12882 + }, + { + "epoch": 1.0623402325389626, + "grad_norm": 2.0182268619537354, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7770431041717529, + "num_tokens": 277429309.0, + "step": 12883 + }, + { + "epoch": 1.0624226931640142, + "grad_norm": 1.9651259183883667, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7470746636390686, + "num_tokens": 277459339.0, + "step": 12884 + }, + { + "epoch": 1.0625051537890657, + "grad_norm": 1.7224284410476685, + "learning_rate": 1e-06, + "loss": 0.5315, + "mean_token_accuracy": 0.8136461973190308, + "num_tokens": 277494501.0, + "step": 12885 + }, + { + "epoch": 1.0625876144141173, + "grad_norm": 1.9508236646652222, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7740462422370911, + "num_tokens": 277524428.0, + "step": 12886 + }, + { + "epoch": 1.0626700750391689, + "grad_norm": 1.876983404159546, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7596419453620911, + "num_tokens": 277557418.0, + "step": 12887 + }, + { + "epoch": 1.0627525356642202, + "grad_norm": 1.9068045616149902, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.7867363095283508, + "num_tokens": 277586372.0, + "step": 12888 + }, + { + "epoch": 1.0628349962892718, + "grad_norm": 2.166752576828003, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7287341356277466, + "num_tokens": 277614145.0, + "step": 12889 + }, + { + "epoch": 1.0629174569143234, + "grad_norm": 1.962896704673767, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7678982615470886, + "num_tokens": 277645147.0, + "step": 12890 + }, + { + "epoch": 1.062999917539375, + "grad_norm": 1.7596468925476074, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7598869800567627, + "num_tokens": 277680505.0, + "step": 12891 + }, + { + "epoch": 1.0630823781644265, + "grad_norm": 1.9077211618423462, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7447839975357056, + "num_tokens": 277711586.0, + "step": 12892 + }, + { + "epoch": 1.063164838789478, + "grad_norm": 1.963787317276001, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7619742155075073, + "num_tokens": 277739051.0, + "step": 12893 + }, + { + "epoch": 1.0632472994145297, + "grad_norm": 2.0268394947052, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7401882410049438, + "num_tokens": 277768556.0, + "step": 12894 + }, + { + "epoch": 1.063329760039581, + "grad_norm": 1.8322386741638184, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7272093296051025, + "num_tokens": 277803838.0, + "step": 12895 + }, + { + "epoch": 1.0634122206646326, + "grad_norm": 2.0970911979675293, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7269520163536072, + "num_tokens": 277832296.0, + "step": 12896 + }, + { + "epoch": 1.0634946812896842, + "grad_norm": 1.9998615980148315, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7791757583618164, + "num_tokens": 277863568.0, + "step": 12897 + }, + { + "epoch": 1.0635771419147357, + "grad_norm": 1.9594717025756836, + "learning_rate": 1e-06, + "loss": 0.6021, + "mean_token_accuracy": 0.7910431027412415, + "num_tokens": 277894356.0, + "step": 12898 + }, + { + "epoch": 1.0636596025397873, + "grad_norm": 1.8827602863311768, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.7900590300559998, + "num_tokens": 277926523.0, + "step": 12899 + }, + { + "epoch": 1.0637420631648389, + "grad_norm": 2.1023967266082764, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7207080125808716, + "num_tokens": 277958871.0, + "step": 12900 + }, + { + "epoch": 1.0638245237898902, + "grad_norm": 1.8447167873382568, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7673912048339844, + "num_tokens": 277989501.0, + "step": 12901 + }, + { + "epoch": 1.0639069844149418, + "grad_norm": 1.9864436388015747, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.7950109243392944, + "num_tokens": 278019313.0, + "step": 12902 + }, + { + "epoch": 1.0639894450399934, + "grad_norm": 1.6920913457870483, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.787245512008667, + "num_tokens": 278056376.0, + "step": 12903 + }, + { + "epoch": 1.064071905665045, + "grad_norm": 1.905058741569519, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7769390344619751, + "num_tokens": 278088545.0, + "step": 12904 + }, + { + "epoch": 1.0641543662900965, + "grad_norm": 1.8142918348312378, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7880266904830933, + "num_tokens": 278124156.0, + "step": 12905 + }, + { + "epoch": 1.064236826915148, + "grad_norm": 2.0122897624969482, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7648724317550659, + "num_tokens": 278151988.0, + "step": 12906 + }, + { + "epoch": 1.0643192875401994, + "grad_norm": 1.8614088296890259, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7592513561248779, + "num_tokens": 278187086.0, + "step": 12907 + }, + { + "epoch": 1.064401748165251, + "grad_norm": 2.1702675819396973, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.740648627281189, + "num_tokens": 278213093.0, + "step": 12908 + }, + { + "epoch": 1.0644842087903026, + "grad_norm": 2.0448460578918457, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7596776485443115, + "num_tokens": 278242336.0, + "step": 12909 + }, + { + "epoch": 1.0645666694153542, + "grad_norm": 2.1032180786132812, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.771572470664978, + "num_tokens": 278267233.0, + "step": 12910 + }, + { + "epoch": 1.0646491300404057, + "grad_norm": 1.900473713874817, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7475396990776062, + "num_tokens": 278299768.0, + "step": 12911 + }, + { + "epoch": 1.0647315906654573, + "grad_norm": 1.9620625972747803, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7546940445899963, + "num_tokens": 278329041.0, + "step": 12912 + }, + { + "epoch": 1.0648140512905089, + "grad_norm": 1.9131425619125366, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7314392328262329, + "num_tokens": 278362319.0, + "step": 12913 + }, + { + "epoch": 1.0648965119155602, + "grad_norm": 1.884453535079956, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7720802426338196, + "num_tokens": 278393719.0, + "step": 12914 + }, + { + "epoch": 1.0649789725406118, + "grad_norm": 2.094630479812622, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.774077296257019, + "num_tokens": 278421374.0, + "step": 12915 + }, + { + "epoch": 1.0650614331656634, + "grad_norm": 2.1011104583740234, + "learning_rate": 1e-06, + "loss": 0.784, + "mean_token_accuracy": 0.7472591400146484, + "num_tokens": 278447156.0, + "step": 12916 + }, + { + "epoch": 1.065143893790715, + "grad_norm": 2.0250258445739746, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.716394305229187, + "num_tokens": 278477138.0, + "step": 12917 + }, + { + "epoch": 1.0652263544157665, + "grad_norm": 2.1112682819366455, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7834370136260986, + "num_tokens": 278503185.0, + "step": 12918 + }, + { + "epoch": 1.065308815040818, + "grad_norm": 2.1272764205932617, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7596032619476318, + "num_tokens": 278531117.0, + "step": 12919 + }, + { + "epoch": 1.0653912756658697, + "grad_norm": 2.168238878250122, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7688955664634705, + "num_tokens": 278557183.0, + "step": 12920 + }, + { + "epoch": 1.065473736290921, + "grad_norm": 1.8338667154312134, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7869175672531128, + "num_tokens": 278591389.0, + "step": 12921 + }, + { + "epoch": 1.0655561969159726, + "grad_norm": 2.0285849571228027, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7712368965148926, + "num_tokens": 278619733.0, + "step": 12922 + }, + { + "epoch": 1.0656386575410242, + "grad_norm": 2.124544620513916, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7511025667190552, + "num_tokens": 278652176.0, + "step": 12923 + }, + { + "epoch": 1.0657211181660757, + "grad_norm": 1.883987545967102, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.744697630405426, + "num_tokens": 278687963.0, + "step": 12924 + }, + { + "epoch": 1.0658035787911273, + "grad_norm": 2.212989330291748, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.7889034748077393, + "num_tokens": 278710848.0, + "step": 12925 + }, + { + "epoch": 1.0658860394161789, + "grad_norm": 1.8852425813674927, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7382040023803711, + "num_tokens": 278745682.0, + "step": 12926 + }, + { + "epoch": 1.0659685000412302, + "grad_norm": 2.3827784061431885, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7380844354629517, + "num_tokens": 278769109.0, + "step": 12927 + }, + { + "epoch": 1.0660509606662818, + "grad_norm": 1.871459722518921, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7749572992324829, + "num_tokens": 278805886.0, + "step": 12928 + }, + { + "epoch": 1.0661334212913334, + "grad_norm": 2.0146498680114746, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7811217308044434, + "num_tokens": 278836875.0, + "step": 12929 + }, + { + "epoch": 1.066215881916385, + "grad_norm": 2.293445348739624, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7600626349449158, + "num_tokens": 278861855.0, + "step": 12930 + }, + { + "epoch": 1.0662983425414365, + "grad_norm": 1.8932876586914062, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7286407947540283, + "num_tokens": 278893825.0, + "step": 12931 + }, + { + "epoch": 1.066380803166488, + "grad_norm": 2.155852794647217, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.729454755783081, + "num_tokens": 278920608.0, + "step": 12932 + }, + { + "epoch": 1.0664632637915394, + "grad_norm": 2.0176823139190674, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7631052732467651, + "num_tokens": 278947468.0, + "step": 12933 + }, + { + "epoch": 1.066545724416591, + "grad_norm": 1.899438738822937, + "learning_rate": 1e-06, + "loss": 0.5886, + "mean_token_accuracy": 0.8033504486083984, + "num_tokens": 278976534.0, + "step": 12934 + }, + { + "epoch": 1.0666281850416426, + "grad_norm": 1.8663362264633179, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7343944311141968, + "num_tokens": 279014206.0, + "step": 12935 + }, + { + "epoch": 1.0667106456666942, + "grad_norm": 1.8977928161621094, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.7973104119300842, + "num_tokens": 279046284.0, + "step": 12936 + }, + { + "epoch": 1.0667931062917457, + "grad_norm": 2.1952755451202393, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7289649248123169, + "num_tokens": 279072989.0, + "step": 12937 + }, + { + "epoch": 1.0668755669167973, + "grad_norm": 1.8788501024246216, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7656382918357849, + "num_tokens": 279106990.0, + "step": 12938 + }, + { + "epoch": 1.0669580275418489, + "grad_norm": 2.0728724002838135, + "learning_rate": 1e-06, + "loss": 0.5945, + "mean_token_accuracy": 0.8058551549911499, + "num_tokens": 279134638.0, + "step": 12939 + }, + { + "epoch": 1.0670404881669002, + "grad_norm": 2.1063640117645264, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7392463684082031, + "num_tokens": 279166578.0, + "step": 12940 + }, + { + "epoch": 1.0671229487919518, + "grad_norm": 1.9062974452972412, + "learning_rate": 1e-06, + "loss": 0.5447, + "mean_token_accuracy": 0.8159288167953491, + "num_tokens": 279194030.0, + "step": 12941 + }, + { + "epoch": 1.0672054094170034, + "grad_norm": 2.2411258220672607, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7680283784866333, + "num_tokens": 279218107.0, + "step": 12942 + }, + { + "epoch": 1.067287870042055, + "grad_norm": 1.9616529941558838, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7811614274978638, + "num_tokens": 279248331.0, + "step": 12943 + }, + { + "epoch": 1.0673703306671065, + "grad_norm": 2.1744790077209473, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7636289000511169, + "num_tokens": 279273422.0, + "step": 12944 + }, + { + "epoch": 1.067452791292158, + "grad_norm": 2.123399019241333, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7471877932548523, + "num_tokens": 279302358.0, + "step": 12945 + }, + { + "epoch": 1.0675352519172094, + "grad_norm": 2.574272871017456, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7399771809577942, + "num_tokens": 279321448.0, + "step": 12946 + }, + { + "epoch": 1.067617712542261, + "grad_norm": 1.8388972282409668, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7782227993011475, + "num_tokens": 279356222.0, + "step": 12947 + }, + { + "epoch": 1.0677001731673126, + "grad_norm": 2.0858538150787354, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7843118906021118, + "num_tokens": 279381969.0, + "step": 12948 + }, + { + "epoch": 1.0677826337923642, + "grad_norm": 2.292693614959717, + "learning_rate": 1e-06, + "loss": 0.5959, + "mean_token_accuracy": 0.7942777276039124, + "num_tokens": 279406115.0, + "step": 12949 + }, + { + "epoch": 1.0678650944174157, + "grad_norm": 1.8873658180236816, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7519737482070923, + "num_tokens": 279439627.0, + "step": 12950 + }, + { + "epoch": 1.0679475550424673, + "grad_norm": 1.9864037036895752, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7821522355079651, + "num_tokens": 279468602.0, + "step": 12951 + }, + { + "epoch": 1.0680300156675187, + "grad_norm": 2.0264992713928223, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7680402398109436, + "num_tokens": 279497566.0, + "step": 12952 + }, + { + "epoch": 1.0681124762925702, + "grad_norm": 2.0460808277130127, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7344520688056946, + "num_tokens": 279526868.0, + "step": 12953 + }, + { + "epoch": 1.0681949369176218, + "grad_norm": 1.9655157327651978, + "learning_rate": 1e-06, + "loss": 0.5849, + "mean_token_accuracy": 0.8010656833648682, + "num_tokens": 279556716.0, + "step": 12954 + }, + { + "epoch": 1.0682773975426734, + "grad_norm": 2.3523128032684326, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7215372323989868, + "num_tokens": 279582942.0, + "step": 12955 + }, + { + "epoch": 1.068359858167725, + "grad_norm": 2.0040013790130615, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7642004489898682, + "num_tokens": 279611693.0, + "step": 12956 + }, + { + "epoch": 1.0684423187927765, + "grad_norm": 1.7922706604003906, + "learning_rate": 1e-06, + "loss": 0.5462, + "mean_token_accuracy": 0.8181192278862, + "num_tokens": 279641934.0, + "step": 12957 + }, + { + "epoch": 1.068524779417828, + "grad_norm": 2.2887215614318848, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7225815057754517, + "num_tokens": 279668363.0, + "step": 12958 + }, + { + "epoch": 1.0686072400428794, + "grad_norm": 1.9312382936477661, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7686376571655273, + "num_tokens": 279697071.0, + "step": 12959 + }, + { + "epoch": 1.068689700667931, + "grad_norm": 2.0770153999328613, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7429196238517761, + "num_tokens": 279727948.0, + "step": 12960 + }, + { + "epoch": 1.0687721612929826, + "grad_norm": 1.943853735923767, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7328847050666809, + "num_tokens": 279761747.0, + "step": 12961 + }, + { + "epoch": 1.0688546219180342, + "grad_norm": 1.991211175918579, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7533907890319824, + "num_tokens": 279794851.0, + "step": 12962 + }, + { + "epoch": 1.0689370825430857, + "grad_norm": 2.0635111331939697, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7758697271347046, + "num_tokens": 279822386.0, + "step": 12963 + }, + { + "epoch": 1.0690195431681373, + "grad_norm": 1.8000599145889282, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7486505508422852, + "num_tokens": 279861279.0, + "step": 12964 + }, + { + "epoch": 1.0691020037931886, + "grad_norm": 1.760657548904419, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.7821710109710693, + "num_tokens": 279896862.0, + "step": 12965 + }, + { + "epoch": 1.0691844644182402, + "grad_norm": 1.9661670923233032, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7847964763641357, + "num_tokens": 279925070.0, + "step": 12966 + }, + { + "epoch": 1.0692669250432918, + "grad_norm": 1.6937503814697266, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7855280637741089, + "num_tokens": 279961333.0, + "step": 12967 + }, + { + "epoch": 1.0693493856683434, + "grad_norm": 2.0377063751220703, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7688242197036743, + "num_tokens": 279988605.0, + "step": 12968 + }, + { + "epoch": 1.069431846293395, + "grad_norm": 2.111854076385498, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7333337068557739, + "num_tokens": 280015538.0, + "step": 12969 + }, + { + "epoch": 1.0695143069184465, + "grad_norm": 1.881737232208252, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7816845178604126, + "num_tokens": 280046181.0, + "step": 12970 + }, + { + "epoch": 1.069596767543498, + "grad_norm": 2.140171527862549, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7286391258239746, + "num_tokens": 280070752.0, + "step": 12971 + }, + { + "epoch": 1.0696792281685494, + "grad_norm": 1.7646130323410034, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.7680131196975708, + "num_tokens": 280108690.0, + "step": 12972 + }, + { + "epoch": 1.069761688793601, + "grad_norm": 2.181584119796753, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7334091067314148, + "num_tokens": 280134948.0, + "step": 12973 + }, + { + "epoch": 1.0698441494186526, + "grad_norm": 2.051995277404785, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.742116391658783, + "num_tokens": 280166147.0, + "step": 12974 + }, + { + "epoch": 1.0699266100437042, + "grad_norm": 1.799034595489502, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7690621614456177, + "num_tokens": 280199409.0, + "step": 12975 + }, + { + "epoch": 1.0700090706687557, + "grad_norm": 1.9144865274429321, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7429741621017456, + "num_tokens": 280232627.0, + "step": 12976 + }, + { + "epoch": 1.0700915312938073, + "grad_norm": 2.1073408126831055, + "learning_rate": 1e-06, + "loss": 0.6046, + "mean_token_accuracy": 0.7961044311523438, + "num_tokens": 280259358.0, + "step": 12977 + }, + { + "epoch": 1.0701739919188586, + "grad_norm": 2.0092835426330566, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7661528587341309, + "num_tokens": 280289801.0, + "step": 12978 + }, + { + "epoch": 1.0702564525439102, + "grad_norm": 2.0820610523223877, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7629595994949341, + "num_tokens": 280314861.0, + "step": 12979 + }, + { + "epoch": 1.0703389131689618, + "grad_norm": 2.059504747390747, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7642083168029785, + "num_tokens": 280343489.0, + "step": 12980 + }, + { + "epoch": 1.0704213737940134, + "grad_norm": 2.283722162246704, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7512337565422058, + "num_tokens": 280368233.0, + "step": 12981 + }, + { + "epoch": 1.070503834419065, + "grad_norm": 1.797940731048584, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7451313734054565, + "num_tokens": 280402886.0, + "step": 12982 + }, + { + "epoch": 1.0705862950441165, + "grad_norm": 1.9899781942367554, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7668901681900024, + "num_tokens": 280435316.0, + "step": 12983 + }, + { + "epoch": 1.070668755669168, + "grad_norm": 2.076469898223877, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7534201145172119, + "num_tokens": 280463949.0, + "step": 12984 + }, + { + "epoch": 1.0707512162942194, + "grad_norm": 2.1003923416137695, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7497917413711548, + "num_tokens": 280491738.0, + "step": 12985 + }, + { + "epoch": 1.070833676919271, + "grad_norm": 1.986093282699585, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7892379760742188, + "num_tokens": 280520340.0, + "step": 12986 + }, + { + "epoch": 1.0709161375443226, + "grad_norm": 2.2928125858306885, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7828898429870605, + "num_tokens": 280545595.0, + "step": 12987 + }, + { + "epoch": 1.0709985981693741, + "grad_norm": 1.987321138381958, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7811437249183655, + "num_tokens": 280574219.0, + "step": 12988 + }, + { + "epoch": 1.0710810587944257, + "grad_norm": 2.166367769241333, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7583351135253906, + "num_tokens": 280599670.0, + "step": 12989 + }, + { + "epoch": 1.0711635194194773, + "grad_norm": 1.8998535871505737, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7383110523223877, + "num_tokens": 280631610.0, + "step": 12990 + }, + { + "epoch": 1.0712459800445286, + "grad_norm": 1.9977076053619385, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7664583921432495, + "num_tokens": 280661570.0, + "step": 12991 + }, + { + "epoch": 1.0713284406695802, + "grad_norm": 1.9867663383483887, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7385525703430176, + "num_tokens": 280691898.0, + "step": 12992 + }, + { + "epoch": 1.0714109012946318, + "grad_norm": 1.8861992359161377, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.7476370334625244, + "num_tokens": 280725718.0, + "step": 12993 + }, + { + "epoch": 1.0714933619196834, + "grad_norm": 1.8419381380081177, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.8055779933929443, + "num_tokens": 280758233.0, + "step": 12994 + }, + { + "epoch": 1.071575822544735, + "grad_norm": 1.818038821220398, + "learning_rate": 1e-06, + "loss": 0.6195, + "mean_token_accuracy": 0.7874830365180969, + "num_tokens": 280792740.0, + "step": 12995 + }, + { + "epoch": 1.0716582831697865, + "grad_norm": 1.8008689880371094, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7513108253479004, + "num_tokens": 280828159.0, + "step": 12996 + }, + { + "epoch": 1.0717407437948379, + "grad_norm": 2.0161943435668945, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7525769472122192, + "num_tokens": 280858682.0, + "step": 12997 + }, + { + "epoch": 1.0718232044198894, + "grad_norm": 1.9826326370239258, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7246915102005005, + "num_tokens": 280888168.0, + "step": 12998 + }, + { + "epoch": 1.071905665044941, + "grad_norm": 2.0429046154022217, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7577333450317383, + "num_tokens": 280916238.0, + "step": 12999 + }, + { + "epoch": 1.0719881256699926, + "grad_norm": 1.9699676036834717, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7161365151405334, + "num_tokens": 280948234.0, + "step": 13000 + }, + { + "epoch": 1.0720705862950441, + "grad_norm": 2.0965542793273926, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7229915857315063, + "num_tokens": 280977382.0, + "step": 13001 + }, + { + "epoch": 1.0721530469200957, + "grad_norm": 2.1002702713012695, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7262090444564819, + "num_tokens": 281006415.0, + "step": 13002 + }, + { + "epoch": 1.0722355075451473, + "grad_norm": 2.0284547805786133, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.7577987313270569, + "num_tokens": 281036857.0, + "step": 13003 + }, + { + "epoch": 1.0723179681701986, + "grad_norm": 1.9982837438583374, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7471499443054199, + "num_tokens": 281066741.0, + "step": 13004 + }, + { + "epoch": 1.0724004287952502, + "grad_norm": 1.9645636081695557, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7460111379623413, + "num_tokens": 281097943.0, + "step": 13005 + }, + { + "epoch": 1.0724828894203018, + "grad_norm": 1.91240656375885, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7466710805892944, + "num_tokens": 281131787.0, + "step": 13006 + }, + { + "epoch": 1.0725653500453534, + "grad_norm": 1.934317946434021, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.7920501232147217, + "num_tokens": 281162116.0, + "step": 13007 + }, + { + "epoch": 1.072647810670405, + "grad_norm": 1.8117294311523438, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7456345558166504, + "num_tokens": 281196544.0, + "step": 13008 + }, + { + "epoch": 1.0727302712954565, + "grad_norm": 1.7470359802246094, + "learning_rate": 1e-06, + "loss": 0.5937, + "mean_token_accuracy": 0.807765007019043, + "num_tokens": 281230416.0, + "step": 13009 + }, + { + "epoch": 1.0728127319205079, + "grad_norm": 1.8691635131835938, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7018570303916931, + "num_tokens": 281264816.0, + "step": 13010 + }, + { + "epoch": 1.0728951925455594, + "grad_norm": 2.1529152393341064, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7650426626205444, + "num_tokens": 281289177.0, + "step": 13011 + }, + { + "epoch": 1.072977653170611, + "grad_norm": 2.088552951812744, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7433461546897888, + "num_tokens": 281317741.0, + "step": 13012 + }, + { + "epoch": 1.0730601137956626, + "grad_norm": 1.8927720785140991, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.746627688407898, + "num_tokens": 281351317.0, + "step": 13013 + }, + { + "epoch": 1.0731425744207141, + "grad_norm": 2.0956692695617676, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7206876873970032, + "num_tokens": 281379631.0, + "step": 13014 + }, + { + "epoch": 1.0732250350457657, + "grad_norm": 1.8065361976623535, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7515230178833008, + "num_tokens": 281412578.0, + "step": 13015 + }, + { + "epoch": 1.073307495670817, + "grad_norm": 2.350637674331665, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7170892357826233, + "num_tokens": 281435640.0, + "step": 13016 + }, + { + "epoch": 1.0733899562958686, + "grad_norm": 1.956578016281128, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7279481887817383, + "num_tokens": 281467869.0, + "step": 13017 + }, + { + "epoch": 1.0734724169209202, + "grad_norm": 1.792466640472412, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7647361159324646, + "num_tokens": 281503147.0, + "step": 13018 + }, + { + "epoch": 1.0735548775459718, + "grad_norm": 1.9465214014053345, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7858459949493408, + "num_tokens": 281530737.0, + "step": 13019 + }, + { + "epoch": 1.0736373381710234, + "grad_norm": 1.9242738485336304, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7638247609138489, + "num_tokens": 281562331.0, + "step": 13020 + }, + { + "epoch": 1.073719798796075, + "grad_norm": 2.0682930946350098, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.766435980796814, + "num_tokens": 281592329.0, + "step": 13021 + }, + { + "epoch": 1.0738022594211265, + "grad_norm": 1.711243748664856, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7455365657806396, + "num_tokens": 281630919.0, + "step": 13022 + }, + { + "epoch": 1.0738847200461779, + "grad_norm": 1.9349921941757202, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7706054449081421, + "num_tokens": 281661577.0, + "step": 13023 + }, + { + "epoch": 1.0739671806712294, + "grad_norm": 2.063735246658325, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7418016195297241, + "num_tokens": 281692343.0, + "step": 13024 + }, + { + "epoch": 1.074049641296281, + "grad_norm": 2.281022787094116, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7561898827552795, + "num_tokens": 281717253.0, + "step": 13025 + }, + { + "epoch": 1.0741321019213326, + "grad_norm": 2.189927816390991, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7647736072540283, + "num_tokens": 281746995.0, + "step": 13026 + }, + { + "epoch": 1.0742145625463841, + "grad_norm": 2.0920629501342773, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8002326488494873, + "num_tokens": 281770209.0, + "step": 13027 + }, + { + "epoch": 1.0742970231714357, + "grad_norm": 1.8412171602249146, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7501198053359985, + "num_tokens": 281804516.0, + "step": 13028 + }, + { + "epoch": 1.0743794837964873, + "grad_norm": 2.3578100204467773, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7585297226905823, + "num_tokens": 281825663.0, + "step": 13029 + }, + { + "epoch": 1.0744619444215386, + "grad_norm": 2.190917730331421, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7332624793052673, + "num_tokens": 281851531.0, + "step": 13030 + }, + { + "epoch": 1.0745444050465902, + "grad_norm": 2.06524920463562, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7367724180221558, + "num_tokens": 281882086.0, + "step": 13031 + }, + { + "epoch": 1.0746268656716418, + "grad_norm": 1.8783785104751587, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7786403894424438, + "num_tokens": 281913446.0, + "step": 13032 + }, + { + "epoch": 1.0747093262966934, + "grad_norm": 1.8061034679412842, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7523698806762695, + "num_tokens": 281948525.0, + "step": 13033 + }, + { + "epoch": 1.074791786921745, + "grad_norm": 1.9997413158416748, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7795668244361877, + "num_tokens": 281976611.0, + "step": 13034 + }, + { + "epoch": 1.0748742475467965, + "grad_norm": 2.080183267593384, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7654718160629272, + "num_tokens": 282004345.0, + "step": 13035 + }, + { + "epoch": 1.0749567081718479, + "grad_norm": 1.9018317461013794, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.729384183883667, + "num_tokens": 282036114.0, + "step": 13036 + }, + { + "epoch": 1.0750391687968994, + "grad_norm": 2.1166741847991943, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7626079320907593, + "num_tokens": 282061938.0, + "step": 13037 + }, + { + "epoch": 1.075121629421951, + "grad_norm": 2.130202293395996, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7709900140762329, + "num_tokens": 282089166.0, + "step": 13038 + }, + { + "epoch": 1.0752040900470026, + "grad_norm": 2.2050812244415283, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.7934072017669678, + "num_tokens": 282117310.0, + "step": 13039 + }, + { + "epoch": 1.0752865506720541, + "grad_norm": 1.8715413808822632, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.7522733211517334, + "num_tokens": 282153232.0, + "step": 13040 + }, + { + "epoch": 1.0753690112971057, + "grad_norm": 2.5817811489105225, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7395761609077454, + "num_tokens": 282174950.0, + "step": 13041 + }, + { + "epoch": 1.075451471922157, + "grad_norm": 2.0814452171325684, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.740966260433197, + "num_tokens": 282204202.0, + "step": 13042 + }, + { + "epoch": 1.0755339325472086, + "grad_norm": 1.967017412185669, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7582210302352905, + "num_tokens": 282232366.0, + "step": 13043 + }, + { + "epoch": 1.0756163931722602, + "grad_norm": 1.7501951456069946, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7695607542991638, + "num_tokens": 282269729.0, + "step": 13044 + }, + { + "epoch": 1.0756988537973118, + "grad_norm": 1.675570011138916, + "learning_rate": 1e-06, + "loss": 0.5926, + "mean_token_accuracy": 0.8008293509483337, + "num_tokens": 282305780.0, + "step": 13045 + }, + { + "epoch": 1.0757813144223634, + "grad_norm": 1.7687385082244873, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7701541781425476, + "num_tokens": 282341394.0, + "step": 13046 + }, + { + "epoch": 1.075863775047415, + "grad_norm": 1.9185636043548584, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7846555113792419, + "num_tokens": 282371449.0, + "step": 13047 + }, + { + "epoch": 1.0759462356724665, + "grad_norm": 2.0235698223114014, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7581733465194702, + "num_tokens": 282399851.0, + "step": 13048 + }, + { + "epoch": 1.0760286962975179, + "grad_norm": 2.2204620838165283, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7677829265594482, + "num_tokens": 282426343.0, + "step": 13049 + }, + { + "epoch": 1.0761111569225694, + "grad_norm": 1.946374535560608, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7828245162963867, + "num_tokens": 282455508.0, + "step": 13050 + }, + { + "epoch": 1.076193617547621, + "grad_norm": 2.000338077545166, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7574889659881592, + "num_tokens": 282485580.0, + "step": 13051 + }, + { + "epoch": 1.0762760781726726, + "grad_norm": 2.0601463317871094, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.752628743648529, + "num_tokens": 282514315.0, + "step": 13052 + }, + { + "epoch": 1.0763585387977241, + "grad_norm": 2.079883337020874, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7518458962440491, + "num_tokens": 282542275.0, + "step": 13053 + }, + { + "epoch": 1.0764409994227757, + "grad_norm": 2.241712808609009, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7479698061943054, + "num_tokens": 282566838.0, + "step": 13054 + }, + { + "epoch": 1.076523460047827, + "grad_norm": 2.000809907913208, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7584208250045776, + "num_tokens": 282596890.0, + "step": 13055 + }, + { + "epoch": 1.0766059206728786, + "grad_norm": 2.1233291625976562, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.70393967628479, + "num_tokens": 282626735.0, + "step": 13056 + }, + { + "epoch": 1.0766883812979302, + "grad_norm": 2.1077375411987305, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.7804313898086548, + "num_tokens": 282655954.0, + "step": 13057 + }, + { + "epoch": 1.0767708419229818, + "grad_norm": 2.063934087753296, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7499953508377075, + "num_tokens": 282683410.0, + "step": 13058 + }, + { + "epoch": 1.0768533025480334, + "grad_norm": 2.0172791481018066, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7488048076629639, + "num_tokens": 282716391.0, + "step": 13059 + }, + { + "epoch": 1.076935763173085, + "grad_norm": 2.260125160217285, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7258391380310059, + "num_tokens": 282740841.0, + "step": 13060 + }, + { + "epoch": 1.0770182237981363, + "grad_norm": 1.8167731761932373, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7480294108390808, + "num_tokens": 282778218.0, + "step": 13061 + }, + { + "epoch": 1.0771006844231878, + "grad_norm": 2.313344717025757, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7793363928794861, + "num_tokens": 282800834.0, + "step": 13062 + }, + { + "epoch": 1.0771831450482394, + "grad_norm": 1.8906455039978027, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7441942691802979, + "num_tokens": 282834227.0, + "step": 13063 + }, + { + "epoch": 1.077265605673291, + "grad_norm": 2.0173206329345703, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.748294472694397, + "num_tokens": 282861175.0, + "step": 13064 + }, + { + "epoch": 1.0773480662983426, + "grad_norm": 1.7824621200561523, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7605960369110107, + "num_tokens": 282895822.0, + "step": 13065 + }, + { + "epoch": 1.0774305269233941, + "grad_norm": 2.5002379417419434, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7353291511535645, + "num_tokens": 282916799.0, + "step": 13066 + }, + { + "epoch": 1.0775129875484457, + "grad_norm": 2.1949095726013184, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7651523947715759, + "num_tokens": 282941986.0, + "step": 13067 + }, + { + "epoch": 1.077595448173497, + "grad_norm": 2.027409076690674, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7726373076438904, + "num_tokens": 282969383.0, + "step": 13068 + }, + { + "epoch": 1.0776779087985486, + "grad_norm": 2.0141632556915283, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7321804165840149, + "num_tokens": 282998549.0, + "step": 13069 + }, + { + "epoch": 1.0777603694236002, + "grad_norm": 1.7888374328613281, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7668548822402954, + "num_tokens": 283033513.0, + "step": 13070 + }, + { + "epoch": 1.0778428300486518, + "grad_norm": 1.7366615533828735, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7726114988327026, + "num_tokens": 283067513.0, + "step": 13071 + }, + { + "epoch": 1.0779252906737034, + "grad_norm": 2.2649059295654297, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7281357049942017, + "num_tokens": 283093582.0, + "step": 13072 + }, + { + "epoch": 1.078007751298755, + "grad_norm": 2.1520190238952637, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7302855253219604, + "num_tokens": 283119761.0, + "step": 13073 + }, + { + "epoch": 1.0780902119238065, + "grad_norm": 1.953267216682434, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7485309839248657, + "num_tokens": 283152190.0, + "step": 13074 + }, + { + "epoch": 1.0781726725488578, + "grad_norm": 2.190370559692383, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7500144243240356, + "num_tokens": 283179481.0, + "step": 13075 + }, + { + "epoch": 1.0782551331739094, + "grad_norm": 2.1584858894348145, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.741032600402832, + "num_tokens": 283206293.0, + "step": 13076 + }, + { + "epoch": 1.078337593798961, + "grad_norm": 1.9819209575653076, + "learning_rate": 1e-06, + "loss": 0.7877, + "mean_token_accuracy": 0.7541110515594482, + "num_tokens": 283243472.0, + "step": 13077 + }, + { + "epoch": 1.0784200544240126, + "grad_norm": 2.0129013061523438, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7807546854019165, + "num_tokens": 283272363.0, + "step": 13078 + }, + { + "epoch": 1.0785025150490641, + "grad_norm": 1.964830994606018, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7564966082572937, + "num_tokens": 283302136.0, + "step": 13079 + }, + { + "epoch": 1.0785849756741157, + "grad_norm": 2.339898109436035, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.758407711982727, + "num_tokens": 283324023.0, + "step": 13080 + }, + { + "epoch": 1.078667436299167, + "grad_norm": 1.7823517322540283, + "learning_rate": 1e-06, + "loss": 0.5613, + "mean_token_accuracy": 0.8150975704193115, + "num_tokens": 283360208.0, + "step": 13081 + }, + { + "epoch": 1.0787498969242186, + "grad_norm": 2.1187002658843994, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7653558850288391, + "num_tokens": 283386200.0, + "step": 13082 + }, + { + "epoch": 1.0788323575492702, + "grad_norm": 1.9758853912353516, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7744242548942566, + "num_tokens": 283417637.0, + "step": 13083 + }, + { + "epoch": 1.0789148181743218, + "grad_norm": 1.701370358467102, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.7995647192001343, + "num_tokens": 283453537.0, + "step": 13084 + }, + { + "epoch": 1.0789972787993733, + "grad_norm": 2.0544240474700928, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.755014955997467, + "num_tokens": 283480573.0, + "step": 13085 + }, + { + "epoch": 1.079079739424425, + "grad_norm": 1.8004156351089478, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.7997444868087769, + "num_tokens": 283513512.0, + "step": 13086 + }, + { + "epoch": 1.0791622000494763, + "grad_norm": 1.9675978422164917, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7616119980812073, + "num_tokens": 283544248.0, + "step": 13087 + }, + { + "epoch": 1.0792446606745278, + "grad_norm": 1.9692171812057495, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7411454916000366, + "num_tokens": 283572795.0, + "step": 13088 + }, + { + "epoch": 1.0793271212995794, + "grad_norm": 2.2628355026245117, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7325807809829712, + "num_tokens": 283601524.0, + "step": 13089 + }, + { + "epoch": 1.079409581924631, + "grad_norm": 2.0179061889648438, + "learning_rate": 1e-06, + "loss": 0.5414, + "mean_token_accuracy": 0.8104085922241211, + "num_tokens": 283630157.0, + "step": 13090 + }, + { + "epoch": 1.0794920425496826, + "grad_norm": 2.548591136932373, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7182029485702515, + "num_tokens": 283655190.0, + "step": 13091 + }, + { + "epoch": 1.0795745031747341, + "grad_norm": 2.026553153991699, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.791554868221283, + "num_tokens": 283681246.0, + "step": 13092 + }, + { + "epoch": 1.0796569637997857, + "grad_norm": 1.904555320739746, + "learning_rate": 1e-06, + "loss": 0.6085, + "mean_token_accuracy": 0.7933657169342041, + "num_tokens": 283710047.0, + "step": 13093 + }, + { + "epoch": 1.079739424424837, + "grad_norm": 2.14569354057312, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7747280597686768, + "num_tokens": 283738923.0, + "step": 13094 + }, + { + "epoch": 1.0798218850498886, + "grad_norm": 1.9023021459579468, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.7824676036834717, + "num_tokens": 283768764.0, + "step": 13095 + }, + { + "epoch": 1.0799043456749402, + "grad_norm": 2.315521001815796, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.784075140953064, + "num_tokens": 283792175.0, + "step": 13096 + }, + { + "epoch": 1.0799868062999918, + "grad_norm": 2.339449882507324, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7521852254867554, + "num_tokens": 283816943.0, + "step": 13097 + }, + { + "epoch": 1.0800692669250433, + "grad_norm": 2.1562206745147705, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7483898401260376, + "num_tokens": 283842774.0, + "step": 13098 + }, + { + "epoch": 1.080151727550095, + "grad_norm": 2.1208248138427734, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7543647289276123, + "num_tokens": 283870024.0, + "step": 13099 + }, + { + "epoch": 1.0802341881751463, + "grad_norm": 2.0923562049865723, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7439970374107361, + "num_tokens": 283901251.0, + "step": 13100 + }, + { + "epoch": 1.0803166488001978, + "grad_norm": 2.1051642894744873, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.761066198348999, + "num_tokens": 283930734.0, + "step": 13101 + }, + { + "epoch": 1.0803991094252494, + "grad_norm": 1.8730688095092773, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7459896802902222, + "num_tokens": 283964315.0, + "step": 13102 + }, + { + "epoch": 1.080481570050301, + "grad_norm": 2.323857069015503, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7509142160415649, + "num_tokens": 283986992.0, + "step": 13103 + }, + { + "epoch": 1.0805640306753526, + "grad_norm": 1.9771188497543335, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7476895451545715, + "num_tokens": 284019670.0, + "step": 13104 + }, + { + "epoch": 1.0806464913004041, + "grad_norm": 1.8895553350448608, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7804371118545532, + "num_tokens": 284051852.0, + "step": 13105 + }, + { + "epoch": 1.0807289519254555, + "grad_norm": 1.9860455989837646, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7150229215621948, + "num_tokens": 284085339.0, + "step": 13106 + }, + { + "epoch": 1.080811412550507, + "grad_norm": 2.101276397705078, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.7795276641845703, + "num_tokens": 284109204.0, + "step": 13107 + }, + { + "epoch": 1.0808938731755586, + "grad_norm": 1.8715705871582031, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7343586087226868, + "num_tokens": 284141871.0, + "step": 13108 + }, + { + "epoch": 1.0809763338006102, + "grad_norm": 2.0494611263275146, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.757553219795227, + "num_tokens": 284168325.0, + "step": 13109 + }, + { + "epoch": 1.0810587944256618, + "grad_norm": 2.0475516319274902, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7282131910324097, + "num_tokens": 284197114.0, + "step": 13110 + }, + { + "epoch": 1.0811412550507133, + "grad_norm": 1.9744164943695068, + "learning_rate": 1e-06, + "loss": 0.7775, + "mean_token_accuracy": 0.7562495470046997, + "num_tokens": 284227388.0, + "step": 13111 + }, + { + "epoch": 1.081223715675765, + "grad_norm": 2.114877700805664, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7391166687011719, + "num_tokens": 284257311.0, + "step": 13112 + }, + { + "epoch": 1.0813061763008163, + "grad_norm": 1.9300274848937988, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7040241956710815, + "num_tokens": 284292941.0, + "step": 13113 + }, + { + "epoch": 1.0813886369258678, + "grad_norm": 1.9398322105407715, + "learning_rate": 1e-06, + "loss": 0.5927, + "mean_token_accuracy": 0.7937322854995728, + "num_tokens": 284321197.0, + "step": 13114 + }, + { + "epoch": 1.0814710975509194, + "grad_norm": 2.2436790466308594, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7211811542510986, + "num_tokens": 284346628.0, + "step": 13115 + }, + { + "epoch": 1.081553558175971, + "grad_norm": 1.7254377603530884, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7867562174797058, + "num_tokens": 284383258.0, + "step": 13116 + }, + { + "epoch": 1.0816360188010226, + "grad_norm": 2.028140068054199, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.777605414390564, + "num_tokens": 284410808.0, + "step": 13117 + }, + { + "epoch": 1.0817184794260741, + "grad_norm": 2.1306705474853516, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7395272850990295, + "num_tokens": 284442096.0, + "step": 13118 + }, + { + "epoch": 1.0818009400511257, + "grad_norm": 1.8779864311218262, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7591758370399475, + "num_tokens": 284477699.0, + "step": 13119 + }, + { + "epoch": 1.081883400676177, + "grad_norm": 1.7920398712158203, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.7918088436126709, + "num_tokens": 284512301.0, + "step": 13120 + }, + { + "epoch": 1.0819658613012286, + "grad_norm": 1.990384817123413, + "learning_rate": 1e-06, + "loss": 0.5893, + "mean_token_accuracy": 0.8016682863235474, + "num_tokens": 284541636.0, + "step": 13121 + }, + { + "epoch": 1.0820483219262802, + "grad_norm": 1.9158402681350708, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7616439461708069, + "num_tokens": 284573478.0, + "step": 13122 + }, + { + "epoch": 1.0821307825513318, + "grad_norm": 2.1684517860412598, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.7679338455200195, + "num_tokens": 284599878.0, + "step": 13123 + }, + { + "epoch": 1.0822132431763833, + "grad_norm": 1.9123567342758179, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7479140758514404, + "num_tokens": 284635935.0, + "step": 13124 + }, + { + "epoch": 1.082295703801435, + "grad_norm": 2.0401480197906494, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.746850848197937, + "num_tokens": 284666389.0, + "step": 13125 + }, + { + "epoch": 1.0823781644264863, + "grad_norm": 2.05366587638855, + "learning_rate": 1e-06, + "loss": 0.6002, + "mean_token_accuracy": 0.7989922761917114, + "num_tokens": 284691897.0, + "step": 13126 + }, + { + "epoch": 1.0824606250515378, + "grad_norm": 1.8729904890060425, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.755168616771698, + "num_tokens": 284726027.0, + "step": 13127 + }, + { + "epoch": 1.0825430856765894, + "grad_norm": 1.8927667140960693, + "learning_rate": 1e-06, + "loss": 0.5734, + "mean_token_accuracy": 0.8037976622581482, + "num_tokens": 284756154.0, + "step": 13128 + }, + { + "epoch": 1.082625546301641, + "grad_norm": 2.3130359649658203, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7422306537628174, + "num_tokens": 284781449.0, + "step": 13129 + }, + { + "epoch": 1.0827080069266926, + "grad_norm": 1.8333618640899658, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.780909538269043, + "num_tokens": 284816041.0, + "step": 13130 + }, + { + "epoch": 1.0827904675517441, + "grad_norm": 1.8540112972259521, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.758610188961029, + "num_tokens": 284851510.0, + "step": 13131 + }, + { + "epoch": 1.0828729281767955, + "grad_norm": 2.1248812675476074, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.7882173657417297, + "num_tokens": 284877333.0, + "step": 13132 + }, + { + "epoch": 1.082955388801847, + "grad_norm": 1.6636080741882324, + "learning_rate": 1e-06, + "loss": 0.5903, + "mean_token_accuracy": 0.80656898021698, + "num_tokens": 284917010.0, + "step": 13133 + }, + { + "epoch": 1.0830378494268986, + "grad_norm": 1.9176849126815796, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7408497333526611, + "num_tokens": 284948280.0, + "step": 13134 + }, + { + "epoch": 1.0831203100519502, + "grad_norm": 1.8624485731124878, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7491477727890015, + "num_tokens": 284982266.0, + "step": 13135 + }, + { + "epoch": 1.0832027706770018, + "grad_norm": 2.2940165996551514, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7311696410179138, + "num_tokens": 285007181.0, + "step": 13136 + }, + { + "epoch": 1.0832852313020533, + "grad_norm": 1.9764469861984253, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7719265222549438, + "num_tokens": 285037379.0, + "step": 13137 + }, + { + "epoch": 1.083367691927105, + "grad_norm": 1.8533496856689453, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7471602559089661, + "num_tokens": 285071437.0, + "step": 13138 + }, + { + "epoch": 1.0834501525521563, + "grad_norm": 2.256571054458618, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7549294233322144, + "num_tokens": 285094518.0, + "step": 13139 + }, + { + "epoch": 1.0835326131772078, + "grad_norm": 2.117936134338379, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.725674033164978, + "num_tokens": 285123380.0, + "step": 13140 + }, + { + "epoch": 1.0836150738022594, + "grad_norm": 1.841233730316162, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7369265556335449, + "num_tokens": 285160996.0, + "step": 13141 + }, + { + "epoch": 1.083697534427311, + "grad_norm": 2.0660345554351807, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.733485221862793, + "num_tokens": 285190821.0, + "step": 13142 + }, + { + "epoch": 1.0837799950523626, + "grad_norm": 1.7922197580337524, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7288503050804138, + "num_tokens": 285228425.0, + "step": 13143 + }, + { + "epoch": 1.0838624556774141, + "grad_norm": 2.0797553062438965, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7713655233383179, + "num_tokens": 285256853.0, + "step": 13144 + }, + { + "epoch": 1.0839449163024655, + "grad_norm": 2.1039202213287354, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7490094900131226, + "num_tokens": 285286161.0, + "step": 13145 + }, + { + "epoch": 1.084027376927517, + "grad_norm": 2.1301863193511963, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7402930855751038, + "num_tokens": 285314270.0, + "step": 13146 + }, + { + "epoch": 1.0841098375525686, + "grad_norm": 2.1014084815979004, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.7458179593086243, + "num_tokens": 285340508.0, + "step": 13147 + }, + { + "epoch": 1.0841922981776202, + "grad_norm": 2.3619744777679443, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.735226035118103, + "num_tokens": 285362102.0, + "step": 13148 + }, + { + "epoch": 1.0842747588026718, + "grad_norm": 1.7975952625274658, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7244514226913452, + "num_tokens": 285398479.0, + "step": 13149 + }, + { + "epoch": 1.0843572194277233, + "grad_norm": 1.9175529479980469, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7824801206588745, + "num_tokens": 285431736.0, + "step": 13150 + }, + { + "epoch": 1.0844396800527747, + "grad_norm": 1.920600414276123, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.7806335687637329, + "num_tokens": 285464257.0, + "step": 13151 + }, + { + "epoch": 1.0845221406778263, + "grad_norm": 2.074359893798828, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.7894302606582642, + "num_tokens": 285490380.0, + "step": 13152 + }, + { + "epoch": 1.0846046013028778, + "grad_norm": 2.041966438293457, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7481502294540405, + "num_tokens": 285518522.0, + "step": 13153 + }, + { + "epoch": 1.0846870619279294, + "grad_norm": 1.9218748807907104, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7339995503425598, + "num_tokens": 285552281.0, + "step": 13154 + }, + { + "epoch": 1.084769522552981, + "grad_norm": 2.4518587589263916, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7381420135498047, + "num_tokens": 285574003.0, + "step": 13155 + }, + { + "epoch": 1.0848519831780326, + "grad_norm": 1.9161288738250732, + "learning_rate": 1e-06, + "loss": 0.6025, + "mean_token_accuracy": 0.8068617582321167, + "num_tokens": 285602438.0, + "step": 13156 + }, + { + "epoch": 1.0849344438030841, + "grad_norm": 2.1308326721191406, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7315007448196411, + "num_tokens": 285631681.0, + "step": 13157 + }, + { + "epoch": 1.0850169044281355, + "grad_norm": 1.732340931892395, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.790563702583313, + "num_tokens": 285666475.0, + "step": 13158 + }, + { + "epoch": 1.085099365053187, + "grad_norm": 1.9705432653427124, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7459875345230103, + "num_tokens": 285698184.0, + "step": 13159 + }, + { + "epoch": 1.0851818256782386, + "grad_norm": 1.9366337060928345, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.776971697807312, + "num_tokens": 285730520.0, + "step": 13160 + }, + { + "epoch": 1.0852642863032902, + "grad_norm": 1.9266735315322876, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7349622249603271, + "num_tokens": 285764416.0, + "step": 13161 + }, + { + "epoch": 1.0853467469283418, + "grad_norm": 1.8178858757019043, + "learning_rate": 1e-06, + "loss": 0.5911, + "mean_token_accuracy": 0.808611273765564, + "num_tokens": 285799540.0, + "step": 13162 + }, + { + "epoch": 1.0854292075533933, + "grad_norm": 1.9324527978897095, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7927485108375549, + "num_tokens": 285834282.0, + "step": 13163 + }, + { + "epoch": 1.0855116681784447, + "grad_norm": 1.8670762777328491, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7758538126945496, + "num_tokens": 285866623.0, + "step": 13164 + }, + { + "epoch": 1.0855941288034963, + "grad_norm": 1.9557099342346191, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.7971253991127014, + "num_tokens": 285898419.0, + "step": 13165 + }, + { + "epoch": 1.0856765894285478, + "grad_norm": 2.214948892593384, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7570708990097046, + "num_tokens": 285925178.0, + "step": 13166 + }, + { + "epoch": 1.0857590500535994, + "grad_norm": 2.0866551399230957, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7580323815345764, + "num_tokens": 285953036.0, + "step": 13167 + }, + { + "epoch": 1.085841510678651, + "grad_norm": 1.9682366847991943, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7388429641723633, + "num_tokens": 285987508.0, + "step": 13168 + }, + { + "epoch": 1.0859239713037026, + "grad_norm": 2.1865556240081787, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7877439260482788, + "num_tokens": 286010452.0, + "step": 13169 + }, + { + "epoch": 1.0860064319287541, + "grad_norm": 2.6885809898376465, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7629632949829102, + "num_tokens": 286037532.0, + "step": 13170 + }, + { + "epoch": 1.0860888925538055, + "grad_norm": 2.055539608001709, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7519338130950928, + "num_tokens": 286066205.0, + "step": 13171 + }, + { + "epoch": 1.086171353178857, + "grad_norm": 1.9421879053115845, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.768358051776886, + "num_tokens": 286097344.0, + "step": 13172 + }, + { + "epoch": 1.0862538138039086, + "grad_norm": 1.984869122505188, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7418441772460938, + "num_tokens": 286130007.0, + "step": 13173 + }, + { + "epoch": 1.0863362744289602, + "grad_norm": 2.312319755554199, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.72752845287323, + "num_tokens": 286157442.0, + "step": 13174 + }, + { + "epoch": 1.0864187350540118, + "grad_norm": 1.8680967092514038, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7217678427696228, + "num_tokens": 286192003.0, + "step": 13175 + }, + { + "epoch": 1.0865011956790633, + "grad_norm": 2.0179574489593506, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7725114226341248, + "num_tokens": 286218933.0, + "step": 13176 + }, + { + "epoch": 1.0865836563041147, + "grad_norm": 2.0029993057250977, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7419871687889099, + "num_tokens": 286249903.0, + "step": 13177 + }, + { + "epoch": 1.0866661169291663, + "grad_norm": 1.982306718826294, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7352139353752136, + "num_tokens": 286281525.0, + "step": 13178 + }, + { + "epoch": 1.0867485775542178, + "grad_norm": 1.8499281406402588, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7445703744888306, + "num_tokens": 286316790.0, + "step": 13179 + }, + { + "epoch": 1.0868310381792694, + "grad_norm": 2.0681164264678955, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7506751418113708, + "num_tokens": 286343196.0, + "step": 13180 + }, + { + "epoch": 1.086913498804321, + "grad_norm": 2.0074238777160645, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7726738452911377, + "num_tokens": 286373338.0, + "step": 13181 + }, + { + "epoch": 1.0869959594293725, + "grad_norm": 2.135850191116333, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7789140939712524, + "num_tokens": 286398824.0, + "step": 13182 + }, + { + "epoch": 1.0870784200544241, + "grad_norm": 1.8867472410202026, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7629145383834839, + "num_tokens": 286432717.0, + "step": 13183 + }, + { + "epoch": 1.0871608806794755, + "grad_norm": 2.0638742446899414, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7218977212905884, + "num_tokens": 286464072.0, + "step": 13184 + }, + { + "epoch": 1.087243341304527, + "grad_norm": 2.19732403755188, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7497106790542603, + "num_tokens": 286489294.0, + "step": 13185 + }, + { + "epoch": 1.0873258019295786, + "grad_norm": 2.0175392627716064, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7768889665603638, + "num_tokens": 286520185.0, + "step": 13186 + }, + { + "epoch": 1.0874082625546302, + "grad_norm": 1.7326351404190063, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.769450306892395, + "num_tokens": 286552211.0, + "step": 13187 + }, + { + "epoch": 1.0874907231796818, + "grad_norm": 2.0388500690460205, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7303509712219238, + "num_tokens": 286582455.0, + "step": 13188 + }, + { + "epoch": 1.0875731838047333, + "grad_norm": 2.293344020843506, + "learning_rate": 1e-06, + "loss": 0.734, + "mean_token_accuracy": 0.7646273374557495, + "num_tokens": 286607485.0, + "step": 13189 + }, + { + "epoch": 1.0876556444297847, + "grad_norm": 2.0733799934387207, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7247363328933716, + "num_tokens": 286637551.0, + "step": 13190 + }, + { + "epoch": 1.0877381050548363, + "grad_norm": 2.027153730392456, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7549566626548767, + "num_tokens": 286667632.0, + "step": 13191 + }, + { + "epoch": 1.0878205656798878, + "grad_norm": 2.0388662815093994, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7348374128341675, + "num_tokens": 286698530.0, + "step": 13192 + }, + { + "epoch": 1.0879030263049394, + "grad_norm": 2.039729595184326, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.780357837677002, + "num_tokens": 286725230.0, + "step": 13193 + }, + { + "epoch": 1.087985486929991, + "grad_norm": 1.9859331846237183, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.7826506495475769, + "num_tokens": 286755717.0, + "step": 13194 + }, + { + "epoch": 1.0880679475550425, + "grad_norm": 2.2709383964538574, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7472666501998901, + "num_tokens": 286780335.0, + "step": 13195 + }, + { + "epoch": 1.088150408180094, + "grad_norm": 2.2585177421569824, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7607759237289429, + "num_tokens": 286804538.0, + "step": 13196 + }, + { + "epoch": 1.0882328688051455, + "grad_norm": 1.991902470588684, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7571942806243896, + "num_tokens": 286833104.0, + "step": 13197 + }, + { + "epoch": 1.088315329430197, + "grad_norm": 1.855970859527588, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.7514951229095459, + "num_tokens": 286869778.0, + "step": 13198 + }, + { + "epoch": 1.0883977900552486, + "grad_norm": 1.8244792222976685, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7274723052978516, + "num_tokens": 286906116.0, + "step": 13199 + }, + { + "epoch": 1.0884802506803002, + "grad_norm": 1.7122637033462524, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.7961046695709229, + "num_tokens": 286942766.0, + "step": 13200 + }, + { + "epoch": 1.0885627113053518, + "grad_norm": 2.0407187938690186, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7521100044250488, + "num_tokens": 286971495.0, + "step": 13201 + }, + { + "epoch": 1.0886451719304033, + "grad_norm": 2.039769172668457, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7474408149719238, + "num_tokens": 287001934.0, + "step": 13202 + }, + { + "epoch": 1.0887276325554547, + "grad_norm": 2.061599016189575, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7420576810836792, + "num_tokens": 287032814.0, + "step": 13203 + }, + { + "epoch": 1.0888100931805063, + "grad_norm": 2.2428202629089355, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7766283750534058, + "num_tokens": 287057160.0, + "step": 13204 + }, + { + "epoch": 1.0888925538055578, + "grad_norm": 2.10302472114563, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7786681652069092, + "num_tokens": 287082729.0, + "step": 13205 + }, + { + "epoch": 1.0889750144306094, + "grad_norm": 1.9383814334869385, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7547919750213623, + "num_tokens": 287116068.0, + "step": 13206 + }, + { + "epoch": 1.089057475055661, + "grad_norm": 2.498253345489502, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7542464733123779, + "num_tokens": 287137177.0, + "step": 13207 + }, + { + "epoch": 1.0891399356807125, + "grad_norm": 2.019920825958252, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7812819480895996, + "num_tokens": 287165747.0, + "step": 13208 + }, + { + "epoch": 1.089222396305764, + "grad_norm": 1.9898396730422974, + "learning_rate": 1e-06, + "loss": 0.5877, + "mean_token_accuracy": 0.808836042881012, + "num_tokens": 287191431.0, + "step": 13209 + }, + { + "epoch": 1.0893048569308155, + "grad_norm": 2.0633535385131836, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7672943472862244, + "num_tokens": 287218678.0, + "step": 13210 + }, + { + "epoch": 1.089387317555867, + "grad_norm": 1.897913932800293, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7457993626594543, + "num_tokens": 287252854.0, + "step": 13211 + }, + { + "epoch": 1.0894697781809186, + "grad_norm": 1.912324070930481, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.7503674626350403, + "num_tokens": 287285285.0, + "step": 13212 + }, + { + "epoch": 1.0895522388059702, + "grad_norm": 1.8980351686477661, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7501855492591858, + "num_tokens": 287316433.0, + "step": 13213 + }, + { + "epoch": 1.0896346994310218, + "grad_norm": 2.0775341987609863, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7631747722625732, + "num_tokens": 287345090.0, + "step": 13214 + }, + { + "epoch": 1.089717160056073, + "grad_norm": 2.005876302719116, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.752062201499939, + "num_tokens": 287374986.0, + "step": 13215 + }, + { + "epoch": 1.0897996206811247, + "grad_norm": 1.8471364974975586, + "learning_rate": 1e-06, + "loss": 0.7614, + "mean_token_accuracy": 0.7565629482269287, + "num_tokens": 287411163.0, + "step": 13216 + }, + { + "epoch": 1.0898820813061763, + "grad_norm": 2.1348283290863037, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7658129930496216, + "num_tokens": 287437013.0, + "step": 13217 + }, + { + "epoch": 1.0899645419312278, + "grad_norm": 2.224360942840576, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7480953335762024, + "num_tokens": 287465516.0, + "step": 13218 + }, + { + "epoch": 1.0900470025562794, + "grad_norm": 2.0059337615966797, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7863556146621704, + "num_tokens": 287493672.0, + "step": 13219 + }, + { + "epoch": 1.090129463181331, + "grad_norm": 2.35996150970459, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.719927191734314, + "num_tokens": 287518121.0, + "step": 13220 + }, + { + "epoch": 1.0902119238063825, + "grad_norm": 2.007256269454956, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.7773512005805969, + "num_tokens": 287545635.0, + "step": 13221 + }, + { + "epoch": 1.090294384431434, + "grad_norm": 1.924906849861145, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7641774415969849, + "num_tokens": 287580867.0, + "step": 13222 + }, + { + "epoch": 1.0903768450564855, + "grad_norm": 2.155006170272827, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7381349802017212, + "num_tokens": 287608257.0, + "step": 13223 + }, + { + "epoch": 1.090459305681537, + "grad_norm": 1.6465085744857788, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7509236335754395, + "num_tokens": 287650121.0, + "step": 13224 + }, + { + "epoch": 1.0905417663065886, + "grad_norm": 2.086627244949341, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7548208236694336, + "num_tokens": 287677753.0, + "step": 13225 + }, + { + "epoch": 1.0906242269316402, + "grad_norm": 2.1518707275390625, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7802598476409912, + "num_tokens": 287703072.0, + "step": 13226 + }, + { + "epoch": 1.0907066875566918, + "grad_norm": 1.8449149131774902, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7800760269165039, + "num_tokens": 287732545.0, + "step": 13227 + }, + { + "epoch": 1.0907891481817433, + "grad_norm": 2.0746021270751953, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.7941837310791016, + "num_tokens": 287759264.0, + "step": 13228 + }, + { + "epoch": 1.0908716088067947, + "grad_norm": 1.9000343084335327, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7486509084701538, + "num_tokens": 287790013.0, + "step": 13229 + }, + { + "epoch": 1.0909540694318463, + "grad_norm": 1.7891626358032227, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7820554971694946, + "num_tokens": 287826149.0, + "step": 13230 + }, + { + "epoch": 1.0910365300568978, + "grad_norm": 2.1199543476104736, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7489904165267944, + "num_tokens": 287855857.0, + "step": 13231 + }, + { + "epoch": 1.0911189906819494, + "grad_norm": 2.148033380508423, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7291997671127319, + "num_tokens": 287884268.0, + "step": 13232 + }, + { + "epoch": 1.091201451307001, + "grad_norm": 2.121553659439087, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.759797215461731, + "num_tokens": 287913351.0, + "step": 13233 + }, + { + "epoch": 1.0912839119320525, + "grad_norm": 1.8187190294265747, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7437386512756348, + "num_tokens": 287950913.0, + "step": 13234 + }, + { + "epoch": 1.091366372557104, + "grad_norm": 1.84718656539917, + "learning_rate": 1e-06, + "loss": 0.5461, + "mean_token_accuracy": 0.8091164827346802, + "num_tokens": 287978375.0, + "step": 13235 + }, + { + "epoch": 1.0914488331821555, + "grad_norm": 2.071126937866211, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7715078592300415, + "num_tokens": 288005475.0, + "step": 13236 + }, + { + "epoch": 1.091531293807207, + "grad_norm": 2.0939338207244873, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7483752965927124, + "num_tokens": 288033120.0, + "step": 13237 + }, + { + "epoch": 1.0916137544322586, + "grad_norm": 2.139540672302246, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7120743989944458, + "num_tokens": 288065325.0, + "step": 13238 + }, + { + "epoch": 1.0916962150573102, + "grad_norm": 2.0090794563293457, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7614178657531738, + "num_tokens": 288094529.0, + "step": 13239 + }, + { + "epoch": 1.0917786756823618, + "grad_norm": 2.0718131065368652, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.763847291469574, + "num_tokens": 288122173.0, + "step": 13240 + }, + { + "epoch": 1.091861136307413, + "grad_norm": 2.0896596908569336, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7273303270339966, + "num_tokens": 288150100.0, + "step": 13241 + }, + { + "epoch": 1.0919435969324647, + "grad_norm": 1.906455159187317, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.772815465927124, + "num_tokens": 288181628.0, + "step": 13242 + }, + { + "epoch": 1.0920260575575162, + "grad_norm": 1.971853256225586, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7811165452003479, + "num_tokens": 288213076.0, + "step": 13243 + }, + { + "epoch": 1.0921085181825678, + "grad_norm": 2.1661810874938965, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7155144810676575, + "num_tokens": 288241839.0, + "step": 13244 + }, + { + "epoch": 1.0921909788076194, + "grad_norm": 2.3028979301452637, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.744325578212738, + "num_tokens": 288265873.0, + "step": 13245 + }, + { + "epoch": 1.092273439432671, + "grad_norm": 1.8805267810821533, + "learning_rate": 1e-06, + "loss": 0.5866, + "mean_token_accuracy": 0.8068941831588745, + "num_tokens": 288296480.0, + "step": 13246 + }, + { + "epoch": 1.0923559000577225, + "grad_norm": 2.0453319549560547, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7479925155639648, + "num_tokens": 288325800.0, + "step": 13247 + }, + { + "epoch": 1.092438360682774, + "grad_norm": 2.1371099948883057, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.780263364315033, + "num_tokens": 288353219.0, + "step": 13248 + }, + { + "epoch": 1.0925208213078255, + "grad_norm": 1.8202497959136963, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7557625770568848, + "num_tokens": 288390130.0, + "step": 13249 + }, + { + "epoch": 1.092603281932877, + "grad_norm": 1.931098222732544, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7615042328834534, + "num_tokens": 288421538.0, + "step": 13250 + }, + { + "epoch": 1.0926857425579286, + "grad_norm": 2.154662847518921, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7586559653282166, + "num_tokens": 288446406.0, + "step": 13251 + }, + { + "epoch": 1.0927682031829802, + "grad_norm": 1.8677558898925781, + "learning_rate": 1e-06, + "loss": 0.5938, + "mean_token_accuracy": 0.8056095838546753, + "num_tokens": 288478952.0, + "step": 13252 + }, + { + "epoch": 1.0928506638080318, + "grad_norm": 1.6689547300338745, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.7890227437019348, + "num_tokens": 288520591.0, + "step": 13253 + }, + { + "epoch": 1.092933124433083, + "grad_norm": 2.067426919937134, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7282368540763855, + "num_tokens": 288551243.0, + "step": 13254 + }, + { + "epoch": 1.0930155850581347, + "grad_norm": 1.7666212320327759, + "learning_rate": 1e-06, + "loss": 0.6003, + "mean_token_accuracy": 0.8063660860061646, + "num_tokens": 288588624.0, + "step": 13255 + }, + { + "epoch": 1.0930980456831862, + "grad_norm": 2.1612021923065186, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7687633037567139, + "num_tokens": 288614405.0, + "step": 13256 + }, + { + "epoch": 1.0931805063082378, + "grad_norm": 2.0096824169158936, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7835807800292969, + "num_tokens": 288642568.0, + "step": 13257 + }, + { + "epoch": 1.0932629669332894, + "grad_norm": 1.8518577814102173, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7870867252349854, + "num_tokens": 288679879.0, + "step": 13258 + }, + { + "epoch": 1.093345427558341, + "grad_norm": 1.9495460987091064, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.7895435094833374, + "num_tokens": 288712785.0, + "step": 13259 + }, + { + "epoch": 1.0934278881833923, + "grad_norm": 1.8208805322647095, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7633147239685059, + "num_tokens": 288748588.0, + "step": 13260 + }, + { + "epoch": 1.0935103488084439, + "grad_norm": 1.8692541122436523, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7708355188369751, + "num_tokens": 288784409.0, + "step": 13261 + }, + { + "epoch": 1.0935928094334955, + "grad_norm": 2.071331024169922, + "learning_rate": 1e-06, + "loss": 0.7842, + "mean_token_accuracy": 0.7518613934516907, + "num_tokens": 288812969.0, + "step": 13262 + }, + { + "epoch": 1.093675270058547, + "grad_norm": 1.959636926651001, + "learning_rate": 1e-06, + "loss": 0.7598, + "mean_token_accuracy": 0.7514730095863342, + "num_tokens": 288844700.0, + "step": 13263 + }, + { + "epoch": 1.0937577306835986, + "grad_norm": 1.8857377767562866, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.7936797142028809, + "num_tokens": 288876905.0, + "step": 13264 + }, + { + "epoch": 1.0938401913086502, + "grad_norm": 1.8979510068893433, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.7862206697463989, + "num_tokens": 288905272.0, + "step": 13265 + }, + { + "epoch": 1.0939226519337018, + "grad_norm": 2.120331287384033, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7445670366287231, + "num_tokens": 288935444.0, + "step": 13266 + }, + { + "epoch": 1.094005112558753, + "grad_norm": 1.8910367488861084, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7547240257263184, + "num_tokens": 288970091.0, + "step": 13267 + }, + { + "epoch": 1.0940875731838047, + "grad_norm": 2.372035026550293, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7485236525535583, + "num_tokens": 288992776.0, + "step": 13268 + }, + { + "epoch": 1.0941700338088562, + "grad_norm": 2.1337101459503174, + "learning_rate": 1e-06, + "loss": 0.6006, + "mean_token_accuracy": 0.8000506162643433, + "num_tokens": 289020063.0, + "step": 13269 + }, + { + "epoch": 1.0942524944339078, + "grad_norm": 1.9739367961883545, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.778333842754364, + "num_tokens": 289048962.0, + "step": 13270 + }, + { + "epoch": 1.0943349550589594, + "grad_norm": 2.226388692855835, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7491346597671509, + "num_tokens": 289074276.0, + "step": 13271 + }, + { + "epoch": 1.094417415684011, + "grad_norm": 2.040184736251831, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7581204175949097, + "num_tokens": 289106453.0, + "step": 13272 + }, + { + "epoch": 1.0944998763090625, + "grad_norm": 1.975763201713562, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7406731247901917, + "num_tokens": 289139319.0, + "step": 13273 + }, + { + "epoch": 1.0945823369341139, + "grad_norm": 2.1495354175567627, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7406919598579407, + "num_tokens": 289166040.0, + "step": 13274 + }, + { + "epoch": 1.0946647975591655, + "grad_norm": 1.9697027206420898, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.7884321212768555, + "num_tokens": 289193765.0, + "step": 13275 + }, + { + "epoch": 1.094747258184217, + "grad_norm": 1.915967583656311, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7559230327606201, + "num_tokens": 289225878.0, + "step": 13276 + }, + { + "epoch": 1.0948297188092686, + "grad_norm": 2.004767894744873, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7545469403266907, + "num_tokens": 289256705.0, + "step": 13277 + }, + { + "epoch": 1.0949121794343202, + "grad_norm": 1.8529260158538818, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.788938045501709, + "num_tokens": 289289542.0, + "step": 13278 + }, + { + "epoch": 1.0949946400593717, + "grad_norm": 1.965781569480896, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.8007864356040955, + "num_tokens": 289317789.0, + "step": 13279 + }, + { + "epoch": 1.095077100684423, + "grad_norm": 1.910498023033142, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.774657666683197, + "num_tokens": 289349268.0, + "step": 13280 + }, + { + "epoch": 1.0951595613094747, + "grad_norm": 2.036039113998413, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7686976194381714, + "num_tokens": 289379873.0, + "step": 13281 + }, + { + "epoch": 1.0952420219345262, + "grad_norm": 1.9394464492797852, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7293860912322998, + "num_tokens": 289412388.0, + "step": 13282 + }, + { + "epoch": 1.0953244825595778, + "grad_norm": 1.8382134437561035, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7767466306686401, + "num_tokens": 289445302.0, + "step": 13283 + }, + { + "epoch": 1.0954069431846294, + "grad_norm": 2.182100772857666, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7684904336929321, + "num_tokens": 289472806.0, + "step": 13284 + }, + { + "epoch": 1.095489403809681, + "grad_norm": 2.199104070663452, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7515024542808533, + "num_tokens": 289498366.0, + "step": 13285 + }, + { + "epoch": 1.0955718644347323, + "grad_norm": 1.8234601020812988, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.7982886433601379, + "num_tokens": 289530995.0, + "step": 13286 + }, + { + "epoch": 1.0956543250597839, + "grad_norm": 1.8782967329025269, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7409985661506653, + "num_tokens": 289563891.0, + "step": 13287 + }, + { + "epoch": 1.0957367856848355, + "grad_norm": 1.9126427173614502, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.8001813292503357, + "num_tokens": 289595149.0, + "step": 13288 + }, + { + "epoch": 1.095819246309887, + "grad_norm": 1.8694446086883545, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.76958167552948, + "num_tokens": 289625352.0, + "step": 13289 + }, + { + "epoch": 1.0959017069349386, + "grad_norm": 2.2294352054595947, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.712323784828186, + "num_tokens": 289654476.0, + "step": 13290 + }, + { + "epoch": 1.0959841675599902, + "grad_norm": 2.025517463684082, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.7844037413597107, + "num_tokens": 289682035.0, + "step": 13291 + }, + { + "epoch": 1.0960666281850417, + "grad_norm": 2.0646607875823975, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7650878429412842, + "num_tokens": 289711716.0, + "step": 13292 + }, + { + "epoch": 1.096149088810093, + "grad_norm": 2.0990548133850098, + "learning_rate": 1e-06, + "loss": 0.7868, + "mean_token_accuracy": 0.7509298920631409, + "num_tokens": 289740235.0, + "step": 13293 + }, + { + "epoch": 1.0962315494351447, + "grad_norm": 2.190375804901123, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7258791923522949, + "num_tokens": 289765678.0, + "step": 13294 + }, + { + "epoch": 1.0963140100601962, + "grad_norm": 1.6676329374313354, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.7694496512413025, + "num_tokens": 289807912.0, + "step": 13295 + }, + { + "epoch": 1.0963964706852478, + "grad_norm": 1.9677990674972534, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7699075937271118, + "num_tokens": 289835885.0, + "step": 13296 + }, + { + "epoch": 1.0964789313102994, + "grad_norm": 2.1621358394622803, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7522686719894409, + "num_tokens": 289861729.0, + "step": 13297 + }, + { + "epoch": 1.096561391935351, + "grad_norm": 1.8065274953842163, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7695307731628418, + "num_tokens": 289893492.0, + "step": 13298 + }, + { + "epoch": 1.0966438525604023, + "grad_norm": 2.136174201965332, + "learning_rate": 1e-06, + "loss": 0.5872, + "mean_token_accuracy": 0.8018472790718079, + "num_tokens": 289917681.0, + "step": 13299 + }, + { + "epoch": 1.0967263131854539, + "grad_norm": 2.1029484272003174, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7659595012664795, + "num_tokens": 289944565.0, + "step": 13300 + }, + { + "epoch": 1.0968087738105055, + "grad_norm": 2.0597331523895264, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7526736259460449, + "num_tokens": 289974575.0, + "step": 13301 + }, + { + "epoch": 1.096891234435557, + "grad_norm": 2.388716697692871, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7355031967163086, + "num_tokens": 289996624.0, + "step": 13302 + }, + { + "epoch": 1.0969736950606086, + "grad_norm": 1.9675542116165161, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.7608540058135986, + "num_tokens": 290029577.0, + "step": 13303 + }, + { + "epoch": 1.0970561556856602, + "grad_norm": 1.9908250570297241, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7567622065544128, + "num_tokens": 290059346.0, + "step": 13304 + }, + { + "epoch": 1.0971386163107115, + "grad_norm": 1.9871984720230103, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7486668825149536, + "num_tokens": 290088105.0, + "step": 13305 + }, + { + "epoch": 1.097221076935763, + "grad_norm": 1.881994605064392, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7822526693344116, + "num_tokens": 290120813.0, + "step": 13306 + }, + { + "epoch": 1.0973035375608147, + "grad_norm": 2.196206569671631, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7083594799041748, + "num_tokens": 290148317.0, + "step": 13307 + }, + { + "epoch": 1.0973859981858662, + "grad_norm": 2.2245163917541504, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7272493243217468, + "num_tokens": 290174023.0, + "step": 13308 + }, + { + "epoch": 1.0974684588109178, + "grad_norm": 2.059293508529663, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7784996032714844, + "num_tokens": 290198825.0, + "step": 13309 + }, + { + "epoch": 1.0975509194359694, + "grad_norm": 2.1348421573638916, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7362426519393921, + "num_tokens": 290226027.0, + "step": 13310 + }, + { + "epoch": 1.097633380061021, + "grad_norm": 1.866131067276001, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7475276589393616, + "num_tokens": 290258545.0, + "step": 13311 + }, + { + "epoch": 1.0977158406860723, + "grad_norm": 1.9898182153701782, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7765496969223022, + "num_tokens": 290287389.0, + "step": 13312 + }, + { + "epoch": 1.0977983013111239, + "grad_norm": 1.9496022462844849, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.7541309595108032, + "num_tokens": 290316587.0, + "step": 13313 + }, + { + "epoch": 1.0978807619361755, + "grad_norm": 1.7315161228179932, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7691665291786194, + "num_tokens": 290355084.0, + "step": 13314 + }, + { + "epoch": 1.097963222561227, + "grad_norm": 1.8355016708374023, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.760165810585022, + "num_tokens": 290389418.0, + "step": 13315 + }, + { + "epoch": 1.0980456831862786, + "grad_norm": 2.1063334941864014, + "learning_rate": 1e-06, + "loss": 0.5657, + "mean_token_accuracy": 0.8060956001281738, + "num_tokens": 290413415.0, + "step": 13316 + }, + { + "epoch": 1.0981281438113302, + "grad_norm": 1.7561721801757812, + "learning_rate": 1e-06, + "loss": 0.7774, + "mean_token_accuracy": 0.7578848600387573, + "num_tokens": 290451901.0, + "step": 13317 + }, + { + "epoch": 1.0982106044363817, + "grad_norm": 2.1160504817962646, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7680724859237671, + "num_tokens": 290484536.0, + "step": 13318 + }, + { + "epoch": 1.098293065061433, + "grad_norm": 2.041498899459839, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7797843217849731, + "num_tokens": 290514145.0, + "step": 13319 + }, + { + "epoch": 1.0983755256864847, + "grad_norm": 2.110800266265869, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.7570762634277344, + "num_tokens": 290542860.0, + "step": 13320 + }, + { + "epoch": 1.0984579863115362, + "grad_norm": 2.114703416824341, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7604862451553345, + "num_tokens": 290571951.0, + "step": 13321 + }, + { + "epoch": 1.0985404469365878, + "grad_norm": 2.366213321685791, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.7840451002120972, + "num_tokens": 290592884.0, + "step": 13322 + }, + { + "epoch": 1.0986229075616394, + "grad_norm": 1.808685541152954, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7807194590568542, + "num_tokens": 290629715.0, + "step": 13323 + }, + { + "epoch": 1.098705368186691, + "grad_norm": 1.9661142826080322, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7408583164215088, + "num_tokens": 290662858.0, + "step": 13324 + }, + { + "epoch": 1.0987878288117423, + "grad_norm": 2.195039987564087, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7470911145210266, + "num_tokens": 290690103.0, + "step": 13325 + }, + { + "epoch": 1.0988702894367939, + "grad_norm": 2.357508897781372, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7484565377235413, + "num_tokens": 290712370.0, + "step": 13326 + }, + { + "epoch": 1.0989527500618455, + "grad_norm": 2.0170140266418457, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7130958437919617, + "num_tokens": 290743909.0, + "step": 13327 + }, + { + "epoch": 1.099035210686897, + "grad_norm": 1.7100056409835815, + "learning_rate": 1e-06, + "loss": 0.5884, + "mean_token_accuracy": 0.8069673776626587, + "num_tokens": 290777691.0, + "step": 13328 + }, + { + "epoch": 1.0991176713119486, + "grad_norm": 2.1613800525665283, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7657449841499329, + "num_tokens": 290803101.0, + "step": 13329 + }, + { + "epoch": 1.0992001319370002, + "grad_norm": 2.334700584411621, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7531819343566895, + "num_tokens": 290826429.0, + "step": 13330 + }, + { + "epoch": 1.0992825925620515, + "grad_norm": 1.894973635673523, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.763116717338562, + "num_tokens": 290856835.0, + "step": 13331 + }, + { + "epoch": 1.099365053187103, + "grad_norm": 1.9551819562911987, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7760439515113831, + "num_tokens": 290889271.0, + "step": 13332 + }, + { + "epoch": 1.0994475138121547, + "grad_norm": 1.812126636505127, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7551652789115906, + "num_tokens": 290926196.0, + "step": 13333 + }, + { + "epoch": 1.0995299744372062, + "grad_norm": 2.2345166206359863, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.7935974597930908, + "num_tokens": 290951734.0, + "step": 13334 + }, + { + "epoch": 1.0996124350622578, + "grad_norm": 1.8452198505401611, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7574031352996826, + "num_tokens": 290988381.0, + "step": 13335 + }, + { + "epoch": 1.0996948956873094, + "grad_norm": 2.0367987155914307, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7229362726211548, + "num_tokens": 291020184.0, + "step": 13336 + }, + { + "epoch": 1.099777356312361, + "grad_norm": 1.9702913761138916, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7674652934074402, + "num_tokens": 291048060.0, + "step": 13337 + }, + { + "epoch": 1.0998598169374123, + "grad_norm": 1.9888712167739868, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7746903300285339, + "num_tokens": 291075794.0, + "step": 13338 + }, + { + "epoch": 1.0999422775624639, + "grad_norm": 2.035778760910034, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.747715950012207, + "num_tokens": 291105126.0, + "step": 13339 + }, + { + "epoch": 1.1000247381875154, + "grad_norm": 1.9410321712493896, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7220169901847839, + "num_tokens": 291140631.0, + "step": 13340 + }, + { + "epoch": 1.100107198812567, + "grad_norm": 2.182563543319702, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7128931283950806, + "num_tokens": 291171413.0, + "step": 13341 + }, + { + "epoch": 1.1001896594376186, + "grad_norm": 2.1782281398773193, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7713871002197266, + "num_tokens": 291200346.0, + "step": 13342 + }, + { + "epoch": 1.1002721200626702, + "grad_norm": 2.16467022895813, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7455451488494873, + "num_tokens": 291226044.0, + "step": 13343 + }, + { + "epoch": 1.1003545806877215, + "grad_norm": 2.335505723953247, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7080502510070801, + "num_tokens": 291251601.0, + "step": 13344 + }, + { + "epoch": 1.100437041312773, + "grad_norm": 2.006269693374634, + "learning_rate": 1e-06, + "loss": 0.56, + "mean_token_accuracy": 0.8118035793304443, + "num_tokens": 291279820.0, + "step": 13345 + }, + { + "epoch": 1.1005195019378247, + "grad_norm": 2.158586263656616, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7745423913002014, + "num_tokens": 291307971.0, + "step": 13346 + }, + { + "epoch": 1.1006019625628762, + "grad_norm": 2.2497775554656982, + "learning_rate": 1e-06, + "loss": 0.7617, + "mean_token_accuracy": 0.7468512058258057, + "num_tokens": 291332280.0, + "step": 13347 + }, + { + "epoch": 1.1006844231879278, + "grad_norm": 2.0582211017608643, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7584094405174255, + "num_tokens": 291364869.0, + "step": 13348 + }, + { + "epoch": 1.1007668838129794, + "grad_norm": 2.0799031257629395, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7819803953170776, + "num_tokens": 291391854.0, + "step": 13349 + }, + { + "epoch": 1.1008493444380307, + "grad_norm": 2.1745686531066895, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7531560659408569, + "num_tokens": 291417716.0, + "step": 13350 + }, + { + "epoch": 1.1009318050630823, + "grad_norm": 2.3597779273986816, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7748392820358276, + "num_tokens": 291446619.0, + "step": 13351 + }, + { + "epoch": 1.1010142656881339, + "grad_norm": 2.4851737022399902, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7354283332824707, + "num_tokens": 291468782.0, + "step": 13352 + }, + { + "epoch": 1.1010967263131854, + "grad_norm": 2.0688257217407227, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7232958078384399, + "num_tokens": 291498838.0, + "step": 13353 + }, + { + "epoch": 1.101179186938237, + "grad_norm": 1.773197889328003, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7681093215942383, + "num_tokens": 291534821.0, + "step": 13354 + }, + { + "epoch": 1.1012616475632886, + "grad_norm": 2.100442886352539, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.732541024684906, + "num_tokens": 291564280.0, + "step": 13355 + }, + { + "epoch": 1.1013441081883402, + "grad_norm": 2.015418767929077, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7664831876754761, + "num_tokens": 291592815.0, + "step": 13356 + }, + { + "epoch": 1.1014265688133915, + "grad_norm": 2.3460044860839844, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7291480302810669, + "num_tokens": 291618422.0, + "step": 13357 + }, + { + "epoch": 1.101509029438443, + "grad_norm": 2.301429510116577, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7718987464904785, + "num_tokens": 291642937.0, + "step": 13358 + }, + { + "epoch": 1.1015914900634947, + "grad_norm": 1.818454384803772, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.7939803600311279, + "num_tokens": 291679046.0, + "step": 13359 + }, + { + "epoch": 1.1016739506885462, + "grad_norm": 2.044083595275879, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.746234655380249, + "num_tokens": 291709650.0, + "step": 13360 + }, + { + "epoch": 1.1017564113135978, + "grad_norm": 2.015713691711426, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7729855179786682, + "num_tokens": 291739958.0, + "step": 13361 + }, + { + "epoch": 1.1018388719386494, + "grad_norm": 2.2066221237182617, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7245945930480957, + "num_tokens": 291766807.0, + "step": 13362 + }, + { + "epoch": 1.101921332563701, + "grad_norm": 1.9309672117233276, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.747435450553894, + "num_tokens": 291799057.0, + "step": 13363 + }, + { + "epoch": 1.1020037931887523, + "grad_norm": 1.778139591217041, + "learning_rate": 1e-06, + "loss": 0.5698, + "mean_token_accuracy": 0.804772138595581, + "num_tokens": 291831254.0, + "step": 13364 + }, + { + "epoch": 1.1020862538138039, + "grad_norm": 1.8913705348968506, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7723445296287537, + "num_tokens": 291864190.0, + "step": 13365 + }, + { + "epoch": 1.1021687144388554, + "grad_norm": 2.013566493988037, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7701781988143921, + "num_tokens": 291893181.0, + "step": 13366 + }, + { + "epoch": 1.102251175063907, + "grad_norm": 2.178281545639038, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7321300506591797, + "num_tokens": 291920146.0, + "step": 13367 + }, + { + "epoch": 1.1023336356889586, + "grad_norm": 1.8042325973510742, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7766170501708984, + "num_tokens": 291953424.0, + "step": 13368 + }, + { + "epoch": 1.1024160963140102, + "grad_norm": 1.7908527851104736, + "learning_rate": 1e-06, + "loss": 0.5949, + "mean_token_accuracy": 0.8004706501960754, + "num_tokens": 291986479.0, + "step": 13369 + }, + { + "epoch": 1.1024985569390615, + "grad_norm": 2.048468589782715, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7397953867912292, + "num_tokens": 292015809.0, + "step": 13370 + }, + { + "epoch": 1.102581017564113, + "grad_norm": 2.0392351150512695, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7484578490257263, + "num_tokens": 292047269.0, + "step": 13371 + }, + { + "epoch": 1.1026634781891647, + "grad_norm": 2.119731903076172, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7122650742530823, + "num_tokens": 292076494.0, + "step": 13372 + }, + { + "epoch": 1.1027459388142162, + "grad_norm": 1.8172943592071533, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.7977533340454102, + "num_tokens": 292108668.0, + "step": 13373 + }, + { + "epoch": 1.1028283994392678, + "grad_norm": 1.9452320337295532, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7550970911979675, + "num_tokens": 292138475.0, + "step": 13374 + }, + { + "epoch": 1.1029108600643194, + "grad_norm": 2.3034117221832275, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.693729043006897, + "num_tokens": 292163808.0, + "step": 13375 + }, + { + "epoch": 1.1029933206893707, + "grad_norm": 1.6161174774169922, + "learning_rate": 1e-06, + "loss": 0.5485, + "mean_token_accuracy": 0.8317257761955261, + "num_tokens": 292203306.0, + "step": 13376 + }, + { + "epoch": 1.1030757813144223, + "grad_norm": 1.9831868410110474, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7761930227279663, + "num_tokens": 292231831.0, + "step": 13377 + }, + { + "epoch": 1.1031582419394739, + "grad_norm": 1.893681526184082, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7568050026893616, + "num_tokens": 292265204.0, + "step": 13378 + }, + { + "epoch": 1.1032407025645254, + "grad_norm": 2.009544849395752, + "learning_rate": 1e-06, + "loss": 0.5723, + "mean_token_accuracy": 0.8044965863227844, + "num_tokens": 292294005.0, + "step": 13379 + }, + { + "epoch": 1.103323163189577, + "grad_norm": 1.8619433641433716, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7640399932861328, + "num_tokens": 292327494.0, + "step": 13380 + }, + { + "epoch": 1.1034056238146286, + "grad_norm": 2.0084116458892822, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.7866042852401733, + "num_tokens": 292357972.0, + "step": 13381 + }, + { + "epoch": 1.1034880844396802, + "grad_norm": 1.9194855690002441, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7784693241119385, + "num_tokens": 292388543.0, + "step": 13382 + }, + { + "epoch": 1.1035705450647315, + "grad_norm": 1.878016471862793, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7695688009262085, + "num_tokens": 292422810.0, + "step": 13383 + }, + { + "epoch": 1.103653005689783, + "grad_norm": 1.8952252864837646, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7757806777954102, + "num_tokens": 292451895.0, + "step": 13384 + }, + { + "epoch": 1.1037354663148347, + "grad_norm": 2.206050395965576, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7194181084632874, + "num_tokens": 292479350.0, + "step": 13385 + }, + { + "epoch": 1.1038179269398862, + "grad_norm": 2.0412378311157227, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7509475946426392, + "num_tokens": 292506730.0, + "step": 13386 + }, + { + "epoch": 1.1039003875649378, + "grad_norm": 2.050419569015503, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7517108917236328, + "num_tokens": 292536646.0, + "step": 13387 + }, + { + "epoch": 1.1039828481899894, + "grad_norm": 2.2311739921569824, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7316094636917114, + "num_tokens": 292561425.0, + "step": 13388 + }, + { + "epoch": 1.1040653088150407, + "grad_norm": 2.007441997528076, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7845272421836853, + "num_tokens": 292587712.0, + "step": 13389 + }, + { + "epoch": 1.1041477694400923, + "grad_norm": 1.7912670373916626, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7452658414840698, + "num_tokens": 292624652.0, + "step": 13390 + }, + { + "epoch": 1.1042302300651439, + "grad_norm": 2.1275603771209717, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7183305621147156, + "num_tokens": 292653419.0, + "step": 13391 + }, + { + "epoch": 1.1043126906901954, + "grad_norm": 1.9816560745239258, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7416547536849976, + "num_tokens": 292687125.0, + "step": 13392 + }, + { + "epoch": 1.104395151315247, + "grad_norm": 2.349210500717163, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7472320795059204, + "num_tokens": 292710163.0, + "step": 13393 + }, + { + "epoch": 1.1044776119402986, + "grad_norm": 1.9038361310958862, + "learning_rate": 1e-06, + "loss": 0.5465, + "mean_token_accuracy": 0.8172852993011475, + "num_tokens": 292739062.0, + "step": 13394 + }, + { + "epoch": 1.10456007256535, + "grad_norm": 2.046079397201538, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7587426900863647, + "num_tokens": 292766061.0, + "step": 13395 + }, + { + "epoch": 1.1046425331904015, + "grad_norm": 2.099438190460205, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.6982055902481079, + "num_tokens": 292795166.0, + "step": 13396 + }, + { + "epoch": 1.104724993815453, + "grad_norm": 2.0128204822540283, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.79400235414505, + "num_tokens": 292823916.0, + "step": 13397 + }, + { + "epoch": 1.1048074544405047, + "grad_norm": 1.8814706802368164, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7735459208488464, + "num_tokens": 292860122.0, + "step": 13398 + }, + { + "epoch": 1.1048899150655562, + "grad_norm": 1.9065570831298828, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.765539824962616, + "num_tokens": 292892814.0, + "step": 13399 + }, + { + "epoch": 1.1049723756906078, + "grad_norm": 2.334688186645508, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7559762001037598, + "num_tokens": 292914133.0, + "step": 13400 + }, + { + "epoch": 1.1050548363156594, + "grad_norm": 2.1447536945343018, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7322103381156921, + "num_tokens": 292941613.0, + "step": 13401 + }, + { + "epoch": 1.1051372969407107, + "grad_norm": 1.961786150932312, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7717800736427307, + "num_tokens": 292971001.0, + "step": 13402 + }, + { + "epoch": 1.1052197575657623, + "grad_norm": 2.245396852493286, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7518749237060547, + "num_tokens": 292995339.0, + "step": 13403 + }, + { + "epoch": 1.1053022181908139, + "grad_norm": 1.8847259283065796, + "learning_rate": 1e-06, + "loss": 0.5727, + "mean_token_accuracy": 0.805972158908844, + "num_tokens": 293023339.0, + "step": 13404 + }, + { + "epoch": 1.1053846788158654, + "grad_norm": 2.0451738834381104, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7585849165916443, + "num_tokens": 293051293.0, + "step": 13405 + }, + { + "epoch": 1.105467139440917, + "grad_norm": 1.7783925533294678, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.8026913404464722, + "num_tokens": 293084016.0, + "step": 13406 + }, + { + "epoch": 1.1055496000659686, + "grad_norm": 2.039161443710327, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.7882341742515564, + "num_tokens": 293116058.0, + "step": 13407 + }, + { + "epoch": 1.10563206069102, + "grad_norm": 1.812275767326355, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7864779829978943, + "num_tokens": 293150030.0, + "step": 13408 + }, + { + "epoch": 1.1057145213160715, + "grad_norm": 2.0554378032684326, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.7889584302902222, + "num_tokens": 293176990.0, + "step": 13409 + }, + { + "epoch": 1.105796981941123, + "grad_norm": 2.0993361473083496, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7396520376205444, + "num_tokens": 293204099.0, + "step": 13410 + }, + { + "epoch": 1.1058794425661747, + "grad_norm": 2.085293769836426, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7599402070045471, + "num_tokens": 293232705.0, + "step": 13411 + }, + { + "epoch": 1.1059619031912262, + "grad_norm": 2.422222852706909, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7622867822647095, + "num_tokens": 293255116.0, + "step": 13412 + }, + { + "epoch": 1.1060443638162778, + "grad_norm": 2.2629613876342773, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7621563076972961, + "num_tokens": 293276861.0, + "step": 13413 + }, + { + "epoch": 1.1061268244413291, + "grad_norm": 1.8381942510604858, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.790673017501831, + "num_tokens": 293309356.0, + "step": 13414 + }, + { + "epoch": 1.1062092850663807, + "grad_norm": 2.0026979446411133, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7451826333999634, + "num_tokens": 293339849.0, + "step": 13415 + }, + { + "epoch": 1.1062917456914323, + "grad_norm": 1.7926281690597534, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.8007463812828064, + "num_tokens": 293372635.0, + "step": 13416 + }, + { + "epoch": 1.1063742063164839, + "grad_norm": 1.8755674362182617, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7413245439529419, + "num_tokens": 293408156.0, + "step": 13417 + }, + { + "epoch": 1.1064566669415354, + "grad_norm": 1.7167352437973022, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.7866815328598022, + "num_tokens": 293444181.0, + "step": 13418 + }, + { + "epoch": 1.106539127566587, + "grad_norm": 1.737677812576294, + "learning_rate": 1e-06, + "loss": 0.5207, + "mean_token_accuracy": 0.8215188980102539, + "num_tokens": 293479659.0, + "step": 13419 + }, + { + "epoch": 1.1066215881916386, + "grad_norm": 2.1375043392181396, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.7635214924812317, + "num_tokens": 293508365.0, + "step": 13420 + }, + { + "epoch": 1.10670404881669, + "grad_norm": 1.9872872829437256, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7215896844863892, + "num_tokens": 293541309.0, + "step": 13421 + }, + { + "epoch": 1.1067865094417415, + "grad_norm": 2.42563796043396, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7416438460350037, + "num_tokens": 293563071.0, + "step": 13422 + }, + { + "epoch": 1.106868970066793, + "grad_norm": 2.166196584701538, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7057350873947144, + "num_tokens": 293591325.0, + "step": 13423 + }, + { + "epoch": 1.1069514306918447, + "grad_norm": 2.328646421432495, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.766879677772522, + "num_tokens": 293613636.0, + "step": 13424 + }, + { + "epoch": 1.1070338913168962, + "grad_norm": 1.9143115282058716, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7675600051879883, + "num_tokens": 293645513.0, + "step": 13425 + }, + { + "epoch": 1.1071163519419478, + "grad_norm": 2.2636094093322754, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.753732442855835, + "num_tokens": 293670560.0, + "step": 13426 + }, + { + "epoch": 1.1071988125669994, + "grad_norm": 2.22786808013916, + "learning_rate": 1e-06, + "loss": 0.5704, + "mean_token_accuracy": 0.8124653100967407, + "num_tokens": 293693656.0, + "step": 13427 + }, + { + "epoch": 1.1072812731920507, + "grad_norm": 2.009605646133423, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7189958095550537, + "num_tokens": 293724602.0, + "step": 13428 + }, + { + "epoch": 1.1073637338171023, + "grad_norm": 1.933398962020874, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7831703424453735, + "num_tokens": 293754381.0, + "step": 13429 + }, + { + "epoch": 1.1074461944421539, + "grad_norm": 2.064351797103882, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7562077045440674, + "num_tokens": 293783244.0, + "step": 13430 + }, + { + "epoch": 1.1075286550672054, + "grad_norm": 1.865890622138977, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.744840145111084, + "num_tokens": 293817643.0, + "step": 13431 + }, + { + "epoch": 1.107611115692257, + "grad_norm": 1.8558053970336914, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.7841915488243103, + "num_tokens": 293849657.0, + "step": 13432 + }, + { + "epoch": 1.1076935763173086, + "grad_norm": 2.076113700866699, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.780829906463623, + "num_tokens": 293876318.0, + "step": 13433 + }, + { + "epoch": 1.10777603694236, + "grad_norm": 1.9859426021575928, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7482468485832214, + "num_tokens": 293907673.0, + "step": 13434 + }, + { + "epoch": 1.1078584975674115, + "grad_norm": 2.045102119445801, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7782095074653625, + "num_tokens": 293937552.0, + "step": 13435 + }, + { + "epoch": 1.107940958192463, + "grad_norm": 2.277026414871216, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.6979559659957886, + "num_tokens": 293966152.0, + "step": 13436 + }, + { + "epoch": 1.1080234188175146, + "grad_norm": 1.608678936958313, + "learning_rate": 1e-06, + "loss": 0.543, + "mean_token_accuracy": 0.8177357912063599, + "num_tokens": 294003219.0, + "step": 13437 + }, + { + "epoch": 1.1081058794425662, + "grad_norm": 2.0201635360717773, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7559417486190796, + "num_tokens": 294035294.0, + "step": 13438 + }, + { + "epoch": 1.1081883400676178, + "grad_norm": 1.855088233947754, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7427348494529724, + "num_tokens": 294072021.0, + "step": 13439 + }, + { + "epoch": 1.1082708006926691, + "grad_norm": 1.674192190170288, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7622876763343811, + "num_tokens": 294110620.0, + "step": 13440 + }, + { + "epoch": 1.1083532613177207, + "grad_norm": 2.3278298377990723, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7608158588409424, + "num_tokens": 294133986.0, + "step": 13441 + }, + { + "epoch": 1.1084357219427723, + "grad_norm": 2.100939989089966, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7786844372749329, + "num_tokens": 294161763.0, + "step": 13442 + }, + { + "epoch": 1.1085181825678239, + "grad_norm": 2.003453254699707, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7329716682434082, + "num_tokens": 294190022.0, + "step": 13443 + }, + { + "epoch": 1.1086006431928754, + "grad_norm": 1.9593634605407715, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.7840995788574219, + "num_tokens": 294222255.0, + "step": 13444 + }, + { + "epoch": 1.108683103817927, + "grad_norm": 2.1346073150634766, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7548450231552124, + "num_tokens": 294253692.0, + "step": 13445 + }, + { + "epoch": 1.1087655644429786, + "grad_norm": 2.3274171352386475, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7262586951255798, + "num_tokens": 294277357.0, + "step": 13446 + }, + { + "epoch": 1.10884802506803, + "grad_norm": 2.2457962036132812, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.748866081237793, + "num_tokens": 294303250.0, + "step": 13447 + }, + { + "epoch": 1.1089304856930815, + "grad_norm": 1.9317626953125, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7286345958709717, + "num_tokens": 294334446.0, + "step": 13448 + }, + { + "epoch": 1.109012946318133, + "grad_norm": 1.8510451316833496, + "learning_rate": 1e-06, + "loss": 0.5658, + "mean_token_accuracy": 0.8108777403831482, + "num_tokens": 294364858.0, + "step": 13449 + }, + { + "epoch": 1.1090954069431846, + "grad_norm": 1.9237732887268066, + "learning_rate": 1e-06, + "loss": 0.5848, + "mean_token_accuracy": 0.7835984230041504, + "num_tokens": 294392648.0, + "step": 13450 + }, + { + "epoch": 1.1091778675682362, + "grad_norm": 2.098869800567627, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7585250735282898, + "num_tokens": 294418936.0, + "step": 13451 + }, + { + "epoch": 1.1092603281932878, + "grad_norm": 1.9689931869506836, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7686859369277954, + "num_tokens": 294451227.0, + "step": 13452 + }, + { + "epoch": 1.1093427888183391, + "grad_norm": 2.096255302429199, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7640196084976196, + "num_tokens": 294478863.0, + "step": 13453 + }, + { + "epoch": 1.1094252494433907, + "grad_norm": 1.9066368341445923, + "learning_rate": 1e-06, + "loss": 0.5719, + "mean_token_accuracy": 0.7955467700958252, + "num_tokens": 294508653.0, + "step": 13454 + }, + { + "epoch": 1.1095077100684423, + "grad_norm": 2.0778326988220215, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7755453586578369, + "num_tokens": 294534859.0, + "step": 13455 + }, + { + "epoch": 1.1095901706934939, + "grad_norm": 1.9162918329238892, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7461743354797363, + "num_tokens": 294564835.0, + "step": 13456 + }, + { + "epoch": 1.1096726313185454, + "grad_norm": 1.7529511451721191, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7541325092315674, + "num_tokens": 294602943.0, + "step": 13457 + }, + { + "epoch": 1.109755091943597, + "grad_norm": 1.7360137701034546, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7891761064529419, + "num_tokens": 294641528.0, + "step": 13458 + }, + { + "epoch": 1.1098375525686484, + "grad_norm": 2.1504507064819336, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7616586685180664, + "num_tokens": 294671737.0, + "step": 13459 + }, + { + "epoch": 1.1099200131937, + "grad_norm": 2.113034248352051, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7697975039482117, + "num_tokens": 294699049.0, + "step": 13460 + }, + { + "epoch": 1.1100024738187515, + "grad_norm": 1.8248122930526733, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7764806747436523, + "num_tokens": 294731758.0, + "step": 13461 + }, + { + "epoch": 1.110084934443803, + "grad_norm": 2.019179105758667, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.7526895999908447, + "num_tokens": 294761397.0, + "step": 13462 + }, + { + "epoch": 1.1101673950688546, + "grad_norm": 1.9026293754577637, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7458281517028809, + "num_tokens": 294794084.0, + "step": 13463 + }, + { + "epoch": 1.1102498556939062, + "grad_norm": 1.852805256843567, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7530702948570251, + "num_tokens": 294829593.0, + "step": 13464 + }, + { + "epoch": 1.1103323163189578, + "grad_norm": 2.083249092102051, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7576693296432495, + "num_tokens": 294858810.0, + "step": 13465 + }, + { + "epoch": 1.1104147769440091, + "grad_norm": 1.9963246583938599, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7656010389328003, + "num_tokens": 294890131.0, + "step": 13466 + }, + { + "epoch": 1.1104972375690607, + "grad_norm": 2.055962085723877, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7094929218292236, + "num_tokens": 294921197.0, + "step": 13467 + }, + { + "epoch": 1.1105796981941123, + "grad_norm": 2.0954296588897705, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7417483925819397, + "num_tokens": 294951990.0, + "step": 13468 + }, + { + "epoch": 1.1106621588191639, + "grad_norm": 1.8834514617919922, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7576429843902588, + "num_tokens": 294986854.0, + "step": 13469 + }, + { + "epoch": 1.1107446194442154, + "grad_norm": 2.056291103363037, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7647980451583862, + "num_tokens": 295015783.0, + "step": 13470 + }, + { + "epoch": 1.110827080069267, + "grad_norm": 2.0835940837860107, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7762842774391174, + "num_tokens": 295039617.0, + "step": 13471 + }, + { + "epoch": 1.1109095406943186, + "grad_norm": 1.7504464387893677, + "learning_rate": 1e-06, + "loss": 0.5707, + "mean_token_accuracy": 0.8080676198005676, + "num_tokens": 295072823.0, + "step": 13472 + }, + { + "epoch": 1.11099200131937, + "grad_norm": 1.8083295822143555, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.750626266002655, + "num_tokens": 295108893.0, + "step": 13473 + }, + { + "epoch": 1.1110744619444215, + "grad_norm": 2.316784381866455, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.740134060382843, + "num_tokens": 295135882.0, + "step": 13474 + }, + { + "epoch": 1.111156922569473, + "grad_norm": 1.8180382251739502, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7523549795150757, + "num_tokens": 295172240.0, + "step": 13475 + }, + { + "epoch": 1.1112393831945246, + "grad_norm": 2.0853471755981445, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7601001262664795, + "num_tokens": 295200070.0, + "step": 13476 + }, + { + "epoch": 1.1113218438195762, + "grad_norm": 1.9510831832885742, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.7593523859977722, + "num_tokens": 295231609.0, + "step": 13477 + }, + { + "epoch": 1.1114043044446278, + "grad_norm": 1.931945562362671, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.764007568359375, + "num_tokens": 295263857.0, + "step": 13478 + }, + { + "epoch": 1.1114867650696791, + "grad_norm": 2.0006213188171387, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7617328763008118, + "num_tokens": 295292957.0, + "step": 13479 + }, + { + "epoch": 1.1115692256947307, + "grad_norm": 1.959458351135254, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7708448171615601, + "num_tokens": 295324643.0, + "step": 13480 + }, + { + "epoch": 1.1116516863197823, + "grad_norm": 1.7534945011138916, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.784363865852356, + "num_tokens": 295360046.0, + "step": 13481 + }, + { + "epoch": 1.1117341469448339, + "grad_norm": 2.0286865234375, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7598946690559387, + "num_tokens": 295385983.0, + "step": 13482 + }, + { + "epoch": 1.1118166075698854, + "grad_norm": 1.846785068511963, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7573099136352539, + "num_tokens": 295420821.0, + "step": 13483 + }, + { + "epoch": 1.111899068194937, + "grad_norm": 2.1034996509552, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7733473777770996, + "num_tokens": 295449576.0, + "step": 13484 + }, + { + "epoch": 1.1119815288199884, + "grad_norm": 2.1187329292297363, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7349150776863098, + "num_tokens": 295476379.0, + "step": 13485 + }, + { + "epoch": 1.11206398944504, + "grad_norm": 1.7332284450531006, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7672425508499146, + "num_tokens": 295514990.0, + "step": 13486 + }, + { + "epoch": 1.1121464500700915, + "grad_norm": 2.0370209217071533, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7545654773712158, + "num_tokens": 295542665.0, + "step": 13487 + }, + { + "epoch": 1.112228910695143, + "grad_norm": 2.2132809162139893, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7298229336738586, + "num_tokens": 295568101.0, + "step": 13488 + }, + { + "epoch": 1.1123113713201946, + "grad_norm": 2.1992909908294678, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7573907375335693, + "num_tokens": 295590375.0, + "step": 13489 + }, + { + "epoch": 1.1123938319452462, + "grad_norm": 2.1602203845977783, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7197881937026978, + "num_tokens": 295617990.0, + "step": 13490 + }, + { + "epoch": 1.1124762925702978, + "grad_norm": 1.9902875423431396, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.762199342250824, + "num_tokens": 295649293.0, + "step": 13491 + }, + { + "epoch": 1.1125587531953491, + "grad_norm": 1.844146966934204, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7748788595199585, + "num_tokens": 295682249.0, + "step": 13492 + }, + { + "epoch": 1.1126412138204007, + "grad_norm": 1.9739362001419067, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7321344614028931, + "num_tokens": 295716064.0, + "step": 13493 + }, + { + "epoch": 1.1127236744454523, + "grad_norm": 1.7800308465957642, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7390585541725159, + "num_tokens": 295757726.0, + "step": 13494 + }, + { + "epoch": 1.1128061350705039, + "grad_norm": 1.8159167766571045, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.795508861541748, + "num_tokens": 295791097.0, + "step": 13495 + }, + { + "epoch": 1.1128885956955554, + "grad_norm": 1.9148072004318237, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7778401374816895, + "num_tokens": 295824963.0, + "step": 13496 + }, + { + "epoch": 1.112971056320607, + "grad_norm": 2.034043073654175, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.77657151222229, + "num_tokens": 295851667.0, + "step": 13497 + }, + { + "epoch": 1.1130535169456583, + "grad_norm": 1.9519742727279663, + "learning_rate": 1e-06, + "loss": 0.7722, + "mean_token_accuracy": 0.749762773513794, + "num_tokens": 295885042.0, + "step": 13498 + }, + { + "epoch": 1.11313597757071, + "grad_norm": 1.7472957372665405, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7802834510803223, + "num_tokens": 295921086.0, + "step": 13499 + }, + { + "epoch": 1.1132184381957615, + "grad_norm": 1.985689640045166, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7645865678787231, + "num_tokens": 295954442.0, + "step": 13500 + }, + { + "epoch": 1.113300898820813, + "grad_norm": 1.808204174041748, + "learning_rate": 1e-06, + "loss": 0.545, + "mean_token_accuracy": 0.796709418296814, + "num_tokens": 295987041.0, + "step": 13501 + }, + { + "epoch": 1.1133833594458646, + "grad_norm": 2.190598249435425, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.73931884765625, + "num_tokens": 296015461.0, + "step": 13502 + }, + { + "epoch": 1.1134658200709162, + "grad_norm": 2.0329487323760986, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7612985372543335, + "num_tokens": 296041938.0, + "step": 13503 + }, + { + "epoch": 1.1135482806959676, + "grad_norm": 2.203087568283081, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7275264263153076, + "num_tokens": 296071916.0, + "step": 13504 + }, + { + "epoch": 1.1136307413210191, + "grad_norm": 2.418631076812744, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7245571613311768, + "num_tokens": 296094753.0, + "step": 13505 + }, + { + "epoch": 1.1137132019460707, + "grad_norm": 1.8910149335861206, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7436355948448181, + "num_tokens": 296128308.0, + "step": 13506 + }, + { + "epoch": 1.1137956625711223, + "grad_norm": 2.1016032695770264, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.7868210077285767, + "num_tokens": 296155066.0, + "step": 13507 + }, + { + "epoch": 1.1138781231961739, + "grad_norm": 2.1253838539123535, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7690723538398743, + "num_tokens": 296178555.0, + "step": 13508 + }, + { + "epoch": 1.1139605838212254, + "grad_norm": 2.1733787059783936, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7691705226898193, + "num_tokens": 296203659.0, + "step": 13509 + }, + { + "epoch": 1.114043044446277, + "grad_norm": 1.7298250198364258, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.786786675453186, + "num_tokens": 296239258.0, + "step": 13510 + }, + { + "epoch": 1.1141255050713283, + "grad_norm": 2.2520713806152344, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7141361236572266, + "num_tokens": 296265791.0, + "step": 13511 + }, + { + "epoch": 1.11420796569638, + "grad_norm": 1.7616058588027954, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.748341977596283, + "num_tokens": 296303198.0, + "step": 13512 + }, + { + "epoch": 1.1142904263214315, + "grad_norm": 2.0397450923919678, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7490246295928955, + "num_tokens": 296331703.0, + "step": 13513 + }, + { + "epoch": 1.114372886946483, + "grad_norm": 2.047686815261841, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7736458778381348, + "num_tokens": 296357441.0, + "step": 13514 + }, + { + "epoch": 1.1144553475715346, + "grad_norm": 1.8663618564605713, + "learning_rate": 1e-06, + "loss": 0.5958, + "mean_token_accuracy": 0.8098772764205933, + "num_tokens": 296389897.0, + "step": 13515 + }, + { + "epoch": 1.1145378081965862, + "grad_norm": 1.738712191581726, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.7952574491500854, + "num_tokens": 296425712.0, + "step": 13516 + }, + { + "epoch": 1.1146202688216378, + "grad_norm": 2.3921077251434326, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7307860851287842, + "num_tokens": 296451003.0, + "step": 13517 + }, + { + "epoch": 1.1147027294466891, + "grad_norm": 2.0939531326293945, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7660195231437683, + "num_tokens": 296478873.0, + "step": 13518 + }, + { + "epoch": 1.1147851900717407, + "grad_norm": 1.8528883457183838, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7551583051681519, + "num_tokens": 296512205.0, + "step": 13519 + }, + { + "epoch": 1.1148676506967923, + "grad_norm": 2.015286684036255, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7628380060195923, + "num_tokens": 296546937.0, + "step": 13520 + }, + { + "epoch": 1.1149501113218439, + "grad_norm": 2.072026491165161, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7529516816139221, + "num_tokens": 296576910.0, + "step": 13521 + }, + { + "epoch": 1.1150325719468954, + "grad_norm": 2.0070409774780273, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7465510368347168, + "num_tokens": 296606750.0, + "step": 13522 + }, + { + "epoch": 1.115115032571947, + "grad_norm": 2.1541872024536133, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7391040921211243, + "num_tokens": 296631761.0, + "step": 13523 + }, + { + "epoch": 1.1151974931969983, + "grad_norm": 1.976962924003601, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7722439169883728, + "num_tokens": 296657985.0, + "step": 13524 + }, + { + "epoch": 1.11527995382205, + "grad_norm": 2.1648902893066406, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7276292443275452, + "num_tokens": 296686521.0, + "step": 13525 + }, + { + "epoch": 1.1153624144471015, + "grad_norm": 2.3804121017456055, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7360965013504028, + "num_tokens": 296710226.0, + "step": 13526 + }, + { + "epoch": 1.115444875072153, + "grad_norm": 2.024799108505249, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.758556604385376, + "num_tokens": 296739278.0, + "step": 13527 + }, + { + "epoch": 1.1155273356972046, + "grad_norm": 1.8498085737228394, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.763740062713623, + "num_tokens": 296772232.0, + "step": 13528 + }, + { + "epoch": 1.1156097963222562, + "grad_norm": 1.940724492073059, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7724933624267578, + "num_tokens": 296801526.0, + "step": 13529 + }, + { + "epoch": 1.1156922569473076, + "grad_norm": 2.159714937210083, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7685941457748413, + "num_tokens": 296826689.0, + "step": 13530 + }, + { + "epoch": 1.1157747175723591, + "grad_norm": 1.9452275037765503, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7808213829994202, + "num_tokens": 296856571.0, + "step": 13531 + }, + { + "epoch": 1.1158571781974107, + "grad_norm": 2.227816343307495, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7374666333198547, + "num_tokens": 296883532.0, + "step": 13532 + }, + { + "epoch": 1.1159396388224623, + "grad_norm": 2.177607774734497, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.7896282076835632, + "num_tokens": 296907710.0, + "step": 13533 + }, + { + "epoch": 1.1160220994475138, + "grad_norm": 1.9260470867156982, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7513746619224548, + "num_tokens": 296939504.0, + "step": 13534 + }, + { + "epoch": 1.1161045600725654, + "grad_norm": 2.0526816844940186, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7414522171020508, + "num_tokens": 296970650.0, + "step": 13535 + }, + { + "epoch": 1.116187020697617, + "grad_norm": 1.8729467391967773, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7757977247238159, + "num_tokens": 297001958.0, + "step": 13536 + }, + { + "epoch": 1.1162694813226683, + "grad_norm": 2.2365925312042236, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7389293909072876, + "num_tokens": 297027469.0, + "step": 13537 + }, + { + "epoch": 1.11635194194772, + "grad_norm": 2.1940369606018066, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7814487218856812, + "num_tokens": 297053325.0, + "step": 13538 + }, + { + "epoch": 1.1164344025727715, + "grad_norm": 2.0178847312927246, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7215733528137207, + "num_tokens": 297084040.0, + "step": 13539 + }, + { + "epoch": 1.116516863197823, + "grad_norm": 1.8725926876068115, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.791193425655365, + "num_tokens": 297113207.0, + "step": 13540 + }, + { + "epoch": 1.1165993238228746, + "grad_norm": 1.9221433401107788, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7889345288276672, + "num_tokens": 297143679.0, + "step": 13541 + }, + { + "epoch": 1.1166817844479262, + "grad_norm": 2.210458517074585, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7810285091400146, + "num_tokens": 297167626.0, + "step": 13542 + }, + { + "epoch": 1.1167642450729776, + "grad_norm": 1.924394130706787, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7728671431541443, + "num_tokens": 297200304.0, + "step": 13543 + }, + { + "epoch": 1.1168467056980291, + "grad_norm": 2.0102946758270264, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7395684719085693, + "num_tokens": 297230626.0, + "step": 13544 + }, + { + "epoch": 1.1169291663230807, + "grad_norm": 1.811800241470337, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7800534963607788, + "num_tokens": 297267294.0, + "step": 13545 + }, + { + "epoch": 1.1170116269481323, + "grad_norm": 1.7108787298202515, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7798199653625488, + "num_tokens": 297302069.0, + "step": 13546 + }, + { + "epoch": 1.1170940875731838, + "grad_norm": 1.8897961378097534, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7478665113449097, + "num_tokens": 297335926.0, + "step": 13547 + }, + { + "epoch": 1.1171765481982354, + "grad_norm": 2.0677731037139893, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7346720099449158, + "num_tokens": 297366387.0, + "step": 13548 + }, + { + "epoch": 1.1172590088232868, + "grad_norm": 1.7856303453445435, + "learning_rate": 1e-06, + "loss": 0.5807, + "mean_token_accuracy": 0.8006317019462585, + "num_tokens": 297400762.0, + "step": 13549 + }, + { + "epoch": 1.1173414694483383, + "grad_norm": 1.943770170211792, + "learning_rate": 1e-06, + "loss": 0.5804, + "mean_token_accuracy": 0.799629807472229, + "num_tokens": 297431665.0, + "step": 13550 + }, + { + "epoch": 1.11742393007339, + "grad_norm": 2.0868420600891113, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7548918128013611, + "num_tokens": 297460670.0, + "step": 13551 + }, + { + "epoch": 1.1175063906984415, + "grad_norm": 1.7166805267333984, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7701347470283508, + "num_tokens": 297496300.0, + "step": 13552 + }, + { + "epoch": 1.117588851323493, + "grad_norm": 1.9595586061477661, + "learning_rate": 1e-06, + "loss": 0.5489, + "mean_token_accuracy": 0.817105233669281, + "num_tokens": 297523732.0, + "step": 13553 + }, + { + "epoch": 1.1176713119485446, + "grad_norm": 2.041679620742798, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.7821512222290039, + "num_tokens": 297550211.0, + "step": 13554 + }, + { + "epoch": 1.1177537725735962, + "grad_norm": 1.8251891136169434, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7788615822792053, + "num_tokens": 297585304.0, + "step": 13555 + }, + { + "epoch": 1.1178362331986476, + "grad_norm": 1.6661278009414673, + "learning_rate": 1e-06, + "loss": 0.54, + "mean_token_accuracy": 0.8210252523422241, + "num_tokens": 297619954.0, + "step": 13556 + }, + { + "epoch": 1.1179186938236991, + "grad_norm": 1.922938585281372, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7720248699188232, + "num_tokens": 297652892.0, + "step": 13557 + }, + { + "epoch": 1.1180011544487507, + "grad_norm": 1.8360227346420288, + "learning_rate": 1e-06, + "loss": 0.5579, + "mean_token_accuracy": 0.81922847032547, + "num_tokens": 297684500.0, + "step": 13558 + }, + { + "epoch": 1.1180836150738023, + "grad_norm": 1.8907771110534668, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.7940534353256226, + "num_tokens": 297716568.0, + "step": 13559 + }, + { + "epoch": 1.1181660756988538, + "grad_norm": 1.7351272106170654, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7658306360244751, + "num_tokens": 297755115.0, + "step": 13560 + }, + { + "epoch": 1.1182485363239054, + "grad_norm": 1.8807328939437866, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7852509021759033, + "num_tokens": 297787997.0, + "step": 13561 + }, + { + "epoch": 1.118330996948957, + "grad_norm": 1.909518837928772, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7670392990112305, + "num_tokens": 297820260.0, + "step": 13562 + }, + { + "epoch": 1.1184134575740083, + "grad_norm": 2.0625336170196533, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7611876726150513, + "num_tokens": 297849305.0, + "step": 13563 + }, + { + "epoch": 1.11849591819906, + "grad_norm": 1.9871548414230347, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7743890881538391, + "num_tokens": 297877058.0, + "step": 13564 + }, + { + "epoch": 1.1185783788241115, + "grad_norm": 1.804593801498413, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.7861522436141968, + "num_tokens": 297910549.0, + "step": 13565 + }, + { + "epoch": 1.118660839449163, + "grad_norm": 1.8545565605163574, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7450218200683594, + "num_tokens": 297945422.0, + "step": 13566 + }, + { + "epoch": 1.1187433000742146, + "grad_norm": 2.0429611206054688, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7333425283432007, + "num_tokens": 297974962.0, + "step": 13567 + }, + { + "epoch": 1.1188257606992662, + "grad_norm": 2.1450307369232178, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7267767786979675, + "num_tokens": 298000755.0, + "step": 13568 + }, + { + "epoch": 1.1189082213243176, + "grad_norm": 1.996603012084961, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7595224380493164, + "num_tokens": 298031781.0, + "step": 13569 + }, + { + "epoch": 1.1189906819493691, + "grad_norm": 1.798846960067749, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7498875856399536, + "num_tokens": 298066286.0, + "step": 13570 + }, + { + "epoch": 1.1190731425744207, + "grad_norm": 2.029433250427246, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7818369269371033, + "num_tokens": 298095336.0, + "step": 13571 + }, + { + "epoch": 1.1191556031994723, + "grad_norm": 2.087899923324585, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7670048475265503, + "num_tokens": 298123726.0, + "step": 13572 + }, + { + "epoch": 1.1192380638245238, + "grad_norm": 1.9983700513839722, + "learning_rate": 1e-06, + "loss": 0.5763, + "mean_token_accuracy": 0.802729606628418, + "num_tokens": 298148828.0, + "step": 13573 + }, + { + "epoch": 1.1193205244495754, + "grad_norm": 2.41121244430542, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7485460638999939, + "num_tokens": 298171656.0, + "step": 13574 + }, + { + "epoch": 1.1194029850746268, + "grad_norm": 2.236942768096924, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7308098077774048, + "num_tokens": 298197035.0, + "step": 13575 + }, + { + "epoch": 1.1194854456996783, + "grad_norm": 1.8668400049209595, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7685046792030334, + "num_tokens": 298227101.0, + "step": 13576 + }, + { + "epoch": 1.11956790632473, + "grad_norm": 1.8063369989395142, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.791171669960022, + "num_tokens": 298261568.0, + "step": 13577 + }, + { + "epoch": 1.1196503669497815, + "grad_norm": 2.0586941242218018, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7394821047782898, + "num_tokens": 298291295.0, + "step": 13578 + }, + { + "epoch": 1.119732827574833, + "grad_norm": 1.8175451755523682, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7765514850616455, + "num_tokens": 298324291.0, + "step": 13579 + }, + { + "epoch": 1.1198152881998846, + "grad_norm": 1.8285638093948364, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.791547954082489, + "num_tokens": 298359406.0, + "step": 13580 + }, + { + "epoch": 1.1198977488249362, + "grad_norm": 1.7913622856140137, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.773998498916626, + "num_tokens": 298396501.0, + "step": 13581 + }, + { + "epoch": 1.1199802094499876, + "grad_norm": 1.6923714876174927, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.7829941511154175, + "num_tokens": 298433988.0, + "step": 13582 + }, + { + "epoch": 1.1200626700750391, + "grad_norm": 1.9355354309082031, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7664890289306641, + "num_tokens": 298464073.0, + "step": 13583 + }, + { + "epoch": 1.1201451307000907, + "grad_norm": 1.8363417387008667, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7834417819976807, + "num_tokens": 298499061.0, + "step": 13584 + }, + { + "epoch": 1.1202275913251423, + "grad_norm": 1.7866326570510864, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7073478698730469, + "num_tokens": 298540621.0, + "step": 13585 + }, + { + "epoch": 1.1203100519501938, + "grad_norm": 1.9644664525985718, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7209962606430054, + "num_tokens": 298571533.0, + "step": 13586 + }, + { + "epoch": 1.1203925125752454, + "grad_norm": 1.9613564014434814, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.760116457939148, + "num_tokens": 298601875.0, + "step": 13587 + }, + { + "epoch": 1.1204749732002968, + "grad_norm": 2.349795341491699, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7304348945617676, + "num_tokens": 298629581.0, + "step": 13588 + }, + { + "epoch": 1.1205574338253483, + "grad_norm": 1.907852053642273, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7536821365356445, + "num_tokens": 298661741.0, + "step": 13589 + }, + { + "epoch": 1.1206398944504, + "grad_norm": 1.8038952350616455, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7292155027389526, + "num_tokens": 298698557.0, + "step": 13590 + }, + { + "epoch": 1.1207223550754515, + "grad_norm": 1.9642473459243774, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7592108249664307, + "num_tokens": 298729178.0, + "step": 13591 + }, + { + "epoch": 1.120804815700503, + "grad_norm": 1.6232883930206299, + "learning_rate": 1e-06, + "loss": 0.5427, + "mean_token_accuracy": 0.8183819055557251, + "num_tokens": 298768175.0, + "step": 13592 + }, + { + "epoch": 1.1208872763255546, + "grad_norm": 2.004149913787842, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7626775503158569, + "num_tokens": 298799692.0, + "step": 13593 + }, + { + "epoch": 1.120969736950606, + "grad_norm": 1.9615727663040161, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7578778266906738, + "num_tokens": 298829561.0, + "step": 13594 + }, + { + "epoch": 1.1210521975756575, + "grad_norm": 2.1838066577911377, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7426990270614624, + "num_tokens": 298858094.0, + "step": 13595 + }, + { + "epoch": 1.1211346582007091, + "grad_norm": 1.7945369482040405, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7710567712783813, + "num_tokens": 298893923.0, + "step": 13596 + }, + { + "epoch": 1.1212171188257607, + "grad_norm": 1.9205538034439087, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7811914682388306, + "num_tokens": 298924950.0, + "step": 13597 + }, + { + "epoch": 1.1212995794508123, + "grad_norm": 1.9556797742843628, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7539213299751282, + "num_tokens": 298956309.0, + "step": 13598 + }, + { + "epoch": 1.1213820400758638, + "grad_norm": 1.7626060247421265, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7740210294723511, + "num_tokens": 298992121.0, + "step": 13599 + }, + { + "epoch": 1.1214645007009154, + "grad_norm": 1.9686278104782104, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7327516674995422, + "num_tokens": 299020526.0, + "step": 13600 + }, + { + "epoch": 1.1215469613259668, + "grad_norm": 1.8716354370117188, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.773361325263977, + "num_tokens": 299053794.0, + "step": 13601 + }, + { + "epoch": 1.1216294219510183, + "grad_norm": 2.1154556274414062, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7561817765235901, + "num_tokens": 299079358.0, + "step": 13602 + }, + { + "epoch": 1.12171188257607, + "grad_norm": 1.9776029586791992, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7721531987190247, + "num_tokens": 299108932.0, + "step": 13603 + }, + { + "epoch": 1.1217943432011215, + "grad_norm": 2.3169522285461426, + "learning_rate": 1e-06, + "loss": 0.5864, + "mean_token_accuracy": 0.7998148202896118, + "num_tokens": 299132206.0, + "step": 13604 + }, + { + "epoch": 1.121876803826173, + "grad_norm": 1.748111605644226, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7801670432090759, + "num_tokens": 299170244.0, + "step": 13605 + }, + { + "epoch": 1.1219592644512246, + "grad_norm": 1.9577369689941406, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7551059722900391, + "num_tokens": 299202699.0, + "step": 13606 + }, + { + "epoch": 1.122041725076276, + "grad_norm": 1.926308512687683, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7419642210006714, + "num_tokens": 299237763.0, + "step": 13607 + }, + { + "epoch": 1.1221241857013275, + "grad_norm": 1.9933463335037231, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7621681690216064, + "num_tokens": 299268236.0, + "step": 13608 + }, + { + "epoch": 1.1222066463263791, + "grad_norm": 1.8998552560806274, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.732697606086731, + "num_tokens": 299304476.0, + "step": 13609 + }, + { + "epoch": 1.1222891069514307, + "grad_norm": 2.1155552864074707, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7340517640113831, + "num_tokens": 299335078.0, + "step": 13610 + }, + { + "epoch": 1.1223715675764823, + "grad_norm": 2.0399229526519775, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7731172442436218, + "num_tokens": 299363946.0, + "step": 13611 + }, + { + "epoch": 1.1224540282015338, + "grad_norm": 1.986167073249817, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7351397275924683, + "num_tokens": 299394245.0, + "step": 13612 + }, + { + "epoch": 1.1225364888265852, + "grad_norm": 2.204066514968872, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7702751159667969, + "num_tokens": 299421379.0, + "step": 13613 + }, + { + "epoch": 1.1226189494516368, + "grad_norm": 1.971238136291504, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7307350635528564, + "num_tokens": 299453268.0, + "step": 13614 + }, + { + "epoch": 1.1227014100766883, + "grad_norm": 2.2564191818237305, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7110347151756287, + "num_tokens": 299481066.0, + "step": 13615 + }, + { + "epoch": 1.12278387070174, + "grad_norm": 2.5082080364227295, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7424697279930115, + "num_tokens": 299504039.0, + "step": 13616 + }, + { + "epoch": 1.1228663313267915, + "grad_norm": 1.9233497381210327, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7341796159744263, + "num_tokens": 299539378.0, + "step": 13617 + }, + { + "epoch": 1.122948791951843, + "grad_norm": 2.0966129302978516, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7617737650871277, + "num_tokens": 299566881.0, + "step": 13618 + }, + { + "epoch": 1.1230312525768946, + "grad_norm": 1.8890178203582764, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.78521329164505, + "num_tokens": 299600170.0, + "step": 13619 + }, + { + "epoch": 1.123113713201946, + "grad_norm": 2.1425793170928955, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7849541902542114, + "num_tokens": 299624824.0, + "step": 13620 + }, + { + "epoch": 1.1231961738269975, + "grad_norm": 1.9103811979293823, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7395045757293701, + "num_tokens": 299657019.0, + "step": 13621 + }, + { + "epoch": 1.1232786344520491, + "grad_norm": 2.0746123790740967, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7355135083198547, + "num_tokens": 299687230.0, + "step": 13622 + }, + { + "epoch": 1.1233610950771007, + "grad_norm": 2.0343050956726074, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7249854207038879, + "num_tokens": 299715893.0, + "step": 13623 + }, + { + "epoch": 1.1234435557021523, + "grad_norm": 2.787567377090454, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7498779296875, + "num_tokens": 299739590.0, + "step": 13624 + }, + { + "epoch": 1.1235260163272038, + "grad_norm": 2.105583429336548, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.771450936794281, + "num_tokens": 299766021.0, + "step": 13625 + }, + { + "epoch": 1.1236084769522554, + "grad_norm": 1.9964983463287354, + "learning_rate": 1e-06, + "loss": 0.584, + "mean_token_accuracy": 0.7969925403594971, + "num_tokens": 299793782.0, + "step": 13626 + }, + { + "epoch": 1.1236909375773068, + "grad_norm": 2.0060925483703613, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7624388933181763, + "num_tokens": 299821328.0, + "step": 13627 + }, + { + "epoch": 1.1237733982023583, + "grad_norm": 1.9175196886062622, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7506582140922546, + "num_tokens": 299855413.0, + "step": 13628 + }, + { + "epoch": 1.12385585882741, + "grad_norm": 1.8108067512512207, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7745527625083923, + "num_tokens": 299889149.0, + "step": 13629 + }, + { + "epoch": 1.1239383194524615, + "grad_norm": 2.288482666015625, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7573824524879456, + "num_tokens": 299914361.0, + "step": 13630 + }, + { + "epoch": 1.124020780077513, + "grad_norm": 2.191667079925537, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7551652789115906, + "num_tokens": 299940287.0, + "step": 13631 + }, + { + "epoch": 1.1241032407025646, + "grad_norm": 1.8698418140411377, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.773889422416687, + "num_tokens": 299973404.0, + "step": 13632 + }, + { + "epoch": 1.124185701327616, + "grad_norm": 1.7220252752304077, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7662701606750488, + "num_tokens": 300010445.0, + "step": 13633 + }, + { + "epoch": 1.1242681619526675, + "grad_norm": 1.9050551652908325, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.756136953830719, + "num_tokens": 300045462.0, + "step": 13634 + }, + { + "epoch": 1.1243506225777191, + "grad_norm": 1.6719517707824707, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.783900797367096, + "num_tokens": 300085537.0, + "step": 13635 + }, + { + "epoch": 1.1244330832027707, + "grad_norm": 1.845708966255188, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7143343687057495, + "num_tokens": 300125682.0, + "step": 13636 + }, + { + "epoch": 1.1245155438278223, + "grad_norm": 1.940610408782959, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7861555218696594, + "num_tokens": 300155853.0, + "step": 13637 + }, + { + "epoch": 1.1245980044528738, + "grad_norm": 2.0822675228118896, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7350329756736755, + "num_tokens": 300185043.0, + "step": 13638 + }, + { + "epoch": 1.1246804650779252, + "grad_norm": 2.1019887924194336, + "learning_rate": 1e-06, + "loss": 0.5998, + "mean_token_accuracy": 0.8082789182662964, + "num_tokens": 300210659.0, + "step": 13639 + }, + { + "epoch": 1.1247629257029768, + "grad_norm": 1.9502906799316406, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7347533106803894, + "num_tokens": 300241500.0, + "step": 13640 + }, + { + "epoch": 1.1248453863280283, + "grad_norm": 2.172109842300415, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7686787843704224, + "num_tokens": 300266128.0, + "step": 13641 + }, + { + "epoch": 1.12492784695308, + "grad_norm": 1.944236159324646, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7739500999450684, + "num_tokens": 300297381.0, + "step": 13642 + }, + { + "epoch": 1.1250103075781315, + "grad_norm": 2.3309919834136963, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7643246650695801, + "num_tokens": 300319668.0, + "step": 13643 + }, + { + "epoch": 1.125092768203183, + "grad_norm": 2.3148248195648193, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7475462555885315, + "num_tokens": 300343824.0, + "step": 13644 + }, + { + "epoch": 1.1251752288282346, + "grad_norm": 2.1002097129821777, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.7898320555686951, + "num_tokens": 300370642.0, + "step": 13645 + }, + { + "epoch": 1.125257689453286, + "grad_norm": 2.026165008544922, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7071919441223145, + "num_tokens": 300404282.0, + "step": 13646 + }, + { + "epoch": 1.1253401500783375, + "grad_norm": 1.8575539588928223, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7840610146522522, + "num_tokens": 300434624.0, + "step": 13647 + }, + { + "epoch": 1.1254226107033891, + "grad_norm": 2.0034589767456055, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7415850162506104, + "num_tokens": 300465727.0, + "step": 13648 + }, + { + "epoch": 1.1255050713284407, + "grad_norm": 1.915716528892517, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7219222784042358, + "num_tokens": 300498504.0, + "step": 13649 + }, + { + "epoch": 1.1255875319534923, + "grad_norm": 1.9723163843154907, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.7826581597328186, + "num_tokens": 300530924.0, + "step": 13650 + }, + { + "epoch": 1.1256699925785438, + "grad_norm": 1.8934736251831055, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7874216437339783, + "num_tokens": 300563098.0, + "step": 13651 + }, + { + "epoch": 1.1257524532035954, + "grad_norm": 1.927061676979065, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7395691871643066, + "num_tokens": 300598890.0, + "step": 13652 + }, + { + "epoch": 1.1258349138286468, + "grad_norm": 2.0591042041778564, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7420109510421753, + "num_tokens": 300627210.0, + "step": 13653 + }, + { + "epoch": 1.1259173744536983, + "grad_norm": 1.9779695272445679, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7550075054168701, + "num_tokens": 300656024.0, + "step": 13654 + }, + { + "epoch": 1.12599983507875, + "grad_norm": 1.745194435119629, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.7843179702758789, + "num_tokens": 300693896.0, + "step": 13655 + }, + { + "epoch": 1.1260822957038015, + "grad_norm": 1.9691389799118042, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7762230634689331, + "num_tokens": 300722571.0, + "step": 13656 + }, + { + "epoch": 1.126164756328853, + "grad_norm": 1.7871707677841187, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.7934052348136902, + "num_tokens": 300756709.0, + "step": 13657 + }, + { + "epoch": 1.1262472169539044, + "grad_norm": 1.804284930229187, + "learning_rate": 1e-06, + "loss": 0.5511, + "mean_token_accuracy": 0.8194608688354492, + "num_tokens": 300790378.0, + "step": 13658 + }, + { + "epoch": 1.126329677578956, + "grad_norm": 1.6566352844238281, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7744042277336121, + "num_tokens": 300825944.0, + "step": 13659 + }, + { + "epoch": 1.1264121382040075, + "grad_norm": 2.0341756343841553, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7711145877838135, + "num_tokens": 300851895.0, + "step": 13660 + }, + { + "epoch": 1.1264945988290591, + "grad_norm": 1.7816423177719116, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.7857601642608643, + "num_tokens": 300886972.0, + "step": 13661 + }, + { + "epoch": 1.1265770594541107, + "grad_norm": 2.123272657394409, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.755164384841919, + "num_tokens": 300913648.0, + "step": 13662 + }, + { + "epoch": 1.1266595200791623, + "grad_norm": 2.0061066150665283, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7500481605529785, + "num_tokens": 300943613.0, + "step": 13663 + }, + { + "epoch": 1.1267419807042138, + "grad_norm": 2.1696722507476807, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7696478962898254, + "num_tokens": 300966669.0, + "step": 13664 + }, + { + "epoch": 1.1268244413292652, + "grad_norm": 2.150458812713623, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7342214584350586, + "num_tokens": 300993194.0, + "step": 13665 + }, + { + "epoch": 1.1269069019543168, + "grad_norm": 2.208369493484497, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7809209823608398, + "num_tokens": 301016063.0, + "step": 13666 + }, + { + "epoch": 1.1269893625793683, + "grad_norm": 2.0442521572113037, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7448334097862244, + "num_tokens": 301044423.0, + "step": 13667 + }, + { + "epoch": 1.12707182320442, + "grad_norm": 1.9380451440811157, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7465870380401611, + "num_tokens": 301076029.0, + "step": 13668 + }, + { + "epoch": 1.1271542838294715, + "grad_norm": 1.9407395124435425, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7609763741493225, + "num_tokens": 301106609.0, + "step": 13669 + }, + { + "epoch": 1.127236744454523, + "grad_norm": 1.729875087738037, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.7936168313026428, + "num_tokens": 301144376.0, + "step": 13670 + }, + { + "epoch": 1.1273192050795746, + "grad_norm": 2.010446071624756, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.7661991119384766, + "num_tokens": 301173507.0, + "step": 13671 + }, + { + "epoch": 1.127401665704626, + "grad_norm": 2.142239570617676, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.7859772443771362, + "num_tokens": 301198207.0, + "step": 13672 + }, + { + "epoch": 1.1274841263296775, + "grad_norm": 2.169713020324707, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7536282539367676, + "num_tokens": 301225678.0, + "step": 13673 + }, + { + "epoch": 1.127566586954729, + "grad_norm": 1.9818873405456543, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7807098627090454, + "num_tokens": 301261783.0, + "step": 13674 + }, + { + "epoch": 1.1276490475797807, + "grad_norm": 2.3618409633636475, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7251633405685425, + "num_tokens": 301287460.0, + "step": 13675 + }, + { + "epoch": 1.1277315082048323, + "grad_norm": 1.78293776512146, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7637891173362732, + "num_tokens": 301327319.0, + "step": 13676 + }, + { + "epoch": 1.1278139688298836, + "grad_norm": 2.221301555633545, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7597389221191406, + "num_tokens": 301351088.0, + "step": 13677 + }, + { + "epoch": 1.1278964294549352, + "grad_norm": 1.8070321083068848, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7569131851196289, + "num_tokens": 301388061.0, + "step": 13678 + }, + { + "epoch": 1.1279788900799868, + "grad_norm": 1.939014196395874, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7377156615257263, + "num_tokens": 301419479.0, + "step": 13679 + }, + { + "epoch": 1.1280613507050383, + "grad_norm": 1.9384592771530151, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7463514804840088, + "num_tokens": 301449075.0, + "step": 13680 + }, + { + "epoch": 1.12814381133009, + "grad_norm": 1.997928500175476, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7103710174560547, + "num_tokens": 301480818.0, + "step": 13681 + }, + { + "epoch": 1.1282262719551415, + "grad_norm": 1.844206690788269, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7513235807418823, + "num_tokens": 301512837.0, + "step": 13682 + }, + { + "epoch": 1.128308732580193, + "grad_norm": 1.924418568611145, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7313842177391052, + "num_tokens": 301544383.0, + "step": 13683 + }, + { + "epoch": 1.1283911932052444, + "grad_norm": 1.6978163719177246, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7817025184631348, + "num_tokens": 301580494.0, + "step": 13684 + }, + { + "epoch": 1.128473653830296, + "grad_norm": 1.9597232341766357, + "learning_rate": 1e-06, + "loss": 0.5884, + "mean_token_accuracy": 0.8013763427734375, + "num_tokens": 301609346.0, + "step": 13685 + }, + { + "epoch": 1.1285561144553475, + "grad_norm": 1.9982576370239258, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7460397481918335, + "num_tokens": 301637197.0, + "step": 13686 + }, + { + "epoch": 1.128638575080399, + "grad_norm": 1.8630468845367432, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7637813091278076, + "num_tokens": 301671721.0, + "step": 13687 + }, + { + "epoch": 1.1287210357054507, + "grad_norm": 1.860884428024292, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.8040479421615601, + "num_tokens": 301703151.0, + "step": 13688 + }, + { + "epoch": 1.1288034963305023, + "grad_norm": 2.14988374710083, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.761432945728302, + "num_tokens": 301728187.0, + "step": 13689 + }, + { + "epoch": 1.1288859569555538, + "grad_norm": 2.2335989475250244, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7206284403800964, + "num_tokens": 301753247.0, + "step": 13690 + }, + { + "epoch": 1.1289684175806052, + "grad_norm": 2.075068473815918, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.753361165523529, + "num_tokens": 301781206.0, + "step": 13691 + }, + { + "epoch": 1.1290508782056567, + "grad_norm": 1.6813751459121704, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.7893710732460022, + "num_tokens": 301822634.0, + "step": 13692 + }, + { + "epoch": 1.1291333388307083, + "grad_norm": 2.0818443298339844, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7400656938552856, + "num_tokens": 301851551.0, + "step": 13693 + }, + { + "epoch": 1.12921579945576, + "grad_norm": 2.162681818008423, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.732078492641449, + "num_tokens": 301880531.0, + "step": 13694 + }, + { + "epoch": 1.1292982600808115, + "grad_norm": 1.8224183320999146, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7341573238372803, + "num_tokens": 301917748.0, + "step": 13695 + }, + { + "epoch": 1.129380720705863, + "grad_norm": 1.8923956155776978, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7568432688713074, + "num_tokens": 301952645.0, + "step": 13696 + }, + { + "epoch": 1.1294631813309146, + "grad_norm": 1.9114247560501099, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7592111825942993, + "num_tokens": 301983108.0, + "step": 13697 + }, + { + "epoch": 1.129545641955966, + "grad_norm": 1.8446253538131714, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7654368877410889, + "num_tokens": 302016688.0, + "step": 13698 + }, + { + "epoch": 1.1296281025810175, + "grad_norm": 2.0164999961853027, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7662277221679688, + "num_tokens": 302047011.0, + "step": 13699 + }, + { + "epoch": 1.129710563206069, + "grad_norm": 1.8382642269134521, + "learning_rate": 1e-06, + "loss": 0.7585, + "mean_token_accuracy": 0.7564178705215454, + "num_tokens": 302081236.0, + "step": 13700 + }, + { + "epoch": 1.1297930238311207, + "grad_norm": 1.9253352880477905, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7617701292037964, + "num_tokens": 302117214.0, + "step": 13701 + }, + { + "epoch": 1.1298754844561723, + "grad_norm": 2.1759793758392334, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7541964054107666, + "num_tokens": 302142547.0, + "step": 13702 + }, + { + "epoch": 1.1299579450812236, + "grad_norm": 2.425780773162842, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7192714214324951, + "num_tokens": 302165453.0, + "step": 13703 + }, + { + "epoch": 1.1300404057062752, + "grad_norm": 1.9191803932189941, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.7540832161903381, + "num_tokens": 302197946.0, + "step": 13704 + }, + { + "epoch": 1.1301228663313267, + "grad_norm": 1.9798107147216797, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.762662410736084, + "num_tokens": 302229489.0, + "step": 13705 + }, + { + "epoch": 1.1302053269563783, + "grad_norm": 2.2733654975891113, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7671383619308472, + "num_tokens": 302255240.0, + "step": 13706 + }, + { + "epoch": 1.13028778758143, + "grad_norm": 2.1961281299591064, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7427471876144409, + "num_tokens": 302280603.0, + "step": 13707 + }, + { + "epoch": 1.1303702482064815, + "grad_norm": 1.848271131515503, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7458176612854004, + "num_tokens": 302314387.0, + "step": 13708 + }, + { + "epoch": 1.130452708831533, + "grad_norm": 2.0536484718322754, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7578818798065186, + "num_tokens": 302344379.0, + "step": 13709 + }, + { + "epoch": 1.1305351694565844, + "grad_norm": 2.1441729068756104, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7653810381889343, + "num_tokens": 302371252.0, + "step": 13710 + }, + { + "epoch": 1.130617630081636, + "grad_norm": 1.9571473598480225, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7586517930030823, + "num_tokens": 302400060.0, + "step": 13711 + }, + { + "epoch": 1.1307000907066875, + "grad_norm": 1.864625334739685, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7708152532577515, + "num_tokens": 302435339.0, + "step": 13712 + }, + { + "epoch": 1.130782551331739, + "grad_norm": 1.8024137020111084, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.7610985040664673, + "num_tokens": 302472903.0, + "step": 13713 + }, + { + "epoch": 1.1308650119567907, + "grad_norm": 1.851128339767456, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7750595211982727, + "num_tokens": 302508345.0, + "step": 13714 + }, + { + "epoch": 1.1309474725818423, + "grad_norm": 1.951562523841858, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7728375196456909, + "num_tokens": 302539391.0, + "step": 13715 + }, + { + "epoch": 1.1310299332068938, + "grad_norm": 2.091353178024292, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7396774291992188, + "num_tokens": 302565198.0, + "step": 13716 + }, + { + "epoch": 1.1311123938319452, + "grad_norm": 1.955488681793213, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7654427886009216, + "num_tokens": 302595218.0, + "step": 13717 + }, + { + "epoch": 1.1311948544569967, + "grad_norm": 2.1620213985443115, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7551793456077576, + "num_tokens": 302620424.0, + "step": 13718 + }, + { + "epoch": 1.1312773150820483, + "grad_norm": 2.1418161392211914, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.7843733429908752, + "num_tokens": 302645847.0, + "step": 13719 + }, + { + "epoch": 1.1313597757071, + "grad_norm": 1.813331127166748, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7751983404159546, + "num_tokens": 302678933.0, + "step": 13720 + }, + { + "epoch": 1.1314422363321515, + "grad_norm": 2.0040881633758545, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7768009901046753, + "num_tokens": 302704054.0, + "step": 13721 + }, + { + "epoch": 1.1315246969572028, + "grad_norm": 2.299903631210327, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7388597726821899, + "num_tokens": 302730688.0, + "step": 13722 + }, + { + "epoch": 1.1316071575822544, + "grad_norm": 2.2147634029388428, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7583111524581909, + "num_tokens": 302759037.0, + "step": 13723 + }, + { + "epoch": 1.131689618207306, + "grad_norm": 2.0251684188842773, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7394120693206787, + "num_tokens": 302790987.0, + "step": 13724 + }, + { + "epoch": 1.1317720788323575, + "grad_norm": 2.3967320919036865, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7441655993461609, + "num_tokens": 302815704.0, + "step": 13725 + }, + { + "epoch": 1.131854539457409, + "grad_norm": 1.8786364793777466, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7709158062934875, + "num_tokens": 302849504.0, + "step": 13726 + }, + { + "epoch": 1.1319370000824607, + "grad_norm": 1.999730110168457, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.7969422340393066, + "num_tokens": 302876213.0, + "step": 13727 + }, + { + "epoch": 1.1320194607075122, + "grad_norm": 2.230137348175049, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7459871172904968, + "num_tokens": 302901722.0, + "step": 13728 + }, + { + "epoch": 1.1321019213325636, + "grad_norm": 1.878884196281433, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.7851674556732178, + "num_tokens": 302933299.0, + "step": 13729 + }, + { + "epoch": 1.1321843819576152, + "grad_norm": 2.0168888568878174, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7660714387893677, + "num_tokens": 302964761.0, + "step": 13730 + }, + { + "epoch": 1.1322668425826667, + "grad_norm": 2.0417873859405518, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7581639885902405, + "num_tokens": 302994184.0, + "step": 13731 + }, + { + "epoch": 1.1323493032077183, + "grad_norm": 2.0381712913513184, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.770817756652832, + "num_tokens": 303022664.0, + "step": 13732 + }, + { + "epoch": 1.1324317638327699, + "grad_norm": 1.8937017917633057, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7875458598136902, + "num_tokens": 303054549.0, + "step": 13733 + }, + { + "epoch": 1.1325142244578215, + "grad_norm": 2.166957139968872, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7370604872703552, + "num_tokens": 303081958.0, + "step": 13734 + }, + { + "epoch": 1.132596685082873, + "grad_norm": 1.9995033740997314, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7626566290855408, + "num_tokens": 303110656.0, + "step": 13735 + }, + { + "epoch": 1.1326791457079244, + "grad_norm": 1.9453481435775757, + "learning_rate": 1e-06, + "loss": 0.6093, + "mean_token_accuracy": 0.7949466109275818, + "num_tokens": 303138794.0, + "step": 13736 + }, + { + "epoch": 1.132761606332976, + "grad_norm": 1.9445991516113281, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7644596099853516, + "num_tokens": 303168790.0, + "step": 13737 + }, + { + "epoch": 1.1328440669580275, + "grad_norm": 1.8329490423202515, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7734638452529907, + "num_tokens": 303203606.0, + "step": 13738 + }, + { + "epoch": 1.132926527583079, + "grad_norm": 1.8809096813201904, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7799580693244934, + "num_tokens": 303235598.0, + "step": 13739 + }, + { + "epoch": 1.1330089882081307, + "grad_norm": 1.830147624015808, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7651679515838623, + "num_tokens": 303269226.0, + "step": 13740 + }, + { + "epoch": 1.1330914488331822, + "grad_norm": 2.2431857585906982, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7603758573532104, + "num_tokens": 303293784.0, + "step": 13741 + }, + { + "epoch": 1.1331739094582336, + "grad_norm": 1.9158309698104858, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.7531793117523193, + "num_tokens": 303323930.0, + "step": 13742 + }, + { + "epoch": 1.1332563700832852, + "grad_norm": 1.9270896911621094, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7597840428352356, + "num_tokens": 303356487.0, + "step": 13743 + }, + { + "epoch": 1.1333388307083367, + "grad_norm": 1.9052015542984009, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.740903377532959, + "num_tokens": 303390309.0, + "step": 13744 + }, + { + "epoch": 1.1334212913333883, + "grad_norm": 2.0180201530456543, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7861255407333374, + "num_tokens": 303420024.0, + "step": 13745 + }, + { + "epoch": 1.1335037519584399, + "grad_norm": 2.091536283493042, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7712165713310242, + "num_tokens": 303451630.0, + "step": 13746 + }, + { + "epoch": 1.1335862125834915, + "grad_norm": 2.2409164905548096, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7457878589630127, + "num_tokens": 303479675.0, + "step": 13747 + }, + { + "epoch": 1.1336686732085428, + "grad_norm": 2.0196194648742676, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7692077159881592, + "num_tokens": 303506100.0, + "step": 13748 + }, + { + "epoch": 1.1337511338335944, + "grad_norm": 2.120439052581787, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7280197739601135, + "num_tokens": 303535089.0, + "step": 13749 + }, + { + "epoch": 1.133833594458646, + "grad_norm": 1.8261473178863525, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.7997238636016846, + "num_tokens": 303568600.0, + "step": 13750 + }, + { + "epoch": 1.1339160550836975, + "grad_norm": 2.105888605117798, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7644233703613281, + "num_tokens": 303597244.0, + "step": 13751 + }, + { + "epoch": 1.133998515708749, + "grad_norm": 2.114319086074829, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7331558465957642, + "num_tokens": 303624888.0, + "step": 13752 + }, + { + "epoch": 1.1340809763338007, + "grad_norm": 1.9447203874588013, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7089587450027466, + "num_tokens": 303657256.0, + "step": 13753 + }, + { + "epoch": 1.1341634369588522, + "grad_norm": 2.3388357162475586, + "learning_rate": 1e-06, + "loss": 0.786, + "mean_token_accuracy": 0.7428205013275146, + "num_tokens": 303682159.0, + "step": 13754 + }, + { + "epoch": 1.1342458975839036, + "grad_norm": 2.0950472354888916, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7680863738059998, + "num_tokens": 303713157.0, + "step": 13755 + }, + { + "epoch": 1.1343283582089552, + "grad_norm": 2.04630446434021, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7652695178985596, + "num_tokens": 303740931.0, + "step": 13756 + }, + { + "epoch": 1.1344108188340067, + "grad_norm": 2.2420663833618164, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7426294684410095, + "num_tokens": 303768912.0, + "step": 13757 + }, + { + "epoch": 1.1344932794590583, + "grad_norm": 1.9096195697784424, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7714449763298035, + "num_tokens": 303799262.0, + "step": 13758 + }, + { + "epoch": 1.1345757400841099, + "grad_norm": 1.9255014657974243, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7443727254867554, + "num_tokens": 303831375.0, + "step": 13759 + }, + { + "epoch": 1.1346582007091615, + "grad_norm": 1.8829655647277832, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7751827239990234, + "num_tokens": 303863941.0, + "step": 13760 + }, + { + "epoch": 1.134740661334213, + "grad_norm": 2.1685683727264404, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7060821056365967, + "num_tokens": 303892353.0, + "step": 13761 + }, + { + "epoch": 1.1348231219592644, + "grad_norm": 2.00219988822937, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7305988073348999, + "num_tokens": 303921810.0, + "step": 13762 + }, + { + "epoch": 1.134905582584316, + "grad_norm": 1.8713393211364746, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.777881383895874, + "num_tokens": 303953608.0, + "step": 13763 + }, + { + "epoch": 1.1349880432093675, + "grad_norm": 1.955600619316101, + "learning_rate": 1e-06, + "loss": 0.5747, + "mean_token_accuracy": 0.8108484148979187, + "num_tokens": 303981642.0, + "step": 13764 + }, + { + "epoch": 1.135070503834419, + "grad_norm": 2.159834146499634, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.76036536693573, + "num_tokens": 304008873.0, + "step": 13765 + }, + { + "epoch": 1.1351529644594707, + "grad_norm": 1.816988468170166, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7670491933822632, + "num_tokens": 304041762.0, + "step": 13766 + }, + { + "epoch": 1.135235425084522, + "grad_norm": 2.0856261253356934, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7502110600471497, + "num_tokens": 304073850.0, + "step": 13767 + }, + { + "epoch": 1.1353178857095736, + "grad_norm": 1.989797830581665, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.7845597267150879, + "num_tokens": 304110915.0, + "step": 13768 + }, + { + "epoch": 1.1354003463346252, + "grad_norm": 1.8451515436172485, + "learning_rate": 1e-06, + "loss": 0.8014, + "mean_token_accuracy": 0.7522773742675781, + "num_tokens": 304142440.0, + "step": 13769 + }, + { + "epoch": 1.1354828069596767, + "grad_norm": 2.1118574142456055, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7559502720832825, + "num_tokens": 304169645.0, + "step": 13770 + }, + { + "epoch": 1.1355652675847283, + "grad_norm": 2.0135843753814697, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7458171844482422, + "num_tokens": 304204139.0, + "step": 13771 + }, + { + "epoch": 1.1356477282097799, + "grad_norm": 2.121342897415161, + "learning_rate": 1e-06, + "loss": 0.5797, + "mean_token_accuracy": 0.8039964437484741, + "num_tokens": 304234167.0, + "step": 13772 + }, + { + "epoch": 1.1357301888348315, + "grad_norm": 1.9667831659317017, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7782779335975647, + "num_tokens": 304264518.0, + "step": 13773 + }, + { + "epoch": 1.1358126494598828, + "grad_norm": 2.1234853267669678, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7465552091598511, + "num_tokens": 304296224.0, + "step": 13774 + }, + { + "epoch": 1.1358951100849344, + "grad_norm": 1.8535550832748413, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7073962688446045, + "num_tokens": 304330855.0, + "step": 13775 + }, + { + "epoch": 1.135977570709986, + "grad_norm": 1.899095058441162, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7825716733932495, + "num_tokens": 304364511.0, + "step": 13776 + }, + { + "epoch": 1.1360600313350375, + "grad_norm": 1.9355603456497192, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.7855072021484375, + "num_tokens": 304395414.0, + "step": 13777 + }, + { + "epoch": 1.136142491960089, + "grad_norm": 2.089444160461426, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7366402745246887, + "num_tokens": 304422334.0, + "step": 13778 + }, + { + "epoch": 1.1362249525851407, + "grad_norm": 2.010666847229004, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7502282857894897, + "num_tokens": 304452414.0, + "step": 13779 + }, + { + "epoch": 1.1363074132101922, + "grad_norm": 1.9008405208587646, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.797359049320221, + "num_tokens": 304483902.0, + "step": 13780 + }, + { + "epoch": 1.1363898738352436, + "grad_norm": 2.0822384357452393, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7399004697799683, + "num_tokens": 304511268.0, + "step": 13781 + }, + { + "epoch": 1.1364723344602952, + "grad_norm": 2.066648244857788, + "learning_rate": 1e-06, + "loss": 0.6072, + "mean_token_accuracy": 0.796634316444397, + "num_tokens": 304535932.0, + "step": 13782 + }, + { + "epoch": 1.1365547950853467, + "grad_norm": 1.885582447052002, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7722522616386414, + "num_tokens": 304569415.0, + "step": 13783 + }, + { + "epoch": 1.1366372557103983, + "grad_norm": 2.296337127685547, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.737675666809082, + "num_tokens": 304593146.0, + "step": 13784 + }, + { + "epoch": 1.1367197163354499, + "grad_norm": 1.7393622398376465, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7611583471298218, + "num_tokens": 304628303.0, + "step": 13785 + }, + { + "epoch": 1.1368021769605015, + "grad_norm": 2.67085599899292, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7444993257522583, + "num_tokens": 304648263.0, + "step": 13786 + }, + { + "epoch": 1.1368846375855528, + "grad_norm": 1.826006531715393, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7606858611106873, + "num_tokens": 304686227.0, + "step": 13787 + }, + { + "epoch": 1.1369670982106044, + "grad_norm": 2.347459316253662, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7305710911750793, + "num_tokens": 304713798.0, + "step": 13788 + }, + { + "epoch": 1.137049558835656, + "grad_norm": 1.87531578540802, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7636896371841431, + "num_tokens": 304747234.0, + "step": 13789 + }, + { + "epoch": 1.1371320194607075, + "grad_norm": 1.8435999155044556, + "learning_rate": 1e-06, + "loss": 0.5952, + "mean_token_accuracy": 0.7872939109802246, + "num_tokens": 304779470.0, + "step": 13790 + }, + { + "epoch": 1.137214480085759, + "grad_norm": 2.0861928462982178, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.733059823513031, + "num_tokens": 304810234.0, + "step": 13791 + }, + { + "epoch": 1.1372969407108107, + "grad_norm": 2.2442545890808105, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7225968837738037, + "num_tokens": 304838054.0, + "step": 13792 + }, + { + "epoch": 1.137379401335862, + "grad_norm": 2.13131046295166, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.7597343921661377, + "num_tokens": 304862864.0, + "step": 13793 + }, + { + "epoch": 1.1374618619609136, + "grad_norm": 1.941162109375, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.72325199842453, + "num_tokens": 304894263.0, + "step": 13794 + }, + { + "epoch": 1.1375443225859652, + "grad_norm": 1.9908798933029175, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7324187755584717, + "num_tokens": 304925819.0, + "step": 13795 + }, + { + "epoch": 1.1376267832110167, + "grad_norm": 1.8675634860992432, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7752772569656372, + "num_tokens": 304956217.0, + "step": 13796 + }, + { + "epoch": 1.1377092438360683, + "grad_norm": 2.5882420539855957, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.742095947265625, + "num_tokens": 304976711.0, + "step": 13797 + }, + { + "epoch": 1.1377917044611199, + "grad_norm": 1.939509630203247, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7232353687286377, + "num_tokens": 305007207.0, + "step": 13798 + }, + { + "epoch": 1.1378741650861715, + "grad_norm": 1.836686134338379, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.7925050258636475, + "num_tokens": 305038069.0, + "step": 13799 + }, + { + "epoch": 1.1379566257112228, + "grad_norm": 2.110712766647339, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7264477610588074, + "num_tokens": 305065154.0, + "step": 13800 + }, + { + "epoch": 1.1380390863362744, + "grad_norm": 2.1021878719329834, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7694327235221863, + "num_tokens": 305094468.0, + "step": 13801 + }, + { + "epoch": 1.138121546961326, + "grad_norm": 2.068479299545288, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.7855956554412842, + "num_tokens": 305120031.0, + "step": 13802 + }, + { + "epoch": 1.1382040075863775, + "grad_norm": 1.823149561882019, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7788868546485901, + "num_tokens": 305154708.0, + "step": 13803 + }, + { + "epoch": 1.138286468211429, + "grad_norm": 2.104336738586426, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7638511061668396, + "num_tokens": 305182011.0, + "step": 13804 + }, + { + "epoch": 1.1383689288364807, + "grad_norm": 1.8391859531402588, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.780985951423645, + "num_tokens": 305213835.0, + "step": 13805 + }, + { + "epoch": 1.1384513894615322, + "grad_norm": 2.118896961212158, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7729928493499756, + "num_tokens": 305244926.0, + "step": 13806 + }, + { + "epoch": 1.1385338500865836, + "grad_norm": 1.941016435623169, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7354425191879272, + "num_tokens": 305275022.0, + "step": 13807 + }, + { + "epoch": 1.1386163107116352, + "grad_norm": 1.9855940341949463, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7707461714744568, + "num_tokens": 305304973.0, + "step": 13808 + }, + { + "epoch": 1.1386987713366867, + "grad_norm": 2.1484005451202393, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7755424976348877, + "num_tokens": 305331874.0, + "step": 13809 + }, + { + "epoch": 1.1387812319617383, + "grad_norm": 1.9635881185531616, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7585344910621643, + "num_tokens": 305362033.0, + "step": 13810 + }, + { + "epoch": 1.1388636925867899, + "grad_norm": 2.0703463554382324, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7543692588806152, + "num_tokens": 305390859.0, + "step": 13811 + }, + { + "epoch": 1.1389461532118412, + "grad_norm": 1.910005807876587, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7719335556030273, + "num_tokens": 305425168.0, + "step": 13812 + }, + { + "epoch": 1.1390286138368928, + "grad_norm": 1.7653435468673706, + "learning_rate": 1e-06, + "loss": 0.5661, + "mean_token_accuracy": 0.8102537393569946, + "num_tokens": 305460759.0, + "step": 13813 + }, + { + "epoch": 1.1391110744619444, + "grad_norm": 1.6689010858535767, + "learning_rate": 1e-06, + "loss": 0.574, + "mean_token_accuracy": 0.8079373836517334, + "num_tokens": 305496590.0, + "step": 13814 + }, + { + "epoch": 1.139193535086996, + "grad_norm": 2.0505359172821045, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7497447729110718, + "num_tokens": 305527792.0, + "step": 13815 + }, + { + "epoch": 1.1392759957120475, + "grad_norm": 1.8669437170028687, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7324119806289673, + "num_tokens": 305562966.0, + "step": 13816 + }, + { + "epoch": 1.139358456337099, + "grad_norm": 1.7441811561584473, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7714598178863525, + "num_tokens": 305600000.0, + "step": 13817 + }, + { + "epoch": 1.1394409169621507, + "grad_norm": 1.9632823467254639, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7788101434707642, + "num_tokens": 305629356.0, + "step": 13818 + }, + { + "epoch": 1.139523377587202, + "grad_norm": 1.9706029891967773, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7674700617790222, + "num_tokens": 305661886.0, + "step": 13819 + }, + { + "epoch": 1.1396058382122536, + "grad_norm": 1.8804742097854614, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7604629993438721, + "num_tokens": 305692483.0, + "step": 13820 + }, + { + "epoch": 1.1396882988373052, + "grad_norm": 2.1589622497558594, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7477959394454956, + "num_tokens": 305720194.0, + "step": 13821 + }, + { + "epoch": 1.1397707594623567, + "grad_norm": 1.9591782093048096, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.736751914024353, + "num_tokens": 305750888.0, + "step": 13822 + }, + { + "epoch": 1.1398532200874083, + "grad_norm": 2.2307240962982178, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.734773576259613, + "num_tokens": 305775662.0, + "step": 13823 + }, + { + "epoch": 1.1399356807124599, + "grad_norm": 2.0976619720458984, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7627596855163574, + "num_tokens": 305800965.0, + "step": 13824 + }, + { + "epoch": 1.1400181413375114, + "grad_norm": 1.8202016353607178, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.747423529624939, + "num_tokens": 305836978.0, + "step": 13825 + }, + { + "epoch": 1.1401006019625628, + "grad_norm": 2.094949722290039, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7267360687255859, + "num_tokens": 305865899.0, + "step": 13826 + }, + { + "epoch": 1.1401830625876144, + "grad_norm": 1.738256573677063, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.7954386472702026, + "num_tokens": 305901564.0, + "step": 13827 + }, + { + "epoch": 1.140265523212666, + "grad_norm": 2.349520444869995, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7497155070304871, + "num_tokens": 305924973.0, + "step": 13828 + }, + { + "epoch": 1.1403479838377175, + "grad_norm": 2.3180086612701416, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7555179595947266, + "num_tokens": 305948826.0, + "step": 13829 + }, + { + "epoch": 1.140430444462769, + "grad_norm": 1.92594575881958, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7159289121627808, + "num_tokens": 305983807.0, + "step": 13830 + }, + { + "epoch": 1.1405129050878207, + "grad_norm": 1.8456356525421143, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7507601380348206, + "num_tokens": 306016887.0, + "step": 13831 + }, + { + "epoch": 1.140595365712872, + "grad_norm": 1.723189115524292, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7614530324935913, + "num_tokens": 306056623.0, + "step": 13832 + }, + { + "epoch": 1.1406778263379236, + "grad_norm": 2.165300130844116, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7564411163330078, + "num_tokens": 306084250.0, + "step": 13833 + }, + { + "epoch": 1.1407602869629752, + "grad_norm": 1.8528871536254883, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7208806872367859, + "num_tokens": 306119005.0, + "step": 13834 + }, + { + "epoch": 1.1408427475880267, + "grad_norm": 2.007640838623047, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7647382616996765, + "num_tokens": 306147518.0, + "step": 13835 + }, + { + "epoch": 1.1409252082130783, + "grad_norm": 2.070878028869629, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.76739501953125, + "num_tokens": 306175040.0, + "step": 13836 + }, + { + "epoch": 1.1410076688381299, + "grad_norm": 2.085484027862549, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7690067887306213, + "num_tokens": 306202069.0, + "step": 13837 + }, + { + "epoch": 1.1410901294631812, + "grad_norm": 2.0852503776550293, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7722587585449219, + "num_tokens": 306230297.0, + "step": 13838 + }, + { + "epoch": 1.1411725900882328, + "grad_norm": 2.367960214614868, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.750545859336853, + "num_tokens": 306252095.0, + "step": 13839 + }, + { + "epoch": 1.1412550507132844, + "grad_norm": 1.9542338848114014, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7460352182388306, + "num_tokens": 306283045.0, + "step": 13840 + }, + { + "epoch": 1.141337511338336, + "grad_norm": 1.955840826034546, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7800708413124084, + "num_tokens": 306311765.0, + "step": 13841 + }, + { + "epoch": 1.1414199719633875, + "grad_norm": 1.9128631353378296, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7694697976112366, + "num_tokens": 306345274.0, + "step": 13842 + }, + { + "epoch": 1.141502432588439, + "grad_norm": 2.0457210540771484, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7361664772033691, + "num_tokens": 306376013.0, + "step": 13843 + }, + { + "epoch": 1.1415848932134907, + "grad_norm": 1.8147149085998535, + "learning_rate": 1e-06, + "loss": 0.7799, + "mean_token_accuracy": 0.7530692219734192, + "num_tokens": 306411108.0, + "step": 13844 + }, + { + "epoch": 1.141667353838542, + "grad_norm": 1.9026272296905518, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7567378282546997, + "num_tokens": 306442450.0, + "step": 13845 + }, + { + "epoch": 1.1417498144635936, + "grad_norm": 2.0409133434295654, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7594212293624878, + "num_tokens": 306469151.0, + "step": 13846 + }, + { + "epoch": 1.1418322750886452, + "grad_norm": 1.7325199842453003, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7703638672828674, + "num_tokens": 306508408.0, + "step": 13847 + }, + { + "epoch": 1.1419147357136967, + "grad_norm": 2.100250244140625, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.752906322479248, + "num_tokens": 306536032.0, + "step": 13848 + }, + { + "epoch": 1.1419971963387483, + "grad_norm": 2.0347511768341064, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.782702624797821, + "num_tokens": 306564734.0, + "step": 13849 + }, + { + "epoch": 1.1420796569637999, + "grad_norm": 1.9329431056976318, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7569648623466492, + "num_tokens": 306594556.0, + "step": 13850 + }, + { + "epoch": 1.1421621175888514, + "grad_norm": 2.0375375747680664, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7755904197692871, + "num_tokens": 306622222.0, + "step": 13851 + }, + { + "epoch": 1.1422445782139028, + "grad_norm": 2.1370701789855957, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7354612350463867, + "num_tokens": 306649534.0, + "step": 13852 + }, + { + "epoch": 1.1423270388389544, + "grad_norm": 1.9551334381103516, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7779123187065125, + "num_tokens": 306677324.0, + "step": 13853 + }, + { + "epoch": 1.142409499464006, + "grad_norm": 1.9906071424484253, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7566218376159668, + "num_tokens": 306713966.0, + "step": 13854 + }, + { + "epoch": 1.1424919600890575, + "grad_norm": 2.135385274887085, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7690862417221069, + "num_tokens": 306738480.0, + "step": 13855 + }, + { + "epoch": 1.142574420714109, + "grad_norm": 1.7966499328613281, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7413747310638428, + "num_tokens": 306772066.0, + "step": 13856 + }, + { + "epoch": 1.1426568813391604, + "grad_norm": 1.9330378770828247, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7740663886070251, + "num_tokens": 306801039.0, + "step": 13857 + }, + { + "epoch": 1.142739341964212, + "grad_norm": 2.5371055603027344, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7631372213363647, + "num_tokens": 306821545.0, + "step": 13858 + }, + { + "epoch": 1.1428218025892636, + "grad_norm": 2.116138458251953, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7432866096496582, + "num_tokens": 306848722.0, + "step": 13859 + }, + { + "epoch": 1.1429042632143152, + "grad_norm": 2.1414413452148438, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.7943108677864075, + "num_tokens": 306873627.0, + "step": 13860 + }, + { + "epoch": 1.1429867238393667, + "grad_norm": 2.0617847442626953, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7387983202934265, + "num_tokens": 306903383.0, + "step": 13861 + }, + { + "epoch": 1.1430691844644183, + "grad_norm": 2.096412420272827, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7765688896179199, + "num_tokens": 306928873.0, + "step": 13862 + }, + { + "epoch": 1.1431516450894699, + "grad_norm": 1.8215786218643188, + "learning_rate": 1e-06, + "loss": 0.5696, + "mean_token_accuracy": 0.8143495917320251, + "num_tokens": 306961852.0, + "step": 13863 + }, + { + "epoch": 1.1432341057145212, + "grad_norm": 2.2588186264038086, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7267675995826721, + "num_tokens": 306986872.0, + "step": 13864 + }, + { + "epoch": 1.1433165663395728, + "grad_norm": 2.0306508541107178, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7541804313659668, + "num_tokens": 307015012.0, + "step": 13865 + }, + { + "epoch": 1.1433990269646244, + "grad_norm": 1.9698278903961182, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7757605910301208, + "num_tokens": 307053828.0, + "step": 13866 + }, + { + "epoch": 1.143481487589676, + "grad_norm": 2.129748582839966, + "learning_rate": 1e-06, + "loss": 0.5638, + "mean_token_accuracy": 0.8129326701164246, + "num_tokens": 307079486.0, + "step": 13867 + }, + { + "epoch": 1.1435639482147275, + "grad_norm": 1.9652440547943115, + "learning_rate": 1e-06, + "loss": 0.7623, + "mean_token_accuracy": 0.7627179026603699, + "num_tokens": 307112571.0, + "step": 13868 + }, + { + "epoch": 1.143646408839779, + "grad_norm": 2.011430263519287, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7778885364532471, + "num_tokens": 307142766.0, + "step": 13869 + }, + { + "epoch": 1.1437288694648307, + "grad_norm": 2.0687313079833984, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7582683563232422, + "num_tokens": 307171819.0, + "step": 13870 + }, + { + "epoch": 1.143811330089882, + "grad_norm": 2.0429739952087402, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7619559168815613, + "num_tokens": 307202471.0, + "step": 13871 + }, + { + "epoch": 1.1438937907149336, + "grad_norm": 2.0359418392181396, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7346230745315552, + "num_tokens": 307234417.0, + "step": 13872 + }, + { + "epoch": 1.1439762513399852, + "grad_norm": 2.418280601501465, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7479243278503418, + "num_tokens": 307261819.0, + "step": 13873 + }, + { + "epoch": 1.1440587119650367, + "grad_norm": 2.4555039405822754, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7529301047325134, + "num_tokens": 307284912.0, + "step": 13874 + }, + { + "epoch": 1.1441411725900883, + "grad_norm": 2.2930727005004883, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7405616044998169, + "num_tokens": 307309052.0, + "step": 13875 + }, + { + "epoch": 1.1442236332151396, + "grad_norm": 1.6365538835525513, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.720660924911499, + "num_tokens": 307350578.0, + "step": 13876 + }, + { + "epoch": 1.1443060938401912, + "grad_norm": 2.4076528549194336, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7321738004684448, + "num_tokens": 307375124.0, + "step": 13877 + }, + { + "epoch": 1.1443885544652428, + "grad_norm": 2.05122709274292, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7631889581680298, + "num_tokens": 307404117.0, + "step": 13878 + }, + { + "epoch": 1.1444710150902944, + "grad_norm": 2.1266977787017822, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7454657554626465, + "num_tokens": 307434021.0, + "step": 13879 + }, + { + "epoch": 1.144553475715346, + "grad_norm": 1.9671281576156616, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7566966414451599, + "num_tokens": 307465849.0, + "step": 13880 + }, + { + "epoch": 1.1446359363403975, + "grad_norm": 2.0394539833068848, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6962573528289795, + "num_tokens": 307497739.0, + "step": 13881 + }, + { + "epoch": 1.144718396965449, + "grad_norm": 2.3673386573791504, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7435219287872314, + "num_tokens": 307522745.0, + "step": 13882 + }, + { + "epoch": 1.1448008575905004, + "grad_norm": 1.989102840423584, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7561361789703369, + "num_tokens": 307554931.0, + "step": 13883 + }, + { + "epoch": 1.144883318215552, + "grad_norm": 1.769381046295166, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7637913227081299, + "num_tokens": 307590698.0, + "step": 13884 + }, + { + "epoch": 1.1449657788406036, + "grad_norm": 2.2758283615112305, + "learning_rate": 1e-06, + "loss": 0.6112, + "mean_token_accuracy": 0.790357232093811, + "num_tokens": 307612341.0, + "step": 13885 + }, + { + "epoch": 1.1450482394656551, + "grad_norm": 1.9692645072937012, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.7667879462242126, + "num_tokens": 307641451.0, + "step": 13886 + }, + { + "epoch": 1.1451307000907067, + "grad_norm": 1.999327301979065, + "learning_rate": 1e-06, + "loss": 0.5424, + "mean_token_accuracy": 0.8120452761650085, + "num_tokens": 307668330.0, + "step": 13887 + }, + { + "epoch": 1.1452131607157583, + "grad_norm": 2.1704444885253906, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7606872320175171, + "num_tokens": 307692434.0, + "step": 13888 + }, + { + "epoch": 1.1452956213408099, + "grad_norm": 1.9662747383117676, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7385308742523193, + "num_tokens": 307724831.0, + "step": 13889 + }, + { + "epoch": 1.1453780819658612, + "grad_norm": 2.0245814323425293, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7508695721626282, + "num_tokens": 307757338.0, + "step": 13890 + }, + { + "epoch": 1.1454605425909128, + "grad_norm": 2.0720996856689453, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.7866458892822266, + "num_tokens": 307781774.0, + "step": 13891 + }, + { + "epoch": 1.1455430032159644, + "grad_norm": 1.8318934440612793, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7753680944442749, + "num_tokens": 307813155.0, + "step": 13892 + }, + { + "epoch": 1.145625463841016, + "grad_norm": 1.8650286197662354, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.7956918478012085, + "num_tokens": 307843788.0, + "step": 13893 + }, + { + "epoch": 1.1457079244660675, + "grad_norm": 2.253167152404785, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7507920265197754, + "num_tokens": 307879667.0, + "step": 13894 + }, + { + "epoch": 1.145790385091119, + "grad_norm": 1.9962431192398071, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7465808987617493, + "num_tokens": 307909710.0, + "step": 13895 + }, + { + "epoch": 1.1458728457161707, + "grad_norm": 1.9305739402770996, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.792269766330719, + "num_tokens": 307941972.0, + "step": 13896 + }, + { + "epoch": 1.145955306341222, + "grad_norm": 1.7950040102005005, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7803071737289429, + "num_tokens": 307978023.0, + "step": 13897 + }, + { + "epoch": 1.1460377669662736, + "grad_norm": 1.8341845273971558, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7751369476318359, + "num_tokens": 308012141.0, + "step": 13898 + }, + { + "epoch": 1.1461202275913251, + "grad_norm": 2.1699588298797607, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7711837291717529, + "num_tokens": 308040658.0, + "step": 13899 + }, + { + "epoch": 1.1462026882163767, + "grad_norm": 1.8956091403961182, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7734096646308899, + "num_tokens": 308073139.0, + "step": 13900 + }, + { + "epoch": 1.1462851488414283, + "grad_norm": 1.815634846687317, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7461397051811218, + "num_tokens": 308107719.0, + "step": 13901 + }, + { + "epoch": 1.1463676094664796, + "grad_norm": 2.018828868865967, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.760918378829956, + "num_tokens": 308137442.0, + "step": 13902 + }, + { + "epoch": 1.1464500700915312, + "grad_norm": 1.9413998126983643, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7626586556434631, + "num_tokens": 308167861.0, + "step": 13903 + }, + { + "epoch": 1.1465325307165828, + "grad_norm": 1.8770060539245605, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.7888748049736023, + "num_tokens": 308201181.0, + "step": 13904 + }, + { + "epoch": 1.1466149913416344, + "grad_norm": 2.1069111824035645, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7684821486473083, + "num_tokens": 308227689.0, + "step": 13905 + }, + { + "epoch": 1.146697451966686, + "grad_norm": 1.9581069946289062, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7802391648292542, + "num_tokens": 308262319.0, + "step": 13906 + }, + { + "epoch": 1.1467799125917375, + "grad_norm": 2.078511953353882, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.777468740940094, + "num_tokens": 308289534.0, + "step": 13907 + }, + { + "epoch": 1.146862373216789, + "grad_norm": 2.356144666671753, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7238304615020752, + "num_tokens": 308313396.0, + "step": 13908 + }, + { + "epoch": 1.1469448338418404, + "grad_norm": 2.274660348892212, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7553391456604004, + "num_tokens": 308337527.0, + "step": 13909 + }, + { + "epoch": 1.147027294466892, + "grad_norm": 1.9204710721969604, + "learning_rate": 1e-06, + "loss": 0.5877, + "mean_token_accuracy": 0.7932605147361755, + "num_tokens": 308366040.0, + "step": 13910 + }, + { + "epoch": 1.1471097550919436, + "grad_norm": 2.1059298515319824, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.7914533615112305, + "num_tokens": 308394132.0, + "step": 13911 + }, + { + "epoch": 1.1471922157169951, + "grad_norm": 1.8640179634094238, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7517389059066772, + "num_tokens": 308428496.0, + "step": 13912 + }, + { + "epoch": 1.1472746763420467, + "grad_norm": 2.021374464035034, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7880383729934692, + "num_tokens": 308459260.0, + "step": 13913 + }, + { + "epoch": 1.1473571369670983, + "grad_norm": 1.9621168375015259, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7258660793304443, + "num_tokens": 308494865.0, + "step": 13914 + }, + { + "epoch": 1.1474395975921499, + "grad_norm": 2.20957088470459, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7371432185173035, + "num_tokens": 308521207.0, + "step": 13915 + }, + { + "epoch": 1.1475220582172012, + "grad_norm": 2.024258852005005, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7577915191650391, + "num_tokens": 308553377.0, + "step": 13916 + }, + { + "epoch": 1.1476045188422528, + "grad_norm": 1.9492548704147339, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7591285109519958, + "num_tokens": 308583134.0, + "step": 13917 + }, + { + "epoch": 1.1476869794673044, + "grad_norm": 1.6403512954711914, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7713428139686584, + "num_tokens": 308624113.0, + "step": 13918 + }, + { + "epoch": 1.147769440092356, + "grad_norm": 2.033811092376709, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7801054120063782, + "num_tokens": 308652082.0, + "step": 13919 + }, + { + "epoch": 1.1478519007174075, + "grad_norm": 1.941112995147705, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.7908746004104614, + "num_tokens": 308683769.0, + "step": 13920 + }, + { + "epoch": 1.1479343613424589, + "grad_norm": 1.870240330696106, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7735078930854797, + "num_tokens": 308713868.0, + "step": 13921 + }, + { + "epoch": 1.1480168219675104, + "grad_norm": 2.1102867126464844, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7698287963867188, + "num_tokens": 308740571.0, + "step": 13922 + }, + { + "epoch": 1.148099282592562, + "grad_norm": 1.6277506351470947, + "learning_rate": 1e-06, + "loss": 0.5847, + "mean_token_accuracy": 0.8080394864082336, + "num_tokens": 308781268.0, + "step": 13923 + }, + { + "epoch": 1.1481817432176136, + "grad_norm": 2.022643804550171, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.7839798927307129, + "num_tokens": 308810819.0, + "step": 13924 + }, + { + "epoch": 1.1482642038426651, + "grad_norm": 1.806060552597046, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.716649055480957, + "num_tokens": 308848187.0, + "step": 13925 + }, + { + "epoch": 1.1483466644677167, + "grad_norm": 1.892224907875061, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7522683143615723, + "num_tokens": 308881472.0, + "step": 13926 + }, + { + "epoch": 1.1484291250927683, + "grad_norm": 2.196721076965332, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7262111306190491, + "num_tokens": 308908887.0, + "step": 13927 + }, + { + "epoch": 1.1485115857178196, + "grad_norm": 1.8264117240905762, + "learning_rate": 1e-06, + "loss": 0.5854, + "mean_token_accuracy": 0.8064683675765991, + "num_tokens": 308942848.0, + "step": 13928 + }, + { + "epoch": 1.1485940463428712, + "grad_norm": 2.335577964782715, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7728722095489502, + "num_tokens": 308965424.0, + "step": 13929 + }, + { + "epoch": 1.1486765069679228, + "grad_norm": 1.8165221214294434, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7745264768600464, + "num_tokens": 309002440.0, + "step": 13930 + }, + { + "epoch": 1.1487589675929744, + "grad_norm": 1.97456693649292, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7685264348983765, + "num_tokens": 309033396.0, + "step": 13931 + }, + { + "epoch": 1.148841428218026, + "grad_norm": 2.287212610244751, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7556788325309753, + "num_tokens": 309056698.0, + "step": 13932 + }, + { + "epoch": 1.1489238888430775, + "grad_norm": 1.8672564029693604, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.76957768201828, + "num_tokens": 309091096.0, + "step": 13933 + }, + { + "epoch": 1.149006349468129, + "grad_norm": 1.8537235260009766, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7770118117332458, + "num_tokens": 309124657.0, + "step": 13934 + }, + { + "epoch": 1.1490888100931804, + "grad_norm": 2.3162693977355957, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7325146198272705, + "num_tokens": 309149805.0, + "step": 13935 + }, + { + "epoch": 1.149171270718232, + "grad_norm": 2.2239599227905273, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.784712553024292, + "num_tokens": 309174416.0, + "step": 13936 + }, + { + "epoch": 1.1492537313432836, + "grad_norm": 2.430717706680298, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7339307069778442, + "num_tokens": 309198393.0, + "step": 13937 + }, + { + "epoch": 1.1493361919683351, + "grad_norm": 1.7431813478469849, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.7929739952087402, + "num_tokens": 309230944.0, + "step": 13938 + }, + { + "epoch": 1.1494186525933867, + "grad_norm": 1.978863000869751, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.773186206817627, + "num_tokens": 309260854.0, + "step": 13939 + }, + { + "epoch": 1.1495011132184383, + "grad_norm": 1.7504836320877075, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7454038858413696, + "num_tokens": 309298755.0, + "step": 13940 + }, + { + "epoch": 1.1495835738434896, + "grad_norm": 1.8590224981307983, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7530144453048706, + "num_tokens": 309331870.0, + "step": 13941 + }, + { + "epoch": 1.1496660344685412, + "grad_norm": 2.231252908706665, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7677111625671387, + "num_tokens": 309357155.0, + "step": 13942 + }, + { + "epoch": 1.1497484950935928, + "grad_norm": 1.9660931825637817, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7577271461486816, + "num_tokens": 309389551.0, + "step": 13943 + }, + { + "epoch": 1.1498309557186444, + "grad_norm": 2.300525426864624, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7453784346580505, + "num_tokens": 309418513.0, + "step": 13944 + }, + { + "epoch": 1.149913416343696, + "grad_norm": 1.9702945947647095, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7692183256149292, + "num_tokens": 309451315.0, + "step": 13945 + }, + { + "epoch": 1.1499958769687475, + "grad_norm": 2.0308361053466797, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.75994873046875, + "num_tokens": 309480589.0, + "step": 13946 + }, + { + "epoch": 1.1500783375937988, + "grad_norm": 1.8852876424789429, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.767838716506958, + "num_tokens": 309513681.0, + "step": 13947 + }, + { + "epoch": 1.1501607982188504, + "grad_norm": 2.2513740062713623, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7312257289886475, + "num_tokens": 309536695.0, + "step": 13948 + }, + { + "epoch": 1.150243258843902, + "grad_norm": 2.0325002670288086, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7467919588088989, + "num_tokens": 309566599.0, + "step": 13949 + }, + { + "epoch": 1.1503257194689536, + "grad_norm": 1.7051293849945068, + "learning_rate": 1e-06, + "loss": 0.6041, + "mean_token_accuracy": 0.79850172996521, + "num_tokens": 309601130.0, + "step": 13950 + }, + { + "epoch": 1.1504081800940051, + "grad_norm": 1.8655966520309448, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.7999477386474609, + "num_tokens": 309631898.0, + "step": 13951 + }, + { + "epoch": 1.1504906407190567, + "grad_norm": 2.078925848007202, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7571333050727844, + "num_tokens": 309660702.0, + "step": 13952 + }, + { + "epoch": 1.1505731013441083, + "grad_norm": 1.796323299407959, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.78443843126297, + "num_tokens": 309693774.0, + "step": 13953 + }, + { + "epoch": 1.1506555619691596, + "grad_norm": 1.983173131942749, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7395551204681396, + "num_tokens": 309726190.0, + "step": 13954 + }, + { + "epoch": 1.1507380225942112, + "grad_norm": 1.9935632944107056, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.7924737930297852, + "num_tokens": 309763274.0, + "step": 13955 + }, + { + "epoch": 1.1508204832192628, + "grad_norm": 2.0720980167388916, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.8027890920639038, + "num_tokens": 309789701.0, + "step": 13956 + }, + { + "epoch": 1.1509029438443144, + "grad_norm": 2.0594263076782227, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7788341045379639, + "num_tokens": 309818042.0, + "step": 13957 + }, + { + "epoch": 1.150985404469366, + "grad_norm": 1.9967231750488281, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7537850737571716, + "num_tokens": 309847457.0, + "step": 13958 + }, + { + "epoch": 1.1510678650944175, + "grad_norm": 1.9089571237564087, + "learning_rate": 1e-06, + "loss": 0.7808, + "mean_token_accuracy": 0.7540998458862305, + "num_tokens": 309878830.0, + "step": 13959 + }, + { + "epoch": 1.151150325719469, + "grad_norm": 2.0412662029266357, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.775118350982666, + "num_tokens": 309906375.0, + "step": 13960 + }, + { + "epoch": 1.1512327863445204, + "grad_norm": 2.0848946571350098, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7758639454841614, + "num_tokens": 309932689.0, + "step": 13961 + }, + { + "epoch": 1.151315246969572, + "grad_norm": 2.0235016345977783, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7667233943939209, + "num_tokens": 309959958.0, + "step": 13962 + }, + { + "epoch": 1.1513977075946236, + "grad_norm": 2.012742042541504, + "learning_rate": 1e-06, + "loss": 0.784, + "mean_token_accuracy": 0.7596180438995361, + "num_tokens": 309990879.0, + "step": 13963 + }, + { + "epoch": 1.1514801682196751, + "grad_norm": 1.9420092105865479, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.752549409866333, + "num_tokens": 310022423.0, + "step": 13964 + }, + { + "epoch": 1.1515626288447267, + "grad_norm": 2.2147281169891357, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7441543936729431, + "num_tokens": 310050153.0, + "step": 13965 + }, + { + "epoch": 1.151645089469778, + "grad_norm": 1.7720274925231934, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7478383779525757, + "num_tokens": 310089703.0, + "step": 13966 + }, + { + "epoch": 1.1517275500948296, + "grad_norm": 2.0197432041168213, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7449352741241455, + "num_tokens": 310117369.0, + "step": 13967 + }, + { + "epoch": 1.1518100107198812, + "grad_norm": 1.6812552213668823, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7740795612335205, + "num_tokens": 310155833.0, + "step": 13968 + }, + { + "epoch": 1.1518924713449328, + "grad_norm": 1.892369270324707, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7487298846244812, + "num_tokens": 310191312.0, + "step": 13969 + }, + { + "epoch": 1.1519749319699844, + "grad_norm": 2.1030147075653076, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.7503105998039246, + "num_tokens": 310218780.0, + "step": 13970 + }, + { + "epoch": 1.152057392595036, + "grad_norm": 1.9448024034500122, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7522981762886047, + "num_tokens": 310248268.0, + "step": 13971 + }, + { + "epoch": 1.1521398532200875, + "grad_norm": 2.208099842071533, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7618340253829956, + "num_tokens": 310272488.0, + "step": 13972 + }, + { + "epoch": 1.1522223138451388, + "grad_norm": 1.985459804534912, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7375756502151489, + "num_tokens": 310303233.0, + "step": 13973 + }, + { + "epoch": 1.1523047744701904, + "grad_norm": 2.285079002380371, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7404593229293823, + "num_tokens": 310327227.0, + "step": 13974 + }, + { + "epoch": 1.152387235095242, + "grad_norm": 1.771166205406189, + "learning_rate": 1e-06, + "loss": 0.4502, + "mean_token_accuracy": 0.8425367474555969, + "num_tokens": 310358020.0, + "step": 13975 + }, + { + "epoch": 1.1524696957202936, + "grad_norm": 2.050297975540161, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.747612476348877, + "num_tokens": 310388063.0, + "step": 13976 + }, + { + "epoch": 1.1525521563453451, + "grad_norm": 1.9465190172195435, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7371140718460083, + "num_tokens": 310420145.0, + "step": 13977 + }, + { + "epoch": 1.1526346169703967, + "grad_norm": 2.0418851375579834, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7376805543899536, + "num_tokens": 310450358.0, + "step": 13978 + }, + { + "epoch": 1.1527170775954483, + "grad_norm": 2.626793622970581, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7511315941810608, + "num_tokens": 310471983.0, + "step": 13979 + }, + { + "epoch": 1.1527995382204996, + "grad_norm": 1.9462060928344727, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7132951617240906, + "num_tokens": 310505681.0, + "step": 13980 + }, + { + "epoch": 1.1528819988455512, + "grad_norm": 1.7755436897277832, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7432129979133606, + "num_tokens": 310546883.0, + "step": 13981 + }, + { + "epoch": 1.1529644594706028, + "grad_norm": 1.9662730693817139, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.746525228023529, + "num_tokens": 310576721.0, + "step": 13982 + }, + { + "epoch": 1.1530469200956543, + "grad_norm": 2.061495542526245, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.774498462677002, + "num_tokens": 310604358.0, + "step": 13983 + }, + { + "epoch": 1.153129380720706, + "grad_norm": 1.7357178926467896, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.775610625743866, + "num_tokens": 310641720.0, + "step": 13984 + }, + { + "epoch": 1.1532118413457575, + "grad_norm": 2.0899174213409424, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7581663727760315, + "num_tokens": 310668603.0, + "step": 13985 + }, + { + "epoch": 1.1532943019708088, + "grad_norm": 1.880592703819275, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7718034386634827, + "num_tokens": 310700245.0, + "step": 13986 + }, + { + "epoch": 1.1533767625958604, + "grad_norm": 1.744751214981079, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.7579984664916992, + "num_tokens": 310735591.0, + "step": 13987 + }, + { + "epoch": 1.153459223220912, + "grad_norm": 2.1355700492858887, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7141933441162109, + "num_tokens": 310764299.0, + "step": 13988 + }, + { + "epoch": 1.1535416838459636, + "grad_norm": 1.9163330793380737, + "learning_rate": 1e-06, + "loss": 0.6221, + "mean_token_accuracy": 0.7912311553955078, + "num_tokens": 310797213.0, + "step": 13989 + }, + { + "epoch": 1.1536241444710151, + "grad_norm": 2.1941094398498535, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7551330924034119, + "num_tokens": 310823579.0, + "step": 13990 + }, + { + "epoch": 1.1537066050960667, + "grad_norm": 2.1066884994506836, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7687408924102783, + "num_tokens": 310852953.0, + "step": 13991 + }, + { + "epoch": 1.153789065721118, + "grad_norm": 2.4219765663146973, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7339168190956116, + "num_tokens": 310872489.0, + "step": 13992 + }, + { + "epoch": 1.1538715263461696, + "grad_norm": 1.9423655271530151, + "learning_rate": 1e-06, + "loss": 0.5815, + "mean_token_accuracy": 0.8112802505493164, + "num_tokens": 310901011.0, + "step": 13993 + }, + { + "epoch": 1.1539539869712212, + "grad_norm": 1.845393180847168, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7525749206542969, + "num_tokens": 310935948.0, + "step": 13994 + }, + { + "epoch": 1.1540364475962728, + "grad_norm": 2.1904525756835938, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7219182252883911, + "num_tokens": 310962979.0, + "step": 13995 + }, + { + "epoch": 1.1541189082213243, + "grad_norm": 1.9035652875900269, + "learning_rate": 1e-06, + "loss": 0.5802, + "mean_token_accuracy": 0.8007653951644897, + "num_tokens": 310997186.0, + "step": 13996 + }, + { + "epoch": 1.154201368846376, + "grad_norm": 2.036583423614502, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.7935574054718018, + "num_tokens": 311029517.0, + "step": 13997 + }, + { + "epoch": 1.1542838294714275, + "grad_norm": 1.6738802194595337, + "learning_rate": 1e-06, + "loss": 0.6077, + "mean_token_accuracy": 0.7971897721290588, + "num_tokens": 311067465.0, + "step": 13998 + }, + { + "epoch": 1.1543662900964788, + "grad_norm": 1.8413785696029663, + "learning_rate": 1e-06, + "loss": 0.587, + "mean_token_accuracy": 0.8073403835296631, + "num_tokens": 311096272.0, + "step": 13999 + }, + { + "epoch": 1.1544487507215304, + "grad_norm": 1.75746750831604, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7664414048194885, + "num_tokens": 311133793.0, + "step": 14000 + }, + { + "epoch": 1.154531211346582, + "grad_norm": 2.3041484355926514, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.726809024810791, + "num_tokens": 311158463.0, + "step": 14001 + }, + { + "epoch": 1.1546136719716336, + "grad_norm": 1.9131118059158325, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7609525918960571, + "num_tokens": 311191388.0, + "step": 14002 + }, + { + "epoch": 1.1546961325966851, + "grad_norm": 1.923203706741333, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7537092566490173, + "num_tokens": 311224195.0, + "step": 14003 + }, + { + "epoch": 1.1547785932217367, + "grad_norm": 2.024423837661743, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7649553418159485, + "num_tokens": 311254089.0, + "step": 14004 + }, + { + "epoch": 1.1548610538467883, + "grad_norm": 2.051091432571411, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7612524032592773, + "num_tokens": 311284552.0, + "step": 14005 + }, + { + "epoch": 1.1549435144718396, + "grad_norm": 1.9273420572280884, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7393783330917358, + "num_tokens": 311317693.0, + "step": 14006 + }, + { + "epoch": 1.1550259750968912, + "grad_norm": 2.1344635486602783, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.7555075883865356, + "num_tokens": 311346803.0, + "step": 14007 + }, + { + "epoch": 1.1551084357219428, + "grad_norm": 2.178798198699951, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7542941570281982, + "num_tokens": 311374092.0, + "step": 14008 + }, + { + "epoch": 1.1551908963469943, + "grad_norm": 1.8594915866851807, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7637743353843689, + "num_tokens": 311411344.0, + "step": 14009 + }, + { + "epoch": 1.155273356972046, + "grad_norm": 1.8765102624893188, + "learning_rate": 1e-06, + "loss": 0.5765, + "mean_token_accuracy": 0.8010674715042114, + "num_tokens": 311446586.0, + "step": 14010 + }, + { + "epoch": 1.1553558175970973, + "grad_norm": 2.0325119495391846, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7622227668762207, + "num_tokens": 311478651.0, + "step": 14011 + }, + { + "epoch": 1.1554382782221488, + "grad_norm": 2.1477978229522705, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.7903938293457031, + "num_tokens": 311504026.0, + "step": 14012 + }, + { + "epoch": 1.1555207388472004, + "grad_norm": 1.957224726676941, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7516003847122192, + "num_tokens": 311535850.0, + "step": 14013 + }, + { + "epoch": 1.155603199472252, + "grad_norm": 2.0527517795562744, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7549811601638794, + "num_tokens": 311567543.0, + "step": 14014 + }, + { + "epoch": 1.1556856600973036, + "grad_norm": 1.9324181079864502, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7450711131095886, + "num_tokens": 311600764.0, + "step": 14015 + }, + { + "epoch": 1.1557681207223551, + "grad_norm": 1.7969672679901123, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7828267812728882, + "num_tokens": 311634852.0, + "step": 14016 + }, + { + "epoch": 1.1558505813474067, + "grad_norm": 1.9983594417572021, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.7555484175682068, + "num_tokens": 311663224.0, + "step": 14017 + }, + { + "epoch": 1.155933041972458, + "grad_norm": 1.6663415431976318, + "learning_rate": 1e-06, + "loss": 0.5855, + "mean_token_accuracy": 0.8030106425285339, + "num_tokens": 311700309.0, + "step": 14018 + }, + { + "epoch": 1.1560155025975096, + "grad_norm": 1.8862279653549194, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7244319915771484, + "num_tokens": 311731887.0, + "step": 14019 + }, + { + "epoch": 1.1560979632225612, + "grad_norm": 1.8500183820724487, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.773679792881012, + "num_tokens": 311762242.0, + "step": 14020 + }, + { + "epoch": 1.1561804238476128, + "grad_norm": 2.0696187019348145, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.754885196685791, + "num_tokens": 311791129.0, + "step": 14021 + }, + { + "epoch": 1.1562628844726643, + "grad_norm": 2.185045003890991, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7490466237068176, + "num_tokens": 311816710.0, + "step": 14022 + }, + { + "epoch": 1.156345345097716, + "grad_norm": 1.8937695026397705, + "learning_rate": 1e-06, + "loss": 0.7883, + "mean_token_accuracy": 0.7508851289749146, + "num_tokens": 311852939.0, + "step": 14023 + }, + { + "epoch": 1.1564278057227675, + "grad_norm": 2.0516505241394043, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7541525363922119, + "num_tokens": 311883889.0, + "step": 14024 + }, + { + "epoch": 1.1565102663478188, + "grad_norm": 2.025879383087158, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7309533953666687, + "num_tokens": 311914782.0, + "step": 14025 + }, + { + "epoch": 1.1565927269728704, + "grad_norm": 1.7966262102127075, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7562590837478638, + "num_tokens": 311950283.0, + "step": 14026 + }, + { + "epoch": 1.156675187597922, + "grad_norm": 2.2048492431640625, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7516089677810669, + "num_tokens": 311978956.0, + "step": 14027 + }, + { + "epoch": 1.1567576482229736, + "grad_norm": 2.113492727279663, + "learning_rate": 1e-06, + "loss": 0.5877, + "mean_token_accuracy": 0.798240602016449, + "num_tokens": 312004312.0, + "step": 14028 + }, + { + "epoch": 1.1568401088480251, + "grad_norm": 1.8184764385223389, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.764858067035675, + "num_tokens": 312038296.0, + "step": 14029 + }, + { + "epoch": 1.1569225694730767, + "grad_norm": 1.7837331295013428, + "learning_rate": 1e-06, + "loss": 0.5212, + "mean_token_accuracy": 0.8182963132858276, + "num_tokens": 312073920.0, + "step": 14030 + }, + { + "epoch": 1.157005030098128, + "grad_norm": 2.25936222076416, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7490096092224121, + "num_tokens": 312097856.0, + "step": 14031 + }, + { + "epoch": 1.1570874907231796, + "grad_norm": 2.0008182525634766, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7773140668869019, + "num_tokens": 312126600.0, + "step": 14032 + }, + { + "epoch": 1.1571699513482312, + "grad_norm": 1.9852101802825928, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.7687944173812866, + "num_tokens": 312155879.0, + "step": 14033 + }, + { + "epoch": 1.1572524119732828, + "grad_norm": 1.847500205039978, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7470636367797852, + "num_tokens": 312190893.0, + "step": 14034 + }, + { + "epoch": 1.1573348725983343, + "grad_norm": 1.893673062324524, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7482982277870178, + "num_tokens": 312226495.0, + "step": 14035 + }, + { + "epoch": 1.157417333223386, + "grad_norm": 2.2521588802337646, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7377891540527344, + "num_tokens": 312250212.0, + "step": 14036 + }, + { + "epoch": 1.1574997938484373, + "grad_norm": 2.301107406616211, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7568589448928833, + "num_tokens": 312275782.0, + "step": 14037 + }, + { + "epoch": 1.1575822544734888, + "grad_norm": 2.0323822498321533, + "learning_rate": 1e-06, + "loss": 0.7303, + "mean_token_accuracy": 0.7641658782958984, + "num_tokens": 312304930.0, + "step": 14038 + }, + { + "epoch": 1.1576647150985404, + "grad_norm": 2.315762758255005, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7703355550765991, + "num_tokens": 312327697.0, + "step": 14039 + }, + { + "epoch": 1.157747175723592, + "grad_norm": 2.0851922035217285, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7591419219970703, + "num_tokens": 312355245.0, + "step": 14040 + }, + { + "epoch": 1.1578296363486436, + "grad_norm": 1.9648706912994385, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.7943921685218811, + "num_tokens": 312389038.0, + "step": 14041 + }, + { + "epoch": 1.1579120969736951, + "grad_norm": 2.1909172534942627, + "learning_rate": 1e-06, + "loss": 0.7868, + "mean_token_accuracy": 0.7412782311439514, + "num_tokens": 312416519.0, + "step": 14042 + }, + { + "epoch": 1.1579945575987467, + "grad_norm": 2.0335962772369385, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.769981861114502, + "num_tokens": 312445091.0, + "step": 14043 + }, + { + "epoch": 1.158077018223798, + "grad_norm": 2.3099541664123535, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.732140839099884, + "num_tokens": 312470115.0, + "step": 14044 + }, + { + "epoch": 1.1581594788488496, + "grad_norm": 2.021630048751831, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.7844472527503967, + "num_tokens": 312498919.0, + "step": 14045 + }, + { + "epoch": 1.1582419394739012, + "grad_norm": 1.9585272073745728, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7596022486686707, + "num_tokens": 312529300.0, + "step": 14046 + }, + { + "epoch": 1.1583244000989528, + "grad_norm": 2.192262887954712, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7298185229301453, + "num_tokens": 312554719.0, + "step": 14047 + }, + { + "epoch": 1.1584068607240043, + "grad_norm": 2.1362709999084473, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7820538282394409, + "num_tokens": 312580770.0, + "step": 14048 + }, + { + "epoch": 1.158489321349056, + "grad_norm": 2.0778045654296875, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7649108171463013, + "num_tokens": 312610371.0, + "step": 14049 + }, + { + "epoch": 1.1585717819741075, + "grad_norm": 2.2076451778411865, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7310348749160767, + "num_tokens": 312638792.0, + "step": 14050 + }, + { + "epoch": 1.1586542425991588, + "grad_norm": 1.9530034065246582, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.7969039678573608, + "num_tokens": 312671383.0, + "step": 14051 + }, + { + "epoch": 1.1587367032242104, + "grad_norm": 1.7498987913131714, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.7912998795509338, + "num_tokens": 312704720.0, + "step": 14052 + }, + { + "epoch": 1.158819163849262, + "grad_norm": 1.892778754234314, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7697497010231018, + "num_tokens": 312735992.0, + "step": 14053 + }, + { + "epoch": 1.1589016244743136, + "grad_norm": 1.8406282663345337, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.770982027053833, + "num_tokens": 312768148.0, + "step": 14054 + }, + { + "epoch": 1.1589840850993651, + "grad_norm": 1.91786789894104, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.775433361530304, + "num_tokens": 312800362.0, + "step": 14055 + }, + { + "epoch": 1.1590665457244165, + "grad_norm": 1.844512701034546, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7719923257827759, + "num_tokens": 312837308.0, + "step": 14056 + }, + { + "epoch": 1.159149006349468, + "grad_norm": 2.0958802700042725, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7405293583869934, + "num_tokens": 312863881.0, + "step": 14057 + }, + { + "epoch": 1.1592314669745196, + "grad_norm": 1.8950016498565674, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.753479540348053, + "num_tokens": 312895941.0, + "step": 14058 + }, + { + "epoch": 1.1593139275995712, + "grad_norm": 1.9858452081680298, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7671982049942017, + "num_tokens": 312926334.0, + "step": 14059 + }, + { + "epoch": 1.1593963882246228, + "grad_norm": 2.062981367111206, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7714397311210632, + "num_tokens": 312951978.0, + "step": 14060 + }, + { + "epoch": 1.1594788488496743, + "grad_norm": 1.799227237701416, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7765804529190063, + "num_tokens": 312987820.0, + "step": 14061 + }, + { + "epoch": 1.159561309474726, + "grad_norm": 2.263590097427368, + "learning_rate": 1e-06, + "loss": 0.5574, + "mean_token_accuracy": 0.8004732728004456, + "num_tokens": 313013700.0, + "step": 14062 + }, + { + "epoch": 1.1596437700997773, + "grad_norm": 2.2687132358551025, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7391568422317505, + "num_tokens": 313040488.0, + "step": 14063 + }, + { + "epoch": 1.1597262307248288, + "grad_norm": 1.9025572538375854, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7642382383346558, + "num_tokens": 313077226.0, + "step": 14064 + }, + { + "epoch": 1.1598086913498804, + "grad_norm": 2.2882421016693115, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7175959944725037, + "num_tokens": 313105674.0, + "step": 14065 + }, + { + "epoch": 1.159891151974932, + "grad_norm": 2.1234002113342285, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.7833989858627319, + "num_tokens": 313129026.0, + "step": 14066 + }, + { + "epoch": 1.1599736125999836, + "grad_norm": 2.078848123550415, + "learning_rate": 1e-06, + "loss": 0.7465, + "mean_token_accuracy": 0.7646598815917969, + "num_tokens": 313159013.0, + "step": 14067 + }, + { + "epoch": 1.1600560732250351, + "grad_norm": 1.674079418182373, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7680907249450684, + "num_tokens": 313195753.0, + "step": 14068 + }, + { + "epoch": 1.1601385338500867, + "grad_norm": 1.925734043121338, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7645769715309143, + "num_tokens": 313225010.0, + "step": 14069 + }, + { + "epoch": 1.160220994475138, + "grad_norm": 1.994908332824707, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.7913932800292969, + "num_tokens": 313252617.0, + "step": 14070 + }, + { + "epoch": 1.1603034551001896, + "grad_norm": 2.047100782394409, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7357084155082703, + "num_tokens": 313283684.0, + "step": 14071 + }, + { + "epoch": 1.1603859157252412, + "grad_norm": 2.0215697288513184, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7772715091705322, + "num_tokens": 313315379.0, + "step": 14072 + }, + { + "epoch": 1.1604683763502928, + "grad_norm": 2.2164359092712402, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7681303024291992, + "num_tokens": 313341045.0, + "step": 14073 + }, + { + "epoch": 1.1605508369753443, + "grad_norm": 2.1048738956451416, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7336297035217285, + "num_tokens": 313367817.0, + "step": 14074 + }, + { + "epoch": 1.160633297600396, + "grad_norm": 2.0207877159118652, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7399387955665588, + "num_tokens": 313397793.0, + "step": 14075 + }, + { + "epoch": 1.1607157582254473, + "grad_norm": 1.9897615909576416, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7784912586212158, + "num_tokens": 313427038.0, + "step": 14076 + }, + { + "epoch": 1.1607982188504988, + "grad_norm": 1.8997347354888916, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7568559646606445, + "num_tokens": 313457918.0, + "step": 14077 + }, + { + "epoch": 1.1608806794755504, + "grad_norm": 1.7011735439300537, + "learning_rate": 1e-06, + "loss": 0.6257, + "mean_token_accuracy": 0.7956589460372925, + "num_tokens": 313496488.0, + "step": 14078 + }, + { + "epoch": 1.160963140100602, + "grad_norm": 2.0341603755950928, + "learning_rate": 1e-06, + "loss": 0.7617, + "mean_token_accuracy": 0.7493665814399719, + "num_tokens": 313527749.0, + "step": 14079 + }, + { + "epoch": 1.1610456007256535, + "grad_norm": 1.9993102550506592, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7371516823768616, + "num_tokens": 313557499.0, + "step": 14080 + }, + { + "epoch": 1.1611280613507051, + "grad_norm": 1.8828343152999878, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.7986518144607544, + "num_tokens": 313588731.0, + "step": 14081 + }, + { + "epoch": 1.1612105219757565, + "grad_norm": 1.7631748914718628, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.8004894852638245, + "num_tokens": 313622722.0, + "step": 14082 + }, + { + "epoch": 1.161292982600808, + "grad_norm": 1.9302027225494385, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.7939673662185669, + "num_tokens": 313654963.0, + "step": 14083 + }, + { + "epoch": 1.1613754432258596, + "grad_norm": 1.8336137533187866, + "learning_rate": 1e-06, + "loss": 0.6229, + "mean_token_accuracy": 0.7983538508415222, + "num_tokens": 313685657.0, + "step": 14084 + }, + { + "epoch": 1.1614579038509112, + "grad_norm": 1.8669183254241943, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7576730251312256, + "num_tokens": 313716905.0, + "step": 14085 + }, + { + "epoch": 1.1615403644759628, + "grad_norm": 1.8345087766647339, + "learning_rate": 1e-06, + "loss": 0.5662, + "mean_token_accuracy": 0.8123398423194885, + "num_tokens": 313747973.0, + "step": 14086 + }, + { + "epoch": 1.1616228251010143, + "grad_norm": 2.1142091751098633, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7495681047439575, + "num_tokens": 313774693.0, + "step": 14087 + }, + { + "epoch": 1.161705285726066, + "grad_norm": 1.9629701375961304, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7515295743942261, + "num_tokens": 313805911.0, + "step": 14088 + }, + { + "epoch": 1.1617877463511173, + "grad_norm": 1.7706018686294556, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7455761432647705, + "num_tokens": 313841633.0, + "step": 14089 + }, + { + "epoch": 1.1618702069761688, + "grad_norm": 2.3142971992492676, + "learning_rate": 1e-06, + "loss": 0.7673, + "mean_token_accuracy": 0.7475999593734741, + "num_tokens": 313864449.0, + "step": 14090 + }, + { + "epoch": 1.1619526676012204, + "grad_norm": 1.998407006263733, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7647727727890015, + "num_tokens": 313892088.0, + "step": 14091 + }, + { + "epoch": 1.162035128226272, + "grad_norm": 2.060025930404663, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.74687260389328, + "num_tokens": 313920096.0, + "step": 14092 + }, + { + "epoch": 1.1621175888513235, + "grad_norm": 2.3113551139831543, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.72510826587677, + "num_tokens": 313944186.0, + "step": 14093 + }, + { + "epoch": 1.1622000494763751, + "grad_norm": 2.432298421859741, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7664215564727783, + "num_tokens": 313966055.0, + "step": 14094 + }, + { + "epoch": 1.1622825101014267, + "grad_norm": 2.157514810562134, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7860321998596191, + "num_tokens": 313994851.0, + "step": 14095 + }, + { + "epoch": 1.162364970726478, + "grad_norm": 1.8779546022415161, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.782325267791748, + "num_tokens": 314027005.0, + "step": 14096 + }, + { + "epoch": 1.1624474313515296, + "grad_norm": 2.087855339050293, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7298771142959595, + "num_tokens": 314055895.0, + "step": 14097 + }, + { + "epoch": 1.1625298919765812, + "grad_norm": 1.8584260940551758, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7456767559051514, + "num_tokens": 314090078.0, + "step": 14098 + }, + { + "epoch": 1.1626123526016328, + "grad_norm": 1.936672568321228, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.7875593900680542, + "num_tokens": 314122604.0, + "step": 14099 + }, + { + "epoch": 1.1626948132266843, + "grad_norm": 1.8148193359375, + "learning_rate": 1e-06, + "loss": 0.5943, + "mean_token_accuracy": 0.8059492707252502, + "num_tokens": 314158446.0, + "step": 14100 + }, + { + "epoch": 1.1627772738517357, + "grad_norm": 2.111860513687134, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7369239330291748, + "num_tokens": 314184558.0, + "step": 14101 + }, + { + "epoch": 1.1628597344767873, + "grad_norm": 2.11330509185791, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7424493432044983, + "num_tokens": 314211773.0, + "step": 14102 + }, + { + "epoch": 1.1629421951018388, + "grad_norm": 2.1301565170288086, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.7923609614372253, + "num_tokens": 314238197.0, + "step": 14103 + }, + { + "epoch": 1.1630246557268904, + "grad_norm": 1.9077413082122803, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7648289203643799, + "num_tokens": 314270098.0, + "step": 14104 + }, + { + "epoch": 1.163107116351942, + "grad_norm": 2.1826672554016113, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7414158582687378, + "num_tokens": 314296223.0, + "step": 14105 + }, + { + "epoch": 1.1631895769769935, + "grad_norm": 2.1032402515411377, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7791914939880371, + "num_tokens": 314323806.0, + "step": 14106 + }, + { + "epoch": 1.1632720376020451, + "grad_norm": 2.042581558227539, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.7915740013122559, + "num_tokens": 314350610.0, + "step": 14107 + }, + { + "epoch": 1.1633544982270965, + "grad_norm": 2.140779972076416, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7587188482284546, + "num_tokens": 314378813.0, + "step": 14108 + }, + { + "epoch": 1.163436958852148, + "grad_norm": 2.014969825744629, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7105280756950378, + "num_tokens": 314408055.0, + "step": 14109 + }, + { + "epoch": 1.1635194194771996, + "grad_norm": 2.111076831817627, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7532422542572021, + "num_tokens": 314438672.0, + "step": 14110 + }, + { + "epoch": 1.1636018801022512, + "grad_norm": 1.8339604139328003, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7390259504318237, + "num_tokens": 314474607.0, + "step": 14111 + }, + { + "epoch": 1.1636843407273028, + "grad_norm": 2.1065425872802734, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.757870078086853, + "num_tokens": 314499668.0, + "step": 14112 + }, + { + "epoch": 1.1637668013523543, + "grad_norm": 1.9271447658538818, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7690825462341309, + "num_tokens": 314532951.0, + "step": 14113 + }, + { + "epoch": 1.163849261977406, + "grad_norm": 1.9910329580307007, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7868601083755493, + "num_tokens": 314560755.0, + "step": 14114 + }, + { + "epoch": 1.1639317226024573, + "grad_norm": 2.1133880615234375, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7497353553771973, + "num_tokens": 314589342.0, + "step": 14115 + }, + { + "epoch": 1.1640141832275088, + "grad_norm": 2.0358927249908447, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7187787294387817, + "num_tokens": 314622548.0, + "step": 14116 + }, + { + "epoch": 1.1640966438525604, + "grad_norm": 2.08203125, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.738919734954834, + "num_tokens": 314651813.0, + "step": 14117 + }, + { + "epoch": 1.164179104477612, + "grad_norm": 2.0597903728485107, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7062762975692749, + "num_tokens": 314679043.0, + "step": 14118 + }, + { + "epoch": 1.1642615651026635, + "grad_norm": 2.193490982055664, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7826815843582153, + "num_tokens": 314702582.0, + "step": 14119 + }, + { + "epoch": 1.164344025727715, + "grad_norm": 1.8611794710159302, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7515015602111816, + "num_tokens": 314734907.0, + "step": 14120 + }, + { + "epoch": 1.1644264863527665, + "grad_norm": 2.562026023864746, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7607645988464355, + "num_tokens": 314753498.0, + "step": 14121 + }, + { + "epoch": 1.164508946977818, + "grad_norm": 2.056118965148926, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7198565006256104, + "num_tokens": 314784273.0, + "step": 14122 + }, + { + "epoch": 1.1645914076028696, + "grad_norm": 2.336111545562744, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7266631126403809, + "num_tokens": 314810674.0, + "step": 14123 + }, + { + "epoch": 1.1646738682279212, + "grad_norm": 2.1014552116394043, + "learning_rate": 1e-06, + "loss": 0.7722, + "mean_token_accuracy": 0.7502850890159607, + "num_tokens": 314839624.0, + "step": 14124 + }, + { + "epoch": 1.1647563288529728, + "grad_norm": 2.0730843544006348, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.755443811416626, + "num_tokens": 314867197.0, + "step": 14125 + }, + { + "epoch": 1.1648387894780243, + "grad_norm": 2.071525812149048, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7533085346221924, + "num_tokens": 314894087.0, + "step": 14126 + }, + { + "epoch": 1.1649212501030757, + "grad_norm": 1.8856624364852905, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.7917108535766602, + "num_tokens": 314928964.0, + "step": 14127 + }, + { + "epoch": 1.1650037107281273, + "grad_norm": 2.176861047744751, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.763505756855011, + "num_tokens": 314954673.0, + "step": 14128 + }, + { + "epoch": 1.1650861713531788, + "grad_norm": 2.289750337600708, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7353876233100891, + "num_tokens": 314978575.0, + "step": 14129 + }, + { + "epoch": 1.1651686319782304, + "grad_norm": 2.079411268234253, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7458529472351074, + "num_tokens": 315008506.0, + "step": 14130 + }, + { + "epoch": 1.165251092603282, + "grad_norm": 2.156259059906006, + "learning_rate": 1e-06, + "loss": 0.7762, + "mean_token_accuracy": 0.7551406621932983, + "num_tokens": 315034203.0, + "step": 14131 + }, + { + "epoch": 1.1653335532283335, + "grad_norm": 1.9822890758514404, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.784180760383606, + "num_tokens": 315062603.0, + "step": 14132 + }, + { + "epoch": 1.1654160138533851, + "grad_norm": 2.0709991455078125, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.797681450843811, + "num_tokens": 315090393.0, + "step": 14133 + }, + { + "epoch": 1.1654984744784365, + "grad_norm": 2.0670814514160156, + "learning_rate": 1e-06, + "loss": 0.548, + "mean_token_accuracy": 0.813726544380188, + "num_tokens": 315116462.0, + "step": 14134 + }, + { + "epoch": 1.165580935103488, + "grad_norm": 1.8880987167358398, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.789065957069397, + "num_tokens": 315147267.0, + "step": 14135 + }, + { + "epoch": 1.1656633957285396, + "grad_norm": 2.1850130558013916, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7391977906227112, + "num_tokens": 315176120.0, + "step": 14136 + }, + { + "epoch": 1.1657458563535912, + "grad_norm": 2.240116834640503, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.728035569190979, + "num_tokens": 315206204.0, + "step": 14137 + }, + { + "epoch": 1.1658283169786428, + "grad_norm": 1.7703841924667358, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.7580885887145996, + "num_tokens": 315245300.0, + "step": 14138 + }, + { + "epoch": 1.1659107776036943, + "grad_norm": 2.0469307899475098, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7626574039459229, + "num_tokens": 315274737.0, + "step": 14139 + }, + { + "epoch": 1.165993238228746, + "grad_norm": 2.075637102127075, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.74141526222229, + "num_tokens": 315305889.0, + "step": 14140 + }, + { + "epoch": 1.1660756988537972, + "grad_norm": 1.9884907007217407, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7801351547241211, + "num_tokens": 315337681.0, + "step": 14141 + }, + { + "epoch": 1.1661581594788488, + "grad_norm": 1.9564085006713867, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7812075614929199, + "num_tokens": 315366992.0, + "step": 14142 + }, + { + "epoch": 1.1662406201039004, + "grad_norm": 2.10263729095459, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7500554919242859, + "num_tokens": 315395211.0, + "step": 14143 + }, + { + "epoch": 1.166323080728952, + "grad_norm": 2.1971349716186523, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7730990648269653, + "num_tokens": 315421307.0, + "step": 14144 + }, + { + "epoch": 1.1664055413540035, + "grad_norm": 2.3250434398651123, + "learning_rate": 1e-06, + "loss": 0.5994, + "mean_token_accuracy": 0.7936071157455444, + "num_tokens": 315445418.0, + "step": 14145 + }, + { + "epoch": 1.166488001979055, + "grad_norm": 2.024284601211548, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7906357049942017, + "num_tokens": 315475600.0, + "step": 14146 + }, + { + "epoch": 1.1665704626041065, + "grad_norm": 2.0468456745147705, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7097710371017456, + "num_tokens": 315507147.0, + "step": 14147 + }, + { + "epoch": 1.166652923229158, + "grad_norm": 1.7535125017166138, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.765117883682251, + "num_tokens": 315541364.0, + "step": 14148 + }, + { + "epoch": 1.1667353838542096, + "grad_norm": 1.7435450553894043, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7732961177825928, + "num_tokens": 315575649.0, + "step": 14149 + }, + { + "epoch": 1.1668178444792612, + "grad_norm": 2.1110751628875732, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7381478548049927, + "num_tokens": 315600957.0, + "step": 14150 + }, + { + "epoch": 1.1669003051043128, + "grad_norm": 2.028973340988159, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.768245279788971, + "num_tokens": 315630063.0, + "step": 14151 + }, + { + "epoch": 1.1669827657293643, + "grad_norm": 1.740196943283081, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.7973084449768066, + "num_tokens": 315666605.0, + "step": 14152 + }, + { + "epoch": 1.1670652263544157, + "grad_norm": 1.865417718887329, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7392189502716064, + "num_tokens": 315700924.0, + "step": 14153 + }, + { + "epoch": 1.1671476869794672, + "grad_norm": 2.2522358894348145, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7260724306106567, + "num_tokens": 315732010.0, + "step": 14154 + }, + { + "epoch": 1.1672301476045188, + "grad_norm": 1.9481736421585083, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7241942882537842, + "num_tokens": 315763809.0, + "step": 14155 + }, + { + "epoch": 1.1673126082295704, + "grad_norm": 1.8642632961273193, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.7984558343887329, + "num_tokens": 315795025.0, + "step": 14156 + }, + { + "epoch": 1.167395068854622, + "grad_norm": 2.0319395065307617, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.725287139415741, + "num_tokens": 315826066.0, + "step": 14157 + }, + { + "epoch": 1.1674775294796735, + "grad_norm": 2.235539197921753, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7754091620445251, + "num_tokens": 315850462.0, + "step": 14158 + }, + { + "epoch": 1.167559990104725, + "grad_norm": 1.8530603647232056, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7695460915565491, + "num_tokens": 315882553.0, + "step": 14159 + }, + { + "epoch": 1.1676424507297765, + "grad_norm": 1.8273099660873413, + "learning_rate": 1e-06, + "loss": 0.4958, + "mean_token_accuracy": 0.829396665096283, + "num_tokens": 315913601.0, + "step": 14160 + }, + { + "epoch": 1.167724911354828, + "grad_norm": 1.9838498830795288, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7601289749145508, + "num_tokens": 315945677.0, + "step": 14161 + }, + { + "epoch": 1.1678073719798796, + "grad_norm": 1.969254970550537, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.747470498085022, + "num_tokens": 315975393.0, + "step": 14162 + }, + { + "epoch": 1.1678898326049312, + "grad_norm": 2.0446536540985107, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7740180492401123, + "num_tokens": 316005686.0, + "step": 14163 + }, + { + "epoch": 1.1679722932299827, + "grad_norm": 2.035820484161377, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7312436699867249, + "num_tokens": 316038302.0, + "step": 14164 + }, + { + "epoch": 1.168054753855034, + "grad_norm": 2.108109474182129, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7190789580345154, + "num_tokens": 316066094.0, + "step": 14165 + }, + { + "epoch": 1.1681372144800857, + "grad_norm": 2.021249294281006, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.7823678255081177, + "num_tokens": 316094363.0, + "step": 14166 + }, + { + "epoch": 1.1682196751051372, + "grad_norm": 1.9557102918624878, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7514363527297974, + "num_tokens": 316124189.0, + "step": 14167 + }, + { + "epoch": 1.1683021357301888, + "grad_norm": 2.2538163661956787, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7297838926315308, + "num_tokens": 316149183.0, + "step": 14168 + }, + { + "epoch": 1.1683845963552404, + "grad_norm": 1.8967961072921753, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7636772394180298, + "num_tokens": 316179308.0, + "step": 14169 + }, + { + "epoch": 1.168467056980292, + "grad_norm": 1.7786446809768677, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7431445121765137, + "num_tokens": 316217138.0, + "step": 14170 + }, + { + "epoch": 1.1685495176053435, + "grad_norm": 2.17583966255188, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7567476034164429, + "num_tokens": 316241738.0, + "step": 14171 + }, + { + "epoch": 1.1686319782303949, + "grad_norm": 2.180037498474121, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7464171648025513, + "num_tokens": 316268249.0, + "step": 14172 + }, + { + "epoch": 1.1687144388554465, + "grad_norm": 2.047306776046753, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7709972262382507, + "num_tokens": 316294197.0, + "step": 14173 + }, + { + "epoch": 1.168796899480498, + "grad_norm": 1.9194591045379639, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.7847615480422974, + "num_tokens": 316324803.0, + "step": 14174 + }, + { + "epoch": 1.1688793601055496, + "grad_norm": 2.107011556625366, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.766956627368927, + "num_tokens": 316352419.0, + "step": 14175 + }, + { + "epoch": 1.1689618207306012, + "grad_norm": 2.0930514335632324, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7582036256790161, + "num_tokens": 316378478.0, + "step": 14176 + }, + { + "epoch": 1.1690442813556527, + "grad_norm": 2.141756296157837, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7118854522705078, + "num_tokens": 316408283.0, + "step": 14177 + }, + { + "epoch": 1.1691267419807043, + "grad_norm": 1.8164509534835815, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.792980968952179, + "num_tokens": 316446903.0, + "step": 14178 + }, + { + "epoch": 1.1692092026057557, + "grad_norm": 2.0135104656219482, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7722243070602417, + "num_tokens": 316475670.0, + "step": 14179 + }, + { + "epoch": 1.1692916632308072, + "grad_norm": 2.033980131149292, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7245295643806458, + "num_tokens": 316505571.0, + "step": 14180 + }, + { + "epoch": 1.1693741238558588, + "grad_norm": 1.968977689743042, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7453864812850952, + "num_tokens": 316539276.0, + "step": 14181 + }, + { + "epoch": 1.1694565844809104, + "grad_norm": 2.1496925354003906, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7370367646217346, + "num_tokens": 316566839.0, + "step": 14182 + }, + { + "epoch": 1.169539045105962, + "grad_norm": 2.2567965984344482, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7257623672485352, + "num_tokens": 316593910.0, + "step": 14183 + }, + { + "epoch": 1.1696215057310135, + "grad_norm": 1.890726923942566, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7832940220832825, + "num_tokens": 316626681.0, + "step": 14184 + }, + { + "epoch": 1.1697039663560649, + "grad_norm": 2.1803181171417236, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7252241373062134, + "num_tokens": 316655198.0, + "step": 14185 + }, + { + "epoch": 1.1697864269811165, + "grad_norm": 1.9152214527130127, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.783454418182373, + "num_tokens": 316687936.0, + "step": 14186 + }, + { + "epoch": 1.169868887606168, + "grad_norm": 1.8691750764846802, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7212502956390381, + "num_tokens": 316721504.0, + "step": 14187 + }, + { + "epoch": 1.1699513482312196, + "grad_norm": 2.1388566493988037, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7791959643363953, + "num_tokens": 316750279.0, + "step": 14188 + }, + { + "epoch": 1.1700338088562712, + "grad_norm": 2.414853811264038, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7454907894134521, + "num_tokens": 316773862.0, + "step": 14189 + }, + { + "epoch": 1.1701162694813227, + "grad_norm": 2.04504132270813, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.748499870300293, + "num_tokens": 316804642.0, + "step": 14190 + }, + { + "epoch": 1.170198730106374, + "grad_norm": 1.9417392015457153, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.780492901802063, + "num_tokens": 316835397.0, + "step": 14191 + }, + { + "epoch": 1.1702811907314257, + "grad_norm": 1.9660663604736328, + "learning_rate": 1e-06, + "loss": 0.5955, + "mean_token_accuracy": 0.7964558601379395, + "num_tokens": 316866460.0, + "step": 14192 + }, + { + "epoch": 1.1703636513564772, + "grad_norm": 1.819115161895752, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7854553461074829, + "num_tokens": 316897811.0, + "step": 14193 + }, + { + "epoch": 1.1704461119815288, + "grad_norm": 1.7866621017456055, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7777974605560303, + "num_tokens": 316931429.0, + "step": 14194 + }, + { + "epoch": 1.1705285726065804, + "grad_norm": 2.037961483001709, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7440822124481201, + "num_tokens": 316962113.0, + "step": 14195 + }, + { + "epoch": 1.170611033231632, + "grad_norm": 2.05010986328125, + "learning_rate": 1e-06, + "loss": 0.5522, + "mean_token_accuracy": 0.8181189298629761, + "num_tokens": 316986777.0, + "step": 14196 + }, + { + "epoch": 1.1706934938566835, + "grad_norm": 2.2886459827423096, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7360619306564331, + "num_tokens": 317011887.0, + "step": 14197 + }, + { + "epoch": 1.1707759544817349, + "grad_norm": 1.9370815753936768, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7346555590629578, + "num_tokens": 317043680.0, + "step": 14198 + }, + { + "epoch": 1.1708584151067865, + "grad_norm": 2.1644790172576904, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.7837746739387512, + "num_tokens": 317072531.0, + "step": 14199 + }, + { + "epoch": 1.170940875731838, + "grad_norm": 2.2963831424713135, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7141987681388855, + "num_tokens": 317098445.0, + "step": 14200 + }, + { + "epoch": 1.1710233363568896, + "grad_norm": 2.0719447135925293, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7545269131660461, + "num_tokens": 317128477.0, + "step": 14201 + }, + { + "epoch": 1.1711057969819412, + "grad_norm": 1.9389995336532593, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7288992404937744, + "num_tokens": 317161499.0, + "step": 14202 + }, + { + "epoch": 1.1711882576069927, + "grad_norm": 1.8748267889022827, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7807105779647827, + "num_tokens": 317193696.0, + "step": 14203 + }, + { + "epoch": 1.1712707182320443, + "grad_norm": 2.038271427154541, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7446846961975098, + "num_tokens": 317225117.0, + "step": 14204 + }, + { + "epoch": 1.1713531788570957, + "grad_norm": 2.0264370441436768, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7668601274490356, + "num_tokens": 317255946.0, + "step": 14205 + }, + { + "epoch": 1.1714356394821472, + "grad_norm": 1.9650994539260864, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.7862553000450134, + "num_tokens": 317285740.0, + "step": 14206 + }, + { + "epoch": 1.1715181001071988, + "grad_norm": 1.9220439195632935, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.738768994808197, + "num_tokens": 317319325.0, + "step": 14207 + }, + { + "epoch": 1.1716005607322504, + "grad_norm": 2.4481046199798584, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7587932348251343, + "num_tokens": 317348978.0, + "step": 14208 + }, + { + "epoch": 1.171683021357302, + "grad_norm": 1.9493460655212402, + "learning_rate": 1e-06, + "loss": 0.5831, + "mean_token_accuracy": 0.8092105388641357, + "num_tokens": 317380721.0, + "step": 14209 + }, + { + "epoch": 1.1717654819823533, + "grad_norm": 2.137930393218994, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.7791191339492798, + "num_tokens": 317407265.0, + "step": 14210 + }, + { + "epoch": 1.1718479426074049, + "grad_norm": 1.7179877758026123, + "learning_rate": 1e-06, + "loss": 0.5909, + "mean_token_accuracy": 0.802353024482727, + "num_tokens": 317443954.0, + "step": 14211 + }, + { + "epoch": 1.1719304032324565, + "grad_norm": 1.7982414960861206, + "learning_rate": 1e-06, + "loss": 0.6189, + "mean_token_accuracy": 0.786708414554596, + "num_tokens": 317476275.0, + "step": 14212 + }, + { + "epoch": 1.172012863857508, + "grad_norm": 2.2865183353424072, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7412106394767761, + "num_tokens": 317501017.0, + "step": 14213 + }, + { + "epoch": 1.1720953244825596, + "grad_norm": 1.8468616008758545, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7456239461898804, + "num_tokens": 317535453.0, + "step": 14214 + }, + { + "epoch": 1.1721777851076112, + "grad_norm": 1.8691039085388184, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7238497734069824, + "num_tokens": 317568595.0, + "step": 14215 + }, + { + "epoch": 1.1722602457326627, + "grad_norm": 1.844287633895874, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.7954121828079224, + "num_tokens": 317600880.0, + "step": 14216 + }, + { + "epoch": 1.172342706357714, + "grad_norm": 2.130927324295044, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7342714667320251, + "num_tokens": 317628130.0, + "step": 14217 + }, + { + "epoch": 1.1724251669827657, + "grad_norm": 1.8510236740112305, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7905610799789429, + "num_tokens": 317660727.0, + "step": 14218 + }, + { + "epoch": 1.1725076276078172, + "grad_norm": 2.335484027862549, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.737532377243042, + "num_tokens": 317682101.0, + "step": 14219 + }, + { + "epoch": 1.1725900882328688, + "grad_norm": 2.0571448802948, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7856199145317078, + "num_tokens": 317711643.0, + "step": 14220 + }, + { + "epoch": 1.1726725488579204, + "grad_norm": 1.9746782779693604, + "learning_rate": 1e-06, + "loss": 0.7825, + "mean_token_accuracy": 0.7412019968032837, + "num_tokens": 317742443.0, + "step": 14221 + }, + { + "epoch": 1.172755009482972, + "grad_norm": 2.165168523788452, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7226952910423279, + "num_tokens": 317769620.0, + "step": 14222 + }, + { + "epoch": 1.1728374701080235, + "grad_norm": 2.2117013931274414, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7220622301101685, + "num_tokens": 317796222.0, + "step": 14223 + }, + { + "epoch": 1.1729199307330749, + "grad_norm": 1.81366765499115, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.7625807523727417, + "num_tokens": 317829117.0, + "step": 14224 + }, + { + "epoch": 1.1730023913581265, + "grad_norm": 1.725364327430725, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7586458921432495, + "num_tokens": 317870681.0, + "step": 14225 + }, + { + "epoch": 1.173084851983178, + "grad_norm": 2.1295177936553955, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7788593769073486, + "num_tokens": 317897508.0, + "step": 14226 + }, + { + "epoch": 1.1731673126082296, + "grad_norm": 1.978380799293518, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7702488303184509, + "num_tokens": 317927611.0, + "step": 14227 + }, + { + "epoch": 1.1732497732332812, + "grad_norm": 2.386699676513672, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7408115267753601, + "num_tokens": 317953631.0, + "step": 14228 + }, + { + "epoch": 1.1733322338583327, + "grad_norm": 1.9462628364562988, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7717968225479126, + "num_tokens": 317985277.0, + "step": 14229 + }, + { + "epoch": 1.173414694483384, + "grad_norm": 1.8391298055648804, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7735426425933838, + "num_tokens": 318020503.0, + "step": 14230 + }, + { + "epoch": 1.1734971551084357, + "grad_norm": 1.8673638105392456, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7706280946731567, + "num_tokens": 318053782.0, + "step": 14231 + }, + { + "epoch": 1.1735796157334872, + "grad_norm": 1.853729486465454, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7597354054450989, + "num_tokens": 318089432.0, + "step": 14232 + }, + { + "epoch": 1.1736620763585388, + "grad_norm": 1.9848098754882812, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7517184615135193, + "num_tokens": 318121733.0, + "step": 14233 + }, + { + "epoch": 1.1737445369835904, + "grad_norm": 2.0411336421966553, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7440481781959534, + "num_tokens": 318152985.0, + "step": 14234 + }, + { + "epoch": 1.173826997608642, + "grad_norm": 2.0051465034484863, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7720006108283997, + "num_tokens": 318180734.0, + "step": 14235 + }, + { + "epoch": 1.1739094582336933, + "grad_norm": 1.9154688119888306, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7470409274101257, + "num_tokens": 318215465.0, + "step": 14236 + }, + { + "epoch": 1.1739919188587449, + "grad_norm": 1.8357832431793213, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7683792114257812, + "num_tokens": 318250522.0, + "step": 14237 + }, + { + "epoch": 1.1740743794837964, + "grad_norm": 1.88509202003479, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7842801809310913, + "num_tokens": 318280859.0, + "step": 14238 + }, + { + "epoch": 1.174156840108848, + "grad_norm": 1.895616888999939, + "learning_rate": 1e-06, + "loss": 0.7999, + "mean_token_accuracy": 0.7434307336807251, + "num_tokens": 318313648.0, + "step": 14239 + }, + { + "epoch": 1.1742393007338996, + "grad_norm": 1.838492751121521, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.7911522388458252, + "num_tokens": 318345335.0, + "step": 14240 + }, + { + "epoch": 1.1743217613589512, + "grad_norm": 2.0470917224884033, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7562322616577148, + "num_tokens": 318373952.0, + "step": 14241 + }, + { + "epoch": 1.1744042219840027, + "grad_norm": 1.801697015762329, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7679657936096191, + "num_tokens": 318408356.0, + "step": 14242 + }, + { + "epoch": 1.174486682609054, + "grad_norm": 1.9701142311096191, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7824018597602844, + "num_tokens": 318436180.0, + "step": 14243 + }, + { + "epoch": 1.1745691432341057, + "grad_norm": 2.1328296661376953, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.7947153449058533, + "num_tokens": 318463773.0, + "step": 14244 + }, + { + "epoch": 1.1746516038591572, + "grad_norm": 1.9466595649719238, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7698847651481628, + "num_tokens": 318494731.0, + "step": 14245 + }, + { + "epoch": 1.1747340644842088, + "grad_norm": 2.0775649547576904, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7426191568374634, + "num_tokens": 318525476.0, + "step": 14246 + }, + { + "epoch": 1.1748165251092604, + "grad_norm": 2.17438006401062, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7767501473426819, + "num_tokens": 318549952.0, + "step": 14247 + }, + { + "epoch": 1.174898985734312, + "grad_norm": 2.051194429397583, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7418032288551331, + "num_tokens": 318577000.0, + "step": 14248 + }, + { + "epoch": 1.1749814463593635, + "grad_norm": 1.9613829851150513, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7460222244262695, + "num_tokens": 318608171.0, + "step": 14249 + }, + { + "epoch": 1.1750639069844149, + "grad_norm": 1.895259976387024, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7770572900772095, + "num_tokens": 318642645.0, + "step": 14250 + }, + { + "epoch": 1.1751463676094664, + "grad_norm": 2.0905470848083496, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7349681854248047, + "num_tokens": 318669571.0, + "step": 14251 + }, + { + "epoch": 1.175228828234518, + "grad_norm": 1.9723091125488281, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7550312280654907, + "num_tokens": 318702137.0, + "step": 14252 + }, + { + "epoch": 1.1753112888595696, + "grad_norm": 1.9635262489318848, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.7899187207221985, + "num_tokens": 318730343.0, + "step": 14253 + }, + { + "epoch": 1.1753937494846212, + "grad_norm": 1.9060449600219727, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7829519510269165, + "num_tokens": 318764385.0, + "step": 14254 + }, + { + "epoch": 1.1754762101096725, + "grad_norm": 1.8075194358825684, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7733931541442871, + "num_tokens": 318798070.0, + "step": 14255 + }, + { + "epoch": 1.175558670734724, + "grad_norm": 2.187898874282837, + "learning_rate": 1e-06, + "loss": 0.7887, + "mean_token_accuracy": 0.7556960582733154, + "num_tokens": 318821795.0, + "step": 14256 + }, + { + "epoch": 1.1756411313597757, + "grad_norm": 1.9351842403411865, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7507041096687317, + "num_tokens": 318854090.0, + "step": 14257 + }, + { + "epoch": 1.1757235919848272, + "grad_norm": 1.9063162803649902, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7732856869697571, + "num_tokens": 318885182.0, + "step": 14258 + }, + { + "epoch": 1.1758060526098788, + "grad_norm": 1.8911293745040894, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7553225755691528, + "num_tokens": 318916954.0, + "step": 14259 + }, + { + "epoch": 1.1758885132349304, + "grad_norm": 2.0393412113189697, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7154427766799927, + "num_tokens": 318946227.0, + "step": 14260 + }, + { + "epoch": 1.175970973859982, + "grad_norm": 1.8993672132492065, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7558442950248718, + "num_tokens": 318981232.0, + "step": 14261 + }, + { + "epoch": 1.1760534344850333, + "grad_norm": 1.761216640472412, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7251107692718506, + "num_tokens": 319020438.0, + "step": 14262 + }, + { + "epoch": 1.1761358951100849, + "grad_norm": 1.8335566520690918, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7473085522651672, + "num_tokens": 319054902.0, + "step": 14263 + }, + { + "epoch": 1.1762183557351364, + "grad_norm": 1.807706594467163, + "learning_rate": 1e-06, + "loss": 0.5788, + "mean_token_accuracy": 0.8095896244049072, + "num_tokens": 319091358.0, + "step": 14264 + }, + { + "epoch": 1.176300816360188, + "grad_norm": 2.1577343940734863, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.7916256785392761, + "num_tokens": 319115791.0, + "step": 14265 + }, + { + "epoch": 1.1763832769852396, + "grad_norm": 1.9982705116271973, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7369441986083984, + "num_tokens": 319146519.0, + "step": 14266 + }, + { + "epoch": 1.1764657376102912, + "grad_norm": 2.2078609466552734, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.757196307182312, + "num_tokens": 319171150.0, + "step": 14267 + }, + { + "epoch": 1.1765481982353427, + "grad_norm": 1.9689480066299438, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7725579142570496, + "num_tokens": 319202821.0, + "step": 14268 + }, + { + "epoch": 1.176630658860394, + "grad_norm": 2.2814831733703613, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7659316062927246, + "num_tokens": 319227605.0, + "step": 14269 + }, + { + "epoch": 1.1767131194854457, + "grad_norm": 2.0627875328063965, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7449880838394165, + "num_tokens": 319255720.0, + "step": 14270 + }, + { + "epoch": 1.1767955801104972, + "grad_norm": 2.1008822917938232, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7768760919570923, + "num_tokens": 319280863.0, + "step": 14271 + }, + { + "epoch": 1.1768780407355488, + "grad_norm": 2.1258437633514404, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.7785412073135376, + "num_tokens": 319306093.0, + "step": 14272 + }, + { + "epoch": 1.1769605013606004, + "grad_norm": 2.4522058963775635, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7061386108398438, + "num_tokens": 319328822.0, + "step": 14273 + }, + { + "epoch": 1.177042961985652, + "grad_norm": 1.6879063844680786, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7923715710639954, + "num_tokens": 319368572.0, + "step": 14274 + }, + { + "epoch": 1.1771254226107033, + "grad_norm": 1.7976504564285278, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.7921440601348877, + "num_tokens": 319399902.0, + "step": 14275 + }, + { + "epoch": 1.1772078832357549, + "grad_norm": 1.9725910425186157, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7381292581558228, + "num_tokens": 319429739.0, + "step": 14276 + }, + { + "epoch": 1.1772903438608064, + "grad_norm": 2.028369188308716, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7745429873466492, + "num_tokens": 319457658.0, + "step": 14277 + }, + { + "epoch": 1.177372804485858, + "grad_norm": 1.8246792554855347, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7820019721984863, + "num_tokens": 319490699.0, + "step": 14278 + }, + { + "epoch": 1.1774552651109096, + "grad_norm": 1.9815526008605957, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7760106325149536, + "num_tokens": 319519273.0, + "step": 14279 + }, + { + "epoch": 1.1775377257359612, + "grad_norm": 2.2388510704040527, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7571883201599121, + "num_tokens": 319546362.0, + "step": 14280 + }, + { + "epoch": 1.1776201863610125, + "grad_norm": 2.0832066535949707, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.7965635061264038, + "num_tokens": 319574070.0, + "step": 14281 + }, + { + "epoch": 1.177702646986064, + "grad_norm": 1.8658840656280518, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7702198624610901, + "num_tokens": 319605398.0, + "step": 14282 + }, + { + "epoch": 1.1777851076111157, + "grad_norm": 1.8374528884887695, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7754759788513184, + "num_tokens": 319639637.0, + "step": 14283 + }, + { + "epoch": 1.1778675682361672, + "grad_norm": 1.9246009588241577, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.753749668598175, + "num_tokens": 319674921.0, + "step": 14284 + }, + { + "epoch": 1.1779500288612188, + "grad_norm": 1.7953261137008667, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.7914696931838989, + "num_tokens": 319708148.0, + "step": 14285 + }, + { + "epoch": 1.1780324894862704, + "grad_norm": 2.1194839477539062, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7319637537002563, + "num_tokens": 319736928.0, + "step": 14286 + }, + { + "epoch": 1.178114950111322, + "grad_norm": 1.7133362293243408, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7739274501800537, + "num_tokens": 319771692.0, + "step": 14287 + }, + { + "epoch": 1.1781974107363733, + "grad_norm": 1.7275354862213135, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7792986631393433, + "num_tokens": 319808968.0, + "step": 14288 + }, + { + "epoch": 1.1782798713614249, + "grad_norm": 1.9215306043624878, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7745791673660278, + "num_tokens": 319843517.0, + "step": 14289 + }, + { + "epoch": 1.1783623319864764, + "grad_norm": 1.8287724256515503, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7919331789016724, + "num_tokens": 319876686.0, + "step": 14290 + }, + { + "epoch": 1.178444792611528, + "grad_norm": 2.1881468296051025, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7587050199508667, + "num_tokens": 319902228.0, + "step": 14291 + }, + { + "epoch": 1.1785272532365796, + "grad_norm": 2.0872714519500732, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7599474191665649, + "num_tokens": 319928671.0, + "step": 14292 + }, + { + "epoch": 1.1786097138616312, + "grad_norm": 1.9856231212615967, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7326675653457642, + "num_tokens": 319961689.0, + "step": 14293 + }, + { + "epoch": 1.1786921744866827, + "grad_norm": 2.1278533935546875, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7153099775314331, + "num_tokens": 319991359.0, + "step": 14294 + }, + { + "epoch": 1.178774635111734, + "grad_norm": 2.0201284885406494, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.6993938684463501, + "num_tokens": 320020886.0, + "step": 14295 + }, + { + "epoch": 1.1788570957367857, + "grad_norm": 2.0966460704803467, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7551008462905884, + "num_tokens": 320048023.0, + "step": 14296 + }, + { + "epoch": 1.1789395563618372, + "grad_norm": 1.8037400245666504, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7768231630325317, + "num_tokens": 320083098.0, + "step": 14297 + }, + { + "epoch": 1.1790220169868888, + "grad_norm": 1.8261065483093262, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7505307197570801, + "num_tokens": 320119802.0, + "step": 14298 + }, + { + "epoch": 1.1791044776119404, + "grad_norm": 2.1888184547424316, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7395007610321045, + "num_tokens": 320145535.0, + "step": 14299 + }, + { + "epoch": 1.1791869382369917, + "grad_norm": 1.836694598197937, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7567708492279053, + "num_tokens": 320179508.0, + "step": 14300 + }, + { + "epoch": 1.1792693988620433, + "grad_norm": 2.1643965244293213, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7606735229492188, + "num_tokens": 320204848.0, + "step": 14301 + }, + { + "epoch": 1.1793518594870949, + "grad_norm": 2.1885600090026855, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7628309726715088, + "num_tokens": 320231154.0, + "step": 14302 + }, + { + "epoch": 1.1794343201121464, + "grad_norm": 2.12758731842041, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.726280152797699, + "num_tokens": 320260930.0, + "step": 14303 + }, + { + "epoch": 1.179516780737198, + "grad_norm": 1.8237791061401367, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7538261413574219, + "num_tokens": 320296385.0, + "step": 14304 + }, + { + "epoch": 1.1795992413622496, + "grad_norm": 2.0532429218292236, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7735385894775391, + "num_tokens": 320323490.0, + "step": 14305 + }, + { + "epoch": 1.1796817019873012, + "grad_norm": 1.839693546295166, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7683447599411011, + "num_tokens": 320357817.0, + "step": 14306 + }, + { + "epoch": 1.1797641626123525, + "grad_norm": 1.9647094011306763, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7588019371032715, + "num_tokens": 320387426.0, + "step": 14307 + }, + { + "epoch": 1.179846623237404, + "grad_norm": 1.9314831495285034, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7640020847320557, + "num_tokens": 320417160.0, + "step": 14308 + }, + { + "epoch": 1.1799290838624557, + "grad_norm": 1.9556704759597778, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7713490724563599, + "num_tokens": 320446975.0, + "step": 14309 + }, + { + "epoch": 1.1800115444875072, + "grad_norm": 1.8580844402313232, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7532470226287842, + "num_tokens": 320481481.0, + "step": 14310 + }, + { + "epoch": 1.1800940051125588, + "grad_norm": 1.6909689903259277, + "learning_rate": 1e-06, + "loss": 0.5471, + "mean_token_accuracy": 0.8158107995986938, + "num_tokens": 320516873.0, + "step": 14311 + }, + { + "epoch": 1.1801764657376104, + "grad_norm": 1.8675377368927002, + "learning_rate": 1e-06, + "loss": 0.7724, + "mean_token_accuracy": 0.7585633397102356, + "num_tokens": 320549684.0, + "step": 14312 + }, + { + "epoch": 1.180258926362662, + "grad_norm": 1.9137983322143555, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7347612380981445, + "num_tokens": 320582016.0, + "step": 14313 + }, + { + "epoch": 1.1803413869877133, + "grad_norm": 1.7376149892807007, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.7546826601028442, + "num_tokens": 320620106.0, + "step": 14314 + }, + { + "epoch": 1.1804238476127649, + "grad_norm": 1.8942177295684814, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7479726076126099, + "num_tokens": 320654124.0, + "step": 14315 + }, + { + "epoch": 1.1805063082378164, + "grad_norm": 2.1818020343780518, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.7881358861923218, + "num_tokens": 320677156.0, + "step": 14316 + }, + { + "epoch": 1.180588768862868, + "grad_norm": 1.7550902366638184, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7622784972190857, + "num_tokens": 320715290.0, + "step": 14317 + }, + { + "epoch": 1.1806712294879196, + "grad_norm": 2.27518892288208, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7421766519546509, + "num_tokens": 320740887.0, + "step": 14318 + }, + { + "epoch": 1.180753690112971, + "grad_norm": 2.0172367095947266, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7750478982925415, + "num_tokens": 320769727.0, + "step": 14319 + }, + { + "epoch": 1.1808361507380225, + "grad_norm": 1.9750179052352905, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7668129205703735, + "num_tokens": 320800813.0, + "step": 14320 + }, + { + "epoch": 1.180918611363074, + "grad_norm": 2.247363328933716, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7452998161315918, + "num_tokens": 320826235.0, + "step": 14321 + }, + { + "epoch": 1.1810010719881257, + "grad_norm": 1.9831684827804565, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7374593019485474, + "num_tokens": 320857101.0, + "step": 14322 + }, + { + "epoch": 1.1810835326131772, + "grad_norm": 2.189757823944092, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7838060855865479, + "num_tokens": 320883218.0, + "step": 14323 + }, + { + "epoch": 1.1811659932382288, + "grad_norm": 2.184316873550415, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.7063800692558289, + "num_tokens": 320910313.0, + "step": 14324 + }, + { + "epoch": 1.1812484538632804, + "grad_norm": 1.9095231294631958, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7707353234291077, + "num_tokens": 320941938.0, + "step": 14325 + }, + { + "epoch": 1.1813309144883317, + "grad_norm": 2.282741069793701, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7252483367919922, + "num_tokens": 320967661.0, + "step": 14326 + }, + { + "epoch": 1.1814133751133833, + "grad_norm": 1.7360796928405762, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7805054783821106, + "num_tokens": 321006019.0, + "step": 14327 + }, + { + "epoch": 1.1814958357384349, + "grad_norm": 2.2427186965942383, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7430841326713562, + "num_tokens": 321031042.0, + "step": 14328 + }, + { + "epoch": 1.1815782963634864, + "grad_norm": 1.8702619075775146, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7567934989929199, + "num_tokens": 321061375.0, + "step": 14329 + }, + { + "epoch": 1.181660756988538, + "grad_norm": 2.0440421104431152, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7729802131652832, + "num_tokens": 321089365.0, + "step": 14330 + }, + { + "epoch": 1.1817432176135896, + "grad_norm": 1.9464129209518433, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7163631319999695, + "num_tokens": 321127383.0, + "step": 14331 + }, + { + "epoch": 1.1818256782386412, + "grad_norm": 1.9291306734085083, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7211108803749084, + "num_tokens": 321160966.0, + "step": 14332 + }, + { + "epoch": 1.1819081388636925, + "grad_norm": 1.984756588935852, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7640782594680786, + "num_tokens": 321191012.0, + "step": 14333 + }, + { + "epoch": 1.181990599488744, + "grad_norm": 1.9498059749603271, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.728216290473938, + "num_tokens": 321220919.0, + "step": 14334 + }, + { + "epoch": 1.1820730601137956, + "grad_norm": 2.01347279548645, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.7807929515838623, + "num_tokens": 321249780.0, + "step": 14335 + }, + { + "epoch": 1.1821555207388472, + "grad_norm": 2.0468697547912598, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7302765846252441, + "num_tokens": 321280290.0, + "step": 14336 + }, + { + "epoch": 1.1822379813638988, + "grad_norm": 2.0688488483428955, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.751991868019104, + "num_tokens": 321308041.0, + "step": 14337 + }, + { + "epoch": 1.1823204419889504, + "grad_norm": 2.1304006576538086, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7419123649597168, + "num_tokens": 321334841.0, + "step": 14338 + }, + { + "epoch": 1.182402902614002, + "grad_norm": 1.9337373971939087, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7481156587600708, + "num_tokens": 321366972.0, + "step": 14339 + }, + { + "epoch": 1.1824853632390533, + "grad_norm": 1.9522866010665894, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.7932308912277222, + "num_tokens": 321396294.0, + "step": 14340 + }, + { + "epoch": 1.1825678238641049, + "grad_norm": 2.320106267929077, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7621877193450928, + "num_tokens": 321418601.0, + "step": 14341 + }, + { + "epoch": 1.1826502844891564, + "grad_norm": 2.0131771564483643, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7720184326171875, + "num_tokens": 321447503.0, + "step": 14342 + }, + { + "epoch": 1.182732745114208, + "grad_norm": 2.1070611476898193, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7354816198348999, + "num_tokens": 321477953.0, + "step": 14343 + }, + { + "epoch": 1.1828152057392596, + "grad_norm": 2.1042354106903076, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7744775414466858, + "num_tokens": 321506063.0, + "step": 14344 + }, + { + "epoch": 1.182897666364311, + "grad_norm": 2.244011640548706, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.7800992727279663, + "num_tokens": 321530657.0, + "step": 14345 + }, + { + "epoch": 1.1829801269893625, + "grad_norm": 1.937090277671814, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7605736255645752, + "num_tokens": 321560110.0, + "step": 14346 + }, + { + "epoch": 1.183062587614414, + "grad_norm": 1.7529399394989014, + "learning_rate": 1e-06, + "loss": 0.559, + "mean_token_accuracy": 0.8141301870346069, + "num_tokens": 321592761.0, + "step": 14347 + }, + { + "epoch": 1.1831450482394656, + "grad_norm": 1.9905065298080444, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7587952613830566, + "num_tokens": 321624820.0, + "step": 14348 + }, + { + "epoch": 1.1832275088645172, + "grad_norm": 1.7929702997207642, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.755760669708252, + "num_tokens": 321657645.0, + "step": 14349 + }, + { + "epoch": 1.1833099694895688, + "grad_norm": 1.9518873691558838, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7812515497207642, + "num_tokens": 321687163.0, + "step": 14350 + }, + { + "epoch": 1.1833924301146204, + "grad_norm": 2.0660407543182373, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7536658048629761, + "num_tokens": 321714632.0, + "step": 14351 + }, + { + "epoch": 1.1834748907396717, + "grad_norm": 2.0145580768585205, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7481791973114014, + "num_tokens": 321745867.0, + "step": 14352 + }, + { + "epoch": 1.1835573513647233, + "grad_norm": 2.0384578704833984, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.780121922492981, + "num_tokens": 321776574.0, + "step": 14353 + }, + { + "epoch": 1.1836398119897749, + "grad_norm": 2.078648805618286, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7243610620498657, + "num_tokens": 321805884.0, + "step": 14354 + }, + { + "epoch": 1.1837222726148264, + "grad_norm": 2.0466556549072266, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7317914366722107, + "num_tokens": 321836599.0, + "step": 14355 + }, + { + "epoch": 1.183804733239878, + "grad_norm": 2.1498358249664307, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7311409115791321, + "num_tokens": 321865397.0, + "step": 14356 + }, + { + "epoch": 1.1838871938649296, + "grad_norm": 2.0595781803131104, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7429938316345215, + "num_tokens": 321895421.0, + "step": 14357 + }, + { + "epoch": 1.1839696544899811, + "grad_norm": 2.1393985748291016, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7439092397689819, + "num_tokens": 321924271.0, + "step": 14358 + }, + { + "epoch": 1.1840521151150325, + "grad_norm": 1.9494760036468506, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7479244470596313, + "num_tokens": 321956670.0, + "step": 14359 + }, + { + "epoch": 1.184134575740084, + "grad_norm": 2.0302577018737793, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7780066728591919, + "num_tokens": 321982828.0, + "step": 14360 + }, + { + "epoch": 1.1842170363651356, + "grad_norm": 2.0042057037353516, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7624116539955139, + "num_tokens": 322012619.0, + "step": 14361 + }, + { + "epoch": 1.1842994969901872, + "grad_norm": 1.7731599807739258, + "learning_rate": 1e-06, + "loss": 0.5959, + "mean_token_accuracy": 0.8162039518356323, + "num_tokens": 322046995.0, + "step": 14362 + }, + { + "epoch": 1.1843819576152388, + "grad_norm": 2.1283881664276123, + "learning_rate": 1e-06, + "loss": 0.6063, + "mean_token_accuracy": 0.7936584949493408, + "num_tokens": 322074342.0, + "step": 14363 + }, + { + "epoch": 1.1844644182402901, + "grad_norm": 2.2614686489105225, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7670103311538696, + "num_tokens": 322101075.0, + "step": 14364 + }, + { + "epoch": 1.1845468788653417, + "grad_norm": 1.9917254447937012, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7526206970214844, + "num_tokens": 322132995.0, + "step": 14365 + }, + { + "epoch": 1.1846293394903933, + "grad_norm": 2.4711618423461914, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7716494798660278, + "num_tokens": 322156665.0, + "step": 14366 + }, + { + "epoch": 1.1847118001154449, + "grad_norm": 1.9588415622711182, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7620171308517456, + "num_tokens": 322190322.0, + "step": 14367 + }, + { + "epoch": 1.1847942607404964, + "grad_norm": 1.741117000579834, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7778260111808777, + "num_tokens": 322229559.0, + "step": 14368 + }, + { + "epoch": 1.184876721365548, + "grad_norm": 1.9709322452545166, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.754836916923523, + "num_tokens": 322259831.0, + "step": 14369 + }, + { + "epoch": 1.1849591819905996, + "grad_norm": 1.8216149806976318, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7908073663711548, + "num_tokens": 322294497.0, + "step": 14370 + }, + { + "epoch": 1.185041642615651, + "grad_norm": 1.9828078746795654, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7755154967308044, + "num_tokens": 322324337.0, + "step": 14371 + }, + { + "epoch": 1.1851241032407025, + "grad_norm": 2.1778650283813477, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7058531045913696, + "num_tokens": 322353833.0, + "step": 14372 + }, + { + "epoch": 1.185206563865754, + "grad_norm": 2.2679030895233154, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7089747190475464, + "num_tokens": 322379680.0, + "step": 14373 + }, + { + "epoch": 1.1852890244908056, + "grad_norm": 1.781856656074524, + "learning_rate": 1e-06, + "loss": 0.5867, + "mean_token_accuracy": 0.8054224848747253, + "num_tokens": 322414025.0, + "step": 14374 + }, + { + "epoch": 1.1853714851158572, + "grad_norm": 2.091780185699463, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7682510018348694, + "num_tokens": 322440009.0, + "step": 14375 + }, + { + "epoch": 1.1854539457409088, + "grad_norm": 1.91759192943573, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.7860400676727295, + "num_tokens": 322468340.0, + "step": 14376 + }, + { + "epoch": 1.1855364063659604, + "grad_norm": 2.0132036209106445, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.744331955909729, + "num_tokens": 322498434.0, + "step": 14377 + }, + { + "epoch": 1.1856188669910117, + "grad_norm": 1.9726471900939941, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7245364785194397, + "num_tokens": 322528960.0, + "step": 14378 + }, + { + "epoch": 1.1857013276160633, + "grad_norm": 2.20900559425354, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7608637809753418, + "num_tokens": 322554873.0, + "step": 14379 + }, + { + "epoch": 1.1857837882411149, + "grad_norm": 1.909457802772522, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7724416255950928, + "num_tokens": 322585515.0, + "step": 14380 + }, + { + "epoch": 1.1858662488661664, + "grad_norm": 2.2142791748046875, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.716405987739563, + "num_tokens": 322616105.0, + "step": 14381 + }, + { + "epoch": 1.185948709491218, + "grad_norm": 2.1237142086029053, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7363094091415405, + "num_tokens": 322645622.0, + "step": 14382 + }, + { + "epoch": 1.1860311701162696, + "grad_norm": 2.0231375694274902, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7394958734512329, + "num_tokens": 322677216.0, + "step": 14383 + }, + { + "epoch": 1.186113630741321, + "grad_norm": 2.01218581199646, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7572156190872192, + "num_tokens": 322706251.0, + "step": 14384 + }, + { + "epoch": 1.1861960913663725, + "grad_norm": 1.8997339010238647, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7650666236877441, + "num_tokens": 322739023.0, + "step": 14385 + }, + { + "epoch": 1.186278551991424, + "grad_norm": 2.39609694480896, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.7750638127326965, + "num_tokens": 322760837.0, + "step": 14386 + }, + { + "epoch": 1.1863610126164756, + "grad_norm": 2.0785326957702637, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7625112533569336, + "num_tokens": 322789658.0, + "step": 14387 + }, + { + "epoch": 1.1864434732415272, + "grad_norm": 1.9694609642028809, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7712428569793701, + "num_tokens": 322821324.0, + "step": 14388 + }, + { + "epoch": 1.1865259338665788, + "grad_norm": 2.1705310344696045, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7778832316398621, + "num_tokens": 322846467.0, + "step": 14389 + }, + { + "epoch": 1.1866083944916301, + "grad_norm": 2.1535744667053223, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.7721104621887207, + "num_tokens": 322871512.0, + "step": 14390 + }, + { + "epoch": 1.1866908551166817, + "grad_norm": 1.9065115451812744, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7682079672813416, + "num_tokens": 322905480.0, + "step": 14391 + }, + { + "epoch": 1.1867733157417333, + "grad_norm": 2.0214104652404785, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7619785070419312, + "num_tokens": 322932979.0, + "step": 14392 + }, + { + "epoch": 1.1868557763667849, + "grad_norm": 1.7496458292007446, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7671458721160889, + "num_tokens": 322968586.0, + "step": 14393 + }, + { + "epoch": 1.1869382369918364, + "grad_norm": 2.3441591262817383, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7708643078804016, + "num_tokens": 322996568.0, + "step": 14394 + }, + { + "epoch": 1.187020697616888, + "grad_norm": 1.9804590940475464, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.7885608077049255, + "num_tokens": 323026443.0, + "step": 14395 + }, + { + "epoch": 1.1871031582419396, + "grad_norm": 2.075984239578247, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.7806453704833984, + "num_tokens": 323054559.0, + "step": 14396 + }, + { + "epoch": 1.187185618866991, + "grad_norm": 2.4235270023345947, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7286526560783386, + "num_tokens": 323076385.0, + "step": 14397 + }, + { + "epoch": 1.1872680794920425, + "grad_norm": 2.0037009716033936, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.751706600189209, + "num_tokens": 323104469.0, + "step": 14398 + }, + { + "epoch": 1.187350540117094, + "grad_norm": 1.9441474676132202, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7376836538314819, + "num_tokens": 323136858.0, + "step": 14399 + }, + { + "epoch": 1.1874330007421456, + "grad_norm": 2.1204307079315186, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7502907514572144, + "num_tokens": 323166128.0, + "step": 14400 + }, + { + "epoch": 1.1875154613671972, + "grad_norm": 1.9792200326919556, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7594153881072998, + "num_tokens": 323197909.0, + "step": 14401 + }, + { + "epoch": 1.1875979219922488, + "grad_norm": 1.9955393075942993, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7354329824447632, + "num_tokens": 323229573.0, + "step": 14402 + }, + { + "epoch": 1.1876803826173004, + "grad_norm": 2.1585092544555664, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7632066011428833, + "num_tokens": 323256953.0, + "step": 14403 + }, + { + "epoch": 1.1877628432423517, + "grad_norm": 1.9555015563964844, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.743308961391449, + "num_tokens": 323287601.0, + "step": 14404 + }, + { + "epoch": 1.1878453038674033, + "grad_norm": 2.0363383293151855, + "learning_rate": 1e-06, + "loss": 0.5456, + "mean_token_accuracy": 0.8189108371734619, + "num_tokens": 323314369.0, + "step": 14405 + }, + { + "epoch": 1.1879277644924549, + "grad_norm": 1.9967621564865112, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7814182043075562, + "num_tokens": 323343252.0, + "step": 14406 + }, + { + "epoch": 1.1880102251175064, + "grad_norm": 2.0262911319732666, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7665784358978271, + "num_tokens": 323374701.0, + "step": 14407 + }, + { + "epoch": 1.188092685742558, + "grad_norm": 2.083935260772705, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7223614454269409, + "num_tokens": 323406777.0, + "step": 14408 + }, + { + "epoch": 1.1881751463676093, + "grad_norm": 1.9481779336929321, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.8023331165313721, + "num_tokens": 323436480.0, + "step": 14409 + }, + { + "epoch": 1.188257606992661, + "grad_norm": 2.1607701778411865, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7690311074256897, + "num_tokens": 323465488.0, + "step": 14410 + }, + { + "epoch": 1.1883400676177125, + "grad_norm": 1.9097774028778076, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7522284388542175, + "num_tokens": 323499578.0, + "step": 14411 + }, + { + "epoch": 1.188422528242764, + "grad_norm": 2.0861124992370605, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7012470960617065, + "num_tokens": 323530532.0, + "step": 14412 + }, + { + "epoch": 1.1885049888678156, + "grad_norm": 1.9299477338790894, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7591254115104675, + "num_tokens": 323561575.0, + "step": 14413 + }, + { + "epoch": 1.1885874494928672, + "grad_norm": 2.0202105045318604, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7226814031600952, + "num_tokens": 323591894.0, + "step": 14414 + }, + { + "epoch": 1.1886699101179188, + "grad_norm": 1.8573205471038818, + "learning_rate": 1e-06, + "loss": 0.5687, + "mean_token_accuracy": 0.8064165115356445, + "num_tokens": 323624591.0, + "step": 14415 + }, + { + "epoch": 1.1887523707429701, + "grad_norm": 1.79551362991333, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.788872480392456, + "num_tokens": 323658805.0, + "step": 14416 + }, + { + "epoch": 1.1888348313680217, + "grad_norm": 1.8887879848480225, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.724905788898468, + "num_tokens": 323694124.0, + "step": 14417 + }, + { + "epoch": 1.1889172919930733, + "grad_norm": 2.2228641510009766, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7404361963272095, + "num_tokens": 323721286.0, + "step": 14418 + }, + { + "epoch": 1.1889997526181248, + "grad_norm": 1.6068739891052246, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.8005458116531372, + "num_tokens": 323762718.0, + "step": 14419 + }, + { + "epoch": 1.1890822132431764, + "grad_norm": 2.1035876274108887, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.766182005405426, + "num_tokens": 323790668.0, + "step": 14420 + }, + { + "epoch": 1.189164673868228, + "grad_norm": 1.952871322631836, + "learning_rate": 1e-06, + "loss": 0.6145, + "mean_token_accuracy": 0.7848988771438599, + "num_tokens": 323824142.0, + "step": 14421 + }, + { + "epoch": 1.1892471344932796, + "grad_norm": 2.0196001529693604, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7349449992179871, + "num_tokens": 323855449.0, + "step": 14422 + }, + { + "epoch": 1.189329595118331, + "grad_norm": 2.2441930770874023, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7331073880195618, + "num_tokens": 323881163.0, + "step": 14423 + }, + { + "epoch": 1.1894120557433825, + "grad_norm": 2.2297122478485107, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7522670030593872, + "num_tokens": 323907923.0, + "step": 14424 + }, + { + "epoch": 1.189494516368434, + "grad_norm": 1.855015516281128, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7335185408592224, + "num_tokens": 323941112.0, + "step": 14425 + }, + { + "epoch": 1.1895769769934856, + "grad_norm": 1.887855052947998, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7724921703338623, + "num_tokens": 323973257.0, + "step": 14426 + }, + { + "epoch": 1.1896594376185372, + "grad_norm": 1.920021653175354, + "learning_rate": 1e-06, + "loss": 0.5774, + "mean_token_accuracy": 0.8118617534637451, + "num_tokens": 324005196.0, + "step": 14427 + }, + { + "epoch": 1.1897418982435888, + "grad_norm": 1.8794680833816528, + "learning_rate": 1e-06, + "loss": 0.5422, + "mean_token_accuracy": 0.8211919069290161, + "num_tokens": 324035669.0, + "step": 14428 + }, + { + "epoch": 1.1898243588686401, + "grad_norm": 1.9926190376281738, + "learning_rate": 1e-06, + "loss": 0.7483, + "mean_token_accuracy": 0.7533500790596008, + "num_tokens": 324065990.0, + "step": 14429 + }, + { + "epoch": 1.1899068194936917, + "grad_norm": 2.111504554748535, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7571945786476135, + "num_tokens": 324094324.0, + "step": 14430 + }, + { + "epoch": 1.1899892801187433, + "grad_norm": 1.9273322820663452, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7389784455299377, + "num_tokens": 324127112.0, + "step": 14431 + }, + { + "epoch": 1.1900717407437948, + "grad_norm": 1.8635627031326294, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7672548294067383, + "num_tokens": 324156756.0, + "step": 14432 + }, + { + "epoch": 1.1901542013688464, + "grad_norm": 1.8537442684173584, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7768479585647583, + "num_tokens": 324188744.0, + "step": 14433 + }, + { + "epoch": 1.190236661993898, + "grad_norm": 1.9156429767608643, + "learning_rate": 1e-06, + "loss": 0.5676, + "mean_token_accuracy": 0.8134364485740662, + "num_tokens": 324218687.0, + "step": 14434 + }, + { + "epoch": 1.1903191226189493, + "grad_norm": 2.107140064239502, + "learning_rate": 1e-06, + "loss": 0.6192, + "mean_token_accuracy": 0.7845784425735474, + "num_tokens": 324243177.0, + "step": 14435 + }, + { + "epoch": 1.190401583244001, + "grad_norm": 1.8452484607696533, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7700554132461548, + "num_tokens": 324277197.0, + "step": 14436 + }, + { + "epoch": 1.1904840438690525, + "grad_norm": 1.9854792356491089, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7545354962348938, + "num_tokens": 324308491.0, + "step": 14437 + }, + { + "epoch": 1.190566504494104, + "grad_norm": 1.9543648958206177, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.7982267141342163, + "num_tokens": 324336656.0, + "step": 14438 + }, + { + "epoch": 1.1906489651191556, + "grad_norm": 2.462019443511963, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7433457970619202, + "num_tokens": 324360895.0, + "step": 14439 + }, + { + "epoch": 1.1907314257442072, + "grad_norm": 1.944505214691162, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7691738605499268, + "num_tokens": 324391313.0, + "step": 14440 + }, + { + "epoch": 1.1908138863692588, + "grad_norm": 1.8526660203933716, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7648675441741943, + "num_tokens": 324423164.0, + "step": 14441 + }, + { + "epoch": 1.1908963469943101, + "grad_norm": 2.278517723083496, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7472834587097168, + "num_tokens": 324448336.0, + "step": 14442 + }, + { + "epoch": 1.1909788076193617, + "grad_norm": 1.904713749885559, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7446366548538208, + "num_tokens": 324481034.0, + "step": 14443 + }, + { + "epoch": 1.1910612682444133, + "grad_norm": 1.8067339658737183, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7434122562408447, + "num_tokens": 324517867.0, + "step": 14444 + }, + { + "epoch": 1.1911437288694648, + "grad_norm": 1.9362168312072754, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7445015907287598, + "num_tokens": 324550656.0, + "step": 14445 + }, + { + "epoch": 1.1912261894945164, + "grad_norm": 1.9743093252182007, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7815464735031128, + "num_tokens": 324582907.0, + "step": 14446 + }, + { + "epoch": 1.191308650119568, + "grad_norm": 2.1003472805023193, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7539529800415039, + "num_tokens": 324613059.0, + "step": 14447 + }, + { + "epoch": 1.1913911107446196, + "grad_norm": 2.376291275024414, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7648310661315918, + "num_tokens": 324634059.0, + "step": 14448 + }, + { + "epoch": 1.191473571369671, + "grad_norm": 2.0139386653900146, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.769791841506958, + "num_tokens": 324660919.0, + "step": 14449 + }, + { + "epoch": 1.1915560319947225, + "grad_norm": 2.032841920852661, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7497881650924683, + "num_tokens": 324690641.0, + "step": 14450 + }, + { + "epoch": 1.191638492619774, + "grad_norm": 1.865994930267334, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7712656855583191, + "num_tokens": 324727508.0, + "step": 14451 + }, + { + "epoch": 1.1917209532448256, + "grad_norm": 2.0799803733825684, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.786362886428833, + "num_tokens": 324756081.0, + "step": 14452 + }, + { + "epoch": 1.1918034138698772, + "grad_norm": 1.9824846982955933, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7450776100158691, + "num_tokens": 324788881.0, + "step": 14453 + }, + { + "epoch": 1.1918858744949286, + "grad_norm": 2.1537296772003174, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.7847289443016052, + "num_tokens": 324815935.0, + "step": 14454 + }, + { + "epoch": 1.1919683351199801, + "grad_norm": 2.073730945587158, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7441433668136597, + "num_tokens": 324848048.0, + "step": 14455 + }, + { + "epoch": 1.1920507957450317, + "grad_norm": 1.8659098148345947, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7691276669502258, + "num_tokens": 324882337.0, + "step": 14456 + }, + { + "epoch": 1.1921332563700833, + "grad_norm": 1.8590131998062134, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.7897192239761353, + "num_tokens": 324913108.0, + "step": 14457 + }, + { + "epoch": 1.1922157169951348, + "grad_norm": 2.1588475704193115, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7243362665176392, + "num_tokens": 324942072.0, + "step": 14458 + }, + { + "epoch": 1.1922981776201864, + "grad_norm": 2.0052006244659424, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7737443447113037, + "num_tokens": 324970734.0, + "step": 14459 + }, + { + "epoch": 1.192380638245238, + "grad_norm": 1.9694126844406128, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7734984755516052, + "num_tokens": 325001405.0, + "step": 14460 + }, + { + "epoch": 1.1924630988702893, + "grad_norm": 1.8482264280319214, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7883888483047485, + "num_tokens": 325034029.0, + "step": 14461 + }, + { + "epoch": 1.192545559495341, + "grad_norm": 1.8375575542449951, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7684860229492188, + "num_tokens": 325067752.0, + "step": 14462 + }, + { + "epoch": 1.1926280201203925, + "grad_norm": 1.9687353372573853, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7514625191688538, + "num_tokens": 325100420.0, + "step": 14463 + }, + { + "epoch": 1.192710480745444, + "grad_norm": 2.1521406173706055, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.795074999332428, + "num_tokens": 325126319.0, + "step": 14464 + }, + { + "epoch": 1.1927929413704956, + "grad_norm": 2.2209529876708984, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7411545515060425, + "num_tokens": 325152121.0, + "step": 14465 + }, + { + "epoch": 1.1928754019955472, + "grad_norm": 2.368030309677124, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7420216202735901, + "num_tokens": 325177399.0, + "step": 14466 + }, + { + "epoch": 1.1929578626205988, + "grad_norm": 2.0383834838867188, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7579053640365601, + "num_tokens": 325205765.0, + "step": 14467 + }, + { + "epoch": 1.1930403232456501, + "grad_norm": 1.9464420080184937, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7653732299804688, + "num_tokens": 325237487.0, + "step": 14468 + }, + { + "epoch": 1.1931227838707017, + "grad_norm": 1.9604676961898804, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7493869662284851, + "num_tokens": 325267869.0, + "step": 14469 + }, + { + "epoch": 1.1932052444957533, + "grad_norm": 2.0624704360961914, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7594053149223328, + "num_tokens": 325297762.0, + "step": 14470 + }, + { + "epoch": 1.1932877051208048, + "grad_norm": 1.9122258424758911, + "learning_rate": 1e-06, + "loss": 0.6011, + "mean_token_accuracy": 0.7958219051361084, + "num_tokens": 325325294.0, + "step": 14471 + }, + { + "epoch": 1.1933701657458564, + "grad_norm": 2.291585683822632, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7059757113456726, + "num_tokens": 325350080.0, + "step": 14472 + }, + { + "epoch": 1.193452626370908, + "grad_norm": 1.7541923522949219, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7801153659820557, + "num_tokens": 325385034.0, + "step": 14473 + }, + { + "epoch": 1.1935350869959593, + "grad_norm": 2.0182571411132812, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.761077880859375, + "num_tokens": 325415614.0, + "step": 14474 + }, + { + "epoch": 1.193617547621011, + "grad_norm": 1.9784711599349976, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7344897985458374, + "num_tokens": 325447342.0, + "step": 14475 + }, + { + "epoch": 1.1937000082460625, + "grad_norm": 2.463045120239258, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7754471898078918, + "num_tokens": 325468121.0, + "step": 14476 + }, + { + "epoch": 1.193782468871114, + "grad_norm": 1.95610773563385, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7669833898544312, + "num_tokens": 325501198.0, + "step": 14477 + }, + { + "epoch": 1.1938649294961656, + "grad_norm": 1.9634259939193726, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7799481749534607, + "num_tokens": 325532263.0, + "step": 14478 + }, + { + "epoch": 1.1939473901212172, + "grad_norm": 2.2176618576049805, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7684956192970276, + "num_tokens": 325559175.0, + "step": 14479 + }, + { + "epoch": 1.1940298507462686, + "grad_norm": 2.133223533630371, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7357643842697144, + "num_tokens": 325586660.0, + "step": 14480 + }, + { + "epoch": 1.1941123113713201, + "grad_norm": 1.9710099697113037, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7756813764572144, + "num_tokens": 325617684.0, + "step": 14481 + }, + { + "epoch": 1.1941947719963717, + "grad_norm": 1.9602324962615967, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7575295567512512, + "num_tokens": 325646989.0, + "step": 14482 + }, + { + "epoch": 1.1942772326214233, + "grad_norm": 1.9488369226455688, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.7842736840248108, + "num_tokens": 325678799.0, + "step": 14483 + }, + { + "epoch": 1.1943596932464748, + "grad_norm": 1.9198009967803955, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7693213820457458, + "num_tokens": 325711226.0, + "step": 14484 + }, + { + "epoch": 1.1944421538715264, + "grad_norm": 1.8815138339996338, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7695077657699585, + "num_tokens": 325742807.0, + "step": 14485 + }, + { + "epoch": 1.194524614496578, + "grad_norm": 1.911181926727295, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7736820578575134, + "num_tokens": 325776839.0, + "step": 14486 + }, + { + "epoch": 1.1946070751216293, + "grad_norm": 2.0533223152160645, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7389895915985107, + "num_tokens": 325805578.0, + "step": 14487 + }, + { + "epoch": 1.194689535746681, + "grad_norm": 2.0652382373809814, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7152757048606873, + "num_tokens": 325834521.0, + "step": 14488 + }, + { + "epoch": 1.1947719963717325, + "grad_norm": 1.7577813863754272, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7437547445297241, + "num_tokens": 325869637.0, + "step": 14489 + }, + { + "epoch": 1.194854456996784, + "grad_norm": 1.8284533023834229, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7827340364456177, + "num_tokens": 325903189.0, + "step": 14490 + }, + { + "epoch": 1.1949369176218356, + "grad_norm": 2.059614419937134, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7410627007484436, + "num_tokens": 325933939.0, + "step": 14491 + }, + { + "epoch": 1.1950193782468872, + "grad_norm": 1.889552354812622, + "learning_rate": 1e-06, + "loss": 0.5605, + "mean_token_accuracy": 0.8122745752334595, + "num_tokens": 325962992.0, + "step": 14492 + }, + { + "epoch": 1.1951018388719388, + "grad_norm": 2.0548934936523438, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7729465961456299, + "num_tokens": 325989727.0, + "step": 14493 + }, + { + "epoch": 1.1951842994969901, + "grad_norm": 1.8038510084152222, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7226573824882507, + "num_tokens": 326028410.0, + "step": 14494 + }, + { + "epoch": 1.1952667601220417, + "grad_norm": 2.0026440620422363, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7631158828735352, + "num_tokens": 326058941.0, + "step": 14495 + }, + { + "epoch": 1.1953492207470933, + "grad_norm": 2.1545238494873047, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.7833943367004395, + "num_tokens": 326084107.0, + "step": 14496 + }, + { + "epoch": 1.1954316813721448, + "grad_norm": 1.9630167484283447, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7560921907424927, + "num_tokens": 326114706.0, + "step": 14497 + }, + { + "epoch": 1.1955141419971964, + "grad_norm": 2.6834027767181396, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.772260308265686, + "num_tokens": 326147101.0, + "step": 14498 + }, + { + "epoch": 1.1955966026222478, + "grad_norm": 2.0764944553375244, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7692481279373169, + "num_tokens": 326173587.0, + "step": 14499 + }, + { + "epoch": 1.1956790632472993, + "grad_norm": 1.8512500524520874, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7546524405479431, + "num_tokens": 326207058.0, + "step": 14500 + }, + { + "epoch": 1.195761523872351, + "grad_norm": 2.203404664993286, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7553789615631104, + "num_tokens": 326232311.0, + "step": 14501 + }, + { + "epoch": 1.1958439844974025, + "grad_norm": 2.1279995441436768, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7289924621582031, + "num_tokens": 326259652.0, + "step": 14502 + }, + { + "epoch": 1.195926445122454, + "grad_norm": 1.8963730335235596, + "learning_rate": 1e-06, + "loss": 0.566, + "mean_token_accuracy": 0.8158287405967712, + "num_tokens": 326292074.0, + "step": 14503 + }, + { + "epoch": 1.1960089057475056, + "grad_norm": 2.0022265911102295, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7481406331062317, + "num_tokens": 326321616.0, + "step": 14504 + }, + { + "epoch": 1.1960913663725572, + "grad_norm": 1.87083899974823, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7425992488861084, + "num_tokens": 326354295.0, + "step": 14505 + }, + { + "epoch": 1.1961738269976085, + "grad_norm": 1.9380921125411987, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7692013382911682, + "num_tokens": 326383040.0, + "step": 14506 + }, + { + "epoch": 1.1962562876226601, + "grad_norm": 2.146301507949829, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7615184783935547, + "num_tokens": 326408508.0, + "step": 14507 + }, + { + "epoch": 1.1963387482477117, + "grad_norm": 1.9935688972473145, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7706689238548279, + "num_tokens": 326437676.0, + "step": 14508 + }, + { + "epoch": 1.1964212088727633, + "grad_norm": 1.9496842622756958, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.7887532711029053, + "num_tokens": 326469052.0, + "step": 14509 + }, + { + "epoch": 1.1965036694978148, + "grad_norm": 1.9413433074951172, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7616636753082275, + "num_tokens": 326502601.0, + "step": 14510 + }, + { + "epoch": 1.1965861301228664, + "grad_norm": 1.9023487567901611, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7635400295257568, + "num_tokens": 326537743.0, + "step": 14511 + }, + { + "epoch": 1.196668590747918, + "grad_norm": 1.8520305156707764, + "learning_rate": 1e-06, + "loss": 0.5156, + "mean_token_accuracy": 0.8198823928833008, + "num_tokens": 326570967.0, + "step": 14512 + }, + { + "epoch": 1.1967510513729693, + "grad_norm": 2.011284828186035, + "learning_rate": 1e-06, + "loss": 0.6038, + "mean_token_accuracy": 0.794386625289917, + "num_tokens": 326597466.0, + "step": 14513 + }, + { + "epoch": 1.196833511998021, + "grad_norm": 2.053251028060913, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7430853843688965, + "num_tokens": 326634556.0, + "step": 14514 + }, + { + "epoch": 1.1969159726230725, + "grad_norm": 1.9595022201538086, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7382321953773499, + "num_tokens": 326669632.0, + "step": 14515 + }, + { + "epoch": 1.196998433248124, + "grad_norm": 2.0574843883514404, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7484874725341797, + "num_tokens": 326697173.0, + "step": 14516 + }, + { + "epoch": 1.1970808938731756, + "grad_norm": 2.1928467750549316, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7494161128997803, + "num_tokens": 326723983.0, + "step": 14517 + }, + { + "epoch": 1.197163354498227, + "grad_norm": 2.056919813156128, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7574834823608398, + "num_tokens": 326753954.0, + "step": 14518 + }, + { + "epoch": 1.1972458151232785, + "grad_norm": 1.997118592262268, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7702783942222595, + "num_tokens": 326785921.0, + "step": 14519 + }, + { + "epoch": 1.1973282757483301, + "grad_norm": 1.921313762664795, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7576635479927063, + "num_tokens": 326818585.0, + "step": 14520 + }, + { + "epoch": 1.1974107363733817, + "grad_norm": 2.1612601280212402, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7511790990829468, + "num_tokens": 326845578.0, + "step": 14521 + }, + { + "epoch": 1.1974931969984333, + "grad_norm": 2.0072579383850098, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7432824969291687, + "num_tokens": 326874017.0, + "step": 14522 + }, + { + "epoch": 1.1975756576234848, + "grad_norm": 2.184870719909668, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7417793273925781, + "num_tokens": 326903478.0, + "step": 14523 + }, + { + "epoch": 1.1976581182485364, + "grad_norm": 1.877466082572937, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.76347815990448, + "num_tokens": 326936797.0, + "step": 14524 + }, + { + "epoch": 1.1977405788735878, + "grad_norm": 1.8107149600982666, + "learning_rate": 1e-06, + "loss": 0.536, + "mean_token_accuracy": 0.825153112411499, + "num_tokens": 326970319.0, + "step": 14525 + }, + { + "epoch": 1.1978230394986393, + "grad_norm": 1.9969674348831177, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7373821139335632, + "num_tokens": 327000692.0, + "step": 14526 + }, + { + "epoch": 1.197905500123691, + "grad_norm": 2.094919443130493, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7500560283660889, + "num_tokens": 327029616.0, + "step": 14527 + }, + { + "epoch": 1.1979879607487425, + "grad_norm": 2.067197322845459, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7591472268104553, + "num_tokens": 327062059.0, + "step": 14528 + }, + { + "epoch": 1.198070421373794, + "grad_norm": 1.929652452468872, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7269948720932007, + "num_tokens": 327098683.0, + "step": 14529 + }, + { + "epoch": 1.1981528819988456, + "grad_norm": 2.160217523574829, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7749924659729004, + "num_tokens": 327125344.0, + "step": 14530 + }, + { + "epoch": 1.1982353426238972, + "grad_norm": 2.1221249103546143, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7539932727813721, + "num_tokens": 327153581.0, + "step": 14531 + }, + { + "epoch": 1.1983178032489485, + "grad_norm": 1.8903841972351074, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7331507205963135, + "num_tokens": 327188678.0, + "step": 14532 + }, + { + "epoch": 1.1984002638740001, + "grad_norm": 1.8206349611282349, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7795294523239136, + "num_tokens": 327219301.0, + "step": 14533 + }, + { + "epoch": 1.1984827244990517, + "grad_norm": 2.064955472946167, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7195419073104858, + "num_tokens": 327248908.0, + "step": 14534 + }, + { + "epoch": 1.1985651851241033, + "grad_norm": 2.15812349319458, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7643007040023804, + "num_tokens": 327276611.0, + "step": 14535 + }, + { + "epoch": 1.1986476457491548, + "grad_norm": 1.9197152853012085, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7929950952529907, + "num_tokens": 327307238.0, + "step": 14536 + }, + { + "epoch": 1.1987301063742064, + "grad_norm": 2.1173558235168457, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.7885647416114807, + "num_tokens": 327333899.0, + "step": 14537 + }, + { + "epoch": 1.198812566999258, + "grad_norm": 2.032142162322998, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.769652247428894, + "num_tokens": 327362691.0, + "step": 14538 + }, + { + "epoch": 1.1988950276243093, + "grad_norm": 2.058006525039673, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7776216268539429, + "num_tokens": 327393039.0, + "step": 14539 + }, + { + "epoch": 1.198977488249361, + "grad_norm": 1.8078677654266357, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7489457726478577, + "num_tokens": 327427644.0, + "step": 14540 + }, + { + "epoch": 1.1990599488744125, + "grad_norm": 2.3284661769866943, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7536518573760986, + "num_tokens": 327451127.0, + "step": 14541 + }, + { + "epoch": 1.199142409499464, + "grad_norm": 2.0151350498199463, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7227559089660645, + "num_tokens": 327482276.0, + "step": 14542 + }, + { + "epoch": 1.1992248701245156, + "grad_norm": 2.0351717472076416, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7097187042236328, + "num_tokens": 327514688.0, + "step": 14543 + }, + { + "epoch": 1.199307330749567, + "grad_norm": 2.0571906566619873, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7191684246063232, + "num_tokens": 327547969.0, + "step": 14544 + }, + { + "epoch": 1.1993897913746185, + "grad_norm": 2.2438032627105713, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7394914031028748, + "num_tokens": 327577792.0, + "step": 14545 + }, + { + "epoch": 1.1994722519996701, + "grad_norm": 2.0988075733184814, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.773045003414154, + "num_tokens": 327604810.0, + "step": 14546 + }, + { + "epoch": 1.1995547126247217, + "grad_norm": 1.8146066665649414, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7841508388519287, + "num_tokens": 327637041.0, + "step": 14547 + }, + { + "epoch": 1.1996371732497733, + "grad_norm": 2.0863873958587646, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7505398392677307, + "num_tokens": 327662479.0, + "step": 14548 + }, + { + "epoch": 1.1997196338748248, + "grad_norm": 2.1311187744140625, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7470773458480835, + "num_tokens": 327689826.0, + "step": 14549 + }, + { + "epoch": 1.1998020944998764, + "grad_norm": 2.112971305847168, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7317662239074707, + "num_tokens": 327721424.0, + "step": 14550 + }, + { + "epoch": 1.1998845551249278, + "grad_norm": 1.988264799118042, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7392481565475464, + "num_tokens": 327752695.0, + "step": 14551 + }, + { + "epoch": 1.1999670157499793, + "grad_norm": 1.9881725311279297, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7743887901306152, + "num_tokens": 327783363.0, + "step": 14552 + }, + { + "epoch": 1.200049476375031, + "grad_norm": 2.052605390548706, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7472110390663147, + "num_tokens": 327812687.0, + "step": 14553 + }, + { + "epoch": 1.2001319370000825, + "grad_norm": 1.977358102798462, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7847862839698792, + "num_tokens": 327842371.0, + "step": 14554 + }, + { + "epoch": 1.200214397625134, + "grad_norm": 2.080108404159546, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7677513957023621, + "num_tokens": 327870047.0, + "step": 14555 + }, + { + "epoch": 1.2002968582501856, + "grad_norm": 2.052853584289551, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7540683746337891, + "num_tokens": 327900354.0, + "step": 14556 + }, + { + "epoch": 1.2003793188752372, + "grad_norm": 2.223783254623413, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7330886721611023, + "num_tokens": 327927825.0, + "step": 14557 + }, + { + "epoch": 1.2004617795002885, + "grad_norm": 1.8817166090011597, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7744647264480591, + "num_tokens": 327957548.0, + "step": 14558 + }, + { + "epoch": 1.2005442401253401, + "grad_norm": 2.0266895294189453, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7760316133499146, + "num_tokens": 327987320.0, + "step": 14559 + }, + { + "epoch": 1.2006267007503917, + "grad_norm": 1.8499560356140137, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7383801341056824, + "num_tokens": 328022461.0, + "step": 14560 + }, + { + "epoch": 1.2007091613754433, + "grad_norm": 1.8488982915878296, + "learning_rate": 1e-06, + "loss": 0.5444, + "mean_token_accuracy": 0.8206878900527954, + "num_tokens": 328053833.0, + "step": 14561 + }, + { + "epoch": 1.2007916220004948, + "grad_norm": 1.9799726009368896, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7638580799102783, + "num_tokens": 328084275.0, + "step": 14562 + }, + { + "epoch": 1.2008740826255462, + "grad_norm": 2.1139352321624756, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7662581205368042, + "num_tokens": 328111279.0, + "step": 14563 + }, + { + "epoch": 1.2009565432505978, + "grad_norm": 1.6948541402816772, + "learning_rate": 1e-06, + "loss": 0.5535, + "mean_token_accuracy": 0.8175249099731445, + "num_tokens": 328149281.0, + "step": 14564 + }, + { + "epoch": 1.2010390038756493, + "grad_norm": 1.8582353591918945, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7713301181793213, + "num_tokens": 328182591.0, + "step": 14565 + }, + { + "epoch": 1.201121464500701, + "grad_norm": 2.0928232669830322, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7566900253295898, + "num_tokens": 328212314.0, + "step": 14566 + }, + { + "epoch": 1.2012039251257525, + "grad_norm": 1.9153456687927246, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7505804896354675, + "num_tokens": 328245312.0, + "step": 14567 + }, + { + "epoch": 1.201286385750804, + "grad_norm": 1.7333784103393555, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7801382541656494, + "num_tokens": 328284298.0, + "step": 14568 + }, + { + "epoch": 1.2013688463758556, + "grad_norm": 2.247849702835083, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.7907077074050903, + "num_tokens": 328307541.0, + "step": 14569 + }, + { + "epoch": 1.201451307000907, + "grad_norm": 1.9650899171829224, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7404977679252625, + "num_tokens": 328335281.0, + "step": 14570 + }, + { + "epoch": 1.2015337676259585, + "grad_norm": 1.9894688129425049, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7774443626403809, + "num_tokens": 328366644.0, + "step": 14571 + }, + { + "epoch": 1.20161622825101, + "grad_norm": 1.6926435232162476, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7857978343963623, + "num_tokens": 328403089.0, + "step": 14572 + }, + { + "epoch": 1.2016986888760617, + "grad_norm": 1.8790730237960815, + "learning_rate": 1e-06, + "loss": 0.7762, + "mean_token_accuracy": 0.7565587759017944, + "num_tokens": 328436101.0, + "step": 14573 + }, + { + "epoch": 1.2017811495011133, + "grad_norm": 2.0211892127990723, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7679909467697144, + "num_tokens": 328465546.0, + "step": 14574 + }, + { + "epoch": 1.2018636101261648, + "grad_norm": 2.178905487060547, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7629048824310303, + "num_tokens": 328490674.0, + "step": 14575 + }, + { + "epoch": 1.2019460707512164, + "grad_norm": 1.932855248451233, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7716268301010132, + "num_tokens": 328523253.0, + "step": 14576 + }, + { + "epoch": 1.2020285313762678, + "grad_norm": 1.9289571046829224, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.769241988658905, + "num_tokens": 328552881.0, + "step": 14577 + }, + { + "epoch": 1.2021109920013193, + "grad_norm": 2.426466941833496, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.7931307554244995, + "num_tokens": 328576272.0, + "step": 14578 + }, + { + "epoch": 1.202193452626371, + "grad_norm": 1.8132668733596802, + "learning_rate": 1e-06, + "loss": 0.7712, + "mean_token_accuracy": 0.7497520446777344, + "num_tokens": 328616752.0, + "step": 14579 + }, + { + "epoch": 1.2022759132514225, + "grad_norm": 2.095583438873291, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7602872848510742, + "num_tokens": 328645303.0, + "step": 14580 + }, + { + "epoch": 1.202358373876474, + "grad_norm": 1.8755207061767578, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7737100124359131, + "num_tokens": 328679704.0, + "step": 14581 + }, + { + "epoch": 1.2024408345015256, + "grad_norm": 1.8663831949234009, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7333479523658752, + "num_tokens": 328711816.0, + "step": 14582 + }, + { + "epoch": 1.202523295126577, + "grad_norm": 2.2246339321136475, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7493718862533569, + "num_tokens": 328737593.0, + "step": 14583 + }, + { + "epoch": 1.2026057557516285, + "grad_norm": 2.112698793411255, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7462158203125, + "num_tokens": 328764871.0, + "step": 14584 + }, + { + "epoch": 1.20268821637668, + "grad_norm": 1.9381678104400635, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7627276182174683, + "num_tokens": 328795929.0, + "step": 14585 + }, + { + "epoch": 1.2027706770017317, + "grad_norm": 2.2968690395355225, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.768934428691864, + "num_tokens": 328817871.0, + "step": 14586 + }, + { + "epoch": 1.2028531376267833, + "grad_norm": 1.8122457265853882, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7816369533538818, + "num_tokens": 328853128.0, + "step": 14587 + }, + { + "epoch": 1.2029355982518348, + "grad_norm": 1.8284372091293335, + "learning_rate": 1e-06, + "loss": 0.6026, + "mean_token_accuracy": 0.8036719560623169, + "num_tokens": 328884702.0, + "step": 14588 + }, + { + "epoch": 1.2030180588768862, + "grad_norm": 2.1533663272857666, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7668981552124023, + "num_tokens": 328908716.0, + "step": 14589 + }, + { + "epoch": 1.2031005195019377, + "grad_norm": 1.8456532955169678, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7900368571281433, + "num_tokens": 328939149.0, + "step": 14590 + }, + { + "epoch": 1.2031829801269893, + "grad_norm": 1.9552704095840454, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7452699542045593, + "num_tokens": 328971378.0, + "step": 14591 + }, + { + "epoch": 1.203265440752041, + "grad_norm": 2.03969669342041, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7364069819450378, + "num_tokens": 329003806.0, + "step": 14592 + }, + { + "epoch": 1.2033479013770925, + "grad_norm": 1.8432470560073853, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7736250162124634, + "num_tokens": 329039571.0, + "step": 14593 + }, + { + "epoch": 1.203430362002144, + "grad_norm": 1.866560697555542, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7821429967880249, + "num_tokens": 329071060.0, + "step": 14594 + }, + { + "epoch": 1.2035128226271956, + "grad_norm": 2.040123462677002, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7238863706588745, + "num_tokens": 329099438.0, + "step": 14595 + }, + { + "epoch": 1.203595283252247, + "grad_norm": 2.0384316444396973, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7554671764373779, + "num_tokens": 329128788.0, + "step": 14596 + }, + { + "epoch": 1.2036777438772985, + "grad_norm": 2.1256325244903564, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7498469352722168, + "num_tokens": 329156845.0, + "step": 14597 + }, + { + "epoch": 1.20376020450235, + "grad_norm": 1.9110058546066284, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7839353084564209, + "num_tokens": 329188948.0, + "step": 14598 + }, + { + "epoch": 1.2038426651274017, + "grad_norm": 2.0308260917663574, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7862573266029358, + "num_tokens": 329217997.0, + "step": 14599 + }, + { + "epoch": 1.2039251257524533, + "grad_norm": 1.979399561882019, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7287578582763672, + "num_tokens": 329247235.0, + "step": 14600 + }, + { + "epoch": 1.2040075863775048, + "grad_norm": 2.144625663757324, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7577730417251587, + "num_tokens": 329272659.0, + "step": 14601 + }, + { + "epoch": 1.2040900470025564, + "grad_norm": 1.802768349647522, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7638671398162842, + "num_tokens": 329306059.0, + "step": 14602 + }, + { + "epoch": 1.2041725076276077, + "grad_norm": 1.7167116403579712, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7729471325874329, + "num_tokens": 329342169.0, + "step": 14603 + }, + { + "epoch": 1.2042549682526593, + "grad_norm": 1.8169441223144531, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7698284387588501, + "num_tokens": 329375169.0, + "step": 14604 + }, + { + "epoch": 1.204337428877711, + "grad_norm": 1.6998276710510254, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7546090483665466, + "num_tokens": 329414422.0, + "step": 14605 + }, + { + "epoch": 1.2044198895027625, + "grad_norm": 2.1328413486480713, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.77766352891922, + "num_tokens": 329443036.0, + "step": 14606 + }, + { + "epoch": 1.204502350127814, + "grad_norm": 1.844562292098999, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7836325168609619, + "num_tokens": 329478245.0, + "step": 14607 + }, + { + "epoch": 1.2045848107528654, + "grad_norm": 1.89639413356781, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.7994709014892578, + "num_tokens": 329509859.0, + "step": 14608 + }, + { + "epoch": 1.204667271377917, + "grad_norm": 1.9832335710525513, + "learning_rate": 1e-06, + "loss": 0.7683, + "mean_token_accuracy": 0.7608652114868164, + "num_tokens": 329540442.0, + "step": 14609 + }, + { + "epoch": 1.2047497320029685, + "grad_norm": 1.9734652042388916, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7555102109909058, + "num_tokens": 329574949.0, + "step": 14610 + }, + { + "epoch": 1.20483219262802, + "grad_norm": 2.337209701538086, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7743602991104126, + "num_tokens": 329600881.0, + "step": 14611 + }, + { + "epoch": 1.2049146532530717, + "grad_norm": 2.093339204788208, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7087420225143433, + "num_tokens": 329629835.0, + "step": 14612 + }, + { + "epoch": 1.2049971138781232, + "grad_norm": 1.9072409868240356, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7048883438110352, + "num_tokens": 329663199.0, + "step": 14613 + }, + { + "epoch": 1.2050795745031748, + "grad_norm": 1.9842760562896729, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7548840641975403, + "num_tokens": 329696424.0, + "step": 14614 + }, + { + "epoch": 1.2051620351282262, + "grad_norm": 2.1217668056488037, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7477865219116211, + "num_tokens": 329723171.0, + "step": 14615 + }, + { + "epoch": 1.2052444957532777, + "grad_norm": 2.096038579940796, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7438178062438965, + "num_tokens": 329751278.0, + "step": 14616 + }, + { + "epoch": 1.2053269563783293, + "grad_norm": 2.058880090713501, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7816793918609619, + "num_tokens": 329776574.0, + "step": 14617 + }, + { + "epoch": 1.205409417003381, + "grad_norm": 1.9779434204101562, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7833971977233887, + "num_tokens": 329806428.0, + "step": 14618 + }, + { + "epoch": 1.2054918776284325, + "grad_norm": 1.7540206909179688, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7862992286682129, + "num_tokens": 329837963.0, + "step": 14619 + }, + { + "epoch": 1.205574338253484, + "grad_norm": 1.90004301071167, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7427934408187866, + "num_tokens": 329870750.0, + "step": 14620 + }, + { + "epoch": 1.2056567988785356, + "grad_norm": 1.915030837059021, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7656521797180176, + "num_tokens": 329902999.0, + "step": 14621 + }, + { + "epoch": 1.205739259503587, + "grad_norm": 2.1390328407287598, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7450976967811584, + "num_tokens": 329931251.0, + "step": 14622 + }, + { + "epoch": 1.2058217201286385, + "grad_norm": 1.9393329620361328, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7276663780212402, + "num_tokens": 329968933.0, + "step": 14623 + }, + { + "epoch": 1.20590418075369, + "grad_norm": 2.3392438888549805, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7643956542015076, + "num_tokens": 329993009.0, + "step": 14624 + }, + { + "epoch": 1.2059866413787417, + "grad_norm": 1.866538643836975, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7530747652053833, + "num_tokens": 330027621.0, + "step": 14625 + }, + { + "epoch": 1.2060691020037932, + "grad_norm": 1.9280085563659668, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7897109985351562, + "num_tokens": 330058583.0, + "step": 14626 + }, + { + "epoch": 1.2061515626288448, + "grad_norm": 1.8815224170684814, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7771760821342468, + "num_tokens": 330092094.0, + "step": 14627 + }, + { + "epoch": 1.2062340232538962, + "grad_norm": 1.880635142326355, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7738297581672668, + "num_tokens": 330124190.0, + "step": 14628 + }, + { + "epoch": 1.2063164838789477, + "grad_norm": 2.1186611652374268, + "learning_rate": 1e-06, + "loss": 0.6029, + "mean_token_accuracy": 0.7995427846908569, + "num_tokens": 330150630.0, + "step": 14629 + }, + { + "epoch": 1.2063989445039993, + "grad_norm": 1.8573118448257446, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.766089677810669, + "num_tokens": 330184339.0, + "step": 14630 + }, + { + "epoch": 1.2064814051290509, + "grad_norm": 2.107809543609619, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7700415253639221, + "num_tokens": 330211218.0, + "step": 14631 + }, + { + "epoch": 1.2065638657541025, + "grad_norm": 1.9915428161621094, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7528355717658997, + "num_tokens": 330242575.0, + "step": 14632 + }, + { + "epoch": 1.206646326379154, + "grad_norm": 1.7667121887207031, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7874125242233276, + "num_tokens": 330277982.0, + "step": 14633 + }, + { + "epoch": 1.2067287870042054, + "grad_norm": 2.0871810913085938, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.720538854598999, + "num_tokens": 330306598.0, + "step": 14634 + }, + { + "epoch": 1.206811247629257, + "grad_norm": 1.7381534576416016, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.765220046043396, + "num_tokens": 330341843.0, + "step": 14635 + }, + { + "epoch": 1.2068937082543085, + "grad_norm": 2.0044174194335938, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7808776497840881, + "num_tokens": 330369214.0, + "step": 14636 + }, + { + "epoch": 1.20697616887936, + "grad_norm": 2.06618332862854, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.713739275932312, + "num_tokens": 330398083.0, + "step": 14637 + }, + { + "epoch": 1.2070586295044117, + "grad_norm": 2.2087485790252686, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.6967813968658447, + "num_tokens": 330425226.0, + "step": 14638 + }, + { + "epoch": 1.2071410901294632, + "grad_norm": 1.9463720321655273, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7595489621162415, + "num_tokens": 330456505.0, + "step": 14639 + }, + { + "epoch": 1.2072235507545148, + "grad_norm": 1.8060320615768433, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7692295908927917, + "num_tokens": 330491823.0, + "step": 14640 + }, + { + "epoch": 1.2073060113795662, + "grad_norm": 2.0643112659454346, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7104417085647583, + "num_tokens": 330522974.0, + "step": 14641 + }, + { + "epoch": 1.2073884720046177, + "grad_norm": 2.126354455947876, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7783178091049194, + "num_tokens": 330549989.0, + "step": 14642 + }, + { + "epoch": 1.2074709326296693, + "grad_norm": 1.8022629022598267, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7284911870956421, + "num_tokens": 330586683.0, + "step": 14643 + }, + { + "epoch": 1.2075533932547209, + "grad_norm": 2.0468339920043945, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7690954804420471, + "num_tokens": 330620511.0, + "step": 14644 + }, + { + "epoch": 1.2076358538797725, + "grad_norm": 1.9171247482299805, + "learning_rate": 1e-06, + "loss": 0.7702, + "mean_token_accuracy": 0.7521708607673645, + "num_tokens": 330655652.0, + "step": 14645 + }, + { + "epoch": 1.207718314504824, + "grad_norm": 1.8834991455078125, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.747185230255127, + "num_tokens": 330689874.0, + "step": 14646 + }, + { + "epoch": 1.2078007751298756, + "grad_norm": 1.9850823879241943, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.7950899600982666, + "num_tokens": 330721792.0, + "step": 14647 + }, + { + "epoch": 1.207883235754927, + "grad_norm": 2.2917063236236572, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.7640684843063354, + "num_tokens": 330746281.0, + "step": 14648 + }, + { + "epoch": 1.2079656963799785, + "grad_norm": 2.1631360054016113, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7821239233016968, + "num_tokens": 330772860.0, + "step": 14649 + }, + { + "epoch": 1.20804815700503, + "grad_norm": 1.800820231437683, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7291468381881714, + "num_tokens": 330809790.0, + "step": 14650 + }, + { + "epoch": 1.2081306176300817, + "grad_norm": 1.9318020343780518, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7442724704742432, + "num_tokens": 330840139.0, + "step": 14651 + }, + { + "epoch": 1.2082130782551332, + "grad_norm": 1.8350719213485718, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.7889679670333862, + "num_tokens": 330871749.0, + "step": 14652 + }, + { + "epoch": 1.2082955388801846, + "grad_norm": 2.0507354736328125, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7547252178192139, + "num_tokens": 330898816.0, + "step": 14653 + }, + { + "epoch": 1.2083779995052362, + "grad_norm": 1.9801386594772339, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7482333183288574, + "num_tokens": 330931452.0, + "step": 14654 + }, + { + "epoch": 1.2084604601302877, + "grad_norm": 2.1622536182403564, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.772041916847229, + "num_tokens": 330955369.0, + "step": 14655 + }, + { + "epoch": 1.2085429207553393, + "grad_norm": 2.0095930099487305, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7209089994430542, + "num_tokens": 330987713.0, + "step": 14656 + }, + { + "epoch": 1.2086253813803909, + "grad_norm": 2.2279160022735596, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7378577589988708, + "num_tokens": 331013753.0, + "step": 14657 + }, + { + "epoch": 1.2087078420054425, + "grad_norm": 2.2380118370056152, + "learning_rate": 1e-06, + "loss": 0.6103, + "mean_token_accuracy": 0.8005236983299255, + "num_tokens": 331035300.0, + "step": 14658 + }, + { + "epoch": 1.208790302630494, + "grad_norm": 2.1197402477264404, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7360554933547974, + "num_tokens": 331064279.0, + "step": 14659 + }, + { + "epoch": 1.2088727632555454, + "grad_norm": 1.915426254272461, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7533339858055115, + "num_tokens": 331098431.0, + "step": 14660 + }, + { + "epoch": 1.208955223880597, + "grad_norm": 1.9788594245910645, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.725961446762085, + "num_tokens": 331130828.0, + "step": 14661 + }, + { + "epoch": 1.2090376845056485, + "grad_norm": 1.774677038192749, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7424442768096924, + "num_tokens": 331164911.0, + "step": 14662 + }, + { + "epoch": 1.2091201451307, + "grad_norm": 2.200336456298828, + "learning_rate": 1e-06, + "loss": 0.545, + "mean_token_accuracy": 0.8135399222373962, + "num_tokens": 331189092.0, + "step": 14663 + }, + { + "epoch": 1.2092026057557517, + "grad_norm": 1.8118854761123657, + "learning_rate": 1e-06, + "loss": 0.6147, + "mean_token_accuracy": 0.7963336706161499, + "num_tokens": 331223897.0, + "step": 14664 + }, + { + "epoch": 1.2092850663808032, + "grad_norm": 2.007185697555542, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7631353735923767, + "num_tokens": 331253587.0, + "step": 14665 + }, + { + "epoch": 1.2093675270058548, + "grad_norm": 2.102984666824341, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.771940290927887, + "num_tokens": 331279794.0, + "step": 14666 + }, + { + "epoch": 1.2094499876309062, + "grad_norm": 1.9456220865249634, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7072086930274963, + "num_tokens": 331311335.0, + "step": 14667 + }, + { + "epoch": 1.2095324482559577, + "grad_norm": 2.0398941040039062, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7709510922431946, + "num_tokens": 331339272.0, + "step": 14668 + }, + { + "epoch": 1.2096149088810093, + "grad_norm": 1.9321047067642212, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7625328898429871, + "num_tokens": 331368119.0, + "step": 14669 + }, + { + "epoch": 1.2096973695060609, + "grad_norm": 2.001884698867798, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7895078063011169, + "num_tokens": 331396584.0, + "step": 14670 + }, + { + "epoch": 1.2097798301311125, + "grad_norm": 2.2686381340026855, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7288305759429932, + "num_tokens": 331425303.0, + "step": 14671 + }, + { + "epoch": 1.209862290756164, + "grad_norm": 1.8709224462509155, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.761655330657959, + "num_tokens": 331461281.0, + "step": 14672 + }, + { + "epoch": 1.2099447513812154, + "grad_norm": 2.345282554626465, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7468392848968506, + "num_tokens": 331483842.0, + "step": 14673 + }, + { + "epoch": 1.210027212006267, + "grad_norm": 1.9821354150772095, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7491530179977417, + "num_tokens": 331515659.0, + "step": 14674 + }, + { + "epoch": 1.2101096726313185, + "grad_norm": 1.7912840843200684, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.7859365940093994, + "num_tokens": 331551325.0, + "step": 14675 + }, + { + "epoch": 1.21019213325637, + "grad_norm": 2.1395249366760254, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.776969313621521, + "num_tokens": 331578188.0, + "step": 14676 + }, + { + "epoch": 1.2102745938814217, + "grad_norm": 2.04372239112854, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7532796859741211, + "num_tokens": 331609293.0, + "step": 14677 + }, + { + "epoch": 1.2103570545064732, + "grad_norm": 2.162236452102661, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7710047960281372, + "num_tokens": 331636328.0, + "step": 14678 + }, + { + "epoch": 1.2104395151315246, + "grad_norm": 1.941710352897644, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7291508913040161, + "num_tokens": 331666248.0, + "step": 14679 + }, + { + "epoch": 1.2105219757565762, + "grad_norm": 2.1034672260284424, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7442219853401184, + "num_tokens": 331693149.0, + "step": 14680 + }, + { + "epoch": 1.2106044363816277, + "grad_norm": 2.196256160736084, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7637653946876526, + "num_tokens": 331716754.0, + "step": 14681 + }, + { + "epoch": 1.2106868970066793, + "grad_norm": 2.100740432739258, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.75389564037323, + "num_tokens": 331741958.0, + "step": 14682 + }, + { + "epoch": 1.2107693576317309, + "grad_norm": 2.09647798538208, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.7463775873184204, + "num_tokens": 331771810.0, + "step": 14683 + }, + { + "epoch": 1.2108518182567825, + "grad_norm": 1.7867766618728638, + "learning_rate": 1e-06, + "loss": 0.5985, + "mean_token_accuracy": 0.7963542342185974, + "num_tokens": 331805208.0, + "step": 14684 + }, + { + "epoch": 1.210934278881834, + "grad_norm": 2.2829782962799072, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.779523491859436, + "num_tokens": 331828319.0, + "step": 14685 + }, + { + "epoch": 1.2110167395068854, + "grad_norm": 2.0897960662841797, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.7707986831665039, + "num_tokens": 331855050.0, + "step": 14686 + }, + { + "epoch": 1.211099200131937, + "grad_norm": 1.9800280332565308, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7588697671890259, + "num_tokens": 331884028.0, + "step": 14687 + }, + { + "epoch": 1.2111816607569885, + "grad_norm": 2.0368363857269287, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7352052927017212, + "num_tokens": 331913692.0, + "step": 14688 + }, + { + "epoch": 1.21126412138204, + "grad_norm": 2.3093039989471436, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7658777236938477, + "num_tokens": 331937430.0, + "step": 14689 + }, + { + "epoch": 1.2113465820070917, + "grad_norm": 1.9644911289215088, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7458436489105225, + "num_tokens": 331968482.0, + "step": 14690 + }, + { + "epoch": 1.2114290426321432, + "grad_norm": 1.9828321933746338, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7271972894668579, + "num_tokens": 331999589.0, + "step": 14691 + }, + { + "epoch": 1.2115115032571948, + "grad_norm": 1.9122432470321655, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.733607292175293, + "num_tokens": 332030465.0, + "step": 14692 + }, + { + "epoch": 1.2115939638822462, + "grad_norm": 2.0960028171539307, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.7834064960479736, + "num_tokens": 332056672.0, + "step": 14693 + }, + { + "epoch": 1.2116764245072977, + "grad_norm": 1.9287358522415161, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.7466199398040771, + "num_tokens": 332089087.0, + "step": 14694 + }, + { + "epoch": 1.2117588851323493, + "grad_norm": 1.775246024131775, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7461857795715332, + "num_tokens": 332125954.0, + "step": 14695 + }, + { + "epoch": 1.2118413457574009, + "grad_norm": 1.943708896636963, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7372753024101257, + "num_tokens": 332153809.0, + "step": 14696 + }, + { + "epoch": 1.2119238063824525, + "grad_norm": 2.4095120429992676, + "learning_rate": 1e-06, + "loss": 0.6193, + "mean_token_accuracy": 0.7971799969673157, + "num_tokens": 332182026.0, + "step": 14697 + }, + { + "epoch": 1.2120062670075038, + "grad_norm": 2.065791606903076, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7734804153442383, + "num_tokens": 332208897.0, + "step": 14698 + }, + { + "epoch": 1.2120887276325554, + "grad_norm": 1.8877111673355103, + "learning_rate": 1e-06, + "loss": 0.7531, + "mean_token_accuracy": 0.7486957311630249, + "num_tokens": 332241792.0, + "step": 14699 + }, + { + "epoch": 1.212171188257607, + "grad_norm": 2.0859785079956055, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.744636058807373, + "num_tokens": 332270202.0, + "step": 14700 + }, + { + "epoch": 1.2122536488826585, + "grad_norm": 2.1631968021392822, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.751158595085144, + "num_tokens": 332298806.0, + "step": 14701 + }, + { + "epoch": 1.21233610950771, + "grad_norm": 2.224228858947754, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7435784339904785, + "num_tokens": 332324783.0, + "step": 14702 + }, + { + "epoch": 1.2124185701327617, + "grad_norm": 1.968198537826538, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.7950743436813354, + "num_tokens": 332354205.0, + "step": 14703 + }, + { + "epoch": 1.2125010307578132, + "grad_norm": 2.136692762374878, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7500520348548889, + "num_tokens": 332384279.0, + "step": 14704 + }, + { + "epoch": 1.2125834913828646, + "grad_norm": 2.1952686309814453, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.7880841493606567, + "num_tokens": 332407328.0, + "step": 14705 + }, + { + "epoch": 1.2126659520079162, + "grad_norm": 1.746098518371582, + "learning_rate": 1e-06, + "loss": 0.5527, + "mean_token_accuracy": 0.8172831535339355, + "num_tokens": 332441208.0, + "step": 14706 + }, + { + "epoch": 1.2127484126329677, + "grad_norm": 2.1849772930145264, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.727555513381958, + "num_tokens": 332467794.0, + "step": 14707 + }, + { + "epoch": 1.2128308732580193, + "grad_norm": 1.845853328704834, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.8041014075279236, + "num_tokens": 332499135.0, + "step": 14708 + }, + { + "epoch": 1.2129133338830709, + "grad_norm": 1.7808105945587158, + "learning_rate": 1e-06, + "loss": 0.5608, + "mean_token_accuracy": 0.8117378354072571, + "num_tokens": 332533834.0, + "step": 14709 + }, + { + "epoch": 1.2129957945081224, + "grad_norm": 2.0081286430358887, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.7903069257736206, + "num_tokens": 332562762.0, + "step": 14710 + }, + { + "epoch": 1.213078255133174, + "grad_norm": 1.92616868019104, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7492730617523193, + "num_tokens": 332592570.0, + "step": 14711 + }, + { + "epoch": 1.2131607157582254, + "grad_norm": 1.959094762802124, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7188928127288818, + "num_tokens": 332629128.0, + "step": 14712 + }, + { + "epoch": 1.213243176383277, + "grad_norm": 1.8591411113739014, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7415558099746704, + "num_tokens": 332662752.0, + "step": 14713 + }, + { + "epoch": 1.2133256370083285, + "grad_norm": 1.7841877937316895, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7716410756111145, + "num_tokens": 332696763.0, + "step": 14714 + }, + { + "epoch": 1.21340809763338, + "grad_norm": 1.87321138381958, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.774847149848938, + "num_tokens": 332728680.0, + "step": 14715 + }, + { + "epoch": 1.2134905582584317, + "grad_norm": 1.8393528461456299, + "learning_rate": 1e-06, + "loss": 0.784, + "mean_token_accuracy": 0.7537339329719543, + "num_tokens": 332765105.0, + "step": 14716 + }, + { + "epoch": 1.213573018883483, + "grad_norm": 1.9572954177856445, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7686428427696228, + "num_tokens": 332794452.0, + "step": 14717 + }, + { + "epoch": 1.2136554795085346, + "grad_norm": 2.121561050415039, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.769205629825592, + "num_tokens": 332821791.0, + "step": 14718 + }, + { + "epoch": 1.2137379401335862, + "grad_norm": 2.2720539569854736, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7248088121414185, + "num_tokens": 332847312.0, + "step": 14719 + }, + { + "epoch": 1.2138204007586377, + "grad_norm": 1.7124290466308594, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.7958145141601562, + "num_tokens": 332885345.0, + "step": 14720 + }, + { + "epoch": 1.2139028613836893, + "grad_norm": 2.0491347312927246, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7304122447967529, + "num_tokens": 332917647.0, + "step": 14721 + }, + { + "epoch": 1.2139853220087409, + "grad_norm": 1.7198712825775146, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7659159898757935, + "num_tokens": 332957451.0, + "step": 14722 + }, + { + "epoch": 1.2140677826337924, + "grad_norm": 1.714179277420044, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.7809199094772339, + "num_tokens": 332994012.0, + "step": 14723 + }, + { + "epoch": 1.2141502432588438, + "grad_norm": 1.992293357849121, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7628044486045837, + "num_tokens": 333027244.0, + "step": 14724 + }, + { + "epoch": 1.2142327038838954, + "grad_norm": 2.0650315284729004, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7801946401596069, + "num_tokens": 333056598.0, + "step": 14725 + }, + { + "epoch": 1.214315164508947, + "grad_norm": 1.7367044687271118, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7673119306564331, + "num_tokens": 333093635.0, + "step": 14726 + }, + { + "epoch": 1.2143976251339985, + "grad_norm": 1.8392918109893799, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7404788732528687, + "num_tokens": 333126100.0, + "step": 14727 + }, + { + "epoch": 1.21448008575905, + "grad_norm": 1.9898467063903809, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7697275876998901, + "num_tokens": 333155646.0, + "step": 14728 + }, + { + "epoch": 1.2145625463841017, + "grad_norm": 2.0182652473449707, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7206323146820068, + "num_tokens": 333185118.0, + "step": 14729 + }, + { + "epoch": 1.2146450070091532, + "grad_norm": 2.0527076721191406, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.771578311920166, + "num_tokens": 333214247.0, + "step": 14730 + }, + { + "epoch": 1.2147274676342046, + "grad_norm": 1.9250577688217163, + "learning_rate": 1e-06, + "loss": 0.7724, + "mean_token_accuracy": 0.750610888004303, + "num_tokens": 333248392.0, + "step": 14731 + }, + { + "epoch": 1.2148099282592562, + "grad_norm": 1.7772471904754639, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7517633438110352, + "num_tokens": 333283783.0, + "step": 14732 + }, + { + "epoch": 1.2148923888843077, + "grad_norm": 1.898211121559143, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7721256017684937, + "num_tokens": 333314056.0, + "step": 14733 + }, + { + "epoch": 1.2149748495093593, + "grad_norm": 1.7639687061309814, + "learning_rate": 1e-06, + "loss": 0.5711, + "mean_token_accuracy": 0.8128830194473267, + "num_tokens": 333347131.0, + "step": 14734 + }, + { + "epoch": 1.2150573101344109, + "grad_norm": 1.692538857460022, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7266721129417419, + "num_tokens": 333387858.0, + "step": 14735 + }, + { + "epoch": 1.2151397707594624, + "grad_norm": 2.034918785095215, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7849677801132202, + "num_tokens": 333414692.0, + "step": 14736 + }, + { + "epoch": 1.215222231384514, + "grad_norm": 2.036339282989502, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7659069895744324, + "num_tokens": 333444628.0, + "step": 14737 + }, + { + "epoch": 1.2153046920095654, + "grad_norm": 1.9144185781478882, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7712527513504028, + "num_tokens": 333475019.0, + "step": 14738 + }, + { + "epoch": 1.215387152634617, + "grad_norm": 2.0304527282714844, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7912795543670654, + "num_tokens": 333500441.0, + "step": 14739 + }, + { + "epoch": 1.2154696132596685, + "grad_norm": 2.1657321453094482, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7399352788925171, + "num_tokens": 333527647.0, + "step": 14740 + }, + { + "epoch": 1.21555207388472, + "grad_norm": 2.0369744300842285, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.723944902420044, + "num_tokens": 333559542.0, + "step": 14741 + }, + { + "epoch": 1.2156345345097717, + "grad_norm": 1.8508707284927368, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7721927165985107, + "num_tokens": 333591612.0, + "step": 14742 + }, + { + "epoch": 1.215716995134823, + "grad_norm": 2.0121257305145264, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7613658905029297, + "num_tokens": 333620072.0, + "step": 14743 + }, + { + "epoch": 1.2157994557598746, + "grad_norm": 1.7428054809570312, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7588620185852051, + "num_tokens": 333657965.0, + "step": 14744 + }, + { + "epoch": 1.2158819163849262, + "grad_norm": 1.9352154731750488, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.744591474533081, + "num_tokens": 333692006.0, + "step": 14745 + }, + { + "epoch": 1.2159643770099777, + "grad_norm": 2.3801984786987305, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7077689170837402, + "num_tokens": 333717999.0, + "step": 14746 + }, + { + "epoch": 1.2160468376350293, + "grad_norm": 1.8878803253173828, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7261338829994202, + "num_tokens": 333754197.0, + "step": 14747 + }, + { + "epoch": 1.2161292982600809, + "grad_norm": 2.2527973651885986, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7361673712730408, + "num_tokens": 333778700.0, + "step": 14748 + }, + { + "epoch": 1.2162117588851324, + "grad_norm": 2.0793209075927734, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7309285402297974, + "num_tokens": 333808206.0, + "step": 14749 + }, + { + "epoch": 1.2162942195101838, + "grad_norm": 1.902654767036438, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7787808179855347, + "num_tokens": 333840527.0, + "step": 14750 + }, + { + "epoch": 1.2163766801352354, + "grad_norm": 1.782680869102478, + "learning_rate": 1e-06, + "loss": 0.5632, + "mean_token_accuracy": 0.8206366300582886, + "num_tokens": 333875749.0, + "step": 14751 + }, + { + "epoch": 1.216459140760287, + "grad_norm": 2.06915283203125, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7549343109130859, + "num_tokens": 333907068.0, + "step": 14752 + }, + { + "epoch": 1.2165416013853385, + "grad_norm": 2.234602212905884, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7525131106376648, + "num_tokens": 333932105.0, + "step": 14753 + }, + { + "epoch": 1.21662406201039, + "grad_norm": 1.9432016611099243, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7825099229812622, + "num_tokens": 333964912.0, + "step": 14754 + }, + { + "epoch": 1.2167065226354417, + "grad_norm": 1.815068244934082, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7588458061218262, + "num_tokens": 333999192.0, + "step": 14755 + }, + { + "epoch": 1.2167889832604932, + "grad_norm": 1.8662114143371582, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7667500972747803, + "num_tokens": 334032487.0, + "step": 14756 + }, + { + "epoch": 1.2168714438855446, + "grad_norm": 2.2251484394073486, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7589905858039856, + "num_tokens": 334056680.0, + "step": 14757 + }, + { + "epoch": 1.2169539045105962, + "grad_norm": 2.318535089492798, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7239338159561157, + "num_tokens": 334080920.0, + "step": 14758 + }, + { + "epoch": 1.2170363651356477, + "grad_norm": 2.0802018642425537, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7509545087814331, + "num_tokens": 334108440.0, + "step": 14759 + }, + { + "epoch": 1.2171188257606993, + "grad_norm": 1.9209836721420288, + "learning_rate": 1e-06, + "loss": 0.7563, + "mean_token_accuracy": 0.7508108615875244, + "num_tokens": 334138857.0, + "step": 14760 + }, + { + "epoch": 1.2172012863857509, + "grad_norm": 2.0028016567230225, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7429704666137695, + "num_tokens": 334169096.0, + "step": 14761 + }, + { + "epoch": 1.2172837470108022, + "grad_norm": 1.9542218446731567, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7598429918289185, + "num_tokens": 334200957.0, + "step": 14762 + }, + { + "epoch": 1.2173662076358538, + "grad_norm": 1.963975191116333, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7567402124404907, + "num_tokens": 334230333.0, + "step": 14763 + }, + { + "epoch": 1.2174486682609054, + "grad_norm": 1.9437823295593262, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7325763702392578, + "num_tokens": 334264838.0, + "step": 14764 + }, + { + "epoch": 1.217531128885957, + "grad_norm": 1.938078761100769, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7735385894775391, + "num_tokens": 334295621.0, + "step": 14765 + }, + { + "epoch": 1.2176135895110085, + "grad_norm": 2.1673026084899902, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7634830474853516, + "num_tokens": 334320917.0, + "step": 14766 + }, + { + "epoch": 1.21769605013606, + "grad_norm": 2.138174295425415, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.7918590307235718, + "num_tokens": 334348331.0, + "step": 14767 + }, + { + "epoch": 1.2177785107611117, + "grad_norm": 2.391857624053955, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.769187331199646, + "num_tokens": 334371097.0, + "step": 14768 + }, + { + "epoch": 1.217860971386163, + "grad_norm": 2.0973689556121826, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7694206833839417, + "num_tokens": 334396641.0, + "step": 14769 + }, + { + "epoch": 1.2179434320112146, + "grad_norm": 1.8967130184173584, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.772135853767395, + "num_tokens": 334430292.0, + "step": 14770 + }, + { + "epoch": 1.2180258926362661, + "grad_norm": 1.9031850099563599, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7804397344589233, + "num_tokens": 334464639.0, + "step": 14771 + }, + { + "epoch": 1.2181083532613177, + "grad_norm": 2.1703503131866455, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7194849252700806, + "num_tokens": 334493451.0, + "step": 14772 + }, + { + "epoch": 1.2181908138863693, + "grad_norm": 2.1898481845855713, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7559505105018616, + "num_tokens": 334521701.0, + "step": 14773 + }, + { + "epoch": 1.2182732745114209, + "grad_norm": 2.0572543144226074, + "learning_rate": 1e-06, + "loss": 0.5895, + "mean_token_accuracy": 0.7990500926971436, + "num_tokens": 334548946.0, + "step": 14774 + }, + { + "epoch": 1.2183557351364724, + "grad_norm": 2.3852345943450928, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.779153048992157, + "num_tokens": 334574526.0, + "step": 14775 + }, + { + "epoch": 1.2184381957615238, + "grad_norm": 1.8641107082366943, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7623960971832275, + "num_tokens": 334611182.0, + "step": 14776 + }, + { + "epoch": 1.2185206563865754, + "grad_norm": 1.8162918090820312, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7382097244262695, + "num_tokens": 334649241.0, + "step": 14777 + }, + { + "epoch": 1.218603117011627, + "grad_norm": 2.23339581489563, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7205509543418884, + "num_tokens": 334674509.0, + "step": 14778 + }, + { + "epoch": 1.2186855776366785, + "grad_norm": 2.068331241607666, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7433410882949829, + "num_tokens": 334704142.0, + "step": 14779 + }, + { + "epoch": 1.21876803826173, + "grad_norm": 2.1682651042938232, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7680089473724365, + "num_tokens": 334729490.0, + "step": 14780 + }, + { + "epoch": 1.2188504988867817, + "grad_norm": 2.123391628265381, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7316904664039612, + "num_tokens": 334757906.0, + "step": 14781 + }, + { + "epoch": 1.218932959511833, + "grad_norm": 1.7975897789001465, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7473604679107666, + "num_tokens": 334795786.0, + "step": 14782 + }, + { + "epoch": 1.2190154201368846, + "grad_norm": 2.212554931640625, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7804802656173706, + "num_tokens": 334820208.0, + "step": 14783 + }, + { + "epoch": 1.2190978807619361, + "grad_norm": 2.1238274574279785, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7405127286911011, + "num_tokens": 334847792.0, + "step": 14784 + }, + { + "epoch": 1.2191803413869877, + "grad_norm": 2.0224034786224365, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.748530924320221, + "num_tokens": 334878780.0, + "step": 14785 + }, + { + "epoch": 1.2192628020120393, + "grad_norm": 1.794168472290039, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7558762431144714, + "num_tokens": 334916247.0, + "step": 14786 + }, + { + "epoch": 1.2193452626370909, + "grad_norm": 1.9451379776000977, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7669041752815247, + "num_tokens": 334947979.0, + "step": 14787 + }, + { + "epoch": 1.2194277232621422, + "grad_norm": 1.8650832176208496, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7662636041641235, + "num_tokens": 334981345.0, + "step": 14788 + }, + { + "epoch": 1.2195101838871938, + "grad_norm": 1.941153883934021, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7130130529403687, + "num_tokens": 335017727.0, + "step": 14789 + }, + { + "epoch": 1.2195926445122454, + "grad_norm": 1.915534496307373, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7695213556289673, + "num_tokens": 335050851.0, + "step": 14790 + }, + { + "epoch": 1.219675105137297, + "grad_norm": 2.08612322807312, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7446321845054626, + "num_tokens": 335079696.0, + "step": 14791 + }, + { + "epoch": 1.2197575657623485, + "grad_norm": 1.9057927131652832, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7603870630264282, + "num_tokens": 335111186.0, + "step": 14792 + }, + { + "epoch": 1.2198400263874, + "grad_norm": 1.9734201431274414, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7645336985588074, + "num_tokens": 335143801.0, + "step": 14793 + }, + { + "epoch": 1.2199224870124517, + "grad_norm": 2.050527334213257, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7591522336006165, + "num_tokens": 335173844.0, + "step": 14794 + }, + { + "epoch": 1.220004947637503, + "grad_norm": 2.137049436569214, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7552359104156494, + "num_tokens": 335201486.0, + "step": 14795 + }, + { + "epoch": 1.2200874082625546, + "grad_norm": 2.206061840057373, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7683332562446594, + "num_tokens": 335228194.0, + "step": 14796 + }, + { + "epoch": 1.2201698688876061, + "grad_norm": 1.9062458276748657, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7742055654525757, + "num_tokens": 335258711.0, + "step": 14797 + }, + { + "epoch": 1.2202523295126577, + "grad_norm": 1.8729639053344727, + "learning_rate": 1e-06, + "loss": 0.6091, + "mean_token_accuracy": 0.7880606055259705, + "num_tokens": 335294103.0, + "step": 14798 + }, + { + "epoch": 1.2203347901377093, + "grad_norm": 1.9976472854614258, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7530496120452881, + "num_tokens": 335327383.0, + "step": 14799 + }, + { + "epoch": 1.2204172507627609, + "grad_norm": 2.111166000366211, + "learning_rate": 1e-06, + "loss": 0.578, + "mean_token_accuracy": 0.7974400520324707, + "num_tokens": 335350835.0, + "step": 14800 + }, + { + "epoch": 1.2204997113878124, + "grad_norm": 2.0009684562683105, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7628072500228882, + "num_tokens": 335382631.0, + "step": 14801 + }, + { + "epoch": 1.2205821720128638, + "grad_norm": 2.0650224685668945, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.7826393246650696, + "num_tokens": 335409088.0, + "step": 14802 + }, + { + "epoch": 1.2206646326379154, + "grad_norm": 1.77312171459198, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.789759635925293, + "num_tokens": 335441901.0, + "step": 14803 + }, + { + "epoch": 1.220747093262967, + "grad_norm": 2.078653573989868, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7362861633300781, + "num_tokens": 335471133.0, + "step": 14804 + }, + { + "epoch": 1.2208295538880185, + "grad_norm": 2.1060104370117188, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.7878512144088745, + "num_tokens": 335499244.0, + "step": 14805 + }, + { + "epoch": 1.22091201451307, + "grad_norm": 1.9074307680130005, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7378082275390625, + "num_tokens": 335533557.0, + "step": 14806 + }, + { + "epoch": 1.2209944751381214, + "grad_norm": 2.0211181640625, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7758746147155762, + "num_tokens": 335564513.0, + "step": 14807 + }, + { + "epoch": 1.221076935763173, + "grad_norm": 1.8220065832138062, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.8023233413696289, + "num_tokens": 335595825.0, + "step": 14808 + }, + { + "epoch": 1.2211593963882246, + "grad_norm": 2.038020372390747, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7450634241104126, + "num_tokens": 335629244.0, + "step": 14809 + }, + { + "epoch": 1.2212418570132761, + "grad_norm": 2.2576749324798584, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7256861925125122, + "num_tokens": 335653847.0, + "step": 14810 + }, + { + "epoch": 1.2213243176383277, + "grad_norm": 2.0582187175750732, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7735261917114258, + "num_tokens": 335682261.0, + "step": 14811 + }, + { + "epoch": 1.2214067782633793, + "grad_norm": 2.0939581394195557, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7528789043426514, + "num_tokens": 335712242.0, + "step": 14812 + }, + { + "epoch": 1.2214892388884309, + "grad_norm": 1.9040178060531616, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.741340696811676, + "num_tokens": 335747808.0, + "step": 14813 + }, + { + "epoch": 1.2215716995134822, + "grad_norm": 2.0708959102630615, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7704814672470093, + "num_tokens": 335774509.0, + "step": 14814 + }, + { + "epoch": 1.2216541601385338, + "grad_norm": 1.800720453262329, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7748059630393982, + "num_tokens": 335808056.0, + "step": 14815 + }, + { + "epoch": 1.2217366207635854, + "grad_norm": 2.159329891204834, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7685894966125488, + "num_tokens": 335834322.0, + "step": 14816 + }, + { + "epoch": 1.221819081388637, + "grad_norm": 2.0264487266540527, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7233438491821289, + "num_tokens": 335865036.0, + "step": 14817 + }, + { + "epoch": 1.2219015420136885, + "grad_norm": 1.709639549255371, + "learning_rate": 1e-06, + "loss": 0.6047, + "mean_token_accuracy": 0.8083404302597046, + "num_tokens": 335901887.0, + "step": 14818 + }, + { + "epoch": 1.22198400263874, + "grad_norm": 2.211127996444702, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7642673850059509, + "num_tokens": 335926218.0, + "step": 14819 + }, + { + "epoch": 1.2220664632637916, + "grad_norm": 2.6919779777526855, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7434749007225037, + "num_tokens": 335954013.0, + "step": 14820 + }, + { + "epoch": 1.222148923888843, + "grad_norm": 2.011936664581299, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7769936919212341, + "num_tokens": 335984467.0, + "step": 14821 + }, + { + "epoch": 1.2222313845138946, + "grad_norm": 2.307201385498047, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7667958736419678, + "num_tokens": 336006480.0, + "step": 14822 + }, + { + "epoch": 1.2223138451389461, + "grad_norm": 2.0343406200408936, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.7501646280288696, + "num_tokens": 336036036.0, + "step": 14823 + }, + { + "epoch": 1.2223963057639977, + "grad_norm": 2.096179723739624, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7462745904922485, + "num_tokens": 336064178.0, + "step": 14824 + }, + { + "epoch": 1.2224787663890493, + "grad_norm": 1.8627628087997437, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7266794443130493, + "num_tokens": 336099990.0, + "step": 14825 + }, + { + "epoch": 1.2225612270141009, + "grad_norm": 2.06482195854187, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7617281675338745, + "num_tokens": 336129260.0, + "step": 14826 + }, + { + "epoch": 1.2226436876391522, + "grad_norm": 1.7234140634536743, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7758769989013672, + "num_tokens": 336167509.0, + "step": 14827 + }, + { + "epoch": 1.2227261482642038, + "grad_norm": 1.939449667930603, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7373062372207642, + "num_tokens": 336200773.0, + "step": 14828 + }, + { + "epoch": 1.2228086088892554, + "grad_norm": 1.8029950857162476, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7728481292724609, + "num_tokens": 336236540.0, + "step": 14829 + }, + { + "epoch": 1.222891069514307, + "grad_norm": 2.070600986480713, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7617233395576477, + "num_tokens": 336262869.0, + "step": 14830 + }, + { + "epoch": 1.2229735301393585, + "grad_norm": 2.0344398021698, + "learning_rate": 1e-06, + "loss": 0.6096, + "mean_token_accuracy": 0.8078310489654541, + "num_tokens": 336287441.0, + "step": 14831 + }, + { + "epoch": 1.22305599076441, + "grad_norm": 1.8857511281967163, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.7960361242294312, + "num_tokens": 336320031.0, + "step": 14832 + }, + { + "epoch": 1.2231384513894614, + "grad_norm": 2.2372145652770996, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7852286100387573, + "num_tokens": 336345203.0, + "step": 14833 + }, + { + "epoch": 1.223220912014513, + "grad_norm": 2.394270420074463, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7722159624099731, + "num_tokens": 336365659.0, + "step": 14834 + }, + { + "epoch": 1.2233033726395646, + "grad_norm": 2.016772508621216, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7566766738891602, + "num_tokens": 336394140.0, + "step": 14835 + }, + { + "epoch": 1.2233858332646161, + "grad_norm": 2.038832902908325, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.7815923094749451, + "num_tokens": 336420837.0, + "step": 14836 + }, + { + "epoch": 1.2234682938896677, + "grad_norm": 2.00386905670166, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7499871850013733, + "num_tokens": 336449240.0, + "step": 14837 + }, + { + "epoch": 1.2235507545147193, + "grad_norm": 2.1084439754486084, + "learning_rate": 1e-06, + "loss": 0.7886, + "mean_token_accuracy": 0.7417517900466919, + "num_tokens": 336481515.0, + "step": 14838 + }, + { + "epoch": 1.2236332151397709, + "grad_norm": 2.030459403991699, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.77034592628479, + "num_tokens": 336507985.0, + "step": 14839 + }, + { + "epoch": 1.2237156757648222, + "grad_norm": 1.8129655122756958, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.7994762063026428, + "num_tokens": 336546270.0, + "step": 14840 + }, + { + "epoch": 1.2237981363898738, + "grad_norm": 1.9155405759811401, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7560091018676758, + "num_tokens": 336579487.0, + "step": 14841 + }, + { + "epoch": 1.2238805970149254, + "grad_norm": 1.9590725898742676, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.772194504737854, + "num_tokens": 336612775.0, + "step": 14842 + }, + { + "epoch": 1.223963057639977, + "grad_norm": 1.6939548254013062, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.771903932094574, + "num_tokens": 336651123.0, + "step": 14843 + }, + { + "epoch": 1.2240455182650285, + "grad_norm": 1.9036967754364014, + "learning_rate": 1e-06, + "loss": 0.5922, + "mean_token_accuracy": 0.7968907356262207, + "num_tokens": 336685463.0, + "step": 14844 + }, + { + "epoch": 1.22412797889008, + "grad_norm": 1.8322997093200684, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.743187665939331, + "num_tokens": 336722245.0, + "step": 14845 + }, + { + "epoch": 1.2242104395151316, + "grad_norm": 2.1442322731018066, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.774486780166626, + "num_tokens": 336752104.0, + "step": 14846 + }, + { + "epoch": 1.224292900140183, + "grad_norm": 1.9541953802108765, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7630182504653931, + "num_tokens": 336781205.0, + "step": 14847 + }, + { + "epoch": 1.2243753607652346, + "grad_norm": 1.8201831579208374, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7569926977157593, + "num_tokens": 336818825.0, + "step": 14848 + }, + { + "epoch": 1.2244578213902861, + "grad_norm": 2.117197275161743, + "learning_rate": 1e-06, + "loss": 0.5761, + "mean_token_accuracy": 0.7980567812919617, + "num_tokens": 336843926.0, + "step": 14849 + }, + { + "epoch": 1.2245402820153377, + "grad_norm": 1.945549488067627, + "learning_rate": 1e-06, + "loss": 0.5935, + "mean_token_accuracy": 0.7961777448654175, + "num_tokens": 336875668.0, + "step": 14850 + }, + { + "epoch": 1.2246227426403893, + "grad_norm": 2.129390239715576, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.7832425832748413, + "num_tokens": 336902025.0, + "step": 14851 + }, + { + "epoch": 1.2247052032654406, + "grad_norm": 2.027164936065674, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7746462821960449, + "num_tokens": 336930271.0, + "step": 14852 + }, + { + "epoch": 1.2247876638904922, + "grad_norm": 1.8008193969726562, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7834104299545288, + "num_tokens": 336963926.0, + "step": 14853 + }, + { + "epoch": 1.2248701245155438, + "grad_norm": 2.0718514919281006, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7684271931648254, + "num_tokens": 336993308.0, + "step": 14854 + }, + { + "epoch": 1.2249525851405954, + "grad_norm": 1.8329405784606934, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7480288743972778, + "num_tokens": 337027172.0, + "step": 14855 + }, + { + "epoch": 1.225035045765647, + "grad_norm": 1.8136030435562134, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.8018207550048828, + "num_tokens": 337060819.0, + "step": 14856 + }, + { + "epoch": 1.2251175063906985, + "grad_norm": 1.8355768918991089, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7533669471740723, + "num_tokens": 337095383.0, + "step": 14857 + }, + { + "epoch": 1.22519996701575, + "grad_norm": 2.42305064201355, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7777589559555054, + "num_tokens": 337116142.0, + "step": 14858 + }, + { + "epoch": 1.2252824276408014, + "grad_norm": 2.0459580421447754, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7817314863204956, + "num_tokens": 337145027.0, + "step": 14859 + }, + { + "epoch": 1.225364888265853, + "grad_norm": 2.4822165966033936, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7442144155502319, + "num_tokens": 337167192.0, + "step": 14860 + }, + { + "epoch": 1.2254473488909046, + "grad_norm": 2.055345296859741, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7831683158874512, + "num_tokens": 337193145.0, + "step": 14861 + }, + { + "epoch": 1.2255298095159561, + "grad_norm": 2.140010118484497, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7317247986793518, + "num_tokens": 337219566.0, + "step": 14862 + }, + { + "epoch": 1.2256122701410077, + "grad_norm": 1.921796441078186, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7613869309425354, + "num_tokens": 337255686.0, + "step": 14863 + }, + { + "epoch": 1.2256947307660593, + "grad_norm": 1.9540600776672363, + "learning_rate": 1e-06, + "loss": 0.6057, + "mean_token_accuracy": 0.8033105134963989, + "num_tokens": 337283418.0, + "step": 14864 + }, + { + "epoch": 1.2257771913911109, + "grad_norm": 2.284834861755371, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7272894382476807, + "num_tokens": 337306992.0, + "step": 14865 + }, + { + "epoch": 1.2258596520161622, + "grad_norm": 1.9369690418243408, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.7833069562911987, + "num_tokens": 337339795.0, + "step": 14866 + }, + { + "epoch": 1.2259421126412138, + "grad_norm": 1.9861364364624023, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7766460180282593, + "num_tokens": 337367948.0, + "step": 14867 + }, + { + "epoch": 1.2260245732662653, + "grad_norm": 2.101287841796875, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7144688367843628, + "num_tokens": 337396273.0, + "step": 14868 + }, + { + "epoch": 1.226107033891317, + "grad_norm": 1.9889402389526367, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7394658923149109, + "num_tokens": 337429268.0, + "step": 14869 + }, + { + "epoch": 1.2261894945163685, + "grad_norm": 2.1639039516448975, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7676609754562378, + "num_tokens": 337455603.0, + "step": 14870 + }, + { + "epoch": 1.22627195514142, + "grad_norm": 1.9223968982696533, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.744178295135498, + "num_tokens": 337487316.0, + "step": 14871 + }, + { + "epoch": 1.2263544157664714, + "grad_norm": 1.9046298265457153, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7857237458229065, + "num_tokens": 337516811.0, + "step": 14872 + }, + { + "epoch": 1.226436876391523, + "grad_norm": 1.9284828901290894, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7554716467857361, + "num_tokens": 337551325.0, + "step": 14873 + }, + { + "epoch": 1.2265193370165746, + "grad_norm": 2.2544500827789307, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7446772456169128, + "num_tokens": 337577151.0, + "step": 14874 + }, + { + "epoch": 1.2266017976416261, + "grad_norm": 2.119990587234497, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7303225994110107, + "num_tokens": 337605376.0, + "step": 14875 + }, + { + "epoch": 1.2266842582666777, + "grad_norm": 1.8907742500305176, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7766715884208679, + "num_tokens": 337637035.0, + "step": 14876 + }, + { + "epoch": 1.2267667188917293, + "grad_norm": 2.0886452198028564, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7292455434799194, + "num_tokens": 337664770.0, + "step": 14877 + }, + { + "epoch": 1.2268491795167806, + "grad_norm": 1.948779821395874, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7643996477127075, + "num_tokens": 337695097.0, + "step": 14878 + }, + { + "epoch": 1.2269316401418322, + "grad_norm": 1.8516122102737427, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7792131900787354, + "num_tokens": 337727909.0, + "step": 14879 + }, + { + "epoch": 1.2270141007668838, + "grad_norm": 2.178356885910034, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7310333847999573, + "num_tokens": 337755046.0, + "step": 14880 + }, + { + "epoch": 1.2270965613919353, + "grad_norm": 1.9146850109100342, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7599702477455139, + "num_tokens": 337787909.0, + "step": 14881 + }, + { + "epoch": 1.227179022016987, + "grad_norm": 1.95987868309021, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7402685880661011, + "num_tokens": 337820297.0, + "step": 14882 + }, + { + "epoch": 1.2272614826420385, + "grad_norm": 2.183253288269043, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.7865618467330933, + "num_tokens": 337847033.0, + "step": 14883 + }, + { + "epoch": 1.22734394326709, + "grad_norm": 2.1791000366210938, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7707113027572632, + "num_tokens": 337872005.0, + "step": 14884 + }, + { + "epoch": 1.2274264038921414, + "grad_norm": 1.8930628299713135, + "learning_rate": 1e-06, + "loss": 0.6147, + "mean_token_accuracy": 0.7949317693710327, + "num_tokens": 337904319.0, + "step": 14885 + }, + { + "epoch": 1.227508864517193, + "grad_norm": 1.9663610458374023, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7727125883102417, + "num_tokens": 337935196.0, + "step": 14886 + }, + { + "epoch": 1.2275913251422446, + "grad_norm": 2.1466166973114014, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7529337406158447, + "num_tokens": 337961157.0, + "step": 14887 + }, + { + "epoch": 1.2276737857672961, + "grad_norm": 2.0834574699401855, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7676803469657898, + "num_tokens": 337986926.0, + "step": 14888 + }, + { + "epoch": 1.2277562463923477, + "grad_norm": 1.8125702142715454, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.784954309463501, + "num_tokens": 338020466.0, + "step": 14889 + }, + { + "epoch": 1.2278387070173993, + "grad_norm": 1.8727836608886719, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7715545892715454, + "num_tokens": 338053723.0, + "step": 14890 + }, + { + "epoch": 1.2279211676424509, + "grad_norm": 1.9911632537841797, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7787411212921143, + "num_tokens": 338082212.0, + "step": 14891 + }, + { + "epoch": 1.2280036282675022, + "grad_norm": 1.8946141004562378, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.8024392127990723, + "num_tokens": 338111889.0, + "step": 14892 + }, + { + "epoch": 1.2280860888925538, + "grad_norm": 1.6614423990249634, + "learning_rate": 1e-06, + "loss": 0.6034, + "mean_token_accuracy": 0.8071080446243286, + "num_tokens": 338150084.0, + "step": 14893 + }, + { + "epoch": 1.2281685495176053, + "grad_norm": 1.9423333406448364, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7146705389022827, + "num_tokens": 338181503.0, + "step": 14894 + }, + { + "epoch": 1.228251010142657, + "grad_norm": 1.922132134437561, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7802204489707947, + "num_tokens": 338213939.0, + "step": 14895 + }, + { + "epoch": 1.2283334707677085, + "grad_norm": 1.970089316368103, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7686001062393188, + "num_tokens": 338244454.0, + "step": 14896 + }, + { + "epoch": 1.2284159313927598, + "grad_norm": 1.8563610315322876, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.7985070943832397, + "num_tokens": 338279203.0, + "step": 14897 + }, + { + "epoch": 1.2284983920178114, + "grad_norm": 1.7855147123336792, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7829259634017944, + "num_tokens": 338312868.0, + "step": 14898 + }, + { + "epoch": 1.228580852642863, + "grad_norm": 2.0913267135620117, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7771517038345337, + "num_tokens": 338341778.0, + "step": 14899 + }, + { + "epoch": 1.2286633132679146, + "grad_norm": 1.9886904954910278, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.781536340713501, + "num_tokens": 338371649.0, + "step": 14900 + }, + { + "epoch": 1.2287457738929661, + "grad_norm": 2.144749164581299, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7592657804489136, + "num_tokens": 338400206.0, + "step": 14901 + }, + { + "epoch": 1.2288282345180177, + "grad_norm": 1.7662169933319092, + "learning_rate": 1e-06, + "loss": 0.5628, + "mean_token_accuracy": 0.796124279499054, + "num_tokens": 338432469.0, + "step": 14902 + }, + { + "epoch": 1.2289106951430693, + "grad_norm": 1.9578713178634644, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7799451351165771, + "num_tokens": 338464636.0, + "step": 14903 + }, + { + "epoch": 1.2289931557681206, + "grad_norm": 2.157942295074463, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7627573013305664, + "num_tokens": 338491351.0, + "step": 14904 + }, + { + "epoch": 1.2290756163931722, + "grad_norm": 2.1582162380218506, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7612802982330322, + "num_tokens": 338515988.0, + "step": 14905 + }, + { + "epoch": 1.2291580770182238, + "grad_norm": 1.9380292892456055, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7648494243621826, + "num_tokens": 338552069.0, + "step": 14906 + }, + { + "epoch": 1.2292405376432753, + "grad_norm": 1.986833095550537, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.779434859752655, + "num_tokens": 338580294.0, + "step": 14907 + }, + { + "epoch": 1.229322998268327, + "grad_norm": 1.903866171836853, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7497495412826538, + "num_tokens": 338613404.0, + "step": 14908 + }, + { + "epoch": 1.2294054588933785, + "grad_norm": 2.1056370735168457, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7246066331863403, + "num_tokens": 338641672.0, + "step": 14909 + }, + { + "epoch": 1.22948791951843, + "grad_norm": 1.889579176902771, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.7972666025161743, + "num_tokens": 338675313.0, + "step": 14910 + }, + { + "epoch": 1.2295703801434814, + "grad_norm": 1.8238251209259033, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7680980563163757, + "num_tokens": 338707353.0, + "step": 14911 + }, + { + "epoch": 1.229652840768533, + "grad_norm": 1.6922961473464966, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7695857882499695, + "num_tokens": 338746379.0, + "step": 14912 + }, + { + "epoch": 1.2297353013935846, + "grad_norm": 2.839627742767334, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7363426685333252, + "num_tokens": 338768936.0, + "step": 14913 + }, + { + "epoch": 1.2298177620186361, + "grad_norm": 1.8397573232650757, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.7874037027359009, + "num_tokens": 338801602.0, + "step": 14914 + }, + { + "epoch": 1.2299002226436877, + "grad_norm": 2.176025629043579, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7531777620315552, + "num_tokens": 338826642.0, + "step": 14915 + }, + { + "epoch": 1.2299826832687393, + "grad_norm": 1.9659940004348755, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7811169624328613, + "num_tokens": 338859569.0, + "step": 14916 + }, + { + "epoch": 1.2300651438937906, + "grad_norm": 2.2642862796783447, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.725956916809082, + "num_tokens": 338885790.0, + "step": 14917 + }, + { + "epoch": 1.2301476045188422, + "grad_norm": 2.0688343048095703, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.7828603386878967, + "num_tokens": 338915104.0, + "step": 14918 + }, + { + "epoch": 1.2302300651438938, + "grad_norm": 1.9305306673049927, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7771623730659485, + "num_tokens": 338944875.0, + "step": 14919 + }, + { + "epoch": 1.2303125257689453, + "grad_norm": 1.8599666357040405, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.7928289175033569, + "num_tokens": 338976279.0, + "step": 14920 + }, + { + "epoch": 1.230394986393997, + "grad_norm": 1.879629373550415, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7689734697341919, + "num_tokens": 339009530.0, + "step": 14921 + }, + { + "epoch": 1.2304774470190485, + "grad_norm": 1.9970812797546387, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7528668642044067, + "num_tokens": 339042003.0, + "step": 14922 + }, + { + "epoch": 1.2305599076440998, + "grad_norm": 2.0866165161132812, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7629439234733582, + "num_tokens": 339071647.0, + "step": 14923 + }, + { + "epoch": 1.2306423682691514, + "grad_norm": 2.1037824153900146, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.742975115776062, + "num_tokens": 339100137.0, + "step": 14924 + }, + { + "epoch": 1.230724828894203, + "grad_norm": 1.9423983097076416, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7526734471321106, + "num_tokens": 339131060.0, + "step": 14925 + }, + { + "epoch": 1.2308072895192546, + "grad_norm": 1.9136543273925781, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7567431330680847, + "num_tokens": 339167817.0, + "step": 14926 + }, + { + "epoch": 1.2308897501443061, + "grad_norm": 2.2422444820404053, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.7857376337051392, + "num_tokens": 339194595.0, + "step": 14927 + }, + { + "epoch": 1.2309722107693577, + "grad_norm": 1.8962106704711914, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7378939390182495, + "num_tokens": 339229889.0, + "step": 14928 + }, + { + "epoch": 1.2310546713944093, + "grad_norm": 1.9231714010238647, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.7909305095672607, + "num_tokens": 339261444.0, + "step": 14929 + }, + { + "epoch": 1.2311371320194606, + "grad_norm": 1.9822406768798828, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7852203249931335, + "num_tokens": 339289653.0, + "step": 14930 + }, + { + "epoch": 1.2312195926445122, + "grad_norm": 2.040518283843994, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7376968860626221, + "num_tokens": 339319240.0, + "step": 14931 + }, + { + "epoch": 1.2313020532695638, + "grad_norm": 1.7840490341186523, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7603359818458557, + "num_tokens": 339354541.0, + "step": 14932 + }, + { + "epoch": 1.2313845138946153, + "grad_norm": 2.0430476665496826, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7543932795524597, + "num_tokens": 339382560.0, + "step": 14933 + }, + { + "epoch": 1.231466974519667, + "grad_norm": 1.9510630369186401, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7163064479827881, + "num_tokens": 339416490.0, + "step": 14934 + }, + { + "epoch": 1.2315494351447185, + "grad_norm": 2.4540822505950928, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7861003279685974, + "num_tokens": 339447844.0, + "step": 14935 + }, + { + "epoch": 1.23163189576977, + "grad_norm": 2.303755283355713, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7545099258422852, + "num_tokens": 339473865.0, + "step": 14936 + }, + { + "epoch": 1.2317143563948214, + "grad_norm": 1.8320657014846802, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.7644308805465698, + "num_tokens": 339508841.0, + "step": 14937 + }, + { + "epoch": 1.231796817019873, + "grad_norm": 1.8768277168273926, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.7931612730026245, + "num_tokens": 339541186.0, + "step": 14938 + }, + { + "epoch": 1.2318792776449246, + "grad_norm": 2.030529499053955, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7348535656929016, + "num_tokens": 339570347.0, + "step": 14939 + }, + { + "epoch": 1.2319617382699761, + "grad_norm": 2.221031904220581, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.725704550743103, + "num_tokens": 339597019.0, + "step": 14940 + }, + { + "epoch": 1.2320441988950277, + "grad_norm": 1.9205628633499146, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7684614658355713, + "num_tokens": 339626599.0, + "step": 14941 + }, + { + "epoch": 1.232126659520079, + "grad_norm": 2.0466225147247314, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7613844275474548, + "num_tokens": 339654010.0, + "step": 14942 + }, + { + "epoch": 1.2322091201451306, + "grad_norm": 1.829202651977539, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.7941151261329651, + "num_tokens": 339688285.0, + "step": 14943 + }, + { + "epoch": 1.2322915807701822, + "grad_norm": 2.042912721633911, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7655029296875, + "num_tokens": 339720596.0, + "step": 14944 + }, + { + "epoch": 1.2323740413952338, + "grad_norm": 2.175175189971924, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.752373456954956, + "num_tokens": 339750350.0, + "step": 14945 + }, + { + "epoch": 1.2324565020202853, + "grad_norm": 2.3070943355560303, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7579401135444641, + "num_tokens": 339773489.0, + "step": 14946 + }, + { + "epoch": 1.232538962645337, + "grad_norm": 2.0196399688720703, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.7833945751190186, + "num_tokens": 339806142.0, + "step": 14947 + }, + { + "epoch": 1.2326214232703885, + "grad_norm": 1.8748202323913574, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7796076536178589, + "num_tokens": 339840494.0, + "step": 14948 + }, + { + "epoch": 1.2327038838954398, + "grad_norm": 1.885349154472351, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7790073156356812, + "num_tokens": 339875447.0, + "step": 14949 + }, + { + "epoch": 1.2327863445204914, + "grad_norm": 1.8689484596252441, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.7583197355270386, + "num_tokens": 339912174.0, + "step": 14950 + }, + { + "epoch": 1.232868805145543, + "grad_norm": 2.081984758377075, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.7803714275360107, + "num_tokens": 339938105.0, + "step": 14951 + }, + { + "epoch": 1.2329512657705946, + "grad_norm": 1.95670485496521, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7592102289199829, + "num_tokens": 339972297.0, + "step": 14952 + }, + { + "epoch": 1.2330337263956461, + "grad_norm": 2.1077992916107178, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.726661205291748, + "num_tokens": 340001871.0, + "step": 14953 + }, + { + "epoch": 1.2331161870206977, + "grad_norm": 2.247354745864868, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7790507078170776, + "num_tokens": 340032393.0, + "step": 14954 + }, + { + "epoch": 1.2331986476457493, + "grad_norm": 2.0469586849212646, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7341626286506653, + "num_tokens": 340062565.0, + "step": 14955 + }, + { + "epoch": 1.2332811082708006, + "grad_norm": 2.184805154800415, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7574034929275513, + "num_tokens": 340088424.0, + "step": 14956 + }, + { + "epoch": 1.2333635688958522, + "grad_norm": 2.4073915481567383, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.741930365562439, + "num_tokens": 340110120.0, + "step": 14957 + }, + { + "epoch": 1.2334460295209038, + "grad_norm": 2.20451283454895, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7712419629096985, + "num_tokens": 340136042.0, + "step": 14958 + }, + { + "epoch": 1.2335284901459553, + "grad_norm": 2.143545150756836, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7563105225563049, + "num_tokens": 340162145.0, + "step": 14959 + }, + { + "epoch": 1.233610950771007, + "grad_norm": 1.90456223487854, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7708461880683899, + "num_tokens": 340194588.0, + "step": 14960 + }, + { + "epoch": 1.2336934113960583, + "grad_norm": 2.1597933769226074, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7732080221176147, + "num_tokens": 340220370.0, + "step": 14961 + }, + { + "epoch": 1.2337758720211098, + "grad_norm": 1.745214819908142, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.773775041103363, + "num_tokens": 340258458.0, + "step": 14962 + }, + { + "epoch": 1.2338583326461614, + "grad_norm": 2.0745718479156494, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.7917709946632385, + "num_tokens": 340286502.0, + "step": 14963 + }, + { + "epoch": 1.233940793271213, + "grad_norm": 1.9092535972595215, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7895797491073608, + "num_tokens": 340317596.0, + "step": 14964 + }, + { + "epoch": 1.2340232538962645, + "grad_norm": 1.9626423120498657, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.7946324348449707, + "num_tokens": 340347775.0, + "step": 14965 + }, + { + "epoch": 1.2341057145213161, + "grad_norm": 2.0517616271972656, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7694324851036072, + "num_tokens": 340380128.0, + "step": 14966 + }, + { + "epoch": 1.2341881751463677, + "grad_norm": 1.8638392686843872, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7531825304031372, + "num_tokens": 340413329.0, + "step": 14967 + }, + { + "epoch": 1.234270635771419, + "grad_norm": 1.8442169427871704, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7676288485527039, + "num_tokens": 340446927.0, + "step": 14968 + }, + { + "epoch": 1.2343530963964706, + "grad_norm": 1.7566171884536743, + "learning_rate": 1e-06, + "loss": 0.5718, + "mean_token_accuracy": 0.8079661130905151, + "num_tokens": 340488414.0, + "step": 14969 + }, + { + "epoch": 1.2344355570215222, + "grad_norm": 2.318498373031616, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7388558983802795, + "num_tokens": 340513120.0, + "step": 14970 + }, + { + "epoch": 1.2345180176465738, + "grad_norm": 2.2620246410369873, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7341057062149048, + "num_tokens": 340539938.0, + "step": 14971 + }, + { + "epoch": 1.2346004782716253, + "grad_norm": 1.9385124444961548, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7790464758872986, + "num_tokens": 340573273.0, + "step": 14972 + }, + { + "epoch": 1.234682938896677, + "grad_norm": 2.517010450363159, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7683203816413879, + "num_tokens": 340598243.0, + "step": 14973 + }, + { + "epoch": 1.2347653995217285, + "grad_norm": 1.95500910282135, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7190775871276855, + "num_tokens": 340631111.0, + "step": 14974 + }, + { + "epoch": 1.2348478601467798, + "grad_norm": 1.9277503490447998, + "learning_rate": 1e-06, + "loss": 0.6048, + "mean_token_accuracy": 0.793642520904541, + "num_tokens": 340663370.0, + "step": 14975 + }, + { + "epoch": 1.2349303207718314, + "grad_norm": 1.9520612955093384, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.775256872177124, + "num_tokens": 340692182.0, + "step": 14976 + }, + { + "epoch": 1.235012781396883, + "grad_norm": 1.6867917776107788, + "learning_rate": 1e-06, + "loss": 0.5948, + "mean_token_accuracy": 0.8013443946838379, + "num_tokens": 340728276.0, + "step": 14977 + }, + { + "epoch": 1.2350952420219345, + "grad_norm": 2.1853740215301514, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7624677419662476, + "num_tokens": 340756469.0, + "step": 14978 + }, + { + "epoch": 1.2351777026469861, + "grad_norm": 1.846244215965271, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7548391222953796, + "num_tokens": 340790546.0, + "step": 14979 + }, + { + "epoch": 1.2352601632720377, + "grad_norm": 2.020055055618286, + "learning_rate": 1e-06, + "loss": 0.7396, + "mean_token_accuracy": 0.7635310888290405, + "num_tokens": 340822165.0, + "step": 14980 + }, + { + "epoch": 1.2353426238970893, + "grad_norm": 1.91269052028656, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7682009935379028, + "num_tokens": 340852646.0, + "step": 14981 + }, + { + "epoch": 1.2354250845221406, + "grad_norm": 2.3704774379730225, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7508774399757385, + "num_tokens": 340874284.0, + "step": 14982 + }, + { + "epoch": 1.2355075451471922, + "grad_norm": 2.1931166648864746, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7498545050621033, + "num_tokens": 340901321.0, + "step": 14983 + }, + { + "epoch": 1.2355900057722438, + "grad_norm": 1.874313235282898, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7882564067840576, + "num_tokens": 340935389.0, + "step": 14984 + }, + { + "epoch": 1.2356724663972953, + "grad_norm": 1.9073940515518188, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.8034269213676453, + "num_tokens": 340966122.0, + "step": 14985 + }, + { + "epoch": 1.235754927022347, + "grad_norm": 1.9512217044830322, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.763530433177948, + "num_tokens": 340997505.0, + "step": 14986 + }, + { + "epoch": 1.2358373876473983, + "grad_norm": 1.7793720960617065, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.786165714263916, + "num_tokens": 341031461.0, + "step": 14987 + }, + { + "epoch": 1.2359198482724498, + "grad_norm": 1.6229344606399536, + "learning_rate": 1e-06, + "loss": 0.5916, + "mean_token_accuracy": 0.8051486015319824, + "num_tokens": 341070206.0, + "step": 14988 + }, + { + "epoch": 1.2360023088975014, + "grad_norm": 2.0878241062164307, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7433677315711975, + "num_tokens": 341099502.0, + "step": 14989 + }, + { + "epoch": 1.236084769522553, + "grad_norm": 1.9440072774887085, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7814092040061951, + "num_tokens": 341131721.0, + "step": 14990 + }, + { + "epoch": 1.2361672301476045, + "grad_norm": 1.7671427726745605, + "learning_rate": 1e-06, + "loss": 0.5464, + "mean_token_accuracy": 0.8187816143035889, + "num_tokens": 341163126.0, + "step": 14991 + }, + { + "epoch": 1.2362496907726561, + "grad_norm": 2.0335702896118164, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7485878467559814, + "num_tokens": 341190874.0, + "step": 14992 + }, + { + "epoch": 1.2363321513977077, + "grad_norm": 1.6637401580810547, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7915922999382019, + "num_tokens": 341229177.0, + "step": 14993 + }, + { + "epoch": 1.236414612022759, + "grad_norm": 2.090585470199585, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.7825037240982056, + "num_tokens": 341257910.0, + "step": 14994 + }, + { + "epoch": 1.2364970726478106, + "grad_norm": 2.049856424331665, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7697939872741699, + "num_tokens": 341292004.0, + "step": 14995 + }, + { + "epoch": 1.2365795332728622, + "grad_norm": 2.1624200344085693, + "learning_rate": 1e-06, + "loss": 0.7498, + "mean_token_accuracy": 0.7667875289916992, + "num_tokens": 341321020.0, + "step": 14996 + }, + { + "epoch": 1.2366619938979138, + "grad_norm": 2.046187400817871, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.78807532787323, + "num_tokens": 341346872.0, + "step": 14997 + }, + { + "epoch": 1.2367444545229653, + "grad_norm": 1.8041987419128418, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7840607762336731, + "num_tokens": 341380103.0, + "step": 14998 + }, + { + "epoch": 1.236826915148017, + "grad_norm": 2.1627259254455566, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7489757537841797, + "num_tokens": 341406512.0, + "step": 14999 + }, + { + "epoch": 1.2369093757730685, + "grad_norm": 2.110029697418213, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7592899799346924, + "num_tokens": 341433171.0, + "step": 15000 + }, + { + "epoch": 1.2369918363981198, + "grad_norm": 2.097153425216675, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7729743123054504, + "num_tokens": 341462590.0, + "step": 15001 + }, + { + "epoch": 1.2370742970231714, + "grad_norm": 1.941326379776001, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7801698446273804, + "num_tokens": 341493568.0, + "step": 15002 + }, + { + "epoch": 1.237156757648223, + "grad_norm": 1.8985012769699097, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7753185033798218, + "num_tokens": 341528399.0, + "step": 15003 + }, + { + "epoch": 1.2372392182732745, + "grad_norm": 2.045427083969116, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7251095771789551, + "num_tokens": 341557007.0, + "step": 15004 + }, + { + "epoch": 1.2373216788983261, + "grad_norm": 1.9265458583831787, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.755901038646698, + "num_tokens": 341590646.0, + "step": 15005 + }, + { + "epoch": 1.2374041395233775, + "grad_norm": 1.9596794843673706, + "learning_rate": 1e-06, + "loss": 0.5867, + "mean_token_accuracy": 0.815168023109436, + "num_tokens": 341620564.0, + "step": 15006 + }, + { + "epoch": 1.237486600148429, + "grad_norm": 1.8443193435668945, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.7901384830474854, + "num_tokens": 341653221.0, + "step": 15007 + }, + { + "epoch": 1.2375690607734806, + "grad_norm": 1.9527848958969116, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7657104134559631, + "num_tokens": 341682653.0, + "step": 15008 + }, + { + "epoch": 1.2376515213985322, + "grad_norm": 1.9678981304168701, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7833703756332397, + "num_tokens": 341716289.0, + "step": 15009 + }, + { + "epoch": 1.2377339820235838, + "grad_norm": 2.173675060272217, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7371071577072144, + "num_tokens": 341744514.0, + "step": 15010 + }, + { + "epoch": 1.2378164426486353, + "grad_norm": 1.7861289978027344, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7861087322235107, + "num_tokens": 341779959.0, + "step": 15011 + }, + { + "epoch": 1.237898903273687, + "grad_norm": 1.8326424360275269, + "learning_rate": 1e-06, + "loss": 0.5052, + "mean_token_accuracy": 0.8292567729949951, + "num_tokens": 341811249.0, + "step": 15012 + }, + { + "epoch": 1.2379813638987383, + "grad_norm": 1.9432212114334106, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7632901668548584, + "num_tokens": 341841650.0, + "step": 15013 + }, + { + "epoch": 1.2380638245237898, + "grad_norm": 2.1475658416748047, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7403877377510071, + "num_tokens": 341868184.0, + "step": 15014 + }, + { + "epoch": 1.2381462851488414, + "grad_norm": 2.1401796340942383, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.7967034578323364, + "num_tokens": 341891212.0, + "step": 15015 + }, + { + "epoch": 1.238228745773893, + "grad_norm": 1.8976179361343384, + "learning_rate": 1e-06, + "loss": 0.6084, + "mean_token_accuracy": 0.8001266717910767, + "num_tokens": 341920083.0, + "step": 15016 + }, + { + "epoch": 1.2383112063989445, + "grad_norm": 1.9390369653701782, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7451770901679993, + "num_tokens": 341952084.0, + "step": 15017 + }, + { + "epoch": 1.2383936670239961, + "grad_norm": 1.9939610958099365, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7329502701759338, + "num_tokens": 341983553.0, + "step": 15018 + }, + { + "epoch": 1.2384761276490477, + "grad_norm": 1.7927982807159424, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.7887767553329468, + "num_tokens": 342017578.0, + "step": 15019 + }, + { + "epoch": 1.238558588274099, + "grad_norm": 1.9886236190795898, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7399376630783081, + "num_tokens": 342049854.0, + "step": 15020 + }, + { + "epoch": 1.2386410488991506, + "grad_norm": 2.1274778842926025, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7714741230010986, + "num_tokens": 342076184.0, + "step": 15021 + }, + { + "epoch": 1.2387235095242022, + "grad_norm": 2.2484130859375, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7690281867980957, + "num_tokens": 342100200.0, + "step": 15022 + }, + { + "epoch": 1.2388059701492538, + "grad_norm": 1.8428566455841064, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.753349244594574, + "num_tokens": 342134098.0, + "step": 15023 + }, + { + "epoch": 1.2388884307743053, + "grad_norm": 1.9526180028915405, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7389328479766846, + "num_tokens": 342165528.0, + "step": 15024 + }, + { + "epoch": 1.238970891399357, + "grad_norm": 2.2918622493743896, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.7624236345291138, + "num_tokens": 342190379.0, + "step": 15025 + }, + { + "epoch": 1.2390533520244082, + "grad_norm": 2.1627299785614014, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7566623091697693, + "num_tokens": 342215201.0, + "step": 15026 + }, + { + "epoch": 1.2391358126494598, + "grad_norm": 1.8952434062957764, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7623359560966492, + "num_tokens": 342246589.0, + "step": 15027 + }, + { + "epoch": 1.2392182732745114, + "grad_norm": 2.033522844314575, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7805037498474121, + "num_tokens": 342277870.0, + "step": 15028 + }, + { + "epoch": 1.239300733899563, + "grad_norm": 1.9031425714492798, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7808265686035156, + "num_tokens": 342311896.0, + "step": 15029 + }, + { + "epoch": 1.2393831945246145, + "grad_norm": 2.0566909313201904, + "learning_rate": 1e-06, + "loss": 0.7704, + "mean_token_accuracy": 0.7539797425270081, + "num_tokens": 342343012.0, + "step": 15030 + }, + { + "epoch": 1.2394656551496661, + "grad_norm": 2.033369779586792, + "learning_rate": 1e-06, + "loss": 0.6251, + "mean_token_accuracy": 0.7898682355880737, + "num_tokens": 342371429.0, + "step": 15031 + }, + { + "epoch": 1.2395481157747175, + "grad_norm": 1.8579015731811523, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.786202073097229, + "num_tokens": 342402639.0, + "step": 15032 + }, + { + "epoch": 1.239630576399769, + "grad_norm": 2.703843593597412, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.7810713052749634, + "num_tokens": 342426981.0, + "step": 15033 + }, + { + "epoch": 1.2397130370248206, + "grad_norm": 2.1857383251190186, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7275399565696716, + "num_tokens": 342454932.0, + "step": 15034 + }, + { + "epoch": 1.2397954976498722, + "grad_norm": 1.9491908550262451, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7546771168708801, + "num_tokens": 342486626.0, + "step": 15035 + }, + { + "epoch": 1.2398779582749238, + "grad_norm": 2.1361448764801025, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7830458283424377, + "num_tokens": 342514082.0, + "step": 15036 + }, + { + "epoch": 1.2399604188999753, + "grad_norm": 2.034895181655884, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7692126035690308, + "num_tokens": 342544364.0, + "step": 15037 + }, + { + "epoch": 1.240042879525027, + "grad_norm": 1.7719430923461914, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7372218370437622, + "num_tokens": 342582865.0, + "step": 15038 + }, + { + "epoch": 1.2401253401500782, + "grad_norm": 1.8307468891143799, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.790097713470459, + "num_tokens": 342614947.0, + "step": 15039 + }, + { + "epoch": 1.2402078007751298, + "grad_norm": 1.9029611349105835, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7763420343399048, + "num_tokens": 342650250.0, + "step": 15040 + }, + { + "epoch": 1.2402902614001814, + "grad_norm": 2.01723575592041, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7545287013053894, + "num_tokens": 342678747.0, + "step": 15041 + }, + { + "epoch": 1.240372722025233, + "grad_norm": 1.7556242942810059, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7680037021636963, + "num_tokens": 342713606.0, + "step": 15042 + }, + { + "epoch": 1.2404551826502845, + "grad_norm": 2.202425479888916, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7519311904907227, + "num_tokens": 342740142.0, + "step": 15043 + }, + { + "epoch": 1.240537643275336, + "grad_norm": 1.9827768802642822, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.776663064956665, + "num_tokens": 342769980.0, + "step": 15044 + }, + { + "epoch": 1.2406201039003877, + "grad_norm": 2.01977801322937, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7651156187057495, + "num_tokens": 342798606.0, + "step": 15045 + }, + { + "epoch": 1.240702564525439, + "grad_norm": 2.100342273712158, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7641671299934387, + "num_tokens": 342825518.0, + "step": 15046 + }, + { + "epoch": 1.2407850251504906, + "grad_norm": 2.184197187423706, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7298581004142761, + "num_tokens": 342853642.0, + "step": 15047 + }, + { + "epoch": 1.2408674857755422, + "grad_norm": 1.9495426416397095, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.777915358543396, + "num_tokens": 342884567.0, + "step": 15048 + }, + { + "epoch": 1.2409499464005938, + "grad_norm": 1.9498854875564575, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.7835776209831238, + "num_tokens": 342915415.0, + "step": 15049 + }, + { + "epoch": 1.2410324070256453, + "grad_norm": 1.7138214111328125, + "learning_rate": 1e-06, + "loss": 0.5785, + "mean_token_accuracy": 0.8032056093215942, + "num_tokens": 342950253.0, + "step": 15050 + }, + { + "epoch": 1.2411148676506967, + "grad_norm": 1.8990445137023926, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7620595693588257, + "num_tokens": 342983720.0, + "step": 15051 + }, + { + "epoch": 1.2411973282757482, + "grad_norm": 2.020982265472412, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7545607686042786, + "num_tokens": 343014043.0, + "step": 15052 + }, + { + "epoch": 1.2412797889007998, + "grad_norm": 1.8929970264434814, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7748509645462036, + "num_tokens": 343046851.0, + "step": 15053 + }, + { + "epoch": 1.2413622495258514, + "grad_norm": 1.828744649887085, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7669509649276733, + "num_tokens": 343078878.0, + "step": 15054 + }, + { + "epoch": 1.241444710150903, + "grad_norm": 2.10457181930542, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7300006747245789, + "num_tokens": 343106341.0, + "step": 15055 + }, + { + "epoch": 1.2415271707759545, + "grad_norm": 2.290527582168579, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7485062479972839, + "num_tokens": 343129398.0, + "step": 15056 + }, + { + "epoch": 1.241609631401006, + "grad_norm": 1.8368574380874634, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7251411080360413, + "num_tokens": 343164429.0, + "step": 15057 + }, + { + "epoch": 1.2416920920260575, + "grad_norm": 2.116288423538208, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7444453239440918, + "num_tokens": 343192472.0, + "step": 15058 + }, + { + "epoch": 1.241774552651109, + "grad_norm": 1.949713945388794, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.7916311025619507, + "num_tokens": 343223970.0, + "step": 15059 + }, + { + "epoch": 1.2418570132761606, + "grad_norm": 2.008437395095825, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7560667991638184, + "num_tokens": 343250982.0, + "step": 15060 + }, + { + "epoch": 1.2419394739012122, + "grad_norm": 1.867611289024353, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7641204595565796, + "num_tokens": 343284619.0, + "step": 15061 + }, + { + "epoch": 1.2420219345262637, + "grad_norm": 1.8834187984466553, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7626718282699585, + "num_tokens": 343319960.0, + "step": 15062 + }, + { + "epoch": 1.2421043951513153, + "grad_norm": 1.8761985301971436, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7459706664085388, + "num_tokens": 343354214.0, + "step": 15063 + }, + { + "epoch": 1.242186855776367, + "grad_norm": 1.91251802444458, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7758978009223938, + "num_tokens": 343386368.0, + "step": 15064 + }, + { + "epoch": 1.2422693164014182, + "grad_norm": 2.064520835876465, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7864500284194946, + "num_tokens": 343415682.0, + "step": 15065 + }, + { + "epoch": 1.2423517770264698, + "grad_norm": 1.9312924146652222, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7676836848258972, + "num_tokens": 343445706.0, + "step": 15066 + }, + { + "epoch": 1.2424342376515214, + "grad_norm": 2.0549263954162598, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7560700178146362, + "num_tokens": 343472775.0, + "step": 15067 + }, + { + "epoch": 1.242516698276573, + "grad_norm": 1.8686072826385498, + "learning_rate": 1e-06, + "loss": 0.6221, + "mean_token_accuracy": 0.7832379341125488, + "num_tokens": 343502978.0, + "step": 15068 + }, + { + "epoch": 1.2425991589016245, + "grad_norm": 1.8755550384521484, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7832549810409546, + "num_tokens": 343534093.0, + "step": 15069 + }, + { + "epoch": 1.242681619526676, + "grad_norm": 1.8166794776916504, + "learning_rate": 1e-06, + "loss": 0.6036, + "mean_token_accuracy": 0.8039106130599976, + "num_tokens": 343567813.0, + "step": 15070 + }, + { + "epoch": 1.2427640801517275, + "grad_norm": 1.9687104225158691, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7566384077072144, + "num_tokens": 343599204.0, + "step": 15071 + }, + { + "epoch": 1.242846540776779, + "grad_norm": 2.2452573776245117, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7753587961196899, + "num_tokens": 343624094.0, + "step": 15072 + }, + { + "epoch": 1.2429290014018306, + "grad_norm": 1.9160562753677368, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.7966532707214355, + "num_tokens": 343654013.0, + "step": 15073 + }, + { + "epoch": 1.2430114620268822, + "grad_norm": 1.8738147020339966, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7724684476852417, + "num_tokens": 343682419.0, + "step": 15074 + }, + { + "epoch": 1.2430939226519337, + "grad_norm": 2.0026659965515137, + "learning_rate": 1e-06, + "loss": 0.5975, + "mean_token_accuracy": 0.8000617623329163, + "num_tokens": 343710190.0, + "step": 15075 + }, + { + "epoch": 1.2431763832769853, + "grad_norm": 1.640066385269165, + "learning_rate": 1e-06, + "loss": 0.5926, + "mean_token_accuracy": 0.8108201026916504, + "num_tokens": 343745330.0, + "step": 15076 + }, + { + "epoch": 1.2432588439020367, + "grad_norm": 2.1204276084899902, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.7795822024345398, + "num_tokens": 343772897.0, + "step": 15077 + }, + { + "epoch": 1.2433413045270882, + "grad_norm": 1.8304778337478638, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7441149950027466, + "num_tokens": 343810388.0, + "step": 15078 + }, + { + "epoch": 1.2434237651521398, + "grad_norm": 2.0231337547302246, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7868776321411133, + "num_tokens": 343838294.0, + "step": 15079 + }, + { + "epoch": 1.2435062257771914, + "grad_norm": 1.8213138580322266, + "learning_rate": 1e-06, + "loss": 0.5972, + "mean_token_accuracy": 0.8052651882171631, + "num_tokens": 343870387.0, + "step": 15080 + }, + { + "epoch": 1.243588686402243, + "grad_norm": 2.2146198749542236, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7637941837310791, + "num_tokens": 343894662.0, + "step": 15081 + }, + { + "epoch": 1.2436711470272945, + "grad_norm": 1.8679440021514893, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7604801654815674, + "num_tokens": 343930925.0, + "step": 15082 + }, + { + "epoch": 1.243753607652346, + "grad_norm": 2.1918692588806152, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7163715362548828, + "num_tokens": 343958815.0, + "step": 15083 + }, + { + "epoch": 1.2438360682773975, + "grad_norm": 1.8948538303375244, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7577056884765625, + "num_tokens": 343993088.0, + "step": 15084 + }, + { + "epoch": 1.243918528902449, + "grad_norm": 2.0423784255981445, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.740162193775177, + "num_tokens": 344022743.0, + "step": 15085 + }, + { + "epoch": 1.2440009895275006, + "grad_norm": 1.948245882987976, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7588834762573242, + "num_tokens": 344053382.0, + "step": 15086 + }, + { + "epoch": 1.2440834501525522, + "grad_norm": 1.9648934602737427, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7602512836456299, + "num_tokens": 344082265.0, + "step": 15087 + }, + { + "epoch": 1.2441659107776037, + "grad_norm": 1.997244954109192, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7762939929962158, + "num_tokens": 344111040.0, + "step": 15088 + }, + { + "epoch": 1.2442483714026553, + "grad_norm": 2.0247437953948975, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7740777134895325, + "num_tokens": 344136773.0, + "step": 15089 + }, + { + "epoch": 1.244330832027707, + "grad_norm": 2.125575065612793, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7472759485244751, + "num_tokens": 344163845.0, + "step": 15090 + }, + { + "epoch": 1.2444132926527582, + "grad_norm": 1.899265170097351, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7128920555114746, + "num_tokens": 344200387.0, + "step": 15091 + }, + { + "epoch": 1.2444957532778098, + "grad_norm": 1.8979277610778809, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.7875157594680786, + "num_tokens": 344232111.0, + "step": 15092 + }, + { + "epoch": 1.2445782139028614, + "grad_norm": 1.734776258468628, + "learning_rate": 1e-06, + "loss": 0.4888, + "mean_token_accuracy": 0.8332297205924988, + "num_tokens": 344263825.0, + "step": 15093 + }, + { + "epoch": 1.244660674527913, + "grad_norm": 1.917486310005188, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.771828293800354, + "num_tokens": 344293224.0, + "step": 15094 + }, + { + "epoch": 1.2447431351529645, + "grad_norm": 2.0456161499023438, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.7481398582458496, + "num_tokens": 344320115.0, + "step": 15095 + }, + { + "epoch": 1.2448255957780159, + "grad_norm": 1.9809378385543823, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7786660194396973, + "num_tokens": 344348527.0, + "step": 15096 + }, + { + "epoch": 1.2449080564030675, + "grad_norm": 2.2601253986358643, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7559071183204651, + "num_tokens": 344373886.0, + "step": 15097 + }, + { + "epoch": 1.244990517028119, + "grad_norm": 2.15425443649292, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7794657945632935, + "num_tokens": 344402019.0, + "step": 15098 + }, + { + "epoch": 1.2450729776531706, + "grad_norm": 1.8900667428970337, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.7877243757247925, + "num_tokens": 344432714.0, + "step": 15099 + }, + { + "epoch": 1.2451554382782222, + "grad_norm": 1.8915175199508667, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7887834310531616, + "num_tokens": 344461221.0, + "step": 15100 + }, + { + "epoch": 1.2452378989032737, + "grad_norm": 2.2127788066864014, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7205063700675964, + "num_tokens": 344488115.0, + "step": 15101 + }, + { + "epoch": 1.2453203595283253, + "grad_norm": 2.063406467437744, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.783873975276947, + "num_tokens": 344518597.0, + "step": 15102 + }, + { + "epoch": 1.2454028201533767, + "grad_norm": 1.9304782152175903, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.75639408826828, + "num_tokens": 344548568.0, + "step": 15103 + }, + { + "epoch": 1.2454852807784282, + "grad_norm": 2.077803134918213, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7740010023117065, + "num_tokens": 344576031.0, + "step": 15104 + }, + { + "epoch": 1.2455677414034798, + "grad_norm": 2.1148128509521484, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7354279160499573, + "num_tokens": 344603166.0, + "step": 15105 + }, + { + "epoch": 1.2456502020285314, + "grad_norm": 1.9444326162338257, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7749713063240051, + "num_tokens": 344632679.0, + "step": 15106 + }, + { + "epoch": 1.245732662653583, + "grad_norm": 2.4536867141723633, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7613808512687683, + "num_tokens": 344656427.0, + "step": 15107 + }, + { + "epoch": 1.2458151232786345, + "grad_norm": 1.916920781135559, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7461006045341492, + "num_tokens": 344691648.0, + "step": 15108 + }, + { + "epoch": 1.245897583903686, + "grad_norm": 1.8420867919921875, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7188515663146973, + "num_tokens": 344728403.0, + "step": 15109 + }, + { + "epoch": 1.2459800445287375, + "grad_norm": 2.3670175075531006, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.749401330947876, + "num_tokens": 344750397.0, + "step": 15110 + }, + { + "epoch": 1.246062505153789, + "grad_norm": 2.1036434173583984, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.7575603127479553, + "num_tokens": 344779695.0, + "step": 15111 + }, + { + "epoch": 1.2461449657788406, + "grad_norm": 1.9691741466522217, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7604150176048279, + "num_tokens": 344810317.0, + "step": 15112 + }, + { + "epoch": 1.2462274264038922, + "grad_norm": 1.985762119293213, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7676482200622559, + "num_tokens": 344841734.0, + "step": 15113 + }, + { + "epoch": 1.2463098870289437, + "grad_norm": 2.148165702819824, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.6982612013816833, + "num_tokens": 344871383.0, + "step": 15114 + }, + { + "epoch": 1.2463923476539953, + "grad_norm": 1.9049971103668213, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7899982929229736, + "num_tokens": 344904217.0, + "step": 15115 + }, + { + "epoch": 1.2464748082790467, + "grad_norm": 2.1547274589538574, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7371782064437866, + "num_tokens": 344932470.0, + "step": 15116 + }, + { + "epoch": 1.2465572689040982, + "grad_norm": 1.8432925939559937, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7591781616210938, + "num_tokens": 344967143.0, + "step": 15117 + }, + { + "epoch": 1.2466397295291498, + "grad_norm": 1.9456318616867065, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7807546854019165, + "num_tokens": 344999260.0, + "step": 15118 + }, + { + "epoch": 1.2467221901542014, + "grad_norm": 1.9818384647369385, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7376251816749573, + "num_tokens": 345029397.0, + "step": 15119 + }, + { + "epoch": 1.246804650779253, + "grad_norm": 2.023306131362915, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7688137292861938, + "num_tokens": 345057693.0, + "step": 15120 + }, + { + "epoch": 1.2468871114043045, + "grad_norm": 2.0299782752990723, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7349570393562317, + "num_tokens": 345091369.0, + "step": 15121 + }, + { + "epoch": 1.2469695720293559, + "grad_norm": 2.032287120819092, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7422850728034973, + "num_tokens": 345120432.0, + "step": 15122 + }, + { + "epoch": 1.2470520326544074, + "grad_norm": 2.232830286026001, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7665737867355347, + "num_tokens": 345144471.0, + "step": 15123 + }, + { + "epoch": 1.247134493279459, + "grad_norm": 1.8963923454284668, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7628652453422546, + "num_tokens": 345179538.0, + "step": 15124 + }, + { + "epoch": 1.2472169539045106, + "grad_norm": 1.9404157400131226, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.7948426008224487, + "num_tokens": 345208222.0, + "step": 15125 + }, + { + "epoch": 1.2472994145295622, + "grad_norm": 1.8520926237106323, + "learning_rate": 1e-06, + "loss": 0.7774, + "mean_token_accuracy": 0.7551932334899902, + "num_tokens": 345242396.0, + "step": 15126 + }, + { + "epoch": 1.2473818751546137, + "grad_norm": 2.121872901916504, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7300047874450684, + "num_tokens": 345272463.0, + "step": 15127 + }, + { + "epoch": 1.2474643357796653, + "grad_norm": 1.8008618354797363, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7741571664810181, + "num_tokens": 345308697.0, + "step": 15128 + }, + { + "epoch": 1.2475467964047167, + "grad_norm": 2.1088826656341553, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7882907390594482, + "num_tokens": 345336303.0, + "step": 15129 + }, + { + "epoch": 1.2476292570297682, + "grad_norm": 1.9553877115249634, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.7873324155807495, + "num_tokens": 345370923.0, + "step": 15130 + }, + { + "epoch": 1.2477117176548198, + "grad_norm": 2.1110103130340576, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7673367261886597, + "num_tokens": 345400274.0, + "step": 15131 + }, + { + "epoch": 1.2477941782798714, + "grad_norm": 1.7883552312850952, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7707424759864807, + "num_tokens": 345436122.0, + "step": 15132 + }, + { + "epoch": 1.247876638904923, + "grad_norm": 2.160290241241455, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.741899847984314, + "num_tokens": 345464764.0, + "step": 15133 + }, + { + "epoch": 1.2479590995299745, + "grad_norm": 1.7486610412597656, + "learning_rate": 1e-06, + "loss": 0.564, + "mean_token_accuracy": 0.811285138130188, + "num_tokens": 345497712.0, + "step": 15134 + }, + { + "epoch": 1.248041560155026, + "grad_norm": 1.8919999599456787, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7392380237579346, + "num_tokens": 345531632.0, + "step": 15135 + }, + { + "epoch": 1.2481240207800774, + "grad_norm": 2.241853952407837, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.7863520383834839, + "num_tokens": 345554056.0, + "step": 15136 + }, + { + "epoch": 1.248206481405129, + "grad_norm": 1.9872610569000244, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7798465490341187, + "num_tokens": 345591109.0, + "step": 15137 + }, + { + "epoch": 1.2482889420301806, + "grad_norm": 1.9827489852905273, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7532920241355896, + "num_tokens": 345619551.0, + "step": 15138 + }, + { + "epoch": 1.2483714026552322, + "grad_norm": 2.097208023071289, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7799054384231567, + "num_tokens": 345646992.0, + "step": 15139 + }, + { + "epoch": 1.2484538632802837, + "grad_norm": 2.1450400352478027, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7409873604774475, + "num_tokens": 345677990.0, + "step": 15140 + }, + { + "epoch": 1.248536323905335, + "grad_norm": 1.9761016368865967, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7598851323127747, + "num_tokens": 345707223.0, + "step": 15141 + }, + { + "epoch": 1.2486187845303867, + "grad_norm": 2.287729501724243, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7695656418800354, + "num_tokens": 345732487.0, + "step": 15142 + }, + { + "epoch": 1.2487012451554382, + "grad_norm": 1.9351844787597656, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7765328288078308, + "num_tokens": 345763278.0, + "step": 15143 + }, + { + "epoch": 1.2487837057804898, + "grad_norm": 1.9390977621078491, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7721562385559082, + "num_tokens": 345793458.0, + "step": 15144 + }, + { + "epoch": 1.2488661664055414, + "grad_norm": 1.991890788078308, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7695691585540771, + "num_tokens": 345823894.0, + "step": 15145 + }, + { + "epoch": 1.248948627030593, + "grad_norm": 2.149697780609131, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7462767362594604, + "num_tokens": 345851375.0, + "step": 15146 + }, + { + "epoch": 1.2490310876556445, + "grad_norm": 1.876047968864441, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.770747184753418, + "num_tokens": 345884081.0, + "step": 15147 + }, + { + "epoch": 1.2491135482806959, + "grad_norm": 2.099369764328003, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7432503700256348, + "num_tokens": 345912722.0, + "step": 15148 + }, + { + "epoch": 1.2491960089057474, + "grad_norm": 2.0268826484680176, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7404190897941589, + "num_tokens": 345942630.0, + "step": 15149 + }, + { + "epoch": 1.249278469530799, + "grad_norm": 2.1189663410186768, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.736907958984375, + "num_tokens": 345970616.0, + "step": 15150 + }, + { + "epoch": 1.2493609301558506, + "grad_norm": 2.1366183757781982, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7478376626968384, + "num_tokens": 345997838.0, + "step": 15151 + }, + { + "epoch": 1.2494433907809022, + "grad_norm": 2.0659241676330566, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7365658283233643, + "num_tokens": 346025355.0, + "step": 15152 + }, + { + "epoch": 1.2495258514059537, + "grad_norm": 1.9849767684936523, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7758309245109558, + "num_tokens": 346056554.0, + "step": 15153 + }, + { + "epoch": 1.2496083120310053, + "grad_norm": 1.9025945663452148, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7626349329948425, + "num_tokens": 346086967.0, + "step": 15154 + }, + { + "epoch": 1.2496907726560567, + "grad_norm": 1.8541491031646729, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7756491899490356, + "num_tokens": 346116328.0, + "step": 15155 + }, + { + "epoch": 1.2497732332811082, + "grad_norm": 1.804040551185608, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.7974870204925537, + "num_tokens": 346149385.0, + "step": 15156 + }, + { + "epoch": 1.2498556939061598, + "grad_norm": 1.9664806127548218, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7546446323394775, + "num_tokens": 346182952.0, + "step": 15157 + }, + { + "epoch": 1.2499381545312114, + "grad_norm": 1.943013310432434, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7495067715644836, + "num_tokens": 346215048.0, + "step": 15158 + }, + { + "epoch": 1.250020615156263, + "grad_norm": 1.8080908060073853, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7773398160934448, + "num_tokens": 346251885.0, + "step": 15159 + }, + { + "epoch": 1.2501030757813143, + "grad_norm": 2.100658416748047, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6982016563415527, + "num_tokens": 346283313.0, + "step": 15160 + }, + { + "epoch": 1.250185536406366, + "grad_norm": 2.2105579376220703, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7643968462944031, + "num_tokens": 346308109.0, + "step": 15161 + }, + { + "epoch": 1.2502679970314174, + "grad_norm": 1.9137364625930786, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7719795107841492, + "num_tokens": 346340975.0, + "step": 15162 + }, + { + "epoch": 1.250350457656469, + "grad_norm": 2.0002918243408203, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.7926272749900818, + "num_tokens": 346370370.0, + "step": 15163 + }, + { + "epoch": 1.2504329182815206, + "grad_norm": 2.1596083641052246, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7255723476409912, + "num_tokens": 346399607.0, + "step": 15164 + }, + { + "epoch": 1.2505153789065722, + "grad_norm": 2.1021316051483154, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7358596324920654, + "num_tokens": 346427269.0, + "step": 15165 + }, + { + "epoch": 1.2505978395316237, + "grad_norm": 1.8190557956695557, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7601075768470764, + "num_tokens": 346460941.0, + "step": 15166 + }, + { + "epoch": 1.250680300156675, + "grad_norm": 1.92720627784729, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7843964695930481, + "num_tokens": 346491228.0, + "step": 15167 + }, + { + "epoch": 1.2507627607817267, + "grad_norm": 2.1175742149353027, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.7916361093521118, + "num_tokens": 346515868.0, + "step": 15168 + }, + { + "epoch": 1.2508452214067782, + "grad_norm": 1.8480442762374878, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.7868221998214722, + "num_tokens": 346546892.0, + "step": 15169 + }, + { + "epoch": 1.2509276820318298, + "grad_norm": 1.8342065811157227, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7748293876647949, + "num_tokens": 346581900.0, + "step": 15170 + }, + { + "epoch": 1.2510101426568814, + "grad_norm": 1.825087070465088, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7810338735580444, + "num_tokens": 346617192.0, + "step": 15171 + }, + { + "epoch": 1.251092603281933, + "grad_norm": 1.9754499197006226, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.73525071144104, + "num_tokens": 346649444.0, + "step": 15172 + }, + { + "epoch": 1.2511750639069845, + "grad_norm": 2.2417850494384766, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7333145141601562, + "num_tokens": 346675705.0, + "step": 15173 + }, + { + "epoch": 1.2512575245320359, + "grad_norm": 2.0271265506744385, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7620609998703003, + "num_tokens": 346704444.0, + "step": 15174 + }, + { + "epoch": 1.2513399851570874, + "grad_norm": 2.0809881687164307, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7397300601005554, + "num_tokens": 346736069.0, + "step": 15175 + }, + { + "epoch": 1.251422445782139, + "grad_norm": 1.9062877893447876, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.7905995845794678, + "num_tokens": 346768102.0, + "step": 15176 + }, + { + "epoch": 1.2515049064071906, + "grad_norm": 1.9104108810424805, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.770826518535614, + "num_tokens": 346801097.0, + "step": 15177 + }, + { + "epoch": 1.2515873670322422, + "grad_norm": 1.9346635341644287, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7213891744613647, + "num_tokens": 346833021.0, + "step": 15178 + }, + { + "epoch": 1.2516698276572935, + "grad_norm": 1.9939686059951782, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7533422708511353, + "num_tokens": 346863442.0, + "step": 15179 + }, + { + "epoch": 1.2517522882823453, + "grad_norm": 1.9189112186431885, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.775794506072998, + "num_tokens": 346896026.0, + "step": 15180 + }, + { + "epoch": 1.2518347489073967, + "grad_norm": 1.8793166875839233, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7192996144294739, + "num_tokens": 346929610.0, + "step": 15181 + }, + { + "epoch": 1.2519172095324482, + "grad_norm": 1.861963152885437, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.771249532699585, + "num_tokens": 346967057.0, + "step": 15182 + }, + { + "epoch": 1.2519996701574998, + "grad_norm": 1.8994303941726685, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7429423928260803, + "num_tokens": 346999464.0, + "step": 15183 + }, + { + "epoch": 1.2520821307825514, + "grad_norm": 2.034489870071411, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7846478223800659, + "num_tokens": 347027746.0, + "step": 15184 + }, + { + "epoch": 1.252164591407603, + "grad_norm": 2.138277769088745, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7428984642028809, + "num_tokens": 347054356.0, + "step": 15185 + }, + { + "epoch": 1.2522470520326543, + "grad_norm": 2.01367449760437, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.7789429426193237, + "num_tokens": 347084068.0, + "step": 15186 + }, + { + "epoch": 1.2523295126577059, + "grad_norm": 2.1836390495300293, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7702090740203857, + "num_tokens": 347108985.0, + "step": 15187 + }, + { + "epoch": 1.2524119732827574, + "grad_norm": 1.921819806098938, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7584972977638245, + "num_tokens": 347140253.0, + "step": 15188 + }, + { + "epoch": 1.252494433907809, + "grad_norm": 2.0591721534729004, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7811814546585083, + "num_tokens": 347168423.0, + "step": 15189 + }, + { + "epoch": 1.2525768945328606, + "grad_norm": 1.7862169742584229, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.7843648195266724, + "num_tokens": 347201375.0, + "step": 15190 + }, + { + "epoch": 1.2526593551579122, + "grad_norm": 2.0048093795776367, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7524635791778564, + "num_tokens": 347233743.0, + "step": 15191 + }, + { + "epoch": 1.2527418157829637, + "grad_norm": 2.0813231468200684, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7326721549034119, + "num_tokens": 347263123.0, + "step": 15192 + }, + { + "epoch": 1.252824276408015, + "grad_norm": 1.9940799474716187, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7471091151237488, + "num_tokens": 347295449.0, + "step": 15193 + }, + { + "epoch": 1.2529067370330667, + "grad_norm": 1.631248116493225, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7814216613769531, + "num_tokens": 347336257.0, + "step": 15194 + }, + { + "epoch": 1.2529891976581182, + "grad_norm": 1.8157906532287598, + "learning_rate": 1e-06, + "loss": 0.6062, + "mean_token_accuracy": 0.8073856830596924, + "num_tokens": 347371531.0, + "step": 15195 + }, + { + "epoch": 1.2530716582831698, + "grad_norm": 1.9468235969543457, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.775959849357605, + "num_tokens": 347403290.0, + "step": 15196 + }, + { + "epoch": 1.2531541189082214, + "grad_norm": 1.972676157951355, + "learning_rate": 1e-06, + "loss": 0.5854, + "mean_token_accuracy": 0.8013125061988831, + "num_tokens": 347430122.0, + "step": 15197 + }, + { + "epoch": 1.253236579533273, + "grad_norm": 1.794832706451416, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.8001308441162109, + "num_tokens": 347463730.0, + "step": 15198 + }, + { + "epoch": 1.2533190401583245, + "grad_norm": 2.180954694747925, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7762573957443237, + "num_tokens": 347490101.0, + "step": 15199 + }, + { + "epoch": 1.2534015007833759, + "grad_norm": 2.078042507171631, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7750871777534485, + "num_tokens": 347516069.0, + "step": 15200 + }, + { + "epoch": 1.2534839614084274, + "grad_norm": 1.747381567955017, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7569325566291809, + "num_tokens": 347554047.0, + "step": 15201 + }, + { + "epoch": 1.253566422033479, + "grad_norm": 2.2347774505615234, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7506105899810791, + "num_tokens": 347581942.0, + "step": 15202 + }, + { + "epoch": 1.2536488826585306, + "grad_norm": 2.058549642562866, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.748856782913208, + "num_tokens": 347610443.0, + "step": 15203 + }, + { + "epoch": 1.2537313432835822, + "grad_norm": 2.0127975940704346, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.7973047494888306, + "num_tokens": 347636262.0, + "step": 15204 + }, + { + "epoch": 1.2538138039086335, + "grad_norm": 2.1438817977905273, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7709516286849976, + "num_tokens": 347661608.0, + "step": 15205 + }, + { + "epoch": 1.2538962645336853, + "grad_norm": 1.7265467643737793, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7395570278167725, + "num_tokens": 347704831.0, + "step": 15206 + }, + { + "epoch": 1.2539787251587367, + "grad_norm": 2.099151134490967, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7775570154190063, + "num_tokens": 347731352.0, + "step": 15207 + }, + { + "epoch": 1.2540611857837882, + "grad_norm": 2.145251989364624, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7461590766906738, + "num_tokens": 347758515.0, + "step": 15208 + }, + { + "epoch": 1.2541436464088398, + "grad_norm": 2.015079975128174, + "learning_rate": 1e-06, + "loss": 0.5658, + "mean_token_accuracy": 0.809383749961853, + "num_tokens": 347785738.0, + "step": 15209 + }, + { + "epoch": 1.2542261070338914, + "grad_norm": 1.9356811046600342, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7496506571769714, + "num_tokens": 347818857.0, + "step": 15210 + }, + { + "epoch": 1.254308567658943, + "grad_norm": 1.9909898042678833, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7265099883079529, + "num_tokens": 347850084.0, + "step": 15211 + }, + { + "epoch": 1.2543910282839943, + "grad_norm": 2.0270822048187256, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7579175233840942, + "num_tokens": 347879311.0, + "step": 15212 + }, + { + "epoch": 1.2544734889090459, + "grad_norm": 2.135011672973633, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7669841051101685, + "num_tokens": 347903436.0, + "step": 15213 + }, + { + "epoch": 1.2545559495340974, + "grad_norm": 2.2089405059814453, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7703264951705933, + "num_tokens": 347929766.0, + "step": 15214 + }, + { + "epoch": 1.254638410159149, + "grad_norm": 2.1608948707580566, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7385445237159729, + "num_tokens": 347957172.0, + "step": 15215 + }, + { + "epoch": 1.2547208707842006, + "grad_norm": 1.767842173576355, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7812371253967285, + "num_tokens": 347993235.0, + "step": 15216 + }, + { + "epoch": 1.2548033314092522, + "grad_norm": 1.764784812927246, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7443637251853943, + "num_tokens": 348031682.0, + "step": 15217 + }, + { + "epoch": 1.2548857920343037, + "grad_norm": 1.884447693824768, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.7946872711181641, + "num_tokens": 348062299.0, + "step": 15218 + }, + { + "epoch": 1.254968252659355, + "grad_norm": 2.1186153888702393, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.781444251537323, + "num_tokens": 348088144.0, + "step": 15219 + }, + { + "epoch": 1.2550507132844066, + "grad_norm": 2.2881247997283936, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.7922136783599854, + "num_tokens": 348107942.0, + "step": 15220 + }, + { + "epoch": 1.2551331739094582, + "grad_norm": 2.0375492572784424, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.7844452857971191, + "num_tokens": 348135863.0, + "step": 15221 + }, + { + "epoch": 1.2552156345345098, + "grad_norm": 2.0074949264526367, + "learning_rate": 1e-06, + "loss": 0.5658, + "mean_token_accuracy": 0.8065854907035828, + "num_tokens": 348167195.0, + "step": 15222 + }, + { + "epoch": 1.2552980951595614, + "grad_norm": 2.165436029434204, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.7514320611953735, + "num_tokens": 348195114.0, + "step": 15223 + }, + { + "epoch": 1.2553805557846127, + "grad_norm": 1.8505228757858276, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.7865426540374756, + "num_tokens": 348226457.0, + "step": 15224 + }, + { + "epoch": 1.2554630164096645, + "grad_norm": 2.013328790664673, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.756772518157959, + "num_tokens": 348253561.0, + "step": 15225 + }, + { + "epoch": 1.2555454770347159, + "grad_norm": 2.2188069820404053, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.7644674777984619, + "num_tokens": 348281212.0, + "step": 15226 + }, + { + "epoch": 1.2556279376597674, + "grad_norm": 1.8670493364334106, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7807587385177612, + "num_tokens": 348315908.0, + "step": 15227 + }, + { + "epoch": 1.255710398284819, + "grad_norm": 1.9752486944198608, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.733547568321228, + "num_tokens": 348345053.0, + "step": 15228 + }, + { + "epoch": 1.2557928589098706, + "grad_norm": 1.963649034500122, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7710376381874084, + "num_tokens": 348377690.0, + "step": 15229 + }, + { + "epoch": 1.2558753195349222, + "grad_norm": 2.0789084434509277, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.7964382171630859, + "num_tokens": 348404066.0, + "step": 15230 + }, + { + "epoch": 1.2559577801599735, + "grad_norm": 2.088838815689087, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7647700309753418, + "num_tokens": 348432882.0, + "step": 15231 + }, + { + "epoch": 1.256040240785025, + "grad_norm": 1.9380698204040527, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.7541634440422058, + "num_tokens": 348465582.0, + "step": 15232 + }, + { + "epoch": 1.2561227014100766, + "grad_norm": 1.9295654296875, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7487187385559082, + "num_tokens": 348499914.0, + "step": 15233 + }, + { + "epoch": 1.2562051620351282, + "grad_norm": 2.0022506713867188, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7342826724052429, + "num_tokens": 348532420.0, + "step": 15234 + }, + { + "epoch": 1.2562876226601798, + "grad_norm": 1.7774983644485474, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7759739756584167, + "num_tokens": 348569105.0, + "step": 15235 + }, + { + "epoch": 1.2563700832852314, + "grad_norm": 1.9150371551513672, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.7977430820465088, + "num_tokens": 348598993.0, + "step": 15236 + }, + { + "epoch": 1.256452543910283, + "grad_norm": 1.845410704612732, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7842118144035339, + "num_tokens": 348629778.0, + "step": 15237 + }, + { + "epoch": 1.2565350045353343, + "grad_norm": 1.9485846757888794, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.783966064453125, + "num_tokens": 348656377.0, + "step": 15238 + }, + { + "epoch": 1.2566174651603859, + "grad_norm": 1.9004898071289062, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7287849187850952, + "num_tokens": 348692209.0, + "step": 15239 + }, + { + "epoch": 1.2566999257854374, + "grad_norm": 1.7636510133743286, + "learning_rate": 1e-06, + "loss": 0.5441, + "mean_token_accuracy": 0.81449294090271, + "num_tokens": 348724794.0, + "step": 15240 + }, + { + "epoch": 1.256782386410489, + "grad_norm": 1.8220981359481812, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7781733274459839, + "num_tokens": 348757663.0, + "step": 15241 + }, + { + "epoch": 1.2568648470355406, + "grad_norm": 1.9162566661834717, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.761742115020752, + "num_tokens": 348794305.0, + "step": 15242 + }, + { + "epoch": 1.2569473076605922, + "grad_norm": 1.9328092336654663, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.750512421131134, + "num_tokens": 348826181.0, + "step": 15243 + }, + { + "epoch": 1.2570297682856437, + "grad_norm": 2.328627586364746, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7479338645935059, + "num_tokens": 348849658.0, + "step": 15244 + }, + { + "epoch": 1.257112228910695, + "grad_norm": 1.9917218685150146, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7153770923614502, + "num_tokens": 348881281.0, + "step": 15245 + }, + { + "epoch": 1.2571946895357466, + "grad_norm": 1.9633762836456299, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7664793729782104, + "num_tokens": 348915247.0, + "step": 15246 + }, + { + "epoch": 1.2572771501607982, + "grad_norm": 1.983238697052002, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7387474775314331, + "num_tokens": 348947545.0, + "step": 15247 + }, + { + "epoch": 1.2573596107858498, + "grad_norm": 2.1681039333343506, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7131194472312927, + "num_tokens": 348983188.0, + "step": 15248 + }, + { + "epoch": 1.2574420714109014, + "grad_norm": 2.311441421508789, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7394996881484985, + "num_tokens": 349008068.0, + "step": 15249 + }, + { + "epoch": 1.2575245320359527, + "grad_norm": 1.766812801361084, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.7928487062454224, + "num_tokens": 349045579.0, + "step": 15250 + }, + { + "epoch": 1.2576069926610045, + "grad_norm": 1.9186986684799194, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7756377458572388, + "num_tokens": 349076182.0, + "step": 15251 + }, + { + "epoch": 1.2576894532860559, + "grad_norm": 2.0760674476623535, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7682414054870605, + "num_tokens": 349104471.0, + "step": 15252 + }, + { + "epoch": 1.2577719139111074, + "grad_norm": 2.0078697204589844, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.729651927947998, + "num_tokens": 349137224.0, + "step": 15253 + }, + { + "epoch": 1.257854374536159, + "grad_norm": 2.1455764770507812, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7813518047332764, + "num_tokens": 349161729.0, + "step": 15254 + }, + { + "epoch": 1.2579368351612106, + "grad_norm": 1.9658697843551636, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7627050876617432, + "num_tokens": 349193125.0, + "step": 15255 + }, + { + "epoch": 1.2580192957862621, + "grad_norm": 1.927038550376892, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7754704356193542, + "num_tokens": 349223934.0, + "step": 15256 + }, + { + "epoch": 1.2581017564113135, + "grad_norm": 2.0286595821380615, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7452785968780518, + "num_tokens": 349254342.0, + "step": 15257 + }, + { + "epoch": 1.258184217036365, + "grad_norm": 1.8925676345825195, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.785173237323761, + "num_tokens": 349287877.0, + "step": 15258 + }, + { + "epoch": 1.2582666776614166, + "grad_norm": 1.927811861038208, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7579742074012756, + "num_tokens": 349320626.0, + "step": 15259 + }, + { + "epoch": 1.2583491382864682, + "grad_norm": 2.1557323932647705, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7494743466377258, + "num_tokens": 349346844.0, + "step": 15260 + }, + { + "epoch": 1.2584315989115198, + "grad_norm": 1.7448960542678833, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.7929496765136719, + "num_tokens": 349383674.0, + "step": 15261 + }, + { + "epoch": 1.2585140595365714, + "grad_norm": 1.7682853937149048, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7708303928375244, + "num_tokens": 349418534.0, + "step": 15262 + }, + { + "epoch": 1.258596520161623, + "grad_norm": 1.7631680965423584, + "learning_rate": 1e-06, + "loss": 0.5941, + "mean_token_accuracy": 0.797838568687439, + "num_tokens": 349454020.0, + "step": 15263 + }, + { + "epoch": 1.2586789807866743, + "grad_norm": 1.7431899309158325, + "learning_rate": 1e-06, + "loss": 0.5788, + "mean_token_accuracy": 0.8137435913085938, + "num_tokens": 349487119.0, + "step": 15264 + }, + { + "epoch": 1.2587614414117259, + "grad_norm": 1.8644113540649414, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7520074844360352, + "num_tokens": 349522645.0, + "step": 15265 + }, + { + "epoch": 1.2588439020367774, + "grad_norm": 1.924350619316101, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7482066750526428, + "num_tokens": 349554273.0, + "step": 15266 + }, + { + "epoch": 1.258926362661829, + "grad_norm": 1.9664549827575684, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7455202341079712, + "num_tokens": 349585476.0, + "step": 15267 + }, + { + "epoch": 1.2590088232868806, + "grad_norm": 1.8105316162109375, + "learning_rate": 1e-06, + "loss": 0.5792, + "mean_token_accuracy": 0.8078294992446899, + "num_tokens": 349616870.0, + "step": 15268 + }, + { + "epoch": 1.259091283911932, + "grad_norm": 1.541091799736023, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7775540351867676, + "num_tokens": 349661344.0, + "step": 15269 + }, + { + "epoch": 1.2591737445369837, + "grad_norm": 1.965807318687439, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7735950350761414, + "num_tokens": 349690227.0, + "step": 15270 + }, + { + "epoch": 1.259256205162035, + "grad_norm": 1.8745405673980713, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.7798901796340942, + "num_tokens": 349720534.0, + "step": 15271 + }, + { + "epoch": 1.2593386657870866, + "grad_norm": 1.942932367324829, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7674600481987, + "num_tokens": 349753950.0, + "step": 15272 + }, + { + "epoch": 1.2594211264121382, + "grad_norm": 1.8811343908309937, + "learning_rate": 1e-06, + "loss": 0.5117, + "mean_token_accuracy": 0.8217480182647705, + "num_tokens": 349786421.0, + "step": 15273 + }, + { + "epoch": 1.2595035870371898, + "grad_norm": 2.1255757808685303, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7752472162246704, + "num_tokens": 349810454.0, + "step": 15274 + }, + { + "epoch": 1.2595860476622414, + "grad_norm": 1.9599143266677856, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7686888575553894, + "num_tokens": 349838717.0, + "step": 15275 + }, + { + "epoch": 1.2596685082872927, + "grad_norm": 1.951663851737976, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7754160165786743, + "num_tokens": 349872874.0, + "step": 15276 + }, + { + "epoch": 1.2597509689123443, + "grad_norm": 2.1190438270568848, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7643741965293884, + "num_tokens": 349898663.0, + "step": 15277 + }, + { + "epoch": 1.2598334295373959, + "grad_norm": 1.8895295858383179, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7715153098106384, + "num_tokens": 349929763.0, + "step": 15278 + }, + { + "epoch": 1.2599158901624474, + "grad_norm": 2.002035617828369, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7647218704223633, + "num_tokens": 349957008.0, + "step": 15279 + }, + { + "epoch": 1.259998350787499, + "grad_norm": 1.9579676389694214, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7508430480957031, + "num_tokens": 349991909.0, + "step": 15280 + }, + { + "epoch": 1.2600808114125506, + "grad_norm": 1.9431772232055664, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.782090425491333, + "num_tokens": 350020073.0, + "step": 15281 + }, + { + "epoch": 1.2601632720376021, + "grad_norm": 2.093763828277588, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7727331519126892, + "num_tokens": 350048169.0, + "step": 15282 + }, + { + "epoch": 1.2602457326626535, + "grad_norm": 1.9775398969650269, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7366164922714233, + "num_tokens": 350082258.0, + "step": 15283 + }, + { + "epoch": 1.260328193287705, + "grad_norm": 1.9914630651474, + "learning_rate": 1e-06, + "loss": 0.5093, + "mean_token_accuracy": 0.815741240978241, + "num_tokens": 350107831.0, + "step": 15284 + }, + { + "epoch": 1.2604106539127566, + "grad_norm": 1.982419729232788, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7401463985443115, + "num_tokens": 350139011.0, + "step": 15285 + }, + { + "epoch": 1.2604931145378082, + "grad_norm": 1.8494784832000732, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7857422828674316, + "num_tokens": 350173151.0, + "step": 15286 + }, + { + "epoch": 1.2605755751628598, + "grad_norm": 2.0438966751098633, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7668375968933105, + "num_tokens": 350200655.0, + "step": 15287 + }, + { + "epoch": 1.2606580357879111, + "grad_norm": 1.8233791589736938, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7702043056488037, + "num_tokens": 350233195.0, + "step": 15288 + }, + { + "epoch": 1.260740496412963, + "grad_norm": 1.8030765056610107, + "learning_rate": 1e-06, + "loss": 0.5758, + "mean_token_accuracy": 0.8104267120361328, + "num_tokens": 350267280.0, + "step": 15289 + }, + { + "epoch": 1.2608229570380143, + "grad_norm": 2.0354814529418945, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7661024332046509, + "num_tokens": 350297454.0, + "step": 15290 + }, + { + "epoch": 1.2609054176630659, + "grad_norm": 2.1195883750915527, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7852245569229126, + "num_tokens": 350323508.0, + "step": 15291 + }, + { + "epoch": 1.2609878782881174, + "grad_norm": 2.01753830909729, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.783011257648468, + "num_tokens": 350354281.0, + "step": 15292 + }, + { + "epoch": 1.261070338913169, + "grad_norm": 2.22099232673645, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7741885781288147, + "num_tokens": 350380875.0, + "step": 15293 + }, + { + "epoch": 1.2611527995382206, + "grad_norm": 2.074923515319824, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7525920867919922, + "num_tokens": 350412225.0, + "step": 15294 + }, + { + "epoch": 1.261235260163272, + "grad_norm": 2.0516116619110107, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7391903400421143, + "num_tokens": 350442008.0, + "step": 15295 + }, + { + "epoch": 1.2613177207883235, + "grad_norm": 1.937101125717163, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.7738964557647705, + "num_tokens": 350473261.0, + "step": 15296 + }, + { + "epoch": 1.261400181413375, + "grad_norm": 1.6837927103042603, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7696307897567749, + "num_tokens": 350516589.0, + "step": 15297 + }, + { + "epoch": 1.2614826420384266, + "grad_norm": 1.9096605777740479, + "learning_rate": 1e-06, + "loss": 0.5933, + "mean_token_accuracy": 0.8005828857421875, + "num_tokens": 350546368.0, + "step": 15298 + }, + { + "epoch": 1.2615651026634782, + "grad_norm": 2.0551352500915527, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7630171775817871, + "num_tokens": 350574880.0, + "step": 15299 + }, + { + "epoch": 1.2616475632885298, + "grad_norm": 1.9608124494552612, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.7939844131469727, + "num_tokens": 350605015.0, + "step": 15300 + }, + { + "epoch": 1.2617300239135814, + "grad_norm": 1.8074358701705933, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.781154453754425, + "num_tokens": 350643765.0, + "step": 15301 + }, + { + "epoch": 1.2618124845386327, + "grad_norm": 2.13486909866333, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.73399817943573, + "num_tokens": 350672592.0, + "step": 15302 + }, + { + "epoch": 1.2618949451636843, + "grad_norm": 1.9553261995315552, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.7899295687675476, + "num_tokens": 350700873.0, + "step": 15303 + }, + { + "epoch": 1.2619774057887359, + "grad_norm": 2.051527976989746, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7730402946472168, + "num_tokens": 350730552.0, + "step": 15304 + }, + { + "epoch": 1.2620598664137874, + "grad_norm": 2.03432297706604, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7786320447921753, + "num_tokens": 350761471.0, + "step": 15305 + }, + { + "epoch": 1.262142327038839, + "grad_norm": 1.9591681957244873, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.791288435459137, + "num_tokens": 350788189.0, + "step": 15306 + }, + { + "epoch": 1.2622247876638906, + "grad_norm": 1.860593318939209, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7464905977249146, + "num_tokens": 350820024.0, + "step": 15307 + }, + { + "epoch": 1.2623072482889421, + "grad_norm": 1.897796392440796, + "learning_rate": 1e-06, + "loss": 0.5733, + "mean_token_accuracy": 0.8102793097496033, + "num_tokens": 350851245.0, + "step": 15308 + }, + { + "epoch": 1.2623897089139935, + "grad_norm": 1.9253114461898804, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7622050046920776, + "num_tokens": 350882984.0, + "step": 15309 + }, + { + "epoch": 1.262472169539045, + "grad_norm": 1.9231891632080078, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.7958399057388306, + "num_tokens": 350911639.0, + "step": 15310 + }, + { + "epoch": 1.2625546301640966, + "grad_norm": 1.9035594463348389, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7680166959762573, + "num_tokens": 350946531.0, + "step": 15311 + }, + { + "epoch": 1.2626370907891482, + "grad_norm": 1.9307117462158203, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.7914543151855469, + "num_tokens": 350978389.0, + "step": 15312 + }, + { + "epoch": 1.2627195514141998, + "grad_norm": 1.8048449754714966, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7735642790794373, + "num_tokens": 351012523.0, + "step": 15313 + }, + { + "epoch": 1.2628020120392511, + "grad_norm": 2.0317652225494385, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7366036176681519, + "num_tokens": 351042631.0, + "step": 15314 + }, + { + "epoch": 1.262884472664303, + "grad_norm": 1.8414169549942017, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7371797561645508, + "num_tokens": 351077644.0, + "step": 15315 + }, + { + "epoch": 1.2629669332893543, + "grad_norm": 1.8928260803222656, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7361130118370056, + "num_tokens": 351112253.0, + "step": 15316 + }, + { + "epoch": 1.2630493939144058, + "grad_norm": 1.889946460723877, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.7780988812446594, + "num_tokens": 351146808.0, + "step": 15317 + }, + { + "epoch": 1.2631318545394574, + "grad_norm": 1.908279299736023, + "learning_rate": 1e-06, + "loss": 0.7724, + "mean_token_accuracy": 0.7519963979721069, + "num_tokens": 351177768.0, + "step": 15318 + }, + { + "epoch": 1.263214315164509, + "grad_norm": 2.0961883068084717, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7460777759552002, + "num_tokens": 351205609.0, + "step": 15319 + }, + { + "epoch": 1.2632967757895606, + "grad_norm": 1.9197466373443604, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7656093835830688, + "num_tokens": 351236517.0, + "step": 15320 + }, + { + "epoch": 1.263379236414612, + "grad_norm": 1.7212419509887695, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.7987436056137085, + "num_tokens": 351271561.0, + "step": 15321 + }, + { + "epoch": 1.2634616970396635, + "grad_norm": 1.897897481918335, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.776226282119751, + "num_tokens": 351301820.0, + "step": 15322 + }, + { + "epoch": 1.263544157664715, + "grad_norm": 1.925217866897583, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7697359323501587, + "num_tokens": 351333951.0, + "step": 15323 + }, + { + "epoch": 1.2636266182897666, + "grad_norm": 1.9109163284301758, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7518059015274048, + "num_tokens": 351365053.0, + "step": 15324 + }, + { + "epoch": 1.2637090789148182, + "grad_norm": 1.9487905502319336, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7603415250778198, + "num_tokens": 351394134.0, + "step": 15325 + }, + { + "epoch": 1.2637915395398698, + "grad_norm": 1.8591821193695068, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.7875959277153015, + "num_tokens": 351424239.0, + "step": 15326 + }, + { + "epoch": 1.2638740001649214, + "grad_norm": 1.978299617767334, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7722630500793457, + "num_tokens": 351452119.0, + "step": 15327 + }, + { + "epoch": 1.2639564607899727, + "grad_norm": 2.152261972427368, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7102847099304199, + "num_tokens": 351480723.0, + "step": 15328 + }, + { + "epoch": 1.2640389214150243, + "grad_norm": 1.7615728378295898, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7798613905906677, + "num_tokens": 351515803.0, + "step": 15329 + }, + { + "epoch": 1.2641213820400758, + "grad_norm": 2.1414616107940674, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.7925375699996948, + "num_tokens": 351542924.0, + "step": 15330 + }, + { + "epoch": 1.2642038426651274, + "grad_norm": 1.934336543083191, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7804264426231384, + "num_tokens": 351570864.0, + "step": 15331 + }, + { + "epoch": 1.264286303290179, + "grad_norm": 1.937188982963562, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.743398904800415, + "num_tokens": 351605327.0, + "step": 15332 + }, + { + "epoch": 1.2643687639152303, + "grad_norm": 2.047074317932129, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7760765552520752, + "num_tokens": 351633632.0, + "step": 15333 + }, + { + "epoch": 1.2644512245402821, + "grad_norm": 1.716173529624939, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.7978375554084778, + "num_tokens": 351668264.0, + "step": 15334 + }, + { + "epoch": 1.2645336851653335, + "grad_norm": 2.0732877254486084, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7830329537391663, + "num_tokens": 351696621.0, + "step": 15335 + }, + { + "epoch": 1.264616145790385, + "grad_norm": 1.8416032791137695, + "learning_rate": 1e-06, + "loss": 0.6007, + "mean_token_accuracy": 0.8003134727478027, + "num_tokens": 351728619.0, + "step": 15336 + }, + { + "epoch": 1.2646986064154366, + "grad_norm": 1.8857187032699585, + "learning_rate": 1e-06, + "loss": 0.5601, + "mean_token_accuracy": 0.8002827763557434, + "num_tokens": 351759996.0, + "step": 15337 + }, + { + "epoch": 1.2647810670404882, + "grad_norm": 1.9575958251953125, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7609366178512573, + "num_tokens": 351791038.0, + "step": 15338 + }, + { + "epoch": 1.2648635276655398, + "grad_norm": 1.8781110048294067, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7454069256782532, + "num_tokens": 351825000.0, + "step": 15339 + }, + { + "epoch": 1.2649459882905911, + "grad_norm": 2.0013418197631836, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7819875478744507, + "num_tokens": 351853878.0, + "step": 15340 + }, + { + "epoch": 1.2650284489156427, + "grad_norm": 2.048332929611206, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7263791561126709, + "num_tokens": 351883929.0, + "step": 15341 + }, + { + "epoch": 1.2651109095406943, + "grad_norm": 1.857418417930603, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.788806140422821, + "num_tokens": 351918469.0, + "step": 15342 + }, + { + "epoch": 1.2651933701657458, + "grad_norm": 1.6958295106887817, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7640012502670288, + "num_tokens": 351957305.0, + "step": 15343 + }, + { + "epoch": 1.2652758307907974, + "grad_norm": 1.8073108196258545, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7657261490821838, + "num_tokens": 351992307.0, + "step": 15344 + }, + { + "epoch": 1.265358291415849, + "grad_norm": 2.166048288345337, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.732266902923584, + "num_tokens": 352020785.0, + "step": 15345 + }, + { + "epoch": 1.2654407520409006, + "grad_norm": 2.0032169818878174, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7753349542617798, + "num_tokens": 352049972.0, + "step": 15346 + }, + { + "epoch": 1.265523212665952, + "grad_norm": 1.9822990894317627, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7545006275177002, + "num_tokens": 352082412.0, + "step": 15347 + }, + { + "epoch": 1.2656056732910035, + "grad_norm": 1.930283546447754, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7543500661849976, + "num_tokens": 352112398.0, + "step": 15348 + }, + { + "epoch": 1.265688133916055, + "grad_norm": 1.852287769317627, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7557597756385803, + "num_tokens": 352145375.0, + "step": 15349 + }, + { + "epoch": 1.2657705945411066, + "grad_norm": 2.002544641494751, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7609114646911621, + "num_tokens": 352176013.0, + "step": 15350 + }, + { + "epoch": 1.2658530551661582, + "grad_norm": 1.8373795747756958, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7505007982254028, + "num_tokens": 352211512.0, + "step": 15351 + }, + { + "epoch": 1.2659355157912098, + "grad_norm": 1.9927093982696533, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7616477012634277, + "num_tokens": 352240775.0, + "step": 15352 + }, + { + "epoch": 1.2660179764162613, + "grad_norm": 2.4587953090667725, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7355605363845825, + "num_tokens": 352263988.0, + "step": 15353 + }, + { + "epoch": 1.2661004370413127, + "grad_norm": 1.9569039344787598, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.7870157957077026, + "num_tokens": 352290951.0, + "step": 15354 + }, + { + "epoch": 1.2661828976663643, + "grad_norm": 1.9531482458114624, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.7854496836662292, + "num_tokens": 352322401.0, + "step": 15355 + }, + { + "epoch": 1.2662653582914158, + "grad_norm": 2.130901575088501, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7505687475204468, + "num_tokens": 352353382.0, + "step": 15356 + }, + { + "epoch": 1.2663478189164674, + "grad_norm": 1.912531852722168, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7492278814315796, + "num_tokens": 352386535.0, + "step": 15357 + }, + { + "epoch": 1.266430279541519, + "grad_norm": 2.2939083576202393, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7663747072219849, + "num_tokens": 352408328.0, + "step": 15358 + }, + { + "epoch": 1.2665127401665703, + "grad_norm": 1.8458155393600464, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7621392607688904, + "num_tokens": 352440287.0, + "step": 15359 + }, + { + "epoch": 1.2665952007916221, + "grad_norm": 1.8754682540893555, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7759158611297607, + "num_tokens": 352473051.0, + "step": 15360 + }, + { + "epoch": 1.2666776614166735, + "grad_norm": 1.8511137962341309, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7397342324256897, + "num_tokens": 352509598.0, + "step": 15361 + }, + { + "epoch": 1.266760122041725, + "grad_norm": 2.1753084659576416, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7696678042411804, + "num_tokens": 352533796.0, + "step": 15362 + }, + { + "epoch": 1.2668425826667766, + "grad_norm": 2.2699642181396484, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7767394185066223, + "num_tokens": 352557720.0, + "step": 15363 + }, + { + "epoch": 1.2669250432918282, + "grad_norm": 2.0889601707458496, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7275763750076294, + "num_tokens": 352589083.0, + "step": 15364 + }, + { + "epoch": 1.2670075039168798, + "grad_norm": 2.120262384414673, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7379281520843506, + "num_tokens": 352616684.0, + "step": 15365 + }, + { + "epoch": 1.2670899645419311, + "grad_norm": 2.1531801223754883, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7490156888961792, + "num_tokens": 352641197.0, + "step": 15366 + }, + { + "epoch": 1.2671724251669827, + "grad_norm": 2.0628442764282227, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7616046667098999, + "num_tokens": 352668112.0, + "step": 15367 + }, + { + "epoch": 1.2672548857920343, + "grad_norm": 1.9680473804473877, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7600655555725098, + "num_tokens": 352700080.0, + "step": 15368 + }, + { + "epoch": 1.2673373464170858, + "grad_norm": 2.348844051361084, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.751333475112915, + "num_tokens": 352724837.0, + "step": 15369 + }, + { + "epoch": 1.2674198070421374, + "grad_norm": 2.019366502761841, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7494274973869324, + "num_tokens": 352753693.0, + "step": 15370 + }, + { + "epoch": 1.267502267667189, + "grad_norm": 2.15360426902771, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7505306601524353, + "num_tokens": 352787394.0, + "step": 15371 + }, + { + "epoch": 1.2675847282922406, + "grad_norm": 1.8117597103118896, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7669029235839844, + "num_tokens": 352821452.0, + "step": 15372 + }, + { + "epoch": 1.267667188917292, + "grad_norm": 1.7408443689346313, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.7496235966682434, + "num_tokens": 352860368.0, + "step": 15373 + }, + { + "epoch": 1.2677496495423435, + "grad_norm": 2.1297683715820312, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7457576990127563, + "num_tokens": 352887642.0, + "step": 15374 + }, + { + "epoch": 1.267832110167395, + "grad_norm": 2.255695104598999, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7509663105010986, + "num_tokens": 352912311.0, + "step": 15375 + }, + { + "epoch": 1.2679145707924466, + "grad_norm": 2.150080680847168, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.755211591720581, + "num_tokens": 352938106.0, + "step": 15376 + }, + { + "epoch": 1.2679970314174982, + "grad_norm": 1.9443515539169312, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.7857945561408997, + "num_tokens": 352967957.0, + "step": 15377 + }, + { + "epoch": 1.2680794920425495, + "grad_norm": 2.0875015258789062, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7570083737373352, + "num_tokens": 352994926.0, + "step": 15378 + }, + { + "epoch": 1.2681619526676013, + "grad_norm": 2.1533515453338623, + "learning_rate": 1e-06, + "loss": 0.564, + "mean_token_accuracy": 0.804688572883606, + "num_tokens": 353022782.0, + "step": 15379 + }, + { + "epoch": 1.2682444132926527, + "grad_norm": 1.9459245204925537, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7168833017349243, + "num_tokens": 353054460.0, + "step": 15380 + }, + { + "epoch": 1.2683268739177043, + "grad_norm": 1.809828519821167, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.7876852750778198, + "num_tokens": 353090736.0, + "step": 15381 + }, + { + "epoch": 1.2684093345427558, + "grad_norm": 1.8339940309524536, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.7919660210609436, + "num_tokens": 353123250.0, + "step": 15382 + }, + { + "epoch": 1.2684917951678074, + "grad_norm": 2.1566264629364014, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7569944262504578, + "num_tokens": 353151161.0, + "step": 15383 + }, + { + "epoch": 1.268574255792859, + "grad_norm": 1.8417613506317139, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.7922102212905884, + "num_tokens": 353184512.0, + "step": 15384 + }, + { + "epoch": 1.2686567164179103, + "grad_norm": 2.1347835063934326, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7537803053855896, + "num_tokens": 353213251.0, + "step": 15385 + }, + { + "epoch": 1.268739177042962, + "grad_norm": 1.6923388242721558, + "learning_rate": 1e-06, + "loss": 0.4779, + "mean_token_accuracy": 0.8412165641784668, + "num_tokens": 353250089.0, + "step": 15386 + }, + { + "epoch": 1.2688216376680135, + "grad_norm": 2.1788766384124756, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7620705962181091, + "num_tokens": 353277256.0, + "step": 15387 + }, + { + "epoch": 1.268904098293065, + "grad_norm": 1.8357208967208862, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7717705368995667, + "num_tokens": 353310671.0, + "step": 15388 + }, + { + "epoch": 1.2689865589181166, + "grad_norm": 1.816394329071045, + "learning_rate": 1e-06, + "loss": 0.6087, + "mean_token_accuracy": 0.7923946976661682, + "num_tokens": 353344672.0, + "step": 15389 + }, + { + "epoch": 1.2690690195431682, + "grad_norm": 1.7236396074295044, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7610323429107666, + "num_tokens": 353385610.0, + "step": 15390 + }, + { + "epoch": 1.2691514801682198, + "grad_norm": 2.0470426082611084, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7743161916732788, + "num_tokens": 353413250.0, + "step": 15391 + }, + { + "epoch": 1.2692339407932711, + "grad_norm": 2.063936710357666, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.767174243927002, + "num_tokens": 353438975.0, + "step": 15392 + }, + { + "epoch": 1.2693164014183227, + "grad_norm": 1.9561924934387207, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7712287902832031, + "num_tokens": 353469131.0, + "step": 15393 + }, + { + "epoch": 1.2693988620433743, + "grad_norm": 1.943078875541687, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7410331964492798, + "num_tokens": 353501711.0, + "step": 15394 + }, + { + "epoch": 1.2694813226684258, + "grad_norm": 2.1597321033477783, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7360129356384277, + "num_tokens": 353530613.0, + "step": 15395 + }, + { + "epoch": 1.2695637832934774, + "grad_norm": 1.7085442543029785, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7821159362792969, + "num_tokens": 353566410.0, + "step": 15396 + }, + { + "epoch": 1.269646243918529, + "grad_norm": 1.9361572265625, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.758826494216919, + "num_tokens": 353595467.0, + "step": 15397 + }, + { + "epoch": 1.2697287045435806, + "grad_norm": 1.988207459449768, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7580251693725586, + "num_tokens": 353625324.0, + "step": 15398 + }, + { + "epoch": 1.269811165168632, + "grad_norm": 1.7203274965286255, + "learning_rate": 1e-06, + "loss": 0.5586, + "mean_token_accuracy": 0.8123171329498291, + "num_tokens": 353661782.0, + "step": 15399 + }, + { + "epoch": 1.2698936257936835, + "grad_norm": 2.0821845531463623, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.7985706329345703, + "num_tokens": 353688148.0, + "step": 15400 + }, + { + "epoch": 1.269976086418735, + "grad_norm": 2.051499605178833, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7688122987747192, + "num_tokens": 353716143.0, + "step": 15401 + }, + { + "epoch": 1.2700585470437866, + "grad_norm": 2.0437746047973633, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.777586042881012, + "num_tokens": 353743853.0, + "step": 15402 + }, + { + "epoch": 1.2701410076688382, + "grad_norm": 2.0081512928009033, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7339000105857849, + "num_tokens": 353774475.0, + "step": 15403 + }, + { + "epoch": 1.2702234682938895, + "grad_norm": 2.0109364986419678, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7354026436805725, + "num_tokens": 353805369.0, + "step": 15404 + }, + { + "epoch": 1.2703059289189413, + "grad_norm": 1.8513662815093994, + "learning_rate": 1e-06, + "loss": 0.5922, + "mean_token_accuracy": 0.8012683987617493, + "num_tokens": 353838072.0, + "step": 15405 + }, + { + "epoch": 1.2703883895439927, + "grad_norm": 2.2974231243133545, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.736483633518219, + "num_tokens": 353864579.0, + "step": 15406 + }, + { + "epoch": 1.2704708501690443, + "grad_norm": 2.136874198913574, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7457691431045532, + "num_tokens": 353892593.0, + "step": 15407 + }, + { + "epoch": 1.2705533107940958, + "grad_norm": 1.8681068420410156, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.8030946850776672, + "num_tokens": 353926061.0, + "step": 15408 + }, + { + "epoch": 1.2706357714191474, + "grad_norm": 1.7750016450881958, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7514607906341553, + "num_tokens": 353963181.0, + "step": 15409 + }, + { + "epoch": 1.270718232044199, + "grad_norm": 1.9867337942123413, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7325498461723328, + "num_tokens": 353994838.0, + "step": 15410 + }, + { + "epoch": 1.2708006926692503, + "grad_norm": 1.9272172451019287, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7505311369895935, + "num_tokens": 354030540.0, + "step": 15411 + }, + { + "epoch": 1.270883153294302, + "grad_norm": 1.9131361246109009, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.749528169631958, + "num_tokens": 354065644.0, + "step": 15412 + }, + { + "epoch": 1.2709656139193535, + "grad_norm": 2.194934129714966, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7645437717437744, + "num_tokens": 354090288.0, + "step": 15413 + }, + { + "epoch": 1.271048074544405, + "grad_norm": 2.042402744293213, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.7799543738365173, + "num_tokens": 354118613.0, + "step": 15414 + }, + { + "epoch": 1.2711305351694566, + "grad_norm": 1.9926477670669556, + "learning_rate": 1e-06, + "loss": 0.6239, + "mean_token_accuracy": 0.790006160736084, + "num_tokens": 354144678.0, + "step": 15415 + }, + { + "epoch": 1.2712129957945082, + "grad_norm": 1.9631969928741455, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7665256857872009, + "num_tokens": 354173942.0, + "step": 15416 + }, + { + "epoch": 1.2712954564195598, + "grad_norm": 2.213618278503418, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7088572382926941, + "num_tokens": 354198851.0, + "step": 15417 + }, + { + "epoch": 1.2713779170446111, + "grad_norm": 1.8101398944854736, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.7804505825042725, + "num_tokens": 354233683.0, + "step": 15418 + }, + { + "epoch": 1.2714603776696627, + "grad_norm": 2.037851333618164, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.7945674657821655, + "num_tokens": 354261539.0, + "step": 15419 + }, + { + "epoch": 1.2715428382947143, + "grad_norm": 1.743307113647461, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.7981521487236023, + "num_tokens": 354299431.0, + "step": 15420 + }, + { + "epoch": 1.2716252989197658, + "grad_norm": 2.0217859745025635, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.77863609790802, + "num_tokens": 354325986.0, + "step": 15421 + }, + { + "epoch": 1.2717077595448174, + "grad_norm": 2.2175774574279785, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7277669906616211, + "num_tokens": 354354814.0, + "step": 15422 + }, + { + "epoch": 1.2717902201698688, + "grad_norm": 2.0293006896972656, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7010759115219116, + "num_tokens": 354383436.0, + "step": 15423 + }, + { + "epoch": 1.2718726807949206, + "grad_norm": 1.8461939096450806, + "learning_rate": 1e-06, + "loss": 0.6062, + "mean_token_accuracy": 0.8069866299629211, + "num_tokens": 354415402.0, + "step": 15424 + }, + { + "epoch": 1.271955141419972, + "grad_norm": 1.9331855773925781, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7398854494094849, + "num_tokens": 354448187.0, + "step": 15425 + }, + { + "epoch": 1.2720376020450235, + "grad_norm": 2.0651960372924805, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.763039767742157, + "num_tokens": 354478438.0, + "step": 15426 + }, + { + "epoch": 1.272120062670075, + "grad_norm": 2.0884604454040527, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7536665797233582, + "num_tokens": 354505367.0, + "step": 15427 + }, + { + "epoch": 1.2722025232951266, + "grad_norm": 2.105161428451538, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.7554647922515869, + "num_tokens": 354532436.0, + "step": 15428 + }, + { + "epoch": 1.2722849839201782, + "grad_norm": 2.046581506729126, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7368065118789673, + "num_tokens": 354561701.0, + "step": 15429 + }, + { + "epoch": 1.2723674445452295, + "grad_norm": 2.495293617248535, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7791526317596436, + "num_tokens": 354582660.0, + "step": 15430 + }, + { + "epoch": 1.2724499051702811, + "grad_norm": 1.949837327003479, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7748212218284607, + "num_tokens": 354613946.0, + "step": 15431 + }, + { + "epoch": 1.2725323657953327, + "grad_norm": 2.1487858295440674, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7685033679008484, + "num_tokens": 354640452.0, + "step": 15432 + }, + { + "epoch": 1.2726148264203843, + "grad_norm": 2.280015707015991, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7411683797836304, + "num_tokens": 354664430.0, + "step": 15433 + }, + { + "epoch": 1.2726972870454358, + "grad_norm": 1.943070888519287, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7468151450157166, + "num_tokens": 354695842.0, + "step": 15434 + }, + { + "epoch": 1.2727797476704874, + "grad_norm": 1.8335309028625488, + "learning_rate": 1e-06, + "loss": 0.5874, + "mean_token_accuracy": 0.794766902923584, + "num_tokens": 354729226.0, + "step": 15435 + }, + { + "epoch": 1.272862208295539, + "grad_norm": 2.2145190238952637, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.757341206073761, + "num_tokens": 354754018.0, + "step": 15436 + }, + { + "epoch": 1.2729446689205903, + "grad_norm": 1.995367407798767, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7264232635498047, + "num_tokens": 354785033.0, + "step": 15437 + }, + { + "epoch": 1.273027129545642, + "grad_norm": 1.7950907945632935, + "learning_rate": 1e-06, + "loss": 0.6077, + "mean_token_accuracy": 0.7948805093765259, + "num_tokens": 354820202.0, + "step": 15438 + }, + { + "epoch": 1.2731095901706935, + "grad_norm": 1.9818596839904785, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7393803596496582, + "num_tokens": 354849772.0, + "step": 15439 + }, + { + "epoch": 1.273192050795745, + "grad_norm": 1.9617750644683838, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7695837616920471, + "num_tokens": 354876619.0, + "step": 15440 + }, + { + "epoch": 1.2732745114207966, + "grad_norm": 2.063892126083374, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.7856047749519348, + "num_tokens": 354907186.0, + "step": 15441 + }, + { + "epoch": 1.2733569720458482, + "grad_norm": 1.852425217628479, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7628797292709351, + "num_tokens": 354941881.0, + "step": 15442 + }, + { + "epoch": 1.2734394326708998, + "grad_norm": 1.7458689212799072, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7622287273406982, + "num_tokens": 354976862.0, + "step": 15443 + }, + { + "epoch": 1.2735218932959511, + "grad_norm": 2.0866920948028564, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7451667189598083, + "num_tokens": 355005718.0, + "step": 15444 + }, + { + "epoch": 1.2736043539210027, + "grad_norm": 1.8436092138290405, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8052804470062256, + "num_tokens": 355035612.0, + "step": 15445 + }, + { + "epoch": 1.2736868145460543, + "grad_norm": 1.873520016670227, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7319572567939758, + "num_tokens": 355070264.0, + "step": 15446 + }, + { + "epoch": 1.2737692751711058, + "grad_norm": 1.7702480554580688, + "learning_rate": 1e-06, + "loss": 0.5713, + "mean_token_accuracy": 0.8157556056976318, + "num_tokens": 355102522.0, + "step": 15447 + }, + { + "epoch": 1.2738517357961574, + "grad_norm": 1.8000270128250122, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7729306221008301, + "num_tokens": 355139545.0, + "step": 15448 + }, + { + "epoch": 1.2739341964212088, + "grad_norm": 1.9971317052841187, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.744619607925415, + "num_tokens": 355171537.0, + "step": 15449 + }, + { + "epoch": 1.2740166570462605, + "grad_norm": 2.0665652751922607, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7727034687995911, + "num_tokens": 355201054.0, + "step": 15450 + }, + { + "epoch": 1.274099117671312, + "grad_norm": 2.1235103607177734, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7870168685913086, + "num_tokens": 355226856.0, + "step": 15451 + }, + { + "epoch": 1.2741815782963635, + "grad_norm": 1.9861775636672974, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7793790102005005, + "num_tokens": 355257582.0, + "step": 15452 + }, + { + "epoch": 1.274264038921415, + "grad_norm": 2.198780059814453, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7509627342224121, + "num_tokens": 355285264.0, + "step": 15453 + }, + { + "epoch": 1.2743464995464666, + "grad_norm": 2.041249990463257, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7244995832443237, + "num_tokens": 355313605.0, + "step": 15454 + }, + { + "epoch": 1.2744289601715182, + "grad_norm": 1.9393820762634277, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7571374773979187, + "num_tokens": 355348087.0, + "step": 15455 + }, + { + "epoch": 1.2745114207965695, + "grad_norm": 2.1475653648376465, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7727360725402832, + "num_tokens": 355374777.0, + "step": 15456 + }, + { + "epoch": 1.274593881421621, + "grad_norm": 2.0702145099639893, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7409895658493042, + "num_tokens": 355403917.0, + "step": 15457 + }, + { + "epoch": 1.2746763420466727, + "grad_norm": 2.058140754699707, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7233800888061523, + "num_tokens": 355436206.0, + "step": 15458 + }, + { + "epoch": 1.2747588026717243, + "grad_norm": 1.8987812995910645, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7821853756904602, + "num_tokens": 355468527.0, + "step": 15459 + }, + { + "epoch": 1.2748412632967758, + "grad_norm": 2.65492582321167, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7713435888290405, + "num_tokens": 355487120.0, + "step": 15460 + }, + { + "epoch": 1.2749237239218274, + "grad_norm": 2.221722364425659, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7760970592498779, + "num_tokens": 355510840.0, + "step": 15461 + }, + { + "epoch": 1.275006184546879, + "grad_norm": 2.132239818572998, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7763926982879639, + "num_tokens": 355537275.0, + "step": 15462 + }, + { + "epoch": 1.2750886451719303, + "grad_norm": 1.8833187818527222, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.7965699434280396, + "num_tokens": 355567817.0, + "step": 15463 + }, + { + "epoch": 1.275171105796982, + "grad_norm": 1.8205097913742065, + "learning_rate": 1e-06, + "loss": 0.5802, + "mean_token_accuracy": 0.8017296195030212, + "num_tokens": 355597075.0, + "step": 15464 + }, + { + "epoch": 1.2752535664220335, + "grad_norm": 1.909837245941162, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7593386173248291, + "num_tokens": 355628529.0, + "step": 15465 + }, + { + "epoch": 1.275336027047085, + "grad_norm": 1.8479024171829224, + "learning_rate": 1e-06, + "loss": 0.7531, + "mean_token_accuracy": 0.7617631554603577, + "num_tokens": 355663963.0, + "step": 15466 + }, + { + "epoch": 1.2754184876721366, + "grad_norm": 1.7660164833068848, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7789665460586548, + "num_tokens": 355697163.0, + "step": 15467 + }, + { + "epoch": 1.275500948297188, + "grad_norm": 1.9740256071090698, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7561461925506592, + "num_tokens": 355727126.0, + "step": 15468 + }, + { + "epoch": 1.2755834089222398, + "grad_norm": 1.9751955270767212, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.7934920787811279, + "num_tokens": 355754810.0, + "step": 15469 + }, + { + "epoch": 1.275665869547291, + "grad_norm": 2.15643310546875, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.785078763961792, + "num_tokens": 355784610.0, + "step": 15470 + }, + { + "epoch": 1.2757483301723427, + "grad_norm": 1.9906986951828003, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.7963557839393616, + "num_tokens": 355811190.0, + "step": 15471 + }, + { + "epoch": 1.2758307907973943, + "grad_norm": 1.8909953832626343, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7631567120552063, + "num_tokens": 355843626.0, + "step": 15472 + }, + { + "epoch": 1.2759132514224458, + "grad_norm": 1.9309662580490112, + "learning_rate": 1e-06, + "loss": 0.5541, + "mean_token_accuracy": 0.8097507357597351, + "num_tokens": 355874497.0, + "step": 15473 + }, + { + "epoch": 1.2759957120474974, + "grad_norm": 1.9917304515838623, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7754278182983398, + "num_tokens": 355904410.0, + "step": 15474 + }, + { + "epoch": 1.2760781726725487, + "grad_norm": 1.946190357208252, + "learning_rate": 1e-06, + "loss": 0.6114, + "mean_token_accuracy": 0.7853108644485474, + "num_tokens": 355936583.0, + "step": 15475 + }, + { + "epoch": 1.2761606332976003, + "grad_norm": 2.0424985885620117, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7565666437149048, + "num_tokens": 355968422.0, + "step": 15476 + }, + { + "epoch": 1.276243093922652, + "grad_norm": 2.03619384765625, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7142523527145386, + "num_tokens": 356001634.0, + "step": 15477 + }, + { + "epoch": 1.2763255545477035, + "grad_norm": 1.8960421085357666, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7713482975959778, + "num_tokens": 356032077.0, + "step": 15478 + }, + { + "epoch": 1.276408015172755, + "grad_norm": 2.287034511566162, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7348130941390991, + "num_tokens": 356057358.0, + "step": 15479 + }, + { + "epoch": 1.2764904757978066, + "grad_norm": 1.9547420740127563, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7675659656524658, + "num_tokens": 356086567.0, + "step": 15480 + }, + { + "epoch": 1.2765729364228582, + "grad_norm": 1.8920276165008545, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.7890565395355225, + "num_tokens": 356117393.0, + "step": 15481 + }, + { + "epoch": 1.2766553970479095, + "grad_norm": 2.1813857555389404, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7217508554458618, + "num_tokens": 356146153.0, + "step": 15482 + }, + { + "epoch": 1.276737857672961, + "grad_norm": 1.7931063175201416, + "learning_rate": 1e-06, + "loss": 0.6239, + "mean_token_accuracy": 0.7943427562713623, + "num_tokens": 356181307.0, + "step": 15483 + }, + { + "epoch": 1.2768203182980127, + "grad_norm": 2.223106622695923, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7311418652534485, + "num_tokens": 356208750.0, + "step": 15484 + }, + { + "epoch": 1.2769027789230643, + "grad_norm": 1.89740788936615, + "learning_rate": 1e-06, + "loss": 0.6126, + "mean_token_accuracy": 0.792730450630188, + "num_tokens": 356238488.0, + "step": 15485 + }, + { + "epoch": 1.2769852395481158, + "grad_norm": 2.1182596683502197, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7358561158180237, + "num_tokens": 356265776.0, + "step": 15486 + }, + { + "epoch": 1.2770677001731672, + "grad_norm": 1.9895468950271606, + "learning_rate": 1e-06, + "loss": 0.6118, + "mean_token_accuracy": 0.7955880761146545, + "num_tokens": 356295728.0, + "step": 15487 + }, + { + "epoch": 1.277150160798219, + "grad_norm": 2.2128546237945557, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7662667036056519, + "num_tokens": 356319948.0, + "step": 15488 + }, + { + "epoch": 1.2772326214232703, + "grad_norm": 2.2255659103393555, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7726030349731445, + "num_tokens": 356343380.0, + "step": 15489 + }, + { + "epoch": 1.277315082048322, + "grad_norm": 1.843556523323059, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7696596384048462, + "num_tokens": 356374765.0, + "step": 15490 + }, + { + "epoch": 1.2773975426733735, + "grad_norm": 1.8497635126113892, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.7901919484138489, + "num_tokens": 356407102.0, + "step": 15491 + }, + { + "epoch": 1.277480003298425, + "grad_norm": 1.9562681913375854, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7662525177001953, + "num_tokens": 356441316.0, + "step": 15492 + }, + { + "epoch": 1.2775624639234766, + "grad_norm": 2.0001766681671143, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7813169956207275, + "num_tokens": 356470740.0, + "step": 15493 + }, + { + "epoch": 1.277644924548528, + "grad_norm": 1.8253239393234253, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.737370491027832, + "num_tokens": 356510221.0, + "step": 15494 + }, + { + "epoch": 1.2777273851735795, + "grad_norm": 2.0292930603027344, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7709904909133911, + "num_tokens": 356537028.0, + "step": 15495 + }, + { + "epoch": 1.277809845798631, + "grad_norm": 1.9107942581176758, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7628809213638306, + "num_tokens": 356571824.0, + "step": 15496 + }, + { + "epoch": 1.2778923064236827, + "grad_norm": 2.22408127784729, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7611252069473267, + "num_tokens": 356597360.0, + "step": 15497 + }, + { + "epoch": 1.2779747670487343, + "grad_norm": 2.1568586826324463, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.767113983631134, + "num_tokens": 356622791.0, + "step": 15498 + }, + { + "epoch": 1.2780572276737858, + "grad_norm": 2.1701393127441406, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7338358163833618, + "num_tokens": 356648390.0, + "step": 15499 + }, + { + "epoch": 1.2781396882988374, + "grad_norm": 2.259145498275757, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.7922695279121399, + "num_tokens": 356669395.0, + "step": 15500 + }, + { + "epoch": 1.2782221489238887, + "grad_norm": 1.9828574657440186, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7683255672454834, + "num_tokens": 356697723.0, + "step": 15501 + }, + { + "epoch": 1.2783046095489403, + "grad_norm": 2.127985715866089, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.71306312084198, + "num_tokens": 356727081.0, + "step": 15502 + }, + { + "epoch": 1.278387070173992, + "grad_norm": 2.0486104488372803, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7704782485961914, + "num_tokens": 356754955.0, + "step": 15503 + }, + { + "epoch": 1.2784695307990435, + "grad_norm": 2.039926290512085, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7448956966400146, + "num_tokens": 356782996.0, + "step": 15504 + }, + { + "epoch": 1.278551991424095, + "grad_norm": 2.0102131366729736, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6827489137649536, + "num_tokens": 356812988.0, + "step": 15505 + }, + { + "epoch": 1.2786344520491466, + "grad_norm": 1.9866985082626343, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7859152555465698, + "num_tokens": 356841693.0, + "step": 15506 + }, + { + "epoch": 1.2787169126741982, + "grad_norm": 1.919190526008606, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7653309106826782, + "num_tokens": 356871421.0, + "step": 15507 + }, + { + "epoch": 1.2787993732992495, + "grad_norm": 2.1835975646972656, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.7799699306488037, + "num_tokens": 356897373.0, + "step": 15508 + }, + { + "epoch": 1.278881833924301, + "grad_norm": 2.0156502723693848, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7219749093055725, + "num_tokens": 356929679.0, + "step": 15509 + }, + { + "epoch": 1.2789642945493527, + "grad_norm": 2.1725547313690186, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7471610307693481, + "num_tokens": 356953510.0, + "step": 15510 + }, + { + "epoch": 1.2790467551744042, + "grad_norm": 1.7807284593582153, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7771196365356445, + "num_tokens": 356988575.0, + "step": 15511 + }, + { + "epoch": 1.2791292157994558, + "grad_norm": 2.0704245567321777, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.753054678440094, + "num_tokens": 357018830.0, + "step": 15512 + }, + { + "epoch": 1.2792116764245072, + "grad_norm": 1.844380497932434, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7683557271957397, + "num_tokens": 357051816.0, + "step": 15513 + }, + { + "epoch": 1.279294137049559, + "grad_norm": 2.1429696083068848, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.758594274520874, + "num_tokens": 357077154.0, + "step": 15514 + }, + { + "epoch": 1.2793765976746103, + "grad_norm": 1.9619935750961304, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.736130952835083, + "num_tokens": 357108787.0, + "step": 15515 + }, + { + "epoch": 1.2794590582996619, + "grad_norm": 1.7746977806091309, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.7889418601989746, + "num_tokens": 357143167.0, + "step": 15516 + }, + { + "epoch": 1.2795415189247135, + "grad_norm": 1.7349084615707397, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7830973863601685, + "num_tokens": 357178105.0, + "step": 15517 + }, + { + "epoch": 1.279623979549765, + "grad_norm": 2.1636533737182617, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7241462469100952, + "num_tokens": 357204626.0, + "step": 15518 + }, + { + "epoch": 1.2797064401748166, + "grad_norm": 1.8796133995056152, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7456478476524353, + "num_tokens": 357237389.0, + "step": 15519 + }, + { + "epoch": 1.279788900799868, + "grad_norm": 1.8991731405258179, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7682746648788452, + "num_tokens": 357269812.0, + "step": 15520 + }, + { + "epoch": 1.2798713614249195, + "grad_norm": 2.113842010498047, + "learning_rate": 1e-06, + "loss": 0.6118, + "mean_token_accuracy": 0.7827062606811523, + "num_tokens": 357294136.0, + "step": 15521 + }, + { + "epoch": 1.279953822049971, + "grad_norm": 1.9922032356262207, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7312005758285522, + "num_tokens": 357327363.0, + "step": 15522 + }, + { + "epoch": 1.2800362826750227, + "grad_norm": 1.849995493888855, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7319369912147522, + "num_tokens": 357361861.0, + "step": 15523 + }, + { + "epoch": 1.2801187433000742, + "grad_norm": 2.222433567047119, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7136727571487427, + "num_tokens": 357390637.0, + "step": 15524 + }, + { + "epoch": 1.2802012039251258, + "grad_norm": 1.76923668384552, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.7671043872833252, + "num_tokens": 357426134.0, + "step": 15525 + }, + { + "epoch": 1.2802836645501774, + "grad_norm": 2.06108021736145, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7643988728523254, + "num_tokens": 357456160.0, + "step": 15526 + }, + { + "epoch": 1.2803661251752287, + "grad_norm": 1.9054286479949951, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7723813652992249, + "num_tokens": 357486460.0, + "step": 15527 + }, + { + "epoch": 1.2804485858002803, + "grad_norm": 2.1330726146698, + "learning_rate": 1e-06, + "loss": 0.784, + "mean_token_accuracy": 0.7530086636543274, + "num_tokens": 357512998.0, + "step": 15528 + }, + { + "epoch": 1.2805310464253319, + "grad_norm": 2.1119024753570557, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7732751369476318, + "num_tokens": 357543300.0, + "step": 15529 + }, + { + "epoch": 1.2806135070503835, + "grad_norm": 1.9728704690933228, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7756292223930359, + "num_tokens": 357572775.0, + "step": 15530 + }, + { + "epoch": 1.280695967675435, + "grad_norm": 2.262406826019287, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7745745778083801, + "num_tokens": 357599026.0, + "step": 15531 + }, + { + "epoch": 1.2807784283004864, + "grad_norm": 1.8498839139938354, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7199419736862183, + "num_tokens": 357635919.0, + "step": 15532 + }, + { + "epoch": 1.2808608889255382, + "grad_norm": 1.8653802871704102, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7302709817886353, + "num_tokens": 357673453.0, + "step": 15533 + }, + { + "epoch": 1.2809433495505895, + "grad_norm": 1.9063283205032349, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7867692708969116, + "num_tokens": 357703183.0, + "step": 15534 + }, + { + "epoch": 1.281025810175641, + "grad_norm": 2.147509813308716, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7644432783126831, + "num_tokens": 357729674.0, + "step": 15535 + }, + { + "epoch": 1.2811082708006927, + "grad_norm": 2.207749366760254, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.762062668800354, + "num_tokens": 357757020.0, + "step": 15536 + }, + { + "epoch": 1.2811907314257442, + "grad_norm": 1.97740638256073, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.797279953956604, + "num_tokens": 357786404.0, + "step": 15537 + }, + { + "epoch": 1.2812731920507958, + "grad_norm": 2.3275089263916016, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7729865312576294, + "num_tokens": 357808494.0, + "step": 15538 + }, + { + "epoch": 1.2813556526758472, + "grad_norm": 2.2530086040496826, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7440121173858643, + "num_tokens": 357837707.0, + "step": 15539 + }, + { + "epoch": 1.2814381133008987, + "grad_norm": 1.8675271272659302, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7816102504730225, + "num_tokens": 357869098.0, + "step": 15540 + }, + { + "epoch": 1.2815205739259503, + "grad_norm": 2.1218175888061523, + "learning_rate": 1e-06, + "loss": 0.6088, + "mean_token_accuracy": 0.7907629609107971, + "num_tokens": 357895520.0, + "step": 15541 + }, + { + "epoch": 1.2816030345510019, + "grad_norm": 2.106870412826538, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7627239227294922, + "num_tokens": 357923013.0, + "step": 15542 + }, + { + "epoch": 1.2816854951760535, + "grad_norm": 1.8968380689620972, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7671458721160889, + "num_tokens": 357955279.0, + "step": 15543 + }, + { + "epoch": 1.281767955801105, + "grad_norm": 1.9108771085739136, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7766333222389221, + "num_tokens": 357990138.0, + "step": 15544 + }, + { + "epoch": 1.2818504164261566, + "grad_norm": 1.9319572448730469, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.7902494072914124, + "num_tokens": 358019231.0, + "step": 15545 + }, + { + "epoch": 1.281932877051208, + "grad_norm": 2.3984410762786865, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7538160681724548, + "num_tokens": 358042072.0, + "step": 15546 + }, + { + "epoch": 1.2820153376762595, + "grad_norm": 2.184656858444214, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7491456866264343, + "num_tokens": 358067526.0, + "step": 15547 + }, + { + "epoch": 1.282097798301311, + "grad_norm": 2.1557185649871826, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7514110207557678, + "num_tokens": 358093536.0, + "step": 15548 + }, + { + "epoch": 1.2821802589263627, + "grad_norm": 2.1702072620391846, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7451575994491577, + "num_tokens": 358118582.0, + "step": 15549 + }, + { + "epoch": 1.2822627195514142, + "grad_norm": 1.9674795866012573, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7593749761581421, + "num_tokens": 358146818.0, + "step": 15550 + }, + { + "epoch": 1.2823451801764658, + "grad_norm": 2.0380611419677734, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7579445838928223, + "num_tokens": 358176175.0, + "step": 15551 + }, + { + "epoch": 1.2824276408015174, + "grad_norm": 1.954817295074463, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7590362429618835, + "num_tokens": 358206639.0, + "step": 15552 + }, + { + "epoch": 1.2825101014265687, + "grad_norm": 2.1342122554779053, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7702232599258423, + "num_tokens": 358233752.0, + "step": 15553 + }, + { + "epoch": 1.2825925620516203, + "grad_norm": 2.0282275676727295, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7302380800247192, + "num_tokens": 358265825.0, + "step": 15554 + }, + { + "epoch": 1.2826750226766719, + "grad_norm": 1.9940202236175537, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.7577265501022339, + "num_tokens": 358294958.0, + "step": 15555 + }, + { + "epoch": 1.2827574833017235, + "grad_norm": 2.061363935470581, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.754896879196167, + "num_tokens": 358321377.0, + "step": 15556 + }, + { + "epoch": 1.282839943926775, + "grad_norm": 1.9118008613586426, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7627548575401306, + "num_tokens": 358355493.0, + "step": 15557 + }, + { + "epoch": 1.2829224045518264, + "grad_norm": 2.03544545173645, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7598099708557129, + "num_tokens": 358386348.0, + "step": 15558 + }, + { + "epoch": 1.2830048651768782, + "grad_norm": 2.096398115158081, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7655088901519775, + "num_tokens": 358414939.0, + "step": 15559 + }, + { + "epoch": 1.2830873258019295, + "grad_norm": 1.8640565872192383, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7689969539642334, + "num_tokens": 358452088.0, + "step": 15560 + }, + { + "epoch": 1.283169786426981, + "grad_norm": 2.0313448905944824, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7326304316520691, + "num_tokens": 358484859.0, + "step": 15561 + }, + { + "epoch": 1.2832522470520327, + "grad_norm": 1.8799277544021606, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.731300413608551, + "num_tokens": 358516543.0, + "step": 15562 + }, + { + "epoch": 1.2833347076770842, + "grad_norm": 1.7825113534927368, + "learning_rate": 1e-06, + "loss": 0.5337, + "mean_token_accuracy": 0.8140732645988464, + "num_tokens": 358549695.0, + "step": 15563 + }, + { + "epoch": 1.2834171683021358, + "grad_norm": 2.049269199371338, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7448722124099731, + "num_tokens": 358580986.0, + "step": 15564 + }, + { + "epoch": 1.2834996289271872, + "grad_norm": 2.073552370071411, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.7929288744926453, + "num_tokens": 358608459.0, + "step": 15565 + }, + { + "epoch": 1.2835820895522387, + "grad_norm": 2.1904377937316895, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7590401768684387, + "num_tokens": 358637550.0, + "step": 15566 + }, + { + "epoch": 1.2836645501772903, + "grad_norm": 1.8537328243255615, + "learning_rate": 1e-06, + "loss": 0.5792, + "mean_token_accuracy": 0.8092799186706543, + "num_tokens": 358668377.0, + "step": 15567 + }, + { + "epoch": 1.2837470108023419, + "grad_norm": 1.8427207469940186, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.792460560798645, + "num_tokens": 358701279.0, + "step": 15568 + }, + { + "epoch": 1.2838294714273935, + "grad_norm": 1.7844550609588623, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.7851201295852661, + "num_tokens": 358737567.0, + "step": 15569 + }, + { + "epoch": 1.283911932052445, + "grad_norm": 2.0166783332824707, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7871439456939697, + "num_tokens": 358767293.0, + "step": 15570 + }, + { + "epoch": 1.2839943926774966, + "grad_norm": 2.1594512462615967, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7516080141067505, + "num_tokens": 358795971.0, + "step": 15571 + }, + { + "epoch": 1.284076853302548, + "grad_norm": 1.9470092058181763, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7625632882118225, + "num_tokens": 358826999.0, + "step": 15572 + }, + { + "epoch": 1.2841593139275995, + "grad_norm": 2.0853354930877686, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.7958002090454102, + "num_tokens": 358855538.0, + "step": 15573 + }, + { + "epoch": 1.284241774552651, + "grad_norm": 2.3310258388519287, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7704504132270813, + "num_tokens": 358881205.0, + "step": 15574 + }, + { + "epoch": 1.2843242351777027, + "grad_norm": 1.9518837928771973, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7719427943229675, + "num_tokens": 358913682.0, + "step": 15575 + }, + { + "epoch": 1.2844066958027542, + "grad_norm": 1.948817491531372, + "learning_rate": 1e-06, + "loss": 0.6038, + "mean_token_accuracy": 0.7961282134056091, + "num_tokens": 358942555.0, + "step": 15576 + }, + { + "epoch": 1.2844891564278056, + "grad_norm": 2.179473400115967, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7752755284309387, + "num_tokens": 358967968.0, + "step": 15577 + }, + { + "epoch": 1.2845716170528574, + "grad_norm": 1.8726038932800293, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.7426304221153259, + "num_tokens": 359003601.0, + "step": 15578 + }, + { + "epoch": 1.2846540776779087, + "grad_norm": 1.7570737600326538, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.7959306240081787, + "num_tokens": 359035571.0, + "step": 15579 + }, + { + "epoch": 1.2847365383029603, + "grad_norm": 1.9483586549758911, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7568897604942322, + "num_tokens": 359069071.0, + "step": 15580 + }, + { + "epoch": 1.2848189989280119, + "grad_norm": 2.072598695755005, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7676517963409424, + "num_tokens": 359097939.0, + "step": 15581 + }, + { + "epoch": 1.2849014595530635, + "grad_norm": 1.8881899118423462, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.762714147567749, + "num_tokens": 359129045.0, + "step": 15582 + }, + { + "epoch": 1.284983920178115, + "grad_norm": 1.9154549837112427, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7250992059707642, + "num_tokens": 359163219.0, + "step": 15583 + }, + { + "epoch": 1.2850663808031664, + "grad_norm": 1.8710113763809204, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7521220445632935, + "num_tokens": 359195094.0, + "step": 15584 + }, + { + "epoch": 1.285148841428218, + "grad_norm": 2.06927227973938, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7828495502471924, + "num_tokens": 359224246.0, + "step": 15585 + }, + { + "epoch": 1.2852313020532695, + "grad_norm": 2.255791664123535, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7806264162063599, + "num_tokens": 359251103.0, + "step": 15586 + }, + { + "epoch": 1.285313762678321, + "grad_norm": 1.9092704057693481, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7732031345367432, + "num_tokens": 359284458.0, + "step": 15587 + }, + { + "epoch": 1.2853962233033727, + "grad_norm": 2.3205363750457764, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7186193466186523, + "num_tokens": 359309319.0, + "step": 15588 + }, + { + "epoch": 1.2854786839284242, + "grad_norm": 2.0016067028045654, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7631141543388367, + "num_tokens": 359339085.0, + "step": 15589 + }, + { + "epoch": 1.2855611445534758, + "grad_norm": 1.7997462749481201, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7808465361595154, + "num_tokens": 359373027.0, + "step": 15590 + }, + { + "epoch": 1.2856436051785272, + "grad_norm": 1.8986228704452515, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7460102438926697, + "num_tokens": 359406537.0, + "step": 15591 + }, + { + "epoch": 1.2857260658035787, + "grad_norm": 1.9066414833068848, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7615426778793335, + "num_tokens": 359436368.0, + "step": 15592 + }, + { + "epoch": 1.2858085264286303, + "grad_norm": 1.9393415451049805, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7406325340270996, + "num_tokens": 359469730.0, + "step": 15593 + }, + { + "epoch": 1.2858909870536819, + "grad_norm": 2.0719473361968994, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7259708046913147, + "num_tokens": 359499331.0, + "step": 15594 + }, + { + "epoch": 1.2859734476787335, + "grad_norm": 1.9779514074325562, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.752782940864563, + "num_tokens": 359530338.0, + "step": 15595 + }, + { + "epoch": 1.286055908303785, + "grad_norm": 2.0957443714141846, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7464644908905029, + "num_tokens": 359557608.0, + "step": 15596 + }, + { + "epoch": 1.2861383689288366, + "grad_norm": 2.31897234916687, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7290251851081848, + "num_tokens": 359582934.0, + "step": 15597 + }, + { + "epoch": 1.286220829553888, + "grad_norm": 2.0073909759521484, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.7769140005111694, + "num_tokens": 359611524.0, + "step": 15598 + }, + { + "epoch": 1.2863032901789395, + "grad_norm": 1.8489718437194824, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7662274837493896, + "num_tokens": 359648154.0, + "step": 15599 + }, + { + "epoch": 1.286385750803991, + "grad_norm": 1.9399629831314087, + "learning_rate": 1e-06, + "loss": 0.5357, + "mean_token_accuracy": 0.8216906785964966, + "num_tokens": 359680196.0, + "step": 15600 + }, + { + "epoch": 1.2864682114290427, + "grad_norm": 1.8993420600891113, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7537903785705566, + "num_tokens": 359713760.0, + "step": 15601 + }, + { + "epoch": 1.2865506720540942, + "grad_norm": 1.9523382186889648, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7248530387878418, + "num_tokens": 359746915.0, + "step": 15602 + }, + { + "epoch": 1.2866331326791456, + "grad_norm": 2.0730035305023193, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7403749227523804, + "num_tokens": 359776660.0, + "step": 15603 + }, + { + "epoch": 1.2867155933041974, + "grad_norm": 1.7666985988616943, + "learning_rate": 1e-06, + "loss": 0.5583, + "mean_token_accuracy": 0.8163800835609436, + "num_tokens": 359810108.0, + "step": 15604 + }, + { + "epoch": 1.2867980539292487, + "grad_norm": 1.814520239830017, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7430272102355957, + "num_tokens": 359845654.0, + "step": 15605 + }, + { + "epoch": 1.2868805145543003, + "grad_norm": 2.1518213748931885, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7366999983787537, + "num_tokens": 359873293.0, + "step": 15606 + }, + { + "epoch": 1.2869629751793519, + "grad_norm": 2.0163135528564453, + "learning_rate": 1e-06, + "loss": 0.6263, + "mean_token_accuracy": 0.795642077922821, + "num_tokens": 359903204.0, + "step": 15607 + }, + { + "epoch": 1.2870454358044034, + "grad_norm": 1.8448680639266968, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.8056291341781616, + "num_tokens": 359934365.0, + "step": 15608 + }, + { + "epoch": 1.287127896429455, + "grad_norm": 2.2521286010742188, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7409424781799316, + "num_tokens": 359959979.0, + "step": 15609 + }, + { + "epoch": 1.2872103570545064, + "grad_norm": 2.3294904232025146, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.749624490737915, + "num_tokens": 359982216.0, + "step": 15610 + }, + { + "epoch": 1.287292817679558, + "grad_norm": 2.309809923171997, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7633693218231201, + "num_tokens": 360005804.0, + "step": 15611 + }, + { + "epoch": 1.2873752783046095, + "grad_norm": 2.3084425926208496, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7670021653175354, + "num_tokens": 360029948.0, + "step": 15612 + }, + { + "epoch": 1.287457738929661, + "grad_norm": 1.840361475944519, + "learning_rate": 1e-06, + "loss": 0.5994, + "mean_token_accuracy": 0.7997962832450867, + "num_tokens": 360062516.0, + "step": 15613 + }, + { + "epoch": 1.2875401995547127, + "grad_norm": 2.117161989212036, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7673649191856384, + "num_tokens": 360090744.0, + "step": 15614 + }, + { + "epoch": 1.2876226601797642, + "grad_norm": 1.9455552101135254, + "learning_rate": 1e-06, + "loss": 0.6091, + "mean_token_accuracy": 0.7995920777320862, + "num_tokens": 360119669.0, + "step": 15615 + }, + { + "epoch": 1.2877051208048158, + "grad_norm": 2.0183217525482178, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7715901136398315, + "num_tokens": 360149461.0, + "step": 15616 + }, + { + "epoch": 1.2877875814298672, + "grad_norm": 2.1070241928100586, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7245171070098877, + "num_tokens": 360177529.0, + "step": 15617 + }, + { + "epoch": 1.2878700420549187, + "grad_norm": 2.1387686729431152, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.757239818572998, + "num_tokens": 360206838.0, + "step": 15618 + }, + { + "epoch": 1.2879525026799703, + "grad_norm": 1.9786731004714966, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7606502175331116, + "num_tokens": 360237716.0, + "step": 15619 + }, + { + "epoch": 1.2880349633050219, + "grad_norm": 2.045287609100342, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7336206436157227, + "num_tokens": 360267751.0, + "step": 15620 + }, + { + "epoch": 1.2881174239300734, + "grad_norm": 2.164425849914551, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6939905881881714, + "num_tokens": 360298901.0, + "step": 15621 + }, + { + "epoch": 1.2881998845551248, + "grad_norm": 1.9156337976455688, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7699554562568665, + "num_tokens": 360333405.0, + "step": 15622 + }, + { + "epoch": 1.2882823451801766, + "grad_norm": 2.007991313934326, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7701939344406128, + "num_tokens": 360362393.0, + "step": 15623 + }, + { + "epoch": 1.288364805805228, + "grad_norm": 2.0123379230499268, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7420125603675842, + "num_tokens": 360392882.0, + "step": 15624 + }, + { + "epoch": 1.2884472664302795, + "grad_norm": 2.0801308155059814, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7235825061798096, + "num_tokens": 360419666.0, + "step": 15625 + }, + { + "epoch": 1.288529727055331, + "grad_norm": 2.163238286972046, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7303251028060913, + "num_tokens": 360445950.0, + "step": 15626 + }, + { + "epoch": 1.2886121876803827, + "grad_norm": 2.021667242050171, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7688381671905518, + "num_tokens": 360473849.0, + "step": 15627 + }, + { + "epoch": 1.2886946483054342, + "grad_norm": 1.8047659397125244, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.7838071584701538, + "num_tokens": 360510981.0, + "step": 15628 + }, + { + "epoch": 1.2887771089304856, + "grad_norm": 1.995326042175293, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7705762386322021, + "num_tokens": 360539700.0, + "step": 15629 + }, + { + "epoch": 1.2888595695555372, + "grad_norm": 1.6667813062667847, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7578933238983154, + "num_tokens": 360576484.0, + "step": 15630 + }, + { + "epoch": 1.2889420301805887, + "grad_norm": 2.4581849575042725, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7457008361816406, + "num_tokens": 360597589.0, + "step": 15631 + }, + { + "epoch": 1.2890244908056403, + "grad_norm": 1.8890738487243652, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7602275609970093, + "num_tokens": 360631129.0, + "step": 15632 + }, + { + "epoch": 1.2891069514306919, + "grad_norm": 2.020777702331543, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7660111784934998, + "num_tokens": 360658861.0, + "step": 15633 + }, + { + "epoch": 1.2891894120557434, + "grad_norm": 1.9861600399017334, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7818474769592285, + "num_tokens": 360687546.0, + "step": 15634 + }, + { + "epoch": 1.289271872680795, + "grad_norm": 1.9153252840042114, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7803729772567749, + "num_tokens": 360719052.0, + "step": 15635 + }, + { + "epoch": 1.2893543333058464, + "grad_norm": 1.9453786611557007, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7645409107208252, + "num_tokens": 360748601.0, + "step": 15636 + }, + { + "epoch": 1.289436793930898, + "grad_norm": 1.8688945770263672, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7740393877029419, + "num_tokens": 360782190.0, + "step": 15637 + }, + { + "epoch": 1.2895192545559495, + "grad_norm": 2.115116596221924, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7533617615699768, + "num_tokens": 360810086.0, + "step": 15638 + }, + { + "epoch": 1.289601715181001, + "grad_norm": 2.1383180618286133, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7636846303939819, + "num_tokens": 360833946.0, + "step": 15639 + }, + { + "epoch": 1.2896841758060527, + "grad_norm": 1.9226737022399902, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.8013726472854614, + "num_tokens": 360864254.0, + "step": 15640 + }, + { + "epoch": 1.2897666364311042, + "grad_norm": 2.133104085922241, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7558961510658264, + "num_tokens": 360890045.0, + "step": 15641 + }, + { + "epoch": 1.2898490970561558, + "grad_norm": 1.9129246473312378, + "learning_rate": 1e-06, + "loss": 0.5099, + "mean_token_accuracy": 0.8191869258880615, + "num_tokens": 360916089.0, + "step": 15642 + }, + { + "epoch": 1.2899315576812072, + "grad_norm": 1.946908950805664, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7849119901657104, + "num_tokens": 360945547.0, + "step": 15643 + }, + { + "epoch": 1.2900140183062587, + "grad_norm": 2.2128775119781494, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.7515865564346313, + "num_tokens": 360970610.0, + "step": 15644 + }, + { + "epoch": 1.2900964789313103, + "grad_norm": 2.055903673171997, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7656741142272949, + "num_tokens": 360999990.0, + "step": 15645 + }, + { + "epoch": 1.2901789395563619, + "grad_norm": 2.1082026958465576, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.7562106847763062, + "num_tokens": 361027143.0, + "step": 15646 + }, + { + "epoch": 1.2902614001814134, + "grad_norm": 1.8454850912094116, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.7973767518997192, + "num_tokens": 361061774.0, + "step": 15647 + }, + { + "epoch": 1.2903438608064648, + "grad_norm": 1.8986353874206543, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7702932357788086, + "num_tokens": 361094559.0, + "step": 15648 + }, + { + "epoch": 1.2904263214315166, + "grad_norm": 2.2499096393585205, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7337778806686401, + "num_tokens": 361119775.0, + "step": 15649 + }, + { + "epoch": 1.290508782056568, + "grad_norm": 1.9508845806121826, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.768216609954834, + "num_tokens": 361150334.0, + "step": 15650 + }, + { + "epoch": 1.2905912426816195, + "grad_norm": 1.9478678703308105, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.7856071591377258, + "num_tokens": 361181626.0, + "step": 15651 + }, + { + "epoch": 1.290673703306671, + "grad_norm": 2.031264543533325, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7529405951499939, + "num_tokens": 361211394.0, + "step": 15652 + }, + { + "epoch": 1.2907561639317227, + "grad_norm": 2.3049607276916504, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7259316444396973, + "num_tokens": 361234212.0, + "step": 15653 + }, + { + "epoch": 1.2908386245567742, + "grad_norm": 1.9657254219055176, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7569549083709717, + "num_tokens": 361266941.0, + "step": 15654 + }, + { + "epoch": 1.2909210851818256, + "grad_norm": 2.3819339275360107, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7575325965881348, + "num_tokens": 361292147.0, + "step": 15655 + }, + { + "epoch": 1.2910035458068772, + "grad_norm": 2.1910486221313477, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7433037161827087, + "num_tokens": 361319700.0, + "step": 15656 + }, + { + "epoch": 1.2910860064319287, + "grad_norm": 1.9793622493743896, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7813116312026978, + "num_tokens": 361349023.0, + "step": 15657 + }, + { + "epoch": 1.2911684670569803, + "grad_norm": 2.288386821746826, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7631834745407104, + "num_tokens": 361375949.0, + "step": 15658 + }, + { + "epoch": 1.2912509276820319, + "grad_norm": 1.877583622932434, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.7892012000083923, + "num_tokens": 361408254.0, + "step": 15659 + }, + { + "epoch": 1.2913333883070834, + "grad_norm": 1.9699708223342896, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7731461524963379, + "num_tokens": 361436474.0, + "step": 15660 + }, + { + "epoch": 1.291415848932135, + "grad_norm": 2.1341381072998047, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7526237964630127, + "num_tokens": 361466591.0, + "step": 15661 + }, + { + "epoch": 1.2914983095571864, + "grad_norm": 2.1126396656036377, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7546293139457703, + "num_tokens": 361497417.0, + "step": 15662 + }, + { + "epoch": 1.291580770182238, + "grad_norm": 1.9134223461151123, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7580468654632568, + "num_tokens": 361533474.0, + "step": 15663 + }, + { + "epoch": 1.2916632308072895, + "grad_norm": 2.325482130050659, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.739181399345398, + "num_tokens": 361558381.0, + "step": 15664 + }, + { + "epoch": 1.291745691432341, + "grad_norm": 2.103280544281006, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7527645826339722, + "num_tokens": 361588125.0, + "step": 15665 + }, + { + "epoch": 1.2918281520573927, + "grad_norm": 1.7624757289886475, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7608776092529297, + "num_tokens": 361627773.0, + "step": 15666 + }, + { + "epoch": 1.291910612682444, + "grad_norm": 1.9235949516296387, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7475296258926392, + "num_tokens": 361659926.0, + "step": 15667 + }, + { + "epoch": 1.2919930733074958, + "grad_norm": 2.107150077819824, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7255781888961792, + "num_tokens": 361689502.0, + "step": 15668 + }, + { + "epoch": 1.2920755339325471, + "grad_norm": 2.1180896759033203, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7722162008285522, + "num_tokens": 361716155.0, + "step": 15669 + }, + { + "epoch": 1.2921579945575987, + "grad_norm": 1.9585797786712646, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7475031614303589, + "num_tokens": 361749297.0, + "step": 15670 + }, + { + "epoch": 1.2922404551826503, + "grad_norm": 2.076418399810791, + "learning_rate": 1e-06, + "loss": 0.7738, + "mean_token_accuracy": 0.7519234418869019, + "num_tokens": 361779147.0, + "step": 15671 + }, + { + "epoch": 1.2923229158077019, + "grad_norm": 1.832846760749817, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7426053881645203, + "num_tokens": 361819855.0, + "step": 15672 + }, + { + "epoch": 1.2924053764327534, + "grad_norm": 1.9158523082733154, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7278731465339661, + "num_tokens": 361856265.0, + "step": 15673 + }, + { + "epoch": 1.2924878370578048, + "grad_norm": 2.0555031299591064, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7649131417274475, + "num_tokens": 361885843.0, + "step": 15674 + }, + { + "epoch": 1.2925702976828564, + "grad_norm": 1.7567546367645264, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.750930905342102, + "num_tokens": 361923321.0, + "step": 15675 + }, + { + "epoch": 1.292652758307908, + "grad_norm": 2.1070919036865234, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7204848527908325, + "num_tokens": 361954433.0, + "step": 15676 + }, + { + "epoch": 1.2927352189329595, + "grad_norm": 2.0759027004241943, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7542943954467773, + "num_tokens": 361981555.0, + "step": 15677 + }, + { + "epoch": 1.292817679558011, + "grad_norm": 2.019995927810669, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7358465194702148, + "num_tokens": 362011121.0, + "step": 15678 + }, + { + "epoch": 1.2929001401830627, + "grad_norm": 2.1537482738494873, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7500905990600586, + "num_tokens": 362041678.0, + "step": 15679 + }, + { + "epoch": 1.2929826008081142, + "grad_norm": 2.1819849014282227, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7453137636184692, + "num_tokens": 362068666.0, + "step": 15680 + }, + { + "epoch": 1.2930650614331656, + "grad_norm": 2.1462700366973877, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7270952463150024, + "num_tokens": 362096335.0, + "step": 15681 + }, + { + "epoch": 1.2931475220582171, + "grad_norm": 2.0524396896362305, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7450887560844421, + "num_tokens": 362126184.0, + "step": 15682 + }, + { + "epoch": 1.2932299826832687, + "grad_norm": 1.9340640306472778, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7451804876327515, + "num_tokens": 362161751.0, + "step": 15683 + }, + { + "epoch": 1.2933124433083203, + "grad_norm": 2.3065648078918457, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.761053740978241, + "num_tokens": 362184057.0, + "step": 15684 + }, + { + "epoch": 1.2933949039333719, + "grad_norm": 2.1303014755249023, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7668322324752808, + "num_tokens": 362212488.0, + "step": 15685 + }, + { + "epoch": 1.2934773645584232, + "grad_norm": 2.4234373569488525, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7427166700363159, + "num_tokens": 362237289.0, + "step": 15686 + }, + { + "epoch": 1.293559825183475, + "grad_norm": 1.9266102313995361, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7845510244369507, + "num_tokens": 362267365.0, + "step": 15687 + }, + { + "epoch": 1.2936422858085264, + "grad_norm": 2.156930685043335, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7921788096427917, + "num_tokens": 362290670.0, + "step": 15688 + }, + { + "epoch": 1.293724746433578, + "grad_norm": 2.0277328491210938, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7638818025588989, + "num_tokens": 362318975.0, + "step": 15689 + }, + { + "epoch": 1.2938072070586295, + "grad_norm": 1.9021809101104736, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7889187335968018, + "num_tokens": 362349883.0, + "step": 15690 + }, + { + "epoch": 1.293889667683681, + "grad_norm": 2.0751917362213135, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7704074382781982, + "num_tokens": 362376582.0, + "step": 15691 + }, + { + "epoch": 1.2939721283087326, + "grad_norm": 2.334383010864258, + "learning_rate": 1e-06, + "loss": 0.7465, + "mean_token_accuracy": 0.7527498006820679, + "num_tokens": 362401233.0, + "step": 15692 + }, + { + "epoch": 1.294054588933784, + "grad_norm": 2.0590474605560303, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7293040752410889, + "num_tokens": 362430613.0, + "step": 15693 + }, + { + "epoch": 1.2941370495588358, + "grad_norm": 1.993872880935669, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7438417673110962, + "num_tokens": 362461574.0, + "step": 15694 + }, + { + "epoch": 1.2942195101838871, + "grad_norm": 2.0242538452148438, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7661607265472412, + "num_tokens": 362492010.0, + "step": 15695 + }, + { + "epoch": 1.2943019708089387, + "grad_norm": 1.9139658212661743, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7688166499137878, + "num_tokens": 362524005.0, + "step": 15696 + }, + { + "epoch": 1.2943844314339903, + "grad_norm": 2.041171073913574, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7310101389884949, + "num_tokens": 362550794.0, + "step": 15697 + }, + { + "epoch": 1.2944668920590419, + "grad_norm": 1.9779033660888672, + "learning_rate": 1e-06, + "loss": 0.6165, + "mean_token_accuracy": 0.7974309921264648, + "num_tokens": 362582364.0, + "step": 15698 + }, + { + "epoch": 1.2945493526840934, + "grad_norm": 1.8489954471588135, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7278355360031128, + "num_tokens": 362618506.0, + "step": 15699 + }, + { + "epoch": 1.2946318133091448, + "grad_norm": 1.834054708480835, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7528407573699951, + "num_tokens": 362652936.0, + "step": 15700 + }, + { + "epoch": 1.2947142739341964, + "grad_norm": 1.8407517671585083, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7377870678901672, + "num_tokens": 362689516.0, + "step": 15701 + }, + { + "epoch": 1.294796734559248, + "grad_norm": 1.9120975732803345, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7839803099632263, + "num_tokens": 362719030.0, + "step": 15702 + }, + { + "epoch": 1.2948791951842995, + "grad_norm": 1.9942824840545654, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7683055400848389, + "num_tokens": 362749313.0, + "step": 15703 + }, + { + "epoch": 1.294961655809351, + "grad_norm": 2.1248998641967773, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.7890572547912598, + "num_tokens": 362775216.0, + "step": 15704 + }, + { + "epoch": 1.2950441164344026, + "grad_norm": 2.0968050956726074, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7619009613990784, + "num_tokens": 362805870.0, + "step": 15705 + }, + { + "epoch": 1.2951265770594542, + "grad_norm": 2.023993730545044, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.786771833896637, + "num_tokens": 362835711.0, + "step": 15706 + }, + { + "epoch": 1.2952090376845056, + "grad_norm": 2.3387982845306396, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7420123219490051, + "num_tokens": 362859118.0, + "step": 15707 + }, + { + "epoch": 1.2952914983095571, + "grad_norm": 2.223402738571167, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7246053218841553, + "num_tokens": 362890763.0, + "step": 15708 + }, + { + "epoch": 1.2953739589346087, + "grad_norm": 1.8242378234863281, + "learning_rate": 1e-06, + "loss": 0.6226, + "mean_token_accuracy": 0.7978172302246094, + "num_tokens": 362922728.0, + "step": 15709 + }, + { + "epoch": 1.2954564195596603, + "grad_norm": 1.7611193656921387, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7599697113037109, + "num_tokens": 362957507.0, + "step": 15710 + }, + { + "epoch": 1.2955388801847119, + "grad_norm": 1.8419368267059326, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.7959821224212646, + "num_tokens": 362987611.0, + "step": 15711 + }, + { + "epoch": 1.2956213408097632, + "grad_norm": 1.948259711265564, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7377524971961975, + "num_tokens": 363020934.0, + "step": 15712 + }, + { + "epoch": 1.295703801434815, + "grad_norm": 1.7860958576202393, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.7835974097251892, + "num_tokens": 363057093.0, + "step": 15713 + }, + { + "epoch": 1.2957862620598664, + "grad_norm": 2.252354860305786, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7390623688697815, + "num_tokens": 363082539.0, + "step": 15714 + }, + { + "epoch": 1.295868722684918, + "grad_norm": 2.03790283203125, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.7793071269989014, + "num_tokens": 363113206.0, + "step": 15715 + }, + { + "epoch": 1.2959511833099695, + "grad_norm": 2.177201747894287, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7734612226486206, + "num_tokens": 363136902.0, + "step": 15716 + }, + { + "epoch": 1.296033643935021, + "grad_norm": 1.815097689628601, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7386063933372498, + "num_tokens": 363172576.0, + "step": 15717 + }, + { + "epoch": 1.2961161045600726, + "grad_norm": 1.870271921157837, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.7941855192184448, + "num_tokens": 363205191.0, + "step": 15718 + }, + { + "epoch": 1.296198565185124, + "grad_norm": 1.9859740734100342, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.7555582523345947, + "num_tokens": 363238683.0, + "step": 15719 + }, + { + "epoch": 1.2962810258101756, + "grad_norm": 1.6777716875076294, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.7977874875068665, + "num_tokens": 363275696.0, + "step": 15720 + }, + { + "epoch": 1.2963634864352271, + "grad_norm": 2.10646915435791, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.724618673324585, + "num_tokens": 363305573.0, + "step": 15721 + }, + { + "epoch": 1.2964459470602787, + "grad_norm": 1.9299468994140625, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7818446159362793, + "num_tokens": 363336585.0, + "step": 15722 + }, + { + "epoch": 1.2965284076853303, + "grad_norm": 1.8038910627365112, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7853614091873169, + "num_tokens": 363369912.0, + "step": 15723 + }, + { + "epoch": 1.2966108683103819, + "grad_norm": 1.9597337245941162, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7552720308303833, + "num_tokens": 363401352.0, + "step": 15724 + }, + { + "epoch": 1.2966933289354334, + "grad_norm": 2.1238009929656982, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.7464863657951355, + "num_tokens": 363428597.0, + "step": 15725 + }, + { + "epoch": 1.2967757895604848, + "grad_norm": 1.7438768148422241, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.7457422018051147, + "num_tokens": 363463107.0, + "step": 15726 + }, + { + "epoch": 1.2968582501855364, + "grad_norm": 2.139033079147339, + "learning_rate": 1e-06, + "loss": 0.6239, + "mean_token_accuracy": 0.7971265316009521, + "num_tokens": 363488629.0, + "step": 15727 + }, + { + "epoch": 1.296940710810588, + "grad_norm": 1.8669291734695435, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7660096287727356, + "num_tokens": 363521970.0, + "step": 15728 + }, + { + "epoch": 1.2970231714356395, + "grad_norm": 2.092034339904785, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7741277813911438, + "num_tokens": 363547681.0, + "step": 15729 + }, + { + "epoch": 1.297105632060691, + "grad_norm": 2.0122387409210205, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7796449065208435, + "num_tokens": 363577576.0, + "step": 15730 + }, + { + "epoch": 1.2971880926857424, + "grad_norm": 1.849867582321167, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7264524698257446, + "num_tokens": 363607791.0, + "step": 15731 + }, + { + "epoch": 1.2972705533107942, + "grad_norm": 2.0382962226867676, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7474596500396729, + "num_tokens": 363638390.0, + "step": 15732 + }, + { + "epoch": 1.2973530139358456, + "grad_norm": 2.019775152206421, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7323639392852783, + "num_tokens": 363669089.0, + "step": 15733 + }, + { + "epoch": 1.2974354745608971, + "grad_norm": 2.110722303390503, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7663431167602539, + "num_tokens": 363694722.0, + "step": 15734 + }, + { + "epoch": 1.2975179351859487, + "grad_norm": 2.1903865337371826, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7467764616012573, + "num_tokens": 363721465.0, + "step": 15735 + }, + { + "epoch": 1.2976003958110003, + "grad_norm": 2.057952880859375, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7298949360847473, + "num_tokens": 363750135.0, + "step": 15736 + }, + { + "epoch": 1.2976828564360519, + "grad_norm": 2.0433003902435303, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7218936681747437, + "num_tokens": 363780147.0, + "step": 15737 + }, + { + "epoch": 1.2977653170611032, + "grad_norm": 1.959852695465088, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7400305867195129, + "num_tokens": 363810981.0, + "step": 15738 + }, + { + "epoch": 1.2978477776861548, + "grad_norm": 1.711313009262085, + "learning_rate": 1e-06, + "loss": 0.5444, + "mean_token_accuracy": 0.8239312171936035, + "num_tokens": 363849401.0, + "step": 15739 + }, + { + "epoch": 1.2979302383112064, + "grad_norm": 1.9595955610275269, + "learning_rate": 1e-06, + "loss": 0.7698, + "mean_token_accuracy": 0.7657341361045837, + "num_tokens": 363881801.0, + "step": 15740 + }, + { + "epoch": 1.298012698936258, + "grad_norm": 2.0390639305114746, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7576751112937927, + "num_tokens": 363910017.0, + "step": 15741 + }, + { + "epoch": 1.2980951595613095, + "grad_norm": 1.99354887008667, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7556740045547485, + "num_tokens": 363938471.0, + "step": 15742 + }, + { + "epoch": 1.298177620186361, + "grad_norm": 2.07898211479187, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7592054009437561, + "num_tokens": 363966180.0, + "step": 15743 + }, + { + "epoch": 1.2982600808114126, + "grad_norm": 1.9220492839813232, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7706325650215149, + "num_tokens": 363996711.0, + "step": 15744 + }, + { + "epoch": 1.298342541436464, + "grad_norm": 2.013838529586792, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7679387331008911, + "num_tokens": 364026447.0, + "step": 15745 + }, + { + "epoch": 1.2984250020615156, + "grad_norm": 2.0443994998931885, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7525618076324463, + "num_tokens": 364055628.0, + "step": 15746 + }, + { + "epoch": 1.2985074626865671, + "grad_norm": 2.1707632541656494, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7608374953269958, + "num_tokens": 364081867.0, + "step": 15747 + }, + { + "epoch": 1.2985899233116187, + "grad_norm": 2.0568132400512695, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7396344542503357, + "num_tokens": 364111020.0, + "step": 15748 + }, + { + "epoch": 1.2986723839366703, + "grad_norm": 2.002887010574341, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7581022381782532, + "num_tokens": 364143662.0, + "step": 15749 + }, + { + "epoch": 1.2987548445617219, + "grad_norm": 2.068061351776123, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7517651319503784, + "num_tokens": 364170898.0, + "step": 15750 + }, + { + "epoch": 1.2988373051867734, + "grad_norm": 2.375655174255371, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7541625499725342, + "num_tokens": 364196758.0, + "step": 15751 + }, + { + "epoch": 1.2989197658118248, + "grad_norm": 2.16982102394104, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7398091554641724, + "num_tokens": 364225973.0, + "step": 15752 + }, + { + "epoch": 1.2990022264368764, + "grad_norm": 1.9568382501602173, + "learning_rate": 1e-06, + "loss": 0.5721, + "mean_token_accuracy": 0.8108656406402588, + "num_tokens": 364256040.0, + "step": 15753 + }, + { + "epoch": 1.299084687061928, + "grad_norm": 1.9477137327194214, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.773292064666748, + "num_tokens": 364286389.0, + "step": 15754 + }, + { + "epoch": 1.2991671476869795, + "grad_norm": 1.9009745121002197, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7510644197463989, + "num_tokens": 364320464.0, + "step": 15755 + }, + { + "epoch": 1.299249608312031, + "grad_norm": 2.2258474826812744, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7696692943572998, + "num_tokens": 364345178.0, + "step": 15756 + }, + { + "epoch": 1.2993320689370824, + "grad_norm": 1.9816755056381226, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7843921780586243, + "num_tokens": 364376470.0, + "step": 15757 + }, + { + "epoch": 1.2994145295621342, + "grad_norm": 1.8803658485412598, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7611438035964966, + "num_tokens": 364408515.0, + "step": 15758 + }, + { + "epoch": 1.2994969901871856, + "grad_norm": 1.9514490365982056, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7722911238670349, + "num_tokens": 364439138.0, + "step": 15759 + }, + { + "epoch": 1.2995794508122371, + "grad_norm": 2.2687876224517822, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.752116858959198, + "num_tokens": 364463162.0, + "step": 15760 + }, + { + "epoch": 1.2996619114372887, + "grad_norm": 1.7691518068313599, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7626520395278931, + "num_tokens": 364502264.0, + "step": 15761 + }, + { + "epoch": 1.2997443720623403, + "grad_norm": 1.9340131282806396, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.744223415851593, + "num_tokens": 364536721.0, + "step": 15762 + }, + { + "epoch": 1.2998268326873919, + "grad_norm": 2.0132758617401123, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.782381534576416, + "num_tokens": 364566714.0, + "step": 15763 + }, + { + "epoch": 1.2999092933124432, + "grad_norm": 2.0445187091827393, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7481409311294556, + "num_tokens": 364598587.0, + "step": 15764 + }, + { + "epoch": 1.2999917539374948, + "grad_norm": 2.0462353229522705, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.777851939201355, + "num_tokens": 364628052.0, + "step": 15765 + }, + { + "epoch": 1.3000742145625463, + "grad_norm": 2.186941623687744, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7647206783294678, + "num_tokens": 364655537.0, + "step": 15766 + }, + { + "epoch": 1.300156675187598, + "grad_norm": 2.2400033473968506, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7247146368026733, + "num_tokens": 364685731.0, + "step": 15767 + }, + { + "epoch": 1.3002391358126495, + "grad_norm": 1.7882115840911865, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7480517625808716, + "num_tokens": 364722657.0, + "step": 15768 + }, + { + "epoch": 1.300321596437701, + "grad_norm": 1.8806276321411133, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7516415119171143, + "num_tokens": 364754799.0, + "step": 15769 + }, + { + "epoch": 1.3004040570627526, + "grad_norm": 1.8362247943878174, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.790695309638977, + "num_tokens": 364786910.0, + "step": 15770 + }, + { + "epoch": 1.300486517687804, + "grad_norm": 2.0153989791870117, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7368497848510742, + "num_tokens": 364818025.0, + "step": 15771 + }, + { + "epoch": 1.3005689783128556, + "grad_norm": 1.9656671285629272, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.74906986951828, + "num_tokens": 364850729.0, + "step": 15772 + }, + { + "epoch": 1.3006514389379071, + "grad_norm": 2.0110719203948975, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7697145938873291, + "num_tokens": 364880954.0, + "step": 15773 + }, + { + "epoch": 1.3007338995629587, + "grad_norm": 2.051443338394165, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7538123726844788, + "num_tokens": 364908670.0, + "step": 15774 + }, + { + "epoch": 1.3008163601880103, + "grad_norm": 1.9045251607894897, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7853155732154846, + "num_tokens": 364939412.0, + "step": 15775 + }, + { + "epoch": 1.3008988208130616, + "grad_norm": 1.8454962968826294, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7768981456756592, + "num_tokens": 364975095.0, + "step": 15776 + }, + { + "epoch": 1.3009812814381134, + "grad_norm": 1.8352686166763306, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7734179496765137, + "num_tokens": 365008213.0, + "step": 15777 + }, + { + "epoch": 1.3010637420631648, + "grad_norm": 2.039849042892456, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7503340244293213, + "num_tokens": 365036034.0, + "step": 15778 + }, + { + "epoch": 1.3011462026882163, + "grad_norm": 2.00473952293396, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7725265622138977, + "num_tokens": 365066481.0, + "step": 15779 + }, + { + "epoch": 1.301228663313268, + "grad_norm": 1.8941349983215332, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7609168291091919, + "num_tokens": 365101917.0, + "step": 15780 + }, + { + "epoch": 1.3013111239383195, + "grad_norm": 2.082930326461792, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7238376140594482, + "num_tokens": 365134477.0, + "step": 15781 + }, + { + "epoch": 1.301393584563371, + "grad_norm": 1.9910677671432495, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7554558515548706, + "num_tokens": 365166337.0, + "step": 15782 + }, + { + "epoch": 1.3014760451884224, + "grad_norm": 2.303637742996216, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7414970993995667, + "num_tokens": 365191143.0, + "step": 15783 + }, + { + "epoch": 1.301558505813474, + "grad_norm": 1.922439455986023, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7635213732719421, + "num_tokens": 365219458.0, + "step": 15784 + }, + { + "epoch": 1.3016409664385256, + "grad_norm": 2.223123788833618, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7451648116111755, + "num_tokens": 365244811.0, + "step": 15785 + }, + { + "epoch": 1.3017234270635771, + "grad_norm": 2.1725637912750244, + "learning_rate": 1e-06, + "loss": 0.7877, + "mean_token_accuracy": 0.7423347234725952, + "num_tokens": 365274534.0, + "step": 15786 + }, + { + "epoch": 1.3018058876886287, + "grad_norm": 2.0873095989227295, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7783181667327881, + "num_tokens": 365305183.0, + "step": 15787 + }, + { + "epoch": 1.3018883483136803, + "grad_norm": 2.1882810592651367, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7352964878082275, + "num_tokens": 365332531.0, + "step": 15788 + }, + { + "epoch": 1.3019708089387318, + "grad_norm": 2.035431385040283, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7472735643386841, + "num_tokens": 365363913.0, + "step": 15789 + }, + { + "epoch": 1.3020532695637832, + "grad_norm": 2.0087790489196777, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.758525013923645, + "num_tokens": 365394101.0, + "step": 15790 + }, + { + "epoch": 1.3021357301888348, + "grad_norm": 1.9464377164840698, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7509275078773499, + "num_tokens": 365426113.0, + "step": 15791 + }, + { + "epoch": 1.3022181908138863, + "grad_norm": 2.4189696311950684, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7560460567474365, + "num_tokens": 365447016.0, + "step": 15792 + }, + { + "epoch": 1.302300651438938, + "grad_norm": 1.9885648488998413, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7481674551963806, + "num_tokens": 365477125.0, + "step": 15793 + }, + { + "epoch": 1.3023831120639895, + "grad_norm": 2.0173096656799316, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.7916744947433472, + "num_tokens": 365503375.0, + "step": 15794 + }, + { + "epoch": 1.302465572689041, + "grad_norm": 1.9893391132354736, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7449252009391785, + "num_tokens": 365535151.0, + "step": 15795 + }, + { + "epoch": 1.3025480333140926, + "grad_norm": 1.6538376808166504, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7873750925064087, + "num_tokens": 365571491.0, + "step": 15796 + }, + { + "epoch": 1.302630493939144, + "grad_norm": 1.8941651582717896, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7629673480987549, + "num_tokens": 365608536.0, + "step": 15797 + }, + { + "epoch": 1.3027129545641956, + "grad_norm": 2.276268720626831, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7660173177719116, + "num_tokens": 365632339.0, + "step": 15798 + }, + { + "epoch": 1.3027954151892471, + "grad_norm": 2.039844036102295, + "learning_rate": 1e-06, + "loss": 0.765, + "mean_token_accuracy": 0.7514693737030029, + "num_tokens": 365661086.0, + "step": 15799 + }, + { + "epoch": 1.3028778758142987, + "grad_norm": 2.045992136001587, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.748656690120697, + "num_tokens": 365688637.0, + "step": 15800 + }, + { + "epoch": 1.3029603364393503, + "grad_norm": 1.7255561351776123, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7729071378707886, + "num_tokens": 365723292.0, + "step": 15801 + }, + { + "epoch": 1.3030427970644016, + "grad_norm": 2.2409780025482178, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7785553336143494, + "num_tokens": 365746847.0, + "step": 15802 + }, + { + "epoch": 1.3031252576894534, + "grad_norm": 1.8976047039031982, + "learning_rate": 1e-06, + "loss": 0.5722, + "mean_token_accuracy": 0.8036813735961914, + "num_tokens": 365779250.0, + "step": 15803 + }, + { + "epoch": 1.3032077183145048, + "grad_norm": 2.170301675796509, + "learning_rate": 1e-06, + "loss": 0.7452, + "mean_token_accuracy": 0.7588889002799988, + "num_tokens": 365803409.0, + "step": 15804 + }, + { + "epoch": 1.3032901789395563, + "grad_norm": 2.0286307334899902, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.8022595643997192, + "num_tokens": 365833014.0, + "step": 15805 + }, + { + "epoch": 1.303372639564608, + "grad_norm": 1.9997891187667847, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7184415459632874, + "num_tokens": 365866807.0, + "step": 15806 + }, + { + "epoch": 1.3034551001896595, + "grad_norm": 2.410545587539673, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7451153993606567, + "num_tokens": 365890176.0, + "step": 15807 + }, + { + "epoch": 1.303537560814711, + "grad_norm": 1.854435682296753, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7647863030433655, + "num_tokens": 365924245.0, + "step": 15808 + }, + { + "epoch": 1.3036200214397624, + "grad_norm": 2.1314585208892822, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7522491216659546, + "num_tokens": 365952607.0, + "step": 15809 + }, + { + "epoch": 1.303702482064814, + "grad_norm": 2.078324556350708, + "learning_rate": 1e-06, + "loss": 0.6173, + "mean_token_accuracy": 0.7911512851715088, + "num_tokens": 365979910.0, + "step": 15810 + }, + { + "epoch": 1.3037849426898656, + "grad_norm": 1.8708648681640625, + "learning_rate": 1e-06, + "loss": 0.5986, + "mean_token_accuracy": 0.7986558675765991, + "num_tokens": 366009943.0, + "step": 15811 + }, + { + "epoch": 1.3038674033149171, + "grad_norm": 2.395465850830078, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.791885256767273, + "num_tokens": 366035105.0, + "step": 15812 + }, + { + "epoch": 1.3039498639399687, + "grad_norm": 1.8290612697601318, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.7888548374176025, + "num_tokens": 366072819.0, + "step": 15813 + }, + { + "epoch": 1.3040323245650203, + "grad_norm": 2.0420563220977783, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7392816543579102, + "num_tokens": 366102783.0, + "step": 15814 + }, + { + "epoch": 1.3041147851900718, + "grad_norm": 2.101300001144409, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7812871932983398, + "num_tokens": 366131042.0, + "step": 15815 + }, + { + "epoch": 1.3041972458151232, + "grad_norm": 1.8615953922271729, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7713034152984619, + "num_tokens": 366165012.0, + "step": 15816 + }, + { + "epoch": 1.3042797064401748, + "grad_norm": 1.7800803184509277, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7615566849708557, + "num_tokens": 366200186.0, + "step": 15817 + }, + { + "epoch": 1.3043621670652263, + "grad_norm": 1.901122808456421, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7609195709228516, + "num_tokens": 366229764.0, + "step": 15818 + }, + { + "epoch": 1.304444627690278, + "grad_norm": 1.8587124347686768, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.8024742007255554, + "num_tokens": 366261945.0, + "step": 15819 + }, + { + "epoch": 1.3045270883153295, + "grad_norm": 2.0348191261291504, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7716318368911743, + "num_tokens": 366291210.0, + "step": 15820 + }, + { + "epoch": 1.3046095489403808, + "grad_norm": 2.068650960922241, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7373403310775757, + "num_tokens": 366322312.0, + "step": 15821 + }, + { + "epoch": 1.3046920095654326, + "grad_norm": 1.982613444328308, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7508863210678101, + "num_tokens": 366355375.0, + "step": 15822 + }, + { + "epoch": 1.304774470190484, + "grad_norm": 2.4792282581329346, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7087033987045288, + "num_tokens": 366378664.0, + "step": 15823 + }, + { + "epoch": 1.3048569308155356, + "grad_norm": 2.31032657623291, + "learning_rate": 1e-06, + "loss": 0.7486, + "mean_token_accuracy": 0.760993480682373, + "num_tokens": 366404479.0, + "step": 15824 + }, + { + "epoch": 1.3049393914405871, + "grad_norm": 2.069887638092041, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7439829111099243, + "num_tokens": 366436613.0, + "step": 15825 + }, + { + "epoch": 1.3050218520656387, + "grad_norm": 1.7894445657730103, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7735636830329895, + "num_tokens": 366472974.0, + "step": 15826 + }, + { + "epoch": 1.3051043126906903, + "grad_norm": 2.0691962242126465, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7607907652854919, + "num_tokens": 366502708.0, + "step": 15827 + }, + { + "epoch": 1.3051867733157416, + "grad_norm": 1.9745110273361206, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7669821977615356, + "num_tokens": 366530602.0, + "step": 15828 + }, + { + "epoch": 1.3052692339407932, + "grad_norm": 1.9740015268325806, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7441227436065674, + "num_tokens": 366558903.0, + "step": 15829 + }, + { + "epoch": 1.3053516945658448, + "grad_norm": 2.087268590927124, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7422606348991394, + "num_tokens": 366585460.0, + "step": 15830 + }, + { + "epoch": 1.3054341551908963, + "grad_norm": 1.8130834102630615, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.7913558483123779, + "num_tokens": 366619926.0, + "step": 15831 + }, + { + "epoch": 1.305516615815948, + "grad_norm": 1.9253263473510742, + "learning_rate": 1e-06, + "loss": 0.6103, + "mean_token_accuracy": 0.8026456832885742, + "num_tokens": 366652167.0, + "step": 15832 + }, + { + "epoch": 1.3055990764409995, + "grad_norm": 2.2414350509643555, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.7408672571182251, + "num_tokens": 366677128.0, + "step": 15833 + }, + { + "epoch": 1.305681537066051, + "grad_norm": 2.131978750228882, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7511398196220398, + "num_tokens": 366703210.0, + "step": 15834 + }, + { + "epoch": 1.3057639976911024, + "grad_norm": 2.2463841438293457, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7327578067779541, + "num_tokens": 366731213.0, + "step": 15835 + }, + { + "epoch": 1.305846458316154, + "grad_norm": 2.007262706756592, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7576271295547485, + "num_tokens": 366760481.0, + "step": 15836 + }, + { + "epoch": 1.3059289189412056, + "grad_norm": 1.888978123664856, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7622371912002563, + "num_tokens": 366793391.0, + "step": 15837 + }, + { + "epoch": 1.3060113795662571, + "grad_norm": 2.0271756649017334, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7647116780281067, + "num_tokens": 366820906.0, + "step": 15838 + }, + { + "epoch": 1.3060938401913087, + "grad_norm": 2.098031759262085, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7702693343162537, + "num_tokens": 366851238.0, + "step": 15839 + }, + { + "epoch": 1.3061763008163603, + "grad_norm": 1.9481464624404907, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7588787078857422, + "num_tokens": 366883294.0, + "step": 15840 + }, + { + "epoch": 1.3062587614414118, + "grad_norm": 1.905765414237976, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.7919725179672241, + "num_tokens": 366914044.0, + "step": 15841 + }, + { + "epoch": 1.3063412220664632, + "grad_norm": 2.177795886993408, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.763590931892395, + "num_tokens": 366943358.0, + "step": 15842 + }, + { + "epoch": 1.3064236826915148, + "grad_norm": 2.001753330230713, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7753617763519287, + "num_tokens": 366970760.0, + "step": 15843 + }, + { + "epoch": 1.3065061433165663, + "grad_norm": 1.9471744298934937, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7362514734268188, + "num_tokens": 367001714.0, + "step": 15844 + }, + { + "epoch": 1.306588603941618, + "grad_norm": 1.7849440574645996, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7633140683174133, + "num_tokens": 367040162.0, + "step": 15845 + }, + { + "epoch": 1.3066710645666695, + "grad_norm": 1.7225217819213867, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7638506889343262, + "num_tokens": 367075885.0, + "step": 15846 + }, + { + "epoch": 1.3067535251917208, + "grad_norm": 1.8946280479431152, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.7907127141952515, + "num_tokens": 367107783.0, + "step": 15847 + }, + { + "epoch": 1.3068359858167726, + "grad_norm": 2.313523769378662, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7414416670799255, + "num_tokens": 367137548.0, + "step": 15848 + }, + { + "epoch": 1.306918446441824, + "grad_norm": 2.096963882446289, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7402042150497437, + "num_tokens": 367167059.0, + "step": 15849 + }, + { + "epoch": 1.3070009070668756, + "grad_norm": 1.9124761819839478, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7668341398239136, + "num_tokens": 367198960.0, + "step": 15850 + }, + { + "epoch": 1.3070833676919271, + "grad_norm": 2.251657485961914, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.744245171546936, + "num_tokens": 367224614.0, + "step": 15851 + }, + { + "epoch": 1.3071658283169787, + "grad_norm": 2.197380781173706, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7406095266342163, + "num_tokens": 367250730.0, + "step": 15852 + }, + { + "epoch": 1.3072482889420303, + "grad_norm": 2.2928578853607178, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7371580004692078, + "num_tokens": 367276322.0, + "step": 15853 + }, + { + "epoch": 1.3073307495670816, + "grad_norm": 1.778214693069458, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7649964690208435, + "num_tokens": 367311870.0, + "step": 15854 + }, + { + "epoch": 1.3074132101921332, + "grad_norm": 1.983148217201233, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7564613819122314, + "num_tokens": 367342363.0, + "step": 15855 + }, + { + "epoch": 1.3074956708171848, + "grad_norm": 1.862664818763733, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7574548721313477, + "num_tokens": 367377046.0, + "step": 15856 + }, + { + "epoch": 1.3075781314422363, + "grad_norm": 1.9733730554580688, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7492288947105408, + "num_tokens": 367405179.0, + "step": 15857 + }, + { + "epoch": 1.307660592067288, + "grad_norm": 1.8745167255401611, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7465020418167114, + "num_tokens": 367437900.0, + "step": 15858 + }, + { + "epoch": 1.3077430526923395, + "grad_norm": 1.7552224397659302, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7654308676719666, + "num_tokens": 367474121.0, + "step": 15859 + }, + { + "epoch": 1.307825513317391, + "grad_norm": 1.8737056255340576, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7848609685897827, + "num_tokens": 367506942.0, + "step": 15860 + }, + { + "epoch": 1.3079079739424424, + "grad_norm": 2.302565336227417, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7816125154495239, + "num_tokens": 367529322.0, + "step": 15861 + }, + { + "epoch": 1.307990434567494, + "grad_norm": 2.127436637878418, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.7730522155761719, + "num_tokens": 367556185.0, + "step": 15862 + }, + { + "epoch": 1.3080728951925455, + "grad_norm": 2.1104516983032227, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7185031175613403, + "num_tokens": 367583738.0, + "step": 15863 + }, + { + "epoch": 1.3081553558175971, + "grad_norm": 1.9735466241836548, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7900363206863403, + "num_tokens": 367612810.0, + "step": 15864 + }, + { + "epoch": 1.3082378164426487, + "grad_norm": 1.715233564376831, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7565267086029053, + "num_tokens": 367652206.0, + "step": 15865 + }, + { + "epoch": 1.3083202770677, + "grad_norm": 1.711208462715149, + "learning_rate": 1e-06, + "loss": 0.7706, + "mean_token_accuracy": 0.7552992105484009, + "num_tokens": 367688267.0, + "step": 15866 + }, + { + "epoch": 1.3084027376927518, + "grad_norm": 1.8498611450195312, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7307312488555908, + "num_tokens": 367722273.0, + "step": 15867 + }, + { + "epoch": 1.3084851983178032, + "grad_norm": 2.0347869396209717, + "learning_rate": 1e-06, + "loss": 0.5972, + "mean_token_accuracy": 0.807102620601654, + "num_tokens": 367747729.0, + "step": 15868 + }, + { + "epoch": 1.3085676589428548, + "grad_norm": 1.9004099369049072, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7779485583305359, + "num_tokens": 367777470.0, + "step": 15869 + }, + { + "epoch": 1.3086501195679063, + "grad_norm": 1.978798747062683, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.7792938947677612, + "num_tokens": 367803036.0, + "step": 15870 + }, + { + "epoch": 1.308732580192958, + "grad_norm": 1.905051350593567, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7828125357627869, + "num_tokens": 367834430.0, + "step": 15871 + }, + { + "epoch": 1.3088150408180095, + "grad_norm": 2.0063886642456055, + "learning_rate": 1e-06, + "loss": 0.5665, + "mean_token_accuracy": 0.7991917133331299, + "num_tokens": 367861819.0, + "step": 15872 + }, + { + "epoch": 1.3088975014430608, + "grad_norm": 1.944618582725525, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7401549816131592, + "num_tokens": 367895937.0, + "step": 15873 + }, + { + "epoch": 1.3089799620681124, + "grad_norm": 2.1135878562927246, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7766176462173462, + "num_tokens": 367923490.0, + "step": 15874 + }, + { + "epoch": 1.309062422693164, + "grad_norm": 1.919480323791504, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7456892728805542, + "num_tokens": 367957810.0, + "step": 15875 + }, + { + "epoch": 1.3091448833182155, + "grad_norm": 2.135791778564453, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7654139995574951, + "num_tokens": 367983328.0, + "step": 15876 + }, + { + "epoch": 1.3092273439432671, + "grad_norm": 2.0193734169006348, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7440758943557739, + "num_tokens": 368012301.0, + "step": 15877 + }, + { + "epoch": 1.3093098045683187, + "grad_norm": 2.0309836864471436, + "learning_rate": 1e-06, + "loss": 0.5984, + "mean_token_accuracy": 0.8053900003433228, + "num_tokens": 368041962.0, + "step": 15878 + }, + { + "epoch": 1.3093922651933703, + "grad_norm": 2.1284537315368652, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.7922287583351135, + "num_tokens": 368065685.0, + "step": 15879 + }, + { + "epoch": 1.3094747258184216, + "grad_norm": 1.7858892679214478, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7824161648750305, + "num_tokens": 368099048.0, + "step": 15880 + }, + { + "epoch": 1.3095571864434732, + "grad_norm": 1.95490300655365, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7687005400657654, + "num_tokens": 368131351.0, + "step": 15881 + }, + { + "epoch": 1.3096396470685248, + "grad_norm": 1.830121397972107, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.750019371509552, + "num_tokens": 368167421.0, + "step": 15882 + }, + { + "epoch": 1.3097221076935763, + "grad_norm": 1.9535621404647827, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.7867761254310608, + "num_tokens": 368199519.0, + "step": 15883 + }, + { + "epoch": 1.309804568318628, + "grad_norm": 2.162627696990967, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7313849329948425, + "num_tokens": 368226342.0, + "step": 15884 + }, + { + "epoch": 1.3098870289436793, + "grad_norm": 1.8872343301773071, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7325320243835449, + "num_tokens": 368261103.0, + "step": 15885 + }, + { + "epoch": 1.309969489568731, + "grad_norm": 1.8942697048187256, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7360490560531616, + "num_tokens": 368295825.0, + "step": 15886 + }, + { + "epoch": 1.3100519501937824, + "grad_norm": 2.023023843765259, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7576038837432861, + "num_tokens": 368323706.0, + "step": 15887 + }, + { + "epoch": 1.310134410818834, + "grad_norm": 1.8587470054626465, + "learning_rate": 1e-06, + "loss": 0.6239, + "mean_token_accuracy": 0.7956428527832031, + "num_tokens": 368356410.0, + "step": 15888 + }, + { + "epoch": 1.3102168714438855, + "grad_norm": 2.1480872631073, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7787412405014038, + "num_tokens": 368384175.0, + "step": 15889 + }, + { + "epoch": 1.3102993320689371, + "grad_norm": 1.936876893043518, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7646661996841431, + "num_tokens": 368418623.0, + "step": 15890 + }, + { + "epoch": 1.3103817926939887, + "grad_norm": 2.0576155185699463, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.775597333908081, + "num_tokens": 368443550.0, + "step": 15891 + }, + { + "epoch": 1.31046425331904, + "grad_norm": 2.190849542617798, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7559313178062439, + "num_tokens": 368468346.0, + "step": 15892 + }, + { + "epoch": 1.3105467139440918, + "grad_norm": 1.869093656539917, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7598198652267456, + "num_tokens": 368499810.0, + "step": 15893 + }, + { + "epoch": 1.3106291745691432, + "grad_norm": 1.980090618133545, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7370240688323975, + "num_tokens": 368531758.0, + "step": 15894 + }, + { + "epoch": 1.3107116351941948, + "grad_norm": 1.8299593925476074, + "learning_rate": 1e-06, + "loss": 0.5609, + "mean_token_accuracy": 0.8145033121109009, + "num_tokens": 368564330.0, + "step": 15895 + }, + { + "epoch": 1.3107940958192463, + "grad_norm": 1.9017621278762817, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7694786190986633, + "num_tokens": 368596445.0, + "step": 15896 + }, + { + "epoch": 1.310876556444298, + "grad_norm": 1.9610388278961182, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7494792938232422, + "num_tokens": 368628611.0, + "step": 15897 + }, + { + "epoch": 1.3109590170693495, + "grad_norm": 2.3111367225646973, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7298436164855957, + "num_tokens": 368653720.0, + "step": 15898 + }, + { + "epoch": 1.3110414776944008, + "grad_norm": 2.1010637283325195, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7459615468978882, + "num_tokens": 368681915.0, + "step": 15899 + }, + { + "epoch": 1.3111239383194524, + "grad_norm": 2.122126579284668, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7629700899124146, + "num_tokens": 368709181.0, + "step": 15900 + }, + { + "epoch": 1.311206398944504, + "grad_norm": 1.8770661354064941, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7831094264984131, + "num_tokens": 368741603.0, + "step": 15901 + }, + { + "epoch": 1.3112888595695555, + "grad_norm": 2.032665252685547, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.743757963180542, + "num_tokens": 368774451.0, + "step": 15902 + }, + { + "epoch": 1.3113713201946071, + "grad_norm": 1.9107944965362549, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7473329305648804, + "num_tokens": 368805606.0, + "step": 15903 + }, + { + "epoch": 1.3114537808196587, + "grad_norm": 1.9231183528900146, + "learning_rate": 1e-06, + "loss": 0.7702, + "mean_token_accuracy": 0.7543225288391113, + "num_tokens": 368836339.0, + "step": 15904 + }, + { + "epoch": 1.3115362414447103, + "grad_norm": 2.2909862995147705, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.746960461139679, + "num_tokens": 368859529.0, + "step": 15905 + }, + { + "epoch": 1.3116187020697616, + "grad_norm": 2.089479446411133, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7697729468345642, + "num_tokens": 368888361.0, + "step": 15906 + }, + { + "epoch": 1.3117011626948132, + "grad_norm": 1.8940083980560303, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.728723406791687, + "num_tokens": 368920887.0, + "step": 15907 + }, + { + "epoch": 1.3117836233198648, + "grad_norm": 2.099968910217285, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7404953241348267, + "num_tokens": 368948842.0, + "step": 15908 + }, + { + "epoch": 1.3118660839449163, + "grad_norm": 2.045543909072876, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.747530996799469, + "num_tokens": 368977751.0, + "step": 15909 + }, + { + "epoch": 1.311948544569968, + "grad_norm": 1.926262617111206, + "learning_rate": 1e-06, + "loss": 0.5632, + "mean_token_accuracy": 0.8135266900062561, + "num_tokens": 369009287.0, + "step": 15910 + }, + { + "epoch": 1.3120310051950193, + "grad_norm": 1.986763596534729, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.748099684715271, + "num_tokens": 369036709.0, + "step": 15911 + }, + { + "epoch": 1.312113465820071, + "grad_norm": 2.0388247966766357, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7415907979011536, + "num_tokens": 369068059.0, + "step": 15912 + }, + { + "epoch": 1.3121959264451224, + "grad_norm": 1.9076623916625977, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7609584927558899, + "num_tokens": 369100503.0, + "step": 15913 + }, + { + "epoch": 1.312278387070174, + "grad_norm": 2.3125686645507812, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7348476052284241, + "num_tokens": 369125838.0, + "step": 15914 + }, + { + "epoch": 1.3123608476952255, + "grad_norm": 2.0698328018188477, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.723121166229248, + "num_tokens": 369155502.0, + "step": 15915 + }, + { + "epoch": 1.3124433083202771, + "grad_norm": 2.377589702606201, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7350240349769592, + "num_tokens": 369179589.0, + "step": 15916 + }, + { + "epoch": 1.3125257689453287, + "grad_norm": 1.913787603378296, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7563154697418213, + "num_tokens": 369212304.0, + "step": 15917 + }, + { + "epoch": 1.31260822957038, + "grad_norm": 1.999745488166809, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7671191692352295, + "num_tokens": 369241688.0, + "step": 15918 + }, + { + "epoch": 1.3126906901954316, + "grad_norm": 1.9722912311553955, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7628041505813599, + "num_tokens": 369271693.0, + "step": 15919 + }, + { + "epoch": 1.3127731508204832, + "grad_norm": 1.8755396604537964, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7856464982032776, + "num_tokens": 369302902.0, + "step": 15920 + }, + { + "epoch": 1.3128556114455348, + "grad_norm": 1.9706521034240723, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7570416927337646, + "num_tokens": 369334205.0, + "step": 15921 + }, + { + "epoch": 1.3129380720705863, + "grad_norm": 2.0982160568237305, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.7912511229515076, + "num_tokens": 369357199.0, + "step": 15922 + }, + { + "epoch": 1.313020532695638, + "grad_norm": 2.1426076889038086, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7331621646881104, + "num_tokens": 369385894.0, + "step": 15923 + }, + { + "epoch": 1.3131029933206895, + "grad_norm": 2.095858573913574, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7291891574859619, + "num_tokens": 369416425.0, + "step": 15924 + }, + { + "epoch": 1.3131854539457408, + "grad_norm": 2.0168983936309814, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7495424151420593, + "num_tokens": 369446493.0, + "step": 15925 + }, + { + "epoch": 1.3132679145707924, + "grad_norm": 1.8727532625198364, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.759541392326355, + "num_tokens": 369478989.0, + "step": 15926 + }, + { + "epoch": 1.313350375195844, + "grad_norm": 2.099433183670044, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7506982088088989, + "num_tokens": 369508180.0, + "step": 15927 + }, + { + "epoch": 1.3134328358208955, + "grad_norm": 1.959100365638733, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.7831394672393799, + "num_tokens": 369538092.0, + "step": 15928 + }, + { + "epoch": 1.3135152964459471, + "grad_norm": 2.496173620223999, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.756999135017395, + "num_tokens": 369559638.0, + "step": 15929 + }, + { + "epoch": 1.3135977570709985, + "grad_norm": 2.0715904235839844, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7380121946334839, + "num_tokens": 369592129.0, + "step": 15930 + }, + { + "epoch": 1.3136802176960503, + "grad_norm": 2.030778408050537, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.767096996307373, + "num_tokens": 369621299.0, + "step": 15931 + }, + { + "epoch": 1.3137626783211016, + "grad_norm": 1.9782012701034546, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7655543088912964, + "num_tokens": 369652982.0, + "step": 15932 + }, + { + "epoch": 1.3138451389461532, + "grad_norm": 2.1571826934814453, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7186551094055176, + "num_tokens": 369680466.0, + "step": 15933 + }, + { + "epoch": 1.3139275995712048, + "grad_norm": 2.0806124210357666, + "learning_rate": 1e-06, + "loss": 0.6011, + "mean_token_accuracy": 0.802322268486023, + "num_tokens": 369706250.0, + "step": 15934 + }, + { + "epoch": 1.3140100601962563, + "grad_norm": 2.1750428676605225, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.72346031665802, + "num_tokens": 369736625.0, + "step": 15935 + }, + { + "epoch": 1.314092520821308, + "grad_norm": 1.8732281923294067, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.758937418460846, + "num_tokens": 369771681.0, + "step": 15936 + }, + { + "epoch": 1.3141749814463592, + "grad_norm": 1.906541109085083, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7615842223167419, + "num_tokens": 369802603.0, + "step": 15937 + }, + { + "epoch": 1.3142574420714108, + "grad_norm": 2.015052556991577, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7510107159614563, + "num_tokens": 369831696.0, + "step": 15938 + }, + { + "epoch": 1.3143399026964624, + "grad_norm": 2.0464677810668945, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7615342140197754, + "num_tokens": 369858924.0, + "step": 15939 + }, + { + "epoch": 1.314422363321514, + "grad_norm": 1.817914366722107, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7488399744033813, + "num_tokens": 369894934.0, + "step": 15940 + }, + { + "epoch": 1.3145048239465655, + "grad_norm": 1.7429313659667969, + "learning_rate": 1e-06, + "loss": 0.5169, + "mean_token_accuracy": 0.8091936111450195, + "num_tokens": 369928705.0, + "step": 15941 + }, + { + "epoch": 1.314587284571617, + "grad_norm": 2.128568410873413, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7762226462364197, + "num_tokens": 369956772.0, + "step": 15942 + }, + { + "epoch": 1.3146697451966687, + "grad_norm": 1.9546445608139038, + "learning_rate": 1e-06, + "loss": 0.7774, + "mean_token_accuracy": 0.7511705160140991, + "num_tokens": 369985839.0, + "step": 15943 + }, + { + "epoch": 1.31475220582172, + "grad_norm": 2.1887025833129883, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7480243444442749, + "num_tokens": 370011034.0, + "step": 15944 + }, + { + "epoch": 1.3148346664467716, + "grad_norm": 1.85349440574646, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.7856618165969849, + "num_tokens": 370042393.0, + "step": 15945 + }, + { + "epoch": 1.3149171270718232, + "grad_norm": 1.9850480556488037, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7732348442077637, + "num_tokens": 370072073.0, + "step": 15946 + }, + { + "epoch": 1.3149995876968747, + "grad_norm": 1.7691447734832764, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7273308634757996, + "num_tokens": 370109595.0, + "step": 15947 + }, + { + "epoch": 1.3150820483219263, + "grad_norm": 1.7941277027130127, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7581756711006165, + "num_tokens": 370144867.0, + "step": 15948 + }, + { + "epoch": 1.315164508946978, + "grad_norm": 2.1484668254852295, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7414200305938721, + "num_tokens": 370172988.0, + "step": 15949 + }, + { + "epoch": 1.3152469695720295, + "grad_norm": 1.996293306350708, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7676314115524292, + "num_tokens": 370203051.0, + "step": 15950 + }, + { + "epoch": 1.3153294301970808, + "grad_norm": 2.445345163345337, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7562922835350037, + "num_tokens": 370227447.0, + "step": 15951 + }, + { + "epoch": 1.3154118908221324, + "grad_norm": 2.1266329288482666, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7354535460472107, + "num_tokens": 370255835.0, + "step": 15952 + }, + { + "epoch": 1.315494351447184, + "grad_norm": 2.112546920776367, + "learning_rate": 1e-06, + "loss": 0.511, + "mean_token_accuracy": 0.8096096515655518, + "num_tokens": 370280694.0, + "step": 15953 + }, + { + "epoch": 1.3155768120722355, + "grad_norm": 2.005789279937744, + "learning_rate": 1e-06, + "loss": 0.5691, + "mean_token_accuracy": 0.8128635883331299, + "num_tokens": 370313400.0, + "step": 15954 + }, + { + "epoch": 1.315659272697287, + "grad_norm": 2.073394298553467, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7647089958190918, + "num_tokens": 370340346.0, + "step": 15955 + }, + { + "epoch": 1.3157417333223385, + "grad_norm": 1.9347559213638306, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7662608623504639, + "num_tokens": 370372046.0, + "step": 15956 + }, + { + "epoch": 1.3158241939473903, + "grad_norm": 1.92149019241333, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7850671410560608, + "num_tokens": 370400531.0, + "step": 15957 + }, + { + "epoch": 1.3159066545724416, + "grad_norm": 2.0766866207122803, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7473537921905518, + "num_tokens": 370428550.0, + "step": 15958 + }, + { + "epoch": 1.3159891151974932, + "grad_norm": 2.2612648010253906, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7564090490341187, + "num_tokens": 370455519.0, + "step": 15959 + }, + { + "epoch": 1.3160715758225447, + "grad_norm": 1.6749902963638306, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.779682993888855, + "num_tokens": 370499254.0, + "step": 15960 + }, + { + "epoch": 1.3161540364475963, + "grad_norm": 2.004193067550659, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.7989546060562134, + "num_tokens": 370527688.0, + "step": 15961 + }, + { + "epoch": 1.316236497072648, + "grad_norm": 2.145186185836792, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7597886323928833, + "num_tokens": 370555870.0, + "step": 15962 + }, + { + "epoch": 1.3163189576976992, + "grad_norm": 2.126450538635254, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7413849234580994, + "num_tokens": 370583674.0, + "step": 15963 + }, + { + "epoch": 1.3164014183227508, + "grad_norm": 2.0889151096343994, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7624158263206482, + "num_tokens": 370610802.0, + "step": 15964 + }, + { + "epoch": 1.3164838789478024, + "grad_norm": 2.3592159748077393, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7478822469711304, + "num_tokens": 370634220.0, + "step": 15965 + }, + { + "epoch": 1.316566339572854, + "grad_norm": 1.980676293373108, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7567214369773865, + "num_tokens": 370664247.0, + "step": 15966 + }, + { + "epoch": 1.3166488001979055, + "grad_norm": 1.9304462671279907, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7532933950424194, + "num_tokens": 370696691.0, + "step": 15967 + }, + { + "epoch": 1.316731260822957, + "grad_norm": 1.8756691217422485, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7689427733421326, + "num_tokens": 370728648.0, + "step": 15968 + }, + { + "epoch": 1.3168137214480087, + "grad_norm": 2.0477185249328613, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7624284029006958, + "num_tokens": 370757721.0, + "step": 15969 + }, + { + "epoch": 1.31689618207306, + "grad_norm": 1.6543673276901245, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7548279762268066, + "num_tokens": 370798658.0, + "step": 15970 + }, + { + "epoch": 1.3169786426981116, + "grad_norm": 1.7872635126113892, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7652860879898071, + "num_tokens": 370833119.0, + "step": 15971 + }, + { + "epoch": 1.3170611033231632, + "grad_norm": 1.804833173751831, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7702450752258301, + "num_tokens": 370870245.0, + "step": 15972 + }, + { + "epoch": 1.3171435639482147, + "grad_norm": 2.0849409103393555, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7717580795288086, + "num_tokens": 370897522.0, + "step": 15973 + }, + { + "epoch": 1.3172260245732663, + "grad_norm": 2.006140947341919, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7739921808242798, + "num_tokens": 370927472.0, + "step": 15974 + }, + { + "epoch": 1.3173084851983177, + "grad_norm": 2.288215160369873, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7432165741920471, + "num_tokens": 370951167.0, + "step": 15975 + }, + { + "epoch": 1.3173909458233695, + "grad_norm": 2.22025990486145, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7262818813323975, + "num_tokens": 370978304.0, + "step": 15976 + }, + { + "epoch": 1.3174734064484208, + "grad_norm": 2.048567295074463, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7682950496673584, + "num_tokens": 371007731.0, + "step": 15977 + }, + { + "epoch": 1.3175558670734724, + "grad_norm": 1.7591290473937988, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.738111138343811, + "num_tokens": 371048967.0, + "step": 15978 + }, + { + "epoch": 1.317638327698524, + "grad_norm": 2.0013418197631836, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7724410891532898, + "num_tokens": 371081210.0, + "step": 15979 + }, + { + "epoch": 1.3177207883235755, + "grad_norm": 2.164919376373291, + "learning_rate": 1e-06, + "loss": 0.7704, + "mean_token_accuracy": 0.756743311882019, + "num_tokens": 371106686.0, + "step": 15980 + }, + { + "epoch": 1.317803248948627, + "grad_norm": 2.110853672027588, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7520819902420044, + "num_tokens": 371133845.0, + "step": 15981 + }, + { + "epoch": 1.3178857095736785, + "grad_norm": 1.8757665157318115, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7558171153068542, + "num_tokens": 371165243.0, + "step": 15982 + }, + { + "epoch": 1.31796817019873, + "grad_norm": 2.0399298667907715, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7561541199684143, + "num_tokens": 371194949.0, + "step": 15983 + }, + { + "epoch": 1.3180506308237816, + "grad_norm": 1.8779881000518799, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7531108856201172, + "num_tokens": 371228612.0, + "step": 15984 + }, + { + "epoch": 1.3181330914488332, + "grad_norm": 2.0973970890045166, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7579911947250366, + "num_tokens": 371257100.0, + "step": 15985 + }, + { + "epoch": 1.3182155520738847, + "grad_norm": 1.8331466913223267, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7772846221923828, + "num_tokens": 371291421.0, + "step": 15986 + }, + { + "epoch": 1.3182980126989363, + "grad_norm": 2.05399489402771, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7708699703216553, + "num_tokens": 371317696.0, + "step": 15987 + }, + { + "epoch": 1.318380473323988, + "grad_norm": 1.7235820293426514, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.8023220300674438, + "num_tokens": 371354749.0, + "step": 15988 + }, + { + "epoch": 1.3184629339490392, + "grad_norm": 1.8938175439834595, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7587332725524902, + "num_tokens": 371386155.0, + "step": 15989 + }, + { + "epoch": 1.3185453945740908, + "grad_norm": 2.0832765102386475, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.7414084672927856, + "num_tokens": 371416426.0, + "step": 15990 + }, + { + "epoch": 1.3186278551991424, + "grad_norm": 1.878098964691162, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.7935863137245178, + "num_tokens": 371447738.0, + "step": 15991 + }, + { + "epoch": 1.318710315824194, + "grad_norm": 2.2078020572662354, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7387023568153381, + "num_tokens": 371474010.0, + "step": 15992 + }, + { + "epoch": 1.3187927764492455, + "grad_norm": 2.1894145011901855, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7431083917617798, + "num_tokens": 371500289.0, + "step": 15993 + }, + { + "epoch": 1.318875237074297, + "grad_norm": 1.8419816493988037, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7613905072212219, + "num_tokens": 371535021.0, + "step": 15994 + }, + { + "epoch": 1.3189576976993487, + "grad_norm": 1.7858588695526123, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7833685874938965, + "num_tokens": 371565104.0, + "step": 15995 + }, + { + "epoch": 1.3190401583244, + "grad_norm": 2.14752459526062, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7464379668235779, + "num_tokens": 371592297.0, + "step": 15996 + }, + { + "epoch": 1.3191226189494516, + "grad_norm": 1.8493412733078003, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.7525372505187988, + "num_tokens": 371624969.0, + "step": 15997 + }, + { + "epoch": 1.3192050795745032, + "grad_norm": 2.0527377128601074, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.7848248481750488, + "num_tokens": 371653726.0, + "step": 15998 + }, + { + "epoch": 1.3192875401995547, + "grad_norm": 1.8547005653381348, + "learning_rate": 1e-06, + "loss": 0.5572, + "mean_token_accuracy": 0.8023924231529236, + "num_tokens": 371683492.0, + "step": 15999 + }, + { + "epoch": 1.3193700008246063, + "grad_norm": 2.100343942642212, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7793354988098145, + "num_tokens": 371709795.0, + "step": 16000 + }, + { + "epoch": 1.3194524614496577, + "grad_norm": 2.215162515640259, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7600283622741699, + "num_tokens": 371733933.0, + "step": 16001 + }, + { + "epoch": 1.3195349220747095, + "grad_norm": 2.0217809677124023, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7777826189994812, + "num_tokens": 371763589.0, + "step": 16002 + }, + { + "epoch": 1.3196173826997608, + "grad_norm": 1.9467964172363281, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7652322053909302, + "num_tokens": 371795486.0, + "step": 16003 + }, + { + "epoch": 1.3196998433248124, + "grad_norm": 1.8322303295135498, + "learning_rate": 1e-06, + "loss": 0.5152, + "mean_token_accuracy": 0.8275181651115417, + "num_tokens": 371827188.0, + "step": 16004 + }, + { + "epoch": 1.319782303949864, + "grad_norm": 1.8640650510787964, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7735280990600586, + "num_tokens": 371860145.0, + "step": 16005 + }, + { + "epoch": 1.3198647645749155, + "grad_norm": 2.1554274559020996, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.7868831157684326, + "num_tokens": 371884914.0, + "step": 16006 + }, + { + "epoch": 1.319947225199967, + "grad_norm": 2.183150053024292, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7678214311599731, + "num_tokens": 371916086.0, + "step": 16007 + }, + { + "epoch": 1.3200296858250185, + "grad_norm": 2.000051736831665, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7763593196868896, + "num_tokens": 371944034.0, + "step": 16008 + }, + { + "epoch": 1.32011214645007, + "grad_norm": 2.0509631633758545, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7803436517715454, + "num_tokens": 371970040.0, + "step": 16009 + }, + { + "epoch": 1.3201946070751216, + "grad_norm": 2.052478075027466, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7606325149536133, + "num_tokens": 371997728.0, + "step": 16010 + }, + { + "epoch": 1.3202770677001732, + "grad_norm": 2.41813588142395, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7788639068603516, + "num_tokens": 372020453.0, + "step": 16011 + }, + { + "epoch": 1.3203595283252247, + "grad_norm": 2.4831318855285645, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7368086576461792, + "num_tokens": 372043767.0, + "step": 16012 + }, + { + "epoch": 1.3204419889502763, + "grad_norm": 1.678878903388977, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7705588340759277, + "num_tokens": 372081740.0, + "step": 16013 + }, + { + "epoch": 1.3205244495753279, + "grad_norm": 2.063058853149414, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7600502371788025, + "num_tokens": 372112872.0, + "step": 16014 + }, + { + "epoch": 1.3206069102003792, + "grad_norm": 2.2217631340026855, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.759975016117096, + "num_tokens": 372142858.0, + "step": 16015 + }, + { + "epoch": 1.3206893708254308, + "grad_norm": 2.046950101852417, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7780239582061768, + "num_tokens": 372173544.0, + "step": 16016 + }, + { + "epoch": 1.3207718314504824, + "grad_norm": 2.0183794498443604, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.7526737451553345, + "num_tokens": 372201758.0, + "step": 16017 + }, + { + "epoch": 1.320854292075534, + "grad_norm": 1.9110727310180664, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7398809790611267, + "num_tokens": 372232870.0, + "step": 16018 + }, + { + "epoch": 1.3209367527005855, + "grad_norm": 2.177840232849121, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7653448581695557, + "num_tokens": 372257704.0, + "step": 16019 + }, + { + "epoch": 1.3210192133256369, + "grad_norm": 1.8888709545135498, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7611657381057739, + "num_tokens": 372287156.0, + "step": 16020 + }, + { + "epoch": 1.3211016739506887, + "grad_norm": 2.035372257232666, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7560786008834839, + "num_tokens": 372315906.0, + "step": 16021 + }, + { + "epoch": 1.32118413457574, + "grad_norm": 1.910636305809021, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7763528227806091, + "num_tokens": 372345931.0, + "step": 16022 + }, + { + "epoch": 1.3212665952007916, + "grad_norm": 1.9816961288452148, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7289234399795532, + "num_tokens": 372380273.0, + "step": 16023 + }, + { + "epoch": 1.3213490558258432, + "grad_norm": 1.8732483386993408, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7794885635375977, + "num_tokens": 372410976.0, + "step": 16024 + }, + { + "epoch": 1.3214315164508947, + "grad_norm": 1.7902477979660034, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7665925025939941, + "num_tokens": 372447052.0, + "step": 16025 + }, + { + "epoch": 1.3215139770759463, + "grad_norm": 2.021383285522461, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7307350635528564, + "num_tokens": 372477834.0, + "step": 16026 + }, + { + "epoch": 1.3215964377009977, + "grad_norm": 1.8304864168167114, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.7929903268814087, + "num_tokens": 372509160.0, + "step": 16027 + }, + { + "epoch": 1.3216788983260492, + "grad_norm": 2.031365156173706, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7789911031723022, + "num_tokens": 372536946.0, + "step": 16028 + }, + { + "epoch": 1.3217613589511008, + "grad_norm": 2.1266086101531982, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7819955348968506, + "num_tokens": 372564902.0, + "step": 16029 + }, + { + "epoch": 1.3218438195761524, + "grad_norm": 2.032330274581909, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.7924381494522095, + "num_tokens": 372591750.0, + "step": 16030 + }, + { + "epoch": 1.321926280201204, + "grad_norm": 2.1299855709075928, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7569195032119751, + "num_tokens": 372618919.0, + "step": 16031 + }, + { + "epoch": 1.3220087408262555, + "grad_norm": 1.8909804821014404, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7564485669136047, + "num_tokens": 372649315.0, + "step": 16032 + }, + { + "epoch": 1.322091201451307, + "grad_norm": 1.9819228649139404, + "learning_rate": 1e-06, + "loss": 0.5492, + "mean_token_accuracy": 0.8133264183998108, + "num_tokens": 372676605.0, + "step": 16033 + }, + { + "epoch": 1.3221736620763584, + "grad_norm": 1.7789627313613892, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7691251635551453, + "num_tokens": 372713900.0, + "step": 16034 + }, + { + "epoch": 1.32225612270141, + "grad_norm": 2.04494047164917, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7378129363059998, + "num_tokens": 372745031.0, + "step": 16035 + }, + { + "epoch": 1.3223385833264616, + "grad_norm": 1.9464738368988037, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.7937154769897461, + "num_tokens": 372774154.0, + "step": 16036 + }, + { + "epoch": 1.3224210439515132, + "grad_norm": 1.888282299041748, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7712664604187012, + "num_tokens": 372806240.0, + "step": 16037 + }, + { + "epoch": 1.3225035045765647, + "grad_norm": 2.2277376651763916, + "learning_rate": 1e-06, + "loss": 0.6096, + "mean_token_accuracy": 0.8025811314582825, + "num_tokens": 372831776.0, + "step": 16038 + }, + { + "epoch": 1.3225859652016163, + "grad_norm": 2.0438570976257324, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7718378305435181, + "num_tokens": 372860390.0, + "step": 16039 + }, + { + "epoch": 1.3226684258266679, + "grad_norm": 1.973338007926941, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7638791799545288, + "num_tokens": 372891969.0, + "step": 16040 + }, + { + "epoch": 1.3227508864517192, + "grad_norm": 1.787443995475769, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.7890163660049438, + "num_tokens": 372926007.0, + "step": 16041 + }, + { + "epoch": 1.3228333470767708, + "grad_norm": 2.1113522052764893, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7482208609580994, + "num_tokens": 372955859.0, + "step": 16042 + }, + { + "epoch": 1.3229158077018224, + "grad_norm": 2.2317116260528564, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7058680653572083, + "num_tokens": 372983422.0, + "step": 16043 + }, + { + "epoch": 1.322998268326874, + "grad_norm": 2.0041284561157227, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7721319198608398, + "num_tokens": 373013573.0, + "step": 16044 + }, + { + "epoch": 1.3230807289519255, + "grad_norm": 2.0795786380767822, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7710040211677551, + "num_tokens": 373039889.0, + "step": 16045 + }, + { + "epoch": 1.3231631895769769, + "grad_norm": 2.129528045654297, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7476827502250671, + "num_tokens": 373070923.0, + "step": 16046 + }, + { + "epoch": 1.3232456502020287, + "grad_norm": 2.046748638153076, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7319597005844116, + "num_tokens": 373098204.0, + "step": 16047 + }, + { + "epoch": 1.32332811082708, + "grad_norm": 2.220412492752075, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7777993679046631, + "num_tokens": 373122903.0, + "step": 16048 + }, + { + "epoch": 1.3234105714521316, + "grad_norm": 1.9610613584518433, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7263971567153931, + "num_tokens": 373154084.0, + "step": 16049 + }, + { + "epoch": 1.3234930320771832, + "grad_norm": 2.022411823272705, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7737569212913513, + "num_tokens": 373185106.0, + "step": 16050 + }, + { + "epoch": 1.3235754927022347, + "grad_norm": 2.262439012527466, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7656615972518921, + "num_tokens": 373209422.0, + "step": 16051 + }, + { + "epoch": 1.3236579533272863, + "grad_norm": 2.19148588180542, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7356546521186829, + "num_tokens": 373236059.0, + "step": 16052 + }, + { + "epoch": 1.3237404139523377, + "grad_norm": 2.1561834812164307, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7412502765655518, + "num_tokens": 373264845.0, + "step": 16053 + }, + { + "epoch": 1.3238228745773892, + "grad_norm": 2.1144230365753174, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7672753930091858, + "num_tokens": 373292553.0, + "step": 16054 + }, + { + "epoch": 1.3239053352024408, + "grad_norm": 2.045945882797241, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.746512770652771, + "num_tokens": 373326526.0, + "step": 16055 + }, + { + "epoch": 1.3239877958274924, + "grad_norm": 1.8428243398666382, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7673906683921814, + "num_tokens": 373360586.0, + "step": 16056 + }, + { + "epoch": 1.324070256452544, + "grad_norm": 1.9588565826416016, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7403407096862793, + "num_tokens": 373392018.0, + "step": 16057 + }, + { + "epoch": 1.3241527170775955, + "grad_norm": 1.8637378215789795, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7838939428329468, + "num_tokens": 373425139.0, + "step": 16058 + }, + { + "epoch": 1.324235177702647, + "grad_norm": 1.9505584239959717, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7530153393745422, + "num_tokens": 373458537.0, + "step": 16059 + }, + { + "epoch": 1.3243176383276984, + "grad_norm": 1.8976335525512695, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7778682112693787, + "num_tokens": 373490880.0, + "step": 16060 + }, + { + "epoch": 1.32440009895275, + "grad_norm": 2.3069472312927246, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7403992414474487, + "num_tokens": 373514673.0, + "step": 16061 + }, + { + "epoch": 1.3244825595778016, + "grad_norm": 1.8117330074310303, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7553991079330444, + "num_tokens": 373549092.0, + "step": 16062 + }, + { + "epoch": 1.3245650202028532, + "grad_norm": 1.796120285987854, + "learning_rate": 1e-06, + "loss": 0.6251, + "mean_token_accuracy": 0.7900621891021729, + "num_tokens": 373583470.0, + "step": 16063 + }, + { + "epoch": 1.3246474808279047, + "grad_norm": 1.840067744255066, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7870355248451233, + "num_tokens": 373615094.0, + "step": 16064 + }, + { + "epoch": 1.324729941452956, + "grad_norm": 1.8480035066604614, + "learning_rate": 1e-06, + "loss": 0.6054, + "mean_token_accuracy": 0.7970349788665771, + "num_tokens": 373646675.0, + "step": 16065 + }, + { + "epoch": 1.3248124020780079, + "grad_norm": 1.8854992389678955, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7601242661476135, + "num_tokens": 373677705.0, + "step": 16066 + }, + { + "epoch": 1.3248948627030592, + "grad_norm": 2.1040005683898926, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7336956262588501, + "num_tokens": 373706810.0, + "step": 16067 + }, + { + "epoch": 1.3249773233281108, + "grad_norm": 2.337233543395996, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7104317545890808, + "num_tokens": 373729652.0, + "step": 16068 + }, + { + "epoch": 1.3250597839531624, + "grad_norm": 1.695625901222229, + "learning_rate": 1e-06, + "loss": 0.5663, + "mean_token_accuracy": 0.8198016285896301, + "num_tokens": 373769184.0, + "step": 16069 + }, + { + "epoch": 1.325142244578214, + "grad_norm": 1.8634346723556519, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7246215343475342, + "num_tokens": 373803830.0, + "step": 16070 + }, + { + "epoch": 1.3252247052032655, + "grad_norm": 2.09741473197937, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7474784851074219, + "num_tokens": 373831281.0, + "step": 16071 + }, + { + "epoch": 1.3253071658283169, + "grad_norm": 2.1604979038238525, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7602289915084839, + "num_tokens": 373856917.0, + "step": 16072 + }, + { + "epoch": 1.3253896264533684, + "grad_norm": 2.1213624477386475, + "learning_rate": 1e-06, + "loss": 0.6226, + "mean_token_accuracy": 0.7988417744636536, + "num_tokens": 373880690.0, + "step": 16073 + }, + { + "epoch": 1.32547208707842, + "grad_norm": 1.9443869590759277, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7636326551437378, + "num_tokens": 373915233.0, + "step": 16074 + }, + { + "epoch": 1.3255545477034716, + "grad_norm": 1.8257962465286255, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7757867574691772, + "num_tokens": 373952449.0, + "step": 16075 + }, + { + "epoch": 1.3256370083285232, + "grad_norm": 2.1764473915100098, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7379566431045532, + "num_tokens": 373979189.0, + "step": 16076 + }, + { + "epoch": 1.3257194689535747, + "grad_norm": 1.6940481662750244, + "learning_rate": 1e-06, + "loss": 0.6192, + "mean_token_accuracy": 0.7924190759658813, + "num_tokens": 374016857.0, + "step": 16077 + }, + { + "epoch": 1.3258019295786263, + "grad_norm": 2.250525951385498, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7218189239501953, + "num_tokens": 374043268.0, + "step": 16078 + }, + { + "epoch": 1.3258843902036777, + "grad_norm": 1.9450443983078003, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7698566317558289, + "num_tokens": 374073554.0, + "step": 16079 + }, + { + "epoch": 1.3259668508287292, + "grad_norm": 1.8842756748199463, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7628815770149231, + "num_tokens": 374103579.0, + "step": 16080 + }, + { + "epoch": 1.3260493114537808, + "grad_norm": 1.9300026893615723, + "learning_rate": 1e-06, + "loss": 0.5203, + "mean_token_accuracy": 0.8238085508346558, + "num_tokens": 374132427.0, + "step": 16081 + }, + { + "epoch": 1.3261317720788324, + "grad_norm": 2.005187749862671, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7638765573501587, + "num_tokens": 374162972.0, + "step": 16082 + }, + { + "epoch": 1.326214232703884, + "grad_norm": 1.9454833269119263, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7896946668624878, + "num_tokens": 374192949.0, + "step": 16083 + }, + { + "epoch": 1.3262966933289355, + "grad_norm": 2.2086894512176514, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7311367392539978, + "num_tokens": 374217871.0, + "step": 16084 + }, + { + "epoch": 1.326379153953987, + "grad_norm": 1.799828052520752, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7747961282730103, + "num_tokens": 374251827.0, + "step": 16085 + }, + { + "epoch": 1.3264616145790384, + "grad_norm": 1.7805722951889038, + "learning_rate": 1e-06, + "loss": 0.6086, + "mean_token_accuracy": 0.8050161600112915, + "num_tokens": 374285960.0, + "step": 16086 + }, + { + "epoch": 1.32654407520409, + "grad_norm": 1.9949692487716675, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7277016639709473, + "num_tokens": 374316693.0, + "step": 16087 + }, + { + "epoch": 1.3266265358291416, + "grad_norm": 2.1175084114074707, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7467783689498901, + "num_tokens": 374344837.0, + "step": 16088 + }, + { + "epoch": 1.3267089964541932, + "grad_norm": 2.009490728378296, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.798399806022644, + "num_tokens": 374370696.0, + "step": 16089 + }, + { + "epoch": 1.3267914570792447, + "grad_norm": 2.229139566421509, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7488386631011963, + "num_tokens": 374395924.0, + "step": 16090 + }, + { + "epoch": 1.326873917704296, + "grad_norm": 1.8925429582595825, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7614136934280396, + "num_tokens": 374426435.0, + "step": 16091 + }, + { + "epoch": 1.3269563783293479, + "grad_norm": 1.9996721744537354, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7394369840621948, + "num_tokens": 374458098.0, + "step": 16092 + }, + { + "epoch": 1.3270388389543992, + "grad_norm": 2.0013582706451416, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7723982334136963, + "num_tokens": 374488781.0, + "step": 16093 + }, + { + "epoch": 1.3271212995794508, + "grad_norm": 1.937745213508606, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7227485179901123, + "num_tokens": 374521969.0, + "step": 16094 + }, + { + "epoch": 1.3272037602045024, + "grad_norm": 2.226314067840576, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7670634984970093, + "num_tokens": 374547377.0, + "step": 16095 + }, + { + "epoch": 1.327286220829554, + "grad_norm": 2.2333900928497314, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7432395219802856, + "num_tokens": 374572808.0, + "step": 16096 + }, + { + "epoch": 1.3273686814546055, + "grad_norm": 1.7968920469284058, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7419702410697937, + "num_tokens": 374609652.0, + "step": 16097 + }, + { + "epoch": 1.3274511420796569, + "grad_norm": 2.2309579849243164, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7454988956451416, + "num_tokens": 374634328.0, + "step": 16098 + }, + { + "epoch": 1.3275336027047084, + "grad_norm": 2.0097496509552, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7677674293518066, + "num_tokens": 374661879.0, + "step": 16099 + }, + { + "epoch": 1.32761606332976, + "grad_norm": 1.8677865266799927, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7751087546348572, + "num_tokens": 374695229.0, + "step": 16100 + }, + { + "epoch": 1.3276985239548116, + "grad_norm": 2.037757396697998, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7724409103393555, + "num_tokens": 374725162.0, + "step": 16101 + }, + { + "epoch": 1.3277809845798632, + "grad_norm": 2.011958599090576, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7745222449302673, + "num_tokens": 374753974.0, + "step": 16102 + }, + { + "epoch": 1.3278634452049147, + "grad_norm": 1.9223706722259521, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.7887287139892578, + "num_tokens": 374783882.0, + "step": 16103 + }, + { + "epoch": 1.3279459058299663, + "grad_norm": 1.942210078239441, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7659288644790649, + "num_tokens": 374813730.0, + "step": 16104 + }, + { + "epoch": 1.3280283664550177, + "grad_norm": 2.191709280014038, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.7431979179382324, + "num_tokens": 374838703.0, + "step": 16105 + }, + { + "epoch": 1.3281108270800692, + "grad_norm": 2.008406400680542, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7589073181152344, + "num_tokens": 374870752.0, + "step": 16106 + }, + { + "epoch": 1.3281932877051208, + "grad_norm": 2.2803919315338135, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7379358410835266, + "num_tokens": 374895121.0, + "step": 16107 + }, + { + "epoch": 1.3282757483301724, + "grad_norm": 1.9970897436141968, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7680842876434326, + "num_tokens": 374923247.0, + "step": 16108 + }, + { + "epoch": 1.328358208955224, + "grad_norm": 1.7633174657821655, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7416772246360779, + "num_tokens": 374963431.0, + "step": 16109 + }, + { + "epoch": 1.3284406695802753, + "grad_norm": 1.9029603004455566, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7479816675186157, + "num_tokens": 374996851.0, + "step": 16110 + }, + { + "epoch": 1.328523130205327, + "grad_norm": 2.167661666870117, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7609209418296814, + "num_tokens": 375025298.0, + "step": 16111 + }, + { + "epoch": 1.3286055908303784, + "grad_norm": 2.2092490196228027, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.7755230665206909, + "num_tokens": 375050391.0, + "step": 16112 + }, + { + "epoch": 1.32868805145543, + "grad_norm": 2.3389618396759033, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7401729822158813, + "num_tokens": 375074338.0, + "step": 16113 + }, + { + "epoch": 1.3287705120804816, + "grad_norm": 1.8396164178848267, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7777615785598755, + "num_tokens": 375106417.0, + "step": 16114 + }, + { + "epoch": 1.3288529727055332, + "grad_norm": 1.9373329877853394, + "learning_rate": 1e-06, + "loss": 0.7673, + "mean_token_accuracy": 0.7598933577537537, + "num_tokens": 375139069.0, + "step": 16115 + }, + { + "epoch": 1.3289354333305847, + "grad_norm": 1.8895237445831299, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7715145349502563, + "num_tokens": 375170250.0, + "step": 16116 + }, + { + "epoch": 1.329017893955636, + "grad_norm": 2.029231548309326, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7248678803443909, + "num_tokens": 375201295.0, + "step": 16117 + }, + { + "epoch": 1.3291003545806876, + "grad_norm": 2.007232666015625, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.7929813265800476, + "num_tokens": 375227585.0, + "step": 16118 + }, + { + "epoch": 1.3291828152057392, + "grad_norm": 2.4643454551696777, + "learning_rate": 1e-06, + "loss": 0.615, + "mean_token_accuracy": 0.7902206778526306, + "num_tokens": 375249761.0, + "step": 16119 + }, + { + "epoch": 1.3292652758307908, + "grad_norm": 2.5883047580718994, + "learning_rate": 1e-06, + "loss": 0.7723, + "mean_token_accuracy": 0.7483603954315186, + "num_tokens": 375268766.0, + "step": 16120 + }, + { + "epoch": 1.3293477364558424, + "grad_norm": 1.9119879007339478, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7673228979110718, + "num_tokens": 375299558.0, + "step": 16121 + }, + { + "epoch": 1.329430197080894, + "grad_norm": 2.009281635284424, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.729415237903595, + "num_tokens": 375328556.0, + "step": 16122 + }, + { + "epoch": 1.3295126577059455, + "grad_norm": 2.2054808139801025, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7672027349472046, + "num_tokens": 375353014.0, + "step": 16123 + }, + { + "epoch": 1.3295951183309969, + "grad_norm": 2.0485241413116455, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7181363701820374, + "num_tokens": 375383264.0, + "step": 16124 + }, + { + "epoch": 1.3296775789560484, + "grad_norm": 1.9270401000976562, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.7926340103149414, + "num_tokens": 375412383.0, + "step": 16125 + }, + { + "epoch": 1.3297600395811, + "grad_norm": 1.902400016784668, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7364046573638916, + "num_tokens": 375444926.0, + "step": 16126 + }, + { + "epoch": 1.3298425002061516, + "grad_norm": 2.0539937019348145, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7586561441421509, + "num_tokens": 375473830.0, + "step": 16127 + }, + { + "epoch": 1.3299249608312032, + "grad_norm": 2.0628459453582764, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7356888651847839, + "num_tokens": 375506195.0, + "step": 16128 + }, + { + "epoch": 1.3300074214562545, + "grad_norm": 1.8897737264633179, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7478737831115723, + "num_tokens": 375544549.0, + "step": 16129 + }, + { + "epoch": 1.3300898820813063, + "grad_norm": 1.9249132871627808, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7417515516281128, + "num_tokens": 375576422.0, + "step": 16130 + }, + { + "epoch": 1.3301723427063576, + "grad_norm": 2.277839422225952, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7588099241256714, + "num_tokens": 375602222.0, + "step": 16131 + }, + { + "epoch": 1.3302548033314092, + "grad_norm": 2.0496811866760254, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7355936765670776, + "num_tokens": 375633779.0, + "step": 16132 + }, + { + "epoch": 1.3303372639564608, + "grad_norm": 1.920008659362793, + "learning_rate": 1e-06, + "loss": 0.6029, + "mean_token_accuracy": 0.7901322841644287, + "num_tokens": 375662761.0, + "step": 16133 + }, + { + "epoch": 1.3304197245815124, + "grad_norm": 2.113132953643799, + "learning_rate": 1e-06, + "loss": 0.7396, + "mean_token_accuracy": 0.7593555450439453, + "num_tokens": 375694338.0, + "step": 16134 + }, + { + "epoch": 1.330502185206564, + "grad_norm": 1.952429175376892, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7872717380523682, + "num_tokens": 375726776.0, + "step": 16135 + }, + { + "epoch": 1.3305846458316153, + "grad_norm": 2.1965131759643555, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7826626300811768, + "num_tokens": 375751818.0, + "step": 16136 + }, + { + "epoch": 1.3306671064566669, + "grad_norm": 2.037869691848755, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7663739919662476, + "num_tokens": 375779129.0, + "step": 16137 + }, + { + "epoch": 1.3307495670817184, + "grad_norm": 1.9874018430709839, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7811060547828674, + "num_tokens": 375811362.0, + "step": 16138 + }, + { + "epoch": 1.33083202770677, + "grad_norm": 2.003887891769409, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7697997689247131, + "num_tokens": 375838275.0, + "step": 16139 + }, + { + "epoch": 1.3309144883318216, + "grad_norm": 1.9287770986557007, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.7972465753555298, + "num_tokens": 375866789.0, + "step": 16140 + }, + { + "epoch": 1.3309969489568731, + "grad_norm": 1.7202986478805542, + "learning_rate": 1e-06, + "loss": 0.5964, + "mean_token_accuracy": 0.8037662506103516, + "num_tokens": 375900606.0, + "step": 16141 + }, + { + "epoch": 1.3310794095819247, + "grad_norm": 2.208141565322876, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.739241898059845, + "num_tokens": 375927062.0, + "step": 16142 + }, + { + "epoch": 1.331161870206976, + "grad_norm": 1.80305016040802, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7757580280303955, + "num_tokens": 375963653.0, + "step": 16143 + }, + { + "epoch": 1.3312443308320276, + "grad_norm": 2.2149906158447266, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.7897615432739258, + "num_tokens": 375986863.0, + "step": 16144 + }, + { + "epoch": 1.3313267914570792, + "grad_norm": 2.1225836277008057, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7543494701385498, + "num_tokens": 376014620.0, + "step": 16145 + }, + { + "epoch": 1.3314092520821308, + "grad_norm": 1.9197081327438354, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7498323917388916, + "num_tokens": 376045560.0, + "step": 16146 + }, + { + "epoch": 1.3314917127071824, + "grad_norm": 1.884128212928772, + "learning_rate": 1e-06, + "loss": 0.786, + "mean_token_accuracy": 0.7576507329940796, + "num_tokens": 376081297.0, + "step": 16147 + }, + { + "epoch": 1.331574173332234, + "grad_norm": 2.220820665359497, + "learning_rate": 1e-06, + "loss": 0.7821, + "mean_token_accuracy": 0.7500339150428772, + "num_tokens": 376106530.0, + "step": 16148 + }, + { + "epoch": 1.3316566339572855, + "grad_norm": 2.277885675430298, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7733731269836426, + "num_tokens": 376130853.0, + "step": 16149 + }, + { + "epoch": 1.3317390945823369, + "grad_norm": 1.9032962322235107, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7565765380859375, + "num_tokens": 376163846.0, + "step": 16150 + }, + { + "epoch": 1.3318215552073884, + "grad_norm": 2.2080416679382324, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7397117614746094, + "num_tokens": 376189247.0, + "step": 16151 + }, + { + "epoch": 1.33190401583244, + "grad_norm": 1.6475361585617065, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.7735378742218018, + "num_tokens": 376228423.0, + "step": 16152 + }, + { + "epoch": 1.3319864764574916, + "grad_norm": 2.141116142272949, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7769505977630615, + "num_tokens": 376258944.0, + "step": 16153 + }, + { + "epoch": 1.3320689370825431, + "grad_norm": 2.1137149333953857, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7461031079292297, + "num_tokens": 376289461.0, + "step": 16154 + }, + { + "epoch": 1.3321513977075945, + "grad_norm": 2.1522817611694336, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7609585523605347, + "num_tokens": 376316918.0, + "step": 16155 + }, + { + "epoch": 1.3322338583326463, + "grad_norm": 2.125279664993286, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7561444044113159, + "num_tokens": 376343092.0, + "step": 16156 + }, + { + "epoch": 1.3323163189576976, + "grad_norm": 2.112720012664795, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.742072343826294, + "num_tokens": 376371176.0, + "step": 16157 + }, + { + "epoch": 1.3323987795827492, + "grad_norm": 1.819196105003357, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.7827897071838379, + "num_tokens": 376403561.0, + "step": 16158 + }, + { + "epoch": 1.3324812402078008, + "grad_norm": 2.117201805114746, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7448405623435974, + "num_tokens": 376431091.0, + "step": 16159 + }, + { + "epoch": 1.3325637008328524, + "grad_norm": 1.9040213823318481, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7489112615585327, + "num_tokens": 376463585.0, + "step": 16160 + }, + { + "epoch": 1.332646161457904, + "grad_norm": 1.9615050554275513, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7713296413421631, + "num_tokens": 376492999.0, + "step": 16161 + }, + { + "epoch": 1.3327286220829553, + "grad_norm": 1.8074346780776978, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7703132629394531, + "num_tokens": 376526068.0, + "step": 16162 + }, + { + "epoch": 1.3328110827080069, + "grad_norm": 2.136108875274658, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7402398586273193, + "num_tokens": 376556030.0, + "step": 16163 + }, + { + "epoch": 1.3328935433330584, + "grad_norm": 1.9266997575759888, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.7861427068710327, + "num_tokens": 376587204.0, + "step": 16164 + }, + { + "epoch": 1.33297600395811, + "grad_norm": 1.853514313697815, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.7813512086868286, + "num_tokens": 376621151.0, + "step": 16165 + }, + { + "epoch": 1.3330584645831616, + "grad_norm": 2.0407965183258057, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7868820428848267, + "num_tokens": 376649244.0, + "step": 16166 + }, + { + "epoch": 1.3331409252082131, + "grad_norm": 2.251460313796997, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7415441274642944, + "num_tokens": 376676205.0, + "step": 16167 + }, + { + "epoch": 1.3332233858332647, + "grad_norm": 2.162172555923462, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7377437949180603, + "num_tokens": 376703168.0, + "step": 16168 + }, + { + "epoch": 1.333305846458316, + "grad_norm": 1.8861521482467651, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7431304454803467, + "num_tokens": 376737747.0, + "step": 16169 + }, + { + "epoch": 1.3333883070833676, + "grad_norm": 2.07881498336792, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7665070295333862, + "num_tokens": 376764338.0, + "step": 16170 + }, + { + "epoch": 1.3334707677084192, + "grad_norm": 1.87348210811615, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7646852731704712, + "num_tokens": 376800378.0, + "step": 16171 + }, + { + "epoch": 1.3335532283334708, + "grad_norm": 2.040602922439575, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7665332555770874, + "num_tokens": 376827155.0, + "step": 16172 + }, + { + "epoch": 1.3336356889585224, + "grad_norm": 1.727223515510559, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.7873022556304932, + "num_tokens": 376864705.0, + "step": 16173 + }, + { + "epoch": 1.3337181495835737, + "grad_norm": 2.0528993606567383, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.7896118760108948, + "num_tokens": 376890947.0, + "step": 16174 + }, + { + "epoch": 1.3338006102086255, + "grad_norm": 1.7155457735061646, + "learning_rate": 1e-06, + "loss": 0.5436, + "mean_token_accuracy": 0.816166877746582, + "num_tokens": 376925519.0, + "step": 16175 + }, + { + "epoch": 1.3338830708336769, + "grad_norm": 1.7529700994491577, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7865148782730103, + "num_tokens": 376960947.0, + "step": 16176 + }, + { + "epoch": 1.3339655314587284, + "grad_norm": 2.151806116104126, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7540954947471619, + "num_tokens": 376989386.0, + "step": 16177 + }, + { + "epoch": 1.33404799208378, + "grad_norm": 1.7136311531066895, + "learning_rate": 1e-06, + "loss": 0.512, + "mean_token_accuracy": 0.8285800814628601, + "num_tokens": 377023318.0, + "step": 16178 + }, + { + "epoch": 1.3341304527088316, + "grad_norm": 2.0285115242004395, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7391586303710938, + "num_tokens": 377054409.0, + "step": 16179 + }, + { + "epoch": 1.3342129133338831, + "grad_norm": 2.192653179168701, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7223100066184998, + "num_tokens": 377081669.0, + "step": 16180 + }, + { + "epoch": 1.3342953739589345, + "grad_norm": 2.333735227584839, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7302300333976746, + "num_tokens": 377105502.0, + "step": 16181 + }, + { + "epoch": 1.334377834583986, + "grad_norm": 2.0120089054107666, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.7835944294929504, + "num_tokens": 377135800.0, + "step": 16182 + }, + { + "epoch": 1.3344602952090376, + "grad_norm": 1.7810957431793213, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7401079535484314, + "num_tokens": 377174638.0, + "step": 16183 + }, + { + "epoch": 1.3345427558340892, + "grad_norm": 1.8883599042892456, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.7988826632499695, + "num_tokens": 377205570.0, + "step": 16184 + }, + { + "epoch": 1.3346252164591408, + "grad_norm": 2.1124699115753174, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.7801406383514404, + "num_tokens": 377231616.0, + "step": 16185 + }, + { + "epoch": 1.3347076770841924, + "grad_norm": 2.002436399459839, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7448383569717407, + "num_tokens": 377262026.0, + "step": 16186 + }, + { + "epoch": 1.334790137709244, + "grad_norm": 1.9031658172607422, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.7842514514923096, + "num_tokens": 377294671.0, + "step": 16187 + }, + { + "epoch": 1.3348725983342953, + "grad_norm": 2.1874265670776367, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7397168874740601, + "num_tokens": 377322000.0, + "step": 16188 + }, + { + "epoch": 1.3349550589593469, + "grad_norm": 2.4717535972595215, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.702728807926178, + "num_tokens": 377345187.0, + "step": 16189 + }, + { + "epoch": 1.3350375195843984, + "grad_norm": 2.37976336479187, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7685321569442749, + "num_tokens": 377366869.0, + "step": 16190 + }, + { + "epoch": 1.33511998020945, + "grad_norm": 2.260361433029175, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7555568218231201, + "num_tokens": 377391714.0, + "step": 16191 + }, + { + "epoch": 1.3352024408345016, + "grad_norm": 2.1252431869506836, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7712892293930054, + "num_tokens": 377422454.0, + "step": 16192 + }, + { + "epoch": 1.3352849014595531, + "grad_norm": 1.8947436809539795, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.775542676448822, + "num_tokens": 377456122.0, + "step": 16193 + }, + { + "epoch": 1.3353673620846047, + "grad_norm": 2.0280072689056396, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.7630256414413452, + "num_tokens": 377486822.0, + "step": 16194 + }, + { + "epoch": 1.335449822709656, + "grad_norm": 2.061419725418091, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7367824912071228, + "num_tokens": 377514925.0, + "step": 16195 + }, + { + "epoch": 1.3355322833347076, + "grad_norm": 1.9507440328598022, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7742379903793335, + "num_tokens": 377545115.0, + "step": 16196 + }, + { + "epoch": 1.3356147439597592, + "grad_norm": 1.833048939704895, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7631047964096069, + "num_tokens": 377579719.0, + "step": 16197 + }, + { + "epoch": 1.3356972045848108, + "grad_norm": 1.7036527395248413, + "learning_rate": 1e-06, + "loss": 0.5002, + "mean_token_accuracy": 0.8334041237831116, + "num_tokens": 377613277.0, + "step": 16198 + }, + { + "epoch": 1.3357796652098624, + "grad_norm": 2.263664484024048, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7721145153045654, + "num_tokens": 377638014.0, + "step": 16199 + }, + { + "epoch": 1.3358621258349137, + "grad_norm": 2.0859262943267822, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7231703400611877, + "num_tokens": 377669647.0, + "step": 16200 + }, + { + "epoch": 1.3359445864599655, + "grad_norm": 2.1943538188934326, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7537499070167542, + "num_tokens": 377699498.0, + "step": 16201 + }, + { + "epoch": 1.3360270470850169, + "grad_norm": 1.7435461282730103, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7649266719818115, + "num_tokens": 377736324.0, + "step": 16202 + }, + { + "epoch": 1.3361095077100684, + "grad_norm": 1.703573226928711, + "learning_rate": 1e-06, + "loss": 0.6012, + "mean_token_accuracy": 0.8036551475524902, + "num_tokens": 377775616.0, + "step": 16203 + }, + { + "epoch": 1.33619196833512, + "grad_norm": 1.782634973526001, + "learning_rate": 1e-06, + "loss": 0.6077, + "mean_token_accuracy": 0.79996258020401, + "num_tokens": 377809376.0, + "step": 16204 + }, + { + "epoch": 1.3362744289601716, + "grad_norm": 2.310321092605591, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7319734692573547, + "num_tokens": 377832838.0, + "step": 16205 + }, + { + "epoch": 1.3363568895852231, + "grad_norm": 2.0023653507232666, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.771997332572937, + "num_tokens": 377863173.0, + "step": 16206 + }, + { + "epoch": 1.3364393502102745, + "grad_norm": 2.1052088737487793, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7827284336090088, + "num_tokens": 377889168.0, + "step": 16207 + }, + { + "epoch": 1.336521810835326, + "grad_norm": 2.0844168663024902, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7618257999420166, + "num_tokens": 377917304.0, + "step": 16208 + }, + { + "epoch": 1.3366042714603776, + "grad_norm": 2.0949552059173584, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7560598850250244, + "num_tokens": 377944414.0, + "step": 16209 + }, + { + "epoch": 1.3366867320854292, + "grad_norm": 2.157012462615967, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7399081587791443, + "num_tokens": 377972474.0, + "step": 16210 + }, + { + "epoch": 1.3367691927104808, + "grad_norm": 2.1454570293426514, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7650012969970703, + "num_tokens": 377998790.0, + "step": 16211 + }, + { + "epoch": 1.3368516533355324, + "grad_norm": 1.9746981859207153, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7593451142311096, + "num_tokens": 378026453.0, + "step": 16212 + }, + { + "epoch": 1.336934113960584, + "grad_norm": 2.4104456901550293, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.766109824180603, + "num_tokens": 378049443.0, + "step": 16213 + }, + { + "epoch": 1.3370165745856353, + "grad_norm": 2.047053575515747, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7795619368553162, + "num_tokens": 378076329.0, + "step": 16214 + }, + { + "epoch": 1.3370990352106868, + "grad_norm": 1.8120750188827515, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7768963575363159, + "num_tokens": 378110403.0, + "step": 16215 + }, + { + "epoch": 1.3371814958357384, + "grad_norm": 1.8809016942977905, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.744663655757904, + "num_tokens": 378147947.0, + "step": 16216 + }, + { + "epoch": 1.33726395646079, + "grad_norm": 1.925188422203064, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7766205072402954, + "num_tokens": 378178304.0, + "step": 16217 + }, + { + "epoch": 1.3373464170858416, + "grad_norm": 2.2150988578796387, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7673231959342957, + "num_tokens": 378200896.0, + "step": 16218 + }, + { + "epoch": 1.337428877710893, + "grad_norm": 2.0085196495056152, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7288082838058472, + "num_tokens": 378231542.0, + "step": 16219 + }, + { + "epoch": 1.3375113383359447, + "grad_norm": 1.8867584466934204, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7610272169113159, + "num_tokens": 378265655.0, + "step": 16220 + }, + { + "epoch": 1.337593798960996, + "grad_norm": 1.7414910793304443, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7862833738327026, + "num_tokens": 378303638.0, + "step": 16221 + }, + { + "epoch": 1.3376762595860476, + "grad_norm": 2.357818365097046, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.763014554977417, + "num_tokens": 378326503.0, + "step": 16222 + }, + { + "epoch": 1.3377587202110992, + "grad_norm": 2.052826166152954, + "learning_rate": 1e-06, + "loss": 0.6082, + "mean_token_accuracy": 0.7944424152374268, + "num_tokens": 378353523.0, + "step": 16223 + }, + { + "epoch": 1.3378411808361508, + "grad_norm": 2.043365955352783, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7506688833236694, + "num_tokens": 378381676.0, + "step": 16224 + }, + { + "epoch": 1.3379236414612024, + "grad_norm": 2.1292355060577393, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7765823602676392, + "num_tokens": 378407616.0, + "step": 16225 + }, + { + "epoch": 1.3380061020862537, + "grad_norm": 2.4975364208221436, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7186736464500427, + "num_tokens": 378428455.0, + "step": 16226 + }, + { + "epoch": 1.3380885627113053, + "grad_norm": 1.9686453342437744, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7596412897109985, + "num_tokens": 378458800.0, + "step": 16227 + }, + { + "epoch": 1.3381710233363568, + "grad_norm": 2.2405271530151367, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7670893669128418, + "num_tokens": 378484068.0, + "step": 16228 + }, + { + "epoch": 1.3382534839614084, + "grad_norm": 2.218381643295288, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.7558214068412781, + "num_tokens": 378510178.0, + "step": 16229 + }, + { + "epoch": 1.33833594458646, + "grad_norm": 1.9924800395965576, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7464038133621216, + "num_tokens": 378541061.0, + "step": 16230 + }, + { + "epoch": 1.3384184052115116, + "grad_norm": 1.9024230241775513, + "learning_rate": 1e-06, + "loss": 0.6066, + "mean_token_accuracy": 0.7942357659339905, + "num_tokens": 378571940.0, + "step": 16231 + }, + { + "epoch": 1.3385008658365631, + "grad_norm": 1.8826910257339478, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7618420124053955, + "num_tokens": 378607750.0, + "step": 16232 + }, + { + "epoch": 1.3385833264616145, + "grad_norm": 1.9895175695419312, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7672703266143799, + "num_tokens": 378634080.0, + "step": 16233 + }, + { + "epoch": 1.338665787086666, + "grad_norm": 2.0261523723602295, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7396700382232666, + "num_tokens": 378663619.0, + "step": 16234 + }, + { + "epoch": 1.3387482477117176, + "grad_norm": 1.90935480594635, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7528570890426636, + "num_tokens": 378695754.0, + "step": 16235 + }, + { + "epoch": 1.3388307083367692, + "grad_norm": 1.9678133726119995, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7758842706680298, + "num_tokens": 378725891.0, + "step": 16236 + }, + { + "epoch": 1.3389131689618208, + "grad_norm": 2.177063226699829, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7376277446746826, + "num_tokens": 378750250.0, + "step": 16237 + }, + { + "epoch": 1.3389956295868723, + "grad_norm": 2.0694079399108887, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.740900993347168, + "num_tokens": 378779082.0, + "step": 16238 + }, + { + "epoch": 1.339078090211924, + "grad_norm": 2.0379140377044678, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.76862633228302, + "num_tokens": 378809665.0, + "step": 16239 + }, + { + "epoch": 1.3391605508369753, + "grad_norm": 1.9654889106750488, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.708948016166687, + "num_tokens": 378843037.0, + "step": 16240 + }, + { + "epoch": 1.3392430114620268, + "grad_norm": 2.2296574115753174, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7491987347602844, + "num_tokens": 378867894.0, + "step": 16241 + }, + { + "epoch": 1.3393254720870784, + "grad_norm": 2.146904468536377, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7591997981071472, + "num_tokens": 378895996.0, + "step": 16242 + }, + { + "epoch": 1.33940793271213, + "grad_norm": 1.7359483242034912, + "learning_rate": 1e-06, + "loss": 0.5632, + "mean_token_accuracy": 0.8080480694770813, + "num_tokens": 378927653.0, + "step": 16243 + }, + { + "epoch": 1.3394903933371816, + "grad_norm": 2.038175582885742, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7518367767333984, + "num_tokens": 378956463.0, + "step": 16244 + }, + { + "epoch": 1.339572853962233, + "grad_norm": 2.054685354232788, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7818636298179626, + "num_tokens": 378983487.0, + "step": 16245 + }, + { + "epoch": 1.3396553145872847, + "grad_norm": 2.2228283882141113, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7567563056945801, + "num_tokens": 379008674.0, + "step": 16246 + }, + { + "epoch": 1.339737775212336, + "grad_norm": 1.6323139667510986, + "learning_rate": 1e-06, + "loss": 0.5812, + "mean_token_accuracy": 0.8088921308517456, + "num_tokens": 379044621.0, + "step": 16247 + }, + { + "epoch": 1.3398202358373876, + "grad_norm": 1.7210637331008911, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7599343657493591, + "num_tokens": 379080966.0, + "step": 16248 + }, + { + "epoch": 1.3399026964624392, + "grad_norm": 1.8550716638565063, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.736839234828949, + "num_tokens": 379116961.0, + "step": 16249 + }, + { + "epoch": 1.3399851570874908, + "grad_norm": 2.0728061199188232, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7379329800605774, + "num_tokens": 379147700.0, + "step": 16250 + }, + { + "epoch": 1.3400676177125423, + "grad_norm": 2.2127647399902344, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7433837652206421, + "num_tokens": 379174342.0, + "step": 16251 + }, + { + "epoch": 1.3401500783375937, + "grad_norm": 2.058708667755127, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7401151657104492, + "num_tokens": 379206106.0, + "step": 16252 + }, + { + "epoch": 1.3402325389626453, + "grad_norm": 1.9924479722976685, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.7821211814880371, + "num_tokens": 379234339.0, + "step": 16253 + }, + { + "epoch": 1.3403149995876968, + "grad_norm": 1.8029179573059082, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.7881099581718445, + "num_tokens": 379269430.0, + "step": 16254 + }, + { + "epoch": 1.3403974602127484, + "grad_norm": 1.8128392696380615, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.7948420643806458, + "num_tokens": 379302065.0, + "step": 16255 + }, + { + "epoch": 1.3404799208378, + "grad_norm": 1.9682282209396362, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7773450016975403, + "num_tokens": 379333602.0, + "step": 16256 + }, + { + "epoch": 1.3405623814628516, + "grad_norm": 2.0150184631347656, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.753781795501709, + "num_tokens": 379364912.0, + "step": 16257 + }, + { + "epoch": 1.3406448420879031, + "grad_norm": 1.9999819993972778, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7682819366455078, + "num_tokens": 379397538.0, + "step": 16258 + }, + { + "epoch": 1.3407273027129545, + "grad_norm": 2.0597851276397705, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7407699823379517, + "num_tokens": 379427577.0, + "step": 16259 + }, + { + "epoch": 1.340809763338006, + "grad_norm": 1.9612889289855957, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7479916214942932, + "num_tokens": 379459128.0, + "step": 16260 + }, + { + "epoch": 1.3408922239630576, + "grad_norm": 1.9885402917861938, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7694674134254456, + "num_tokens": 379488842.0, + "step": 16261 + }, + { + "epoch": 1.3409746845881092, + "grad_norm": 1.7902894020080566, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7668954133987427, + "num_tokens": 379521483.0, + "step": 16262 + }, + { + "epoch": 1.3410571452131608, + "grad_norm": 1.958086609840393, + "learning_rate": 1e-06, + "loss": 0.5976, + "mean_token_accuracy": 0.8074925541877747, + "num_tokens": 379551385.0, + "step": 16263 + }, + { + "epoch": 1.3411396058382121, + "grad_norm": 1.9448304176330566, + "learning_rate": 1e-06, + "loss": 0.7585, + "mean_token_accuracy": 0.7621323466300964, + "num_tokens": 379582968.0, + "step": 16264 + }, + { + "epoch": 1.341222066463264, + "grad_norm": 1.9342877864837646, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.734923243522644, + "num_tokens": 379612156.0, + "step": 16265 + }, + { + "epoch": 1.3413045270883153, + "grad_norm": 1.9944359064102173, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7445683479309082, + "num_tokens": 379642694.0, + "step": 16266 + }, + { + "epoch": 1.3413869877133668, + "grad_norm": 2.064772367477417, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7393320798873901, + "num_tokens": 379670081.0, + "step": 16267 + }, + { + "epoch": 1.3414694483384184, + "grad_norm": 1.9179770946502686, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.7913969159126282, + "num_tokens": 379700706.0, + "step": 16268 + }, + { + "epoch": 1.34155190896347, + "grad_norm": 2.3280162811279297, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7595955729484558, + "num_tokens": 379726521.0, + "step": 16269 + }, + { + "epoch": 1.3416343695885216, + "grad_norm": 1.9241863489151, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7036842703819275, + "num_tokens": 379763639.0, + "step": 16270 + }, + { + "epoch": 1.341716830213573, + "grad_norm": 1.9007844924926758, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7315553426742554, + "num_tokens": 379797014.0, + "step": 16271 + }, + { + "epoch": 1.3417992908386245, + "grad_norm": 1.9158025979995728, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7663767337799072, + "num_tokens": 379829027.0, + "step": 16272 + }, + { + "epoch": 1.341881751463676, + "grad_norm": 2.177440881729126, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7339181900024414, + "num_tokens": 379855773.0, + "step": 16273 + }, + { + "epoch": 1.3419642120887276, + "grad_norm": 2.1814205646514893, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7661502361297607, + "num_tokens": 379880918.0, + "step": 16274 + }, + { + "epoch": 1.3420466727137792, + "grad_norm": 2.0495150089263916, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7284512519836426, + "num_tokens": 379915315.0, + "step": 16275 + }, + { + "epoch": 1.3421291333388308, + "grad_norm": 1.8144009113311768, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7576923370361328, + "num_tokens": 379952161.0, + "step": 16276 + }, + { + "epoch": 1.3422115939638823, + "grad_norm": 1.8160275220870972, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7580969333648682, + "num_tokens": 379984805.0, + "step": 16277 + }, + { + "epoch": 1.3422940545889337, + "grad_norm": 1.8238370418548584, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7674891352653503, + "num_tokens": 380019062.0, + "step": 16278 + }, + { + "epoch": 1.3423765152139853, + "grad_norm": 2.263439416885376, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7612600326538086, + "num_tokens": 380043343.0, + "step": 16279 + }, + { + "epoch": 1.3424589758390368, + "grad_norm": 2.110258102416992, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7238667011260986, + "num_tokens": 380077451.0, + "step": 16280 + }, + { + "epoch": 1.3425414364640884, + "grad_norm": 1.8714911937713623, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7475599050521851, + "num_tokens": 380109545.0, + "step": 16281 + }, + { + "epoch": 1.34262389708914, + "grad_norm": 2.1548421382904053, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.7802445292472839, + "num_tokens": 380138306.0, + "step": 16282 + }, + { + "epoch": 1.3427063577141916, + "grad_norm": 2.0960769653320312, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7848243117332458, + "num_tokens": 380164247.0, + "step": 16283 + }, + { + "epoch": 1.3427888183392431, + "grad_norm": 1.9628088474273682, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7470911145210266, + "num_tokens": 380192969.0, + "step": 16284 + }, + { + "epoch": 1.3428712789642945, + "grad_norm": 2.153489351272583, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7474353313446045, + "num_tokens": 380217730.0, + "step": 16285 + }, + { + "epoch": 1.342953739589346, + "grad_norm": 2.0646533966064453, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7676125168800354, + "num_tokens": 380243715.0, + "step": 16286 + }, + { + "epoch": 1.3430362002143976, + "grad_norm": 2.0559945106506348, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7690241932868958, + "num_tokens": 380274108.0, + "step": 16287 + }, + { + "epoch": 1.3431186608394492, + "grad_norm": 1.8544763326644897, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7518537044525146, + "num_tokens": 380307960.0, + "step": 16288 + }, + { + "epoch": 1.3432011214645008, + "grad_norm": 1.981626272201538, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.767427921295166, + "num_tokens": 380334702.0, + "step": 16289 + }, + { + "epoch": 1.3432835820895521, + "grad_norm": 2.102071523666382, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7781307697296143, + "num_tokens": 380360675.0, + "step": 16290 + }, + { + "epoch": 1.343366042714604, + "grad_norm": 1.9904823303222656, + "learning_rate": 1e-06, + "loss": 0.5412, + "mean_token_accuracy": 0.8194660544395447, + "num_tokens": 380390771.0, + "step": 16291 + }, + { + "epoch": 1.3434485033396553, + "grad_norm": 2.040609121322632, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.7479925155639648, + "num_tokens": 380419208.0, + "step": 16292 + }, + { + "epoch": 1.3435309639647068, + "grad_norm": 1.8827269077301025, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7802428007125854, + "num_tokens": 380453034.0, + "step": 16293 + }, + { + "epoch": 1.3436134245897584, + "grad_norm": 1.9620215892791748, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7363282442092896, + "num_tokens": 380485868.0, + "step": 16294 + }, + { + "epoch": 1.34369588521481, + "grad_norm": 2.1508703231811523, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7777074575424194, + "num_tokens": 380511388.0, + "step": 16295 + }, + { + "epoch": 1.3437783458398616, + "grad_norm": 2.1142399311065674, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7479071617126465, + "num_tokens": 380538456.0, + "step": 16296 + }, + { + "epoch": 1.343860806464913, + "grad_norm": 1.7700598239898682, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.795560359954834, + "num_tokens": 380573621.0, + "step": 16297 + }, + { + "epoch": 1.3439432670899645, + "grad_norm": 2.09089994430542, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7711520195007324, + "num_tokens": 380601110.0, + "step": 16298 + }, + { + "epoch": 1.344025727715016, + "grad_norm": 1.8381609916687012, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7484582662582397, + "num_tokens": 380636606.0, + "step": 16299 + }, + { + "epoch": 1.3441081883400676, + "grad_norm": 1.9313501119613647, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7512727379798889, + "num_tokens": 380668779.0, + "step": 16300 + }, + { + "epoch": 1.3441906489651192, + "grad_norm": 1.9972254037857056, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7404037714004517, + "num_tokens": 380699016.0, + "step": 16301 + }, + { + "epoch": 1.3442731095901708, + "grad_norm": 1.9237266778945923, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7726792097091675, + "num_tokens": 380729297.0, + "step": 16302 + }, + { + "epoch": 1.3443555702152223, + "grad_norm": 1.7755430936813354, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7622650861740112, + "num_tokens": 380763027.0, + "step": 16303 + }, + { + "epoch": 1.3444380308402737, + "grad_norm": 1.9633384943008423, + "learning_rate": 1e-06, + "loss": 0.7619, + "mean_token_accuracy": 0.7607746124267578, + "num_tokens": 380791948.0, + "step": 16304 + }, + { + "epoch": 1.3445204914653253, + "grad_norm": 1.9580793380737305, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7619025707244873, + "num_tokens": 380822986.0, + "step": 16305 + }, + { + "epoch": 1.3446029520903768, + "grad_norm": 1.6978468894958496, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7379508018493652, + "num_tokens": 380860755.0, + "step": 16306 + }, + { + "epoch": 1.3446854127154284, + "grad_norm": 2.567068338394165, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7223374843597412, + "num_tokens": 380882560.0, + "step": 16307 + }, + { + "epoch": 1.34476787334048, + "grad_norm": 2.24546480178833, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7590720653533936, + "num_tokens": 380906950.0, + "step": 16308 + }, + { + "epoch": 1.3448503339655313, + "grad_norm": 2.176841974258423, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7701402306556702, + "num_tokens": 380934215.0, + "step": 16309 + }, + { + "epoch": 1.3449327945905831, + "grad_norm": 2.179123878479004, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7341386079788208, + "num_tokens": 380960128.0, + "step": 16310 + }, + { + "epoch": 1.3450152552156345, + "grad_norm": 2.001918315887451, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7626622915267944, + "num_tokens": 380989728.0, + "step": 16311 + }, + { + "epoch": 1.345097715840686, + "grad_norm": 1.942278504371643, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7562122344970703, + "num_tokens": 381024408.0, + "step": 16312 + }, + { + "epoch": 1.3451801764657376, + "grad_norm": 2.1438164710998535, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.7824780344963074, + "num_tokens": 381049350.0, + "step": 16313 + }, + { + "epoch": 1.3452626370907892, + "grad_norm": 1.9405596256256104, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.7912498116493225, + "num_tokens": 381079381.0, + "step": 16314 + }, + { + "epoch": 1.3453450977158408, + "grad_norm": 1.9445478916168213, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7706418037414551, + "num_tokens": 381107169.0, + "step": 16315 + }, + { + "epoch": 1.3454275583408921, + "grad_norm": 2.1088171005249023, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7553185820579529, + "num_tokens": 381135456.0, + "step": 16316 + }, + { + "epoch": 1.3455100189659437, + "grad_norm": 1.9350885152816772, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7544091939926147, + "num_tokens": 381165279.0, + "step": 16317 + }, + { + "epoch": 1.3455924795909953, + "grad_norm": 1.976593255996704, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7703962326049805, + "num_tokens": 381195095.0, + "step": 16318 + }, + { + "epoch": 1.3456749402160468, + "grad_norm": 2.0424232482910156, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7634109854698181, + "num_tokens": 381223798.0, + "step": 16319 + }, + { + "epoch": 1.3457574008410984, + "grad_norm": 2.0017428398132324, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7537320852279663, + "num_tokens": 381255057.0, + "step": 16320 + }, + { + "epoch": 1.34583986146615, + "grad_norm": 2.1158814430236816, + "learning_rate": 1e-06, + "loss": 0.5783, + "mean_token_accuracy": 0.7935070991516113, + "num_tokens": 381279210.0, + "step": 16321 + }, + { + "epoch": 1.3459223220912016, + "grad_norm": 2.163510322570801, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.747643768787384, + "num_tokens": 381308166.0, + "step": 16322 + }, + { + "epoch": 1.346004782716253, + "grad_norm": 1.8417304754257202, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7743820548057556, + "num_tokens": 381341501.0, + "step": 16323 + }, + { + "epoch": 1.3460872433413045, + "grad_norm": 1.9166873693466187, + "learning_rate": 1e-06, + "loss": 0.5587, + "mean_token_accuracy": 0.8133648633956909, + "num_tokens": 381370738.0, + "step": 16324 + }, + { + "epoch": 1.346169703966356, + "grad_norm": 1.7837321758270264, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7864413261413574, + "num_tokens": 381407586.0, + "step": 16325 + }, + { + "epoch": 1.3462521645914076, + "grad_norm": 1.8877160549163818, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7769597768783569, + "num_tokens": 381439788.0, + "step": 16326 + }, + { + "epoch": 1.3463346252164592, + "grad_norm": 2.0657103061676025, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7211836576461792, + "num_tokens": 381470824.0, + "step": 16327 + }, + { + "epoch": 1.3464170858415105, + "grad_norm": 1.9377543926239014, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7644781470298767, + "num_tokens": 381500457.0, + "step": 16328 + }, + { + "epoch": 1.3464995464665623, + "grad_norm": 2.167962074279785, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.733241856098175, + "num_tokens": 381530396.0, + "step": 16329 + }, + { + "epoch": 1.3465820070916137, + "grad_norm": 1.9522212743759155, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7479982972145081, + "num_tokens": 381561997.0, + "step": 16330 + }, + { + "epoch": 1.3466644677166653, + "grad_norm": 2.1007542610168457, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7305426001548767, + "num_tokens": 381590850.0, + "step": 16331 + }, + { + "epoch": 1.3467469283417168, + "grad_norm": 2.16227650642395, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7400362491607666, + "num_tokens": 381619260.0, + "step": 16332 + }, + { + "epoch": 1.3468293889667684, + "grad_norm": 1.9375457763671875, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.7859233021736145, + "num_tokens": 381649603.0, + "step": 16333 + }, + { + "epoch": 1.34691184959182, + "grad_norm": 1.934081792831421, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.754307210445404, + "num_tokens": 381684601.0, + "step": 16334 + }, + { + "epoch": 1.3469943102168713, + "grad_norm": 1.989288330078125, + "learning_rate": 1e-06, + "loss": 0.5952, + "mean_token_accuracy": 0.8037383556365967, + "num_tokens": 381714155.0, + "step": 16335 + }, + { + "epoch": 1.347076770841923, + "grad_norm": 1.8427002429962158, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7702704668045044, + "num_tokens": 381746071.0, + "step": 16336 + }, + { + "epoch": 1.3471592314669745, + "grad_norm": 1.9940123558044434, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7692934274673462, + "num_tokens": 381773574.0, + "step": 16337 + }, + { + "epoch": 1.347241692092026, + "grad_norm": 1.9020143747329712, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7565732002258301, + "num_tokens": 381807828.0, + "step": 16338 + }, + { + "epoch": 1.3473241527170776, + "grad_norm": 2.328904867172241, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7267552018165588, + "num_tokens": 381832034.0, + "step": 16339 + }, + { + "epoch": 1.3474066133421292, + "grad_norm": 1.8584115505218506, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.78623366355896, + "num_tokens": 381865357.0, + "step": 16340 + }, + { + "epoch": 1.3474890739671808, + "grad_norm": 1.6637935638427734, + "learning_rate": 1e-06, + "loss": 0.5822, + "mean_token_accuracy": 0.8050556182861328, + "num_tokens": 381903115.0, + "step": 16341 + }, + { + "epoch": 1.3475715345922321, + "grad_norm": 1.9083442687988281, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7610373497009277, + "num_tokens": 381933423.0, + "step": 16342 + }, + { + "epoch": 1.3476539952172837, + "grad_norm": 2.0721707344055176, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7733838558197021, + "num_tokens": 381963163.0, + "step": 16343 + }, + { + "epoch": 1.3477364558423353, + "grad_norm": 1.8835309743881226, + "learning_rate": 1e-06, + "loss": 0.5636, + "mean_token_accuracy": 0.8095675110816956, + "num_tokens": 381992904.0, + "step": 16344 + }, + { + "epoch": 1.3478189164673868, + "grad_norm": 2.1966559886932373, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7534852027893066, + "num_tokens": 382020103.0, + "step": 16345 + }, + { + "epoch": 1.3479013770924384, + "grad_norm": 1.8434292078018188, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.774867057800293, + "num_tokens": 382055630.0, + "step": 16346 + }, + { + "epoch": 1.34798383771749, + "grad_norm": 2.1406338214874268, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7165389060974121, + "num_tokens": 382082464.0, + "step": 16347 + }, + { + "epoch": 1.3480662983425415, + "grad_norm": 2.3021528720855713, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7192883491516113, + "num_tokens": 382106848.0, + "step": 16348 + }, + { + "epoch": 1.348148758967593, + "grad_norm": 1.8382117748260498, + "learning_rate": 1e-06, + "loss": 0.6091, + "mean_token_accuracy": 0.7993643283843994, + "num_tokens": 382140461.0, + "step": 16349 + }, + { + "epoch": 1.3482312195926445, + "grad_norm": 1.744368314743042, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.7768407464027405, + "num_tokens": 382176748.0, + "step": 16350 + }, + { + "epoch": 1.348313680217696, + "grad_norm": 2.1862611770629883, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7575931549072266, + "num_tokens": 382201900.0, + "step": 16351 + }, + { + "epoch": 1.3483961408427476, + "grad_norm": 2.2648401260375977, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7667995095252991, + "num_tokens": 382224823.0, + "step": 16352 + }, + { + "epoch": 1.3484786014677992, + "grad_norm": 1.867913007736206, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.7833399772644043, + "num_tokens": 382256011.0, + "step": 16353 + }, + { + "epoch": 1.3485610620928505, + "grad_norm": 2.008422374725342, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7524820566177368, + "num_tokens": 382284590.0, + "step": 16354 + }, + { + "epoch": 1.3486435227179023, + "grad_norm": 2.2649471759796143, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.75850510597229, + "num_tokens": 382308075.0, + "step": 16355 + }, + { + "epoch": 1.3487259833429537, + "grad_norm": 2.016958713531494, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7298845052719116, + "num_tokens": 382340064.0, + "step": 16356 + }, + { + "epoch": 1.3488084439680053, + "grad_norm": 2.013720750808716, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.7797437310218811, + "num_tokens": 382367046.0, + "step": 16357 + }, + { + "epoch": 1.3488909045930568, + "grad_norm": 1.978652834892273, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7723225355148315, + "num_tokens": 382402486.0, + "step": 16358 + }, + { + "epoch": 1.3489733652181084, + "grad_norm": 2.0265052318573, + "learning_rate": 1e-06, + "loss": 0.6089, + "mean_token_accuracy": 0.7981114387512207, + "num_tokens": 382430055.0, + "step": 16359 + }, + { + "epoch": 1.34905582584316, + "grad_norm": 1.7295974493026733, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7795851230621338, + "num_tokens": 382464215.0, + "step": 16360 + }, + { + "epoch": 1.3491382864682113, + "grad_norm": 2.2662463188171387, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7248643040657043, + "num_tokens": 382490799.0, + "step": 16361 + }, + { + "epoch": 1.349220747093263, + "grad_norm": 1.9859477281570435, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7764515280723572, + "num_tokens": 382519961.0, + "step": 16362 + }, + { + "epoch": 1.3493032077183145, + "grad_norm": 2.161548137664795, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7224989533424377, + "num_tokens": 382547489.0, + "step": 16363 + }, + { + "epoch": 1.349385668343366, + "grad_norm": 1.9763638973236084, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7377811670303345, + "num_tokens": 382578246.0, + "step": 16364 + }, + { + "epoch": 1.3494681289684176, + "grad_norm": 2.1494593620300293, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7463334202766418, + "num_tokens": 382604494.0, + "step": 16365 + }, + { + "epoch": 1.3495505895934692, + "grad_norm": 2.0780718326568604, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7131004333496094, + "num_tokens": 382634195.0, + "step": 16366 + }, + { + "epoch": 1.3496330502185208, + "grad_norm": 1.822014331817627, + "learning_rate": 1e-06, + "loss": 0.6044, + "mean_token_accuracy": 0.7996470928192139, + "num_tokens": 382664409.0, + "step": 16367 + }, + { + "epoch": 1.349715510843572, + "grad_norm": 1.90878164768219, + "learning_rate": 1e-06, + "loss": 0.5685, + "mean_token_accuracy": 0.7971735000610352, + "num_tokens": 382696233.0, + "step": 16368 + }, + { + "epoch": 1.3497979714686237, + "grad_norm": 1.9015647172927856, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.7563492655754089, + "num_tokens": 382728074.0, + "step": 16369 + }, + { + "epoch": 1.3498804320936753, + "grad_norm": 1.8871855735778809, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7714023590087891, + "num_tokens": 382761321.0, + "step": 16370 + }, + { + "epoch": 1.3499628927187268, + "grad_norm": 1.8004488945007324, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.767265796661377, + "num_tokens": 382796618.0, + "step": 16371 + }, + { + "epoch": 1.3500453533437784, + "grad_norm": 1.9841755628585815, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7185884714126587, + "num_tokens": 382828319.0, + "step": 16372 + }, + { + "epoch": 1.3501278139688297, + "grad_norm": 1.8693110942840576, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.7958532571792603, + "num_tokens": 382861423.0, + "step": 16373 + }, + { + "epoch": 1.3502102745938815, + "grad_norm": 2.1435980796813965, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7636341452598572, + "num_tokens": 382889251.0, + "step": 16374 + }, + { + "epoch": 1.350292735218933, + "grad_norm": 1.9890494346618652, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7390682697296143, + "num_tokens": 382922043.0, + "step": 16375 + }, + { + "epoch": 1.3503751958439845, + "grad_norm": 2.0240211486816406, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7534359097480774, + "num_tokens": 382951349.0, + "step": 16376 + }, + { + "epoch": 1.350457656469036, + "grad_norm": 2.0076472759246826, + "learning_rate": 1e-06, + "loss": 0.7999, + "mean_token_accuracy": 0.7394936084747314, + "num_tokens": 382982604.0, + "step": 16377 + }, + { + "epoch": 1.3505401170940876, + "grad_norm": 1.9619402885437012, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7708515524864197, + "num_tokens": 383015788.0, + "step": 16378 + }, + { + "epoch": 1.3506225777191392, + "grad_norm": 1.9779342412948608, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.729073166847229, + "num_tokens": 383047307.0, + "step": 16379 + }, + { + "epoch": 1.3507050383441905, + "grad_norm": 1.806004285812378, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7666706442832947, + "num_tokens": 383080171.0, + "step": 16380 + }, + { + "epoch": 1.350787498969242, + "grad_norm": 1.8598806858062744, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7735958099365234, + "num_tokens": 383114911.0, + "step": 16381 + }, + { + "epoch": 1.3508699595942937, + "grad_norm": 1.9451563358306885, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.746756374835968, + "num_tokens": 383147636.0, + "step": 16382 + }, + { + "epoch": 1.3509524202193453, + "grad_norm": 2.2738070487976074, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7596995830535889, + "num_tokens": 383173789.0, + "step": 16383 + }, + { + "epoch": 1.3510348808443968, + "grad_norm": 2.0488524436950684, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7578551769256592, + "num_tokens": 383201527.0, + "step": 16384 + }, + { + "epoch": 1.3511173414694484, + "grad_norm": 2.1740334033966064, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.6994763016700745, + "num_tokens": 383229420.0, + "step": 16385 + }, + { + "epoch": 1.3511998020945, + "grad_norm": 2.18326735496521, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.764660120010376, + "num_tokens": 383255245.0, + "step": 16386 + }, + { + "epoch": 1.3512822627195513, + "grad_norm": 2.0271337032318115, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7634772062301636, + "num_tokens": 383284490.0, + "step": 16387 + }, + { + "epoch": 1.351364723344603, + "grad_norm": 1.8850252628326416, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7445381879806519, + "num_tokens": 383321814.0, + "step": 16388 + }, + { + "epoch": 1.3514471839696545, + "grad_norm": 1.8577160835266113, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7684732675552368, + "num_tokens": 383353411.0, + "step": 16389 + }, + { + "epoch": 1.351529644594706, + "grad_norm": 1.9934992790222168, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7787889838218689, + "num_tokens": 383383203.0, + "step": 16390 + }, + { + "epoch": 1.3516121052197576, + "grad_norm": 1.8058565855026245, + "learning_rate": 1e-06, + "loss": 0.5998, + "mean_token_accuracy": 0.8031623363494873, + "num_tokens": 383415148.0, + "step": 16391 + }, + { + "epoch": 1.3516945658448092, + "grad_norm": 1.986754298210144, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7671109437942505, + "num_tokens": 383443977.0, + "step": 16392 + }, + { + "epoch": 1.3517770264698608, + "grad_norm": 2.164907455444336, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7677031755447388, + "num_tokens": 383468005.0, + "step": 16393 + }, + { + "epoch": 1.351859487094912, + "grad_norm": 1.8954030275344849, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7851364016532898, + "num_tokens": 383499894.0, + "step": 16394 + }, + { + "epoch": 1.3519419477199637, + "grad_norm": 1.803215503692627, + "learning_rate": 1e-06, + "loss": 0.6165, + "mean_token_accuracy": 0.7963549494743347, + "num_tokens": 383533146.0, + "step": 16395 + }, + { + "epoch": 1.3520244083450152, + "grad_norm": 1.8910226821899414, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7818425893783569, + "num_tokens": 383565562.0, + "step": 16396 + }, + { + "epoch": 1.3521068689700668, + "grad_norm": 1.900688648223877, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.743959367275238, + "num_tokens": 383598721.0, + "step": 16397 + }, + { + "epoch": 1.3521893295951184, + "grad_norm": 2.0239787101745605, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.759402871131897, + "num_tokens": 383628603.0, + "step": 16398 + }, + { + "epoch": 1.3522717902201697, + "grad_norm": 1.941313624382019, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.751287579536438, + "num_tokens": 383660777.0, + "step": 16399 + }, + { + "epoch": 1.3523542508452215, + "grad_norm": 1.7540040016174316, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7658387422561646, + "num_tokens": 383698664.0, + "step": 16400 + }, + { + "epoch": 1.352436711470273, + "grad_norm": 2.1693503856658936, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7342389822006226, + "num_tokens": 383728955.0, + "step": 16401 + }, + { + "epoch": 1.3525191720953245, + "grad_norm": 2.115307331085205, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7548320293426514, + "num_tokens": 383756458.0, + "step": 16402 + }, + { + "epoch": 1.352601632720376, + "grad_norm": 2.0815885066986084, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7741582989692688, + "num_tokens": 383782198.0, + "step": 16403 + }, + { + "epoch": 1.3526840933454276, + "grad_norm": 2.067944049835205, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7549275159835815, + "num_tokens": 383809411.0, + "step": 16404 + }, + { + "epoch": 1.3527665539704792, + "grad_norm": 1.9509644508361816, + "learning_rate": 1e-06, + "loss": 0.5807, + "mean_token_accuracy": 0.7978007197380066, + "num_tokens": 383837901.0, + "step": 16405 + }, + { + "epoch": 1.3528490145955305, + "grad_norm": 2.178192138671875, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.7500149011611938, + "num_tokens": 383865458.0, + "step": 16406 + }, + { + "epoch": 1.352931475220582, + "grad_norm": 1.8455946445465088, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7464232444763184, + "num_tokens": 383901381.0, + "step": 16407 + }, + { + "epoch": 1.3530139358456337, + "grad_norm": 2.0662829875946045, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.751423716545105, + "num_tokens": 383927767.0, + "step": 16408 + }, + { + "epoch": 1.3530963964706852, + "grad_norm": 2.2771096229553223, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7662885785102844, + "num_tokens": 383951367.0, + "step": 16409 + }, + { + "epoch": 1.3531788570957368, + "grad_norm": 2.2470362186431885, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7723044157028198, + "num_tokens": 383977624.0, + "step": 16410 + }, + { + "epoch": 1.3532613177207884, + "grad_norm": 1.9309654235839844, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.7992461919784546, + "num_tokens": 384006641.0, + "step": 16411 + }, + { + "epoch": 1.35334377834584, + "grad_norm": 1.9595837593078613, + "learning_rate": 1e-06, + "loss": 0.7852, + "mean_token_accuracy": 0.768520712852478, + "num_tokens": 384038180.0, + "step": 16412 + }, + { + "epoch": 1.3534262389708913, + "grad_norm": 1.9594393968582153, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7486948370933533, + "num_tokens": 384069448.0, + "step": 16413 + }, + { + "epoch": 1.3535086995959429, + "grad_norm": 1.7614946365356445, + "learning_rate": 1e-06, + "loss": 0.4837, + "mean_token_accuracy": 0.8356962203979492, + "num_tokens": 384102201.0, + "step": 16414 + }, + { + "epoch": 1.3535911602209945, + "grad_norm": 1.8579792976379395, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.775922417640686, + "num_tokens": 384136693.0, + "step": 16415 + }, + { + "epoch": 1.353673620846046, + "grad_norm": 2.039777994155884, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7788373827934265, + "num_tokens": 384164672.0, + "step": 16416 + }, + { + "epoch": 1.3537560814710976, + "grad_norm": 1.9719091653823853, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7786802053451538, + "num_tokens": 384194828.0, + "step": 16417 + }, + { + "epoch": 1.353838542096149, + "grad_norm": 2.1047449111938477, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7409613728523254, + "num_tokens": 384221658.0, + "step": 16418 + }, + { + "epoch": 1.3539210027212008, + "grad_norm": 1.9103058576583862, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7764798402786255, + "num_tokens": 384251185.0, + "step": 16419 + }, + { + "epoch": 1.354003463346252, + "grad_norm": 2.0815300941467285, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7684391736984253, + "num_tokens": 384278812.0, + "step": 16420 + }, + { + "epoch": 1.3540859239713037, + "grad_norm": 2.0067427158355713, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.7512518167495728, + "num_tokens": 384307255.0, + "step": 16421 + }, + { + "epoch": 1.3541683845963552, + "grad_norm": 2.2478859424591064, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7623186111450195, + "num_tokens": 384330754.0, + "step": 16422 + }, + { + "epoch": 1.3542508452214068, + "grad_norm": 1.9256433248519897, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7809617519378662, + "num_tokens": 384362425.0, + "step": 16423 + }, + { + "epoch": 1.3543333058464584, + "grad_norm": 1.8791015148162842, + "learning_rate": 1e-06, + "loss": 0.6085, + "mean_token_accuracy": 0.7984790802001953, + "num_tokens": 384390565.0, + "step": 16424 + }, + { + "epoch": 1.3544157664715097, + "grad_norm": 1.7179758548736572, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.7963166236877441, + "num_tokens": 384425328.0, + "step": 16425 + }, + { + "epoch": 1.3544982270965613, + "grad_norm": 1.9655473232269287, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7757989168167114, + "num_tokens": 384456131.0, + "step": 16426 + }, + { + "epoch": 1.3545806877216129, + "grad_norm": 2.1690425872802734, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7350634336471558, + "num_tokens": 384480972.0, + "step": 16427 + }, + { + "epoch": 1.3546631483466645, + "grad_norm": 1.7409305572509766, + "learning_rate": 1e-06, + "loss": 0.6076, + "mean_token_accuracy": 0.8005014657974243, + "num_tokens": 384516382.0, + "step": 16428 + }, + { + "epoch": 1.354745608971716, + "grad_norm": 2.6337931156158447, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7372408509254456, + "num_tokens": 384537818.0, + "step": 16429 + }, + { + "epoch": 1.3548280695967676, + "grad_norm": 1.6913882493972778, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.7936237454414368, + "num_tokens": 384573599.0, + "step": 16430 + }, + { + "epoch": 1.3549105302218192, + "grad_norm": 2.0938446521759033, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7682769894599915, + "num_tokens": 384603752.0, + "step": 16431 + }, + { + "epoch": 1.3549929908468705, + "grad_norm": 1.940914273262024, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7001355290412903, + "num_tokens": 384635423.0, + "step": 16432 + }, + { + "epoch": 1.355075451471922, + "grad_norm": 1.9277663230895996, + "learning_rate": 1e-06, + "loss": 0.5881, + "mean_token_accuracy": 0.7987889051437378, + "num_tokens": 384664054.0, + "step": 16433 + }, + { + "epoch": 1.3551579120969737, + "grad_norm": 2.3754382133483887, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7480176687240601, + "num_tokens": 384687995.0, + "step": 16434 + }, + { + "epoch": 1.3552403727220252, + "grad_norm": 1.8175793886184692, + "learning_rate": 1e-06, + "loss": 0.5964, + "mean_token_accuracy": 0.797001838684082, + "num_tokens": 384720683.0, + "step": 16435 + }, + { + "epoch": 1.3553228333470768, + "grad_norm": 2.0133209228515625, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.761552095413208, + "num_tokens": 384749105.0, + "step": 16436 + }, + { + "epoch": 1.3554052939721284, + "grad_norm": 1.9161320924758911, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.7655712962150574, + "num_tokens": 384782451.0, + "step": 16437 + }, + { + "epoch": 1.35548775459718, + "grad_norm": 1.98958158493042, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7623171210289001, + "num_tokens": 384814097.0, + "step": 16438 + }, + { + "epoch": 1.3555702152222313, + "grad_norm": 2.061769723892212, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.7882319688796997, + "num_tokens": 384841159.0, + "step": 16439 + }, + { + "epoch": 1.3556526758472829, + "grad_norm": 1.8249117136001587, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.7792143821716309, + "num_tokens": 384874381.0, + "step": 16440 + }, + { + "epoch": 1.3557351364723345, + "grad_norm": 2.0164742469787598, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7432330846786499, + "num_tokens": 384906452.0, + "step": 16441 + }, + { + "epoch": 1.355817597097386, + "grad_norm": 2.267529249191284, + "learning_rate": 1e-06, + "loss": 0.7842, + "mean_token_accuracy": 0.7519803047180176, + "num_tokens": 384931323.0, + "step": 16442 + }, + { + "epoch": 1.3559000577224376, + "grad_norm": 2.0571882724761963, + "learning_rate": 1e-06, + "loss": 0.6165, + "mean_token_accuracy": 0.7851129174232483, + "num_tokens": 384957306.0, + "step": 16443 + }, + { + "epoch": 1.355982518347489, + "grad_norm": 1.9412819147109985, + "learning_rate": 1e-06, + "loss": 0.7531, + "mean_token_accuracy": 0.7627316117286682, + "num_tokens": 384988621.0, + "step": 16444 + }, + { + "epoch": 1.3560649789725407, + "grad_norm": 1.889677882194519, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7566202282905579, + "num_tokens": 385020215.0, + "step": 16445 + }, + { + "epoch": 1.356147439597592, + "grad_norm": 1.888036847114563, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.759527325630188, + "num_tokens": 385053642.0, + "step": 16446 + }, + { + "epoch": 1.3562299002226437, + "grad_norm": 1.9296208620071411, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7791010737419128, + "num_tokens": 385083720.0, + "step": 16447 + }, + { + "epoch": 1.3563123608476952, + "grad_norm": 2.2259397506713867, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.7831112742424011, + "num_tokens": 385107539.0, + "step": 16448 + }, + { + "epoch": 1.3563948214727468, + "grad_norm": 1.841103434562683, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.745470404624939, + "num_tokens": 385140731.0, + "step": 16449 + }, + { + "epoch": 1.3564772820977984, + "grad_norm": 2.185648202896118, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.745404064655304, + "num_tokens": 385169237.0, + "step": 16450 + }, + { + "epoch": 1.3565597427228497, + "grad_norm": 2.2311465740203857, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7756226062774658, + "num_tokens": 385192212.0, + "step": 16451 + }, + { + "epoch": 1.3566422033479013, + "grad_norm": 1.7066566944122314, + "learning_rate": 1e-06, + "loss": 0.5593, + "mean_token_accuracy": 0.8155480623245239, + "num_tokens": 385230728.0, + "step": 16452 + }, + { + "epoch": 1.3567246639729529, + "grad_norm": 1.9651741981506348, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7427549362182617, + "num_tokens": 385263587.0, + "step": 16453 + }, + { + "epoch": 1.3568071245980045, + "grad_norm": 2.0925076007843018, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7341158390045166, + "num_tokens": 385293676.0, + "step": 16454 + }, + { + "epoch": 1.356889585223056, + "grad_norm": 1.7684887647628784, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7747225761413574, + "num_tokens": 385330806.0, + "step": 16455 + }, + { + "epoch": 1.3569720458481076, + "grad_norm": 2.1464200019836426, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7432251572608948, + "num_tokens": 385357514.0, + "step": 16456 + }, + { + "epoch": 1.3570545064731592, + "grad_norm": 2.158003568649292, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7499164938926697, + "num_tokens": 385383955.0, + "step": 16457 + }, + { + "epoch": 1.3571369670982105, + "grad_norm": 1.961029291152954, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7739909291267395, + "num_tokens": 385417327.0, + "step": 16458 + }, + { + "epoch": 1.357219427723262, + "grad_norm": 2.0551822185516357, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7436046004295349, + "num_tokens": 385446358.0, + "step": 16459 + }, + { + "epoch": 1.3573018883483137, + "grad_norm": 1.9696606397628784, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7325279116630554, + "num_tokens": 385477183.0, + "step": 16460 + }, + { + "epoch": 1.3573843489733652, + "grad_norm": 2.011831283569336, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7530001997947693, + "num_tokens": 385508136.0, + "step": 16461 + }, + { + "epoch": 1.3574668095984168, + "grad_norm": 1.8969491720199585, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7759222388267517, + "num_tokens": 385541034.0, + "step": 16462 + }, + { + "epoch": 1.3575492702234682, + "grad_norm": 2.1322104930877686, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7824035882949829, + "num_tokens": 385565198.0, + "step": 16463 + }, + { + "epoch": 1.35763173084852, + "grad_norm": 1.9454177618026733, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7556564807891846, + "num_tokens": 385596249.0, + "step": 16464 + }, + { + "epoch": 1.3577141914735713, + "grad_norm": 1.8625023365020752, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7283861041069031, + "num_tokens": 385632921.0, + "step": 16465 + }, + { + "epoch": 1.3577966520986229, + "grad_norm": 1.9111803770065308, + "learning_rate": 1e-06, + "loss": 0.5528, + "mean_token_accuracy": 0.811551570892334, + "num_tokens": 385662033.0, + "step": 16466 + }, + { + "epoch": 1.3578791127236745, + "grad_norm": 1.8860431909561157, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.764357328414917, + "num_tokens": 385692706.0, + "step": 16467 + }, + { + "epoch": 1.357961573348726, + "grad_norm": 1.8559799194335938, + "learning_rate": 1e-06, + "loss": 0.5905, + "mean_token_accuracy": 0.793591320514679, + "num_tokens": 385725776.0, + "step": 16468 + }, + { + "epoch": 1.3580440339737776, + "grad_norm": 2.16459059715271, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7771056890487671, + "num_tokens": 385753065.0, + "step": 16469 + }, + { + "epoch": 1.358126494598829, + "grad_norm": 1.9739525318145752, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7792858481407166, + "num_tokens": 385780576.0, + "step": 16470 + }, + { + "epoch": 1.3582089552238805, + "grad_norm": 2.3400466442108154, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7465482354164124, + "num_tokens": 385805990.0, + "step": 16471 + }, + { + "epoch": 1.358291415848932, + "grad_norm": 1.9659265279769897, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7734482288360596, + "num_tokens": 385836521.0, + "step": 16472 + }, + { + "epoch": 1.3583738764739837, + "grad_norm": 1.7301617860794067, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7655240297317505, + "num_tokens": 385876170.0, + "step": 16473 + }, + { + "epoch": 1.3584563370990352, + "grad_norm": 2.0405375957489014, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7538350820541382, + "num_tokens": 385907231.0, + "step": 16474 + }, + { + "epoch": 1.3585387977240868, + "grad_norm": 1.927769422531128, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7631218433380127, + "num_tokens": 385941657.0, + "step": 16475 + }, + { + "epoch": 1.3586212583491384, + "grad_norm": 1.8807415962219238, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.785478949546814, + "num_tokens": 385971742.0, + "step": 16476 + }, + { + "epoch": 1.3587037189741897, + "grad_norm": 2.025073289871216, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.695489227771759, + "num_tokens": 386003247.0, + "step": 16477 + }, + { + "epoch": 1.3587861795992413, + "grad_norm": 2.4811806678771973, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7112219333648682, + "num_tokens": 386024634.0, + "step": 16478 + }, + { + "epoch": 1.3588686402242929, + "grad_norm": 2.2615621089935303, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7382474541664124, + "num_tokens": 386052176.0, + "step": 16479 + }, + { + "epoch": 1.3589511008493445, + "grad_norm": 2.2918877601623535, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7345849871635437, + "num_tokens": 386076841.0, + "step": 16480 + }, + { + "epoch": 1.359033561474396, + "grad_norm": 1.850469946861267, + "learning_rate": 1e-06, + "loss": 0.5636, + "mean_token_accuracy": 0.8121662139892578, + "num_tokens": 386107790.0, + "step": 16481 + }, + { + "epoch": 1.3591160220994476, + "grad_norm": 1.9254571199417114, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7237540483474731, + "num_tokens": 386140101.0, + "step": 16482 + }, + { + "epoch": 1.3591984827244992, + "grad_norm": 1.8680065870285034, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7766972780227661, + "num_tokens": 386174930.0, + "step": 16483 + }, + { + "epoch": 1.3592809433495505, + "grad_norm": 1.992748498916626, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7535499334335327, + "num_tokens": 386202519.0, + "step": 16484 + }, + { + "epoch": 1.359363403974602, + "grad_norm": 2.2751362323760986, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7405915260314941, + "num_tokens": 386228466.0, + "step": 16485 + }, + { + "epoch": 1.3594458645996537, + "grad_norm": 2.0707430839538574, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7488083839416504, + "num_tokens": 386255540.0, + "step": 16486 + }, + { + "epoch": 1.3595283252247052, + "grad_norm": 2.0472633838653564, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7344172596931458, + "num_tokens": 386285314.0, + "step": 16487 + }, + { + "epoch": 1.3596107858497568, + "grad_norm": 1.837402105331421, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7509453296661377, + "num_tokens": 386319829.0, + "step": 16488 + }, + { + "epoch": 1.3596932464748082, + "grad_norm": 1.8321630954742432, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7633430361747742, + "num_tokens": 386354817.0, + "step": 16489 + }, + { + "epoch": 1.35977570709986, + "grad_norm": 1.9752341508865356, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7613162994384766, + "num_tokens": 386383138.0, + "step": 16490 + }, + { + "epoch": 1.3598581677249113, + "grad_norm": 2.1390089988708496, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7328190207481384, + "num_tokens": 386412222.0, + "step": 16491 + }, + { + "epoch": 1.3599406283499629, + "grad_norm": 2.201665163040161, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7641605734825134, + "num_tokens": 386437818.0, + "step": 16492 + }, + { + "epoch": 1.3600230889750144, + "grad_norm": 2.0959880352020264, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7450432777404785, + "num_tokens": 386465847.0, + "step": 16493 + }, + { + "epoch": 1.360105549600066, + "grad_norm": 2.2332613468170166, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7324544191360474, + "num_tokens": 386489852.0, + "step": 16494 + }, + { + "epoch": 1.3601880102251176, + "grad_norm": 2.289283275604248, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7564694285392761, + "num_tokens": 386513487.0, + "step": 16495 + }, + { + "epoch": 1.360270470850169, + "grad_norm": 1.9587277173995972, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.7811376452445984, + "num_tokens": 386544041.0, + "step": 16496 + }, + { + "epoch": 1.3603529314752205, + "grad_norm": 2.073911428451538, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.708707332611084, + "num_tokens": 386572153.0, + "step": 16497 + }, + { + "epoch": 1.360435392100272, + "grad_norm": 2.0786802768707275, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7225181460380554, + "num_tokens": 386602800.0, + "step": 16498 + }, + { + "epoch": 1.3605178527253237, + "grad_norm": 1.6800127029418945, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7839410305023193, + "num_tokens": 386641584.0, + "step": 16499 + }, + { + "epoch": 1.3606003133503752, + "grad_norm": 2.2041828632354736, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7236162424087524, + "num_tokens": 386668426.0, + "step": 16500 + }, + { + "epoch": 1.3606827739754268, + "grad_norm": 1.8975154161453247, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.728708028793335, + "num_tokens": 386704080.0, + "step": 16501 + }, + { + "epoch": 1.3607652346004784, + "grad_norm": 1.8509687185287476, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.7732142210006714, + "num_tokens": 386735756.0, + "step": 16502 + }, + { + "epoch": 1.3608476952255297, + "grad_norm": 1.7908780574798584, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.7878323793411255, + "num_tokens": 386771654.0, + "step": 16503 + }, + { + "epoch": 1.3609301558505813, + "grad_norm": 2.2524471282958984, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7361509203910828, + "num_tokens": 386798244.0, + "step": 16504 + }, + { + "epoch": 1.3610126164756329, + "grad_norm": 2.1328086853027344, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.785148561000824, + "num_tokens": 386825010.0, + "step": 16505 + }, + { + "epoch": 1.3610950771006844, + "grad_norm": 2.1314759254455566, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7157127857208252, + "num_tokens": 386856834.0, + "step": 16506 + }, + { + "epoch": 1.361177537725736, + "grad_norm": 2.202731132507324, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7221737504005432, + "num_tokens": 386884591.0, + "step": 16507 + }, + { + "epoch": 1.3612599983507874, + "grad_norm": 1.8065489530563354, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7474662661552429, + "num_tokens": 386925190.0, + "step": 16508 + }, + { + "epoch": 1.3613424589758392, + "grad_norm": 2.094393730163574, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7468295693397522, + "num_tokens": 386953950.0, + "step": 16509 + }, + { + "epoch": 1.3614249196008905, + "grad_norm": 1.9622420072555542, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.768814206123352, + "num_tokens": 386983510.0, + "step": 16510 + }, + { + "epoch": 1.361507380225942, + "grad_norm": 1.9243124723434448, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7519840598106384, + "num_tokens": 387014511.0, + "step": 16511 + }, + { + "epoch": 1.3615898408509937, + "grad_norm": 1.8869602680206299, + "learning_rate": 1e-06, + "loss": 0.6019, + "mean_token_accuracy": 0.7957932949066162, + "num_tokens": 387044275.0, + "step": 16512 + }, + { + "epoch": 1.3616723014760452, + "grad_norm": 2.2310636043548584, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.74624103307724, + "num_tokens": 387070578.0, + "step": 16513 + }, + { + "epoch": 1.3617547621010968, + "grad_norm": 1.8433375358581543, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7813323736190796, + "num_tokens": 387103877.0, + "step": 16514 + }, + { + "epoch": 1.3618372227261482, + "grad_norm": 1.7334163188934326, + "learning_rate": 1e-06, + "loss": 0.5457, + "mean_token_accuracy": 0.8192199468612671, + "num_tokens": 387140210.0, + "step": 16515 + }, + { + "epoch": 1.3619196833511997, + "grad_norm": 2.475801467895508, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7168152332305908, + "num_tokens": 387163534.0, + "step": 16516 + }, + { + "epoch": 1.3620021439762513, + "grad_norm": 2.113539457321167, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7573680281639099, + "num_tokens": 387192473.0, + "step": 16517 + }, + { + "epoch": 1.3620846046013029, + "grad_norm": 2.2650411128997803, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7534709572792053, + "num_tokens": 387218442.0, + "step": 16518 + }, + { + "epoch": 1.3621670652263544, + "grad_norm": 2.1950130462646484, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7658131718635559, + "num_tokens": 387243758.0, + "step": 16519 + }, + { + "epoch": 1.362249525851406, + "grad_norm": 1.9896389245986938, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7645567655563354, + "num_tokens": 387272452.0, + "step": 16520 + }, + { + "epoch": 1.3623319864764576, + "grad_norm": 1.6282906532287598, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7399095296859741, + "num_tokens": 387315279.0, + "step": 16521 + }, + { + "epoch": 1.362414447101509, + "grad_norm": 1.9767661094665527, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7875161170959473, + "num_tokens": 387343503.0, + "step": 16522 + }, + { + "epoch": 1.3624969077265605, + "grad_norm": 1.9732578992843628, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7485875487327576, + "num_tokens": 387373056.0, + "step": 16523 + }, + { + "epoch": 1.362579368351612, + "grad_norm": 1.8309154510498047, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7547193765640259, + "num_tokens": 387408551.0, + "step": 16524 + }, + { + "epoch": 1.3626618289766637, + "grad_norm": 2.1385817527770996, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7421414256095886, + "num_tokens": 387433834.0, + "step": 16525 + }, + { + "epoch": 1.3627442896017152, + "grad_norm": 2.002331256866455, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7245032787322998, + "num_tokens": 387461928.0, + "step": 16526 + }, + { + "epoch": 1.3628267502267666, + "grad_norm": 2.20819091796875, + "learning_rate": 1e-06, + "loss": 0.6075, + "mean_token_accuracy": 0.7873624563217163, + "num_tokens": 387487779.0, + "step": 16527 + }, + { + "epoch": 1.3629092108518184, + "grad_norm": 1.7928028106689453, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.7964907884597778, + "num_tokens": 387522231.0, + "step": 16528 + }, + { + "epoch": 1.3629916714768697, + "grad_norm": 2.135894298553467, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.7941137552261353, + "num_tokens": 387548444.0, + "step": 16529 + }, + { + "epoch": 1.3630741321019213, + "grad_norm": 2.215182304382324, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7570540904998779, + "num_tokens": 387575723.0, + "step": 16530 + }, + { + "epoch": 1.3631565927269729, + "grad_norm": 1.8780736923217773, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7689138650894165, + "num_tokens": 387610757.0, + "step": 16531 + }, + { + "epoch": 1.3632390533520244, + "grad_norm": 2.1809513568878174, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.719918966293335, + "num_tokens": 387638965.0, + "step": 16532 + }, + { + "epoch": 1.363321513977076, + "grad_norm": 2.3896477222442627, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.740503191947937, + "num_tokens": 387665298.0, + "step": 16533 + }, + { + "epoch": 1.3634039746021274, + "grad_norm": 2.034158945083618, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7485277652740479, + "num_tokens": 387696419.0, + "step": 16534 + }, + { + "epoch": 1.363486435227179, + "grad_norm": 2.0606625080108643, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7199602723121643, + "num_tokens": 387727079.0, + "step": 16535 + }, + { + "epoch": 1.3635688958522305, + "grad_norm": 1.8640737533569336, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7561038732528687, + "num_tokens": 387762379.0, + "step": 16536 + }, + { + "epoch": 1.363651356477282, + "grad_norm": 2.033700466156006, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7685608863830566, + "num_tokens": 387790766.0, + "step": 16537 + }, + { + "epoch": 1.3637338171023337, + "grad_norm": 1.9320909976959229, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7226940989494324, + "num_tokens": 387825769.0, + "step": 16538 + }, + { + "epoch": 1.3638162777273852, + "grad_norm": 2.104372501373291, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7843411564826965, + "num_tokens": 387854663.0, + "step": 16539 + }, + { + "epoch": 1.3638987383524368, + "grad_norm": 2.441063165664673, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.7938477993011475, + "num_tokens": 387876838.0, + "step": 16540 + }, + { + "epoch": 1.3639811989774882, + "grad_norm": 1.9843264818191528, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7621015310287476, + "num_tokens": 387906481.0, + "step": 16541 + }, + { + "epoch": 1.3640636596025397, + "grad_norm": 2.1648993492126465, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.7835773229598999, + "num_tokens": 387931498.0, + "step": 16542 + }, + { + "epoch": 1.3641461202275913, + "grad_norm": 1.9596716165542603, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7288883924484253, + "num_tokens": 387963140.0, + "step": 16543 + }, + { + "epoch": 1.3642285808526429, + "grad_norm": 2.0104193687438965, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7495037317276001, + "num_tokens": 387995308.0, + "step": 16544 + }, + { + "epoch": 1.3643110414776944, + "grad_norm": 2.3267555236816406, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7421656250953674, + "num_tokens": 388020680.0, + "step": 16545 + }, + { + "epoch": 1.364393502102746, + "grad_norm": 1.8802618980407715, + "learning_rate": 1e-06, + "loss": 0.597, + "mean_token_accuracy": 0.7955948114395142, + "num_tokens": 388053912.0, + "step": 16546 + }, + { + "epoch": 1.3644759627277976, + "grad_norm": 2.0028204917907715, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7446908950805664, + "num_tokens": 388085312.0, + "step": 16547 + }, + { + "epoch": 1.364558423352849, + "grad_norm": 2.2094414234161377, + "learning_rate": 1e-06, + "loss": 0.5943, + "mean_token_accuracy": 0.7942470908164978, + "num_tokens": 388107037.0, + "step": 16548 + }, + { + "epoch": 1.3646408839779005, + "grad_norm": 2.0695269107818604, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7564035654067993, + "num_tokens": 388135662.0, + "step": 16549 + }, + { + "epoch": 1.364723344602952, + "grad_norm": 2.0702192783355713, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7762564420700073, + "num_tokens": 388165146.0, + "step": 16550 + }, + { + "epoch": 1.3648058052280037, + "grad_norm": 1.780169129371643, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.7924020290374756, + "num_tokens": 388200389.0, + "step": 16551 + }, + { + "epoch": 1.3648882658530552, + "grad_norm": 2.153050661087036, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7539867758750916, + "num_tokens": 388225402.0, + "step": 16552 + }, + { + "epoch": 1.3649707264781066, + "grad_norm": 2.09647274017334, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7497389316558838, + "num_tokens": 388252781.0, + "step": 16553 + }, + { + "epoch": 1.3650531871031584, + "grad_norm": 1.8196115493774414, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7628054618835449, + "num_tokens": 388289350.0, + "step": 16554 + }, + { + "epoch": 1.3651356477282097, + "grad_norm": 2.122567892074585, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7738662958145142, + "num_tokens": 388316392.0, + "step": 16555 + }, + { + "epoch": 1.3652181083532613, + "grad_norm": 2.177037000656128, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7544633150100708, + "num_tokens": 388340636.0, + "step": 16556 + }, + { + "epoch": 1.3653005689783129, + "grad_norm": 1.9321988821029663, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.7951681613922119, + "num_tokens": 388372658.0, + "step": 16557 + }, + { + "epoch": 1.3653830296033644, + "grad_norm": 2.1531481742858887, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7745739221572876, + "num_tokens": 388398320.0, + "step": 16558 + }, + { + "epoch": 1.365465490228416, + "grad_norm": 2.126638174057007, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7390570640563965, + "num_tokens": 388426659.0, + "step": 16559 + }, + { + "epoch": 1.3655479508534674, + "grad_norm": 2.2022182941436768, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7386784553527832, + "num_tokens": 388451232.0, + "step": 16560 + }, + { + "epoch": 1.365630411478519, + "grad_norm": 2.1045000553131104, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7541074752807617, + "num_tokens": 388482301.0, + "step": 16561 + }, + { + "epoch": 1.3657128721035705, + "grad_norm": 2.2048187255859375, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7626693844795227, + "num_tokens": 388508158.0, + "step": 16562 + }, + { + "epoch": 1.365795332728622, + "grad_norm": 1.8964813947677612, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.758155345916748, + "num_tokens": 388540229.0, + "step": 16563 + }, + { + "epoch": 1.3658777933536737, + "grad_norm": 2.023132085800171, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.7947379350662231, + "num_tokens": 388566935.0, + "step": 16564 + }, + { + "epoch": 1.3659602539787252, + "grad_norm": 1.9364445209503174, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7648986577987671, + "num_tokens": 388599841.0, + "step": 16565 + }, + { + "epoch": 1.3660427146037768, + "grad_norm": 1.9579492807388306, + "learning_rate": 1e-06, + "loss": 0.5625, + "mean_token_accuracy": 0.8086538314819336, + "num_tokens": 388627977.0, + "step": 16566 + }, + { + "epoch": 1.3661251752288281, + "grad_norm": 2.0478951930999756, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.7849413156509399, + "num_tokens": 388653554.0, + "step": 16567 + }, + { + "epoch": 1.3662076358538797, + "grad_norm": 1.84027099609375, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.7899876832962036, + "num_tokens": 388687263.0, + "step": 16568 + }, + { + "epoch": 1.3662900964789313, + "grad_norm": 1.9973666667938232, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7823168635368347, + "num_tokens": 388717269.0, + "step": 16569 + }, + { + "epoch": 1.3663725571039829, + "grad_norm": 2.229630708694458, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.771676778793335, + "num_tokens": 388744970.0, + "step": 16570 + }, + { + "epoch": 1.3664550177290344, + "grad_norm": 2.248445987701416, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7140596508979797, + "num_tokens": 388770236.0, + "step": 16571 + }, + { + "epoch": 1.3665374783540858, + "grad_norm": 1.7385976314544678, + "learning_rate": 1e-06, + "loss": 0.5749, + "mean_token_accuracy": 0.8108229637145996, + "num_tokens": 388802070.0, + "step": 16572 + }, + { + "epoch": 1.3666199389791376, + "grad_norm": 2.090034008026123, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7754634618759155, + "num_tokens": 388828139.0, + "step": 16573 + }, + { + "epoch": 1.366702399604189, + "grad_norm": 2.052157163619995, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.760046124458313, + "num_tokens": 388858647.0, + "step": 16574 + }, + { + "epoch": 1.3667848602292405, + "grad_norm": 1.9439870119094849, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7391257286071777, + "num_tokens": 388888548.0, + "step": 16575 + }, + { + "epoch": 1.366867320854292, + "grad_norm": 1.880029320716858, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7825996279716492, + "num_tokens": 388921294.0, + "step": 16576 + }, + { + "epoch": 1.3669497814793437, + "grad_norm": 2.018446207046509, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.7880285978317261, + "num_tokens": 388949639.0, + "step": 16577 + }, + { + "epoch": 1.3670322421043952, + "grad_norm": 1.9296921491622925, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7130610942840576, + "num_tokens": 388984608.0, + "step": 16578 + }, + { + "epoch": 1.3671147027294466, + "grad_norm": 2.0250613689422607, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7553561329841614, + "num_tokens": 389015204.0, + "step": 16579 + }, + { + "epoch": 1.3671971633544981, + "grad_norm": 2.0015783309936523, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.778254508972168, + "num_tokens": 389046721.0, + "step": 16580 + }, + { + "epoch": 1.3672796239795497, + "grad_norm": 2.130983352661133, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7770994305610657, + "num_tokens": 389072496.0, + "step": 16581 + }, + { + "epoch": 1.3673620846046013, + "grad_norm": 2.2676472663879395, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7559635639190674, + "num_tokens": 389097876.0, + "step": 16582 + }, + { + "epoch": 1.3674445452296529, + "grad_norm": 2.031416893005371, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7598288059234619, + "num_tokens": 389125479.0, + "step": 16583 + }, + { + "epoch": 1.3675270058547044, + "grad_norm": 2.2417590618133545, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7509377002716064, + "num_tokens": 389151765.0, + "step": 16584 + }, + { + "epoch": 1.367609466479756, + "grad_norm": 2.1620113849639893, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7714824676513672, + "num_tokens": 389179563.0, + "step": 16585 + }, + { + "epoch": 1.3676919271048074, + "grad_norm": 1.985369086265564, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.7818870544433594, + "num_tokens": 389206292.0, + "step": 16586 + }, + { + "epoch": 1.367774387729859, + "grad_norm": 2.05755615234375, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7713649868965149, + "num_tokens": 389233922.0, + "step": 16587 + }, + { + "epoch": 1.3678568483549105, + "grad_norm": 1.8725894689559937, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7780629396438599, + "num_tokens": 389265176.0, + "step": 16588 + }, + { + "epoch": 1.367939308979962, + "grad_norm": 1.9200737476348877, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.7897387742996216, + "num_tokens": 389298379.0, + "step": 16589 + }, + { + "epoch": 1.3680217696050136, + "grad_norm": 1.898484706878662, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7236558198928833, + "num_tokens": 389334795.0, + "step": 16590 + }, + { + "epoch": 1.3681042302300652, + "grad_norm": 2.1807665824890137, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7187371253967285, + "num_tokens": 389359990.0, + "step": 16591 + }, + { + "epoch": 1.3681866908551168, + "grad_norm": 2.1859216690063477, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7680668830871582, + "num_tokens": 389384588.0, + "step": 16592 + }, + { + "epoch": 1.3682691514801681, + "grad_norm": 2.08246111869812, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7589035034179688, + "num_tokens": 389413123.0, + "step": 16593 + }, + { + "epoch": 1.3683516121052197, + "grad_norm": 2.2110753059387207, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7387592792510986, + "num_tokens": 389440994.0, + "step": 16594 + }, + { + "epoch": 1.3684340727302713, + "grad_norm": 1.9595005512237549, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.790880560874939, + "num_tokens": 389472804.0, + "step": 16595 + }, + { + "epoch": 1.3685165333553229, + "grad_norm": 1.786376714706421, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7821270823478699, + "num_tokens": 389509106.0, + "step": 16596 + }, + { + "epoch": 1.3685989939803744, + "grad_norm": 2.2144718170166016, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7799962162971497, + "num_tokens": 389534184.0, + "step": 16597 + }, + { + "epoch": 1.3686814546054258, + "grad_norm": 2.3374714851379395, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7316553592681885, + "num_tokens": 389558622.0, + "step": 16598 + }, + { + "epoch": 1.3687639152304776, + "grad_norm": 1.955381155014038, + "learning_rate": 1e-06, + "loss": 0.5612, + "mean_token_accuracy": 0.8136597871780396, + "num_tokens": 389586998.0, + "step": 16599 + }, + { + "epoch": 1.368846375855529, + "grad_norm": 1.9909613132476807, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7508784532546997, + "num_tokens": 389626143.0, + "step": 16600 + }, + { + "epoch": 1.3689288364805805, + "grad_norm": 1.838249683380127, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7448848485946655, + "num_tokens": 389660199.0, + "step": 16601 + }, + { + "epoch": 1.369011297105632, + "grad_norm": 2.1110312938690186, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7906665802001953, + "num_tokens": 389687714.0, + "step": 16602 + }, + { + "epoch": 1.3690937577306836, + "grad_norm": 2.230825424194336, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7821654081344604, + "num_tokens": 389711321.0, + "step": 16603 + }, + { + "epoch": 1.3691762183557352, + "grad_norm": 2.037714719772339, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.7672053575515747, + "num_tokens": 389740860.0, + "step": 16604 + }, + { + "epoch": 1.3692586789807866, + "grad_norm": 1.914973258972168, + "learning_rate": 1e-06, + "loss": 0.5354, + "mean_token_accuracy": 0.8213216066360474, + "num_tokens": 389772409.0, + "step": 16605 + }, + { + "epoch": 1.3693411396058381, + "grad_norm": 2.385476589202881, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7424913644790649, + "num_tokens": 389795626.0, + "step": 16606 + }, + { + "epoch": 1.3694236002308897, + "grad_norm": 1.781028151512146, + "learning_rate": 1e-06, + "loss": 0.6006, + "mean_token_accuracy": 0.8006950616836548, + "num_tokens": 389827877.0, + "step": 16607 + }, + { + "epoch": 1.3695060608559413, + "grad_norm": 1.9314137697219849, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7705460786819458, + "num_tokens": 389858814.0, + "step": 16608 + }, + { + "epoch": 1.3695885214809929, + "grad_norm": 2.142446279525757, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7777768969535828, + "num_tokens": 389885569.0, + "step": 16609 + }, + { + "epoch": 1.3696709821060444, + "grad_norm": 2.018315315246582, + "learning_rate": 1e-06, + "loss": 0.5757, + "mean_token_accuracy": 0.8061779737472534, + "num_tokens": 389910801.0, + "step": 16610 + }, + { + "epoch": 1.369753442731096, + "grad_norm": 1.9943474531173706, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7351104021072388, + "num_tokens": 389944804.0, + "step": 16611 + }, + { + "epoch": 1.3698359033561474, + "grad_norm": 1.818268895149231, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7515530586242676, + "num_tokens": 389979649.0, + "step": 16612 + }, + { + "epoch": 1.369918363981199, + "grad_norm": 2.0791447162628174, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7918485999107361, + "num_tokens": 390007954.0, + "step": 16613 + }, + { + "epoch": 1.3700008246062505, + "grad_norm": 2.092633008956909, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7539002895355225, + "num_tokens": 390036502.0, + "step": 16614 + }, + { + "epoch": 1.370083285231302, + "grad_norm": 1.8088668584823608, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7702522277832031, + "num_tokens": 390073435.0, + "step": 16615 + }, + { + "epoch": 1.3701657458563536, + "grad_norm": 1.979811429977417, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.7821550369262695, + "num_tokens": 390105987.0, + "step": 16616 + }, + { + "epoch": 1.370248206481405, + "grad_norm": 2.165584087371826, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7607665061950684, + "num_tokens": 390133892.0, + "step": 16617 + }, + { + "epoch": 1.3703306671064568, + "grad_norm": 2.415283679962158, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7498769164085388, + "num_tokens": 390154629.0, + "step": 16618 + }, + { + "epoch": 1.3704131277315081, + "grad_norm": 2.343611240386963, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7713643312454224, + "num_tokens": 390176533.0, + "step": 16619 + }, + { + "epoch": 1.3704955883565597, + "grad_norm": 1.711963176727295, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7682332992553711, + "num_tokens": 390215077.0, + "step": 16620 + }, + { + "epoch": 1.3705780489816113, + "grad_norm": 2.023310899734497, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7584929466247559, + "num_tokens": 390247031.0, + "step": 16621 + }, + { + "epoch": 1.3706605096066629, + "grad_norm": 2.2009379863739014, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.737967848777771, + "num_tokens": 390275100.0, + "step": 16622 + }, + { + "epoch": 1.3707429702317144, + "grad_norm": 1.9487289190292358, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7099040746688843, + "num_tokens": 390306582.0, + "step": 16623 + }, + { + "epoch": 1.3708254308567658, + "grad_norm": 2.1982154846191406, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7702881097793579, + "num_tokens": 390330796.0, + "step": 16624 + }, + { + "epoch": 1.3709078914818174, + "grad_norm": 1.9127393960952759, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7582948803901672, + "num_tokens": 390359732.0, + "step": 16625 + }, + { + "epoch": 1.370990352106869, + "grad_norm": 2.0154268741607666, + "learning_rate": 1e-06, + "loss": 0.5663, + "mean_token_accuracy": 0.8097438812255859, + "num_tokens": 390387643.0, + "step": 16626 + }, + { + "epoch": 1.3710728127319205, + "grad_norm": 2.0494167804718018, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7651153206825256, + "num_tokens": 390416533.0, + "step": 16627 + }, + { + "epoch": 1.371155273356972, + "grad_norm": 2.1019325256347656, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7286686897277832, + "num_tokens": 390444987.0, + "step": 16628 + }, + { + "epoch": 1.3712377339820236, + "grad_norm": 1.851015567779541, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7614508867263794, + "num_tokens": 390480847.0, + "step": 16629 + }, + { + "epoch": 1.3713201946070752, + "grad_norm": 2.1423935890197754, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7738564014434814, + "num_tokens": 390507990.0, + "step": 16630 + }, + { + "epoch": 1.3714026552321266, + "grad_norm": 2.004899024963379, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.7744734287261963, + "num_tokens": 390539900.0, + "step": 16631 + }, + { + "epoch": 1.3714851158571781, + "grad_norm": 1.905532717704773, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7166444063186646, + "num_tokens": 390575058.0, + "step": 16632 + }, + { + "epoch": 1.3715675764822297, + "grad_norm": 2.1160712242126465, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7747184038162231, + "num_tokens": 390600304.0, + "step": 16633 + }, + { + "epoch": 1.3716500371072813, + "grad_norm": 2.430631637573242, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.7372415065765381, + "num_tokens": 390622268.0, + "step": 16634 + }, + { + "epoch": 1.3717324977323329, + "grad_norm": 2.005331516265869, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7754320502281189, + "num_tokens": 390653628.0, + "step": 16635 + }, + { + "epoch": 1.3718149583573844, + "grad_norm": 2.0950677394866943, + "learning_rate": 1e-06, + "loss": 0.5952, + "mean_token_accuracy": 0.7944473028182983, + "num_tokens": 390677461.0, + "step": 16636 + }, + { + "epoch": 1.371897418982436, + "grad_norm": 1.7050960063934326, + "learning_rate": 1e-06, + "loss": 0.5639, + "mean_token_accuracy": 0.8082667589187622, + "num_tokens": 390713261.0, + "step": 16637 + }, + { + "epoch": 1.3719798796074874, + "grad_norm": 2.0416247844696045, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7267646789550781, + "num_tokens": 390743861.0, + "step": 16638 + }, + { + "epoch": 1.372062340232539, + "grad_norm": 2.4051456451416016, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7164257764816284, + "num_tokens": 390767068.0, + "step": 16639 + }, + { + "epoch": 1.3721448008575905, + "grad_norm": 1.8190997838974, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7374424934387207, + "num_tokens": 390807283.0, + "step": 16640 + }, + { + "epoch": 1.372227261482642, + "grad_norm": 1.7430886030197144, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7834789752960205, + "num_tokens": 390844345.0, + "step": 16641 + }, + { + "epoch": 1.3723097221076936, + "grad_norm": 2.035823345184326, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7587646245956421, + "num_tokens": 390874647.0, + "step": 16642 + }, + { + "epoch": 1.372392182732745, + "grad_norm": 2.126577138900757, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7558606863021851, + "num_tokens": 390903012.0, + "step": 16643 + }, + { + "epoch": 1.3724746433577968, + "grad_norm": 2.115227222442627, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.7574162483215332, + "num_tokens": 390929256.0, + "step": 16644 + }, + { + "epoch": 1.3725571039828481, + "grad_norm": 2.138356924057007, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7744625806808472, + "num_tokens": 390956533.0, + "step": 16645 + }, + { + "epoch": 1.3726395646078997, + "grad_norm": 2.0767345428466797, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7359000444412231, + "num_tokens": 390985312.0, + "step": 16646 + }, + { + "epoch": 1.3727220252329513, + "grad_norm": 1.8905141353607178, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7763519883155823, + "num_tokens": 391016751.0, + "step": 16647 + }, + { + "epoch": 1.3728044858580029, + "grad_norm": 1.9891058206558228, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7493358254432678, + "num_tokens": 391046382.0, + "step": 16648 + }, + { + "epoch": 1.3728869464830544, + "grad_norm": 2.182281732559204, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7814643979072571, + "num_tokens": 391071698.0, + "step": 16649 + }, + { + "epoch": 1.3729694071081058, + "grad_norm": 1.8681565523147583, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7717236280441284, + "num_tokens": 391104583.0, + "step": 16650 + }, + { + "epoch": 1.3730518677331573, + "grad_norm": 2.1986637115478516, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7600061893463135, + "num_tokens": 391130891.0, + "step": 16651 + }, + { + "epoch": 1.373134328358209, + "grad_norm": 1.9873571395874023, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7737880349159241, + "num_tokens": 391160283.0, + "step": 16652 + }, + { + "epoch": 1.3732167889832605, + "grad_norm": 2.234555244445801, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7494096755981445, + "num_tokens": 391185877.0, + "step": 16653 + }, + { + "epoch": 1.373299249608312, + "grad_norm": 2.132066488265991, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7488298416137695, + "num_tokens": 391213499.0, + "step": 16654 + }, + { + "epoch": 1.3733817102333636, + "grad_norm": 2.1480138301849365, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7607648372650146, + "num_tokens": 391241613.0, + "step": 16655 + }, + { + "epoch": 1.3734641708584152, + "grad_norm": 2.1745190620422363, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7534148693084717, + "num_tokens": 391267706.0, + "step": 16656 + }, + { + "epoch": 1.3735466314834666, + "grad_norm": 1.83269202709198, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.785273551940918, + "num_tokens": 391300066.0, + "step": 16657 + }, + { + "epoch": 1.3736290921085181, + "grad_norm": 1.894200325012207, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.74843430519104, + "num_tokens": 391335589.0, + "step": 16658 + }, + { + "epoch": 1.3737115527335697, + "grad_norm": 2.2090353965759277, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7452964782714844, + "num_tokens": 391361979.0, + "step": 16659 + }, + { + "epoch": 1.3737940133586213, + "grad_norm": 2.217942714691162, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7473561763763428, + "num_tokens": 391389590.0, + "step": 16660 + }, + { + "epoch": 1.3738764739836729, + "grad_norm": 2.1117103099823, + "learning_rate": 1e-06, + "loss": 0.5773, + "mean_token_accuracy": 0.7982640266418457, + "num_tokens": 391414790.0, + "step": 16661 + }, + { + "epoch": 1.3739589346087242, + "grad_norm": 2.15283465385437, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7661905884742737, + "num_tokens": 391440105.0, + "step": 16662 + }, + { + "epoch": 1.374041395233776, + "grad_norm": 1.709689736366272, + "learning_rate": 1e-06, + "loss": 0.6169, + "mean_token_accuracy": 0.8037972450256348, + "num_tokens": 391475497.0, + "step": 16663 + }, + { + "epoch": 1.3741238558588273, + "grad_norm": 1.9164698123931885, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.752426028251648, + "num_tokens": 391512960.0, + "step": 16664 + }, + { + "epoch": 1.374206316483879, + "grad_norm": 1.9832606315612793, + "learning_rate": 1e-06, + "loss": 0.5938, + "mean_token_accuracy": 0.8078281879425049, + "num_tokens": 391542908.0, + "step": 16665 + }, + { + "epoch": 1.3742887771089305, + "grad_norm": 1.9807034730911255, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7178077697753906, + "num_tokens": 391574217.0, + "step": 16666 + }, + { + "epoch": 1.374371237733982, + "grad_norm": 1.8755981922149658, + "learning_rate": 1e-06, + "loss": 0.5373, + "mean_token_accuracy": 0.8167935609817505, + "num_tokens": 391602231.0, + "step": 16667 + }, + { + "epoch": 1.3744536983590336, + "grad_norm": 2.60063099861145, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7100234627723694, + "num_tokens": 391624867.0, + "step": 16668 + }, + { + "epoch": 1.374536158984085, + "grad_norm": 1.9104304313659668, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.7428345680236816, + "num_tokens": 391658930.0, + "step": 16669 + }, + { + "epoch": 1.3746186196091366, + "grad_norm": 2.0226268768310547, + "learning_rate": 1e-06, + "loss": 0.7803, + "mean_token_accuracy": 0.7588532567024231, + "num_tokens": 391694662.0, + "step": 16670 + }, + { + "epoch": 1.3747010802341881, + "grad_norm": 1.9469443559646606, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7643433809280396, + "num_tokens": 391726239.0, + "step": 16671 + }, + { + "epoch": 1.3747835408592397, + "grad_norm": 1.9607298374176025, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7804881930351257, + "num_tokens": 391756557.0, + "step": 16672 + }, + { + "epoch": 1.3748660014842913, + "grad_norm": 2.064218282699585, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7754875421524048, + "num_tokens": 391787324.0, + "step": 16673 + }, + { + "epoch": 1.3749484621093429, + "grad_norm": 2.110447883605957, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7708073854446411, + "num_tokens": 391814158.0, + "step": 16674 + }, + { + "epoch": 1.3750309227343944, + "grad_norm": 1.9047152996063232, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7394794225692749, + "num_tokens": 391847377.0, + "step": 16675 + }, + { + "epoch": 1.3751133833594458, + "grad_norm": 1.7387235164642334, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7653903961181641, + "num_tokens": 391886077.0, + "step": 16676 + }, + { + "epoch": 1.3751958439844973, + "grad_norm": 1.9376673698425293, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.7874971628189087, + "num_tokens": 391916375.0, + "step": 16677 + }, + { + "epoch": 1.375278304609549, + "grad_norm": 1.7825430631637573, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.7914050817489624, + "num_tokens": 391950105.0, + "step": 16678 + }, + { + "epoch": 1.3753607652346005, + "grad_norm": 1.8557606935501099, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7649803757667542, + "num_tokens": 391984399.0, + "step": 16679 + }, + { + "epoch": 1.375443225859652, + "grad_norm": 2.6882174015045166, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7314191460609436, + "num_tokens": 392004164.0, + "step": 16680 + }, + { + "epoch": 1.3755256864847036, + "grad_norm": 1.7576684951782227, + "learning_rate": 1e-06, + "loss": 0.5786, + "mean_token_accuracy": 0.7897821068763733, + "num_tokens": 392034329.0, + "step": 16681 + }, + { + "epoch": 1.3756081471097552, + "grad_norm": 1.823551058769226, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.7877750396728516, + "num_tokens": 392065937.0, + "step": 16682 + }, + { + "epoch": 1.3756906077348066, + "grad_norm": 1.9316030740737915, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7703989744186401, + "num_tokens": 392097317.0, + "step": 16683 + }, + { + "epoch": 1.3757730683598581, + "grad_norm": 1.9823611974716187, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7685656547546387, + "num_tokens": 392124630.0, + "step": 16684 + }, + { + "epoch": 1.3758555289849097, + "grad_norm": 1.89362370967865, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7485959529876709, + "num_tokens": 392158826.0, + "step": 16685 + }, + { + "epoch": 1.3759379896099613, + "grad_norm": 1.974252700805664, + "learning_rate": 1e-06, + "loss": 0.6024, + "mean_token_accuracy": 0.7993450164794922, + "num_tokens": 392187967.0, + "step": 16686 + }, + { + "epoch": 1.3760204502350128, + "grad_norm": 1.9664386510849, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7615523338317871, + "num_tokens": 392219406.0, + "step": 16687 + }, + { + "epoch": 1.3761029108600642, + "grad_norm": 2.0660500526428223, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7620762586593628, + "num_tokens": 392250393.0, + "step": 16688 + }, + { + "epoch": 1.376185371485116, + "grad_norm": 1.8170888423919678, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7800358533859253, + "num_tokens": 392285750.0, + "step": 16689 + }, + { + "epoch": 1.3762678321101673, + "grad_norm": 1.7483621835708618, + "learning_rate": 1e-06, + "loss": 0.5466, + "mean_token_accuracy": 0.822496235370636, + "num_tokens": 392318040.0, + "step": 16690 + }, + { + "epoch": 1.376350292735219, + "grad_norm": 1.9677884578704834, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7305595874786377, + "num_tokens": 392348955.0, + "step": 16691 + }, + { + "epoch": 1.3764327533602705, + "grad_norm": 2.3799502849578857, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7367528676986694, + "num_tokens": 392373160.0, + "step": 16692 + }, + { + "epoch": 1.376515213985322, + "grad_norm": 1.9715851545333862, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7297669053077698, + "num_tokens": 392408106.0, + "step": 16693 + }, + { + "epoch": 1.3765976746103736, + "grad_norm": 1.9196312427520752, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7676211595535278, + "num_tokens": 392440351.0, + "step": 16694 + }, + { + "epoch": 1.376680135235425, + "grad_norm": 1.7896984815597534, + "learning_rate": 1e-06, + "loss": 0.5887, + "mean_token_accuracy": 0.8038276433944702, + "num_tokens": 392473952.0, + "step": 16695 + }, + { + "epoch": 1.3767625958604766, + "grad_norm": 1.8725414276123047, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7811741828918457, + "num_tokens": 392507432.0, + "step": 16696 + }, + { + "epoch": 1.3768450564855281, + "grad_norm": 2.370351791381836, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7563451528549194, + "num_tokens": 392531630.0, + "step": 16697 + }, + { + "epoch": 1.3769275171105797, + "grad_norm": 1.7704962491989136, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7774844169616699, + "num_tokens": 392565861.0, + "step": 16698 + }, + { + "epoch": 1.3770099777356313, + "grad_norm": 2.0996384620666504, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7479221820831299, + "num_tokens": 392594488.0, + "step": 16699 + }, + { + "epoch": 1.3770924383606828, + "grad_norm": 1.8725416660308838, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7731291055679321, + "num_tokens": 392625698.0, + "step": 16700 + }, + { + "epoch": 1.3771748989857344, + "grad_norm": 1.7939661741256714, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.7856981754302979, + "num_tokens": 392659029.0, + "step": 16701 + }, + { + "epoch": 1.3772573596107858, + "grad_norm": 2.552272319793701, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.734014630317688, + "num_tokens": 392678668.0, + "step": 16702 + }, + { + "epoch": 1.3773398202358373, + "grad_norm": 2.0195136070251465, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7485711574554443, + "num_tokens": 392707734.0, + "step": 16703 + }, + { + "epoch": 1.377422280860889, + "grad_norm": 1.8314666748046875, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7778952121734619, + "num_tokens": 392740514.0, + "step": 16704 + }, + { + "epoch": 1.3775047414859405, + "grad_norm": 1.7678741216659546, + "learning_rate": 1e-06, + "loss": 0.5911, + "mean_token_accuracy": 0.792421817779541, + "num_tokens": 392772467.0, + "step": 16705 + }, + { + "epoch": 1.377587202110992, + "grad_norm": 1.7893476486206055, + "learning_rate": 1e-06, + "loss": 0.7774, + "mean_token_accuracy": 0.7494410276412964, + "num_tokens": 392809310.0, + "step": 16706 + }, + { + "epoch": 1.3776696627360434, + "grad_norm": 2.058790445327759, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7702507376670837, + "num_tokens": 392837456.0, + "step": 16707 + }, + { + "epoch": 1.3777521233610952, + "grad_norm": 2.0945680141448975, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7117248177528381, + "num_tokens": 392865617.0, + "step": 16708 + }, + { + "epoch": 1.3778345839861466, + "grad_norm": 1.8143213987350464, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7274243235588074, + "num_tokens": 392901856.0, + "step": 16709 + }, + { + "epoch": 1.3779170446111981, + "grad_norm": 2.0459556579589844, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7746874094009399, + "num_tokens": 392928118.0, + "step": 16710 + }, + { + "epoch": 1.3779995052362497, + "grad_norm": 2.013306140899658, + "learning_rate": 1e-06, + "loss": 0.5666, + "mean_token_accuracy": 0.8026812672615051, + "num_tokens": 392955883.0, + "step": 16711 + }, + { + "epoch": 1.3780819658613013, + "grad_norm": 2.083387851715088, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7476233243942261, + "num_tokens": 392985169.0, + "step": 16712 + }, + { + "epoch": 1.3781644264863528, + "grad_norm": 1.87149977684021, + "learning_rate": 1e-06, + "loss": 0.5866, + "mean_token_accuracy": 0.8059597015380859, + "num_tokens": 393020034.0, + "step": 16713 + }, + { + "epoch": 1.3782468871114042, + "grad_norm": 1.8737088441848755, + "learning_rate": 1e-06, + "loss": 0.4931, + "mean_token_accuracy": 0.8340978622436523, + "num_tokens": 393053031.0, + "step": 16714 + }, + { + "epoch": 1.3783293477364558, + "grad_norm": 1.9173754453659058, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.7681122422218323, + "num_tokens": 393085978.0, + "step": 16715 + }, + { + "epoch": 1.3784118083615073, + "grad_norm": 2.104372262954712, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.8009123206138611, + "num_tokens": 393112251.0, + "step": 16716 + }, + { + "epoch": 1.378494268986559, + "grad_norm": 1.952049732208252, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7728478908538818, + "num_tokens": 393142980.0, + "step": 16717 + }, + { + "epoch": 1.3785767296116105, + "grad_norm": 2.008594036102295, + "learning_rate": 1e-06, + "loss": 0.7498, + "mean_token_accuracy": 0.7539016604423523, + "num_tokens": 393172500.0, + "step": 16718 + }, + { + "epoch": 1.378659190236662, + "grad_norm": 2.1878128051757812, + "learning_rate": 1e-06, + "loss": 0.5775, + "mean_token_accuracy": 0.8031786680221558, + "num_tokens": 393199418.0, + "step": 16719 + }, + { + "epoch": 1.3787416508617136, + "grad_norm": 2.477617025375366, + "learning_rate": 1e-06, + "loss": 0.7702, + "mean_token_accuracy": 0.7483761310577393, + "num_tokens": 393220433.0, + "step": 16720 + }, + { + "epoch": 1.378824111486765, + "grad_norm": 2.1023881435394287, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.75260990858078, + "num_tokens": 393246736.0, + "step": 16721 + }, + { + "epoch": 1.3789065721118166, + "grad_norm": 1.7481062412261963, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7448772192001343, + "num_tokens": 393284043.0, + "step": 16722 + }, + { + "epoch": 1.3789890327368681, + "grad_norm": 1.9180916547775269, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.744143009185791, + "num_tokens": 393317257.0, + "step": 16723 + }, + { + "epoch": 1.3790714933619197, + "grad_norm": 1.9710899591445923, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7365174293518066, + "num_tokens": 393348922.0, + "step": 16724 + }, + { + "epoch": 1.3791539539869713, + "grad_norm": 2.0470101833343506, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7595133185386658, + "num_tokens": 393377208.0, + "step": 16725 + }, + { + "epoch": 1.3792364146120226, + "grad_norm": 1.963019847869873, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.7883859276771545, + "num_tokens": 393408134.0, + "step": 16726 + }, + { + "epoch": 1.3793188752370744, + "grad_norm": 1.996288537979126, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.7686809301376343, + "num_tokens": 393438882.0, + "step": 16727 + }, + { + "epoch": 1.3794013358621258, + "grad_norm": 2.0720834732055664, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.761004626750946, + "num_tokens": 393470678.0, + "step": 16728 + }, + { + "epoch": 1.3794837964871773, + "grad_norm": 2.0531435012817383, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7549774646759033, + "num_tokens": 393504101.0, + "step": 16729 + }, + { + "epoch": 1.379566257112229, + "grad_norm": 2.0329697132110596, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.724772572517395, + "num_tokens": 393534236.0, + "step": 16730 + }, + { + "epoch": 1.3796487177372805, + "grad_norm": 2.0416157245635986, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7456477880477905, + "num_tokens": 393565524.0, + "step": 16731 + }, + { + "epoch": 1.379731178362332, + "grad_norm": 1.8237017393112183, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7470987439155579, + "num_tokens": 393599106.0, + "step": 16732 + }, + { + "epoch": 1.3798136389873834, + "grad_norm": 1.877778172492981, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.763208270072937, + "num_tokens": 393632249.0, + "step": 16733 + }, + { + "epoch": 1.3798960996124352, + "grad_norm": 2.0979082584381104, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7573082447052002, + "num_tokens": 393658587.0, + "step": 16734 + }, + { + "epoch": 1.3799785602374866, + "grad_norm": 2.255204677581787, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7649781703948975, + "num_tokens": 393683112.0, + "step": 16735 + }, + { + "epoch": 1.3800610208625381, + "grad_norm": 2.3190858364105225, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7516652345657349, + "num_tokens": 393707294.0, + "step": 16736 + }, + { + "epoch": 1.3801434814875897, + "grad_norm": 1.675911784172058, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7626363039016724, + "num_tokens": 393744374.0, + "step": 16737 + }, + { + "epoch": 1.3802259421126413, + "grad_norm": 2.1803183555603027, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7642350196838379, + "num_tokens": 393770665.0, + "step": 16738 + }, + { + "epoch": 1.3803084027376928, + "grad_norm": 2.384312868118286, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7385629415512085, + "num_tokens": 393795067.0, + "step": 16739 + }, + { + "epoch": 1.3803908633627442, + "grad_norm": 2.12846302986145, + "learning_rate": 1e-06, + "loss": 0.5663, + "mean_token_accuracy": 0.7901334166526794, + "num_tokens": 393820766.0, + "step": 16740 + }, + { + "epoch": 1.3804733239877958, + "grad_norm": 1.9856057167053223, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7771733403205872, + "num_tokens": 393851606.0, + "step": 16741 + }, + { + "epoch": 1.3805557846128473, + "grad_norm": 2.122018337249756, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7708290815353394, + "num_tokens": 393877823.0, + "step": 16742 + }, + { + "epoch": 1.380638245237899, + "grad_norm": 2.0032877922058105, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7221196889877319, + "num_tokens": 393910720.0, + "step": 16743 + }, + { + "epoch": 1.3807207058629505, + "grad_norm": 1.8012264966964722, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7750956416130066, + "num_tokens": 393947621.0, + "step": 16744 + }, + { + "epoch": 1.380803166488002, + "grad_norm": 2.018911838531494, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7851963639259338, + "num_tokens": 393977935.0, + "step": 16745 + }, + { + "epoch": 1.3808856271130536, + "grad_norm": 1.9637559652328491, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7649672031402588, + "num_tokens": 394010511.0, + "step": 16746 + }, + { + "epoch": 1.380968087738105, + "grad_norm": 1.9487524032592773, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.7879037857055664, + "num_tokens": 394037802.0, + "step": 16747 + }, + { + "epoch": 1.3810505483631565, + "grad_norm": 1.83603835105896, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8008537292480469, + "num_tokens": 394070086.0, + "step": 16748 + }, + { + "epoch": 1.3811330089882081, + "grad_norm": 1.8516517877578735, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7286221981048584, + "num_tokens": 394105877.0, + "step": 16749 + }, + { + "epoch": 1.3812154696132597, + "grad_norm": 1.817705512046814, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7343883514404297, + "num_tokens": 394142071.0, + "step": 16750 + }, + { + "epoch": 1.3812979302383113, + "grad_norm": 1.8104898929595947, + "learning_rate": 1e-06, + "loss": 0.4391, + "mean_token_accuracy": 0.8494929075241089, + "num_tokens": 394171031.0, + "step": 16751 + }, + { + "epoch": 1.3813803908633626, + "grad_norm": 1.8007732629776, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7619824409484863, + "num_tokens": 394205891.0, + "step": 16752 + }, + { + "epoch": 1.3814628514884144, + "grad_norm": 2.239367961883545, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7201949954032898, + "num_tokens": 394231493.0, + "step": 16753 + }, + { + "epoch": 1.3815453121134658, + "grad_norm": 2.0427348613739014, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7555158138275146, + "num_tokens": 394261312.0, + "step": 16754 + }, + { + "epoch": 1.3816277727385173, + "grad_norm": 2.0310475826263428, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.7917864322662354, + "num_tokens": 394289976.0, + "step": 16755 + }, + { + "epoch": 1.381710233363569, + "grad_norm": 2.052098274230957, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7740128636360168, + "num_tokens": 394317761.0, + "step": 16756 + }, + { + "epoch": 1.3817926939886205, + "grad_norm": 1.9454425573349, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.7916052341461182, + "num_tokens": 394346238.0, + "step": 16757 + }, + { + "epoch": 1.381875154613672, + "grad_norm": 1.905746340751648, + "learning_rate": 1e-06, + "loss": 0.5161, + "mean_token_accuracy": 0.8208548426628113, + "num_tokens": 394375815.0, + "step": 16758 + }, + { + "epoch": 1.3819576152387234, + "grad_norm": 2.1127684116363525, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7285830974578857, + "num_tokens": 394405785.0, + "step": 16759 + }, + { + "epoch": 1.382040075863775, + "grad_norm": 1.750327467918396, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.7967016696929932, + "num_tokens": 394440134.0, + "step": 16760 + }, + { + "epoch": 1.3821225364888265, + "grad_norm": 2.090876817703247, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7721739411354065, + "num_tokens": 394465309.0, + "step": 16761 + }, + { + "epoch": 1.3822049971138781, + "grad_norm": 1.8617218732833862, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7715184688568115, + "num_tokens": 394497718.0, + "step": 16762 + }, + { + "epoch": 1.3822874577389297, + "grad_norm": 2.2942376136779785, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7303941249847412, + "num_tokens": 394523382.0, + "step": 16763 + }, + { + "epoch": 1.3823699183639813, + "grad_norm": 2.0271642208099365, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7410504817962646, + "num_tokens": 394552365.0, + "step": 16764 + }, + { + "epoch": 1.3824523789890328, + "grad_norm": 2.0210776329040527, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7486828565597534, + "num_tokens": 394583268.0, + "step": 16765 + }, + { + "epoch": 1.3825348396140842, + "grad_norm": 2.060741901397705, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7481999397277832, + "num_tokens": 394614168.0, + "step": 16766 + }, + { + "epoch": 1.3826173002391358, + "grad_norm": 1.748260498046875, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7281442880630493, + "num_tokens": 394650202.0, + "step": 16767 + }, + { + "epoch": 1.3826997608641873, + "grad_norm": 2.152189016342163, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7557626962661743, + "num_tokens": 394677474.0, + "step": 16768 + }, + { + "epoch": 1.382782221489239, + "grad_norm": 1.9208070039749146, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.7807720899581909, + "num_tokens": 394707637.0, + "step": 16769 + }, + { + "epoch": 1.3828646821142905, + "grad_norm": 1.7627421617507935, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7804876565933228, + "num_tokens": 394741893.0, + "step": 16770 + }, + { + "epoch": 1.3829471427393418, + "grad_norm": 2.116766929626465, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7755531668663025, + "num_tokens": 394770263.0, + "step": 16771 + }, + { + "epoch": 1.3830296033643936, + "grad_norm": 2.0686192512512207, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.7815076112747192, + "num_tokens": 394797515.0, + "step": 16772 + }, + { + "epoch": 1.383112063989445, + "grad_norm": 1.8628567457199097, + "learning_rate": 1e-06, + "loss": 0.6032, + "mean_token_accuracy": 0.8034290075302124, + "num_tokens": 394830319.0, + "step": 16773 + }, + { + "epoch": 1.3831945246144965, + "grad_norm": 1.952950358390808, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7573085427284241, + "num_tokens": 394863397.0, + "step": 16774 + }, + { + "epoch": 1.3832769852395481, + "grad_norm": 2.016902208328247, + "learning_rate": 1e-06, + "loss": 0.6124, + "mean_token_accuracy": 0.800782322883606, + "num_tokens": 394893119.0, + "step": 16775 + }, + { + "epoch": 1.3833594458645997, + "grad_norm": 2.043405532836914, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.769239068031311, + "num_tokens": 394923844.0, + "step": 16776 + }, + { + "epoch": 1.3834419064896513, + "grad_norm": 2.044445037841797, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7392721176147461, + "num_tokens": 394955519.0, + "step": 16777 + }, + { + "epoch": 1.3835243671147026, + "grad_norm": 1.9303773641586304, + "learning_rate": 1e-06, + "loss": 0.6062, + "mean_token_accuracy": 0.7972074747085571, + "num_tokens": 394984610.0, + "step": 16778 + }, + { + "epoch": 1.3836068277397542, + "grad_norm": 1.9513224363327026, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7548056840896606, + "num_tokens": 395014814.0, + "step": 16779 + }, + { + "epoch": 1.3836892883648058, + "grad_norm": 2.1048808097839355, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.738080620765686, + "num_tokens": 395041596.0, + "step": 16780 + }, + { + "epoch": 1.3837717489898573, + "grad_norm": 2.2732813358306885, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7264013290405273, + "num_tokens": 395067046.0, + "step": 16781 + }, + { + "epoch": 1.383854209614909, + "grad_norm": 1.8642066717147827, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.7976658344268799, + "num_tokens": 395099865.0, + "step": 16782 + }, + { + "epoch": 1.3839366702399605, + "grad_norm": 1.6888436079025269, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.7864623069763184, + "num_tokens": 395136621.0, + "step": 16783 + }, + { + "epoch": 1.384019130865012, + "grad_norm": 1.9438821077346802, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.750779390335083, + "num_tokens": 395167136.0, + "step": 16784 + }, + { + "epoch": 1.3841015914900634, + "grad_norm": 2.2490134239196777, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7542884349822998, + "num_tokens": 395191922.0, + "step": 16785 + }, + { + "epoch": 1.384184052115115, + "grad_norm": 1.7828365564346313, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7294360399246216, + "num_tokens": 395226794.0, + "step": 16786 + }, + { + "epoch": 1.3842665127401665, + "grad_norm": 1.9613600969314575, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.758208692073822, + "num_tokens": 395257141.0, + "step": 16787 + }, + { + "epoch": 1.3843489733652181, + "grad_norm": 1.9761810302734375, + "learning_rate": 1e-06, + "loss": 0.5788, + "mean_token_accuracy": 0.8054250478744507, + "num_tokens": 395284192.0, + "step": 16788 + }, + { + "epoch": 1.3844314339902697, + "grad_norm": 1.9622045755386353, + "learning_rate": 1e-06, + "loss": 0.6155, + "mean_token_accuracy": 0.7943670153617859, + "num_tokens": 395313362.0, + "step": 16789 + }, + { + "epoch": 1.3845138946153213, + "grad_norm": 2.149845838546753, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.7804527282714844, + "num_tokens": 395338126.0, + "step": 16790 + }, + { + "epoch": 1.3845963552403728, + "grad_norm": 2.404723882675171, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7497702240943909, + "num_tokens": 395360277.0, + "step": 16791 + }, + { + "epoch": 1.3846788158654242, + "grad_norm": 2.003356456756592, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7682821750640869, + "num_tokens": 395388008.0, + "step": 16792 + }, + { + "epoch": 1.3847612764904758, + "grad_norm": 1.8138304948806763, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7611250877380371, + "num_tokens": 395425463.0, + "step": 16793 + }, + { + "epoch": 1.3848437371155273, + "grad_norm": 1.937121868133545, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7640700936317444, + "num_tokens": 395457981.0, + "step": 16794 + }, + { + "epoch": 1.384926197740579, + "grad_norm": 1.95884370803833, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.764521062374115, + "num_tokens": 395487774.0, + "step": 16795 + }, + { + "epoch": 1.3850086583656305, + "grad_norm": 2.4692373275756836, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7587268948554993, + "num_tokens": 395510609.0, + "step": 16796 + }, + { + "epoch": 1.3850911189906818, + "grad_norm": 2.0864617824554443, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7566585540771484, + "num_tokens": 395539017.0, + "step": 16797 + }, + { + "epoch": 1.3851735796157336, + "grad_norm": 2.0081896781921387, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7396570444107056, + "num_tokens": 395568862.0, + "step": 16798 + }, + { + "epoch": 1.385256040240785, + "grad_norm": 2.0017573833465576, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7515357732772827, + "num_tokens": 395598165.0, + "step": 16799 + }, + { + "epoch": 1.3853385008658365, + "grad_norm": 2.0479519367218018, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7233378291130066, + "num_tokens": 395626610.0, + "step": 16800 + }, + { + "epoch": 1.3854209614908881, + "grad_norm": 2.0228962898254395, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7152239084243774, + "num_tokens": 395656242.0, + "step": 16801 + }, + { + "epoch": 1.3855034221159397, + "grad_norm": 1.8819783926010132, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7801798582077026, + "num_tokens": 395690340.0, + "step": 16802 + }, + { + "epoch": 1.3855858827409913, + "grad_norm": 1.8429672718048096, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7516441345214844, + "num_tokens": 395727823.0, + "step": 16803 + }, + { + "epoch": 1.3856683433660426, + "grad_norm": 1.7132792472839355, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7734119296073914, + "num_tokens": 395762561.0, + "step": 16804 + }, + { + "epoch": 1.3857508039910942, + "grad_norm": 1.7938945293426514, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7657749652862549, + "num_tokens": 395797703.0, + "step": 16805 + }, + { + "epoch": 1.3858332646161458, + "grad_norm": 1.8295180797576904, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7527909874916077, + "num_tokens": 395833671.0, + "step": 16806 + }, + { + "epoch": 1.3859157252411973, + "grad_norm": 1.9057323932647705, + "learning_rate": 1e-06, + "loss": 0.5659, + "mean_token_accuracy": 0.8130254745483398, + "num_tokens": 395864660.0, + "step": 16807 + }, + { + "epoch": 1.385998185866249, + "grad_norm": 1.9220093488693237, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7387732267379761, + "num_tokens": 395897070.0, + "step": 16808 + }, + { + "epoch": 1.3860806464913005, + "grad_norm": 1.8800697326660156, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.7934640645980835, + "num_tokens": 395927530.0, + "step": 16809 + }, + { + "epoch": 1.386163107116352, + "grad_norm": 1.9220637083053589, + "learning_rate": 1e-06, + "loss": 0.5948, + "mean_token_accuracy": 0.802081823348999, + "num_tokens": 395957232.0, + "step": 16810 + }, + { + "epoch": 1.3862455677414034, + "grad_norm": 1.7175509929656982, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7711145877838135, + "num_tokens": 395998311.0, + "step": 16811 + }, + { + "epoch": 1.386328028366455, + "grad_norm": 1.5940176248550415, + "learning_rate": 1e-06, + "loss": 0.5755, + "mean_token_accuracy": 0.8047749996185303, + "num_tokens": 396039180.0, + "step": 16812 + }, + { + "epoch": 1.3864104889915065, + "grad_norm": 2.25236439704895, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7331797480583191, + "num_tokens": 396062340.0, + "step": 16813 + }, + { + "epoch": 1.3864929496165581, + "grad_norm": 1.9746657609939575, + "learning_rate": 1e-06, + "loss": 0.5955, + "mean_token_accuracy": 0.7951374053955078, + "num_tokens": 396092238.0, + "step": 16814 + }, + { + "epoch": 1.3865754102416097, + "grad_norm": 1.6914849281311035, + "learning_rate": 1e-06, + "loss": 0.5745, + "mean_token_accuracy": 0.8094390034675598, + "num_tokens": 396130996.0, + "step": 16815 + }, + { + "epoch": 1.386657870866661, + "grad_norm": 1.9707762002944946, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7340709567070007, + "num_tokens": 396161694.0, + "step": 16816 + }, + { + "epoch": 1.3867403314917128, + "grad_norm": 2.3776772022247314, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7440159320831299, + "num_tokens": 396186651.0, + "step": 16817 + }, + { + "epoch": 1.3868227921167642, + "grad_norm": 2.1596832275390625, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7320756912231445, + "num_tokens": 396213358.0, + "step": 16818 + }, + { + "epoch": 1.3869052527418158, + "grad_norm": 2.0317587852478027, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7687700986862183, + "num_tokens": 396240439.0, + "step": 16819 + }, + { + "epoch": 1.3869877133668673, + "grad_norm": 2.3489508628845215, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7263972759246826, + "num_tokens": 396264017.0, + "step": 16820 + }, + { + "epoch": 1.387070173991919, + "grad_norm": 1.9663139581680298, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7716515064239502, + "num_tokens": 396296547.0, + "step": 16821 + }, + { + "epoch": 1.3871526346169705, + "grad_norm": 1.9621134996414185, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.7594548463821411, + "num_tokens": 396329677.0, + "step": 16822 + }, + { + "epoch": 1.3872350952420218, + "grad_norm": 1.9351129531860352, + "learning_rate": 1e-06, + "loss": 0.5962, + "mean_token_accuracy": 0.800521731376648, + "num_tokens": 396359775.0, + "step": 16823 + }, + { + "epoch": 1.3873175558670734, + "grad_norm": 2.060206174850464, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7558834552764893, + "num_tokens": 396390788.0, + "step": 16824 + }, + { + "epoch": 1.387400016492125, + "grad_norm": 1.9082279205322266, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7808046340942383, + "num_tokens": 396420836.0, + "step": 16825 + }, + { + "epoch": 1.3874824771171765, + "grad_norm": 1.8850879669189453, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7626306414604187, + "num_tokens": 396454149.0, + "step": 16826 + }, + { + "epoch": 1.387564937742228, + "grad_norm": 2.1487269401550293, + "learning_rate": 1e-06, + "loss": 0.5719, + "mean_token_accuracy": 0.7956283092498779, + "num_tokens": 396478911.0, + "step": 16827 + }, + { + "epoch": 1.3876473983672797, + "grad_norm": 1.8550070524215698, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7592898607254028, + "num_tokens": 396511335.0, + "step": 16828 + }, + { + "epoch": 1.3877298589923313, + "grad_norm": 2.029669761657715, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.7768852710723877, + "num_tokens": 396540136.0, + "step": 16829 + }, + { + "epoch": 1.3878123196173826, + "grad_norm": 1.8448430299758911, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.7729228734970093, + "num_tokens": 396574094.0, + "step": 16830 + }, + { + "epoch": 1.3878947802424342, + "grad_norm": 1.9639972448349, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7733163237571716, + "num_tokens": 396603206.0, + "step": 16831 + }, + { + "epoch": 1.3879772408674858, + "grad_norm": 1.8062975406646729, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7583820819854736, + "num_tokens": 396637174.0, + "step": 16832 + }, + { + "epoch": 1.3880597014925373, + "grad_norm": 2.2376790046691895, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.7771569490432739, + "num_tokens": 396660139.0, + "step": 16833 + }, + { + "epoch": 1.388142162117589, + "grad_norm": 2.0046920776367188, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7505483031272888, + "num_tokens": 396689781.0, + "step": 16834 + }, + { + "epoch": 1.3882246227426405, + "grad_norm": 1.985849142074585, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.775399386882782, + "num_tokens": 396718843.0, + "step": 16835 + }, + { + "epoch": 1.388307083367692, + "grad_norm": 1.7991913557052612, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7897523045539856, + "num_tokens": 396751125.0, + "step": 16836 + }, + { + "epoch": 1.3883895439927434, + "grad_norm": 1.978369116783142, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.759620189666748, + "num_tokens": 396779809.0, + "step": 16837 + }, + { + "epoch": 1.388472004617795, + "grad_norm": 1.9635891914367676, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7376595735549927, + "num_tokens": 396814791.0, + "step": 16838 + }, + { + "epoch": 1.3885544652428465, + "grad_norm": 1.9279344081878662, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7649054527282715, + "num_tokens": 396842379.0, + "step": 16839 + }, + { + "epoch": 1.388636925867898, + "grad_norm": 2.005625009536743, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7350713014602661, + "num_tokens": 396873805.0, + "step": 16840 + }, + { + "epoch": 1.3887193864929497, + "grad_norm": 2.1275579929351807, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7294735908508301, + "num_tokens": 396900337.0, + "step": 16841 + }, + { + "epoch": 1.388801847118001, + "grad_norm": 2.023005247116089, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7643797397613525, + "num_tokens": 396929231.0, + "step": 16842 + }, + { + "epoch": 1.3888843077430528, + "grad_norm": 2.164001703262329, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.7757213115692139, + "num_tokens": 396956494.0, + "step": 16843 + }, + { + "epoch": 1.3889667683681042, + "grad_norm": 2.0201852321624756, + "learning_rate": 1e-06, + "loss": 0.5931, + "mean_token_accuracy": 0.8050706386566162, + "num_tokens": 396984067.0, + "step": 16844 + }, + { + "epoch": 1.3890492289931557, + "grad_norm": 1.9141004085540771, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.7898494601249695, + "num_tokens": 397016049.0, + "step": 16845 + }, + { + "epoch": 1.3891316896182073, + "grad_norm": 2.3186023235321045, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7658575773239136, + "num_tokens": 397041048.0, + "step": 16846 + }, + { + "epoch": 1.389214150243259, + "grad_norm": 1.8232378959655762, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7775717377662659, + "num_tokens": 397074984.0, + "step": 16847 + }, + { + "epoch": 1.3892966108683105, + "grad_norm": 2.235835075378418, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7392844557762146, + "num_tokens": 397102212.0, + "step": 16848 + }, + { + "epoch": 1.3893790714933618, + "grad_norm": 2.1651206016540527, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.769331693649292, + "num_tokens": 397127824.0, + "step": 16849 + }, + { + "epoch": 1.3894615321184134, + "grad_norm": 1.8406810760498047, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.741026759147644, + "num_tokens": 397163407.0, + "step": 16850 + }, + { + "epoch": 1.389543992743465, + "grad_norm": 2.1620092391967773, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.758772611618042, + "num_tokens": 397188723.0, + "step": 16851 + }, + { + "epoch": 1.3896264533685165, + "grad_norm": 1.8423856496810913, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.7966310977935791, + "num_tokens": 397220145.0, + "step": 16852 + }, + { + "epoch": 1.389708913993568, + "grad_norm": 2.063288450241089, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7564705610275269, + "num_tokens": 397248670.0, + "step": 16853 + }, + { + "epoch": 1.3897913746186197, + "grad_norm": 1.8645515441894531, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7715096473693848, + "num_tokens": 397280682.0, + "step": 16854 + }, + { + "epoch": 1.3898738352436713, + "grad_norm": 1.7937341928482056, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.7900882363319397, + "num_tokens": 397314457.0, + "step": 16855 + }, + { + "epoch": 1.3899562958687226, + "grad_norm": 1.819392442703247, + "learning_rate": 1e-06, + "loss": 0.5675, + "mean_token_accuracy": 0.8101987242698669, + "num_tokens": 397345712.0, + "step": 16856 + }, + { + "epoch": 1.3900387564937742, + "grad_norm": 1.9118949174880981, + "learning_rate": 1e-06, + "loss": 0.552, + "mean_token_accuracy": 0.8182377219200134, + "num_tokens": 397375214.0, + "step": 16857 + }, + { + "epoch": 1.3901212171188257, + "grad_norm": 1.7347848415374756, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7238278985023499, + "num_tokens": 397412806.0, + "step": 16858 + }, + { + "epoch": 1.3902036777438773, + "grad_norm": 2.0990607738494873, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7482742071151733, + "num_tokens": 397441638.0, + "step": 16859 + }, + { + "epoch": 1.390286138368929, + "grad_norm": 1.7236665487289429, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.7853679656982422, + "num_tokens": 397475935.0, + "step": 16860 + }, + { + "epoch": 1.3903685989939802, + "grad_norm": 1.8307114839553833, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.7902052998542786, + "num_tokens": 397505237.0, + "step": 16861 + }, + { + "epoch": 1.390451059619032, + "grad_norm": 2.332367420196533, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.732473611831665, + "num_tokens": 397531274.0, + "step": 16862 + }, + { + "epoch": 1.3905335202440834, + "grad_norm": 2.0237720012664795, + "learning_rate": 1e-06, + "loss": 0.5395, + "mean_token_accuracy": 0.815139889717102, + "num_tokens": 397559257.0, + "step": 16863 + }, + { + "epoch": 1.390615980869135, + "grad_norm": 1.9896411895751953, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.7883623838424683, + "num_tokens": 397588096.0, + "step": 16864 + }, + { + "epoch": 1.3906984414941865, + "grad_norm": 1.964945912361145, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7768512964248657, + "num_tokens": 397619489.0, + "step": 16865 + }, + { + "epoch": 1.390780902119238, + "grad_norm": 1.8621643781661987, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7617055177688599, + "num_tokens": 397650926.0, + "step": 16866 + }, + { + "epoch": 1.3908633627442897, + "grad_norm": 2.0219414234161377, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7364850044250488, + "num_tokens": 397680029.0, + "step": 16867 + }, + { + "epoch": 1.390945823369341, + "grad_norm": 1.7789667844772339, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7822664976119995, + "num_tokens": 397713496.0, + "step": 16868 + }, + { + "epoch": 1.3910282839943926, + "grad_norm": 2.192357301712036, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7138912677764893, + "num_tokens": 397741672.0, + "step": 16869 + }, + { + "epoch": 1.3911107446194442, + "grad_norm": 1.8702079057693481, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.779641330242157, + "num_tokens": 397772538.0, + "step": 16870 + }, + { + "epoch": 1.3911932052444957, + "grad_norm": 2.0215795040130615, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7285851240158081, + "num_tokens": 397805924.0, + "step": 16871 + }, + { + "epoch": 1.3912756658695473, + "grad_norm": 1.9315911531448364, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.758484423160553, + "num_tokens": 397838767.0, + "step": 16872 + }, + { + "epoch": 1.391358126494599, + "grad_norm": 1.8987231254577637, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7255290746688843, + "num_tokens": 397873087.0, + "step": 16873 + }, + { + "epoch": 1.3914405871196505, + "grad_norm": 2.0083200931549072, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7693694829940796, + "num_tokens": 397903408.0, + "step": 16874 + }, + { + "epoch": 1.3915230477447018, + "grad_norm": 2.0842957496643066, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7628884315490723, + "num_tokens": 397932603.0, + "step": 16875 + }, + { + "epoch": 1.3916055083697534, + "grad_norm": 1.8834308385849, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7753843665122986, + "num_tokens": 397964567.0, + "step": 16876 + }, + { + "epoch": 1.391687968994805, + "grad_norm": 1.8532586097717285, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.742415189743042, + "num_tokens": 397996626.0, + "step": 16877 + }, + { + "epoch": 1.3917704296198565, + "grad_norm": 1.8588755130767822, + "learning_rate": 1e-06, + "loss": 0.5969, + "mean_token_accuracy": 0.7835631966590881, + "num_tokens": 398026064.0, + "step": 16878 + }, + { + "epoch": 1.391852890244908, + "grad_norm": 2.0849173069000244, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7564035058021545, + "num_tokens": 398053051.0, + "step": 16879 + }, + { + "epoch": 1.3919353508699597, + "grad_norm": 1.9239569902420044, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7366843223571777, + "num_tokens": 398085665.0, + "step": 16880 + }, + { + "epoch": 1.3920178114950112, + "grad_norm": 2.0034704208374023, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7424089312553406, + "num_tokens": 398116092.0, + "step": 16881 + }, + { + "epoch": 1.3921002721200626, + "grad_norm": 1.9343945980072021, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7455478310585022, + "num_tokens": 398145776.0, + "step": 16882 + }, + { + "epoch": 1.3921827327451142, + "grad_norm": 2.3180196285247803, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7258327007293701, + "num_tokens": 398173188.0, + "step": 16883 + }, + { + "epoch": 1.3922651933701657, + "grad_norm": 2.1653482913970947, + "learning_rate": 1e-06, + "loss": 0.588, + "mean_token_accuracy": 0.7999007701873779, + "num_tokens": 398198773.0, + "step": 16884 + }, + { + "epoch": 1.3923476539952173, + "grad_norm": 1.8908915519714355, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7666845917701721, + "num_tokens": 398229188.0, + "step": 16885 + }, + { + "epoch": 1.3924301146202689, + "grad_norm": 2.1990134716033936, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.7570433020591736, + "num_tokens": 398254173.0, + "step": 16886 + }, + { + "epoch": 1.3925125752453202, + "grad_norm": 1.8207945823669434, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.780574381351471, + "num_tokens": 398287066.0, + "step": 16887 + }, + { + "epoch": 1.392595035870372, + "grad_norm": 1.9918107986450195, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.767418384552002, + "num_tokens": 398316355.0, + "step": 16888 + }, + { + "epoch": 1.3926774964954234, + "grad_norm": 2.2221810817718506, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7287002801895142, + "num_tokens": 398344122.0, + "step": 16889 + }, + { + "epoch": 1.392759957120475, + "grad_norm": 2.524221897125244, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.744475781917572, + "num_tokens": 398364626.0, + "step": 16890 + }, + { + "epoch": 1.3928424177455265, + "grad_norm": 1.8229398727416992, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7590450048446655, + "num_tokens": 398400127.0, + "step": 16891 + }, + { + "epoch": 1.392924878370578, + "grad_norm": 2.1067495346069336, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7200398445129395, + "num_tokens": 398429501.0, + "step": 16892 + }, + { + "epoch": 1.3930073389956297, + "grad_norm": 2.341254711151123, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7235949039459229, + "num_tokens": 398453604.0, + "step": 16893 + }, + { + "epoch": 1.393089799620681, + "grad_norm": 1.9497650861740112, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7531212568283081, + "num_tokens": 398485527.0, + "step": 16894 + }, + { + "epoch": 1.3931722602457326, + "grad_norm": 2.1176185607910156, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7831676006317139, + "num_tokens": 398510501.0, + "step": 16895 + }, + { + "epoch": 1.3932547208707842, + "grad_norm": 1.8626765012741089, + "learning_rate": 1e-06, + "loss": 0.5863, + "mean_token_accuracy": 0.8081164956092834, + "num_tokens": 398542123.0, + "step": 16896 + }, + { + "epoch": 1.3933371814958357, + "grad_norm": 1.8697651624679565, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.74416184425354, + "num_tokens": 398576474.0, + "step": 16897 + }, + { + "epoch": 1.3934196421208873, + "grad_norm": 2.1373589038848877, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7739858627319336, + "num_tokens": 398603920.0, + "step": 16898 + }, + { + "epoch": 1.3935021027459389, + "grad_norm": 1.8805404901504517, + "learning_rate": 1e-06, + "loss": 0.5197, + "mean_token_accuracy": 0.8253597617149353, + "num_tokens": 398634504.0, + "step": 16899 + }, + { + "epoch": 1.3935845633709905, + "grad_norm": 2.1268093585968018, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7554624080657959, + "num_tokens": 398662811.0, + "step": 16900 + }, + { + "epoch": 1.3936670239960418, + "grad_norm": 1.8703312873840332, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7727062106132507, + "num_tokens": 398694652.0, + "step": 16901 + }, + { + "epoch": 1.3937494846210934, + "grad_norm": 1.8849241733551025, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7616548538208008, + "num_tokens": 398726317.0, + "step": 16902 + }, + { + "epoch": 1.393831945246145, + "grad_norm": 1.96514093875885, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7329633235931396, + "num_tokens": 398762361.0, + "step": 16903 + }, + { + "epoch": 1.3939144058711965, + "grad_norm": 1.9265419244766235, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.7883491516113281, + "num_tokens": 398791313.0, + "step": 16904 + }, + { + "epoch": 1.393996866496248, + "grad_norm": 1.8879386186599731, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7512032985687256, + "num_tokens": 398825571.0, + "step": 16905 + }, + { + "epoch": 1.3940793271212994, + "grad_norm": 2.1448452472686768, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.776731014251709, + "num_tokens": 398851314.0, + "step": 16906 + }, + { + "epoch": 1.3941617877463512, + "grad_norm": 2.1842405796051025, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7046586871147156, + "num_tokens": 398881272.0, + "step": 16907 + }, + { + "epoch": 1.3942442483714026, + "grad_norm": 2.1187596321105957, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7548494338989258, + "num_tokens": 398908921.0, + "step": 16908 + }, + { + "epoch": 1.3943267089964542, + "grad_norm": 2.0105581283569336, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7750512361526489, + "num_tokens": 398938826.0, + "step": 16909 + }, + { + "epoch": 1.3944091696215057, + "grad_norm": 1.9038140773773193, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.776328980922699, + "num_tokens": 398970471.0, + "step": 16910 + }, + { + "epoch": 1.3944916302465573, + "grad_norm": 1.925254464149475, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7825658917427063, + "num_tokens": 399005501.0, + "step": 16911 + }, + { + "epoch": 1.3945740908716089, + "grad_norm": 2.358414649963379, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.7515578866004944, + "num_tokens": 399032302.0, + "step": 16912 + }, + { + "epoch": 1.3946565514966602, + "grad_norm": 1.8812587261199951, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7429500222206116, + "num_tokens": 399067732.0, + "step": 16913 + }, + { + "epoch": 1.3947390121217118, + "grad_norm": 1.9694143533706665, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7689357995986938, + "num_tokens": 399095247.0, + "step": 16914 + }, + { + "epoch": 1.3948214727467634, + "grad_norm": 1.969395637512207, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7560673952102661, + "num_tokens": 399125797.0, + "step": 16915 + }, + { + "epoch": 1.394903933371815, + "grad_norm": 2.02773380279541, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.7834838032722473, + "num_tokens": 399159007.0, + "step": 16916 + }, + { + "epoch": 1.3949863939968665, + "grad_norm": 1.741701364517212, + "learning_rate": 1e-06, + "loss": 0.6074, + "mean_token_accuracy": 0.8026627898216248, + "num_tokens": 399190580.0, + "step": 16917 + }, + { + "epoch": 1.395068854621918, + "grad_norm": 2.087522506713867, + "learning_rate": 1e-06, + "loss": 0.7698, + "mean_token_accuracy": 0.7528811693191528, + "num_tokens": 399217093.0, + "step": 16918 + }, + { + "epoch": 1.3951513152469697, + "grad_norm": 1.8417595624923706, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7920567989349365, + "num_tokens": 399248817.0, + "step": 16919 + }, + { + "epoch": 1.395233775872021, + "grad_norm": 1.634894609451294, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.7890917658805847, + "num_tokens": 399291155.0, + "step": 16920 + }, + { + "epoch": 1.3953162364970726, + "grad_norm": 1.9388632774353027, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.759406566619873, + "num_tokens": 399322210.0, + "step": 16921 + }, + { + "epoch": 1.3953986971221242, + "grad_norm": 2.009579658508301, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7390279769897461, + "num_tokens": 399353986.0, + "step": 16922 + }, + { + "epoch": 1.3954811577471757, + "grad_norm": 2.2751152515411377, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7172595858573914, + "num_tokens": 399382256.0, + "step": 16923 + }, + { + "epoch": 1.3955636183722273, + "grad_norm": 1.7965795993804932, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7893832325935364, + "num_tokens": 399416322.0, + "step": 16924 + }, + { + "epoch": 1.3956460789972787, + "grad_norm": 2.1546361446380615, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7748545408248901, + "num_tokens": 399439470.0, + "step": 16925 + }, + { + "epoch": 1.3957285396223305, + "grad_norm": 2.3517274856567383, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7631021738052368, + "num_tokens": 399462093.0, + "step": 16926 + }, + { + "epoch": 1.3958110002473818, + "grad_norm": 1.939390778541565, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7290915846824646, + "num_tokens": 399496284.0, + "step": 16927 + }, + { + "epoch": 1.3958934608724334, + "grad_norm": 2.2641043663024902, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7582067251205444, + "num_tokens": 399519147.0, + "step": 16928 + }, + { + "epoch": 1.395975921497485, + "grad_norm": 2.067152500152588, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7777723073959351, + "num_tokens": 399548021.0, + "step": 16929 + }, + { + "epoch": 1.3960583821225365, + "grad_norm": 1.857068419456482, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.771977424621582, + "num_tokens": 399578574.0, + "step": 16930 + }, + { + "epoch": 1.396140842747588, + "grad_norm": 2.003753662109375, + "learning_rate": 1e-06, + "loss": 0.5991, + "mean_token_accuracy": 0.8005807995796204, + "num_tokens": 399608748.0, + "step": 16931 + }, + { + "epoch": 1.3962233033726394, + "grad_norm": 1.8826956748962402, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7814527750015259, + "num_tokens": 399641370.0, + "step": 16932 + }, + { + "epoch": 1.3963057639976912, + "grad_norm": 2.0771024227142334, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7695409059524536, + "num_tokens": 399669055.0, + "step": 16933 + }, + { + "epoch": 1.3963882246227426, + "grad_norm": 2.2620246410369873, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7271031141281128, + "num_tokens": 399693421.0, + "step": 16934 + }, + { + "epoch": 1.3964706852477942, + "grad_norm": 2.120692491531372, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7330403327941895, + "num_tokens": 399724058.0, + "step": 16935 + }, + { + "epoch": 1.3965531458728457, + "grad_norm": 2.096113920211792, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7016779184341431, + "num_tokens": 399755525.0, + "step": 16936 + }, + { + "epoch": 1.3966356064978973, + "grad_norm": 2.404402732849121, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7314195036888123, + "num_tokens": 399780183.0, + "step": 16937 + }, + { + "epoch": 1.3967180671229489, + "grad_norm": 2.2983922958374023, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.763446569442749, + "num_tokens": 399805606.0, + "step": 16938 + }, + { + "epoch": 1.3968005277480002, + "grad_norm": 2.1520299911499023, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7382210493087769, + "num_tokens": 399839875.0, + "step": 16939 + }, + { + "epoch": 1.3968829883730518, + "grad_norm": 1.7368077039718628, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7746486067771912, + "num_tokens": 399875378.0, + "step": 16940 + }, + { + "epoch": 1.3969654489981034, + "grad_norm": 2.213865280151367, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7713032364845276, + "num_tokens": 399899762.0, + "step": 16941 + }, + { + "epoch": 1.397047909623155, + "grad_norm": 1.830075740814209, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7497050762176514, + "num_tokens": 399934295.0, + "step": 16942 + }, + { + "epoch": 1.3971303702482065, + "grad_norm": 2.023333787918091, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7444883584976196, + "num_tokens": 399964739.0, + "step": 16943 + }, + { + "epoch": 1.397212830873258, + "grad_norm": 2.1177873611450195, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.768330454826355, + "num_tokens": 399993321.0, + "step": 16944 + }, + { + "epoch": 1.3972952914983097, + "grad_norm": 1.9738249778747559, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7597002983093262, + "num_tokens": 400024239.0, + "step": 16945 + }, + { + "epoch": 1.397377752123361, + "grad_norm": 1.9502980709075928, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7653380036354065, + "num_tokens": 400058930.0, + "step": 16946 + }, + { + "epoch": 1.3974602127484126, + "grad_norm": 1.9325507879257202, + "learning_rate": 1e-06, + "loss": 0.6031, + "mean_token_accuracy": 0.7943935990333557, + "num_tokens": 400088657.0, + "step": 16947 + }, + { + "epoch": 1.3975426733734642, + "grad_norm": 1.76450777053833, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7632350325584412, + "num_tokens": 400128313.0, + "step": 16948 + }, + { + "epoch": 1.3976251339985157, + "grad_norm": 2.0215003490448, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.7909715175628662, + "num_tokens": 400158716.0, + "step": 16949 + }, + { + "epoch": 1.3977075946235673, + "grad_norm": 1.9982455968856812, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7286925315856934, + "num_tokens": 400190297.0, + "step": 16950 + }, + { + "epoch": 1.3977900552486187, + "grad_norm": 1.9065476655960083, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7707674503326416, + "num_tokens": 400222510.0, + "step": 16951 + }, + { + "epoch": 1.3978725158736705, + "grad_norm": 1.9362974166870117, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7699128985404968, + "num_tokens": 400253564.0, + "step": 16952 + }, + { + "epoch": 1.3979549764987218, + "grad_norm": 2.0676512718200684, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7586742043495178, + "num_tokens": 400280847.0, + "step": 16953 + }, + { + "epoch": 1.3980374371237734, + "grad_norm": 2.068267822265625, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7779289484024048, + "num_tokens": 400307972.0, + "step": 16954 + }, + { + "epoch": 1.398119897748825, + "grad_norm": 2.1068103313446045, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.7569445371627808, + "num_tokens": 400337169.0, + "step": 16955 + }, + { + "epoch": 1.3982023583738765, + "grad_norm": 2.1922576427459717, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7537107467651367, + "num_tokens": 400365017.0, + "step": 16956 + }, + { + "epoch": 1.398284818998928, + "grad_norm": 2.0939998626708984, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7730469703674316, + "num_tokens": 400392381.0, + "step": 16957 + }, + { + "epoch": 1.3983672796239794, + "grad_norm": 2.05491304397583, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7745250463485718, + "num_tokens": 400420413.0, + "step": 16958 + }, + { + "epoch": 1.398449740249031, + "grad_norm": 2.1213691234588623, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7666139602661133, + "num_tokens": 400449079.0, + "step": 16959 + }, + { + "epoch": 1.3985322008740826, + "grad_norm": 1.7385177612304688, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.782372236251831, + "num_tokens": 400485827.0, + "step": 16960 + }, + { + "epoch": 1.3986146614991342, + "grad_norm": 1.9654548168182373, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7306458353996277, + "num_tokens": 400517099.0, + "step": 16961 + }, + { + "epoch": 1.3986971221241857, + "grad_norm": 1.894713044166565, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.7885154485702515, + "num_tokens": 400548030.0, + "step": 16962 + }, + { + "epoch": 1.3987795827492373, + "grad_norm": 2.2331206798553467, + "learning_rate": 1e-06, + "loss": 0.5459, + "mean_token_accuracy": 0.8030308485031128, + "num_tokens": 400570053.0, + "step": 16963 + }, + { + "epoch": 1.3988620433742889, + "grad_norm": 2.036388874053955, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7042170166969299, + "num_tokens": 400600834.0, + "step": 16964 + }, + { + "epoch": 1.3989445039993402, + "grad_norm": 2.0679779052734375, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7708734273910522, + "num_tokens": 400626784.0, + "step": 16965 + }, + { + "epoch": 1.3990269646243918, + "grad_norm": 2.2878222465515137, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7361618280410767, + "num_tokens": 400651998.0, + "step": 16966 + }, + { + "epoch": 1.3991094252494434, + "grad_norm": 1.9805008172988892, + "learning_rate": 1e-06, + "loss": 0.6098, + "mean_token_accuracy": 0.8019510507583618, + "num_tokens": 400683472.0, + "step": 16967 + }, + { + "epoch": 1.399191885874495, + "grad_norm": 2.274566888809204, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7427172660827637, + "num_tokens": 400706631.0, + "step": 16968 + }, + { + "epoch": 1.3992743464995465, + "grad_norm": 2.333749294281006, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7593826055526733, + "num_tokens": 400728870.0, + "step": 16969 + }, + { + "epoch": 1.3993568071245979, + "grad_norm": 1.7652090787887573, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.7936859130859375, + "num_tokens": 400762419.0, + "step": 16970 + }, + { + "epoch": 1.3994392677496497, + "grad_norm": 1.7670000791549683, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7745552062988281, + "num_tokens": 400800380.0, + "step": 16971 + }, + { + "epoch": 1.399521728374701, + "grad_norm": 2.115269660949707, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7394888997077942, + "num_tokens": 400826853.0, + "step": 16972 + }, + { + "epoch": 1.3996041889997526, + "grad_norm": 1.8254107236862183, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7694035768508911, + "num_tokens": 400864319.0, + "step": 16973 + }, + { + "epoch": 1.3996866496248042, + "grad_norm": 2.250293016433716, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7629433870315552, + "num_tokens": 400889671.0, + "step": 16974 + }, + { + "epoch": 1.3997691102498557, + "grad_norm": 2.2774031162261963, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7568049430847168, + "num_tokens": 400913693.0, + "step": 16975 + }, + { + "epoch": 1.3998515708749073, + "grad_norm": 2.031674385070801, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7631955742835999, + "num_tokens": 400943631.0, + "step": 16976 + }, + { + "epoch": 1.3999340314999587, + "grad_norm": 2.234583854675293, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7172954678535461, + "num_tokens": 400970381.0, + "step": 16977 + }, + { + "epoch": 1.4000164921250102, + "grad_norm": 1.9822427034378052, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.7942245006561279, + "num_tokens": 400997729.0, + "step": 16978 + }, + { + "epoch": 1.4000989527500618, + "grad_norm": 2.1146063804626465, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.756550669670105, + "num_tokens": 401027400.0, + "step": 16979 + }, + { + "epoch": 1.4001814133751134, + "grad_norm": 1.9822300672531128, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7442768812179565, + "num_tokens": 401056757.0, + "step": 16980 + }, + { + "epoch": 1.400263874000165, + "grad_norm": 1.9276134967803955, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7505990266799927, + "num_tokens": 401088558.0, + "step": 16981 + }, + { + "epoch": 1.4003463346252165, + "grad_norm": 2.2272121906280518, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7601990103721619, + "num_tokens": 401115977.0, + "step": 16982 + }, + { + "epoch": 1.400428795250268, + "grad_norm": 1.8208855390548706, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7523348331451416, + "num_tokens": 401150571.0, + "step": 16983 + }, + { + "epoch": 1.4005112558753194, + "grad_norm": 2.246624708175659, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7469347715377808, + "num_tokens": 401174803.0, + "step": 16984 + }, + { + "epoch": 1.400593716500371, + "grad_norm": 2.1511850357055664, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7578834295272827, + "num_tokens": 401203056.0, + "step": 16985 + }, + { + "epoch": 1.4006761771254226, + "grad_norm": 2.217799186706543, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7499160766601562, + "num_tokens": 401228062.0, + "step": 16986 + }, + { + "epoch": 1.4007586377504742, + "grad_norm": 2.339388132095337, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7236957550048828, + "num_tokens": 401251593.0, + "step": 16987 + }, + { + "epoch": 1.4008410983755257, + "grad_norm": 2.0522336959838867, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.8025741577148438, + "num_tokens": 401276919.0, + "step": 16988 + }, + { + "epoch": 1.4009235590005773, + "grad_norm": 2.424260377883911, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7387685775756836, + "num_tokens": 401298270.0, + "step": 16989 + }, + { + "epoch": 1.4010060196256289, + "grad_norm": 2.216730833053589, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7788649797439575, + "num_tokens": 401326353.0, + "step": 16990 + }, + { + "epoch": 1.4010884802506802, + "grad_norm": 2.0015981197357178, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.755172848701477, + "num_tokens": 401358417.0, + "step": 16991 + }, + { + "epoch": 1.4011709408757318, + "grad_norm": 2.067030191421509, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7731149196624756, + "num_tokens": 401386733.0, + "step": 16992 + }, + { + "epoch": 1.4012534015007834, + "grad_norm": 1.9075281620025635, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.7816692590713501, + "num_tokens": 401418484.0, + "step": 16993 + }, + { + "epoch": 1.401335862125835, + "grad_norm": 2.1266398429870605, + "learning_rate": 1e-06, + "loss": 0.7738, + "mean_token_accuracy": 0.76160728931427, + "num_tokens": 401445224.0, + "step": 16994 + }, + { + "epoch": 1.4014183227508865, + "grad_norm": 1.8192944526672363, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7693292498588562, + "num_tokens": 401477296.0, + "step": 16995 + }, + { + "epoch": 1.4015007833759379, + "grad_norm": 2.1490187644958496, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7472883462905884, + "num_tokens": 401503726.0, + "step": 16996 + }, + { + "epoch": 1.4015832440009897, + "grad_norm": 1.8699707984924316, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7265126705169678, + "num_tokens": 401541462.0, + "step": 16997 + }, + { + "epoch": 1.401665704626041, + "grad_norm": 1.7369016408920288, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7803676128387451, + "num_tokens": 401576468.0, + "step": 16998 + }, + { + "epoch": 1.4017481652510926, + "grad_norm": 2.1688244342803955, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7643547058105469, + "num_tokens": 401604318.0, + "step": 16999 + }, + { + "epoch": 1.4018306258761442, + "grad_norm": 2.100203514099121, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7350854873657227, + "num_tokens": 401633690.0, + "step": 17000 + }, + { + "epoch": 1.4019130865011957, + "grad_norm": 2.212411403656006, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.7831295132637024, + "num_tokens": 401658707.0, + "step": 17001 + }, + { + "epoch": 1.4019955471262473, + "grad_norm": 1.9855183362960815, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7425148487091064, + "num_tokens": 401687751.0, + "step": 17002 + }, + { + "epoch": 1.4020780077512986, + "grad_norm": 1.9513466358184814, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.7916911840438843, + "num_tokens": 401712589.0, + "step": 17003 + }, + { + "epoch": 1.4021604683763502, + "grad_norm": 2.1977477073669434, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7257852554321289, + "num_tokens": 401741411.0, + "step": 17004 + }, + { + "epoch": 1.4022429290014018, + "grad_norm": 1.9477466344833374, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7473982572555542, + "num_tokens": 401771319.0, + "step": 17005 + }, + { + "epoch": 1.4023253896264534, + "grad_norm": 1.8760021924972534, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.745174765586853, + "num_tokens": 401804368.0, + "step": 17006 + }, + { + "epoch": 1.402407850251505, + "grad_norm": 1.887951135635376, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.79012131690979, + "num_tokens": 401835077.0, + "step": 17007 + }, + { + "epoch": 1.4024903108765565, + "grad_norm": 1.9113034009933472, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7766048908233643, + "num_tokens": 401867631.0, + "step": 17008 + }, + { + "epoch": 1.402572771501608, + "grad_norm": 1.9819600582122803, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.7441320419311523, + "num_tokens": 401897451.0, + "step": 17009 + }, + { + "epoch": 1.4026552321266594, + "grad_norm": 2.0545740127563477, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7787890434265137, + "num_tokens": 401925134.0, + "step": 17010 + }, + { + "epoch": 1.402737692751711, + "grad_norm": 2.114060878753662, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7253627777099609, + "num_tokens": 401952929.0, + "step": 17011 + }, + { + "epoch": 1.4028201533767626, + "grad_norm": 1.7158913612365723, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7659524083137512, + "num_tokens": 401986904.0, + "step": 17012 + }, + { + "epoch": 1.4029026140018142, + "grad_norm": 2.265977382659912, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7105516791343689, + "num_tokens": 402012613.0, + "step": 17013 + }, + { + "epoch": 1.4029850746268657, + "grad_norm": 1.8931338787078857, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.744728684425354, + "num_tokens": 402044306.0, + "step": 17014 + }, + { + "epoch": 1.403067535251917, + "grad_norm": 1.9543068408966064, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7752352356910706, + "num_tokens": 402073179.0, + "step": 17015 + }, + { + "epoch": 1.4031499958769689, + "grad_norm": 2.205082654953003, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7569159865379333, + "num_tokens": 402098944.0, + "step": 17016 + }, + { + "epoch": 1.4032324565020202, + "grad_norm": 1.9068795442581177, + "learning_rate": 1e-06, + "loss": 0.5552, + "mean_token_accuracy": 0.8103771209716797, + "num_tokens": 402130316.0, + "step": 17017 + }, + { + "epoch": 1.4033149171270718, + "grad_norm": 1.8518308401107788, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.7949315309524536, + "num_tokens": 402163149.0, + "step": 17018 + }, + { + "epoch": 1.4033973777521234, + "grad_norm": 1.8409355878829956, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7451967000961304, + "num_tokens": 402196760.0, + "step": 17019 + }, + { + "epoch": 1.403479838377175, + "grad_norm": 2.072110652923584, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7699318528175354, + "num_tokens": 402223980.0, + "step": 17020 + }, + { + "epoch": 1.4035622990022265, + "grad_norm": 2.1948366165161133, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.7915453910827637, + "num_tokens": 402245327.0, + "step": 17021 + }, + { + "epoch": 1.4036447596272779, + "grad_norm": 1.8663395643234253, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7574844360351562, + "num_tokens": 402279222.0, + "step": 17022 + }, + { + "epoch": 1.4037272202523294, + "grad_norm": 2.0100834369659424, + "learning_rate": 1e-06, + "loss": 0.7774, + "mean_token_accuracy": 0.761796772480011, + "num_tokens": 402311456.0, + "step": 17023 + }, + { + "epoch": 1.403809680877381, + "grad_norm": 2.163506507873535, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7268129587173462, + "num_tokens": 402339803.0, + "step": 17024 + }, + { + "epoch": 1.4038921415024326, + "grad_norm": 2.1383328437805176, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.7919160723686218, + "num_tokens": 402367552.0, + "step": 17025 + }, + { + "epoch": 1.4039746021274842, + "grad_norm": 1.849493145942688, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.793377697467804, + "num_tokens": 402397374.0, + "step": 17026 + }, + { + "epoch": 1.4040570627525357, + "grad_norm": 1.9834449291229248, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.7896517515182495, + "num_tokens": 402424896.0, + "step": 17027 + }, + { + "epoch": 1.4041395233775873, + "grad_norm": 2.328951597213745, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7862961292266846, + "num_tokens": 402449037.0, + "step": 17028 + }, + { + "epoch": 1.4042219840026386, + "grad_norm": 1.8308161497116089, + "learning_rate": 1e-06, + "loss": 0.5774, + "mean_token_accuracy": 0.8120678663253784, + "num_tokens": 402480077.0, + "step": 17029 + }, + { + "epoch": 1.4043044446276902, + "grad_norm": 2.1216516494750977, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7213321328163147, + "num_tokens": 402511500.0, + "step": 17030 + }, + { + "epoch": 1.4043869052527418, + "grad_norm": 2.0972352027893066, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7513095140457153, + "num_tokens": 402537007.0, + "step": 17031 + }, + { + "epoch": 1.4044693658777934, + "grad_norm": 2.1163437366485596, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.71771639585495, + "num_tokens": 402565830.0, + "step": 17032 + }, + { + "epoch": 1.404551826502845, + "grad_norm": 1.781860589981079, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.7957497835159302, + "num_tokens": 402599132.0, + "step": 17033 + }, + { + "epoch": 1.4046342871278965, + "grad_norm": 1.897487759590149, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7655394673347473, + "num_tokens": 402632555.0, + "step": 17034 + }, + { + "epoch": 1.404716747752948, + "grad_norm": 1.7714776992797852, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.7951191663742065, + "num_tokens": 402668242.0, + "step": 17035 + }, + { + "epoch": 1.4047992083779994, + "grad_norm": 1.9582240581512451, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7703824639320374, + "num_tokens": 402698668.0, + "step": 17036 + }, + { + "epoch": 1.404881669003051, + "grad_norm": 2.042226791381836, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7688693404197693, + "num_tokens": 402728040.0, + "step": 17037 + }, + { + "epoch": 1.4049641296281026, + "grad_norm": 2.087545394897461, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7573301792144775, + "num_tokens": 402756997.0, + "step": 17038 + }, + { + "epoch": 1.4050465902531541, + "grad_norm": 1.917944312095642, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.753600537776947, + "num_tokens": 402790054.0, + "step": 17039 + }, + { + "epoch": 1.4051290508782057, + "grad_norm": 1.9982775449752808, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7626731395721436, + "num_tokens": 402821552.0, + "step": 17040 + }, + { + "epoch": 1.405211511503257, + "grad_norm": 2.138251781463623, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.778857409954071, + "num_tokens": 402845775.0, + "step": 17041 + }, + { + "epoch": 1.4052939721283089, + "grad_norm": 1.8821167945861816, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7508391737937927, + "num_tokens": 402878375.0, + "step": 17042 + }, + { + "epoch": 1.4053764327533602, + "grad_norm": 2.092325210571289, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7433915138244629, + "num_tokens": 402906272.0, + "step": 17043 + }, + { + "epoch": 1.4054588933784118, + "grad_norm": 2.2121658325195312, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7529993057250977, + "num_tokens": 402931132.0, + "step": 17044 + }, + { + "epoch": 1.4055413540034634, + "grad_norm": 1.8983426094055176, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.7721698880195618, + "num_tokens": 402958645.0, + "step": 17045 + }, + { + "epoch": 1.405623814628515, + "grad_norm": 2.187593936920166, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7506839036941528, + "num_tokens": 402984778.0, + "step": 17046 + }, + { + "epoch": 1.4057062752535665, + "grad_norm": 1.8019181489944458, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.723501980304718, + "num_tokens": 403022626.0, + "step": 17047 + }, + { + "epoch": 1.4057887358786179, + "grad_norm": 2.183391809463501, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.7785301804542542, + "num_tokens": 403047887.0, + "step": 17048 + }, + { + "epoch": 1.4058711965036694, + "grad_norm": 2.047630786895752, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.7895640134811401, + "num_tokens": 403071551.0, + "step": 17049 + }, + { + "epoch": 1.405953657128721, + "grad_norm": 2.112959623336792, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7400298714637756, + "num_tokens": 403099563.0, + "step": 17050 + }, + { + "epoch": 1.4060361177537726, + "grad_norm": 2.4168920516967773, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7574619054794312, + "num_tokens": 403120809.0, + "step": 17051 + }, + { + "epoch": 1.4061185783788241, + "grad_norm": 1.7741705179214478, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8010662794113159, + "num_tokens": 403152245.0, + "step": 17052 + }, + { + "epoch": 1.4062010390038757, + "grad_norm": 1.8244656324386597, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.740975022315979, + "num_tokens": 403186317.0, + "step": 17053 + }, + { + "epoch": 1.4062834996289273, + "grad_norm": 1.9348723888397217, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7509596943855286, + "num_tokens": 403216765.0, + "step": 17054 + }, + { + "epoch": 1.4063659602539786, + "grad_norm": 2.0343613624572754, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.7820865511894226, + "num_tokens": 403243853.0, + "step": 17055 + }, + { + "epoch": 1.4064484208790302, + "grad_norm": 2.1772122383117676, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.763862133026123, + "num_tokens": 403268823.0, + "step": 17056 + }, + { + "epoch": 1.4065308815040818, + "grad_norm": 1.9749977588653564, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7642881870269775, + "num_tokens": 403298129.0, + "step": 17057 + }, + { + "epoch": 1.4066133421291334, + "grad_norm": 2.0092499256134033, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.774691104888916, + "num_tokens": 403327558.0, + "step": 17058 + }, + { + "epoch": 1.406695802754185, + "grad_norm": 1.9643816947937012, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7369653582572937, + "num_tokens": 403358682.0, + "step": 17059 + }, + { + "epoch": 1.4067782633792363, + "grad_norm": 2.0692028999328613, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7717952132225037, + "num_tokens": 403386290.0, + "step": 17060 + }, + { + "epoch": 1.406860724004288, + "grad_norm": 1.8100897073745728, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.770638108253479, + "num_tokens": 403422376.0, + "step": 17061 + }, + { + "epoch": 1.4069431846293394, + "grad_norm": 1.7677346467971802, + "learning_rate": 1e-06, + "loss": 0.5714, + "mean_token_accuracy": 0.8125754594802856, + "num_tokens": 403456428.0, + "step": 17062 + }, + { + "epoch": 1.407025645254391, + "grad_norm": 1.9005182981491089, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7555245161056519, + "num_tokens": 403488617.0, + "step": 17063 + }, + { + "epoch": 1.4071081058794426, + "grad_norm": 1.4773722887039185, + "learning_rate": 1e-06, + "loss": 0.5421, + "mean_token_accuracy": 0.8254650831222534, + "num_tokens": 403529192.0, + "step": 17064 + }, + { + "epoch": 1.4071905665044941, + "grad_norm": 2.1270663738250732, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7814618349075317, + "num_tokens": 403552801.0, + "step": 17065 + }, + { + "epoch": 1.4072730271295457, + "grad_norm": 2.228461980819702, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7274107336997986, + "num_tokens": 403577171.0, + "step": 17066 + }, + { + "epoch": 1.407355487754597, + "grad_norm": 1.978649616241455, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7451448440551758, + "num_tokens": 403607910.0, + "step": 17067 + }, + { + "epoch": 1.4074379483796486, + "grad_norm": 1.944780945777893, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7704693078994751, + "num_tokens": 403640771.0, + "step": 17068 + }, + { + "epoch": 1.4075204090047002, + "grad_norm": 2.054189682006836, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7563644647598267, + "num_tokens": 403671144.0, + "step": 17069 + }, + { + "epoch": 1.4076028696297518, + "grad_norm": 1.9221335649490356, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.7840996980667114, + "num_tokens": 403702147.0, + "step": 17070 + }, + { + "epoch": 1.4076853302548034, + "grad_norm": 2.016392469406128, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7400060892105103, + "num_tokens": 403734047.0, + "step": 17071 + }, + { + "epoch": 1.407767790879855, + "grad_norm": 1.9218511581420898, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7486347556114197, + "num_tokens": 403763609.0, + "step": 17072 + }, + { + "epoch": 1.4078502515049065, + "grad_norm": 1.84565269947052, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7778580188751221, + "num_tokens": 403796621.0, + "step": 17073 + }, + { + "epoch": 1.4079327121299579, + "grad_norm": 1.7617762088775635, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7562983632087708, + "num_tokens": 403832155.0, + "step": 17074 + }, + { + "epoch": 1.4080151727550094, + "grad_norm": 1.7381230592727661, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7470124363899231, + "num_tokens": 403871170.0, + "step": 17075 + }, + { + "epoch": 1.408097633380061, + "grad_norm": 1.7970514297485352, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.77774977684021, + "num_tokens": 403903286.0, + "step": 17076 + }, + { + "epoch": 1.4081800940051126, + "grad_norm": 1.8444486856460571, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7680726051330566, + "num_tokens": 403935564.0, + "step": 17077 + }, + { + "epoch": 1.4082625546301641, + "grad_norm": 2.0029056072235107, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7569501399993896, + "num_tokens": 403966935.0, + "step": 17078 + }, + { + "epoch": 1.4083450152552157, + "grad_norm": 1.6920863389968872, + "learning_rate": 1e-06, + "loss": 0.58, + "mean_token_accuracy": 0.8079743385314941, + "num_tokens": 404006205.0, + "step": 17079 + }, + { + "epoch": 1.4084274758802673, + "grad_norm": 1.8780066967010498, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.7831233739852905, + "num_tokens": 404034716.0, + "step": 17080 + }, + { + "epoch": 1.4085099365053186, + "grad_norm": 2.114938497543335, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7608544826507568, + "num_tokens": 404063509.0, + "step": 17081 + }, + { + "epoch": 1.4085923971303702, + "grad_norm": 2.1494696140289307, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7744120359420776, + "num_tokens": 404088210.0, + "step": 17082 + }, + { + "epoch": 1.4086748577554218, + "grad_norm": 1.763299584388733, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7884666323661804, + "num_tokens": 404123287.0, + "step": 17083 + }, + { + "epoch": 1.4087573183804734, + "grad_norm": 2.1149046421051025, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7342885732650757, + "num_tokens": 404151550.0, + "step": 17084 + }, + { + "epoch": 1.408839779005525, + "grad_norm": 2.113603353500366, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7622891664505005, + "num_tokens": 404178391.0, + "step": 17085 + }, + { + "epoch": 1.4089222396305763, + "grad_norm": 1.9129893779754639, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7599267363548279, + "num_tokens": 404209283.0, + "step": 17086 + }, + { + "epoch": 1.409004700255628, + "grad_norm": 2.371724843978882, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7559351921081543, + "num_tokens": 404232482.0, + "step": 17087 + }, + { + "epoch": 1.4090871608806794, + "grad_norm": 2.1919052600860596, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7837331295013428, + "num_tokens": 404256045.0, + "step": 17088 + }, + { + "epoch": 1.409169621505731, + "grad_norm": 2.218319892883301, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7534350752830505, + "num_tokens": 404282167.0, + "step": 17089 + }, + { + "epoch": 1.4092520821307826, + "grad_norm": 1.9436259269714355, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7680405378341675, + "num_tokens": 404314608.0, + "step": 17090 + }, + { + "epoch": 1.4093345427558341, + "grad_norm": 2.1558847427368164, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7537426352500916, + "num_tokens": 404343161.0, + "step": 17091 + }, + { + "epoch": 1.4094170033808857, + "grad_norm": 2.094313383102417, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7525471448898315, + "num_tokens": 404371459.0, + "step": 17092 + }, + { + "epoch": 1.409499464005937, + "grad_norm": 1.847328543663025, + "learning_rate": 1e-06, + "loss": 0.6151, + "mean_token_accuracy": 0.7959127426147461, + "num_tokens": 404406008.0, + "step": 17093 + }, + { + "epoch": 1.4095819246309886, + "grad_norm": 2.076016664505005, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7682288885116577, + "num_tokens": 404433303.0, + "step": 17094 + }, + { + "epoch": 1.4096643852560402, + "grad_norm": 1.9250991344451904, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.7666143178939819, + "num_tokens": 404464779.0, + "step": 17095 + }, + { + "epoch": 1.4097468458810918, + "grad_norm": 1.938928246498108, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7507643103599548, + "num_tokens": 404495641.0, + "step": 17096 + }, + { + "epoch": 1.4098293065061434, + "grad_norm": 2.286768913269043, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7700032591819763, + "num_tokens": 404527511.0, + "step": 17097 + }, + { + "epoch": 1.409911767131195, + "grad_norm": 1.9376411437988281, + "learning_rate": 1e-06, + "loss": 0.5172, + "mean_token_accuracy": 0.8274524807929993, + "num_tokens": 404556297.0, + "step": 17098 + }, + { + "epoch": 1.4099942277562465, + "grad_norm": 1.8899714946746826, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7591609954833984, + "num_tokens": 404590130.0, + "step": 17099 + }, + { + "epoch": 1.4100766883812978, + "grad_norm": 2.0652711391448975, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.740739107131958, + "num_tokens": 404623990.0, + "step": 17100 + }, + { + "epoch": 1.4101591490063494, + "grad_norm": 1.9244126081466675, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7638290524482727, + "num_tokens": 404655818.0, + "step": 17101 + }, + { + "epoch": 1.410241609631401, + "grad_norm": 1.8817349672317505, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7573943138122559, + "num_tokens": 404688408.0, + "step": 17102 + }, + { + "epoch": 1.4103240702564526, + "grad_norm": 1.876326560974121, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.776757538318634, + "num_tokens": 404722418.0, + "step": 17103 + }, + { + "epoch": 1.4104065308815041, + "grad_norm": 1.8210631608963013, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.7944372892379761, + "num_tokens": 404759431.0, + "step": 17104 + }, + { + "epoch": 1.4104889915065555, + "grad_norm": 1.9489929676055908, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.8002681732177734, + "num_tokens": 404790342.0, + "step": 17105 + }, + { + "epoch": 1.4105714521316073, + "grad_norm": 2.099759817123413, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7443718314170837, + "num_tokens": 404818680.0, + "step": 17106 + }, + { + "epoch": 1.4106539127566586, + "grad_norm": 2.1650278568267822, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7152916193008423, + "num_tokens": 404847278.0, + "step": 17107 + }, + { + "epoch": 1.4107363733817102, + "grad_norm": 1.8066737651824951, + "learning_rate": 1e-06, + "loss": 0.6001, + "mean_token_accuracy": 0.7926453948020935, + "num_tokens": 404880331.0, + "step": 17108 + }, + { + "epoch": 1.4108188340067618, + "grad_norm": 1.8209044933319092, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.79384446144104, + "num_tokens": 404913319.0, + "step": 17109 + }, + { + "epoch": 1.4109012946318134, + "grad_norm": 2.222287654876709, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7785398364067078, + "num_tokens": 404938099.0, + "step": 17110 + }, + { + "epoch": 1.410983755256865, + "grad_norm": 1.9001976251602173, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7457503080368042, + "num_tokens": 404971139.0, + "step": 17111 + }, + { + "epoch": 1.4110662158819163, + "grad_norm": 2.101989507675171, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7344749569892883, + "num_tokens": 404999027.0, + "step": 17112 + }, + { + "epoch": 1.4111486765069678, + "grad_norm": 1.9840106964111328, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.777569055557251, + "num_tokens": 405026782.0, + "step": 17113 + }, + { + "epoch": 1.4112311371320194, + "grad_norm": 1.960314393043518, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.7615324854850769, + "num_tokens": 405056549.0, + "step": 17114 + }, + { + "epoch": 1.411313597757071, + "grad_norm": 1.8049062490463257, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7720708847045898, + "num_tokens": 405091716.0, + "step": 17115 + }, + { + "epoch": 1.4113960583821226, + "grad_norm": 2.38261079788208, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7730484008789062, + "num_tokens": 405113832.0, + "step": 17116 + }, + { + "epoch": 1.4114785190071741, + "grad_norm": 1.6980708837509155, + "learning_rate": 1e-06, + "loss": 0.5644, + "mean_token_accuracy": 0.8196943998336792, + "num_tokens": 405150745.0, + "step": 17117 + }, + { + "epoch": 1.4115609796322257, + "grad_norm": 1.930827021598816, + "learning_rate": 1e-06, + "loss": 0.5856, + "mean_token_accuracy": 0.8095636367797852, + "num_tokens": 405183160.0, + "step": 17118 + }, + { + "epoch": 1.411643440257277, + "grad_norm": 2.0148961544036865, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7766704559326172, + "num_tokens": 405215868.0, + "step": 17119 + }, + { + "epoch": 1.4117259008823286, + "grad_norm": 1.9570432901382446, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7537261843681335, + "num_tokens": 405249975.0, + "step": 17120 + }, + { + "epoch": 1.4118083615073802, + "grad_norm": 1.9091486930847168, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7588409781455994, + "num_tokens": 405282598.0, + "step": 17121 + }, + { + "epoch": 1.4118908221324318, + "grad_norm": 1.8894888162612915, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7597140669822693, + "num_tokens": 405315613.0, + "step": 17122 + }, + { + "epoch": 1.4119732827574834, + "grad_norm": 1.9958473443984985, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7464148998260498, + "num_tokens": 405346473.0, + "step": 17123 + }, + { + "epoch": 1.412055743382535, + "grad_norm": 2.3567254543304443, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7631021738052368, + "num_tokens": 405370285.0, + "step": 17124 + }, + { + "epoch": 1.4121382040075865, + "grad_norm": 1.8960967063903809, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7835609912872314, + "num_tokens": 405402738.0, + "step": 17125 + }, + { + "epoch": 1.4122206646326378, + "grad_norm": 1.7527590990066528, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.7583392858505249, + "num_tokens": 405440670.0, + "step": 17126 + }, + { + "epoch": 1.4123031252576894, + "grad_norm": 1.928085207939148, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.7487031817436218, + "num_tokens": 405472874.0, + "step": 17127 + }, + { + "epoch": 1.412385585882741, + "grad_norm": 2.235546827316284, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7334449291229248, + "num_tokens": 405501470.0, + "step": 17128 + }, + { + "epoch": 1.4124680465077926, + "grad_norm": 1.951077938079834, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7702362537384033, + "num_tokens": 405531443.0, + "step": 17129 + }, + { + "epoch": 1.4125505071328441, + "grad_norm": 1.9332199096679688, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7642011642456055, + "num_tokens": 405563230.0, + "step": 17130 + }, + { + "epoch": 1.4126329677578955, + "grad_norm": 2.0416100025177, + "learning_rate": 1e-06, + "loss": 0.5174, + "mean_token_accuracy": 0.8111569881439209, + "num_tokens": 405589264.0, + "step": 17131 + }, + { + "epoch": 1.4127154283829473, + "grad_norm": 2.0895304679870605, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7368142604827881, + "num_tokens": 405620003.0, + "step": 17132 + }, + { + "epoch": 1.4127978890079986, + "grad_norm": 1.987633228302002, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.765998363494873, + "num_tokens": 405647924.0, + "step": 17133 + }, + { + "epoch": 1.4128803496330502, + "grad_norm": 1.9259978532791138, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7736801505088806, + "num_tokens": 405677880.0, + "step": 17134 + }, + { + "epoch": 1.4129628102581018, + "grad_norm": 2.052241086959839, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.77555251121521, + "num_tokens": 405703533.0, + "step": 17135 + }, + { + "epoch": 1.4130452708831533, + "grad_norm": 1.8487967252731323, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7492364048957825, + "num_tokens": 405738402.0, + "step": 17136 + }, + { + "epoch": 1.413127731508205, + "grad_norm": 2.3294618129730225, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7509839534759521, + "num_tokens": 405760565.0, + "step": 17137 + }, + { + "epoch": 1.4132101921332563, + "grad_norm": 2.093423366546631, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.752750813961029, + "num_tokens": 405787855.0, + "step": 17138 + }, + { + "epoch": 1.4132926527583078, + "grad_norm": 1.8302462100982666, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7659778594970703, + "num_tokens": 405823578.0, + "step": 17139 + }, + { + "epoch": 1.4133751133833594, + "grad_norm": 1.9886912107467651, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.764061689376831, + "num_tokens": 405855329.0, + "step": 17140 + }, + { + "epoch": 1.413457574008411, + "grad_norm": 1.9340953826904297, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.7931584119796753, + "num_tokens": 405887756.0, + "step": 17141 + }, + { + "epoch": 1.4135400346334626, + "grad_norm": 1.727324366569519, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7547396421432495, + "num_tokens": 405925609.0, + "step": 17142 + }, + { + "epoch": 1.4136224952585141, + "grad_norm": 2.2572648525238037, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.779573917388916, + "num_tokens": 405948490.0, + "step": 17143 + }, + { + "epoch": 1.4137049558835657, + "grad_norm": 2.043015718460083, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7593252062797546, + "num_tokens": 405977540.0, + "step": 17144 + }, + { + "epoch": 1.413787416508617, + "grad_norm": 2.251884937286377, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.8009766340255737, + "num_tokens": 406001249.0, + "step": 17145 + }, + { + "epoch": 1.4138698771336686, + "grad_norm": 2.3161075115203857, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7123022675514221, + "num_tokens": 406027157.0, + "step": 17146 + }, + { + "epoch": 1.4139523377587202, + "grad_norm": 2.097391128540039, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.7578520178794861, + "num_tokens": 406053640.0, + "step": 17147 + }, + { + "epoch": 1.4140347983837718, + "grad_norm": 2.011820077896118, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7754154205322266, + "num_tokens": 406083753.0, + "step": 17148 + }, + { + "epoch": 1.4141172590088233, + "grad_norm": 1.9874308109283447, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.7928894758224487, + "num_tokens": 406113349.0, + "step": 17149 + }, + { + "epoch": 1.4141997196338747, + "grad_norm": 2.0547289848327637, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7594691514968872, + "num_tokens": 406140637.0, + "step": 17150 + }, + { + "epoch": 1.4142821802589265, + "grad_norm": 2.105604648590088, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7277398109436035, + "num_tokens": 406168715.0, + "step": 17151 + }, + { + "epoch": 1.4143646408839778, + "grad_norm": 2.035712480545044, + "learning_rate": 1e-06, + "loss": 0.5581, + "mean_token_accuracy": 0.8155759572982788, + "num_tokens": 406195087.0, + "step": 17152 + }, + { + "epoch": 1.4144471015090294, + "grad_norm": 2.0383248329162598, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.768896222114563, + "num_tokens": 406222242.0, + "step": 17153 + }, + { + "epoch": 1.414529562134081, + "grad_norm": 1.9517724514007568, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7872217297554016, + "num_tokens": 406251280.0, + "step": 17154 + }, + { + "epoch": 1.4146120227591326, + "grad_norm": 1.7547334432601929, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.7858515381813049, + "num_tokens": 406286206.0, + "step": 17155 + }, + { + "epoch": 1.4146944833841841, + "grad_norm": 2.0070996284484863, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7629307508468628, + "num_tokens": 406320105.0, + "step": 17156 + }, + { + "epoch": 1.4147769440092355, + "grad_norm": 2.1299867630004883, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.714147686958313, + "num_tokens": 406349452.0, + "step": 17157 + }, + { + "epoch": 1.414859404634287, + "grad_norm": 1.9116846323013306, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.784950315952301, + "num_tokens": 406379791.0, + "step": 17158 + }, + { + "epoch": 1.4149418652593386, + "grad_norm": 1.767024278640747, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.7979927062988281, + "num_tokens": 406416381.0, + "step": 17159 + }, + { + "epoch": 1.4150243258843902, + "grad_norm": 2.125974178314209, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7798253893852234, + "num_tokens": 406440403.0, + "step": 17160 + }, + { + "epoch": 1.4151067865094418, + "grad_norm": 2.141026735305786, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7481124401092529, + "num_tokens": 406468117.0, + "step": 17161 + }, + { + "epoch": 1.4151892471344933, + "grad_norm": 2.0977413654327393, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7355737686157227, + "num_tokens": 406498161.0, + "step": 17162 + }, + { + "epoch": 1.415271707759545, + "grad_norm": 2.4018492698669434, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7558104991912842, + "num_tokens": 406520617.0, + "step": 17163 + }, + { + "epoch": 1.4153541683845963, + "grad_norm": 1.9441454410552979, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7654305100440979, + "num_tokens": 406552256.0, + "step": 17164 + }, + { + "epoch": 1.4154366290096478, + "grad_norm": 1.861809253692627, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7780553102493286, + "num_tokens": 406584917.0, + "step": 17165 + }, + { + "epoch": 1.4155190896346994, + "grad_norm": 1.9666496515274048, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7542258501052856, + "num_tokens": 406616319.0, + "step": 17166 + }, + { + "epoch": 1.415601550259751, + "grad_norm": 2.0410454273223877, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.7606689929962158, + "num_tokens": 406644206.0, + "step": 17167 + }, + { + "epoch": 1.4156840108848026, + "grad_norm": 1.774551272392273, + "learning_rate": 1e-06, + "loss": 0.5421, + "mean_token_accuracy": 0.8168755769729614, + "num_tokens": 406678546.0, + "step": 17168 + }, + { + "epoch": 1.415766471509854, + "grad_norm": 1.9954595565795898, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7613159418106079, + "num_tokens": 406708774.0, + "step": 17169 + }, + { + "epoch": 1.4158489321349057, + "grad_norm": 1.7640140056610107, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7550151348114014, + "num_tokens": 406746117.0, + "step": 17170 + }, + { + "epoch": 1.415931392759957, + "grad_norm": 1.9422816038131714, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.7954623103141785, + "num_tokens": 406776103.0, + "step": 17171 + }, + { + "epoch": 1.4160138533850086, + "grad_norm": 1.9944241046905518, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7377032041549683, + "num_tokens": 406807764.0, + "step": 17172 + }, + { + "epoch": 1.4160963140100602, + "grad_norm": 1.9803617000579834, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.7861986756324768, + "num_tokens": 406840888.0, + "step": 17173 + }, + { + "epoch": 1.4161787746351118, + "grad_norm": 1.997405767440796, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.745304524898529, + "num_tokens": 406870286.0, + "step": 17174 + }, + { + "epoch": 1.4162612352601633, + "grad_norm": 2.204861879348755, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.728133499622345, + "num_tokens": 406895847.0, + "step": 17175 + }, + { + "epoch": 1.4163436958852147, + "grad_norm": 2.160571575164795, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7246769666671753, + "num_tokens": 406923130.0, + "step": 17176 + }, + { + "epoch": 1.4164261565102663, + "grad_norm": 2.1996195316314697, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7159826755523682, + "num_tokens": 406950311.0, + "step": 17177 + }, + { + "epoch": 1.4165086171353178, + "grad_norm": 2.0065062046051025, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.7899975776672363, + "num_tokens": 406978214.0, + "step": 17178 + }, + { + "epoch": 1.4165910777603694, + "grad_norm": 1.9423761367797852, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7375407218933105, + "num_tokens": 407012433.0, + "step": 17179 + }, + { + "epoch": 1.416673538385421, + "grad_norm": 1.8780224323272705, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7534753084182739, + "num_tokens": 407049364.0, + "step": 17180 + }, + { + "epoch": 1.4167559990104726, + "grad_norm": 1.8807098865509033, + "learning_rate": 1e-06, + "loss": 0.5617, + "mean_token_accuracy": 0.8095965385437012, + "num_tokens": 407079266.0, + "step": 17181 + }, + { + "epoch": 1.4168384596355241, + "grad_norm": 2.0605151653289795, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.753359317779541, + "num_tokens": 407108403.0, + "step": 17182 + }, + { + "epoch": 1.4169209202605755, + "grad_norm": 2.083207607269287, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7432445287704468, + "num_tokens": 407136025.0, + "step": 17183 + }, + { + "epoch": 1.417003380885627, + "grad_norm": 2.2669172286987305, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.7842903137207031, + "num_tokens": 407159062.0, + "step": 17184 + }, + { + "epoch": 1.4170858415106786, + "grad_norm": 1.9775267839431763, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7506078481674194, + "num_tokens": 407191630.0, + "step": 17185 + }, + { + "epoch": 1.4171683021357302, + "grad_norm": 1.9143335819244385, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7675302624702454, + "num_tokens": 407224037.0, + "step": 17186 + }, + { + "epoch": 1.4172507627607818, + "grad_norm": 2.2481582164764404, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7738158702850342, + "num_tokens": 407251579.0, + "step": 17187 + }, + { + "epoch": 1.4173332233858333, + "grad_norm": 2.0698413848876953, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7553005218505859, + "num_tokens": 407280626.0, + "step": 17188 + }, + { + "epoch": 1.417415684010885, + "grad_norm": 1.9776440858840942, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7388468980789185, + "num_tokens": 407311267.0, + "step": 17189 + }, + { + "epoch": 1.4174981446359363, + "grad_norm": 2.074448347091675, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7409235239028931, + "num_tokens": 407340821.0, + "step": 17190 + }, + { + "epoch": 1.4175806052609878, + "grad_norm": 1.9729998111724854, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.727968692779541, + "num_tokens": 407373178.0, + "step": 17191 + }, + { + "epoch": 1.4176630658860394, + "grad_norm": 1.7759935855865479, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7886922359466553, + "num_tokens": 407408428.0, + "step": 17192 + }, + { + "epoch": 1.417745526511091, + "grad_norm": 2.0771994590759277, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.7881884574890137, + "num_tokens": 407435878.0, + "step": 17193 + }, + { + "epoch": 1.4178279871361426, + "grad_norm": 2.3530519008636475, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7638677358627319, + "num_tokens": 407460358.0, + "step": 17194 + }, + { + "epoch": 1.417910447761194, + "grad_norm": 2.3449277877807617, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7242667078971863, + "num_tokens": 407487190.0, + "step": 17195 + }, + { + "epoch": 1.4179929083862457, + "grad_norm": 1.7980445623397827, + "learning_rate": 1e-06, + "loss": 0.7688, + "mean_token_accuracy": 0.7564412951469421, + "num_tokens": 407521089.0, + "step": 17196 + }, + { + "epoch": 1.418075369011297, + "grad_norm": 2.0648021697998047, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7539252042770386, + "num_tokens": 407549065.0, + "step": 17197 + }, + { + "epoch": 1.4181578296363486, + "grad_norm": 1.7545084953308105, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7561681270599365, + "num_tokens": 407585069.0, + "step": 17198 + }, + { + "epoch": 1.4182402902614002, + "grad_norm": 2.072298526763916, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7668494582176208, + "num_tokens": 407611685.0, + "step": 17199 + }, + { + "epoch": 1.4183227508864518, + "grad_norm": 1.9611390829086304, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7602347731590271, + "num_tokens": 407643315.0, + "step": 17200 + }, + { + "epoch": 1.4184052115115033, + "grad_norm": 1.9237710237503052, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7664419412612915, + "num_tokens": 407673516.0, + "step": 17201 + }, + { + "epoch": 1.4184876721365547, + "grad_norm": 2.1198618412017822, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7459956407546997, + "num_tokens": 407704329.0, + "step": 17202 + }, + { + "epoch": 1.4185701327616063, + "grad_norm": 1.9103367328643799, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7602770328521729, + "num_tokens": 407736948.0, + "step": 17203 + }, + { + "epoch": 1.4186525933866578, + "grad_norm": 1.8814891576766968, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7622936964035034, + "num_tokens": 407770526.0, + "step": 17204 + }, + { + "epoch": 1.4187350540117094, + "grad_norm": 1.9563850164413452, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7665939331054688, + "num_tokens": 407800120.0, + "step": 17205 + }, + { + "epoch": 1.418817514636761, + "grad_norm": 2.4789671897888184, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7554770708084106, + "num_tokens": 407820779.0, + "step": 17206 + }, + { + "epoch": 1.4188999752618126, + "grad_norm": 1.8734194040298462, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7551976442337036, + "num_tokens": 407853479.0, + "step": 17207 + }, + { + "epoch": 1.4189824358868641, + "grad_norm": 1.9205396175384521, + "learning_rate": 1e-06, + "loss": 0.5688, + "mean_token_accuracy": 0.8068872690200806, + "num_tokens": 407881269.0, + "step": 17208 + }, + { + "epoch": 1.4190648965119155, + "grad_norm": 2.0618176460266113, + "learning_rate": 1e-06, + "loss": 0.624, + "mean_token_accuracy": 0.7949208617210388, + "num_tokens": 407908246.0, + "step": 17209 + }, + { + "epoch": 1.419147357136967, + "grad_norm": 2.075040578842163, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7620681524276733, + "num_tokens": 407934601.0, + "step": 17210 + }, + { + "epoch": 1.4192298177620186, + "grad_norm": 2.3804755210876465, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7653710842132568, + "num_tokens": 407959284.0, + "step": 17211 + }, + { + "epoch": 1.4193122783870702, + "grad_norm": 2.060802936553955, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7510021924972534, + "num_tokens": 407987866.0, + "step": 17212 + }, + { + "epoch": 1.4193947390121218, + "grad_norm": 1.8638259172439575, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.782102108001709, + "num_tokens": 408025527.0, + "step": 17213 + }, + { + "epoch": 1.4194771996371731, + "grad_norm": 2.425912857055664, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7338618636131287, + "num_tokens": 408047528.0, + "step": 17214 + }, + { + "epoch": 1.419559660262225, + "grad_norm": 2.011425495147705, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.7891674041748047, + "num_tokens": 408076206.0, + "step": 17215 + }, + { + "epoch": 1.4196421208872763, + "grad_norm": 2.2784502506256104, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.7463968396186829, + "num_tokens": 408100413.0, + "step": 17216 + }, + { + "epoch": 1.4197245815123278, + "grad_norm": 1.883805513381958, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7766826152801514, + "num_tokens": 408132026.0, + "step": 17217 + }, + { + "epoch": 1.4198070421373794, + "grad_norm": 2.2660646438598633, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7529369592666626, + "num_tokens": 408162887.0, + "step": 17218 + }, + { + "epoch": 1.419889502762431, + "grad_norm": 1.7550363540649414, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7663306593894958, + "num_tokens": 408200465.0, + "step": 17219 + }, + { + "epoch": 1.4199719633874825, + "grad_norm": 2.3225584030151367, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7317957878112793, + "num_tokens": 408222855.0, + "step": 17220 + }, + { + "epoch": 1.420054424012534, + "grad_norm": 2.2153146266937256, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7489362955093384, + "num_tokens": 408247560.0, + "step": 17221 + }, + { + "epoch": 1.4201368846375855, + "grad_norm": 2.0168137550354004, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7654649615287781, + "num_tokens": 408278679.0, + "step": 17222 + }, + { + "epoch": 1.420219345262637, + "grad_norm": 2.1944968700408936, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7686703205108643, + "num_tokens": 408302655.0, + "step": 17223 + }, + { + "epoch": 1.4203018058876886, + "grad_norm": 2.154214859008789, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7711749076843262, + "num_tokens": 408328546.0, + "step": 17224 + }, + { + "epoch": 1.4203842665127402, + "grad_norm": 2.134765863418579, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7601735591888428, + "num_tokens": 408353297.0, + "step": 17225 + }, + { + "epoch": 1.4204667271377918, + "grad_norm": 2.2404680252075195, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.7800114750862122, + "num_tokens": 408377542.0, + "step": 17226 + }, + { + "epoch": 1.4205491877628433, + "grad_norm": 2.3517489433288574, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7533393502235413, + "num_tokens": 408401764.0, + "step": 17227 + }, + { + "epoch": 1.4206316483878947, + "grad_norm": 2.1167352199554443, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7837733030319214, + "num_tokens": 408427731.0, + "step": 17228 + }, + { + "epoch": 1.4207141090129463, + "grad_norm": 1.9149335622787476, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7845920324325562, + "num_tokens": 408458095.0, + "step": 17229 + }, + { + "epoch": 1.4207965696379978, + "grad_norm": 1.8158886432647705, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7877060770988464, + "num_tokens": 408491805.0, + "step": 17230 + }, + { + "epoch": 1.4208790302630494, + "grad_norm": 2.240375280380249, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7074569463729858, + "num_tokens": 408517319.0, + "step": 17231 + }, + { + "epoch": 1.420961490888101, + "grad_norm": 1.8503755331039429, + "learning_rate": 1e-06, + "loss": 0.5811, + "mean_token_accuracy": 0.8033215999603271, + "num_tokens": 408551704.0, + "step": 17232 + }, + { + "epoch": 1.4210439515131525, + "grad_norm": 1.753757357597351, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7627050876617432, + "num_tokens": 408587196.0, + "step": 17233 + }, + { + "epoch": 1.4211264121382041, + "grad_norm": 1.935842752456665, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7684329152107239, + "num_tokens": 408618264.0, + "step": 17234 + }, + { + "epoch": 1.4212088727632555, + "grad_norm": 1.8223875761032104, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7833764553070068, + "num_tokens": 408651023.0, + "step": 17235 + }, + { + "epoch": 1.421291333388307, + "grad_norm": 2.131082534790039, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7813934087753296, + "num_tokens": 408675953.0, + "step": 17236 + }, + { + "epoch": 1.4213737940133586, + "grad_norm": 2.5559258460998535, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7471953630447388, + "num_tokens": 408695448.0, + "step": 17237 + }, + { + "epoch": 1.4214562546384102, + "grad_norm": 1.7707070112228394, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7776782512664795, + "num_tokens": 408730026.0, + "step": 17238 + }, + { + "epoch": 1.4215387152634618, + "grad_norm": 1.9851746559143066, + "learning_rate": 1e-06, + "loss": 0.5955, + "mean_token_accuracy": 0.799007773399353, + "num_tokens": 408760204.0, + "step": 17239 + }, + { + "epoch": 1.4216211758885131, + "grad_norm": 2.0257394313812256, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7584031820297241, + "num_tokens": 408787106.0, + "step": 17240 + }, + { + "epoch": 1.421703636513565, + "grad_norm": 1.7945083379745483, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.731469988822937, + "num_tokens": 408822798.0, + "step": 17241 + }, + { + "epoch": 1.4217860971386163, + "grad_norm": 2.2717788219451904, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7048536539077759, + "num_tokens": 408848623.0, + "step": 17242 + }, + { + "epoch": 1.4218685577636678, + "grad_norm": 2.0517632961273193, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7924888134002686, + "num_tokens": 408878123.0, + "step": 17243 + }, + { + "epoch": 1.4219510183887194, + "grad_norm": 2.211942434310913, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7529587149620056, + "num_tokens": 408905113.0, + "step": 17244 + }, + { + "epoch": 1.422033479013771, + "grad_norm": 2.0512843132019043, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.724779486656189, + "num_tokens": 408936047.0, + "step": 17245 + }, + { + "epoch": 1.4221159396388225, + "grad_norm": 1.7862824201583862, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7683863639831543, + "num_tokens": 408968910.0, + "step": 17246 + }, + { + "epoch": 1.422198400263874, + "grad_norm": 1.9377002716064453, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7605385780334473, + "num_tokens": 409001417.0, + "step": 17247 + }, + { + "epoch": 1.4222808608889255, + "grad_norm": 2.1277456283569336, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.7450140714645386, + "num_tokens": 409029075.0, + "step": 17248 + }, + { + "epoch": 1.422363321513977, + "grad_norm": 2.0863454341888428, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7178051471710205, + "num_tokens": 409060181.0, + "step": 17249 + }, + { + "epoch": 1.4224457821390286, + "grad_norm": 2.124397039413452, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.76585853099823, + "num_tokens": 409089774.0, + "step": 17250 + }, + { + "epoch": 1.4225282427640802, + "grad_norm": 2.032066583633423, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7078014612197876, + "num_tokens": 409120867.0, + "step": 17251 + }, + { + "epoch": 1.4226107033891318, + "grad_norm": 1.9675791263580322, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.7949028611183167, + "num_tokens": 409152430.0, + "step": 17252 + }, + { + "epoch": 1.4226931640141833, + "grad_norm": 1.9183478355407715, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7702081203460693, + "num_tokens": 409181938.0, + "step": 17253 + }, + { + "epoch": 1.4227756246392347, + "grad_norm": 2.135009527206421, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7497873306274414, + "num_tokens": 409208685.0, + "step": 17254 + }, + { + "epoch": 1.4228580852642863, + "grad_norm": 1.7482359409332275, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7627438306808472, + "num_tokens": 409241878.0, + "step": 17255 + }, + { + "epoch": 1.4229405458893378, + "grad_norm": 1.8463972806930542, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7735657691955566, + "num_tokens": 409272759.0, + "step": 17256 + }, + { + "epoch": 1.4230230065143894, + "grad_norm": 1.8309534788131714, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.776391863822937, + "num_tokens": 409307580.0, + "step": 17257 + }, + { + "epoch": 1.423105467139441, + "grad_norm": 2.4002256393432617, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7102564573287964, + "num_tokens": 409331521.0, + "step": 17258 + }, + { + "epoch": 1.4231879277644923, + "grad_norm": 1.8236498832702637, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.771709680557251, + "num_tokens": 409365050.0, + "step": 17259 + }, + { + "epoch": 1.4232703883895441, + "grad_norm": 2.0409657955169678, + "learning_rate": 1e-06, + "loss": 0.7817, + "mean_token_accuracy": 0.7466937303543091, + "num_tokens": 409394361.0, + "step": 17260 + }, + { + "epoch": 1.4233528490145955, + "grad_norm": 1.9889225959777832, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.773564338684082, + "num_tokens": 409422248.0, + "step": 17261 + }, + { + "epoch": 1.423435309639647, + "grad_norm": 1.7302075624465942, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.8020951151847839, + "num_tokens": 409456057.0, + "step": 17262 + }, + { + "epoch": 1.4235177702646986, + "grad_norm": 2.102999687194824, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7748082876205444, + "num_tokens": 409482551.0, + "step": 17263 + }, + { + "epoch": 1.4236002308897502, + "grad_norm": 1.761531949043274, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7535173296928406, + "num_tokens": 409519492.0, + "step": 17264 + }, + { + "epoch": 1.4236826915148018, + "grad_norm": 1.991772174835205, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7543516159057617, + "num_tokens": 409548752.0, + "step": 17265 + }, + { + "epoch": 1.423765152139853, + "grad_norm": 2.155872344970703, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7633787393569946, + "num_tokens": 409575847.0, + "step": 17266 + }, + { + "epoch": 1.4238476127649047, + "grad_norm": 2.110349178314209, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7402245998382568, + "num_tokens": 409605079.0, + "step": 17267 + }, + { + "epoch": 1.4239300733899563, + "grad_norm": 2.0065388679504395, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7789720296859741, + "num_tokens": 409634241.0, + "step": 17268 + }, + { + "epoch": 1.4240125340150078, + "grad_norm": 2.1464684009552, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7370661497116089, + "num_tokens": 409660615.0, + "step": 17269 + }, + { + "epoch": 1.4240949946400594, + "grad_norm": 2.1347947120666504, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7792676687240601, + "num_tokens": 409685427.0, + "step": 17270 + }, + { + "epoch": 1.424177455265111, + "grad_norm": 2.3071138858795166, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.740329384803772, + "num_tokens": 409710774.0, + "step": 17271 + }, + { + "epoch": 1.4242599158901625, + "grad_norm": 2.1881208419799805, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7578433752059937, + "num_tokens": 409735582.0, + "step": 17272 + }, + { + "epoch": 1.424342376515214, + "grad_norm": 2.0206196308135986, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7217495441436768, + "num_tokens": 409766461.0, + "step": 17273 + }, + { + "epoch": 1.4244248371402655, + "grad_norm": 1.9673510789871216, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7597200274467468, + "num_tokens": 409796835.0, + "step": 17274 + }, + { + "epoch": 1.424507297765317, + "grad_norm": 2.088564395904541, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.781477153301239, + "num_tokens": 409826106.0, + "step": 17275 + }, + { + "epoch": 1.4245897583903686, + "grad_norm": 2.07454514503479, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7311151623725891, + "num_tokens": 409852317.0, + "step": 17276 + }, + { + "epoch": 1.4246722190154202, + "grad_norm": 1.7410826683044434, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7736073732376099, + "num_tokens": 409890462.0, + "step": 17277 + }, + { + "epoch": 1.4247546796404718, + "grad_norm": 2.0222301483154297, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7582792043685913, + "num_tokens": 409918106.0, + "step": 17278 + }, + { + "epoch": 1.4248371402655233, + "grad_norm": 2.2841238975524902, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7517380714416504, + "num_tokens": 409942891.0, + "step": 17279 + }, + { + "epoch": 1.4249196008905747, + "grad_norm": 1.9093188047409058, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7763202786445618, + "num_tokens": 409974678.0, + "step": 17280 + }, + { + "epoch": 1.4250020615156263, + "grad_norm": 2.0055267810821533, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7541700601577759, + "num_tokens": 410006453.0, + "step": 17281 + }, + { + "epoch": 1.4250845221406778, + "grad_norm": 2.0249881744384766, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7525863647460938, + "num_tokens": 410033997.0, + "step": 17282 + }, + { + "epoch": 1.4251669827657294, + "grad_norm": 2.194622755050659, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7412126660346985, + "num_tokens": 410061584.0, + "step": 17283 + }, + { + "epoch": 1.425249443390781, + "grad_norm": 2.0484347343444824, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.750461220741272, + "num_tokens": 410089402.0, + "step": 17284 + }, + { + "epoch": 1.4253319040158323, + "grad_norm": 2.0294294357299805, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.75225830078125, + "num_tokens": 410119909.0, + "step": 17285 + }, + { + "epoch": 1.4254143646408841, + "grad_norm": 2.0737102031707764, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7816042900085449, + "num_tokens": 410146071.0, + "step": 17286 + }, + { + "epoch": 1.4254968252659355, + "grad_norm": 1.821935772895813, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.760607123374939, + "num_tokens": 410179977.0, + "step": 17287 + }, + { + "epoch": 1.425579285890987, + "grad_norm": 2.3547286987304688, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7559809684753418, + "num_tokens": 410203965.0, + "step": 17288 + }, + { + "epoch": 1.4256617465160386, + "grad_norm": 1.9440991878509521, + "learning_rate": 1e-06, + "loss": 0.7774, + "mean_token_accuracy": 0.7569887042045593, + "num_tokens": 410235796.0, + "step": 17289 + }, + { + "epoch": 1.4257442071410902, + "grad_norm": 2.0429131984710693, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7448216676712036, + "num_tokens": 410264958.0, + "step": 17290 + }, + { + "epoch": 1.4258266677661418, + "grad_norm": 1.934349536895752, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7711126208305359, + "num_tokens": 410292720.0, + "step": 17291 + }, + { + "epoch": 1.425909128391193, + "grad_norm": 2.136181116104126, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7402336001396179, + "num_tokens": 410320776.0, + "step": 17292 + }, + { + "epoch": 1.4259915890162447, + "grad_norm": 2.024470329284668, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7284011840820312, + "num_tokens": 410349900.0, + "step": 17293 + }, + { + "epoch": 1.4260740496412962, + "grad_norm": 2.0903589725494385, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7362056374549866, + "num_tokens": 410380443.0, + "step": 17294 + }, + { + "epoch": 1.4261565102663478, + "grad_norm": 1.9889585971832275, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7744671106338501, + "num_tokens": 410411984.0, + "step": 17295 + }, + { + "epoch": 1.4262389708913994, + "grad_norm": 1.9187110662460327, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7803769111633301, + "num_tokens": 410441778.0, + "step": 17296 + }, + { + "epoch": 1.426321431516451, + "grad_norm": 1.9397391080856323, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7705096006393433, + "num_tokens": 410471191.0, + "step": 17297 + }, + { + "epoch": 1.4264038921415025, + "grad_norm": 2.141737699508667, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7514764666557312, + "num_tokens": 410497071.0, + "step": 17298 + }, + { + "epoch": 1.4264863527665539, + "grad_norm": 1.9157284498214722, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7518211603164673, + "num_tokens": 410528067.0, + "step": 17299 + }, + { + "epoch": 1.4265688133916055, + "grad_norm": 2.0954527854919434, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.736304521560669, + "num_tokens": 410556126.0, + "step": 17300 + }, + { + "epoch": 1.426651274016657, + "grad_norm": 1.8265525102615356, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7258323431015015, + "num_tokens": 410589830.0, + "step": 17301 + }, + { + "epoch": 1.4267337346417086, + "grad_norm": 1.9983583688735962, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7358626127243042, + "num_tokens": 410618430.0, + "step": 17302 + }, + { + "epoch": 1.4268161952667602, + "grad_norm": 1.8337358236312866, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.7816838026046753, + "num_tokens": 410650509.0, + "step": 17303 + }, + { + "epoch": 1.4268986558918115, + "grad_norm": 2.0879499912261963, + "learning_rate": 1e-06, + "loss": 0.6229, + "mean_token_accuracy": 0.7974993586540222, + "num_tokens": 410674746.0, + "step": 17304 + }, + { + "epoch": 1.4269811165168633, + "grad_norm": 2.056513786315918, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.731319785118103, + "num_tokens": 410707184.0, + "step": 17305 + }, + { + "epoch": 1.4270635771419147, + "grad_norm": 2.153736114501953, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7640955448150635, + "num_tokens": 410733588.0, + "step": 17306 + }, + { + "epoch": 1.4271460377669662, + "grad_norm": 1.8918739557266235, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7745457887649536, + "num_tokens": 410767500.0, + "step": 17307 + }, + { + "epoch": 1.4272284983920178, + "grad_norm": 2.0818474292755127, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7285149693489075, + "num_tokens": 410796434.0, + "step": 17308 + }, + { + "epoch": 1.4273109590170694, + "grad_norm": 1.8052611351013184, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.7920888662338257, + "num_tokens": 410828265.0, + "step": 17309 + }, + { + "epoch": 1.427393419642121, + "grad_norm": 1.9123404026031494, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7416760921478271, + "num_tokens": 410859204.0, + "step": 17310 + }, + { + "epoch": 1.4274758802671723, + "grad_norm": 1.9471193552017212, + "learning_rate": 1e-06, + "loss": 0.5468, + "mean_token_accuracy": 0.8138635158538818, + "num_tokens": 410888681.0, + "step": 17311 + }, + { + "epoch": 1.4275583408922239, + "grad_norm": 1.7665053606033325, + "learning_rate": 1e-06, + "loss": 0.6025, + "mean_token_accuracy": 0.8037109971046448, + "num_tokens": 410924535.0, + "step": 17312 + }, + { + "epoch": 1.4276408015172755, + "grad_norm": 1.9573296308517456, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7768757343292236, + "num_tokens": 410952895.0, + "step": 17313 + }, + { + "epoch": 1.427723262142327, + "grad_norm": 1.9100908041000366, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.7941766381263733, + "num_tokens": 410982889.0, + "step": 17314 + }, + { + "epoch": 1.4278057227673786, + "grad_norm": 2.055048704147339, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.7857139110565186, + "num_tokens": 411010854.0, + "step": 17315 + }, + { + "epoch": 1.4278881833924302, + "grad_norm": 2.029646873474121, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7393521070480347, + "num_tokens": 411041077.0, + "step": 17316 + }, + { + "epoch": 1.4279706440174817, + "grad_norm": 2.054041624069214, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.7938288450241089, + "num_tokens": 411066528.0, + "step": 17317 + }, + { + "epoch": 1.428053104642533, + "grad_norm": 1.9188185930252075, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.7909097075462341, + "num_tokens": 411098161.0, + "step": 17318 + }, + { + "epoch": 1.4281355652675847, + "grad_norm": 1.941156268119812, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.790743350982666, + "num_tokens": 411129463.0, + "step": 17319 + }, + { + "epoch": 1.4282180258926362, + "grad_norm": 2.186286211013794, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7656278014183044, + "num_tokens": 411157331.0, + "step": 17320 + }, + { + "epoch": 1.4283004865176878, + "grad_norm": 2.0014357566833496, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7611895799636841, + "num_tokens": 411188618.0, + "step": 17321 + }, + { + "epoch": 1.4283829471427394, + "grad_norm": 2.064035177230835, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.774509072303772, + "num_tokens": 411214646.0, + "step": 17322 + }, + { + "epoch": 1.428465407767791, + "grad_norm": 1.8907698392868042, + "learning_rate": 1e-06, + "loss": 0.5439, + "mean_token_accuracy": 0.8095046281814575, + "num_tokens": 411245476.0, + "step": 17323 + }, + { + "epoch": 1.4285478683928425, + "grad_norm": 2.1205973625183105, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7309754490852356, + "num_tokens": 411273228.0, + "step": 17324 + }, + { + "epoch": 1.4286303290178939, + "grad_norm": 1.9770761728286743, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.772648811340332, + "num_tokens": 411303096.0, + "step": 17325 + }, + { + "epoch": 1.4287127896429455, + "grad_norm": 2.0287675857543945, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7422959208488464, + "num_tokens": 411334903.0, + "step": 17326 + }, + { + "epoch": 1.428795250267997, + "grad_norm": 2.366082191467285, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.764543890953064, + "num_tokens": 411356413.0, + "step": 17327 + }, + { + "epoch": 1.4288777108930486, + "grad_norm": 1.9461957216262817, + "learning_rate": 1e-06, + "loss": 0.6144, + "mean_token_accuracy": 0.7933838367462158, + "num_tokens": 411386920.0, + "step": 17328 + }, + { + "epoch": 1.4289601715181002, + "grad_norm": 1.716019868850708, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.8031989932060242, + "num_tokens": 411422371.0, + "step": 17329 + }, + { + "epoch": 1.4290426321431515, + "grad_norm": 2.024397850036621, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7458389401435852, + "num_tokens": 411452710.0, + "step": 17330 + }, + { + "epoch": 1.4291250927682033, + "grad_norm": 1.8967899084091187, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7889025807380676, + "num_tokens": 411484751.0, + "step": 17331 + }, + { + "epoch": 1.4292075533932547, + "grad_norm": 1.7322213649749756, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.7927840948104858, + "num_tokens": 411518110.0, + "step": 17332 + }, + { + "epoch": 1.4292900140183062, + "grad_norm": 1.7950422763824463, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7795462608337402, + "num_tokens": 411556712.0, + "step": 17333 + }, + { + "epoch": 1.4293724746433578, + "grad_norm": 1.8517358303070068, + "learning_rate": 1e-06, + "loss": 0.5424, + "mean_token_accuracy": 0.8175386786460876, + "num_tokens": 411588774.0, + "step": 17334 + }, + { + "epoch": 1.4294549352684094, + "grad_norm": 2.1518666744232178, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7290321588516235, + "num_tokens": 411615901.0, + "step": 17335 + }, + { + "epoch": 1.429537395893461, + "grad_norm": 2.043121576309204, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7795710563659668, + "num_tokens": 411641953.0, + "step": 17336 + }, + { + "epoch": 1.4296198565185123, + "grad_norm": 1.7786256074905396, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7645880579948425, + "num_tokens": 411678909.0, + "step": 17337 + }, + { + "epoch": 1.4297023171435639, + "grad_norm": 1.9160465002059937, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7747886180877686, + "num_tokens": 411710324.0, + "step": 17338 + }, + { + "epoch": 1.4297847777686155, + "grad_norm": 1.8869409561157227, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7664139270782471, + "num_tokens": 411745247.0, + "step": 17339 + }, + { + "epoch": 1.429867238393667, + "grad_norm": 2.252851724624634, + "learning_rate": 1e-06, + "loss": 0.7637, + "mean_token_accuracy": 0.7614166736602783, + "num_tokens": 411772109.0, + "step": 17340 + }, + { + "epoch": 1.4299496990187186, + "grad_norm": 2.1368958950042725, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7494910955429077, + "num_tokens": 411800240.0, + "step": 17341 + }, + { + "epoch": 1.4300321596437702, + "grad_norm": 1.9949971437454224, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7658736705780029, + "num_tokens": 411828364.0, + "step": 17342 + }, + { + "epoch": 1.4301146202688217, + "grad_norm": 1.9245891571044922, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7499581575393677, + "num_tokens": 411861678.0, + "step": 17343 + }, + { + "epoch": 1.430197080893873, + "grad_norm": 1.8757312297821045, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7539988160133362, + "num_tokens": 411894186.0, + "step": 17344 + }, + { + "epoch": 1.4302795415189247, + "grad_norm": 1.860243320465088, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7793729305267334, + "num_tokens": 411922171.0, + "step": 17345 + }, + { + "epoch": 1.4303620021439762, + "grad_norm": 1.9745296239852905, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7621000409126282, + "num_tokens": 411952210.0, + "step": 17346 + }, + { + "epoch": 1.4304444627690278, + "grad_norm": 2.3082661628723145, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7530394792556763, + "num_tokens": 411976442.0, + "step": 17347 + }, + { + "epoch": 1.4305269233940794, + "grad_norm": 1.9525864124298096, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7847124338150024, + "num_tokens": 412005676.0, + "step": 17348 + }, + { + "epoch": 1.4306093840191307, + "grad_norm": 2.1625044345855713, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7776581048965454, + "num_tokens": 412034062.0, + "step": 17349 + }, + { + "epoch": 1.4306918446441825, + "grad_norm": 2.190692186355591, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7355775237083435, + "num_tokens": 412059343.0, + "step": 17350 + }, + { + "epoch": 1.4307743052692339, + "grad_norm": 2.118212938308716, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7664939761161804, + "num_tokens": 412085598.0, + "step": 17351 + }, + { + "epoch": 1.4308567658942855, + "grad_norm": 2.5094082355499268, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7412392497062683, + "num_tokens": 412107765.0, + "step": 17352 + }, + { + "epoch": 1.430939226519337, + "grad_norm": 1.7525781393051147, + "learning_rate": 1e-06, + "loss": 0.6132, + "mean_token_accuracy": 0.7975199222564697, + "num_tokens": 412142129.0, + "step": 17353 + }, + { + "epoch": 1.4310216871443886, + "grad_norm": 1.931313395500183, + "learning_rate": 1e-06, + "loss": 0.5412, + "mean_token_accuracy": 0.8211659789085388, + "num_tokens": 412170647.0, + "step": 17354 + }, + { + "epoch": 1.4311041477694402, + "grad_norm": 1.9664667844772339, + "learning_rate": 1e-06, + "loss": 0.784, + "mean_token_accuracy": 0.751003623008728, + "num_tokens": 412199510.0, + "step": 17355 + }, + { + "epoch": 1.4311866083944915, + "grad_norm": 1.9280366897583008, + "learning_rate": 1e-06, + "loss": 0.6128, + "mean_token_accuracy": 0.7865267992019653, + "num_tokens": 412230684.0, + "step": 17356 + }, + { + "epoch": 1.431269069019543, + "grad_norm": 2.172821044921875, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7447570562362671, + "num_tokens": 412259470.0, + "step": 17357 + }, + { + "epoch": 1.4313515296445947, + "grad_norm": 1.9393912553787231, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.7929078936576843, + "num_tokens": 412292237.0, + "step": 17358 + }, + { + "epoch": 1.4314339902696462, + "grad_norm": 2.0344388484954834, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7602136731147766, + "num_tokens": 412323755.0, + "step": 17359 + }, + { + "epoch": 1.4315164508946978, + "grad_norm": 2.107616424560547, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7472154498100281, + "num_tokens": 412355415.0, + "step": 17360 + }, + { + "epoch": 1.4315989115197494, + "grad_norm": 1.729792833328247, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7723650932312012, + "num_tokens": 412393890.0, + "step": 17361 + }, + { + "epoch": 1.431681372144801, + "grad_norm": 1.7571009397506714, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7854627370834351, + "num_tokens": 412430255.0, + "step": 17362 + }, + { + "epoch": 1.4317638327698523, + "grad_norm": 1.7743175029754639, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7683229446411133, + "num_tokens": 412470013.0, + "step": 17363 + }, + { + "epoch": 1.4318462933949039, + "grad_norm": 1.8499560356140137, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7838768362998962, + "num_tokens": 412504046.0, + "step": 17364 + }, + { + "epoch": 1.4319287540199555, + "grad_norm": 2.1169278621673584, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7771081924438477, + "num_tokens": 412531458.0, + "step": 17365 + }, + { + "epoch": 1.432011214645007, + "grad_norm": 2.042299747467041, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7642813920974731, + "num_tokens": 412558873.0, + "step": 17366 + }, + { + "epoch": 1.4320936752700586, + "grad_norm": 1.70842707157135, + "learning_rate": 1e-06, + "loss": 0.6178, + "mean_token_accuracy": 0.7951686978340149, + "num_tokens": 412594101.0, + "step": 17367 + }, + { + "epoch": 1.43217613589511, + "grad_norm": 1.8218320608139038, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.785224437713623, + "num_tokens": 412627408.0, + "step": 17368 + }, + { + "epoch": 1.4322585965201617, + "grad_norm": 2.029590368270874, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7400003671646118, + "num_tokens": 412656710.0, + "step": 17369 + }, + { + "epoch": 1.432341057145213, + "grad_norm": 1.8916654586791992, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7726659178733826, + "num_tokens": 412686644.0, + "step": 17370 + }, + { + "epoch": 1.4324235177702647, + "grad_norm": 1.9037728309631348, + "learning_rate": 1e-06, + "loss": 0.5907, + "mean_token_accuracy": 0.7812305092811584, + "num_tokens": 412717709.0, + "step": 17371 + }, + { + "epoch": 1.4325059783953162, + "grad_norm": 2.120471954345703, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.782992959022522, + "num_tokens": 412745019.0, + "step": 17372 + }, + { + "epoch": 1.4325884390203678, + "grad_norm": 2.0666463375091553, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7661542296409607, + "num_tokens": 412771712.0, + "step": 17373 + }, + { + "epoch": 1.4326708996454194, + "grad_norm": 1.8738802671432495, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7791987061500549, + "num_tokens": 412803045.0, + "step": 17374 + }, + { + "epoch": 1.4327533602704707, + "grad_norm": 2.024296760559082, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7462196350097656, + "num_tokens": 412833122.0, + "step": 17375 + }, + { + "epoch": 1.4328358208955223, + "grad_norm": 1.8102068901062012, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7617166042327881, + "num_tokens": 412867614.0, + "step": 17376 + }, + { + "epoch": 1.4329182815205739, + "grad_norm": 2.3532276153564453, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7651565670967102, + "num_tokens": 412890062.0, + "step": 17377 + }, + { + "epoch": 1.4330007421456255, + "grad_norm": 2.0175273418426514, + "learning_rate": 1e-06, + "loss": 0.539, + "mean_token_accuracy": 0.8144011497497559, + "num_tokens": 412916742.0, + "step": 17378 + }, + { + "epoch": 1.433083202770677, + "grad_norm": 1.9669846296310425, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7841585874557495, + "num_tokens": 412945731.0, + "step": 17379 + }, + { + "epoch": 1.4331656633957286, + "grad_norm": 1.8864978551864624, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7444426417350769, + "num_tokens": 412982923.0, + "step": 17380 + }, + { + "epoch": 1.4332481240207802, + "grad_norm": 2.174823045730591, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7485771179199219, + "num_tokens": 413011743.0, + "step": 17381 + }, + { + "epoch": 1.4333305846458315, + "grad_norm": 1.9870641231536865, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7656461000442505, + "num_tokens": 413043309.0, + "step": 17382 + }, + { + "epoch": 1.433413045270883, + "grad_norm": 1.9574748277664185, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7554699182510376, + "num_tokens": 413074073.0, + "step": 17383 + }, + { + "epoch": 1.4334955058959347, + "grad_norm": 1.8281866312026978, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7672677040100098, + "num_tokens": 413106532.0, + "step": 17384 + }, + { + "epoch": 1.4335779665209862, + "grad_norm": 1.8424255847930908, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7715533375740051, + "num_tokens": 413139284.0, + "step": 17385 + }, + { + "epoch": 1.4336604271460378, + "grad_norm": 2.3415415287017822, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7345791459083557, + "num_tokens": 413164530.0, + "step": 17386 + }, + { + "epoch": 1.4337428877710894, + "grad_norm": 2.0738391876220703, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7436068058013916, + "num_tokens": 413194163.0, + "step": 17387 + }, + { + "epoch": 1.433825348396141, + "grad_norm": 1.9629336595535278, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7557011842727661, + "num_tokens": 413224212.0, + "step": 17388 + }, + { + "epoch": 1.4339078090211923, + "grad_norm": 2.0315029621124268, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7553049325942993, + "num_tokens": 413253888.0, + "step": 17389 + }, + { + "epoch": 1.4339902696462439, + "grad_norm": 2.2267963886260986, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7277628779411316, + "num_tokens": 413281184.0, + "step": 17390 + }, + { + "epoch": 1.4340727302712954, + "grad_norm": 1.945354700088501, + "learning_rate": 1e-06, + "loss": 0.5445, + "mean_token_accuracy": 0.8140074014663696, + "num_tokens": 413311497.0, + "step": 17391 + }, + { + "epoch": 1.434155190896347, + "grad_norm": 1.914959192276001, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7710780501365662, + "num_tokens": 413343629.0, + "step": 17392 + }, + { + "epoch": 1.4342376515213986, + "grad_norm": 1.7326786518096924, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7429631948471069, + "num_tokens": 413381722.0, + "step": 17393 + }, + { + "epoch": 1.43432011214645, + "grad_norm": 1.7939448356628418, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7603400945663452, + "num_tokens": 413417593.0, + "step": 17394 + }, + { + "epoch": 1.4344025727715017, + "grad_norm": 2.0474562644958496, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7526664733886719, + "num_tokens": 413447993.0, + "step": 17395 + }, + { + "epoch": 1.434485033396553, + "grad_norm": 1.8164803981781006, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7416718602180481, + "num_tokens": 413485677.0, + "step": 17396 + }, + { + "epoch": 1.4345674940216047, + "grad_norm": 2.1201131343841553, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7319164872169495, + "num_tokens": 413514484.0, + "step": 17397 + }, + { + "epoch": 1.4346499546466562, + "grad_norm": 1.9405661821365356, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.7631170749664307, + "num_tokens": 413549859.0, + "step": 17398 + }, + { + "epoch": 1.4347324152717078, + "grad_norm": 2.1439363956451416, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7597993612289429, + "num_tokens": 413577853.0, + "step": 17399 + }, + { + "epoch": 1.4348148758967594, + "grad_norm": 1.9881012439727783, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.7895746231079102, + "num_tokens": 413606835.0, + "step": 17400 + }, + { + "epoch": 1.4348973365218107, + "grad_norm": 2.162271022796631, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7493067979812622, + "num_tokens": 413633596.0, + "step": 17401 + }, + { + "epoch": 1.4349797971468623, + "grad_norm": 1.9536887407302856, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7684444189071655, + "num_tokens": 413666049.0, + "step": 17402 + }, + { + "epoch": 1.4350622577719139, + "grad_norm": 2.092679738998413, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7587594389915466, + "num_tokens": 413693075.0, + "step": 17403 + }, + { + "epoch": 1.4351447183969654, + "grad_norm": 1.9020386934280396, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.7572590112686157, + "num_tokens": 413724491.0, + "step": 17404 + }, + { + "epoch": 1.435227179022017, + "grad_norm": 2.5160040855407715, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7148783802986145, + "num_tokens": 413748337.0, + "step": 17405 + }, + { + "epoch": 1.4353096396470686, + "grad_norm": 1.928017258644104, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7763491868972778, + "num_tokens": 413779333.0, + "step": 17406 + }, + { + "epoch": 1.4353921002721202, + "grad_norm": 1.9264452457427979, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7491888403892517, + "num_tokens": 413814537.0, + "step": 17407 + }, + { + "epoch": 1.4354745608971715, + "grad_norm": 2.2319793701171875, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.750412106513977, + "num_tokens": 413839275.0, + "step": 17408 + }, + { + "epoch": 1.435557021522223, + "grad_norm": 2.1612887382507324, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7733606100082397, + "num_tokens": 413866461.0, + "step": 17409 + }, + { + "epoch": 1.4356394821472747, + "grad_norm": 1.9165149927139282, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7538976073265076, + "num_tokens": 413897481.0, + "step": 17410 + }, + { + "epoch": 1.4357219427723262, + "grad_norm": 2.1576647758483887, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7669570446014404, + "num_tokens": 413926227.0, + "step": 17411 + }, + { + "epoch": 1.4358044033973778, + "grad_norm": 1.9317586421966553, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7636880874633789, + "num_tokens": 413958585.0, + "step": 17412 + }, + { + "epoch": 1.4358868640224292, + "grad_norm": 2.1965160369873047, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7719277143478394, + "num_tokens": 413983966.0, + "step": 17413 + }, + { + "epoch": 1.435969324647481, + "grad_norm": 1.891173243522644, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7432201504707336, + "num_tokens": 414017334.0, + "step": 17414 + }, + { + "epoch": 1.4360517852725323, + "grad_norm": 1.9541575908660889, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.7854068279266357, + "num_tokens": 414045847.0, + "step": 17415 + }, + { + "epoch": 1.4361342458975839, + "grad_norm": 1.7025220394134521, + "learning_rate": 1e-06, + "loss": 0.5672, + "mean_token_accuracy": 0.8158361911773682, + "num_tokens": 414082621.0, + "step": 17416 + }, + { + "epoch": 1.4362167065226354, + "grad_norm": 2.1625311374664307, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.7850039601325989, + "num_tokens": 414105841.0, + "step": 17417 + }, + { + "epoch": 1.436299167147687, + "grad_norm": 2.0544774532318115, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7738738059997559, + "num_tokens": 414133476.0, + "step": 17418 + }, + { + "epoch": 1.4363816277727386, + "grad_norm": 1.998982310295105, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7872371077537537, + "num_tokens": 414163433.0, + "step": 17419 + }, + { + "epoch": 1.43646408839779, + "grad_norm": 1.9766227006912231, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.7993360757827759, + "num_tokens": 414193934.0, + "step": 17420 + }, + { + "epoch": 1.4365465490228415, + "grad_norm": 1.9021272659301758, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7380808591842651, + "num_tokens": 414226131.0, + "step": 17421 + }, + { + "epoch": 1.436629009647893, + "grad_norm": 1.9649031162261963, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7606414556503296, + "num_tokens": 414257318.0, + "step": 17422 + }, + { + "epoch": 1.4367114702729447, + "grad_norm": 1.9931639432907104, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7337280511856079, + "num_tokens": 414290500.0, + "step": 17423 + }, + { + "epoch": 1.4367939308979962, + "grad_norm": 1.967659831047058, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7603297233581543, + "num_tokens": 414319766.0, + "step": 17424 + }, + { + "epoch": 1.4368763915230478, + "grad_norm": 2.1923606395721436, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7459543943405151, + "num_tokens": 414348535.0, + "step": 17425 + }, + { + "epoch": 1.4369588521480994, + "grad_norm": 1.8677879571914673, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7300758361816406, + "num_tokens": 414382883.0, + "step": 17426 + }, + { + "epoch": 1.4370413127731507, + "grad_norm": 1.9691743850708008, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7645062208175659, + "num_tokens": 414414327.0, + "step": 17427 + }, + { + "epoch": 1.4371237733982023, + "grad_norm": 1.7982467412948608, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7660597562789917, + "num_tokens": 414454940.0, + "step": 17428 + }, + { + "epoch": 1.4372062340232539, + "grad_norm": 2.067689895629883, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7452808022499084, + "num_tokens": 414486163.0, + "step": 17429 + }, + { + "epoch": 1.4372886946483054, + "grad_norm": 2.041050910949707, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7608689665794373, + "num_tokens": 414517672.0, + "step": 17430 + }, + { + "epoch": 1.437371155273357, + "grad_norm": 1.9880143404006958, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.749245285987854, + "num_tokens": 414546611.0, + "step": 17431 + }, + { + "epoch": 1.4374536158984086, + "grad_norm": 1.9750159978866577, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7838928699493408, + "num_tokens": 414577231.0, + "step": 17432 + }, + { + "epoch": 1.4375360765234602, + "grad_norm": 2.006589651107788, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.7792546153068542, + "num_tokens": 414603944.0, + "step": 17433 + }, + { + "epoch": 1.4376185371485115, + "grad_norm": 1.8066914081573486, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7643508911132812, + "num_tokens": 414640731.0, + "step": 17434 + }, + { + "epoch": 1.437700997773563, + "grad_norm": 1.765773892402649, + "learning_rate": 1e-06, + "loss": 0.4521, + "mean_token_accuracy": 0.8499706983566284, + "num_tokens": 414675656.0, + "step": 17435 + }, + { + "epoch": 1.4377834583986147, + "grad_norm": 2.058952569961548, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.7880740165710449, + "num_tokens": 414703708.0, + "step": 17436 + }, + { + "epoch": 1.4378659190236662, + "grad_norm": 1.9177497625350952, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7728544473648071, + "num_tokens": 414736798.0, + "step": 17437 + }, + { + "epoch": 1.4379483796487178, + "grad_norm": 2.255882501602173, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7576601505279541, + "num_tokens": 414762612.0, + "step": 17438 + }, + { + "epoch": 1.4380308402737692, + "grad_norm": 1.8467456102371216, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7657544016838074, + "num_tokens": 414791952.0, + "step": 17439 + }, + { + "epoch": 1.438113300898821, + "grad_norm": 2.1745095252990723, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7739656567573547, + "num_tokens": 414819663.0, + "step": 17440 + }, + { + "epoch": 1.4381957615238723, + "grad_norm": 1.9563249349594116, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.7911429405212402, + "num_tokens": 414850688.0, + "step": 17441 + }, + { + "epoch": 1.4382782221489239, + "grad_norm": 1.8949228525161743, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7781803607940674, + "num_tokens": 414882653.0, + "step": 17442 + }, + { + "epoch": 1.4383606827739754, + "grad_norm": 2.199502944946289, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7338556051254272, + "num_tokens": 414909487.0, + "step": 17443 + }, + { + "epoch": 1.438443143399027, + "grad_norm": 1.9385254383087158, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.778956949710846, + "num_tokens": 414939442.0, + "step": 17444 + }, + { + "epoch": 1.4385256040240786, + "grad_norm": 1.8198010921478271, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.757848858833313, + "num_tokens": 414973780.0, + "step": 17445 + }, + { + "epoch": 1.43860806464913, + "grad_norm": 2.0180933475494385, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7759179472923279, + "num_tokens": 415006866.0, + "step": 17446 + }, + { + "epoch": 1.4386905252741815, + "grad_norm": 1.9901942014694214, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7621703743934631, + "num_tokens": 415038171.0, + "step": 17447 + }, + { + "epoch": 1.438772985899233, + "grad_norm": 2.2630350589752197, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.745013952255249, + "num_tokens": 415065560.0, + "step": 17448 + }, + { + "epoch": 1.4388554465242847, + "grad_norm": 1.8160947561264038, + "learning_rate": 1e-06, + "loss": 0.604, + "mean_token_accuracy": 0.7996260523796082, + "num_tokens": 415102141.0, + "step": 17449 + }, + { + "epoch": 1.4389379071493362, + "grad_norm": 2.334299325942993, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.748738169670105, + "num_tokens": 415126937.0, + "step": 17450 + }, + { + "epoch": 1.4390203677743878, + "grad_norm": 1.970070481300354, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7492557764053345, + "num_tokens": 415156871.0, + "step": 17451 + }, + { + "epoch": 1.4391028283994394, + "grad_norm": 2.1443402767181396, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7559207677841187, + "num_tokens": 415183102.0, + "step": 17452 + }, + { + "epoch": 1.4391852890244907, + "grad_norm": 1.8721745014190674, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7580713033676147, + "num_tokens": 415215967.0, + "step": 17453 + }, + { + "epoch": 1.4392677496495423, + "grad_norm": 2.0614590644836426, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7099146842956543, + "num_tokens": 415244416.0, + "step": 17454 + }, + { + "epoch": 1.4393502102745939, + "grad_norm": 1.896951675415039, + "learning_rate": 1e-06, + "loss": 0.5183, + "mean_token_accuracy": 0.8074814081192017, + "num_tokens": 415271242.0, + "step": 17455 + }, + { + "epoch": 1.4394326708996454, + "grad_norm": 1.8571711778640747, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7654000520706177, + "num_tokens": 415305763.0, + "step": 17456 + }, + { + "epoch": 1.439515131524697, + "grad_norm": 2.2939412593841553, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7463651895523071, + "num_tokens": 415330013.0, + "step": 17457 + }, + { + "epoch": 1.4395975921497484, + "grad_norm": 2.0495693683624268, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7547030448913574, + "num_tokens": 415360692.0, + "step": 17458 + }, + { + "epoch": 1.4396800527748002, + "grad_norm": 1.835680603981018, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.780143141746521, + "num_tokens": 415394617.0, + "step": 17459 + }, + { + "epoch": 1.4397625133998515, + "grad_norm": 2.3190789222717285, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.764270544052124, + "num_tokens": 415416689.0, + "step": 17460 + }, + { + "epoch": 1.439844974024903, + "grad_norm": 1.9439553022384644, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7560204267501831, + "num_tokens": 415447866.0, + "step": 17461 + }, + { + "epoch": 1.4399274346499547, + "grad_norm": 2.2082626819610596, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7594735622406006, + "num_tokens": 415474217.0, + "step": 17462 + }, + { + "epoch": 1.4400098952750062, + "grad_norm": 2.0640947818756104, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7278577089309692, + "num_tokens": 415505516.0, + "step": 17463 + }, + { + "epoch": 1.4400923559000578, + "grad_norm": 2.2522661685943604, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7712563872337341, + "num_tokens": 415531958.0, + "step": 17464 + }, + { + "epoch": 1.4401748165251091, + "grad_norm": 1.9752471446990967, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.7826782464981079, + "num_tokens": 415562577.0, + "step": 17465 + }, + { + "epoch": 1.4402572771501607, + "grad_norm": 1.9369828701019287, + "learning_rate": 1e-06, + "loss": 0.7798, + "mean_token_accuracy": 0.7599108219146729, + "num_tokens": 415595373.0, + "step": 17466 + }, + { + "epoch": 1.4403397377752123, + "grad_norm": 1.9423348903656006, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7601555585861206, + "num_tokens": 415626651.0, + "step": 17467 + }, + { + "epoch": 1.4404221984002639, + "grad_norm": 2.374596118927002, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7410433292388916, + "num_tokens": 415651920.0, + "step": 17468 + }, + { + "epoch": 1.4405046590253154, + "grad_norm": 1.979466438293457, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7568737268447876, + "num_tokens": 415681978.0, + "step": 17469 + }, + { + "epoch": 1.440587119650367, + "grad_norm": 1.9425183534622192, + "learning_rate": 1e-06, + "loss": 0.5762, + "mean_token_accuracy": 0.8080173134803772, + "num_tokens": 415714145.0, + "step": 17470 + }, + { + "epoch": 1.4406695802754186, + "grad_norm": 1.931696891784668, + "learning_rate": 1e-06, + "loss": 0.7723, + "mean_token_accuracy": 0.7554493546485901, + "num_tokens": 415745663.0, + "step": 17471 + }, + { + "epoch": 1.44075204090047, + "grad_norm": 2.0780346393585205, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7761058807373047, + "num_tokens": 415773055.0, + "step": 17472 + }, + { + "epoch": 1.4408345015255215, + "grad_norm": 2.0421929359436035, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7496201992034912, + "num_tokens": 415802095.0, + "step": 17473 + }, + { + "epoch": 1.440916962150573, + "grad_norm": 1.7236392498016357, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.7822775840759277, + "num_tokens": 415838028.0, + "step": 17474 + }, + { + "epoch": 1.4409994227756246, + "grad_norm": 2.295360803604126, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.749934196472168, + "num_tokens": 415861941.0, + "step": 17475 + }, + { + "epoch": 1.4410818834006762, + "grad_norm": 1.7786322832107544, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7427290678024292, + "num_tokens": 415897662.0, + "step": 17476 + }, + { + "epoch": 1.4411643440257278, + "grad_norm": 1.8619210720062256, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7469647526741028, + "num_tokens": 415932805.0, + "step": 17477 + }, + { + "epoch": 1.4412468046507794, + "grad_norm": 2.5469212532043457, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7508350610733032, + "num_tokens": 415953398.0, + "step": 17478 + }, + { + "epoch": 1.4413292652758307, + "grad_norm": 2.0818729400634766, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7679645419120789, + "num_tokens": 415981299.0, + "step": 17479 + }, + { + "epoch": 1.4414117259008823, + "grad_norm": 1.7712795734405518, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.7870452404022217, + "num_tokens": 416017911.0, + "step": 17480 + }, + { + "epoch": 1.4414941865259339, + "grad_norm": 1.903381586074829, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7320066690444946, + "num_tokens": 416052482.0, + "step": 17481 + }, + { + "epoch": 1.4415766471509854, + "grad_norm": 1.751032829284668, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7591131925582886, + "num_tokens": 416088028.0, + "step": 17482 + }, + { + "epoch": 1.441659107776037, + "grad_norm": 2.2823257446289062, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7591149806976318, + "num_tokens": 416111477.0, + "step": 17483 + }, + { + "epoch": 1.4417415684010884, + "grad_norm": 1.98249351978302, + "learning_rate": 1e-06, + "loss": 0.8328, + "mean_token_accuracy": 0.7393794655799866, + "num_tokens": 416141814.0, + "step": 17484 + }, + { + "epoch": 1.4418240290261402, + "grad_norm": 1.8933031558990479, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7655502557754517, + "num_tokens": 416177611.0, + "step": 17485 + }, + { + "epoch": 1.4419064896511915, + "grad_norm": 1.9293476343154907, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7594344615936279, + "num_tokens": 416213067.0, + "step": 17486 + }, + { + "epoch": 1.441988950276243, + "grad_norm": 1.9268560409545898, + "learning_rate": 1e-06, + "loss": 0.5886, + "mean_token_accuracy": 0.8071200847625732, + "num_tokens": 416242380.0, + "step": 17487 + }, + { + "epoch": 1.4420714109012946, + "grad_norm": 2.351158857345581, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7509289979934692, + "num_tokens": 416263928.0, + "step": 17488 + }, + { + "epoch": 1.4421538715263462, + "grad_norm": 1.7243449687957764, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7522093653678894, + "num_tokens": 416303133.0, + "step": 17489 + }, + { + "epoch": 1.4422363321513978, + "grad_norm": 2.0378036499023438, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7131362557411194, + "num_tokens": 416333541.0, + "step": 17490 + }, + { + "epoch": 1.4423187927764491, + "grad_norm": 2.0373520851135254, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7243537902832031, + "num_tokens": 416364944.0, + "step": 17491 + }, + { + "epoch": 1.4424012534015007, + "grad_norm": 1.9295809268951416, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7559496760368347, + "num_tokens": 416395939.0, + "step": 17492 + }, + { + "epoch": 1.4424837140265523, + "grad_norm": 1.6896432638168335, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.7874243259429932, + "num_tokens": 416433651.0, + "step": 17493 + }, + { + "epoch": 1.4425661746516039, + "grad_norm": 1.990201711654663, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7391759157180786, + "num_tokens": 416464521.0, + "step": 17494 + }, + { + "epoch": 1.4426486352766554, + "grad_norm": 1.776285171508789, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.7830371856689453, + "num_tokens": 416498261.0, + "step": 17495 + }, + { + "epoch": 1.442731095901707, + "grad_norm": 1.829525351524353, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7469329237937927, + "num_tokens": 416533961.0, + "step": 17496 + }, + { + "epoch": 1.4428135565267586, + "grad_norm": 1.8370096683502197, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.7933275699615479, + "num_tokens": 416566091.0, + "step": 17497 + }, + { + "epoch": 1.44289601715181, + "grad_norm": 2.118777275085449, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7621136903762817, + "num_tokens": 416591737.0, + "step": 17498 + }, + { + "epoch": 1.4429784777768615, + "grad_norm": 2.021251678466797, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.792378306388855, + "num_tokens": 416617711.0, + "step": 17499 + }, + { + "epoch": 1.443060938401913, + "grad_norm": 1.628971815109253, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7918558120727539, + "num_tokens": 416658677.0, + "step": 17500 + }, + { + "epoch": 1.4431433990269646, + "grad_norm": 1.8379768133163452, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.789311408996582, + "num_tokens": 416695768.0, + "step": 17501 + }, + { + "epoch": 1.4432258596520162, + "grad_norm": 1.977272391319275, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7658900618553162, + "num_tokens": 416724145.0, + "step": 17502 + }, + { + "epoch": 1.4433083202770676, + "grad_norm": 2.575967311859131, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.745343804359436, + "num_tokens": 416745014.0, + "step": 17503 + }, + { + "epoch": 1.4433907809021194, + "grad_norm": 1.9538304805755615, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.7853620052337646, + "num_tokens": 416775992.0, + "step": 17504 + }, + { + "epoch": 1.4434732415271707, + "grad_norm": 1.660637378692627, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.7971607446670532, + "num_tokens": 416815126.0, + "step": 17505 + }, + { + "epoch": 1.4435557021522223, + "grad_norm": 2.1483561992645264, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7696216106414795, + "num_tokens": 416841218.0, + "step": 17506 + }, + { + "epoch": 1.4436381627772739, + "grad_norm": 1.8772742748260498, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7486340999603271, + "num_tokens": 416875231.0, + "step": 17507 + }, + { + "epoch": 1.4437206234023254, + "grad_norm": 1.9021410942077637, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7635940909385681, + "num_tokens": 416908924.0, + "step": 17508 + }, + { + "epoch": 1.443803084027377, + "grad_norm": 1.8176469802856445, + "learning_rate": 1e-06, + "loss": 0.5468, + "mean_token_accuracy": 0.8090569972991943, + "num_tokens": 416939872.0, + "step": 17509 + }, + { + "epoch": 1.4438855446524284, + "grad_norm": 1.9649296998977661, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7563021183013916, + "num_tokens": 416970236.0, + "step": 17510 + }, + { + "epoch": 1.44396800527748, + "grad_norm": 2.178708076477051, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.7749755382537842, + "num_tokens": 416995164.0, + "step": 17511 + }, + { + "epoch": 1.4440504659025315, + "grad_norm": 1.7748481035232544, + "learning_rate": 1e-06, + "loss": 0.5795, + "mean_token_accuracy": 0.8091917037963867, + "num_tokens": 417027943.0, + "step": 17512 + }, + { + "epoch": 1.444132926527583, + "grad_norm": 2.072101593017578, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7320117354393005, + "num_tokens": 417056377.0, + "step": 17513 + }, + { + "epoch": 1.4442153871526346, + "grad_norm": 1.8587255477905273, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7804088592529297, + "num_tokens": 417090771.0, + "step": 17514 + }, + { + "epoch": 1.4442978477776862, + "grad_norm": 2.0272634029388428, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7613829970359802, + "num_tokens": 417121484.0, + "step": 17515 + }, + { + "epoch": 1.4443803084027378, + "grad_norm": 1.7036699056625366, + "learning_rate": 1e-06, + "loss": 0.5757, + "mean_token_accuracy": 0.8117913007736206, + "num_tokens": 417160536.0, + "step": 17516 + }, + { + "epoch": 1.4444627690277891, + "grad_norm": 1.9188947677612305, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.7889907360076904, + "num_tokens": 417188885.0, + "step": 17517 + }, + { + "epoch": 1.4445452296528407, + "grad_norm": 1.8231326341629028, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.7519452571868896, + "num_tokens": 417222807.0, + "step": 17518 + }, + { + "epoch": 1.4446276902778923, + "grad_norm": 1.907124638557434, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7619985342025757, + "num_tokens": 417253201.0, + "step": 17519 + }, + { + "epoch": 1.4447101509029439, + "grad_norm": 2.0882978439331055, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7557550668716431, + "num_tokens": 417281557.0, + "step": 17520 + }, + { + "epoch": 1.4447926115279954, + "grad_norm": 1.7578542232513428, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.7901253700256348, + "num_tokens": 417314907.0, + "step": 17521 + }, + { + "epoch": 1.444875072153047, + "grad_norm": 2.1860485076904297, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7740695476531982, + "num_tokens": 417337908.0, + "step": 17522 + }, + { + "epoch": 1.4449575327780986, + "grad_norm": 2.4057047367095947, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7279289960861206, + "num_tokens": 417361299.0, + "step": 17523 + }, + { + "epoch": 1.44503999340315, + "grad_norm": 2.1939175128936768, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.766018271446228, + "num_tokens": 417389244.0, + "step": 17524 + }, + { + "epoch": 1.4451224540282015, + "grad_norm": 2.1873104572296143, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7530912160873413, + "num_tokens": 417415019.0, + "step": 17525 + }, + { + "epoch": 1.445204914653253, + "grad_norm": 2.3004586696624756, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7756438255310059, + "num_tokens": 417438004.0, + "step": 17526 + }, + { + "epoch": 1.4452873752783046, + "grad_norm": 1.9050747156143188, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.7831200957298279, + "num_tokens": 417466743.0, + "step": 17527 + }, + { + "epoch": 1.4453698359033562, + "grad_norm": 1.8459821939468384, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.729232668876648, + "num_tokens": 417500257.0, + "step": 17528 + }, + { + "epoch": 1.4454522965284076, + "grad_norm": 1.8986928462982178, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7586213946342468, + "num_tokens": 417533871.0, + "step": 17529 + }, + { + "epoch": 1.4455347571534594, + "grad_norm": 2.0719990730285645, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7312060594558716, + "num_tokens": 417562025.0, + "step": 17530 + }, + { + "epoch": 1.4456172177785107, + "grad_norm": 1.7389215230941772, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7716332674026489, + "num_tokens": 417596909.0, + "step": 17531 + }, + { + "epoch": 1.4456996784035623, + "grad_norm": 2.1045734882354736, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7552673816680908, + "num_tokens": 417624657.0, + "step": 17532 + }, + { + "epoch": 1.4457821390286139, + "grad_norm": 1.8966599702835083, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7411543130874634, + "num_tokens": 417658035.0, + "step": 17533 + }, + { + "epoch": 1.4458645996536654, + "grad_norm": 1.9185034036636353, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7580989599227905, + "num_tokens": 417689522.0, + "step": 17534 + }, + { + "epoch": 1.445947060278717, + "grad_norm": 1.7433857917785645, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.7938269376754761, + "num_tokens": 417722849.0, + "step": 17535 + }, + { + "epoch": 1.4460295209037684, + "grad_norm": 1.8962208032608032, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7798063158988953, + "num_tokens": 417752958.0, + "step": 17536 + }, + { + "epoch": 1.44611198152882, + "grad_norm": 2.221996307373047, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7536201477050781, + "num_tokens": 417779789.0, + "step": 17537 + }, + { + "epoch": 1.4461944421538715, + "grad_norm": 2.1259679794311523, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7322311997413635, + "num_tokens": 417808010.0, + "step": 17538 + }, + { + "epoch": 1.446276902778923, + "grad_norm": 1.9798678159713745, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7482345700263977, + "num_tokens": 417838206.0, + "step": 17539 + }, + { + "epoch": 1.4463593634039746, + "grad_norm": 2.1506710052490234, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7456763982772827, + "num_tokens": 417863476.0, + "step": 17540 + }, + { + "epoch": 1.4464418240290262, + "grad_norm": 2.0590081214904785, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7719359993934631, + "num_tokens": 417890350.0, + "step": 17541 + }, + { + "epoch": 1.4465242846540778, + "grad_norm": 2.0771939754486084, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7349438667297363, + "num_tokens": 417920783.0, + "step": 17542 + }, + { + "epoch": 1.4466067452791291, + "grad_norm": 1.7649037837982178, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7680360078811646, + "num_tokens": 417954999.0, + "step": 17543 + }, + { + "epoch": 1.4466892059041807, + "grad_norm": 2.025360345840454, + "learning_rate": 1e-06, + "loss": 0.7712, + "mean_token_accuracy": 0.7497503757476807, + "num_tokens": 417983340.0, + "step": 17544 + }, + { + "epoch": 1.4467716665292323, + "grad_norm": 1.946915864944458, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.762495219707489, + "num_tokens": 418013150.0, + "step": 17545 + }, + { + "epoch": 1.4468541271542839, + "grad_norm": 2.3472135066986084, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.7808784246444702, + "num_tokens": 418034149.0, + "step": 17546 + }, + { + "epoch": 1.4469365877793354, + "grad_norm": 2.050551652908325, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7515242695808411, + "num_tokens": 418062283.0, + "step": 17547 + }, + { + "epoch": 1.4470190484043868, + "grad_norm": 2.108595848083496, + "learning_rate": 1e-06, + "loss": 0.599, + "mean_token_accuracy": 0.7961041331291199, + "num_tokens": 418089042.0, + "step": 17548 + }, + { + "epoch": 1.4471015090294386, + "grad_norm": 1.9368376731872559, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7370961308479309, + "num_tokens": 418120764.0, + "step": 17549 + }, + { + "epoch": 1.44718396965449, + "grad_norm": 2.172089099884033, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7449972629547119, + "num_tokens": 418146156.0, + "step": 17550 + }, + { + "epoch": 1.4472664302795415, + "grad_norm": 1.7508407831192017, + "learning_rate": 1e-06, + "loss": 0.7738, + "mean_token_accuracy": 0.7584706544876099, + "num_tokens": 418183724.0, + "step": 17551 + }, + { + "epoch": 1.447348890904593, + "grad_norm": 2.054865837097168, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7529736161231995, + "num_tokens": 418211184.0, + "step": 17552 + }, + { + "epoch": 1.4474313515296446, + "grad_norm": 2.1544651985168457, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7492108941078186, + "num_tokens": 418237806.0, + "step": 17553 + }, + { + "epoch": 1.4475138121546962, + "grad_norm": 2.0926930904388428, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7577027082443237, + "num_tokens": 418265476.0, + "step": 17554 + }, + { + "epoch": 1.4475962727797476, + "grad_norm": 1.8966093063354492, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7745325565338135, + "num_tokens": 418294582.0, + "step": 17555 + }, + { + "epoch": 1.4476787334047991, + "grad_norm": 2.034984588623047, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7721884846687317, + "num_tokens": 418321788.0, + "step": 17556 + }, + { + "epoch": 1.4477611940298507, + "grad_norm": 1.9416370391845703, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7769184112548828, + "num_tokens": 418353917.0, + "step": 17557 + }, + { + "epoch": 1.4478436546549023, + "grad_norm": 1.9072784185409546, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7729413509368896, + "num_tokens": 418382823.0, + "step": 17558 + }, + { + "epoch": 1.4479261152799539, + "grad_norm": 2.1463074684143066, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7657546997070312, + "num_tokens": 418409829.0, + "step": 17559 + }, + { + "epoch": 1.4480085759050054, + "grad_norm": 2.116121530532837, + "learning_rate": 1e-06, + "loss": 0.5759, + "mean_token_accuracy": 0.8103563189506531, + "num_tokens": 418435812.0, + "step": 17560 + }, + { + "epoch": 1.448091036530057, + "grad_norm": 1.9519898891448975, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7525689601898193, + "num_tokens": 418469664.0, + "step": 17561 + }, + { + "epoch": 1.4481734971551083, + "grad_norm": 1.943734049797058, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.7931381464004517, + "num_tokens": 418499423.0, + "step": 17562 + }, + { + "epoch": 1.44825595778016, + "grad_norm": 1.9513623714447021, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7485212087631226, + "num_tokens": 418531305.0, + "step": 17563 + }, + { + "epoch": 1.4483384184052115, + "grad_norm": 1.9869413375854492, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7410993576049805, + "num_tokens": 418561750.0, + "step": 17564 + }, + { + "epoch": 1.448420879030263, + "grad_norm": 2.251101016998291, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7384044528007507, + "num_tokens": 418585635.0, + "step": 17565 + }, + { + "epoch": 1.4485033396553146, + "grad_norm": 1.7865432500839233, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.795796275138855, + "num_tokens": 418618634.0, + "step": 17566 + }, + { + "epoch": 1.448585800280366, + "grad_norm": 1.9691544771194458, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.753824770450592, + "num_tokens": 418647674.0, + "step": 17567 + }, + { + "epoch": 1.4486682609054178, + "grad_norm": 2.337263584136963, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7562835812568665, + "num_tokens": 418671924.0, + "step": 17568 + }, + { + "epoch": 1.4487507215304691, + "grad_norm": 1.863213300704956, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7598702907562256, + "num_tokens": 418708931.0, + "step": 17569 + }, + { + "epoch": 1.4488331821555207, + "grad_norm": 1.944016456604004, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7371461391448975, + "num_tokens": 418741614.0, + "step": 17570 + }, + { + "epoch": 1.4489156427805723, + "grad_norm": 2.227461099624634, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7631996870040894, + "num_tokens": 418765571.0, + "step": 17571 + }, + { + "epoch": 1.4489981034056238, + "grad_norm": 1.7818282842636108, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7894619703292847, + "num_tokens": 418800365.0, + "step": 17572 + }, + { + "epoch": 1.4490805640306754, + "grad_norm": 1.7635459899902344, + "learning_rate": 1e-06, + "loss": 0.5873, + "mean_token_accuracy": 0.7987993955612183, + "num_tokens": 418831739.0, + "step": 17573 + }, + { + "epoch": 1.4491630246557268, + "grad_norm": 2.0243918895721436, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7370279431343079, + "num_tokens": 418862765.0, + "step": 17574 + }, + { + "epoch": 1.4492454852807786, + "grad_norm": 2.0269312858581543, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7612367272377014, + "num_tokens": 418893860.0, + "step": 17575 + }, + { + "epoch": 1.44932794590583, + "grad_norm": 1.8622238636016846, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7374247908592224, + "num_tokens": 418926267.0, + "step": 17576 + }, + { + "epoch": 1.4494104065308815, + "grad_norm": 1.759531855583191, + "learning_rate": 1e-06, + "loss": 0.5878, + "mean_token_accuracy": 0.800318717956543, + "num_tokens": 418956987.0, + "step": 17577 + }, + { + "epoch": 1.449492867155933, + "grad_norm": 1.8305163383483887, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7845470905303955, + "num_tokens": 418989009.0, + "step": 17578 + }, + { + "epoch": 1.4495753277809846, + "grad_norm": 2.1254284381866455, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7639569044113159, + "num_tokens": 419017438.0, + "step": 17579 + }, + { + "epoch": 1.4496577884060362, + "grad_norm": 2.395656108856201, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7196149826049805, + "num_tokens": 419043401.0, + "step": 17580 + }, + { + "epoch": 1.4497402490310876, + "grad_norm": 2.2056093215942383, + "learning_rate": 1e-06, + "loss": 0.7877, + "mean_token_accuracy": 0.7489496469497681, + "num_tokens": 419070452.0, + "step": 17581 + }, + { + "epoch": 1.4498227096561391, + "grad_norm": 1.8977634906768799, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7722974419593811, + "num_tokens": 419100842.0, + "step": 17582 + }, + { + "epoch": 1.4499051702811907, + "grad_norm": 2.1481690406799316, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7426702380180359, + "num_tokens": 419128908.0, + "step": 17583 + }, + { + "epoch": 1.4499876309062423, + "grad_norm": 2.4427642822265625, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7345978021621704, + "num_tokens": 419149444.0, + "step": 17584 + }, + { + "epoch": 1.4500700915312938, + "grad_norm": 1.7866555452346802, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.7986032962799072, + "num_tokens": 419183131.0, + "step": 17585 + }, + { + "epoch": 1.4501525521563454, + "grad_norm": 2.0609147548675537, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7608705759048462, + "num_tokens": 419214123.0, + "step": 17586 + }, + { + "epoch": 1.450235012781397, + "grad_norm": 1.8234707117080688, + "learning_rate": 1e-06, + "loss": 0.5857, + "mean_token_accuracy": 0.8027650117874146, + "num_tokens": 419247414.0, + "step": 17587 + }, + { + "epoch": 1.4503174734064483, + "grad_norm": 1.7577729225158691, + "learning_rate": 1e-06, + "loss": 0.6203, + "mean_token_accuracy": 0.795320987701416, + "num_tokens": 419283094.0, + "step": 17588 + }, + { + "epoch": 1.4503999340315, + "grad_norm": 2.110922336578369, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7686508893966675, + "num_tokens": 419311127.0, + "step": 17589 + }, + { + "epoch": 1.4504823946565515, + "grad_norm": 2.2150514125823975, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7686539888381958, + "num_tokens": 419336278.0, + "step": 17590 + }, + { + "epoch": 1.450564855281603, + "grad_norm": 1.7937812805175781, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7758992910385132, + "num_tokens": 419370971.0, + "step": 17591 + }, + { + "epoch": 1.4506473159066546, + "grad_norm": 2.1599268913269043, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7568836808204651, + "num_tokens": 419399170.0, + "step": 17592 + }, + { + "epoch": 1.450729776531706, + "grad_norm": 1.8552318811416626, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7679016590118408, + "num_tokens": 419432167.0, + "step": 17593 + }, + { + "epoch": 1.4508122371567578, + "grad_norm": 2.1413276195526123, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7489254474639893, + "num_tokens": 419456327.0, + "step": 17594 + }, + { + "epoch": 1.4508946977818091, + "grad_norm": 1.9831851720809937, + "learning_rate": 1e-06, + "loss": 0.522, + "mean_token_accuracy": 0.8257156610488892, + "num_tokens": 419482503.0, + "step": 17595 + }, + { + "epoch": 1.4509771584068607, + "grad_norm": 1.9255903959274292, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7610010504722595, + "num_tokens": 419516238.0, + "step": 17596 + }, + { + "epoch": 1.4510596190319123, + "grad_norm": 2.026224136352539, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7500004768371582, + "num_tokens": 419546475.0, + "step": 17597 + }, + { + "epoch": 1.4511420796569638, + "grad_norm": 2.093151330947876, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.7855684161186218, + "num_tokens": 419571175.0, + "step": 17598 + }, + { + "epoch": 1.4512245402820154, + "grad_norm": 1.8275190591812134, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7585106492042542, + "num_tokens": 419608049.0, + "step": 17599 + }, + { + "epoch": 1.4513070009070668, + "grad_norm": 2.083401918411255, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7592039108276367, + "num_tokens": 419638056.0, + "step": 17600 + }, + { + "epoch": 1.4513894615321183, + "grad_norm": 2.0137312412261963, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7552539706230164, + "num_tokens": 419669203.0, + "step": 17601 + }, + { + "epoch": 1.45147192215717, + "grad_norm": 1.9496312141418457, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7340596914291382, + "num_tokens": 419701164.0, + "step": 17602 + }, + { + "epoch": 1.4515543827822215, + "grad_norm": 2.1532180309295654, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7715773582458496, + "num_tokens": 419727336.0, + "step": 17603 + }, + { + "epoch": 1.451636843407273, + "grad_norm": 2.4011852741241455, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7433285713195801, + "num_tokens": 419750152.0, + "step": 17604 + }, + { + "epoch": 1.4517193040323246, + "grad_norm": 1.9188146591186523, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.7945886254310608, + "num_tokens": 419780019.0, + "step": 17605 + }, + { + "epoch": 1.4518017646573762, + "grad_norm": 2.0188076496124268, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.791599690914154, + "num_tokens": 419805310.0, + "step": 17606 + }, + { + "epoch": 1.4518842252824276, + "grad_norm": 1.8640555143356323, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7290723323822021, + "num_tokens": 419837695.0, + "step": 17607 + }, + { + "epoch": 1.4519666859074791, + "grad_norm": 2.033109426498413, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7761377096176147, + "num_tokens": 419865359.0, + "step": 17608 + }, + { + "epoch": 1.4520491465325307, + "grad_norm": 2.0919830799102783, + "learning_rate": 1e-06, + "loss": 0.765, + "mean_token_accuracy": 0.752144455909729, + "num_tokens": 419891170.0, + "step": 17609 + }, + { + "epoch": 1.4521316071575823, + "grad_norm": 1.7828271389007568, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7965134382247925, + "num_tokens": 419924059.0, + "step": 17610 + }, + { + "epoch": 1.4522140677826338, + "grad_norm": 1.7377996444702148, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7676401734352112, + "num_tokens": 419961292.0, + "step": 17611 + }, + { + "epoch": 1.4522965284076852, + "grad_norm": 1.9376219511032104, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7741314768791199, + "num_tokens": 419994012.0, + "step": 17612 + }, + { + "epoch": 1.452378989032737, + "grad_norm": 2.131686210632324, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7653706073760986, + "num_tokens": 420019978.0, + "step": 17613 + }, + { + "epoch": 1.4524614496577883, + "grad_norm": 1.9824802875518799, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7319043874740601, + "num_tokens": 420051397.0, + "step": 17614 + }, + { + "epoch": 1.45254391028284, + "grad_norm": 2.016298294067383, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7876893281936646, + "num_tokens": 420080919.0, + "step": 17615 + }, + { + "epoch": 1.4526263709078915, + "grad_norm": 2.15657114982605, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7459174394607544, + "num_tokens": 420108936.0, + "step": 17616 + }, + { + "epoch": 1.452708831532943, + "grad_norm": 1.940534234046936, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.7946226596832275, + "num_tokens": 420139903.0, + "step": 17617 + }, + { + "epoch": 1.4527912921579946, + "grad_norm": 2.0079879760742188, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7742307782173157, + "num_tokens": 420168915.0, + "step": 17618 + }, + { + "epoch": 1.452873752783046, + "grad_norm": 2.1236190795898438, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7781643867492676, + "num_tokens": 420196052.0, + "step": 17619 + }, + { + "epoch": 1.4529562134080976, + "grad_norm": 2.078042507171631, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7249552011489868, + "num_tokens": 420224369.0, + "step": 17620 + }, + { + "epoch": 1.4530386740331491, + "grad_norm": 1.965985655784607, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7687743902206421, + "num_tokens": 420259535.0, + "step": 17621 + }, + { + "epoch": 1.4531211346582007, + "grad_norm": 2.03073787689209, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7683846950531006, + "num_tokens": 420289875.0, + "step": 17622 + }, + { + "epoch": 1.4532035952832523, + "grad_norm": 1.94712495803833, + "learning_rate": 1e-06, + "loss": 0.6126, + "mean_token_accuracy": 0.7873343825340271, + "num_tokens": 420318041.0, + "step": 17623 + }, + { + "epoch": 1.4532860559083038, + "grad_norm": 1.8967150449752808, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7692334651947021, + "num_tokens": 420350473.0, + "step": 17624 + }, + { + "epoch": 1.4533685165333554, + "grad_norm": 1.9486100673675537, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.7870561480522156, + "num_tokens": 420380495.0, + "step": 17625 + }, + { + "epoch": 1.4534509771584068, + "grad_norm": 1.8271453380584717, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.7600560188293457, + "num_tokens": 420416056.0, + "step": 17626 + }, + { + "epoch": 1.4535334377834583, + "grad_norm": 2.0027685165405273, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7435405254364014, + "num_tokens": 420447997.0, + "step": 17627 + }, + { + "epoch": 1.45361589840851, + "grad_norm": 1.9368083477020264, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7780480980873108, + "num_tokens": 420478012.0, + "step": 17628 + }, + { + "epoch": 1.4536983590335615, + "grad_norm": 1.9184104204177856, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7620715498924255, + "num_tokens": 420507325.0, + "step": 17629 + }, + { + "epoch": 1.453780819658613, + "grad_norm": 2.0048325061798096, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.7959709167480469, + "num_tokens": 420534361.0, + "step": 17630 + }, + { + "epoch": 1.4538632802836646, + "grad_norm": 1.9219157695770264, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7673006653785706, + "num_tokens": 420566016.0, + "step": 17631 + }, + { + "epoch": 1.4539457409087162, + "grad_norm": 1.7628259658813477, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7670538425445557, + "num_tokens": 420603094.0, + "step": 17632 + }, + { + "epoch": 1.4540282015337676, + "grad_norm": 1.9916388988494873, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7728204727172852, + "num_tokens": 420630747.0, + "step": 17633 + }, + { + "epoch": 1.4541106621588191, + "grad_norm": 1.9647002220153809, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7455359697341919, + "num_tokens": 420662645.0, + "step": 17634 + }, + { + "epoch": 1.4541931227838707, + "grad_norm": 1.9327501058578491, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7635696530342102, + "num_tokens": 420695298.0, + "step": 17635 + }, + { + "epoch": 1.4542755834089223, + "grad_norm": 2.1306650638580322, + "learning_rate": 1e-06, + "loss": 0.7808, + "mean_token_accuracy": 0.7510480284690857, + "num_tokens": 420725649.0, + "step": 17636 + }, + { + "epoch": 1.4543580440339738, + "grad_norm": 1.8944214582443237, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7888696193695068, + "num_tokens": 420756222.0, + "step": 17637 + }, + { + "epoch": 1.4544405046590252, + "grad_norm": 2.1732981204986572, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.7942986488342285, + "num_tokens": 420779324.0, + "step": 17638 + }, + { + "epoch": 1.454522965284077, + "grad_norm": 1.9192261695861816, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7804235816001892, + "num_tokens": 420808776.0, + "step": 17639 + }, + { + "epoch": 1.4546054259091283, + "grad_norm": 2.1085314750671387, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7225213050842285, + "num_tokens": 420835307.0, + "step": 17640 + }, + { + "epoch": 1.45468788653418, + "grad_norm": 2.0161044597625732, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7706199884414673, + "num_tokens": 420864120.0, + "step": 17641 + }, + { + "epoch": 1.4547703471592315, + "grad_norm": 1.9907554388046265, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7360408306121826, + "num_tokens": 420897103.0, + "step": 17642 + }, + { + "epoch": 1.454852807784283, + "grad_norm": 2.109976291656494, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.7588689923286438, + "num_tokens": 420923147.0, + "step": 17643 + }, + { + "epoch": 1.4549352684093346, + "grad_norm": 1.9091908931732178, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.7543076276779175, + "num_tokens": 420957547.0, + "step": 17644 + }, + { + "epoch": 1.455017729034386, + "grad_norm": 1.951303243637085, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7355340719223022, + "num_tokens": 420990868.0, + "step": 17645 + }, + { + "epoch": 1.4551001896594375, + "grad_norm": 1.9589941501617432, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7823692560195923, + "num_tokens": 421023143.0, + "step": 17646 + }, + { + "epoch": 1.4551826502844891, + "grad_norm": 1.7600349187850952, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.7806586623191833, + "num_tokens": 421057305.0, + "step": 17647 + }, + { + "epoch": 1.4552651109095407, + "grad_norm": 1.9192792177200317, + "learning_rate": 1e-06, + "loss": 0.6015, + "mean_token_accuracy": 0.8011103868484497, + "num_tokens": 421087011.0, + "step": 17648 + }, + { + "epoch": 1.4553475715345923, + "grad_norm": 1.8433808088302612, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7780894041061401, + "num_tokens": 421119977.0, + "step": 17649 + }, + { + "epoch": 1.4554300321596438, + "grad_norm": 2.123929262161255, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7346974015235901, + "num_tokens": 421148677.0, + "step": 17650 + }, + { + "epoch": 1.4555124927846954, + "grad_norm": 2.2064216136932373, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7707329392433167, + "num_tokens": 421173857.0, + "step": 17651 + }, + { + "epoch": 1.4555949534097468, + "grad_norm": 1.8510868549346924, + "learning_rate": 1e-06, + "loss": 0.7619, + "mean_token_accuracy": 0.7565096616744995, + "num_tokens": 421207583.0, + "step": 17652 + }, + { + "epoch": 1.4556774140347983, + "grad_norm": 2.3176522254943848, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7403322458267212, + "num_tokens": 421231733.0, + "step": 17653 + }, + { + "epoch": 1.45575987465985, + "grad_norm": 1.9394453763961792, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7533074617385864, + "num_tokens": 421262007.0, + "step": 17654 + }, + { + "epoch": 1.4558423352849015, + "grad_norm": 2.2777152061462402, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7197005748748779, + "num_tokens": 421286529.0, + "step": 17655 + }, + { + "epoch": 1.455924795909953, + "grad_norm": 1.8712049722671509, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7466070652008057, + "num_tokens": 421319049.0, + "step": 17656 + }, + { + "epoch": 1.4560072565350044, + "grad_norm": 1.8171727657318115, + "learning_rate": 1e-06, + "loss": 0.5666, + "mean_token_accuracy": 0.8100874423980713, + "num_tokens": 421349977.0, + "step": 17657 + }, + { + "epoch": 1.4560897171600562, + "grad_norm": 1.9634640216827393, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7799054384231567, + "num_tokens": 421381852.0, + "step": 17658 + }, + { + "epoch": 1.4561721777851075, + "grad_norm": 1.9452447891235352, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7789220809936523, + "num_tokens": 421411452.0, + "step": 17659 + }, + { + "epoch": 1.4562546384101591, + "grad_norm": 2.008251667022705, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7613343000411987, + "num_tokens": 421440672.0, + "step": 17660 + }, + { + "epoch": 1.4563370990352107, + "grad_norm": 2.2760610580444336, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7298487424850464, + "num_tokens": 421465325.0, + "step": 17661 + }, + { + "epoch": 1.4564195596602623, + "grad_norm": 2.143720865249634, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.76158607006073, + "num_tokens": 421493049.0, + "step": 17662 + }, + { + "epoch": 1.4565020202853138, + "grad_norm": 1.8629305362701416, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7805132865905762, + "num_tokens": 421524285.0, + "step": 17663 + }, + { + "epoch": 1.4565844809103652, + "grad_norm": 1.7937288284301758, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7850444316864014, + "num_tokens": 421558915.0, + "step": 17664 + }, + { + "epoch": 1.4566669415354168, + "grad_norm": 2.107473611831665, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7559574246406555, + "num_tokens": 421590437.0, + "step": 17665 + }, + { + "epoch": 1.4567494021604683, + "grad_norm": 2.0926132202148438, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7368713617324829, + "num_tokens": 421618573.0, + "step": 17666 + }, + { + "epoch": 1.45683186278552, + "grad_norm": 2.09698748588562, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7233932018280029, + "num_tokens": 421651168.0, + "step": 17667 + }, + { + "epoch": 1.4569143234105715, + "grad_norm": 2.139097213745117, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7208884954452515, + "num_tokens": 421679144.0, + "step": 17668 + }, + { + "epoch": 1.456996784035623, + "grad_norm": 2.129098415374756, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7728068232536316, + "num_tokens": 421705795.0, + "step": 17669 + }, + { + "epoch": 1.4570792446606746, + "grad_norm": 2.0497236251831055, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7174698114395142, + "num_tokens": 421735733.0, + "step": 17670 + }, + { + "epoch": 1.457161705285726, + "grad_norm": 2.0167808532714844, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.7973917126655579, + "num_tokens": 421760989.0, + "step": 17671 + }, + { + "epoch": 1.4572441659107775, + "grad_norm": 1.9495644569396973, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7830474376678467, + "num_tokens": 421790792.0, + "step": 17672 + }, + { + "epoch": 1.4573266265358291, + "grad_norm": 1.856205701828003, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7644561529159546, + "num_tokens": 421824590.0, + "step": 17673 + }, + { + "epoch": 1.4574090871608807, + "grad_norm": 1.9773527383804321, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7538495063781738, + "num_tokens": 421858501.0, + "step": 17674 + }, + { + "epoch": 1.4574915477859323, + "grad_norm": 1.9961884021759033, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7627504467964172, + "num_tokens": 421890344.0, + "step": 17675 + }, + { + "epoch": 1.4575740084109838, + "grad_norm": 1.664887547492981, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7687053084373474, + "num_tokens": 421927811.0, + "step": 17676 + }, + { + "epoch": 1.4576564690360354, + "grad_norm": 2.003100633621216, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7775752544403076, + "num_tokens": 421956720.0, + "step": 17677 + }, + { + "epoch": 1.4577389296610868, + "grad_norm": 2.3119428157806396, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.764333963394165, + "num_tokens": 421979262.0, + "step": 17678 + }, + { + "epoch": 1.4578213902861383, + "grad_norm": 1.935590147972107, + "learning_rate": 1e-06, + "loss": 0.5941, + "mean_token_accuracy": 0.7945162057876587, + "num_tokens": 422009501.0, + "step": 17679 + }, + { + "epoch": 1.45790385091119, + "grad_norm": 2.1209518909454346, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.779029130935669, + "num_tokens": 422034638.0, + "step": 17680 + }, + { + "epoch": 1.4579863115362415, + "grad_norm": 1.8683291673660278, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.7522813677787781, + "num_tokens": 422068362.0, + "step": 17681 + }, + { + "epoch": 1.458068772161293, + "grad_norm": 1.7917470932006836, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7338143587112427, + "num_tokens": 422103440.0, + "step": 17682 + }, + { + "epoch": 1.4581512327863444, + "grad_norm": 1.9021573066711426, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7618299722671509, + "num_tokens": 422136740.0, + "step": 17683 + }, + { + "epoch": 1.4582336934113962, + "grad_norm": 1.9789386987686157, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7860621809959412, + "num_tokens": 422164842.0, + "step": 17684 + }, + { + "epoch": 1.4583161540364475, + "grad_norm": 1.7117722034454346, + "learning_rate": 1e-06, + "loss": 0.5856, + "mean_token_accuracy": 0.8026145696640015, + "num_tokens": 422199102.0, + "step": 17685 + }, + { + "epoch": 1.4583986146614991, + "grad_norm": 1.8577587604522705, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.7925029397010803, + "num_tokens": 422229821.0, + "step": 17686 + }, + { + "epoch": 1.4584810752865507, + "grad_norm": 2.009385108947754, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7932809591293335, + "num_tokens": 422256324.0, + "step": 17687 + }, + { + "epoch": 1.4585635359116023, + "grad_norm": 1.9612585306167603, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.757523238658905, + "num_tokens": 422284939.0, + "step": 17688 + }, + { + "epoch": 1.4586459965366538, + "grad_norm": 1.930919885635376, + "learning_rate": 1e-06, + "loss": 0.5279, + "mean_token_accuracy": 0.8257998824119568, + "num_tokens": 422313401.0, + "step": 17689 + }, + { + "epoch": 1.4587284571617052, + "grad_norm": 2.1449947357177734, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7683895826339722, + "num_tokens": 422339495.0, + "step": 17690 + }, + { + "epoch": 1.4588109177867568, + "grad_norm": 2.0643656253814697, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7567546963691711, + "num_tokens": 422368267.0, + "step": 17691 + }, + { + "epoch": 1.4588933784118083, + "grad_norm": 1.8831746578216553, + "learning_rate": 1e-06, + "loss": 0.7452, + "mean_token_accuracy": 0.7703510522842407, + "num_tokens": 422398391.0, + "step": 17692 + }, + { + "epoch": 1.45897583903686, + "grad_norm": 2.2336490154266357, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7637442350387573, + "num_tokens": 422423938.0, + "step": 17693 + }, + { + "epoch": 1.4590582996619115, + "grad_norm": 2.3752872943878174, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7366489171981812, + "num_tokens": 422446343.0, + "step": 17694 + }, + { + "epoch": 1.459140760286963, + "grad_norm": 2.2106668949127197, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7836723923683167, + "num_tokens": 422471844.0, + "step": 17695 + }, + { + "epoch": 1.4592232209120146, + "grad_norm": 2.0786945819854736, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.7934175729751587, + "num_tokens": 422498203.0, + "step": 17696 + }, + { + "epoch": 1.459305681537066, + "grad_norm": 1.8485628366470337, + "learning_rate": 1e-06, + "loss": 0.7852, + "mean_token_accuracy": 0.7451009750366211, + "num_tokens": 422533610.0, + "step": 17697 + }, + { + "epoch": 1.4593881421621175, + "grad_norm": 2.2499947547912598, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7660946846008301, + "num_tokens": 422557353.0, + "step": 17698 + }, + { + "epoch": 1.4594706027871691, + "grad_norm": 1.8805186748504639, + "learning_rate": 1e-06, + "loss": 0.6014, + "mean_token_accuracy": 0.8018835186958313, + "num_tokens": 422588613.0, + "step": 17699 + }, + { + "epoch": 1.4595530634122207, + "grad_norm": 2.056236982345581, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7469560503959656, + "num_tokens": 422618201.0, + "step": 17700 + }, + { + "epoch": 1.4596355240372723, + "grad_norm": 2.058809995651245, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7229079604148865, + "num_tokens": 422646020.0, + "step": 17701 + }, + { + "epoch": 1.4597179846623236, + "grad_norm": 2.1843230724334717, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7312109470367432, + "num_tokens": 422670840.0, + "step": 17702 + }, + { + "epoch": 1.4598004452873754, + "grad_norm": 2.034151554107666, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7260076999664307, + "num_tokens": 422704952.0, + "step": 17703 + }, + { + "epoch": 1.4598829059124268, + "grad_norm": 1.9290515184402466, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.7848430871963501, + "num_tokens": 422735928.0, + "step": 17704 + }, + { + "epoch": 1.4599653665374783, + "grad_norm": 2.062183380126953, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7454245090484619, + "num_tokens": 422763099.0, + "step": 17705 + }, + { + "epoch": 1.46004782716253, + "grad_norm": 2.105896472930908, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7330330014228821, + "num_tokens": 422792319.0, + "step": 17706 + }, + { + "epoch": 1.4601302877875815, + "grad_norm": 1.9701189994812012, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.7631450295448303, + "num_tokens": 422826870.0, + "step": 17707 + }, + { + "epoch": 1.460212748412633, + "grad_norm": 2.001892328262329, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7835344672203064, + "num_tokens": 422856587.0, + "step": 17708 + }, + { + "epoch": 1.4602952090376844, + "grad_norm": 1.7494908571243286, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7646124362945557, + "num_tokens": 422891551.0, + "step": 17709 + }, + { + "epoch": 1.460377669662736, + "grad_norm": 1.935114860534668, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.751686155796051, + "num_tokens": 422920881.0, + "step": 17710 + }, + { + "epoch": 1.4604601302877875, + "grad_norm": 2.156609296798706, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7331738471984863, + "num_tokens": 422948544.0, + "step": 17711 + }, + { + "epoch": 1.4605425909128391, + "grad_norm": 2.174221992492676, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7728977799415588, + "num_tokens": 422975733.0, + "step": 17712 + }, + { + "epoch": 1.4606250515378907, + "grad_norm": 1.9493281841278076, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7353702783584595, + "num_tokens": 423006549.0, + "step": 17713 + }, + { + "epoch": 1.4607075121629423, + "grad_norm": 1.9033524990081787, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7638168931007385, + "num_tokens": 423038718.0, + "step": 17714 + }, + { + "epoch": 1.4607899727879938, + "grad_norm": 2.0821144580841064, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.77281254529953, + "num_tokens": 423066702.0, + "step": 17715 + }, + { + "epoch": 1.4608724334130452, + "grad_norm": 1.7892624139785767, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7814829349517822, + "num_tokens": 423102882.0, + "step": 17716 + }, + { + "epoch": 1.4609548940380968, + "grad_norm": 2.103100538253784, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7687898874282837, + "num_tokens": 423127000.0, + "step": 17717 + }, + { + "epoch": 1.4610373546631483, + "grad_norm": 1.9583055973052979, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7805391550064087, + "num_tokens": 423156115.0, + "step": 17718 + }, + { + "epoch": 1.4611198152882, + "grad_norm": 1.6924023628234863, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7696389555931091, + "num_tokens": 423193436.0, + "step": 17719 + }, + { + "epoch": 1.4612022759132515, + "grad_norm": 2.165829658508301, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7693653106689453, + "num_tokens": 423218807.0, + "step": 17720 + }, + { + "epoch": 1.461284736538303, + "grad_norm": 2.6807146072387695, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7331670522689819, + "num_tokens": 423237204.0, + "step": 17721 + }, + { + "epoch": 1.4613671971633546, + "grad_norm": 1.9824525117874146, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7616701126098633, + "num_tokens": 423269214.0, + "step": 17722 + }, + { + "epoch": 1.461449657788406, + "grad_norm": 2.244040012359619, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7546466588973999, + "num_tokens": 423294416.0, + "step": 17723 + }, + { + "epoch": 1.4615321184134575, + "grad_norm": 1.9068403244018555, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.8027584552764893, + "num_tokens": 423325009.0, + "step": 17724 + }, + { + "epoch": 1.461614579038509, + "grad_norm": 2.1400656700134277, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7506723999977112, + "num_tokens": 423349737.0, + "step": 17725 + }, + { + "epoch": 1.4616970396635607, + "grad_norm": 2.0890769958496094, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7669453620910645, + "num_tokens": 423376623.0, + "step": 17726 + }, + { + "epoch": 1.4617795002886123, + "grad_norm": 1.7298756837844849, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7685492038726807, + "num_tokens": 423417190.0, + "step": 17727 + }, + { + "epoch": 1.4618619609136636, + "grad_norm": 1.6098039150238037, + "learning_rate": 1e-06, + "loss": 0.5685, + "mean_token_accuracy": 0.810248851776123, + "num_tokens": 423456445.0, + "step": 17728 + }, + { + "epoch": 1.4619444215387154, + "grad_norm": 1.9618291854858398, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.7916491031646729, + "num_tokens": 423486226.0, + "step": 17729 + }, + { + "epoch": 1.4620268821637668, + "grad_norm": 1.9354625940322876, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.768235981464386, + "num_tokens": 423517503.0, + "step": 17730 + }, + { + "epoch": 1.4621093427888183, + "grad_norm": 1.9853852987289429, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.770478367805481, + "num_tokens": 423547053.0, + "step": 17731 + }, + { + "epoch": 1.46219180341387, + "grad_norm": 2.130607843399048, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7662180662155151, + "num_tokens": 423572878.0, + "step": 17732 + }, + { + "epoch": 1.4622742640389215, + "grad_norm": 2.1243090629577637, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7366477251052856, + "num_tokens": 423601673.0, + "step": 17733 + }, + { + "epoch": 1.462356724663973, + "grad_norm": 1.989198088645935, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7705578804016113, + "num_tokens": 423630283.0, + "step": 17734 + }, + { + "epoch": 1.4624391852890244, + "grad_norm": 1.9717580080032349, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7426712512969971, + "num_tokens": 423658398.0, + "step": 17735 + }, + { + "epoch": 1.462521645914076, + "grad_norm": 1.9127907752990723, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7678263187408447, + "num_tokens": 423693922.0, + "step": 17736 + }, + { + "epoch": 1.4626041065391275, + "grad_norm": 1.7983185052871704, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7777240872383118, + "num_tokens": 423727065.0, + "step": 17737 + }, + { + "epoch": 1.462686567164179, + "grad_norm": 1.7496980428695679, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.7978314757347107, + "num_tokens": 423761227.0, + "step": 17738 + }, + { + "epoch": 1.4627690277892307, + "grad_norm": 1.9402543306350708, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.7857216596603394, + "num_tokens": 423791687.0, + "step": 17739 + }, + { + "epoch": 1.4628514884142823, + "grad_norm": 2.0722970962524414, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.7863761186599731, + "num_tokens": 423821168.0, + "step": 17740 + }, + { + "epoch": 1.4629339490393338, + "grad_norm": 1.9872757196426392, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.7974698543548584, + "num_tokens": 423850747.0, + "step": 17741 + }, + { + "epoch": 1.4630164096643852, + "grad_norm": 1.8786873817443848, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7967308759689331, + "num_tokens": 423883530.0, + "step": 17742 + }, + { + "epoch": 1.4630988702894367, + "grad_norm": 1.9786347150802612, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7567948698997498, + "num_tokens": 423911451.0, + "step": 17743 + }, + { + "epoch": 1.4631813309144883, + "grad_norm": 1.9615447521209717, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7661963701248169, + "num_tokens": 423942342.0, + "step": 17744 + }, + { + "epoch": 1.46326379153954, + "grad_norm": 2.0890398025512695, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7465741038322449, + "num_tokens": 423973965.0, + "step": 17745 + }, + { + "epoch": 1.4633462521645915, + "grad_norm": 2.1835579872131348, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7310358881950378, + "num_tokens": 424000315.0, + "step": 17746 + }, + { + "epoch": 1.4634287127896428, + "grad_norm": 1.9338548183441162, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.733228325843811, + "num_tokens": 424033811.0, + "step": 17747 + }, + { + "epoch": 1.4635111734146946, + "grad_norm": 1.905530571937561, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7771297693252563, + "num_tokens": 424064874.0, + "step": 17748 + }, + { + "epoch": 1.463593634039746, + "grad_norm": 1.9896942377090454, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7622621059417725, + "num_tokens": 424094033.0, + "step": 17749 + }, + { + "epoch": 1.4636760946647975, + "grad_norm": 1.833784580230713, + "learning_rate": 1e-06, + "loss": 0.6221, + "mean_token_accuracy": 0.7945952415466309, + "num_tokens": 424124819.0, + "step": 17750 + }, + { + "epoch": 1.463758555289849, + "grad_norm": 1.6967819929122925, + "learning_rate": 1e-06, + "loss": 0.6095, + "mean_token_accuracy": 0.7965506911277771, + "num_tokens": 424157631.0, + "step": 17751 + }, + { + "epoch": 1.4638410159149007, + "grad_norm": 1.8365411758422852, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7394930124282837, + "num_tokens": 424193032.0, + "step": 17752 + }, + { + "epoch": 1.4639234765399523, + "grad_norm": 1.9216907024383545, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7525434494018555, + "num_tokens": 424223967.0, + "step": 17753 + }, + { + "epoch": 1.4640059371650036, + "grad_norm": 2.358377695083618, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7156202793121338, + "num_tokens": 424249629.0, + "step": 17754 + }, + { + "epoch": 1.4640883977900552, + "grad_norm": 1.7803916931152344, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.7595645189285278, + "num_tokens": 424283401.0, + "step": 17755 + }, + { + "epoch": 1.4641708584151067, + "grad_norm": 2.039585590362549, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7624784111976624, + "num_tokens": 424311802.0, + "step": 17756 + }, + { + "epoch": 1.4642533190401583, + "grad_norm": 2.1461215019226074, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7477156519889832, + "num_tokens": 424342023.0, + "step": 17757 + }, + { + "epoch": 1.46433577966521, + "grad_norm": 1.8177088499069214, + "learning_rate": 1e-06, + "loss": 0.6194, + "mean_token_accuracy": 0.7880517840385437, + "num_tokens": 424375781.0, + "step": 17758 + }, + { + "epoch": 1.4644182402902615, + "grad_norm": 2.179943561553955, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7395870685577393, + "num_tokens": 424400733.0, + "step": 17759 + }, + { + "epoch": 1.464500700915313, + "grad_norm": 1.9592660665512085, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7646192908287048, + "num_tokens": 424431410.0, + "step": 17760 + }, + { + "epoch": 1.4645831615403644, + "grad_norm": 2.0024445056915283, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7707598209381104, + "num_tokens": 424459840.0, + "step": 17761 + }, + { + "epoch": 1.464665622165416, + "grad_norm": 1.973341941833496, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7810192108154297, + "num_tokens": 424487460.0, + "step": 17762 + }, + { + "epoch": 1.4647480827904675, + "grad_norm": 2.4137213230133057, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7300623655319214, + "num_tokens": 424510988.0, + "step": 17763 + }, + { + "epoch": 1.464830543415519, + "grad_norm": 2.0785531997680664, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7449249029159546, + "num_tokens": 424541489.0, + "step": 17764 + }, + { + "epoch": 1.4649130040405707, + "grad_norm": 2.0416040420532227, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7464690208435059, + "num_tokens": 424570919.0, + "step": 17765 + }, + { + "epoch": 1.464995464665622, + "grad_norm": 1.9755347967147827, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7438802123069763, + "num_tokens": 424602078.0, + "step": 17766 + }, + { + "epoch": 1.4650779252906738, + "grad_norm": 1.9405423402786255, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.7939595580101013, + "num_tokens": 424629577.0, + "step": 17767 + }, + { + "epoch": 1.4651603859157252, + "grad_norm": 2.158334493637085, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7347114086151123, + "num_tokens": 424656560.0, + "step": 17768 + }, + { + "epoch": 1.4652428465407767, + "grad_norm": 2.062778949737549, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7640542984008789, + "num_tokens": 424684921.0, + "step": 17769 + }, + { + "epoch": 1.4653253071658283, + "grad_norm": 2.055478096008301, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7509132623672485, + "num_tokens": 424716910.0, + "step": 17770 + }, + { + "epoch": 1.46540776779088, + "grad_norm": 1.9460582733154297, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7719429731369019, + "num_tokens": 424747481.0, + "step": 17771 + }, + { + "epoch": 1.4654902284159315, + "grad_norm": 2.1684746742248535, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7669294476509094, + "num_tokens": 424773622.0, + "step": 17772 + }, + { + "epoch": 1.4655726890409828, + "grad_norm": 2.339348077774048, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7809809446334839, + "num_tokens": 424796788.0, + "step": 17773 + }, + { + "epoch": 1.4656551496660346, + "grad_norm": 2.0479507446289062, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7773265838623047, + "num_tokens": 424825144.0, + "step": 17774 + }, + { + "epoch": 1.465737610291086, + "grad_norm": 1.9853612184524536, + "learning_rate": 1e-06, + "loss": 0.6054, + "mean_token_accuracy": 0.8008125424385071, + "num_tokens": 424852769.0, + "step": 17775 + }, + { + "epoch": 1.4658200709161375, + "grad_norm": 2.069444417953491, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.7917566895484924, + "num_tokens": 424881683.0, + "step": 17776 + }, + { + "epoch": 1.465902531541189, + "grad_norm": 1.8748927116394043, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.7929027080535889, + "num_tokens": 424911800.0, + "step": 17777 + }, + { + "epoch": 1.4659849921662407, + "grad_norm": 1.8413137197494507, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.7950210571289062, + "num_tokens": 424945101.0, + "step": 17778 + }, + { + "epoch": 1.4660674527912922, + "grad_norm": 2.5059752464294434, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7477830052375793, + "num_tokens": 424965871.0, + "step": 17779 + }, + { + "epoch": 1.4661499134163436, + "grad_norm": 1.9012233018875122, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7168322801589966, + "num_tokens": 425000864.0, + "step": 17780 + }, + { + "epoch": 1.4662323740413952, + "grad_norm": 2.1365878582000732, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7712670564651489, + "num_tokens": 425028254.0, + "step": 17781 + }, + { + "epoch": 1.4663148346664467, + "grad_norm": 2.055485725402832, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7349587678909302, + "num_tokens": 425057808.0, + "step": 17782 + }, + { + "epoch": 1.4663972952914983, + "grad_norm": 1.9496463537216187, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.8013288974761963, + "num_tokens": 425093079.0, + "step": 17783 + }, + { + "epoch": 1.4664797559165499, + "grad_norm": 2.1087849140167236, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7427030801773071, + "num_tokens": 425120004.0, + "step": 17784 + }, + { + "epoch": 1.4665622165416015, + "grad_norm": 1.8584190607070923, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7712215781211853, + "num_tokens": 425150660.0, + "step": 17785 + }, + { + "epoch": 1.466644677166653, + "grad_norm": 2.2105064392089844, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7809535264968872, + "num_tokens": 425176375.0, + "step": 17786 + }, + { + "epoch": 1.4667271377917044, + "grad_norm": 2.385545253753662, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7603399753570557, + "num_tokens": 425198497.0, + "step": 17787 + }, + { + "epoch": 1.466809598416756, + "grad_norm": 1.7777791023254395, + "learning_rate": 1e-06, + "loss": 0.5726, + "mean_token_accuracy": 0.8084796071052551, + "num_tokens": 425231715.0, + "step": 17788 + }, + { + "epoch": 1.4668920590418075, + "grad_norm": 1.8806259632110596, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7630582451820374, + "num_tokens": 425264918.0, + "step": 17789 + }, + { + "epoch": 1.466974519666859, + "grad_norm": 1.8876256942749023, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7726073861122131, + "num_tokens": 425297287.0, + "step": 17790 + }, + { + "epoch": 1.4670569802919107, + "grad_norm": 2.1637916564941406, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7159096002578735, + "num_tokens": 425324651.0, + "step": 17791 + }, + { + "epoch": 1.467139440916962, + "grad_norm": 2.104800224304199, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7479280829429626, + "num_tokens": 425350863.0, + "step": 17792 + }, + { + "epoch": 1.4672219015420138, + "grad_norm": 1.8641761541366577, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.752556562423706, + "num_tokens": 425385051.0, + "step": 17793 + }, + { + "epoch": 1.4673043621670652, + "grad_norm": 2.212566614151001, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7640160918235779, + "num_tokens": 425411221.0, + "step": 17794 + }, + { + "epoch": 1.4673868227921167, + "grad_norm": 1.9591054916381836, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7782226800918579, + "num_tokens": 425438107.0, + "step": 17795 + }, + { + "epoch": 1.4674692834171683, + "grad_norm": 2.086120128631592, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7352133989334106, + "num_tokens": 425465982.0, + "step": 17796 + }, + { + "epoch": 1.4675517440422199, + "grad_norm": 2.192535161972046, + "learning_rate": 1e-06, + "loss": 0.6101, + "mean_token_accuracy": 0.7976394295692444, + "num_tokens": 425490094.0, + "step": 17797 + }, + { + "epoch": 1.4676342046672715, + "grad_norm": 1.7566726207733154, + "learning_rate": 1e-06, + "loss": 0.568, + "mean_token_accuracy": 0.8083252310752869, + "num_tokens": 425524066.0, + "step": 17798 + }, + { + "epoch": 1.4677166652923228, + "grad_norm": 1.9171690940856934, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7726212739944458, + "num_tokens": 425555306.0, + "step": 17799 + }, + { + "epoch": 1.4677991259173744, + "grad_norm": 2.2812135219573975, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7367864847183228, + "num_tokens": 425581012.0, + "step": 17800 + }, + { + "epoch": 1.467881586542426, + "grad_norm": 1.8472750186920166, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7498350143432617, + "num_tokens": 425617734.0, + "step": 17801 + }, + { + "epoch": 1.4679640471674775, + "grad_norm": 2.040634870529175, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7718569040298462, + "num_tokens": 425647170.0, + "step": 17802 + }, + { + "epoch": 1.468046507792529, + "grad_norm": 1.7668085098266602, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7599462866783142, + "num_tokens": 425687415.0, + "step": 17803 + }, + { + "epoch": 1.4681289684175807, + "grad_norm": 2.1160430908203125, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7552666664123535, + "num_tokens": 425714586.0, + "step": 17804 + }, + { + "epoch": 1.4682114290426322, + "grad_norm": 2.0427086353302, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7656769156455994, + "num_tokens": 425744077.0, + "step": 17805 + }, + { + "epoch": 1.4682938896676836, + "grad_norm": 1.8674875497817993, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7853248119354248, + "num_tokens": 425776107.0, + "step": 17806 + }, + { + "epoch": 1.4683763502927352, + "grad_norm": 1.7216911315917969, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7684615850448608, + "num_tokens": 425816785.0, + "step": 17807 + }, + { + "epoch": 1.4684588109177867, + "grad_norm": 1.7997901439666748, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7123504877090454, + "num_tokens": 425852659.0, + "step": 17808 + }, + { + "epoch": 1.4685412715428383, + "grad_norm": 2.0546045303344727, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7539355754852295, + "num_tokens": 425882314.0, + "step": 17809 + }, + { + "epoch": 1.4686237321678899, + "grad_norm": 1.8759955167770386, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.7770165205001831, + "num_tokens": 425911630.0, + "step": 17810 + }, + { + "epoch": 1.4687061927929412, + "grad_norm": 2.1102294921875, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7359791994094849, + "num_tokens": 425938415.0, + "step": 17811 + }, + { + "epoch": 1.468788653417993, + "grad_norm": 2.0117809772491455, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7479724884033203, + "num_tokens": 425970342.0, + "step": 17812 + }, + { + "epoch": 1.4688711140430444, + "grad_norm": 2.6980319023132324, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7639520168304443, + "num_tokens": 426000836.0, + "step": 17813 + }, + { + "epoch": 1.468953574668096, + "grad_norm": 1.6883357763290405, + "learning_rate": 1e-06, + "loss": 0.5609, + "mean_token_accuracy": 0.8081948161125183, + "num_tokens": 426036039.0, + "step": 17814 + }, + { + "epoch": 1.4690360352931475, + "grad_norm": 2.092047691345215, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.7592999935150146, + "num_tokens": 426062675.0, + "step": 17815 + }, + { + "epoch": 1.469118495918199, + "grad_norm": 1.7838516235351562, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7858726978302002, + "num_tokens": 426097126.0, + "step": 17816 + }, + { + "epoch": 1.4692009565432507, + "grad_norm": 2.1602888107299805, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.7939577698707581, + "num_tokens": 426124046.0, + "step": 17817 + }, + { + "epoch": 1.469283417168302, + "grad_norm": 1.9119908809661865, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.7933974862098694, + "num_tokens": 426155298.0, + "step": 17818 + }, + { + "epoch": 1.4693658777933536, + "grad_norm": 2.1047677993774414, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7822065949440002, + "num_tokens": 426186867.0, + "step": 17819 + }, + { + "epoch": 1.4694483384184052, + "grad_norm": 1.9407124519348145, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7685890197753906, + "num_tokens": 426217228.0, + "step": 17820 + }, + { + "epoch": 1.4695307990434567, + "grad_norm": 1.9215373992919922, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7706578969955444, + "num_tokens": 426250799.0, + "step": 17821 + }, + { + "epoch": 1.4696132596685083, + "grad_norm": 1.9851880073547363, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7661311626434326, + "num_tokens": 426284794.0, + "step": 17822 + }, + { + "epoch": 1.4696957202935599, + "grad_norm": 1.9614977836608887, + "learning_rate": 1e-06, + "loss": 0.6089, + "mean_token_accuracy": 0.7885209321975708, + "num_tokens": 426315172.0, + "step": 17823 + }, + { + "epoch": 1.4697781809186115, + "grad_norm": 1.7902820110321045, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.7974466681480408, + "num_tokens": 426350618.0, + "step": 17824 + }, + { + "epoch": 1.4698606415436628, + "grad_norm": 1.9877758026123047, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7610284090042114, + "num_tokens": 426379531.0, + "step": 17825 + }, + { + "epoch": 1.4699431021687144, + "grad_norm": 1.8801854848861694, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7678389549255371, + "num_tokens": 426411989.0, + "step": 17826 + }, + { + "epoch": 1.470025562793766, + "grad_norm": 1.944154977798462, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7703291773796082, + "num_tokens": 426444785.0, + "step": 17827 + }, + { + "epoch": 1.4701080234188175, + "grad_norm": 2.020963191986084, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7646583914756775, + "num_tokens": 426475617.0, + "step": 17828 + }, + { + "epoch": 1.470190484043869, + "grad_norm": 1.9834277629852295, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7080274820327759, + "num_tokens": 426506907.0, + "step": 17829 + }, + { + "epoch": 1.4702729446689207, + "grad_norm": 2.1454877853393555, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7545287013053894, + "num_tokens": 426533105.0, + "step": 17830 + }, + { + "epoch": 1.4703554052939722, + "grad_norm": 2.1729705333709717, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7584139704704285, + "num_tokens": 426560523.0, + "step": 17831 + }, + { + "epoch": 1.4704378659190236, + "grad_norm": 2.024961233139038, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7395266890525818, + "num_tokens": 426592181.0, + "step": 17832 + }, + { + "epoch": 1.4705203265440752, + "grad_norm": 1.9816936254501343, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.765898585319519, + "num_tokens": 426622535.0, + "step": 17833 + }, + { + "epoch": 1.4706027871691267, + "grad_norm": 1.834973692893982, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7811217308044434, + "num_tokens": 426654395.0, + "step": 17834 + }, + { + "epoch": 1.4706852477941783, + "grad_norm": 1.8829622268676758, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7677125930786133, + "num_tokens": 426691116.0, + "step": 17835 + }, + { + "epoch": 1.4707677084192299, + "grad_norm": 1.8941158056259155, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7682910561561584, + "num_tokens": 426725428.0, + "step": 17836 + }, + { + "epoch": 1.4708501690442812, + "grad_norm": 1.7665932178497314, + "learning_rate": 1e-06, + "loss": 0.6144, + "mean_token_accuracy": 0.799018144607544, + "num_tokens": 426756392.0, + "step": 17837 + }, + { + "epoch": 1.470932629669333, + "grad_norm": 1.960262656211853, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7503466606140137, + "num_tokens": 426789020.0, + "step": 17838 + }, + { + "epoch": 1.4710150902943844, + "grad_norm": 2.1908538341522217, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7590652704238892, + "num_tokens": 426817849.0, + "step": 17839 + }, + { + "epoch": 1.471097550919436, + "grad_norm": 1.979722499847412, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7698581218719482, + "num_tokens": 426848608.0, + "step": 17840 + }, + { + "epoch": 1.4711800115444875, + "grad_norm": 1.9020698070526123, + "learning_rate": 1e-06, + "loss": 0.5281, + "mean_token_accuracy": 0.8197343945503235, + "num_tokens": 426877155.0, + "step": 17841 + }, + { + "epoch": 1.471262472169539, + "grad_norm": 2.1317555904388428, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7329158782958984, + "num_tokens": 426902000.0, + "step": 17842 + }, + { + "epoch": 1.4713449327945907, + "grad_norm": 1.8422704935073853, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.764986515045166, + "num_tokens": 426936834.0, + "step": 17843 + }, + { + "epoch": 1.471427393419642, + "grad_norm": 1.9793949127197266, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7431704998016357, + "num_tokens": 426967543.0, + "step": 17844 + }, + { + "epoch": 1.4715098540446936, + "grad_norm": 1.8705004453659058, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7607376575469971, + "num_tokens": 427001438.0, + "step": 17845 + }, + { + "epoch": 1.4715923146697452, + "grad_norm": 1.981626033782959, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.770160973072052, + "num_tokens": 427031756.0, + "step": 17846 + }, + { + "epoch": 1.4716747752947967, + "grad_norm": 1.8846876621246338, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7720530033111572, + "num_tokens": 427062668.0, + "step": 17847 + }, + { + "epoch": 1.4717572359198483, + "grad_norm": 1.9044538736343384, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.7861617803573608, + "num_tokens": 427094730.0, + "step": 17848 + }, + { + "epoch": 1.4718396965448999, + "grad_norm": 2.4042534828186035, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7786219120025635, + "num_tokens": 427114798.0, + "step": 17849 + }, + { + "epoch": 1.4719221571699515, + "grad_norm": 1.8320987224578857, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7748950719833374, + "num_tokens": 427148919.0, + "step": 17850 + }, + { + "epoch": 1.4720046177950028, + "grad_norm": 2.0198774337768555, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7195289134979248, + "num_tokens": 427180529.0, + "step": 17851 + }, + { + "epoch": 1.4720870784200544, + "grad_norm": 2.0459043979644775, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7165181636810303, + "num_tokens": 427215101.0, + "step": 17852 + }, + { + "epoch": 1.472169539045106, + "grad_norm": 2.110873222351074, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7429404854774475, + "num_tokens": 427241650.0, + "step": 17853 + }, + { + "epoch": 1.4722519996701575, + "grad_norm": 2.1314680576324463, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.769371747970581, + "num_tokens": 427270907.0, + "step": 17854 + }, + { + "epoch": 1.472334460295209, + "grad_norm": 1.9481910467147827, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.7634786367416382, + "num_tokens": 427298388.0, + "step": 17855 + }, + { + "epoch": 1.4724169209202604, + "grad_norm": 2.03861665725708, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7623063325881958, + "num_tokens": 427327289.0, + "step": 17856 + }, + { + "epoch": 1.4724993815453122, + "grad_norm": 1.9478720426559448, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7886433601379395, + "num_tokens": 427355413.0, + "step": 17857 + }, + { + "epoch": 1.4725818421703636, + "grad_norm": 1.7441668510437012, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7741367816925049, + "num_tokens": 427388611.0, + "step": 17858 + }, + { + "epoch": 1.4726643027954152, + "grad_norm": 1.9102168083190918, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7538279294967651, + "num_tokens": 427420064.0, + "step": 17859 + }, + { + "epoch": 1.4727467634204667, + "grad_norm": 2.0724973678588867, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.7529783844947815, + "num_tokens": 427447358.0, + "step": 17860 + }, + { + "epoch": 1.4728292240455183, + "grad_norm": 1.996935248374939, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.798053503036499, + "num_tokens": 427476284.0, + "step": 17861 + }, + { + "epoch": 1.4729116846705699, + "grad_norm": 2.1312756538391113, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7503389716148376, + "num_tokens": 427503745.0, + "step": 17862 + }, + { + "epoch": 1.4729941452956212, + "grad_norm": 2.0070667266845703, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7515826225280762, + "num_tokens": 427534226.0, + "step": 17863 + }, + { + "epoch": 1.4730766059206728, + "grad_norm": 2.031419038772583, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7671257853507996, + "num_tokens": 427561977.0, + "step": 17864 + }, + { + "epoch": 1.4731590665457244, + "grad_norm": 2.045658588409424, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.784454345703125, + "num_tokens": 427588573.0, + "step": 17865 + }, + { + "epoch": 1.473241527170776, + "grad_norm": 2.2465903759002686, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.742242693901062, + "num_tokens": 427614608.0, + "step": 17866 + }, + { + "epoch": 1.4733239877958275, + "grad_norm": 1.891287922859192, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.8026463985443115, + "num_tokens": 427643012.0, + "step": 17867 + }, + { + "epoch": 1.473406448420879, + "grad_norm": 2.1348652839660645, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7482256889343262, + "num_tokens": 427668476.0, + "step": 17868 + }, + { + "epoch": 1.4734889090459307, + "grad_norm": 2.079841375350952, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7783923149108887, + "num_tokens": 427693947.0, + "step": 17869 + }, + { + "epoch": 1.473571369670982, + "grad_norm": 2.01425838470459, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7429106831550598, + "num_tokens": 427723707.0, + "step": 17870 + }, + { + "epoch": 1.4736538302960336, + "grad_norm": 2.182633638381958, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7475508451461792, + "num_tokens": 427749327.0, + "step": 17871 + }, + { + "epoch": 1.4737362909210852, + "grad_norm": 2.205972909927368, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7455141544342041, + "num_tokens": 427773707.0, + "step": 17872 + }, + { + "epoch": 1.4738187515461367, + "grad_norm": 2.064647912979126, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7265794277191162, + "num_tokens": 427806425.0, + "step": 17873 + }, + { + "epoch": 1.4739012121711883, + "grad_norm": 1.688862919807434, + "learning_rate": 1e-06, + "loss": 0.6126, + "mean_token_accuracy": 0.8017981648445129, + "num_tokens": 427843308.0, + "step": 17874 + }, + { + "epoch": 1.4739836727962399, + "grad_norm": 2.053297281265259, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.7905703783035278, + "num_tokens": 427870801.0, + "step": 17875 + }, + { + "epoch": 1.4740661334212914, + "grad_norm": 2.16398286819458, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7247380614280701, + "num_tokens": 427898629.0, + "step": 17876 + }, + { + "epoch": 1.4741485940463428, + "grad_norm": 2.010387420654297, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7291913032531738, + "num_tokens": 427929641.0, + "step": 17877 + }, + { + "epoch": 1.4742310546713944, + "grad_norm": 2.4604783058166504, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7462527751922607, + "num_tokens": 427951899.0, + "step": 17878 + }, + { + "epoch": 1.474313515296446, + "grad_norm": 1.9448626041412354, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7698227763175964, + "num_tokens": 427982700.0, + "step": 17879 + }, + { + "epoch": 1.4743959759214975, + "grad_norm": 2.2412736415863037, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.766643226146698, + "num_tokens": 428007863.0, + "step": 17880 + }, + { + "epoch": 1.474478436546549, + "grad_norm": 1.8252127170562744, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.7918800711631775, + "num_tokens": 428042284.0, + "step": 17881 + }, + { + "epoch": 1.4745608971716004, + "grad_norm": 2.1163291931152344, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.7875438332557678, + "num_tokens": 428069251.0, + "step": 17882 + }, + { + "epoch": 1.4746433577966522, + "grad_norm": 1.8467018604278564, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7704566717147827, + "num_tokens": 428102222.0, + "step": 17883 + }, + { + "epoch": 1.4747258184217036, + "grad_norm": 1.9067859649658203, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.7703447937965393, + "num_tokens": 428133904.0, + "step": 17884 + }, + { + "epoch": 1.4748082790467552, + "grad_norm": 2.14288592338562, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7573742866516113, + "num_tokens": 428156622.0, + "step": 17885 + }, + { + "epoch": 1.4748907396718067, + "grad_norm": 1.9417974948883057, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7475451231002808, + "num_tokens": 428189349.0, + "step": 17886 + }, + { + "epoch": 1.4749732002968583, + "grad_norm": 1.975463628768921, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.7969143986701965, + "num_tokens": 428218620.0, + "step": 17887 + }, + { + "epoch": 1.4750556609219099, + "grad_norm": 2.1050758361816406, + "learning_rate": 1e-06, + "loss": 0.7698, + "mean_token_accuracy": 0.7629642486572266, + "num_tokens": 428249168.0, + "step": 17888 + }, + { + "epoch": 1.4751381215469612, + "grad_norm": 2.0693023204803467, + "learning_rate": 1e-06, + "loss": 0.7483, + "mean_token_accuracy": 0.7619240880012512, + "num_tokens": 428277064.0, + "step": 17889 + }, + { + "epoch": 1.4752205821720128, + "grad_norm": 1.9339224100112915, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7735106945037842, + "num_tokens": 428308407.0, + "step": 17890 + }, + { + "epoch": 1.4753030427970644, + "grad_norm": 2.134248733520508, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7582592964172363, + "num_tokens": 428334614.0, + "step": 17891 + }, + { + "epoch": 1.475385503422116, + "grad_norm": 1.76878821849823, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.7747818231582642, + "num_tokens": 428368754.0, + "step": 17892 + }, + { + "epoch": 1.4754679640471675, + "grad_norm": 2.1964240074157715, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7506521344184875, + "num_tokens": 428395130.0, + "step": 17893 + }, + { + "epoch": 1.475550424672219, + "grad_norm": 2.0584897994995117, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.744888961315155, + "num_tokens": 428428277.0, + "step": 17894 + }, + { + "epoch": 1.4756328852972707, + "grad_norm": 1.7107290029525757, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7752714157104492, + "num_tokens": 428464556.0, + "step": 17895 + }, + { + "epoch": 1.475715345922322, + "grad_norm": 2.110107183456421, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.766318142414093, + "num_tokens": 428491451.0, + "step": 17896 + }, + { + "epoch": 1.4757978065473736, + "grad_norm": 2.088425397872925, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7702416181564331, + "num_tokens": 428519692.0, + "step": 17897 + }, + { + "epoch": 1.4758802671724252, + "grad_norm": 1.9863603115081787, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7489327192306519, + "num_tokens": 428548864.0, + "step": 17898 + }, + { + "epoch": 1.4759627277974767, + "grad_norm": 1.9658344984054565, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7090920805931091, + "num_tokens": 428583192.0, + "step": 17899 + }, + { + "epoch": 1.4760451884225283, + "grad_norm": 2.015716791152954, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.756716787815094, + "num_tokens": 428613584.0, + "step": 17900 + }, + { + "epoch": 1.4761276490475796, + "grad_norm": 1.8936678171157837, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7698850631713867, + "num_tokens": 428645165.0, + "step": 17901 + }, + { + "epoch": 1.4762101096726314, + "grad_norm": 2.0413405895233154, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7243117094039917, + "num_tokens": 428675493.0, + "step": 17902 + }, + { + "epoch": 1.4762925702976828, + "grad_norm": 2.061774969100952, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7629190683364868, + "num_tokens": 428701458.0, + "step": 17903 + }, + { + "epoch": 1.4763750309227344, + "grad_norm": 1.9780728816986084, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7642576694488525, + "num_tokens": 428733522.0, + "step": 17904 + }, + { + "epoch": 1.476457491547786, + "grad_norm": 2.21773362159729, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7514700889587402, + "num_tokens": 428757945.0, + "step": 17905 + }, + { + "epoch": 1.4765399521728375, + "grad_norm": 2.4069249629974365, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7486920356750488, + "num_tokens": 428778419.0, + "step": 17906 + }, + { + "epoch": 1.476622412797889, + "grad_norm": 2.015244245529175, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7669479846954346, + "num_tokens": 428806629.0, + "step": 17907 + }, + { + "epoch": 1.4767048734229404, + "grad_norm": 2.322291135787964, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7229340672492981, + "num_tokens": 428829353.0, + "step": 17908 + }, + { + "epoch": 1.476787334047992, + "grad_norm": 2.097395181655884, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7513554096221924, + "num_tokens": 428856127.0, + "step": 17909 + }, + { + "epoch": 1.4768697946730436, + "grad_norm": 1.8312292098999023, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.758755087852478, + "num_tokens": 428890847.0, + "step": 17910 + }, + { + "epoch": 1.4769522552980952, + "grad_norm": 2.0241453647613525, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7524453401565552, + "num_tokens": 428920187.0, + "step": 17911 + }, + { + "epoch": 1.4770347159231467, + "grad_norm": 2.3476991653442383, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7121827602386475, + "num_tokens": 428945022.0, + "step": 17912 + }, + { + "epoch": 1.4771171765481983, + "grad_norm": 1.9488873481750488, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.718791127204895, + "num_tokens": 428977396.0, + "step": 17913 + }, + { + "epoch": 1.4771996371732499, + "grad_norm": 1.9441206455230713, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7741559743881226, + "num_tokens": 429006045.0, + "step": 17914 + }, + { + "epoch": 1.4772820977983012, + "grad_norm": 2.0938608646392822, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7452525496482849, + "num_tokens": 429034436.0, + "step": 17915 + }, + { + "epoch": 1.4773645584233528, + "grad_norm": 2.173158645629883, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7583838105201721, + "num_tokens": 429058029.0, + "step": 17916 + }, + { + "epoch": 1.4774470190484044, + "grad_norm": 1.9067611694335938, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7202609777450562, + "num_tokens": 429094806.0, + "step": 17917 + }, + { + "epoch": 1.477529479673456, + "grad_norm": 1.7722041606903076, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.763209342956543, + "num_tokens": 429132824.0, + "step": 17918 + }, + { + "epoch": 1.4776119402985075, + "grad_norm": 2.0199859142303467, + "learning_rate": 1e-06, + "loss": 0.5527, + "mean_token_accuracy": 0.8005473017692566, + "num_tokens": 429158795.0, + "step": 17919 + }, + { + "epoch": 1.477694400923559, + "grad_norm": 1.8472001552581787, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7806304693222046, + "num_tokens": 429191992.0, + "step": 17920 + }, + { + "epoch": 1.4777768615486107, + "grad_norm": 1.769432544708252, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.7989014387130737, + "num_tokens": 429226163.0, + "step": 17921 + }, + { + "epoch": 1.477859322173662, + "grad_norm": 2.312255859375, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.7482444047927856, + "num_tokens": 429249416.0, + "step": 17922 + }, + { + "epoch": 1.4779417827987136, + "grad_norm": 1.6778925657272339, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.7890401482582092, + "num_tokens": 429288981.0, + "step": 17923 + }, + { + "epoch": 1.4780242434237651, + "grad_norm": 2.1487460136413574, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7419136762619019, + "num_tokens": 429314503.0, + "step": 17924 + }, + { + "epoch": 1.4781067040488167, + "grad_norm": 1.9441912174224854, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7575417757034302, + "num_tokens": 429343029.0, + "step": 17925 + }, + { + "epoch": 1.4781891646738683, + "grad_norm": 1.9915578365325928, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7159380912780762, + "num_tokens": 429373950.0, + "step": 17926 + }, + { + "epoch": 1.4782716252989196, + "grad_norm": 1.9107521772384644, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7384107112884521, + "num_tokens": 429407305.0, + "step": 17927 + }, + { + "epoch": 1.4783540859239714, + "grad_norm": 1.8820308446884155, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.7610629200935364, + "num_tokens": 429438222.0, + "step": 17928 + }, + { + "epoch": 1.4784365465490228, + "grad_norm": 1.9991296529769897, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7601443529129028, + "num_tokens": 429468768.0, + "step": 17929 + }, + { + "epoch": 1.4785190071740744, + "grad_norm": 1.8062487840652466, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7594118714332581, + "num_tokens": 429503560.0, + "step": 17930 + }, + { + "epoch": 1.478601467799126, + "grad_norm": 1.8001248836517334, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.7995429039001465, + "num_tokens": 429536366.0, + "step": 17931 + }, + { + "epoch": 1.4786839284241775, + "grad_norm": 1.8918397426605225, + "learning_rate": 1e-06, + "loss": 0.7762, + "mean_token_accuracy": 0.7496954202651978, + "num_tokens": 429566356.0, + "step": 17932 + }, + { + "epoch": 1.478766389049229, + "grad_norm": 2.0492358207702637, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7353119850158691, + "num_tokens": 429595233.0, + "step": 17933 + }, + { + "epoch": 1.4788488496742804, + "grad_norm": 2.1151235103607178, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.754045307636261, + "num_tokens": 429624472.0, + "step": 17934 + }, + { + "epoch": 1.478931310299332, + "grad_norm": 2.0116922855377197, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.736904501914978, + "num_tokens": 429656104.0, + "step": 17935 + }, + { + "epoch": 1.4790137709243836, + "grad_norm": 2.002601385116577, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7692919373512268, + "num_tokens": 429684546.0, + "step": 17936 + }, + { + "epoch": 1.4790962315494351, + "grad_norm": 1.6995272636413574, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.7992628812789917, + "num_tokens": 429724019.0, + "step": 17937 + }, + { + "epoch": 1.4791786921744867, + "grad_norm": 2.315887689590454, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7601711750030518, + "num_tokens": 429748088.0, + "step": 17938 + }, + { + "epoch": 1.4792611527995383, + "grad_norm": 1.8469598293304443, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.7786299586296082, + "num_tokens": 429779200.0, + "step": 17939 + }, + { + "epoch": 1.4793436134245899, + "grad_norm": 1.9041708707809448, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7758720517158508, + "num_tokens": 429813351.0, + "step": 17940 + }, + { + "epoch": 1.4794260740496412, + "grad_norm": 2.169196128845215, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7411584854125977, + "num_tokens": 429839953.0, + "step": 17941 + }, + { + "epoch": 1.4795085346746928, + "grad_norm": 1.850365161895752, + "learning_rate": 1e-06, + "loss": 0.7584, + "mean_token_accuracy": 0.7572852969169617, + "num_tokens": 429871020.0, + "step": 17942 + }, + { + "epoch": 1.4795909952997444, + "grad_norm": 2.1241092681884766, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7383480072021484, + "num_tokens": 429897910.0, + "step": 17943 + }, + { + "epoch": 1.479673455924796, + "grad_norm": 1.9642112255096436, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.8016282320022583, + "num_tokens": 429925542.0, + "step": 17944 + }, + { + "epoch": 1.4797559165498475, + "grad_norm": 2.135446786880493, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7475019693374634, + "num_tokens": 429954684.0, + "step": 17945 + }, + { + "epoch": 1.4798383771748989, + "grad_norm": 1.9117084741592407, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7727334499359131, + "num_tokens": 429982419.0, + "step": 17946 + }, + { + "epoch": 1.4799208377999507, + "grad_norm": 2.1219353675842285, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7449666261672974, + "num_tokens": 430009817.0, + "step": 17947 + }, + { + "epoch": 1.480003298425002, + "grad_norm": 1.9294052124023438, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7821229696273804, + "num_tokens": 430042550.0, + "step": 17948 + }, + { + "epoch": 1.4800857590500536, + "grad_norm": 1.8205262422561646, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7808964252471924, + "num_tokens": 430075427.0, + "step": 17949 + }, + { + "epoch": 1.4801682196751051, + "grad_norm": 2.1907589435577393, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7484309673309326, + "num_tokens": 430102958.0, + "step": 17950 + }, + { + "epoch": 1.4802506803001567, + "grad_norm": 1.92366361618042, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7242134809494019, + "num_tokens": 430137933.0, + "step": 17951 + }, + { + "epoch": 1.4803331409252083, + "grad_norm": 2.1718175411224365, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7552570700645447, + "num_tokens": 430164506.0, + "step": 17952 + }, + { + "epoch": 1.4804156015502596, + "grad_norm": 1.9395370483398438, + "learning_rate": 1e-06, + "loss": 0.6214, + "mean_token_accuracy": 0.8013797402381897, + "num_tokens": 430193235.0, + "step": 17953 + }, + { + "epoch": 1.4804980621753112, + "grad_norm": 1.87612783908844, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.770374596118927, + "num_tokens": 430226316.0, + "step": 17954 + }, + { + "epoch": 1.4805805228003628, + "grad_norm": 1.8298258781433105, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.7869789600372314, + "num_tokens": 430257453.0, + "step": 17955 + }, + { + "epoch": 1.4806629834254144, + "grad_norm": 1.8884549140930176, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7521489858627319, + "num_tokens": 430291827.0, + "step": 17956 + }, + { + "epoch": 1.480745444050466, + "grad_norm": 2.0643060207366943, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7375142574310303, + "num_tokens": 430324413.0, + "step": 17957 + }, + { + "epoch": 1.4808279046755175, + "grad_norm": 1.9965108633041382, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7426092028617859, + "num_tokens": 430355567.0, + "step": 17958 + }, + { + "epoch": 1.480910365300569, + "grad_norm": 2.0533406734466553, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7615433931350708, + "num_tokens": 430382351.0, + "step": 17959 + }, + { + "epoch": 1.4809928259256204, + "grad_norm": 1.7180507183074951, + "learning_rate": 1e-06, + "loss": 0.5762, + "mean_token_accuracy": 0.8101613521575928, + "num_tokens": 430417682.0, + "step": 17960 + }, + { + "epoch": 1.481075286550672, + "grad_norm": 2.0498909950256348, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7582759857177734, + "num_tokens": 430444736.0, + "step": 17961 + }, + { + "epoch": 1.4811577471757236, + "grad_norm": 1.951517939567566, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7473883032798767, + "num_tokens": 430476288.0, + "step": 17962 + }, + { + "epoch": 1.4812402078007751, + "grad_norm": 1.763939619064331, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.7899415493011475, + "num_tokens": 430508131.0, + "step": 17963 + }, + { + "epoch": 1.4813226684258267, + "grad_norm": 2.3994839191436768, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7500782012939453, + "num_tokens": 430531763.0, + "step": 17964 + }, + { + "epoch": 1.4814051290508783, + "grad_norm": 2.1724953651428223, + "learning_rate": 1e-06, + "loss": 0.734, + "mean_token_accuracy": 0.7548254132270813, + "num_tokens": 430557788.0, + "step": 17965 + }, + { + "epoch": 1.4814875896759299, + "grad_norm": 2.220100164413452, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7469227313995361, + "num_tokens": 430584342.0, + "step": 17966 + }, + { + "epoch": 1.4815700503009812, + "grad_norm": 1.8275474309921265, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7561542987823486, + "num_tokens": 430620025.0, + "step": 17967 + }, + { + "epoch": 1.4816525109260328, + "grad_norm": 1.8970304727554321, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.733712375164032, + "num_tokens": 430652563.0, + "step": 17968 + }, + { + "epoch": 1.4817349715510844, + "grad_norm": 2.093797206878662, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7465637922286987, + "num_tokens": 430681906.0, + "step": 17969 + }, + { + "epoch": 1.481817432176136, + "grad_norm": 2.0876781940460205, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.775127112865448, + "num_tokens": 430711077.0, + "step": 17970 + }, + { + "epoch": 1.4818998928011875, + "grad_norm": 1.9973925352096558, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7770115733146667, + "num_tokens": 430741606.0, + "step": 17971 + }, + { + "epoch": 1.4819823534262389, + "grad_norm": 2.1871206760406494, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7585245370864868, + "num_tokens": 430770766.0, + "step": 17972 + }, + { + "epoch": 1.4820648140512906, + "grad_norm": 1.9246878623962402, + "learning_rate": 1e-06, + "loss": 0.5957, + "mean_token_accuracy": 0.8049588203430176, + "num_tokens": 430803190.0, + "step": 17973 + }, + { + "epoch": 1.482147274676342, + "grad_norm": 2.058992862701416, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7713395357131958, + "num_tokens": 430829436.0, + "step": 17974 + }, + { + "epoch": 1.4822297353013936, + "grad_norm": 1.7914870977401733, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7859264016151428, + "num_tokens": 430864635.0, + "step": 17975 + }, + { + "epoch": 1.4823121959264451, + "grad_norm": 1.908028244972229, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7545163631439209, + "num_tokens": 430895340.0, + "step": 17976 + }, + { + "epoch": 1.4823946565514967, + "grad_norm": 2.0960166454315186, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7598035931587219, + "num_tokens": 430922316.0, + "step": 17977 + }, + { + "epoch": 1.4824771171765483, + "grad_norm": 2.20100474357605, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7108625173568726, + "num_tokens": 430947335.0, + "step": 17978 + }, + { + "epoch": 1.4825595778015996, + "grad_norm": 1.8351668119430542, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7564484477043152, + "num_tokens": 430982061.0, + "step": 17979 + }, + { + "epoch": 1.4826420384266512, + "grad_norm": 1.906977891921997, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7410421371459961, + "num_tokens": 431013589.0, + "step": 17980 + }, + { + "epoch": 1.4827244990517028, + "grad_norm": 1.956146240234375, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7536978721618652, + "num_tokens": 431046889.0, + "step": 17981 + }, + { + "epoch": 1.4828069596767544, + "grad_norm": 2.0154330730438232, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7500847578048706, + "num_tokens": 431075541.0, + "step": 17982 + }, + { + "epoch": 1.482889420301806, + "grad_norm": 1.695337176322937, + "learning_rate": 1e-06, + "loss": 0.5479, + "mean_token_accuracy": 0.8156399726867676, + "num_tokens": 431111689.0, + "step": 17983 + }, + { + "epoch": 1.4829718809268575, + "grad_norm": 2.1872012615203857, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7619480490684509, + "num_tokens": 431136662.0, + "step": 17984 + }, + { + "epoch": 1.483054341551909, + "grad_norm": 1.9535691738128662, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7275950908660889, + "num_tokens": 431168975.0, + "step": 17985 + }, + { + "epoch": 1.4831368021769604, + "grad_norm": 1.832248330116272, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7610156536102295, + "num_tokens": 431203788.0, + "step": 17986 + }, + { + "epoch": 1.483219262802012, + "grad_norm": 1.8320703506469727, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.7918771505355835, + "num_tokens": 431235452.0, + "step": 17987 + }, + { + "epoch": 1.4833017234270636, + "grad_norm": 2.1940417289733887, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.7865006923675537, + "num_tokens": 431260218.0, + "step": 17988 + }, + { + "epoch": 1.4833841840521151, + "grad_norm": 1.9007099866867065, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7731125354766846, + "num_tokens": 431291697.0, + "step": 17989 + }, + { + "epoch": 1.4834666446771667, + "grad_norm": 1.679303526878357, + "learning_rate": 1e-06, + "loss": 0.5921, + "mean_token_accuracy": 0.7999757528305054, + "num_tokens": 431329517.0, + "step": 17990 + }, + { + "epoch": 1.483549105302218, + "grad_norm": 2.00130033493042, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7721902132034302, + "num_tokens": 431357703.0, + "step": 17991 + }, + { + "epoch": 1.4836315659272699, + "grad_norm": 2.1773569583892822, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7373394966125488, + "num_tokens": 431385508.0, + "step": 17992 + }, + { + "epoch": 1.4837140265523212, + "grad_norm": 1.920779824256897, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.7849329710006714, + "num_tokens": 431413179.0, + "step": 17993 + }, + { + "epoch": 1.4837964871773728, + "grad_norm": 2.030179977416992, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.732670247554779, + "num_tokens": 431442427.0, + "step": 17994 + }, + { + "epoch": 1.4838789478024244, + "grad_norm": 2.0900633335113525, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7571877241134644, + "num_tokens": 431469673.0, + "step": 17995 + }, + { + "epoch": 1.483961408427476, + "grad_norm": 1.9772881269454956, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7867929935455322, + "num_tokens": 431497819.0, + "step": 17996 + }, + { + "epoch": 1.4840438690525275, + "grad_norm": 2.1508970260620117, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7589175701141357, + "num_tokens": 431528270.0, + "step": 17997 + }, + { + "epoch": 1.4841263296775788, + "grad_norm": 2.039626359939575, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7742515802383423, + "num_tokens": 431557575.0, + "step": 17998 + }, + { + "epoch": 1.4842087903026304, + "grad_norm": 1.874936819076538, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7487000226974487, + "num_tokens": 431593528.0, + "step": 17999 + }, + { + "epoch": 1.484291250927682, + "grad_norm": 1.9037981033325195, + "learning_rate": 1e-06, + "loss": 0.6239, + "mean_token_accuracy": 0.7946301698684692, + "num_tokens": 431622099.0, + "step": 18000 + }, + { + "epoch": 1.4843737115527336, + "grad_norm": 1.810150146484375, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7584166526794434, + "num_tokens": 431655705.0, + "step": 18001 + }, + { + "epoch": 1.4844561721777851, + "grad_norm": 1.8666434288024902, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.785707414150238, + "num_tokens": 431687919.0, + "step": 18002 + }, + { + "epoch": 1.4845386328028367, + "grad_norm": 1.7489523887634277, + "learning_rate": 1e-06, + "loss": 0.578, + "mean_token_accuracy": 0.809528112411499, + "num_tokens": 431722515.0, + "step": 18003 + }, + { + "epoch": 1.4846210934278883, + "grad_norm": 1.751771092414856, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7713749408721924, + "num_tokens": 431753634.0, + "step": 18004 + }, + { + "epoch": 1.4847035540529396, + "grad_norm": 2.2833614349365234, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7263105511665344, + "num_tokens": 431779166.0, + "step": 18005 + }, + { + "epoch": 1.4847860146779912, + "grad_norm": 1.9459736347198486, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.8054822683334351, + "num_tokens": 431809404.0, + "step": 18006 + }, + { + "epoch": 1.4848684753030428, + "grad_norm": 1.8013498783111572, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.795275092124939, + "num_tokens": 431839935.0, + "step": 18007 + }, + { + "epoch": 1.4849509359280944, + "grad_norm": 2.123436212539673, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7306064963340759, + "num_tokens": 431867568.0, + "step": 18008 + }, + { + "epoch": 1.485033396553146, + "grad_norm": 1.9001630544662476, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7520686388015747, + "num_tokens": 431899094.0, + "step": 18009 + }, + { + "epoch": 1.4851158571781973, + "grad_norm": 1.9032390117645264, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.7932757139205933, + "num_tokens": 431931011.0, + "step": 18010 + }, + { + "epoch": 1.485198317803249, + "grad_norm": 1.88112211227417, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7110220193862915, + "num_tokens": 431968643.0, + "step": 18011 + }, + { + "epoch": 1.4852807784283004, + "grad_norm": 2.095076322555542, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7777896523475647, + "num_tokens": 432000726.0, + "step": 18012 + }, + { + "epoch": 1.485363239053352, + "grad_norm": 1.990969181060791, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7758649587631226, + "num_tokens": 432032489.0, + "step": 18013 + }, + { + "epoch": 1.4854456996784036, + "grad_norm": 2.1495091915130615, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.75657057762146, + "num_tokens": 432059874.0, + "step": 18014 + }, + { + "epoch": 1.4855281603034551, + "grad_norm": 1.8124033212661743, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7636986374855042, + "num_tokens": 432096288.0, + "step": 18015 + }, + { + "epoch": 1.4856106209285067, + "grad_norm": 2.017153263092041, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7741685509681702, + "num_tokens": 432127551.0, + "step": 18016 + }, + { + "epoch": 1.485693081553558, + "grad_norm": 1.9761512279510498, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7677191495895386, + "num_tokens": 432156031.0, + "step": 18017 + }, + { + "epoch": 1.4857755421786096, + "grad_norm": 2.2712440490722656, + "learning_rate": 1e-06, + "loss": 0.7498, + "mean_token_accuracy": 0.7563095688819885, + "num_tokens": 432182299.0, + "step": 18018 + }, + { + "epoch": 1.4858580028036612, + "grad_norm": 1.8966268301010132, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7612605094909668, + "num_tokens": 432217427.0, + "step": 18019 + }, + { + "epoch": 1.4859404634287128, + "grad_norm": 2.007418632507324, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7816669940948486, + "num_tokens": 432249120.0, + "step": 18020 + }, + { + "epoch": 1.4860229240537643, + "grad_norm": 1.9132061004638672, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7419261932373047, + "num_tokens": 432284002.0, + "step": 18021 + }, + { + "epoch": 1.486105384678816, + "grad_norm": 2.285954236984253, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7581519484519958, + "num_tokens": 432310789.0, + "step": 18022 + }, + { + "epoch": 1.4861878453038675, + "grad_norm": 1.9240305423736572, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.7981973886489868, + "num_tokens": 432339422.0, + "step": 18023 + }, + { + "epoch": 1.4862703059289188, + "grad_norm": 2.0602009296417236, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7505587935447693, + "num_tokens": 432367737.0, + "step": 18024 + }, + { + "epoch": 1.4863527665539704, + "grad_norm": 1.9316548109054565, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7654885053634644, + "num_tokens": 432399803.0, + "step": 18025 + }, + { + "epoch": 1.486435227179022, + "grad_norm": 2.082216739654541, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7169034481048584, + "num_tokens": 432426682.0, + "step": 18026 + }, + { + "epoch": 1.4865176878040736, + "grad_norm": 2.2151877880096436, + "learning_rate": 1e-06, + "loss": 0.7877, + "mean_token_accuracy": 0.7446852922439575, + "num_tokens": 432452076.0, + "step": 18027 + }, + { + "epoch": 1.4866001484291251, + "grad_norm": 1.8952304124832153, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7844426035881042, + "num_tokens": 432485492.0, + "step": 18028 + }, + { + "epoch": 1.4866826090541767, + "grad_norm": 2.155020236968994, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7579285502433777, + "num_tokens": 432510995.0, + "step": 18029 + }, + { + "epoch": 1.4867650696792283, + "grad_norm": 1.9931198358535767, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7297670841217041, + "num_tokens": 432543351.0, + "step": 18030 + }, + { + "epoch": 1.4868475303042796, + "grad_norm": 2.17482852935791, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7096397876739502, + "num_tokens": 432571362.0, + "step": 18031 + }, + { + "epoch": 1.4869299909293312, + "grad_norm": 1.9092392921447754, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7708209753036499, + "num_tokens": 432604025.0, + "step": 18032 + }, + { + "epoch": 1.4870124515543828, + "grad_norm": 1.8908717632293701, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.7812178730964661, + "num_tokens": 432637822.0, + "step": 18033 + }, + { + "epoch": 1.4870949121794343, + "grad_norm": 1.9566303491592407, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7430646419525146, + "num_tokens": 432666767.0, + "step": 18034 + }, + { + "epoch": 1.487177372804486, + "grad_norm": 2.155928373336792, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7342184782028198, + "num_tokens": 432694571.0, + "step": 18035 + }, + { + "epoch": 1.4872598334295373, + "grad_norm": 2.116750717163086, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.771926999092102, + "num_tokens": 432723239.0, + "step": 18036 + }, + { + "epoch": 1.487342294054589, + "grad_norm": 1.7065739631652832, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7561736106872559, + "num_tokens": 432759259.0, + "step": 18037 + }, + { + "epoch": 1.4874247546796404, + "grad_norm": 1.7408264875411987, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7509347796440125, + "num_tokens": 432795387.0, + "step": 18038 + }, + { + "epoch": 1.487507215304692, + "grad_norm": 1.8940281867980957, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7285792231559753, + "num_tokens": 432829393.0, + "step": 18039 + }, + { + "epoch": 1.4875896759297436, + "grad_norm": 2.039463520050049, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7482032775878906, + "num_tokens": 432858448.0, + "step": 18040 + }, + { + "epoch": 1.4876721365547951, + "grad_norm": 1.9002333879470825, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.7855527400970459, + "num_tokens": 432890792.0, + "step": 18041 + }, + { + "epoch": 1.4877545971798467, + "grad_norm": 1.7729148864746094, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.7937459945678711, + "num_tokens": 432921114.0, + "step": 18042 + }, + { + "epoch": 1.487837057804898, + "grad_norm": 2.0677413940429688, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7347078919410706, + "num_tokens": 432952853.0, + "step": 18043 + }, + { + "epoch": 1.4879195184299496, + "grad_norm": 2.0523159503936768, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7447534203529358, + "num_tokens": 432981118.0, + "step": 18044 + }, + { + "epoch": 1.4880019790550012, + "grad_norm": 1.7807420492172241, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7802860140800476, + "num_tokens": 433019657.0, + "step": 18045 + }, + { + "epoch": 1.4880844396800528, + "grad_norm": 1.9910506010055542, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7607795000076294, + "num_tokens": 433050667.0, + "step": 18046 + }, + { + "epoch": 1.4881669003051043, + "grad_norm": 2.2092409133911133, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.75812166929245, + "num_tokens": 433076361.0, + "step": 18047 + }, + { + "epoch": 1.488249360930156, + "grad_norm": 2.3885207176208496, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7802500128746033, + "num_tokens": 433098413.0, + "step": 18048 + }, + { + "epoch": 1.4883318215552075, + "grad_norm": 1.9251724481582642, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7804131507873535, + "num_tokens": 433129857.0, + "step": 18049 + }, + { + "epoch": 1.4884142821802588, + "grad_norm": 1.7472409009933472, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.7883298397064209, + "num_tokens": 433166840.0, + "step": 18050 + }, + { + "epoch": 1.4884967428053104, + "grad_norm": 1.8913819789886475, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.7835014462471008, + "num_tokens": 433198538.0, + "step": 18051 + }, + { + "epoch": 1.488579203430362, + "grad_norm": 2.013653039932251, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7171230316162109, + "num_tokens": 433231411.0, + "step": 18052 + }, + { + "epoch": 1.4886616640554136, + "grad_norm": 1.5638951063156128, + "learning_rate": 1e-06, + "loss": 0.6145, + "mean_token_accuracy": 0.7909284830093384, + "num_tokens": 433268934.0, + "step": 18053 + }, + { + "epoch": 1.4887441246804651, + "grad_norm": 2.1126809120178223, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7676764130592346, + "num_tokens": 433295161.0, + "step": 18054 + }, + { + "epoch": 1.4888265853055165, + "grad_norm": 2.0327107906341553, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7572870850563049, + "num_tokens": 433323164.0, + "step": 18055 + }, + { + "epoch": 1.4889090459305683, + "grad_norm": 1.9523036479949951, + "learning_rate": 1e-06, + "loss": 0.7886, + "mean_token_accuracy": 0.7645788192749023, + "num_tokens": 433354869.0, + "step": 18056 + }, + { + "epoch": 1.4889915065556196, + "grad_norm": 1.9715930223464966, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7487736344337463, + "num_tokens": 433385663.0, + "step": 18057 + }, + { + "epoch": 1.4890739671806712, + "grad_norm": 2.6281208992004395, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7547680735588074, + "num_tokens": 433404294.0, + "step": 18058 + }, + { + "epoch": 1.4891564278057228, + "grad_norm": 2.015657901763916, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.734099268913269, + "num_tokens": 433434708.0, + "step": 18059 + }, + { + "epoch": 1.4892388884307743, + "grad_norm": 2.1066410541534424, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7615869641304016, + "num_tokens": 433464153.0, + "step": 18060 + }, + { + "epoch": 1.489321349055826, + "grad_norm": 2.0729119777679443, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7721071243286133, + "num_tokens": 433492025.0, + "step": 18061 + }, + { + "epoch": 1.4894038096808773, + "grad_norm": 1.8474524021148682, + "learning_rate": 1e-06, + "loss": 0.5793, + "mean_token_accuracy": 0.809001088142395, + "num_tokens": 433520994.0, + "step": 18062 + }, + { + "epoch": 1.4894862703059288, + "grad_norm": 1.925763726234436, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7755333185195923, + "num_tokens": 433551134.0, + "step": 18063 + }, + { + "epoch": 1.4895687309309804, + "grad_norm": 2.317751884460449, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7702881097793579, + "num_tokens": 433574447.0, + "step": 18064 + }, + { + "epoch": 1.489651191556032, + "grad_norm": 2.0076730251312256, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7646419405937195, + "num_tokens": 433605290.0, + "step": 18065 + }, + { + "epoch": 1.4897336521810836, + "grad_norm": 1.662653923034668, + "learning_rate": 1e-06, + "loss": 0.6214, + "mean_token_accuracy": 0.7946256399154663, + "num_tokens": 433639114.0, + "step": 18066 + }, + { + "epoch": 1.4898161128061351, + "grad_norm": 2.000077724456787, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7340430021286011, + "num_tokens": 433671423.0, + "step": 18067 + }, + { + "epoch": 1.4898985734311867, + "grad_norm": 1.8851019144058228, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.7960193753242493, + "num_tokens": 433703547.0, + "step": 18068 + }, + { + "epoch": 1.489981034056238, + "grad_norm": 1.9781054258346558, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7776476144790649, + "num_tokens": 433734142.0, + "step": 18069 + }, + { + "epoch": 1.4900634946812896, + "grad_norm": 1.906083106994629, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7579708099365234, + "num_tokens": 433766252.0, + "step": 18070 + }, + { + "epoch": 1.4901459553063412, + "grad_norm": 1.7867425680160522, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.7905440926551819, + "num_tokens": 433803765.0, + "step": 18071 + }, + { + "epoch": 1.4902284159313928, + "grad_norm": 1.9289048910140991, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.7992976307868958, + "num_tokens": 433834241.0, + "step": 18072 + }, + { + "epoch": 1.4903108765564443, + "grad_norm": 2.11529803276062, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7680632472038269, + "num_tokens": 433858948.0, + "step": 18073 + }, + { + "epoch": 1.490393337181496, + "grad_norm": 1.9362688064575195, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7463350296020508, + "num_tokens": 433891418.0, + "step": 18074 + }, + { + "epoch": 1.4904757978065475, + "grad_norm": 1.9368364810943604, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7655085325241089, + "num_tokens": 433923187.0, + "step": 18075 + }, + { + "epoch": 1.4905582584315988, + "grad_norm": 2.267026901245117, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7540701031684875, + "num_tokens": 433948062.0, + "step": 18076 + }, + { + "epoch": 1.4906407190566504, + "grad_norm": 2.206132173538208, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7728506326675415, + "num_tokens": 433971419.0, + "step": 18077 + }, + { + "epoch": 1.490723179681702, + "grad_norm": 1.7359939813613892, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.740104079246521, + "num_tokens": 434009370.0, + "step": 18078 + }, + { + "epoch": 1.4908056403067536, + "grad_norm": 1.961625099182129, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7665685415267944, + "num_tokens": 434039231.0, + "step": 18079 + }, + { + "epoch": 1.4908881009318051, + "grad_norm": 1.8569364547729492, + "learning_rate": 1e-06, + "loss": 0.5618, + "mean_token_accuracy": 0.8238987326622009, + "num_tokens": 434069047.0, + "step": 18080 + }, + { + "epoch": 1.4909705615568565, + "grad_norm": 2.260275363922119, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7679991722106934, + "num_tokens": 434094887.0, + "step": 18081 + }, + { + "epoch": 1.4910530221819083, + "grad_norm": 2.320823907852173, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7603921294212341, + "num_tokens": 434118452.0, + "step": 18082 + }, + { + "epoch": 1.4911354828069596, + "grad_norm": 2.0899300575256348, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7446681261062622, + "num_tokens": 434146887.0, + "step": 18083 + }, + { + "epoch": 1.4912179434320112, + "grad_norm": 2.345214366912842, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.733626127243042, + "num_tokens": 434171495.0, + "step": 18084 + }, + { + "epoch": 1.4913004040570628, + "grad_norm": 1.9064795970916748, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.7919744849205017, + "num_tokens": 434203909.0, + "step": 18085 + }, + { + "epoch": 1.4913828646821143, + "grad_norm": 2.158665180206299, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7739063501358032, + "num_tokens": 434229080.0, + "step": 18086 + }, + { + "epoch": 1.491465325307166, + "grad_norm": 2.0275442600250244, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7733111381530762, + "num_tokens": 434256701.0, + "step": 18087 + }, + { + "epoch": 1.4915477859322173, + "grad_norm": 1.9946136474609375, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7689359188079834, + "num_tokens": 434284758.0, + "step": 18088 + }, + { + "epoch": 1.4916302465572688, + "grad_norm": 2.2481236457824707, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7465533018112183, + "num_tokens": 434310061.0, + "step": 18089 + }, + { + "epoch": 1.4917127071823204, + "grad_norm": 1.806212067604065, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.777840256690979, + "num_tokens": 434343465.0, + "step": 18090 + }, + { + "epoch": 1.491795167807372, + "grad_norm": 1.9921413660049438, + "learning_rate": 1e-06, + "loss": 0.5758, + "mean_token_accuracy": 0.8049110174179077, + "num_tokens": 434369545.0, + "step": 18091 + }, + { + "epoch": 1.4918776284324236, + "grad_norm": 1.9799315929412842, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.7769773602485657, + "num_tokens": 434396567.0, + "step": 18092 + }, + { + "epoch": 1.4919600890574751, + "grad_norm": 1.9035766124725342, + "learning_rate": 1e-06, + "loss": 0.5165, + "mean_token_accuracy": 0.8230533599853516, + "num_tokens": 434424516.0, + "step": 18093 + }, + { + "epoch": 1.4920425496825267, + "grad_norm": 2.0798840522766113, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7650667428970337, + "num_tokens": 434452881.0, + "step": 18094 + }, + { + "epoch": 1.492125010307578, + "grad_norm": 2.1043803691864014, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7268801927566528, + "num_tokens": 434481876.0, + "step": 18095 + }, + { + "epoch": 1.4922074709326296, + "grad_norm": 2.3446855545043945, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7765542268753052, + "num_tokens": 434504641.0, + "step": 18096 + }, + { + "epoch": 1.4922899315576812, + "grad_norm": 2.094320297241211, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7681251764297485, + "num_tokens": 434530906.0, + "step": 18097 + }, + { + "epoch": 1.4923723921827328, + "grad_norm": 1.870661735534668, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7889397144317627, + "num_tokens": 434564965.0, + "step": 18098 + }, + { + "epoch": 1.4924548528077843, + "grad_norm": 2.0121870040893555, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7795267105102539, + "num_tokens": 434591211.0, + "step": 18099 + }, + { + "epoch": 1.4925373134328357, + "grad_norm": 1.965110182762146, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7743133902549744, + "num_tokens": 434622047.0, + "step": 18100 + }, + { + "epoch": 1.4926197740578875, + "grad_norm": 1.9135713577270508, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.7881107330322266, + "num_tokens": 434651945.0, + "step": 18101 + }, + { + "epoch": 1.4927022346829388, + "grad_norm": 2.274820566177368, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7350775599479675, + "num_tokens": 434678306.0, + "step": 18102 + }, + { + "epoch": 1.4927846953079904, + "grad_norm": 1.945074439048767, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7364621162414551, + "num_tokens": 434713541.0, + "step": 18103 + }, + { + "epoch": 1.492867155933042, + "grad_norm": 2.286341667175293, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7601920962333679, + "num_tokens": 434737124.0, + "step": 18104 + }, + { + "epoch": 1.4929496165580936, + "grad_norm": 2.1190452575683594, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7511072158813477, + "num_tokens": 434764031.0, + "step": 18105 + }, + { + "epoch": 1.4930320771831451, + "grad_norm": 1.9503201246261597, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.7479655742645264, + "num_tokens": 434796643.0, + "step": 18106 + }, + { + "epoch": 1.4931145378081965, + "grad_norm": 2.1606805324554443, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7743195295333862, + "num_tokens": 434822095.0, + "step": 18107 + }, + { + "epoch": 1.493196998433248, + "grad_norm": 2.007084846496582, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7768303751945496, + "num_tokens": 434851825.0, + "step": 18108 + }, + { + "epoch": 1.4932794590582996, + "grad_norm": 1.9727513790130615, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.7767685651779175, + "num_tokens": 434883347.0, + "step": 18109 + }, + { + "epoch": 1.4933619196833512, + "grad_norm": 1.9339901208877563, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7363610863685608, + "num_tokens": 434912760.0, + "step": 18110 + }, + { + "epoch": 1.4934443803084028, + "grad_norm": 2.111159086227417, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7662616968154907, + "num_tokens": 434938726.0, + "step": 18111 + }, + { + "epoch": 1.4935268409334543, + "grad_norm": 1.915270209312439, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7830029129981995, + "num_tokens": 434970010.0, + "step": 18112 + }, + { + "epoch": 1.493609301558506, + "grad_norm": 1.897148847579956, + "learning_rate": 1e-06, + "loss": 0.596, + "mean_token_accuracy": 0.8057211637496948, + "num_tokens": 435003040.0, + "step": 18113 + }, + { + "epoch": 1.4936917621835573, + "grad_norm": 2.358729839324951, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.7579086422920227, + "num_tokens": 435026196.0, + "step": 18114 + }, + { + "epoch": 1.4937742228086088, + "grad_norm": 2.027719497680664, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.776524007320404, + "num_tokens": 435053449.0, + "step": 18115 + }, + { + "epoch": 1.4938566834336604, + "grad_norm": 1.6366145610809326, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7904071807861328, + "num_tokens": 435094763.0, + "step": 18116 + }, + { + "epoch": 1.493939144058712, + "grad_norm": 1.8381991386413574, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7340209484100342, + "num_tokens": 435130547.0, + "step": 18117 + }, + { + "epoch": 1.4940216046837635, + "grad_norm": 1.8161259889602661, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7740969657897949, + "num_tokens": 435163656.0, + "step": 18118 + }, + { + "epoch": 1.4941040653088151, + "grad_norm": 1.9821465015411377, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7808933258056641, + "num_tokens": 435190159.0, + "step": 18119 + }, + { + "epoch": 1.4941865259338667, + "grad_norm": 2.2337470054626465, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7571268081665039, + "num_tokens": 435216956.0, + "step": 18120 + }, + { + "epoch": 1.494268986558918, + "grad_norm": 1.7878904342651367, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.7635188102722168, + "num_tokens": 435250766.0, + "step": 18121 + }, + { + "epoch": 1.4943514471839696, + "grad_norm": 2.0859293937683105, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7343671321868896, + "num_tokens": 435280533.0, + "step": 18122 + }, + { + "epoch": 1.4944339078090212, + "grad_norm": 2.0356550216674805, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7226030826568604, + "num_tokens": 435309145.0, + "step": 18123 + }, + { + "epoch": 1.4945163684340728, + "grad_norm": 2.1236166954040527, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7614154815673828, + "num_tokens": 435335907.0, + "step": 18124 + }, + { + "epoch": 1.4945988290591243, + "grad_norm": 2.144293785095215, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7413800954818726, + "num_tokens": 435362172.0, + "step": 18125 + }, + { + "epoch": 1.4946812896841757, + "grad_norm": 1.901764988899231, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7731201648712158, + "num_tokens": 435394467.0, + "step": 18126 + }, + { + "epoch": 1.4947637503092275, + "grad_norm": 2.069129228591919, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.728284478187561, + "num_tokens": 435423613.0, + "step": 18127 + }, + { + "epoch": 1.4948462109342788, + "grad_norm": 2.013643264770508, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7652854919433594, + "num_tokens": 435454079.0, + "step": 18128 + }, + { + "epoch": 1.4949286715593304, + "grad_norm": 2.074226140975952, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7601872086524963, + "num_tokens": 435480629.0, + "step": 18129 + }, + { + "epoch": 1.495011132184382, + "grad_norm": 2.2417523860931396, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7695299386978149, + "num_tokens": 435507593.0, + "step": 18130 + }, + { + "epoch": 1.4950935928094335, + "grad_norm": 1.8438619375228882, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7882813811302185, + "num_tokens": 435540842.0, + "step": 18131 + }, + { + "epoch": 1.4951760534344851, + "grad_norm": 1.9956315755844116, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7689909338951111, + "num_tokens": 435572048.0, + "step": 18132 + }, + { + "epoch": 1.4952585140595365, + "grad_norm": 2.064440965652466, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7410134077072144, + "num_tokens": 435601226.0, + "step": 18133 + }, + { + "epoch": 1.495340974684588, + "grad_norm": 2.136906862258911, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7427188754081726, + "num_tokens": 435628971.0, + "step": 18134 + }, + { + "epoch": 1.4954234353096396, + "grad_norm": 1.9864202737808228, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7647628784179688, + "num_tokens": 435659668.0, + "step": 18135 + }, + { + "epoch": 1.4955058959346912, + "grad_norm": 2.0418436527252197, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7388555407524109, + "num_tokens": 435689337.0, + "step": 18136 + }, + { + "epoch": 1.4955883565597428, + "grad_norm": 1.7543785572052002, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.7845404148101807, + "num_tokens": 435725516.0, + "step": 18137 + }, + { + "epoch": 1.4956708171847943, + "grad_norm": 1.896405577659607, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7407915592193604, + "num_tokens": 435758298.0, + "step": 18138 + }, + { + "epoch": 1.495753277809846, + "grad_norm": 1.932288408279419, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.774361252784729, + "num_tokens": 435789246.0, + "step": 18139 + }, + { + "epoch": 1.4958357384348973, + "grad_norm": 2.120952844619751, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7414937019348145, + "num_tokens": 435817987.0, + "step": 18140 + }, + { + "epoch": 1.4959181990599488, + "grad_norm": 2.035377264022827, + "learning_rate": 1e-06, + "loss": 0.5597, + "mean_token_accuracy": 0.8065068125724792, + "num_tokens": 435844542.0, + "step": 18141 + }, + { + "epoch": 1.4960006596850004, + "grad_norm": 2.489427089691162, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7381487488746643, + "num_tokens": 435867777.0, + "step": 18142 + }, + { + "epoch": 1.496083120310052, + "grad_norm": 2.007270097732544, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7365643978118896, + "num_tokens": 435897932.0, + "step": 18143 + }, + { + "epoch": 1.4961655809351035, + "grad_norm": 2.078185558319092, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.796474814414978, + "num_tokens": 435926648.0, + "step": 18144 + }, + { + "epoch": 1.496248041560155, + "grad_norm": 1.9201678037643433, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.7910149097442627, + "num_tokens": 435957573.0, + "step": 18145 + }, + { + "epoch": 1.4963305021852067, + "grad_norm": 2.1465377807617188, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7497169971466064, + "num_tokens": 435984744.0, + "step": 18146 + }, + { + "epoch": 1.496412962810258, + "grad_norm": 2.2416698932647705, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7386565804481506, + "num_tokens": 436009283.0, + "step": 18147 + }, + { + "epoch": 1.4964954234353096, + "grad_norm": 2.1366920471191406, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7753303050994873, + "num_tokens": 436034402.0, + "step": 18148 + }, + { + "epoch": 1.4965778840603612, + "grad_norm": 1.8673560619354248, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7654657363891602, + "num_tokens": 436068798.0, + "step": 18149 + }, + { + "epoch": 1.4966603446854128, + "grad_norm": 2.0537407398223877, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7224276065826416, + "num_tokens": 436098784.0, + "step": 18150 + }, + { + "epoch": 1.4967428053104643, + "grad_norm": 1.9650990962982178, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7601351737976074, + "num_tokens": 436129320.0, + "step": 18151 + }, + { + "epoch": 1.4968252659355157, + "grad_norm": 2.036324977874756, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7389951944351196, + "num_tokens": 436159486.0, + "step": 18152 + }, + { + "epoch": 1.4969077265605673, + "grad_norm": 1.8233015537261963, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7736278772354126, + "num_tokens": 436192841.0, + "step": 18153 + }, + { + "epoch": 1.4969901871856188, + "grad_norm": 1.6992970705032349, + "learning_rate": 1e-06, + "loss": 0.5911, + "mean_token_accuracy": 0.8090229034423828, + "num_tokens": 436229319.0, + "step": 18154 + }, + { + "epoch": 1.4970726478106704, + "grad_norm": 2.2573752403259277, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7173815965652466, + "num_tokens": 436258185.0, + "step": 18155 + }, + { + "epoch": 1.497155108435722, + "grad_norm": 2.0164499282836914, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.8039318323135376, + "num_tokens": 436285463.0, + "step": 18156 + }, + { + "epoch": 1.4972375690607735, + "grad_norm": 2.045473098754883, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.7553540468215942, + "num_tokens": 436312513.0, + "step": 18157 + }, + { + "epoch": 1.4973200296858251, + "grad_norm": 2.062469005584717, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.762915849685669, + "num_tokens": 436340505.0, + "step": 18158 + }, + { + "epoch": 1.4974024903108765, + "grad_norm": 1.8463149070739746, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7389281988143921, + "num_tokens": 436374865.0, + "step": 18159 + }, + { + "epoch": 1.497484950935928, + "grad_norm": 2.0939009189605713, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7306454181671143, + "num_tokens": 436405371.0, + "step": 18160 + }, + { + "epoch": 1.4975674115609796, + "grad_norm": 2.0850493907928467, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7472670078277588, + "num_tokens": 436433286.0, + "step": 18161 + }, + { + "epoch": 1.4976498721860312, + "grad_norm": 1.9474925994873047, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7711626291275024, + "num_tokens": 436461404.0, + "step": 18162 + }, + { + "epoch": 1.4977323328110828, + "grad_norm": 1.9918127059936523, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7604640126228333, + "num_tokens": 436491825.0, + "step": 18163 + }, + { + "epoch": 1.4978147934361343, + "grad_norm": 2.3348166942596436, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7803645133972168, + "num_tokens": 436512437.0, + "step": 18164 + }, + { + "epoch": 1.497897254061186, + "grad_norm": 2.055586338043213, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7239675521850586, + "num_tokens": 436541307.0, + "step": 18165 + }, + { + "epoch": 1.4979797146862373, + "grad_norm": 2.034820318222046, + "learning_rate": 1e-06, + "loss": 0.5535, + "mean_token_accuracy": 0.8183420896530151, + "num_tokens": 436567452.0, + "step": 18166 + }, + { + "epoch": 1.4980621753112888, + "grad_norm": 2.0125203132629395, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.7937347292900085, + "num_tokens": 436598237.0, + "step": 18167 + }, + { + "epoch": 1.4981446359363404, + "grad_norm": 1.9684735536575317, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7583999633789062, + "num_tokens": 436626594.0, + "step": 18168 + }, + { + "epoch": 1.498227096561392, + "grad_norm": 2.4252147674560547, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7480511665344238, + "num_tokens": 436647676.0, + "step": 18169 + }, + { + "epoch": 1.4983095571864435, + "grad_norm": 2.041445016860962, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7304888963699341, + "num_tokens": 436678401.0, + "step": 18170 + }, + { + "epoch": 1.498392017811495, + "grad_norm": 2.0216939449310303, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7357286810874939, + "num_tokens": 436709869.0, + "step": 18171 + }, + { + "epoch": 1.4984744784365467, + "grad_norm": 2.1372454166412354, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7356806993484497, + "num_tokens": 436740142.0, + "step": 18172 + }, + { + "epoch": 1.498556939061598, + "grad_norm": 1.743261456489563, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.7843852043151855, + "num_tokens": 436776374.0, + "step": 18173 + }, + { + "epoch": 1.4986393996866496, + "grad_norm": 1.854395866394043, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7303321361541748, + "num_tokens": 436812577.0, + "step": 18174 + }, + { + "epoch": 1.4987218603117012, + "grad_norm": 2.0608019828796387, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7616463899612427, + "num_tokens": 436842313.0, + "step": 18175 + }, + { + "epoch": 1.4988043209367528, + "grad_norm": 2.117659568786621, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7430020570755005, + "num_tokens": 436869763.0, + "step": 18176 + }, + { + "epoch": 1.4988867815618043, + "grad_norm": 2.174712657928467, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7111749649047852, + "num_tokens": 436898439.0, + "step": 18177 + }, + { + "epoch": 1.4989692421868557, + "grad_norm": 2.1235544681549072, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.781907320022583, + "num_tokens": 436926907.0, + "step": 18178 + }, + { + "epoch": 1.4990517028119072, + "grad_norm": 2.065793991088867, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7523894309997559, + "num_tokens": 436953937.0, + "step": 18179 + }, + { + "epoch": 1.4991341634369588, + "grad_norm": 2.3801474571228027, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7400372624397278, + "num_tokens": 436976360.0, + "step": 18180 + }, + { + "epoch": 1.4992166240620104, + "grad_norm": 2.0487735271453857, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7419686317443848, + "num_tokens": 437008188.0, + "step": 18181 + }, + { + "epoch": 1.499299084687062, + "grad_norm": 2.011873245239258, + "learning_rate": 1e-06, + "loss": 0.783, + "mean_token_accuracy": 0.7605007886886597, + "num_tokens": 437038398.0, + "step": 18182 + }, + { + "epoch": 1.4993815453121135, + "grad_norm": 2.2414493560791016, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7719763517379761, + "num_tokens": 437063722.0, + "step": 18183 + }, + { + "epoch": 1.4994640059371651, + "grad_norm": 1.9298617839813232, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.796615719795227, + "num_tokens": 437092630.0, + "step": 18184 + }, + { + "epoch": 1.4995464665622165, + "grad_norm": 1.8827409744262695, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.7525650858879089, + "num_tokens": 437124209.0, + "step": 18185 + }, + { + "epoch": 1.499628927187268, + "grad_norm": 2.069148540496826, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.7379254102706909, + "num_tokens": 437154207.0, + "step": 18186 + }, + { + "epoch": 1.4997113878123196, + "grad_norm": 1.9182054996490479, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7413812875747681, + "num_tokens": 437187197.0, + "step": 18187 + }, + { + "epoch": 1.4997938484373712, + "grad_norm": 1.9466569423675537, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.7938885688781738, + "num_tokens": 437218039.0, + "step": 18188 + }, + { + "epoch": 1.4998763090624228, + "grad_norm": 2.4359521865844727, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7375980615615845, + "num_tokens": 437238707.0, + "step": 18189 + }, + { + "epoch": 1.499958769687474, + "grad_norm": 2.0057878494262695, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7559887766838074, + "num_tokens": 437269407.0, + "step": 18190 + }, + { + "epoch": 1.500041230312526, + "grad_norm": 1.9588817358016968, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7720645070075989, + "num_tokens": 437301348.0, + "step": 18191 + }, + { + "epoch": 1.5001236909375772, + "grad_norm": 2.0075292587280273, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.757452130317688, + "num_tokens": 437329610.0, + "step": 18192 + }, + { + "epoch": 1.5002061515626288, + "grad_norm": 1.85361647605896, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.7536608576774597, + "num_tokens": 437363408.0, + "step": 18193 + }, + { + "epoch": 1.5002886121876804, + "grad_norm": 1.8884073495864868, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7620378732681274, + "num_tokens": 437397112.0, + "step": 18194 + }, + { + "epoch": 1.500371072812732, + "grad_norm": 1.9318045377731323, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7747393846511841, + "num_tokens": 437425885.0, + "step": 18195 + }, + { + "epoch": 1.5004535334377835, + "grad_norm": 2.056772232055664, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.7572484016418457, + "num_tokens": 437452941.0, + "step": 18196 + }, + { + "epoch": 1.5005359940628349, + "grad_norm": 2.0295491218566895, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.7868714928627014, + "num_tokens": 437479983.0, + "step": 18197 + }, + { + "epoch": 1.5006184546878867, + "grad_norm": 1.9804046154022217, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7884378433227539, + "num_tokens": 437510658.0, + "step": 18198 + }, + { + "epoch": 1.500700915312938, + "grad_norm": 2.1192996501922607, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7273977398872375, + "num_tokens": 437540872.0, + "step": 18199 + }, + { + "epoch": 1.5007833759379896, + "grad_norm": 1.9782735109329224, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7215888500213623, + "num_tokens": 437570545.0, + "step": 18200 + }, + { + "epoch": 1.5008658365630412, + "grad_norm": 1.9064137935638428, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7509623765945435, + "num_tokens": 437604136.0, + "step": 18201 + }, + { + "epoch": 1.5009482971880925, + "grad_norm": 1.8346763849258423, + "learning_rate": 1e-06, + "loss": 0.5817, + "mean_token_accuracy": 0.8104413151741028, + "num_tokens": 437634193.0, + "step": 18202 + }, + { + "epoch": 1.5010307578131443, + "grad_norm": 2.064302444458008, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7655439376831055, + "num_tokens": 437660573.0, + "step": 18203 + }, + { + "epoch": 1.5011132184381957, + "grad_norm": 2.117794990539551, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7611010074615479, + "num_tokens": 437688442.0, + "step": 18204 + }, + { + "epoch": 1.5011956790632472, + "grad_norm": 2.2567830085754395, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.7707069516181946, + "num_tokens": 437713118.0, + "step": 18205 + }, + { + "epoch": 1.5012781396882988, + "grad_norm": 2.1462738513946533, + "learning_rate": 1e-06, + "loss": 0.5877, + "mean_token_accuracy": 0.806755781173706, + "num_tokens": 437738358.0, + "step": 18206 + }, + { + "epoch": 1.5013606003133504, + "grad_norm": 2.2456459999084473, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7606652975082397, + "num_tokens": 437763588.0, + "step": 18207 + }, + { + "epoch": 1.501443060938402, + "grad_norm": 2.0825579166412354, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7555210590362549, + "num_tokens": 437793141.0, + "step": 18208 + }, + { + "epoch": 1.5015255215634533, + "grad_norm": 2.2460172176361084, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7294289469718933, + "num_tokens": 437819076.0, + "step": 18209 + }, + { + "epoch": 1.501607982188505, + "grad_norm": 1.7653082609176636, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7706285715103149, + "num_tokens": 437855670.0, + "step": 18210 + }, + { + "epoch": 1.5016904428135565, + "grad_norm": 2.392258882522583, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7519199252128601, + "num_tokens": 437879801.0, + "step": 18211 + }, + { + "epoch": 1.501772903438608, + "grad_norm": 1.9513609409332275, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7427655458450317, + "num_tokens": 437912028.0, + "step": 18212 + }, + { + "epoch": 1.5018553640636596, + "grad_norm": 2.258265733718872, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7517514228820801, + "num_tokens": 437936556.0, + "step": 18213 + }, + { + "epoch": 1.5019378246887112, + "grad_norm": 1.768027424812317, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7526383399963379, + "num_tokens": 437975339.0, + "step": 18214 + }, + { + "epoch": 1.5020202853137627, + "grad_norm": 1.991108775138855, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7257925271987915, + "num_tokens": 438008537.0, + "step": 18215 + }, + { + "epoch": 1.502102745938814, + "grad_norm": 2.1862645149230957, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7658149003982544, + "num_tokens": 438032678.0, + "step": 18216 + }, + { + "epoch": 1.502185206563866, + "grad_norm": 1.84511399269104, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7555401921272278, + "num_tokens": 438066523.0, + "step": 18217 + }, + { + "epoch": 1.5022676671889172, + "grad_norm": 1.926983118057251, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.756751298904419, + "num_tokens": 438098417.0, + "step": 18218 + }, + { + "epoch": 1.5023501278139688, + "grad_norm": 2.3048274517059326, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7345942258834839, + "num_tokens": 438123121.0, + "step": 18219 + }, + { + "epoch": 1.5024325884390204, + "grad_norm": 2.2189176082611084, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7406659722328186, + "num_tokens": 438149634.0, + "step": 18220 + }, + { + "epoch": 1.5025150490640717, + "grad_norm": 2.069679021835327, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7431677579879761, + "num_tokens": 438179789.0, + "step": 18221 + }, + { + "epoch": 1.5025975096891235, + "grad_norm": 1.9780949354171753, + "learning_rate": 1e-06, + "loss": 0.6195, + "mean_token_accuracy": 0.7987245321273804, + "num_tokens": 438210939.0, + "step": 18222 + }, + { + "epoch": 1.5026799703141749, + "grad_norm": 1.8108593225479126, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7827737927436829, + "num_tokens": 438243638.0, + "step": 18223 + }, + { + "epoch": 1.5027624309392267, + "grad_norm": 2.0841903686523438, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7371700406074524, + "num_tokens": 438272176.0, + "step": 18224 + }, + { + "epoch": 1.502844891564278, + "grad_norm": 1.7876311540603638, + "learning_rate": 1e-06, + "loss": 0.574, + "mean_token_accuracy": 0.8173215389251709, + "num_tokens": 438304357.0, + "step": 18225 + }, + { + "epoch": 1.5029273521893296, + "grad_norm": 1.754592776298523, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7757532000541687, + "num_tokens": 438342544.0, + "step": 18226 + }, + { + "epoch": 1.5030098128143812, + "grad_norm": 1.9403494596481323, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7776831388473511, + "num_tokens": 438375305.0, + "step": 18227 + }, + { + "epoch": 1.5030922734394325, + "grad_norm": 2.0256736278533936, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7634975910186768, + "num_tokens": 438403550.0, + "step": 18228 + }, + { + "epoch": 1.5031747340644843, + "grad_norm": 2.0923378467559814, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.739925742149353, + "num_tokens": 438432461.0, + "step": 18229 + }, + { + "epoch": 1.5032571946895357, + "grad_norm": 2.1869893074035645, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7829564809799194, + "num_tokens": 438458685.0, + "step": 18230 + }, + { + "epoch": 1.5033396553145872, + "grad_norm": 1.968759298324585, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7707040309906006, + "num_tokens": 438488541.0, + "step": 18231 + }, + { + "epoch": 1.5034221159396388, + "grad_norm": 1.9626914262771606, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.7916423082351685, + "num_tokens": 438519261.0, + "step": 18232 + }, + { + "epoch": 1.5035045765646904, + "grad_norm": 2.2165145874023438, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7629576921463013, + "num_tokens": 438544006.0, + "step": 18233 + }, + { + "epoch": 1.503587037189742, + "grad_norm": 1.8424897193908691, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7496389746665955, + "num_tokens": 438577253.0, + "step": 18234 + }, + { + "epoch": 1.5036694978147933, + "grad_norm": 1.935030221939087, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7633159160614014, + "num_tokens": 438606223.0, + "step": 18235 + }, + { + "epoch": 1.503751958439845, + "grad_norm": 1.9662961959838867, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.764480710029602, + "num_tokens": 438637948.0, + "step": 18236 + }, + { + "epoch": 1.5038344190648965, + "grad_norm": 2.1397922039031982, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7314767241477966, + "num_tokens": 438666432.0, + "step": 18237 + }, + { + "epoch": 1.503916879689948, + "grad_norm": 1.979897379875183, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7472562789916992, + "num_tokens": 438697683.0, + "step": 18238 + }, + { + "epoch": 1.5039993403149996, + "grad_norm": 1.7061268091201782, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.7909501791000366, + "num_tokens": 438732699.0, + "step": 18239 + }, + { + "epoch": 1.5040818009400512, + "grad_norm": 1.9087363481521606, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.7861946821212769, + "num_tokens": 438763241.0, + "step": 18240 + }, + { + "epoch": 1.5041642615651027, + "grad_norm": 2.197693109512329, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.7910846471786499, + "num_tokens": 438786679.0, + "step": 18241 + }, + { + "epoch": 1.504246722190154, + "grad_norm": 2.373208522796631, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7814446687698364, + "num_tokens": 438806105.0, + "step": 18242 + }, + { + "epoch": 1.504329182815206, + "grad_norm": 1.8132219314575195, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7386576533317566, + "num_tokens": 438844247.0, + "step": 18243 + }, + { + "epoch": 1.5044116434402572, + "grad_norm": 2.0558183193206787, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.733098030090332, + "num_tokens": 438873276.0, + "step": 18244 + }, + { + "epoch": 1.5044941040653088, + "grad_norm": 2.2825238704681396, + "learning_rate": 1e-06, + "loss": 0.7702, + "mean_token_accuracy": 0.7545037269592285, + "num_tokens": 438897965.0, + "step": 18245 + }, + { + "epoch": 1.5045765646903604, + "grad_norm": 2.0599594116210938, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7713455557823181, + "num_tokens": 438926143.0, + "step": 18246 + }, + { + "epoch": 1.5046590253154117, + "grad_norm": 2.034958839416504, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.7548158764839172, + "num_tokens": 438954177.0, + "step": 18247 + }, + { + "epoch": 1.5047414859404635, + "grad_norm": 1.9153387546539307, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.782730758190155, + "num_tokens": 438984351.0, + "step": 18248 + }, + { + "epoch": 1.5048239465655149, + "grad_norm": 2.311927080154419, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7542083263397217, + "num_tokens": 439010987.0, + "step": 18249 + }, + { + "epoch": 1.5049064071905665, + "grad_norm": 2.1471614837646484, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7404332160949707, + "num_tokens": 439037724.0, + "step": 18250 + }, + { + "epoch": 1.504988867815618, + "grad_norm": 1.8867073059082031, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7455393075942993, + "num_tokens": 439070434.0, + "step": 18251 + }, + { + "epoch": 1.5050713284406696, + "grad_norm": 1.931775689125061, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7736893892288208, + "num_tokens": 439101127.0, + "step": 18252 + }, + { + "epoch": 1.5051537890657212, + "grad_norm": 1.8741925954818726, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7685613632202148, + "num_tokens": 439132340.0, + "step": 18253 + }, + { + "epoch": 1.5052362496907725, + "grad_norm": 1.820438265800476, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7660062909126282, + "num_tokens": 439167144.0, + "step": 18254 + }, + { + "epoch": 1.5053187103158243, + "grad_norm": 1.9025418758392334, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.7897687554359436, + "num_tokens": 439197865.0, + "step": 18255 + }, + { + "epoch": 1.5054011709408757, + "grad_norm": 2.433136224746704, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7335140705108643, + "num_tokens": 439221433.0, + "step": 18256 + }, + { + "epoch": 1.5054836315659272, + "grad_norm": 1.9288626909255981, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.747768759727478, + "num_tokens": 439252736.0, + "step": 18257 + }, + { + "epoch": 1.5055660921909788, + "grad_norm": 1.9101762771606445, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7325742244720459, + "num_tokens": 439286435.0, + "step": 18258 + }, + { + "epoch": 1.5056485528160304, + "grad_norm": 2.035884141921997, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7778959274291992, + "num_tokens": 439315282.0, + "step": 18259 + }, + { + "epoch": 1.505731013441082, + "grad_norm": 1.9562511444091797, + "learning_rate": 1e-06, + "loss": 0.6043, + "mean_token_accuracy": 0.7926099896430969, + "num_tokens": 439342211.0, + "step": 18260 + }, + { + "epoch": 1.5058134740661333, + "grad_norm": 2.0551953315734863, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7719988226890564, + "num_tokens": 439371667.0, + "step": 18261 + }, + { + "epoch": 1.505895934691185, + "grad_norm": 1.8134264945983887, + "learning_rate": 1e-06, + "loss": 0.529, + "mean_token_accuracy": 0.8283146619796753, + "num_tokens": 439404816.0, + "step": 18262 + }, + { + "epoch": 1.5059783953162365, + "grad_norm": 1.7874869108200073, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.784051775932312, + "num_tokens": 439436443.0, + "step": 18263 + }, + { + "epoch": 1.506060855941288, + "grad_norm": 1.9915778636932373, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7588478326797485, + "num_tokens": 439466203.0, + "step": 18264 + }, + { + "epoch": 1.5061433165663396, + "grad_norm": 2.133876323699951, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7639564275741577, + "num_tokens": 439494002.0, + "step": 18265 + }, + { + "epoch": 1.506225777191391, + "grad_norm": 1.8003604412078857, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7512627840042114, + "num_tokens": 439526955.0, + "step": 18266 + }, + { + "epoch": 1.5063082378164427, + "grad_norm": 1.8982371091842651, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7667532563209534, + "num_tokens": 439561590.0, + "step": 18267 + }, + { + "epoch": 1.506390698441494, + "grad_norm": 1.7568819522857666, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7438135147094727, + "num_tokens": 439598390.0, + "step": 18268 + }, + { + "epoch": 1.5064731590665459, + "grad_norm": 2.0821008682250977, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7464075088500977, + "num_tokens": 439624689.0, + "step": 18269 + }, + { + "epoch": 1.5065556196915972, + "grad_norm": 1.8230754137039185, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.7946698069572449, + "num_tokens": 439658268.0, + "step": 18270 + }, + { + "epoch": 1.5066380803166488, + "grad_norm": 1.8380773067474365, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.7782987356185913, + "num_tokens": 439691676.0, + "step": 18271 + }, + { + "epoch": 1.5067205409417004, + "grad_norm": 1.8082716464996338, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.7913286089897156, + "num_tokens": 439724550.0, + "step": 18272 + }, + { + "epoch": 1.5068030015667517, + "grad_norm": 1.9309971332550049, + "learning_rate": 1e-06, + "loss": 0.605, + "mean_token_accuracy": 0.8002844452857971, + "num_tokens": 439753858.0, + "step": 18273 + }, + { + "epoch": 1.5068854621918035, + "grad_norm": 1.8234217166900635, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.7793022394180298, + "num_tokens": 439788862.0, + "step": 18274 + }, + { + "epoch": 1.5069679228168549, + "grad_norm": 1.952913522720337, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.7822525501251221, + "num_tokens": 439817319.0, + "step": 18275 + }, + { + "epoch": 1.5070503834419064, + "grad_norm": 2.1109275817871094, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.730366587638855, + "num_tokens": 439844808.0, + "step": 18276 + }, + { + "epoch": 1.507132844066958, + "grad_norm": 1.7393770217895508, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.7961342930793762, + "num_tokens": 439881497.0, + "step": 18277 + }, + { + "epoch": 1.5072153046920096, + "grad_norm": 1.8958834409713745, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7665668725967407, + "num_tokens": 439913565.0, + "step": 18278 + }, + { + "epoch": 1.5072977653170612, + "grad_norm": 2.1164700984954834, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.789229691028595, + "num_tokens": 439939365.0, + "step": 18279 + }, + { + "epoch": 1.5073802259421125, + "grad_norm": 2.0106663703918457, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7428678274154663, + "num_tokens": 439968782.0, + "step": 18280 + }, + { + "epoch": 1.5074626865671643, + "grad_norm": 2.0809199810028076, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.7950177192687988, + "num_tokens": 439996537.0, + "step": 18281 + }, + { + "epoch": 1.5075451471922157, + "grad_norm": 2.157179594039917, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7149671316146851, + "num_tokens": 440024983.0, + "step": 18282 + }, + { + "epoch": 1.5076276078172672, + "grad_norm": 1.923473596572876, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7713242769241333, + "num_tokens": 440056792.0, + "step": 18283 + }, + { + "epoch": 1.5077100684423188, + "grad_norm": 2.1575868129730225, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7421548366546631, + "num_tokens": 440082738.0, + "step": 18284 + }, + { + "epoch": 1.5077925290673704, + "grad_norm": 1.9657942056655884, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7150830030441284, + "num_tokens": 440114613.0, + "step": 18285 + }, + { + "epoch": 1.507874989692422, + "grad_norm": 1.714491844177246, + "learning_rate": 1e-06, + "loss": 0.4895, + "mean_token_accuracy": 0.8288536071777344, + "num_tokens": 440147999.0, + "step": 18286 + }, + { + "epoch": 1.5079574503174733, + "grad_norm": 2.174095630645752, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.758417010307312, + "num_tokens": 440178407.0, + "step": 18287 + }, + { + "epoch": 1.508039910942525, + "grad_norm": 1.9137635231018066, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7795954942703247, + "num_tokens": 440209530.0, + "step": 18288 + }, + { + "epoch": 1.5081223715675764, + "grad_norm": 1.8130358457565308, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.781648576259613, + "num_tokens": 440242946.0, + "step": 18289 + }, + { + "epoch": 1.508204832192628, + "grad_norm": 2.022768497467041, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.7891640663146973, + "num_tokens": 440273316.0, + "step": 18290 + }, + { + "epoch": 1.5082872928176796, + "grad_norm": 1.8996556997299194, + "learning_rate": 1e-06, + "loss": 0.5252, + "mean_token_accuracy": 0.8278359174728394, + "num_tokens": 440303851.0, + "step": 18291 + }, + { + "epoch": 1.508369753442731, + "grad_norm": 1.9851540327072144, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7520331740379333, + "num_tokens": 440335622.0, + "step": 18292 + }, + { + "epoch": 1.5084522140677827, + "grad_norm": 2.2691431045532227, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7450788021087646, + "num_tokens": 440359604.0, + "step": 18293 + }, + { + "epoch": 1.508534674692834, + "grad_norm": 2.0180091857910156, + "learning_rate": 1e-06, + "loss": 0.7722, + "mean_token_accuracy": 0.7555925846099854, + "num_tokens": 440389501.0, + "step": 18294 + }, + { + "epoch": 1.5086171353178857, + "grad_norm": 1.8085129261016846, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7891927361488342, + "num_tokens": 440425109.0, + "step": 18295 + }, + { + "epoch": 1.5086995959429372, + "grad_norm": 1.934948444366455, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7818610668182373, + "num_tokens": 440454780.0, + "step": 18296 + }, + { + "epoch": 1.5087820565679888, + "grad_norm": 2.3878607749938965, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7464205026626587, + "num_tokens": 440479211.0, + "step": 18297 + }, + { + "epoch": 1.5088645171930404, + "grad_norm": 1.9896414279937744, + "learning_rate": 1e-06, + "loss": 0.6059, + "mean_token_accuracy": 0.8026741743087769, + "num_tokens": 440509871.0, + "step": 18298 + }, + { + "epoch": 1.5089469778180917, + "grad_norm": 1.8425079584121704, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7563282251358032, + "num_tokens": 440542020.0, + "step": 18299 + }, + { + "epoch": 1.5090294384431435, + "grad_norm": 2.172088146209717, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7632703185081482, + "num_tokens": 440568018.0, + "step": 18300 + }, + { + "epoch": 1.5091118990681949, + "grad_norm": 1.982149362564087, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7599635124206543, + "num_tokens": 440598578.0, + "step": 18301 + }, + { + "epoch": 1.5091943596932464, + "grad_norm": 1.9682142734527588, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7236478924751282, + "num_tokens": 440630187.0, + "step": 18302 + }, + { + "epoch": 1.509276820318298, + "grad_norm": 2.1270134449005127, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7495380640029907, + "num_tokens": 440656932.0, + "step": 18303 + }, + { + "epoch": 1.5093592809433496, + "grad_norm": 2.0653321743011475, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7842148542404175, + "num_tokens": 440685763.0, + "step": 18304 + }, + { + "epoch": 1.5094417415684012, + "grad_norm": 2.0257797241210938, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7877280712127686, + "num_tokens": 440715897.0, + "step": 18305 + }, + { + "epoch": 1.5095242021934525, + "grad_norm": 1.9836220741271973, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7397555708885193, + "num_tokens": 440747711.0, + "step": 18306 + }, + { + "epoch": 1.5096066628185043, + "grad_norm": 2.23368763923645, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7701916694641113, + "num_tokens": 440771855.0, + "step": 18307 + }, + { + "epoch": 1.5096891234435557, + "grad_norm": 2.0198097229003906, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7768913507461548, + "num_tokens": 440798995.0, + "step": 18308 + }, + { + "epoch": 1.5097715840686072, + "grad_norm": 1.7491085529327393, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.7960302233695984, + "num_tokens": 440833592.0, + "step": 18309 + }, + { + "epoch": 1.5098540446936588, + "grad_norm": 1.97481107711792, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.7861868143081665, + "num_tokens": 440863334.0, + "step": 18310 + }, + { + "epoch": 1.5099365053187102, + "grad_norm": 1.8242684602737427, + "learning_rate": 1e-06, + "loss": 0.6054, + "mean_token_accuracy": 0.7942320108413696, + "num_tokens": 440897465.0, + "step": 18311 + }, + { + "epoch": 1.510018965943762, + "grad_norm": 1.828813910484314, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.7986894249916077, + "num_tokens": 440931010.0, + "step": 18312 + }, + { + "epoch": 1.5101014265688133, + "grad_norm": 1.9855602979660034, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7527540326118469, + "num_tokens": 440964279.0, + "step": 18313 + }, + { + "epoch": 1.510183887193865, + "grad_norm": 2.057978868484497, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7501624822616577, + "num_tokens": 440994092.0, + "step": 18314 + }, + { + "epoch": 1.5102663478189164, + "grad_norm": 1.9901295900344849, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.7846128344535828, + "num_tokens": 441024512.0, + "step": 18315 + }, + { + "epoch": 1.510348808443968, + "grad_norm": 1.8873792886734009, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7463881969451904, + "num_tokens": 441056622.0, + "step": 18316 + }, + { + "epoch": 1.5104312690690196, + "grad_norm": 2.3095881938934326, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.754589319229126, + "num_tokens": 441080294.0, + "step": 18317 + }, + { + "epoch": 1.510513729694071, + "grad_norm": 2.0894956588745117, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7812080979347229, + "num_tokens": 441106378.0, + "step": 18318 + }, + { + "epoch": 1.5105961903191227, + "grad_norm": 1.720055341720581, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7520959377288818, + "num_tokens": 441141473.0, + "step": 18319 + }, + { + "epoch": 1.510678650944174, + "grad_norm": 1.976660966873169, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7737555503845215, + "num_tokens": 441171180.0, + "step": 18320 + }, + { + "epoch": 1.5107611115692257, + "grad_norm": 1.79611074924469, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7629486918449402, + "num_tokens": 441205504.0, + "step": 18321 + }, + { + "epoch": 1.5108435721942772, + "grad_norm": 2.3255720138549805, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.758934497833252, + "num_tokens": 441229165.0, + "step": 18322 + }, + { + "epoch": 1.5109260328193288, + "grad_norm": 2.1384568214416504, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7384523153305054, + "num_tokens": 441256104.0, + "step": 18323 + }, + { + "epoch": 1.5110084934443804, + "grad_norm": 1.8305469751358032, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7749391794204712, + "num_tokens": 441292720.0, + "step": 18324 + }, + { + "epoch": 1.5110909540694317, + "grad_norm": 1.8195219039916992, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7663530111312866, + "num_tokens": 441327151.0, + "step": 18325 + }, + { + "epoch": 1.5111734146944835, + "grad_norm": 2.0032432079315186, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.73388671875, + "num_tokens": 441359087.0, + "step": 18326 + }, + { + "epoch": 1.5112558753195349, + "grad_norm": 2.105748176574707, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.786172091960907, + "num_tokens": 441386750.0, + "step": 18327 + }, + { + "epoch": 1.5113383359445864, + "grad_norm": 2.130366802215576, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7678921222686768, + "num_tokens": 441415813.0, + "step": 18328 + }, + { + "epoch": 1.511420796569638, + "grad_norm": 1.8861743211746216, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7596316337585449, + "num_tokens": 441447424.0, + "step": 18329 + }, + { + "epoch": 1.5115032571946896, + "grad_norm": 2.0664587020874023, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7461071610450745, + "num_tokens": 441477505.0, + "step": 18330 + }, + { + "epoch": 1.5115857178197412, + "grad_norm": 2.1848132610321045, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7313282489776611, + "num_tokens": 441502003.0, + "step": 18331 + }, + { + "epoch": 1.5116681784447925, + "grad_norm": 2.3462250232696533, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7475517988204956, + "num_tokens": 441527762.0, + "step": 18332 + }, + { + "epoch": 1.5117506390698443, + "grad_norm": 2.14749813079834, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.769963264465332, + "num_tokens": 441555044.0, + "step": 18333 + }, + { + "epoch": 1.5118330996948957, + "grad_norm": 2.24246883392334, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.7502925992012024, + "num_tokens": 441579479.0, + "step": 18334 + }, + { + "epoch": 1.5119155603199472, + "grad_norm": 1.9920845031738281, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7780746817588806, + "num_tokens": 441610440.0, + "step": 18335 + }, + { + "epoch": 1.5119980209449988, + "grad_norm": 2.0989394187927246, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7558056116104126, + "num_tokens": 441637586.0, + "step": 18336 + }, + { + "epoch": 1.5120804815700501, + "grad_norm": 2.03134822845459, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.718932032585144, + "num_tokens": 441668483.0, + "step": 18337 + }, + { + "epoch": 1.512162942195102, + "grad_norm": 1.9351167678833008, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.7605367302894592, + "num_tokens": 441700618.0, + "step": 18338 + }, + { + "epoch": 1.5122454028201533, + "grad_norm": 1.9946702718734741, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7677477598190308, + "num_tokens": 441730238.0, + "step": 18339 + }, + { + "epoch": 1.5123278634452049, + "grad_norm": 2.047480583190918, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7775613069534302, + "num_tokens": 441758099.0, + "step": 18340 + }, + { + "epoch": 1.5124103240702564, + "grad_norm": 1.9714127779006958, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7425459027290344, + "num_tokens": 441787530.0, + "step": 18341 + }, + { + "epoch": 1.512492784695308, + "grad_norm": 2.464491128921509, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.779429018497467, + "num_tokens": 441806058.0, + "step": 18342 + }, + { + "epoch": 1.5125752453203596, + "grad_norm": 1.9908190965652466, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.7918342351913452, + "num_tokens": 441836255.0, + "step": 18343 + }, + { + "epoch": 1.512657705945411, + "grad_norm": 2.151703357696533, + "learning_rate": 1e-06, + "loss": 0.7706, + "mean_token_accuracy": 0.758121132850647, + "num_tokens": 441861403.0, + "step": 18344 + }, + { + "epoch": 1.5127401665704627, + "grad_norm": 1.7382616996765137, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.7950105667114258, + "num_tokens": 441898576.0, + "step": 18345 + }, + { + "epoch": 1.512822627195514, + "grad_norm": 2.479243278503418, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7315773963928223, + "num_tokens": 441920697.0, + "step": 18346 + }, + { + "epoch": 1.5129050878205657, + "grad_norm": 2.1050946712493896, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7546277642250061, + "num_tokens": 441947657.0, + "step": 18347 + }, + { + "epoch": 1.5129875484456172, + "grad_norm": 2.0597264766693115, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7278969287872314, + "num_tokens": 441975372.0, + "step": 18348 + }, + { + "epoch": 1.5130700090706688, + "grad_norm": 1.637995958328247, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7845726013183594, + "num_tokens": 442017500.0, + "step": 18349 + }, + { + "epoch": 1.5131524696957204, + "grad_norm": 2.132199287414551, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7814394235610962, + "num_tokens": 442046017.0, + "step": 18350 + }, + { + "epoch": 1.5132349303207717, + "grad_norm": 2.04486346244812, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7650019526481628, + "num_tokens": 442074342.0, + "step": 18351 + }, + { + "epoch": 1.5133173909458235, + "grad_norm": 1.6743215322494507, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.7883885502815247, + "num_tokens": 442114087.0, + "step": 18352 + }, + { + "epoch": 1.5133998515708749, + "grad_norm": 2.143423318862915, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7393780946731567, + "num_tokens": 442142333.0, + "step": 18353 + }, + { + "epoch": 1.5134823121959264, + "grad_norm": 1.9206342697143555, + "learning_rate": 1e-06, + "loss": 0.5814, + "mean_token_accuracy": 0.8094397783279419, + "num_tokens": 442171999.0, + "step": 18354 + }, + { + "epoch": 1.513564772820978, + "grad_norm": 1.75522780418396, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7720158696174622, + "num_tokens": 442207620.0, + "step": 18355 + }, + { + "epoch": 1.5136472334460294, + "grad_norm": 1.7324720621109009, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7574426531791687, + "num_tokens": 442248836.0, + "step": 18356 + }, + { + "epoch": 1.5137296940710812, + "grad_norm": 2.1882944107055664, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7825974225997925, + "num_tokens": 442274861.0, + "step": 18357 + }, + { + "epoch": 1.5138121546961325, + "grad_norm": 1.9048751592636108, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7675499320030212, + "num_tokens": 442307540.0, + "step": 18358 + }, + { + "epoch": 1.5138946153211843, + "grad_norm": 2.086413621902466, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.76406329870224, + "num_tokens": 442337409.0, + "step": 18359 + }, + { + "epoch": 1.5139770759462357, + "grad_norm": 1.9012904167175293, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.781927227973938, + "num_tokens": 442368515.0, + "step": 18360 + }, + { + "epoch": 1.5140595365712872, + "grad_norm": 1.9358205795288086, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7092764377593994, + "num_tokens": 442403285.0, + "step": 18361 + }, + { + "epoch": 1.5141419971963388, + "grad_norm": 1.8511805534362793, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7583981156349182, + "num_tokens": 442436167.0, + "step": 18362 + }, + { + "epoch": 1.5142244578213901, + "grad_norm": 1.7906171083450317, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7624029517173767, + "num_tokens": 442469919.0, + "step": 18363 + }, + { + "epoch": 1.514306918446442, + "grad_norm": 2.010215997695923, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7692910432815552, + "num_tokens": 442501772.0, + "step": 18364 + }, + { + "epoch": 1.5143893790714933, + "grad_norm": 1.9352185726165771, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7876278162002563, + "num_tokens": 442535792.0, + "step": 18365 + }, + { + "epoch": 1.5144718396965449, + "grad_norm": 2.005559206008911, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.7942028045654297, + "num_tokens": 442564456.0, + "step": 18366 + }, + { + "epoch": 1.5145543003215964, + "grad_norm": 1.9383723735809326, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7250995635986328, + "num_tokens": 442597611.0, + "step": 18367 + }, + { + "epoch": 1.514636760946648, + "grad_norm": 1.8264645338058472, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.7887663245201111, + "num_tokens": 442632409.0, + "step": 18368 + }, + { + "epoch": 1.5147192215716996, + "grad_norm": 1.8374648094177246, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7811115980148315, + "num_tokens": 442662803.0, + "step": 18369 + }, + { + "epoch": 1.514801682196751, + "grad_norm": 1.9886705875396729, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.774037778377533, + "num_tokens": 442692440.0, + "step": 18370 + }, + { + "epoch": 1.5148841428218027, + "grad_norm": 2.2817084789276123, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7541227340698242, + "num_tokens": 442718438.0, + "step": 18371 + }, + { + "epoch": 1.514966603446854, + "grad_norm": 1.8506579399108887, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7182429432868958, + "num_tokens": 442752629.0, + "step": 18372 + }, + { + "epoch": 1.5150490640719056, + "grad_norm": 2.201672077178955, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7599077820777893, + "num_tokens": 442778857.0, + "step": 18373 + }, + { + "epoch": 1.5151315246969572, + "grad_norm": 1.7129493951797485, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7451924681663513, + "num_tokens": 442817136.0, + "step": 18374 + }, + { + "epoch": 1.5152139853220088, + "grad_norm": 2.2065227031707764, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7394803166389465, + "num_tokens": 442845009.0, + "step": 18375 + }, + { + "epoch": 1.5152964459470604, + "grad_norm": 1.8312329053878784, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7637888193130493, + "num_tokens": 442880247.0, + "step": 18376 + }, + { + "epoch": 1.5153789065721117, + "grad_norm": 2.584517002105713, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7460100054740906, + "num_tokens": 442900114.0, + "step": 18377 + }, + { + "epoch": 1.5154613671971635, + "grad_norm": 2.062910795211792, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.769594669342041, + "num_tokens": 442926660.0, + "step": 18378 + }, + { + "epoch": 1.5155438278222149, + "grad_norm": 1.8811075687408447, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7712326049804688, + "num_tokens": 442961294.0, + "step": 18379 + }, + { + "epoch": 1.5156262884472664, + "grad_norm": 2.2016286849975586, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7654441595077515, + "num_tokens": 442987425.0, + "step": 18380 + }, + { + "epoch": 1.515708749072318, + "grad_norm": 1.8453165292739868, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.7987565398216248, + "num_tokens": 443018709.0, + "step": 18381 + }, + { + "epoch": 1.5157912096973694, + "grad_norm": 2.0376598834991455, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.7732176780700684, + "num_tokens": 443048660.0, + "step": 18382 + }, + { + "epoch": 1.5158736703224212, + "grad_norm": 2.0366272926330566, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7579104900360107, + "num_tokens": 443077182.0, + "step": 18383 + }, + { + "epoch": 1.5159561309474725, + "grad_norm": 1.9286718368530273, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7851648330688477, + "num_tokens": 443108125.0, + "step": 18384 + }, + { + "epoch": 1.516038591572524, + "grad_norm": 2.244389533996582, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7384362816810608, + "num_tokens": 443133515.0, + "step": 18385 + }, + { + "epoch": 1.5161210521975756, + "grad_norm": 1.8143329620361328, + "learning_rate": 1e-06, + "loss": 0.7821, + "mean_token_accuracy": 0.7514638900756836, + "num_tokens": 443168928.0, + "step": 18386 + }, + { + "epoch": 1.5162035128226272, + "grad_norm": 1.8657275438308716, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7783629298210144, + "num_tokens": 443202020.0, + "step": 18387 + }, + { + "epoch": 1.5162859734476788, + "grad_norm": 2.444624185562134, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7310396432876587, + "num_tokens": 443225712.0, + "step": 18388 + }, + { + "epoch": 1.5163684340727301, + "grad_norm": 2.124659776687622, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7717757225036621, + "num_tokens": 443254147.0, + "step": 18389 + }, + { + "epoch": 1.516450894697782, + "grad_norm": 1.9636787176132202, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7494980692863464, + "num_tokens": 443285597.0, + "step": 18390 + }, + { + "epoch": 1.5165333553228333, + "grad_norm": 2.2976109981536865, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7464694976806641, + "num_tokens": 443311279.0, + "step": 18391 + }, + { + "epoch": 1.5166158159478849, + "grad_norm": 2.1842868328094482, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.777947187423706, + "num_tokens": 443341307.0, + "step": 18392 + }, + { + "epoch": 1.5166982765729364, + "grad_norm": 2.0322325229644775, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7891559600830078, + "num_tokens": 443367696.0, + "step": 18393 + }, + { + "epoch": 1.516780737197988, + "grad_norm": 1.8120388984680176, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7730350494384766, + "num_tokens": 443400323.0, + "step": 18394 + }, + { + "epoch": 1.5168631978230396, + "grad_norm": 1.9134658575057983, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7537186145782471, + "num_tokens": 443432669.0, + "step": 18395 + }, + { + "epoch": 1.516945658448091, + "grad_norm": 1.9560261964797974, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.754468560218811, + "num_tokens": 443465324.0, + "step": 18396 + }, + { + "epoch": 1.5170281190731427, + "grad_norm": 2.180058240890503, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.773674726486206, + "num_tokens": 443492208.0, + "step": 18397 + }, + { + "epoch": 1.517110579698194, + "grad_norm": 1.8896809816360474, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7786718606948853, + "num_tokens": 443523939.0, + "step": 18398 + }, + { + "epoch": 1.5171930403232456, + "grad_norm": 1.790051817893982, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7701537013053894, + "num_tokens": 443556800.0, + "step": 18399 + }, + { + "epoch": 1.5172755009482972, + "grad_norm": 1.8601454496383667, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7548550367355347, + "num_tokens": 443590421.0, + "step": 18400 + }, + { + "epoch": 1.5173579615733486, + "grad_norm": 2.061718702316284, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.758259654045105, + "num_tokens": 443618958.0, + "step": 18401 + }, + { + "epoch": 1.5174404221984004, + "grad_norm": 2.3202693462371826, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7509154081344604, + "num_tokens": 443641869.0, + "step": 18402 + }, + { + "epoch": 1.5175228828234517, + "grad_norm": 1.9734277725219727, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.7956282496452332, + "num_tokens": 443667906.0, + "step": 18403 + }, + { + "epoch": 1.5176053434485033, + "grad_norm": 1.9191100597381592, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7780492901802063, + "num_tokens": 443697589.0, + "step": 18404 + }, + { + "epoch": 1.5176878040735549, + "grad_norm": 2.027517795562744, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7644408941268921, + "num_tokens": 443725508.0, + "step": 18405 + }, + { + "epoch": 1.5177702646986064, + "grad_norm": 1.9274194240570068, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7790168523788452, + "num_tokens": 443757028.0, + "step": 18406 + }, + { + "epoch": 1.517852725323658, + "grad_norm": 2.057921886444092, + "learning_rate": 1e-06, + "loss": 0.6058, + "mean_token_accuracy": 0.7848155498504639, + "num_tokens": 443783834.0, + "step": 18407 + }, + { + "epoch": 1.5179351859487094, + "grad_norm": 1.805971384048462, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.782660722732544, + "num_tokens": 443816381.0, + "step": 18408 + }, + { + "epoch": 1.5180176465737611, + "grad_norm": 1.828539490699768, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7569566965103149, + "num_tokens": 443851160.0, + "step": 18409 + }, + { + "epoch": 1.5181001071988125, + "grad_norm": 2.132549285888672, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.748000979423523, + "num_tokens": 443883195.0, + "step": 18410 + }, + { + "epoch": 1.518182567823864, + "grad_norm": 1.81899893283844, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.7943374514579773, + "num_tokens": 443916610.0, + "step": 18411 + }, + { + "epoch": 1.5182650284489156, + "grad_norm": 2.2442917823791504, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7633098363876343, + "num_tokens": 443942009.0, + "step": 18412 + }, + { + "epoch": 1.5183474890739672, + "grad_norm": 1.988869547843933, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7331379055976868, + "num_tokens": 443973606.0, + "step": 18413 + }, + { + "epoch": 1.5184299496990188, + "grad_norm": 2.046815872192383, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7438898682594299, + "num_tokens": 444001706.0, + "step": 18414 + }, + { + "epoch": 1.5185124103240701, + "grad_norm": 1.8797965049743652, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7605194449424744, + "num_tokens": 444034966.0, + "step": 18415 + }, + { + "epoch": 1.518594870949122, + "grad_norm": 1.9118667840957642, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.7548377513885498, + "num_tokens": 444066092.0, + "step": 18416 + }, + { + "epoch": 1.5186773315741733, + "grad_norm": 2.056161403656006, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7471005916595459, + "num_tokens": 444096369.0, + "step": 18417 + }, + { + "epoch": 1.5187597921992249, + "grad_norm": 2.0605475902557373, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7306789755821228, + "num_tokens": 444128357.0, + "step": 18418 + }, + { + "epoch": 1.5188422528242764, + "grad_norm": 1.9280203580856323, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7535654902458191, + "num_tokens": 444159512.0, + "step": 18419 + }, + { + "epoch": 1.518924713449328, + "grad_norm": 1.918176293373108, + "learning_rate": 1e-06, + "loss": 0.5925, + "mean_token_accuracy": 0.804039716720581, + "num_tokens": 444188886.0, + "step": 18420 + }, + { + "epoch": 1.5190071740743796, + "grad_norm": 1.976365327835083, + "learning_rate": 1e-06, + "loss": 0.7619, + "mean_token_accuracy": 0.761772632598877, + "num_tokens": 444216421.0, + "step": 18421 + }, + { + "epoch": 1.519089634699431, + "grad_norm": 2.296062707901001, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7613092064857483, + "num_tokens": 444238449.0, + "step": 18422 + }, + { + "epoch": 1.5191720953244827, + "grad_norm": 2.099494695663452, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7484890818595886, + "num_tokens": 444266045.0, + "step": 18423 + }, + { + "epoch": 1.519254555949534, + "grad_norm": 2.170238733291626, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.747355580329895, + "num_tokens": 444293394.0, + "step": 18424 + }, + { + "epoch": 1.5193370165745856, + "grad_norm": 1.814211130142212, + "learning_rate": 1e-06, + "loss": 0.6155, + "mean_token_accuracy": 0.7998276352882385, + "num_tokens": 444327682.0, + "step": 18425 + }, + { + "epoch": 1.5194194771996372, + "grad_norm": 1.800620198249817, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7527459859848022, + "num_tokens": 444364725.0, + "step": 18426 + }, + { + "epoch": 1.5195019378246886, + "grad_norm": 1.9234509468078613, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.761286199092865, + "num_tokens": 444396069.0, + "step": 18427 + }, + { + "epoch": 1.5195843984497404, + "grad_norm": 2.202648878097534, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7749420404434204, + "num_tokens": 444423357.0, + "step": 18428 + }, + { + "epoch": 1.5196668590747917, + "grad_norm": 2.2427005767822266, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7347922325134277, + "num_tokens": 444452008.0, + "step": 18429 + }, + { + "epoch": 1.5197493196998433, + "grad_norm": 1.995098352432251, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.7840046286582947, + "num_tokens": 444482202.0, + "step": 18430 + }, + { + "epoch": 1.5198317803248949, + "grad_norm": 2.054511785507202, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7545763254165649, + "num_tokens": 444509717.0, + "step": 18431 + }, + { + "epoch": 1.5199142409499464, + "grad_norm": 1.9856202602386475, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7570065259933472, + "num_tokens": 444540812.0, + "step": 18432 + }, + { + "epoch": 1.519996701574998, + "grad_norm": 2.07940936088562, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7617518901824951, + "num_tokens": 444569603.0, + "step": 18433 + }, + { + "epoch": 1.5200791622000493, + "grad_norm": 1.9315052032470703, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7166677713394165, + "num_tokens": 444603793.0, + "step": 18434 + }, + { + "epoch": 1.5201616228251011, + "grad_norm": 2.135878086090088, + "learning_rate": 1e-06, + "loss": 0.5581, + "mean_token_accuracy": 0.8114644289016724, + "num_tokens": 444630729.0, + "step": 18435 + }, + { + "epoch": 1.5202440834501525, + "grad_norm": 2.068620204925537, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7756277322769165, + "num_tokens": 444661194.0, + "step": 18436 + }, + { + "epoch": 1.520326544075204, + "grad_norm": 1.9791415929794312, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.788327693939209, + "num_tokens": 444688534.0, + "step": 18437 + }, + { + "epoch": 1.5204090047002556, + "grad_norm": 1.9304155111312866, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7757561802864075, + "num_tokens": 444721574.0, + "step": 18438 + }, + { + "epoch": 1.5204914653253072, + "grad_norm": 1.9586633443832397, + "learning_rate": 1e-06, + "loss": 0.5771, + "mean_token_accuracy": 0.8090798854827881, + "num_tokens": 444749975.0, + "step": 18439 + }, + { + "epoch": 1.5205739259503588, + "grad_norm": 2.113507032394409, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7215848565101624, + "num_tokens": 444781579.0, + "step": 18440 + }, + { + "epoch": 1.5206563865754101, + "grad_norm": 2.1207945346832275, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.7688102722167969, + "num_tokens": 444808552.0, + "step": 18441 + }, + { + "epoch": 1.520738847200462, + "grad_norm": 1.8104031085968018, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7413923740386963, + "num_tokens": 444844538.0, + "step": 18442 + }, + { + "epoch": 1.5208213078255133, + "grad_norm": 1.9446500539779663, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7143328785896301, + "num_tokens": 444877257.0, + "step": 18443 + }, + { + "epoch": 1.5209037684505649, + "grad_norm": 2.2044811248779297, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.7591152191162109, + "num_tokens": 444902197.0, + "step": 18444 + }, + { + "epoch": 1.5209862290756164, + "grad_norm": 2.128652572631836, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7615530490875244, + "num_tokens": 444930130.0, + "step": 18445 + }, + { + "epoch": 1.5210686897006678, + "grad_norm": 2.209311008453369, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7812483310699463, + "num_tokens": 444955190.0, + "step": 18446 + }, + { + "epoch": 1.5211511503257196, + "grad_norm": 1.9289066791534424, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.7845776677131653, + "num_tokens": 444984591.0, + "step": 18447 + }, + { + "epoch": 1.521233610950771, + "grad_norm": 1.7385333776474, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.7475156784057617, + "num_tokens": 445025319.0, + "step": 18448 + }, + { + "epoch": 1.5213160715758225, + "grad_norm": 1.8194005489349365, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.766776442527771, + "num_tokens": 445057247.0, + "step": 18449 + }, + { + "epoch": 1.521398532200874, + "grad_norm": 1.7884421348571777, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7739125490188599, + "num_tokens": 445092571.0, + "step": 18450 + }, + { + "epoch": 1.5214809928259256, + "grad_norm": 1.8178085088729858, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7709787487983704, + "num_tokens": 445126149.0, + "step": 18451 + }, + { + "epoch": 1.5215634534509772, + "grad_norm": 2.24780535697937, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7674274444580078, + "num_tokens": 445149335.0, + "step": 18452 + }, + { + "epoch": 1.5216459140760286, + "grad_norm": 1.8550786972045898, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7529782056808472, + "num_tokens": 445183674.0, + "step": 18453 + }, + { + "epoch": 1.5217283747010804, + "grad_norm": 2.103914499282837, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.7839146852493286, + "num_tokens": 445209746.0, + "step": 18454 + }, + { + "epoch": 1.5218108353261317, + "grad_norm": 1.9498850107192993, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7292062044143677, + "num_tokens": 445243021.0, + "step": 18455 + }, + { + "epoch": 1.5218932959511833, + "grad_norm": 1.9353387355804443, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7828604578971863, + "num_tokens": 445276693.0, + "step": 18456 + }, + { + "epoch": 1.5219757565762349, + "grad_norm": 2.029024600982666, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7623401284217834, + "num_tokens": 445306013.0, + "step": 18457 + }, + { + "epoch": 1.5220582172012864, + "grad_norm": 1.8255280256271362, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7742042541503906, + "num_tokens": 445337078.0, + "step": 18458 + }, + { + "epoch": 1.522140677826338, + "grad_norm": 2.030853271484375, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7679711580276489, + "num_tokens": 445366328.0, + "step": 18459 + }, + { + "epoch": 1.5222231384513893, + "grad_norm": 1.9162132740020752, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7861279249191284, + "num_tokens": 445397648.0, + "step": 18460 + }, + { + "epoch": 1.5223055990764411, + "grad_norm": 2.047065019607544, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7513689994812012, + "num_tokens": 445424853.0, + "step": 18461 + }, + { + "epoch": 1.5223880597014925, + "grad_norm": 1.7616068124771118, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7864400148391724, + "num_tokens": 445461898.0, + "step": 18462 + }, + { + "epoch": 1.522470520326544, + "grad_norm": 1.8043028116226196, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7493110299110413, + "num_tokens": 445497600.0, + "step": 18463 + }, + { + "epoch": 1.5225529809515956, + "grad_norm": 1.6682829856872559, + "learning_rate": 1e-06, + "loss": 0.5164, + "mean_token_accuracy": 0.8185160756111145, + "num_tokens": 445534542.0, + "step": 18464 + }, + { + "epoch": 1.522635441576647, + "grad_norm": 1.9562146663665771, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7663959860801697, + "num_tokens": 445563268.0, + "step": 18465 + }, + { + "epoch": 1.5227179022016988, + "grad_norm": 2.2189862728118896, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.759819746017456, + "num_tokens": 445589096.0, + "step": 18466 + }, + { + "epoch": 1.5228003628267501, + "grad_norm": 2.0703165531158447, + "learning_rate": 1e-06, + "loss": 0.6132, + "mean_token_accuracy": 0.7986130714416504, + "num_tokens": 445615729.0, + "step": 18467 + }, + { + "epoch": 1.522882823451802, + "grad_norm": 1.8601986169815063, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7608351707458496, + "num_tokens": 445650180.0, + "step": 18468 + }, + { + "epoch": 1.5229652840768533, + "grad_norm": 2.2083778381347656, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.752789318561554, + "num_tokens": 445677451.0, + "step": 18469 + }, + { + "epoch": 1.5230477447019048, + "grad_norm": 2.1929986476898193, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7638123035430908, + "num_tokens": 445703629.0, + "step": 18470 + }, + { + "epoch": 1.5231302053269564, + "grad_norm": 2.1688289642333984, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7612804174423218, + "num_tokens": 445729440.0, + "step": 18471 + }, + { + "epoch": 1.5232126659520078, + "grad_norm": 2.212669610977173, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7739412188529968, + "num_tokens": 445752839.0, + "step": 18472 + }, + { + "epoch": 1.5232951265770596, + "grad_norm": 2.285569190979004, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7277559041976929, + "num_tokens": 445776167.0, + "step": 18473 + }, + { + "epoch": 1.523377587202111, + "grad_norm": 2.015545606613159, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.7575222253799438, + "num_tokens": 445805749.0, + "step": 18474 + }, + { + "epoch": 1.5234600478271625, + "grad_norm": 1.9309120178222656, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7522684335708618, + "num_tokens": 445838638.0, + "step": 18475 + }, + { + "epoch": 1.523542508452214, + "grad_norm": 1.7255085706710815, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.745452880859375, + "num_tokens": 445877966.0, + "step": 18476 + }, + { + "epoch": 1.5236249690772656, + "grad_norm": 1.9138644933700562, + "learning_rate": 1e-06, + "loss": 0.5608, + "mean_token_accuracy": 0.8106697797775269, + "num_tokens": 445907869.0, + "step": 18477 + }, + { + "epoch": 1.5237074297023172, + "grad_norm": 2.2622292041778564, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7655280232429504, + "num_tokens": 445933672.0, + "step": 18478 + }, + { + "epoch": 1.5237898903273686, + "grad_norm": 1.9143133163452148, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.780921220779419, + "num_tokens": 445962415.0, + "step": 18479 + }, + { + "epoch": 1.5238723509524204, + "grad_norm": 1.8660985231399536, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7620500326156616, + "num_tokens": 445994704.0, + "step": 18480 + }, + { + "epoch": 1.5239548115774717, + "grad_norm": 2.1570053100585938, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7664510011672974, + "num_tokens": 446021806.0, + "step": 18481 + }, + { + "epoch": 1.5240372722025233, + "grad_norm": 1.877025842666626, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.7815510034561157, + "num_tokens": 446056475.0, + "step": 18482 + }, + { + "epoch": 1.5241197328275748, + "grad_norm": 1.8372446298599243, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7733294367790222, + "num_tokens": 446088612.0, + "step": 18483 + }, + { + "epoch": 1.5242021934526264, + "grad_norm": 2.0322349071502686, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7145616412162781, + "num_tokens": 446121449.0, + "step": 18484 + }, + { + "epoch": 1.524284654077678, + "grad_norm": 1.864549160003662, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7790187001228333, + "num_tokens": 446156694.0, + "step": 18485 + }, + { + "epoch": 1.5243671147027293, + "grad_norm": 2.138843297958374, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.7526243925094604, + "num_tokens": 446186210.0, + "step": 18486 + }, + { + "epoch": 1.5244495753277811, + "grad_norm": 2.0038130283355713, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.778373122215271, + "num_tokens": 446215102.0, + "step": 18487 + }, + { + "epoch": 1.5245320359528325, + "grad_norm": 2.0831358432769775, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7726404666900635, + "num_tokens": 446243806.0, + "step": 18488 + }, + { + "epoch": 1.524614496577884, + "grad_norm": 1.9786852598190308, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7604628801345825, + "num_tokens": 446272121.0, + "step": 18489 + }, + { + "epoch": 1.5246969572029356, + "grad_norm": 1.9504592418670654, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7375127077102661, + "num_tokens": 446303168.0, + "step": 18490 + }, + { + "epoch": 1.524779417827987, + "grad_norm": 1.7072951793670654, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.795312762260437, + "num_tokens": 446338233.0, + "step": 18491 + }, + { + "epoch": 1.5248618784530388, + "grad_norm": 1.9641038179397583, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7304978370666504, + "num_tokens": 446374286.0, + "step": 18492 + }, + { + "epoch": 1.5249443390780901, + "grad_norm": 2.1566267013549805, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.785115122795105, + "num_tokens": 446400568.0, + "step": 18493 + }, + { + "epoch": 1.5250267997031417, + "grad_norm": 1.5875632762908936, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.790451169013977, + "num_tokens": 446443884.0, + "step": 18494 + }, + { + "epoch": 1.5251092603281933, + "grad_norm": 1.9993282556533813, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7104488611221313, + "num_tokens": 446475849.0, + "step": 18495 + }, + { + "epoch": 1.5251917209532448, + "grad_norm": 1.863253116607666, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7579220533370972, + "num_tokens": 446511102.0, + "step": 18496 + }, + { + "epoch": 1.5252741815782964, + "grad_norm": 1.9462326765060425, + "learning_rate": 1e-06, + "loss": 0.7712, + "mean_token_accuracy": 0.7552089095115662, + "num_tokens": 446542245.0, + "step": 18497 + }, + { + "epoch": 1.5253566422033478, + "grad_norm": 1.7762179374694824, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7620792984962463, + "num_tokens": 446578248.0, + "step": 18498 + }, + { + "epoch": 1.5254391028283996, + "grad_norm": 2.035900354385376, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7540326714515686, + "num_tokens": 446608279.0, + "step": 18499 + }, + { + "epoch": 1.525521563453451, + "grad_norm": 1.8334966897964478, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.783595085144043, + "num_tokens": 446638845.0, + "step": 18500 + }, + { + "epoch": 1.5256040240785025, + "grad_norm": 2.3543636798858643, + "learning_rate": 1e-06, + "loss": 0.7623, + "mean_token_accuracy": 0.7541990280151367, + "num_tokens": 446661292.0, + "step": 18501 + }, + { + "epoch": 1.525686484703554, + "grad_norm": 2.0397605895996094, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7362804412841797, + "num_tokens": 446692578.0, + "step": 18502 + }, + { + "epoch": 1.5257689453286056, + "grad_norm": 2.032148599624634, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7600226998329163, + "num_tokens": 446723414.0, + "step": 18503 + }, + { + "epoch": 1.5258514059536572, + "grad_norm": 2.01027774810791, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7670137286186218, + "num_tokens": 446753368.0, + "step": 18504 + }, + { + "epoch": 1.5259338665787086, + "grad_norm": 1.9102203845977783, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7724248170852661, + "num_tokens": 446784709.0, + "step": 18505 + }, + { + "epoch": 1.5260163272037603, + "grad_norm": 2.1680736541748047, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7417422533035278, + "num_tokens": 446811908.0, + "step": 18506 + }, + { + "epoch": 1.5260987878288117, + "grad_norm": 2.0438530445098877, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7672767639160156, + "num_tokens": 446838572.0, + "step": 18507 + }, + { + "epoch": 1.5261812484538633, + "grad_norm": 2.0166759490966797, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7619797587394714, + "num_tokens": 446868201.0, + "step": 18508 + }, + { + "epoch": 1.5262637090789148, + "grad_norm": 2.111830234527588, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7392184734344482, + "num_tokens": 446894927.0, + "step": 18509 + }, + { + "epoch": 1.5263461697039662, + "grad_norm": 1.8652604818344116, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7570541501045227, + "num_tokens": 446932013.0, + "step": 18510 + }, + { + "epoch": 1.526428630329018, + "grad_norm": 2.0575809478759766, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.7876967191696167, + "num_tokens": 446958778.0, + "step": 18511 + }, + { + "epoch": 1.5265110909540693, + "grad_norm": 1.9346427917480469, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7469276189804077, + "num_tokens": 446990610.0, + "step": 18512 + }, + { + "epoch": 1.5265935515791211, + "grad_norm": 1.9105123281478882, + "learning_rate": 1e-06, + "loss": 0.7669, + "mean_token_accuracy": 0.7538068294525146, + "num_tokens": 447024745.0, + "step": 18513 + }, + { + "epoch": 1.5266760122041725, + "grad_norm": 2.2413318157196045, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7355340719223022, + "num_tokens": 447055690.0, + "step": 18514 + }, + { + "epoch": 1.526758472829224, + "grad_norm": 2.0304832458496094, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7336555123329163, + "num_tokens": 447085571.0, + "step": 18515 + }, + { + "epoch": 1.5268409334542756, + "grad_norm": 2.078263759613037, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7611031532287598, + "num_tokens": 447110575.0, + "step": 18516 + }, + { + "epoch": 1.526923394079327, + "grad_norm": 1.6769939661026, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.782664954662323, + "num_tokens": 447149300.0, + "step": 18517 + }, + { + "epoch": 1.5270058547043788, + "grad_norm": 1.9949744939804077, + "learning_rate": 1e-06, + "loss": 0.7585, + "mean_token_accuracy": 0.7589836716651917, + "num_tokens": 447179814.0, + "step": 18518 + }, + { + "epoch": 1.5270883153294301, + "grad_norm": 1.720449686050415, + "learning_rate": 1e-06, + "loss": 0.5888, + "mean_token_accuracy": 0.7985457181930542, + "num_tokens": 447217499.0, + "step": 18519 + }, + { + "epoch": 1.5271707759544817, + "grad_norm": 1.8845614194869995, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7390785217285156, + "num_tokens": 447252628.0, + "step": 18520 + }, + { + "epoch": 1.5272532365795333, + "grad_norm": 2.0854761600494385, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7133892774581909, + "num_tokens": 447281295.0, + "step": 18521 + }, + { + "epoch": 1.5273356972045848, + "grad_norm": 2.2687735557556152, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7290194034576416, + "num_tokens": 447308059.0, + "step": 18522 + }, + { + "epoch": 1.5274181578296364, + "grad_norm": 1.8409017324447632, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7793161273002625, + "num_tokens": 447340594.0, + "step": 18523 + }, + { + "epoch": 1.5275006184546878, + "grad_norm": 1.7792696952819824, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.801078200340271, + "num_tokens": 447377577.0, + "step": 18524 + }, + { + "epoch": 1.5275830790797396, + "grad_norm": 2.080881357192993, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7659516334533691, + "num_tokens": 447405289.0, + "step": 18525 + }, + { + "epoch": 1.527665539704791, + "grad_norm": 1.8331236839294434, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.742030143737793, + "num_tokens": 447438306.0, + "step": 18526 + }, + { + "epoch": 1.5277480003298425, + "grad_norm": 2.0006251335144043, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7390655875205994, + "num_tokens": 447471911.0, + "step": 18527 + }, + { + "epoch": 1.527830460954894, + "grad_norm": 2.2083547115325928, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7563724517822266, + "num_tokens": 447495853.0, + "step": 18528 + }, + { + "epoch": 1.5279129215799456, + "grad_norm": 1.9315658807754517, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7584574222564697, + "num_tokens": 447529548.0, + "step": 18529 + }, + { + "epoch": 1.5279953822049972, + "grad_norm": 1.9412729740142822, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.7984820604324341, + "num_tokens": 447560137.0, + "step": 18530 + }, + { + "epoch": 1.5280778428300485, + "grad_norm": 2.1855814456939697, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7229206562042236, + "num_tokens": 447587031.0, + "step": 18531 + }, + { + "epoch": 1.5281603034551003, + "grad_norm": 2.204371213912964, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7675464749336243, + "num_tokens": 447615083.0, + "step": 18532 + }, + { + "epoch": 1.5282427640801517, + "grad_norm": 2.4088964462280273, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7140527963638306, + "num_tokens": 447641346.0, + "step": 18533 + }, + { + "epoch": 1.5283252247052033, + "grad_norm": 1.8677507638931274, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7457571029663086, + "num_tokens": 447676112.0, + "step": 18534 + }, + { + "epoch": 1.5284076853302548, + "grad_norm": 1.9046880006790161, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.743624746799469, + "num_tokens": 447711690.0, + "step": 18535 + }, + { + "epoch": 1.5284901459553062, + "grad_norm": 1.9910259246826172, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7337056994438171, + "num_tokens": 447741722.0, + "step": 18536 + }, + { + "epoch": 1.528572606580358, + "grad_norm": 2.1375322341918945, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7811522483825684, + "num_tokens": 447768520.0, + "step": 18537 + }, + { + "epoch": 1.5286550672054093, + "grad_norm": 2.048936367034912, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7520670890808105, + "num_tokens": 447798018.0, + "step": 18538 + }, + { + "epoch": 1.528737527830461, + "grad_norm": 1.8384712934494019, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7941305041313171, + "num_tokens": 447830459.0, + "step": 18539 + }, + { + "epoch": 1.5288199884555125, + "grad_norm": 2.389768362045288, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7560379505157471, + "num_tokens": 447852873.0, + "step": 18540 + }, + { + "epoch": 1.528902449080564, + "grad_norm": 1.8720636367797852, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8010287880897522, + "num_tokens": 447882598.0, + "step": 18541 + }, + { + "epoch": 1.5289849097056156, + "grad_norm": 1.7308852672576904, + "learning_rate": 1e-06, + "loss": 0.5093, + "mean_token_accuracy": 0.8281559348106384, + "num_tokens": 447918898.0, + "step": 18542 + }, + { + "epoch": 1.529067370330667, + "grad_norm": 1.910825252532959, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7260281443595886, + "num_tokens": 447952339.0, + "step": 18543 + }, + { + "epoch": 1.5291498309557188, + "grad_norm": 1.7973599433898926, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7797882556915283, + "num_tokens": 447985680.0, + "step": 18544 + }, + { + "epoch": 1.5292322915807701, + "grad_norm": 1.875482439994812, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.7991245985031128, + "num_tokens": 448016058.0, + "step": 18545 + }, + { + "epoch": 1.5293147522058217, + "grad_norm": 2.0461764335632324, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7503776550292969, + "num_tokens": 448049211.0, + "step": 18546 + }, + { + "epoch": 1.5293972128308733, + "grad_norm": 2.044059991836548, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7726632356643677, + "num_tokens": 448078169.0, + "step": 18547 + }, + { + "epoch": 1.5294796734559248, + "grad_norm": 1.9992963075637817, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7542637586593628, + "num_tokens": 448111297.0, + "step": 18548 + }, + { + "epoch": 1.5295621340809764, + "grad_norm": 1.93581223487854, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7318882942199707, + "num_tokens": 448142754.0, + "step": 18549 + }, + { + "epoch": 1.5296445947060278, + "grad_norm": 2.1679327487945557, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.785759687423706, + "num_tokens": 448167544.0, + "step": 18550 + }, + { + "epoch": 1.5297270553310796, + "grad_norm": 2.095088005065918, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7433736324310303, + "num_tokens": 448195029.0, + "step": 18551 + }, + { + "epoch": 1.529809515956131, + "grad_norm": 1.7046724557876587, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.785352349281311, + "num_tokens": 448231231.0, + "step": 18552 + }, + { + "epoch": 1.5298919765811825, + "grad_norm": 2.0353429317474365, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7871052622795105, + "num_tokens": 448262189.0, + "step": 18553 + }, + { + "epoch": 1.529974437206234, + "grad_norm": 1.7387042045593262, + "learning_rate": 1e-06, + "loss": 0.5523, + "mean_token_accuracy": 0.8206438422203064, + "num_tokens": 448297753.0, + "step": 18554 + }, + { + "epoch": 1.5300568978312854, + "grad_norm": 2.065981388092041, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.736948549747467, + "num_tokens": 448328575.0, + "step": 18555 + }, + { + "epoch": 1.5301393584563372, + "grad_norm": 2.099729299545288, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.7500417232513428, + "num_tokens": 448356974.0, + "step": 18556 + }, + { + "epoch": 1.5302218190813885, + "grad_norm": 2.050933361053467, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7521481513977051, + "num_tokens": 448386523.0, + "step": 18557 + }, + { + "epoch": 1.5303042797064403, + "grad_norm": 1.8370387554168701, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7648098468780518, + "num_tokens": 448420842.0, + "step": 18558 + }, + { + "epoch": 1.5303867403314917, + "grad_norm": 1.8332436084747314, + "learning_rate": 1e-06, + "loss": 0.4523, + "mean_token_accuracy": 0.8359348177909851, + "num_tokens": 448447747.0, + "step": 18559 + }, + { + "epoch": 1.5304692009565433, + "grad_norm": 2.0122530460357666, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7391777634620667, + "num_tokens": 448479699.0, + "step": 18560 + }, + { + "epoch": 1.5305516615815948, + "grad_norm": 1.8555846214294434, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7138057947158813, + "num_tokens": 448513824.0, + "step": 18561 + }, + { + "epoch": 1.5306341222066462, + "grad_norm": 1.9957531690597534, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7631558775901794, + "num_tokens": 448545614.0, + "step": 18562 + }, + { + "epoch": 1.530716582831698, + "grad_norm": 2.0745136737823486, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7429714798927307, + "num_tokens": 448573065.0, + "step": 18563 + }, + { + "epoch": 1.5307990434567493, + "grad_norm": 1.9747285842895508, + "learning_rate": 1e-06, + "loss": 0.5636, + "mean_token_accuracy": 0.7987993955612183, + "num_tokens": 448599865.0, + "step": 18564 + }, + { + "epoch": 1.530881504081801, + "grad_norm": 2.4085381031036377, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7386224865913391, + "num_tokens": 448623013.0, + "step": 18565 + }, + { + "epoch": 1.5309639647068525, + "grad_norm": 2.187502384185791, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.743722677230835, + "num_tokens": 448648597.0, + "step": 18566 + }, + { + "epoch": 1.531046425331904, + "grad_norm": 2.02620530128479, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7530508637428284, + "num_tokens": 448677640.0, + "step": 18567 + }, + { + "epoch": 1.5311288859569556, + "grad_norm": 1.9200769662857056, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7666395902633667, + "num_tokens": 448710839.0, + "step": 18568 + }, + { + "epoch": 1.531211346582007, + "grad_norm": 1.8862601518630981, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.736179530620575, + "num_tokens": 448746158.0, + "step": 18569 + }, + { + "epoch": 1.5312938072070588, + "grad_norm": 1.9388611316680908, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7733997702598572, + "num_tokens": 448776810.0, + "step": 18570 + }, + { + "epoch": 1.5313762678321101, + "grad_norm": 1.9338492155075073, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7759062051773071, + "num_tokens": 448806570.0, + "step": 18571 + }, + { + "epoch": 1.5314587284571617, + "grad_norm": 2.1385653018951416, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7359036803245544, + "num_tokens": 448834124.0, + "step": 18572 + }, + { + "epoch": 1.5315411890822133, + "grad_norm": 1.9290635585784912, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7811293005943298, + "num_tokens": 448867425.0, + "step": 18573 + }, + { + "epoch": 1.5316236497072648, + "grad_norm": 1.9396876096725464, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7230913639068604, + "num_tokens": 448902710.0, + "step": 18574 + }, + { + "epoch": 1.5317061103323164, + "grad_norm": 1.9051915407180786, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7733309268951416, + "num_tokens": 448934766.0, + "step": 18575 + }, + { + "epoch": 1.5317885709573678, + "grad_norm": 1.9517172574996948, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7468226552009583, + "num_tokens": 448968206.0, + "step": 18576 + }, + { + "epoch": 1.5318710315824196, + "grad_norm": 2.0054709911346436, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7570948004722595, + "num_tokens": 448997856.0, + "step": 18577 + }, + { + "epoch": 1.531953492207471, + "grad_norm": 2.002962350845337, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.761615514755249, + "num_tokens": 449026358.0, + "step": 18578 + }, + { + "epoch": 1.5320359528325225, + "grad_norm": 1.9066885709762573, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.745924174785614, + "num_tokens": 449058034.0, + "step": 18579 + }, + { + "epoch": 1.532118413457574, + "grad_norm": 2.113101005554199, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.777264416217804, + "num_tokens": 449084732.0, + "step": 18580 + }, + { + "epoch": 1.5322008740826254, + "grad_norm": 1.841139793395996, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7300207018852234, + "num_tokens": 449118553.0, + "step": 18581 + }, + { + "epoch": 1.5322833347076772, + "grad_norm": 2.1574056148529053, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7542800307273865, + "num_tokens": 449144726.0, + "step": 18582 + }, + { + "epoch": 1.5323657953327285, + "grad_norm": 1.938032627105713, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7522590756416321, + "num_tokens": 449177830.0, + "step": 18583 + }, + { + "epoch": 1.5324482559577801, + "grad_norm": 1.9034650325775146, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.7891401648521423, + "num_tokens": 449207914.0, + "step": 18584 + }, + { + "epoch": 1.5325307165828317, + "grad_norm": 2.2386858463287354, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7527210116386414, + "num_tokens": 449234261.0, + "step": 18585 + }, + { + "epoch": 1.5326131772078833, + "grad_norm": 2.1623375415802, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7753049731254578, + "num_tokens": 449263890.0, + "step": 18586 + }, + { + "epoch": 1.5326956378329348, + "grad_norm": 2.140141248703003, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7146548628807068, + "num_tokens": 449292422.0, + "step": 18587 + }, + { + "epoch": 1.5327780984579862, + "grad_norm": 2.0101318359375, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7590317726135254, + "num_tokens": 449323074.0, + "step": 18588 + }, + { + "epoch": 1.532860559083038, + "grad_norm": 2.09881591796875, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7619919180870056, + "num_tokens": 449350963.0, + "step": 18589 + }, + { + "epoch": 1.5329430197080893, + "grad_norm": 2.131669521331787, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7646024227142334, + "num_tokens": 449375175.0, + "step": 18590 + }, + { + "epoch": 1.533025480333141, + "grad_norm": 2.3379926681518555, + "learning_rate": 1e-06, + "loss": 0.7656, + "mean_token_accuracy": 0.7453280091285706, + "num_tokens": 449404513.0, + "step": 18591 + }, + { + "epoch": 1.5331079409581925, + "grad_norm": 1.987945556640625, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.757327139377594, + "num_tokens": 449435413.0, + "step": 18592 + }, + { + "epoch": 1.533190401583244, + "grad_norm": 1.7976548671722412, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7535585165023804, + "num_tokens": 449471680.0, + "step": 18593 + }, + { + "epoch": 1.5332728622082956, + "grad_norm": 1.7420759201049805, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7727746367454529, + "num_tokens": 449507320.0, + "step": 18594 + }, + { + "epoch": 1.533355322833347, + "grad_norm": 1.7643327713012695, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.7908095121383667, + "num_tokens": 449542305.0, + "step": 18595 + }, + { + "epoch": 1.5334377834583988, + "grad_norm": 1.9651648998260498, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7589625120162964, + "num_tokens": 449570908.0, + "step": 18596 + }, + { + "epoch": 1.5335202440834501, + "grad_norm": 2.1993563175201416, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7138267755508423, + "num_tokens": 449597956.0, + "step": 18597 + }, + { + "epoch": 1.5336027047085017, + "grad_norm": 2.017673969268799, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7298707365989685, + "num_tokens": 449626540.0, + "step": 18598 + }, + { + "epoch": 1.5336851653335533, + "grad_norm": 1.8366179466247559, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.7910498380661011, + "num_tokens": 449656170.0, + "step": 18599 + }, + { + "epoch": 1.5337676259586046, + "grad_norm": 2.1378135681152344, + "learning_rate": 1e-06, + "loss": 0.7623, + "mean_token_accuracy": 0.7542587518692017, + "num_tokens": 449682238.0, + "step": 18600 + }, + { + "epoch": 1.5338500865836564, + "grad_norm": 2.011098861694336, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7655803561210632, + "num_tokens": 449711864.0, + "step": 18601 + }, + { + "epoch": 1.5339325472087078, + "grad_norm": 1.9450955390930176, + "learning_rate": 1e-06, + "loss": 0.7623, + "mean_token_accuracy": 0.752700686454773, + "num_tokens": 449743568.0, + "step": 18602 + }, + { + "epoch": 1.5340150078337593, + "grad_norm": 2.2624080181121826, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7241377830505371, + "num_tokens": 449769451.0, + "step": 18603 + }, + { + "epoch": 1.534097468458811, + "grad_norm": 2.1690330505371094, + "learning_rate": 1e-06, + "loss": 0.7704, + "mean_token_accuracy": 0.7463490962982178, + "num_tokens": 449796819.0, + "step": 18604 + }, + { + "epoch": 1.5341799290838625, + "grad_norm": 2.282503843307495, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7171751260757446, + "num_tokens": 449824614.0, + "step": 18605 + }, + { + "epoch": 1.534262389708914, + "grad_norm": 2.2059693336486816, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.755841851234436, + "num_tokens": 449849676.0, + "step": 18606 + }, + { + "epoch": 1.5343448503339654, + "grad_norm": 1.8244154453277588, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.7949387431144714, + "num_tokens": 449882336.0, + "step": 18607 + }, + { + "epoch": 1.5344273109590172, + "grad_norm": 2.016418218612671, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7858278751373291, + "num_tokens": 449910208.0, + "step": 18608 + }, + { + "epoch": 1.5345097715840685, + "grad_norm": 2.180478096008301, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7360539436340332, + "num_tokens": 449939362.0, + "step": 18609 + }, + { + "epoch": 1.53459223220912, + "grad_norm": 2.0009822845458984, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7582852840423584, + "num_tokens": 449971310.0, + "step": 18610 + }, + { + "epoch": 1.5346746928341717, + "grad_norm": 1.8875173330307007, + "learning_rate": 1e-06, + "loss": 0.5901, + "mean_token_accuracy": 0.8070366382598877, + "num_tokens": 449999776.0, + "step": 18611 + }, + { + "epoch": 1.5347571534592233, + "grad_norm": 1.8294677734375, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7449034452438354, + "num_tokens": 450035746.0, + "step": 18612 + }, + { + "epoch": 1.5348396140842748, + "grad_norm": 1.731181263923645, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.7789204120635986, + "num_tokens": 450071657.0, + "step": 18613 + }, + { + "epoch": 1.5349220747093262, + "grad_norm": 2.4215917587280273, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7274379730224609, + "num_tokens": 450094153.0, + "step": 18614 + }, + { + "epoch": 1.535004535334378, + "grad_norm": 2.2159371376037598, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7640941143035889, + "num_tokens": 450119137.0, + "step": 18615 + }, + { + "epoch": 1.5350869959594293, + "grad_norm": 2.008357286453247, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7713360786437988, + "num_tokens": 450149441.0, + "step": 18616 + }, + { + "epoch": 1.535169456584481, + "grad_norm": 2.113665819168091, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.7916127443313599, + "num_tokens": 450177005.0, + "step": 18617 + }, + { + "epoch": 1.5352519172095325, + "grad_norm": 2.0574278831481934, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7493442296981812, + "num_tokens": 450207312.0, + "step": 18618 + }, + { + "epoch": 1.535334377834584, + "grad_norm": 2.2021324634552, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7413886189460754, + "num_tokens": 450232159.0, + "step": 18619 + }, + { + "epoch": 1.5354168384596356, + "grad_norm": 2.0380897521972656, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7795857191085815, + "num_tokens": 450259935.0, + "step": 18620 + }, + { + "epoch": 1.535499299084687, + "grad_norm": 2.363454818725586, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.7866731882095337, + "num_tokens": 450293272.0, + "step": 18621 + }, + { + "epoch": 1.5355817597097388, + "grad_norm": 1.9383454322814941, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7681964039802551, + "num_tokens": 450323562.0, + "step": 18622 + }, + { + "epoch": 1.53566422033479, + "grad_norm": 2.1638131141662598, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7430660724639893, + "num_tokens": 450350611.0, + "step": 18623 + }, + { + "epoch": 1.5357466809598417, + "grad_norm": 1.8141252994537354, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7234635353088379, + "num_tokens": 450387290.0, + "step": 18624 + }, + { + "epoch": 1.5358291415848933, + "grad_norm": 2.4514076709747314, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.6985092163085938, + "num_tokens": 450411099.0, + "step": 18625 + }, + { + "epoch": 1.5359116022099446, + "grad_norm": 1.80826997756958, + "learning_rate": 1e-06, + "loss": 0.5299, + "mean_token_accuracy": 0.8217982053756714, + "num_tokens": 450442364.0, + "step": 18626 + }, + { + "epoch": 1.5359940628349964, + "grad_norm": 2.096738338470459, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7621554732322693, + "num_tokens": 450471434.0, + "step": 18627 + }, + { + "epoch": 1.5360765234600477, + "grad_norm": 1.879338026046753, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7452162504196167, + "num_tokens": 450503486.0, + "step": 18628 + }, + { + "epoch": 1.5361589840850993, + "grad_norm": 2.2914087772369385, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7600001096725464, + "num_tokens": 450527028.0, + "step": 18629 + }, + { + "epoch": 1.536241444710151, + "grad_norm": 1.7811524868011475, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.761326789855957, + "num_tokens": 450562250.0, + "step": 18630 + }, + { + "epoch": 1.5363239053352025, + "grad_norm": 1.8657104969024658, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7795830965042114, + "num_tokens": 450595683.0, + "step": 18631 + }, + { + "epoch": 1.536406365960254, + "grad_norm": 1.8720475435256958, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7786120176315308, + "num_tokens": 450627390.0, + "step": 18632 + }, + { + "epoch": 1.5364888265853054, + "grad_norm": 2.1330695152282715, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.7498868703842163, + "num_tokens": 450654606.0, + "step": 18633 + }, + { + "epoch": 1.5365712872103572, + "grad_norm": 2.106370449066162, + "learning_rate": 1e-06, + "loss": 0.5999, + "mean_token_accuracy": 0.79932701587677, + "num_tokens": 450681279.0, + "step": 18634 + }, + { + "epoch": 1.5366537478354085, + "grad_norm": 1.893920660018921, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7633569836616516, + "num_tokens": 450716113.0, + "step": 18635 + }, + { + "epoch": 1.53673620846046, + "grad_norm": 2.244054079055786, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.719605565071106, + "num_tokens": 450742563.0, + "step": 18636 + }, + { + "epoch": 1.5368186690855117, + "grad_norm": 2.3204288482666016, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7450572848320007, + "num_tokens": 450766424.0, + "step": 18637 + }, + { + "epoch": 1.5369011297105633, + "grad_norm": 1.939080834388733, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.785702109336853, + "num_tokens": 450799209.0, + "step": 18638 + }, + { + "epoch": 1.5369835903356148, + "grad_norm": 1.9559316635131836, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7758442163467407, + "num_tokens": 450827637.0, + "step": 18639 + }, + { + "epoch": 1.5370660509606662, + "grad_norm": 1.9322422742843628, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7841655015945435, + "num_tokens": 450855845.0, + "step": 18640 + }, + { + "epoch": 1.537148511585718, + "grad_norm": 2.0541579723358154, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7744302749633789, + "num_tokens": 450883154.0, + "step": 18641 + }, + { + "epoch": 1.5372309722107693, + "grad_norm": 1.8823648691177368, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.7856131792068481, + "num_tokens": 450921204.0, + "step": 18642 + }, + { + "epoch": 1.537313432835821, + "grad_norm": 1.8356636762619019, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7656901478767395, + "num_tokens": 450954842.0, + "step": 18643 + }, + { + "epoch": 1.5373958934608725, + "grad_norm": 2.1460886001586914, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7438740730285645, + "num_tokens": 450980617.0, + "step": 18644 + }, + { + "epoch": 1.5374783540859238, + "grad_norm": 2.1554384231567383, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7665848135948181, + "num_tokens": 451004653.0, + "step": 18645 + }, + { + "epoch": 1.5375608147109756, + "grad_norm": 2.1428701877593994, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7519195079803467, + "num_tokens": 451033537.0, + "step": 18646 + }, + { + "epoch": 1.537643275336027, + "grad_norm": 1.9139288663864136, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7358226776123047, + "num_tokens": 451067368.0, + "step": 18647 + }, + { + "epoch": 1.5377257359610785, + "grad_norm": 1.8331618309020996, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7683768272399902, + "num_tokens": 451101520.0, + "step": 18648 + }, + { + "epoch": 1.53780819658613, + "grad_norm": 2.087229013442993, + "learning_rate": 1e-06, + "loss": 0.5716, + "mean_token_accuracy": 0.8046881556510925, + "num_tokens": 451127711.0, + "step": 18649 + }, + { + "epoch": 1.5378906572111817, + "grad_norm": 1.9237380027770996, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7683227062225342, + "num_tokens": 451159763.0, + "step": 18650 + }, + { + "epoch": 1.5379731178362332, + "grad_norm": 1.7241151332855225, + "learning_rate": 1e-06, + "loss": 0.5576, + "mean_token_accuracy": 0.8106762170791626, + "num_tokens": 451192700.0, + "step": 18651 + }, + { + "epoch": 1.5380555784612846, + "grad_norm": 1.9339927434921265, + "learning_rate": 1e-06, + "loss": 0.5923, + "mean_token_accuracy": 0.8035339117050171, + "num_tokens": 451220293.0, + "step": 18652 + }, + { + "epoch": 1.5381380390863364, + "grad_norm": 2.2395548820495605, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7419671416282654, + "num_tokens": 451244579.0, + "step": 18653 + }, + { + "epoch": 1.5382204997113877, + "grad_norm": 2.149238109588623, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.7783122658729553, + "num_tokens": 451272519.0, + "step": 18654 + }, + { + "epoch": 1.5383029603364393, + "grad_norm": 1.804702877998352, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7699131965637207, + "num_tokens": 451307719.0, + "step": 18655 + }, + { + "epoch": 1.538385420961491, + "grad_norm": 2.134645700454712, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7194101810455322, + "num_tokens": 451334643.0, + "step": 18656 + }, + { + "epoch": 1.5384678815865425, + "grad_norm": 2.0892293453216553, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.7728654146194458, + "num_tokens": 451362794.0, + "step": 18657 + }, + { + "epoch": 1.538550342211594, + "grad_norm": 2.0897743701934814, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7717427611351013, + "num_tokens": 451389183.0, + "step": 18658 + }, + { + "epoch": 1.5386328028366454, + "grad_norm": 1.7582998275756836, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7747001647949219, + "num_tokens": 451423260.0, + "step": 18659 + }, + { + "epoch": 1.5387152634616972, + "grad_norm": 1.9336522817611694, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7423768043518066, + "num_tokens": 451453707.0, + "step": 18660 + }, + { + "epoch": 1.5387977240867485, + "grad_norm": 2.036004066467285, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7669497728347778, + "num_tokens": 451481749.0, + "step": 18661 + }, + { + "epoch": 1.5388801847118, + "grad_norm": 1.7553457021713257, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.7875672578811646, + "num_tokens": 451515036.0, + "step": 18662 + }, + { + "epoch": 1.5389626453368517, + "grad_norm": 1.90568208694458, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7661367654800415, + "num_tokens": 451547414.0, + "step": 18663 + }, + { + "epoch": 1.539045105961903, + "grad_norm": 1.7641586065292358, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.7875750064849854, + "num_tokens": 451582346.0, + "step": 18664 + }, + { + "epoch": 1.5391275665869548, + "grad_norm": 1.921007513999939, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7327041029930115, + "num_tokens": 451616258.0, + "step": 18665 + }, + { + "epoch": 1.5392100272120062, + "grad_norm": 1.9495234489440918, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7429378032684326, + "num_tokens": 451649108.0, + "step": 18666 + }, + { + "epoch": 1.539292487837058, + "grad_norm": 2.2100462913513184, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7447583079338074, + "num_tokens": 451674352.0, + "step": 18667 + }, + { + "epoch": 1.5393749484621093, + "grad_norm": 1.7587916851043701, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7773798704147339, + "num_tokens": 451708726.0, + "step": 18668 + }, + { + "epoch": 1.5394574090871609, + "grad_norm": 1.986277461051941, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.758461594581604, + "num_tokens": 451739739.0, + "step": 18669 + }, + { + "epoch": 1.5395398697122125, + "grad_norm": 1.9549452066421509, + "learning_rate": 1e-06, + "loss": 0.6048, + "mean_token_accuracy": 0.7821881771087646, + "num_tokens": 451769968.0, + "step": 18670 + }, + { + "epoch": 1.5396223303372638, + "grad_norm": 2.054007053375244, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7347626686096191, + "num_tokens": 451799947.0, + "step": 18671 + }, + { + "epoch": 1.5397047909623156, + "grad_norm": 1.8647687435150146, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7723624110221863, + "num_tokens": 451834086.0, + "step": 18672 + }, + { + "epoch": 1.539787251587367, + "grad_norm": 1.8519127368927002, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7346792817115784, + "num_tokens": 451868794.0, + "step": 18673 + }, + { + "epoch": 1.5398697122124185, + "grad_norm": 1.957593321800232, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.783980131149292, + "num_tokens": 451897499.0, + "step": 18674 + }, + { + "epoch": 1.53995217283747, + "grad_norm": 2.0522518157958984, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7476437091827393, + "num_tokens": 451928078.0, + "step": 18675 + }, + { + "epoch": 1.5400346334625217, + "grad_norm": 1.8897902965545654, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7443779706954956, + "num_tokens": 451964145.0, + "step": 18676 + }, + { + "epoch": 1.5401170940875732, + "grad_norm": 2.105867385864258, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7519376277923584, + "num_tokens": 451991362.0, + "step": 18677 + }, + { + "epoch": 1.5401995547126246, + "grad_norm": 1.878740668296814, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.788127601146698, + "num_tokens": 452023589.0, + "step": 18678 + }, + { + "epoch": 1.5402820153376764, + "grad_norm": 1.7568728923797607, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.7886297702789307, + "num_tokens": 452057476.0, + "step": 18679 + }, + { + "epoch": 1.5403644759627277, + "grad_norm": 2.0855538845062256, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.776535153388977, + "num_tokens": 452086780.0, + "step": 18680 + }, + { + "epoch": 1.5404469365877793, + "grad_norm": 2.007744312286377, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7739560604095459, + "num_tokens": 452116747.0, + "step": 18681 + }, + { + "epoch": 1.5405293972128309, + "grad_norm": 1.9644606113433838, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7876021862030029, + "num_tokens": 452144127.0, + "step": 18682 + }, + { + "epoch": 1.5406118578378825, + "grad_norm": 2.0352911949157715, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.716127872467041, + "num_tokens": 452174396.0, + "step": 18683 + }, + { + "epoch": 1.540694318462934, + "grad_norm": 2.3152198791503906, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7380310297012329, + "num_tokens": 452201024.0, + "step": 18684 + }, + { + "epoch": 1.5407767790879854, + "grad_norm": 2.0240797996520996, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7689129114151001, + "num_tokens": 452233564.0, + "step": 18685 + }, + { + "epoch": 1.5408592397130372, + "grad_norm": 2.283583402633667, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7483628988265991, + "num_tokens": 452258090.0, + "step": 18686 + }, + { + "epoch": 1.5409417003380885, + "grad_norm": 1.7308326959609985, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7481338977813721, + "num_tokens": 452294514.0, + "step": 18687 + }, + { + "epoch": 1.54102416096314, + "grad_norm": 1.95579993724823, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.748161792755127, + "num_tokens": 452325127.0, + "step": 18688 + }, + { + "epoch": 1.5411066215881917, + "grad_norm": 2.0646910667419434, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7668534517288208, + "num_tokens": 452353339.0, + "step": 18689 + }, + { + "epoch": 1.541189082213243, + "grad_norm": 1.9271749258041382, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.7900984883308411, + "num_tokens": 452384241.0, + "step": 18690 + }, + { + "epoch": 1.5412715428382948, + "grad_norm": 2.0115277767181396, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7717029452323914, + "num_tokens": 452415115.0, + "step": 18691 + }, + { + "epoch": 1.5413540034633462, + "grad_norm": 1.882154107093811, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7676142454147339, + "num_tokens": 452447772.0, + "step": 18692 + }, + { + "epoch": 1.5414364640883977, + "grad_norm": 1.9405826330184937, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7593491077423096, + "num_tokens": 452478466.0, + "step": 18693 + }, + { + "epoch": 1.5415189247134493, + "grad_norm": 1.7203712463378906, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.783626914024353, + "num_tokens": 452515023.0, + "step": 18694 + }, + { + "epoch": 1.5416013853385009, + "grad_norm": 1.926010251045227, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.734860897064209, + "num_tokens": 452548415.0, + "step": 18695 + }, + { + "epoch": 1.5416838459635525, + "grad_norm": 1.9101417064666748, + "learning_rate": 1e-06, + "loss": 0.591, + "mean_token_accuracy": 0.7973259687423706, + "num_tokens": 452583948.0, + "step": 18696 + }, + { + "epoch": 1.5417663065886038, + "grad_norm": 2.1494359970092773, + "learning_rate": 1e-06, + "loss": 0.7825, + "mean_token_accuracy": 0.7447444200515747, + "num_tokens": 452609012.0, + "step": 18697 + }, + { + "epoch": 1.5418487672136556, + "grad_norm": 2.168530225753784, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7090338468551636, + "num_tokens": 452635235.0, + "step": 18698 + }, + { + "epoch": 1.541931227838707, + "grad_norm": 2.0438873767852783, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.7934132814407349, + "num_tokens": 452661910.0, + "step": 18699 + }, + { + "epoch": 1.5420136884637585, + "grad_norm": 1.9491350650787354, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7668506503105164, + "num_tokens": 452698768.0, + "step": 18700 + }, + { + "epoch": 1.54209614908881, + "grad_norm": 1.8425430059432983, + "learning_rate": 1e-06, + "loss": 0.5527, + "mean_token_accuracy": 0.8036272525787354, + "num_tokens": 452730758.0, + "step": 18701 + }, + { + "epoch": 1.5421786097138617, + "grad_norm": 2.0264158248901367, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7082576751708984, + "num_tokens": 452762699.0, + "step": 18702 + }, + { + "epoch": 1.5422610703389132, + "grad_norm": 2.2641797065734863, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7641655206680298, + "num_tokens": 452790065.0, + "step": 18703 + }, + { + "epoch": 1.5423435309639646, + "grad_norm": 1.8941842317581177, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7389903664588928, + "num_tokens": 452824149.0, + "step": 18704 + }, + { + "epoch": 1.5424259915890164, + "grad_norm": 1.8996031284332275, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7224204540252686, + "num_tokens": 452858252.0, + "step": 18705 + }, + { + "epoch": 1.5425084522140677, + "grad_norm": 2.306971549987793, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7668810486793518, + "num_tokens": 452881077.0, + "step": 18706 + }, + { + "epoch": 1.5425909128391193, + "grad_norm": 2.017148971557617, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7735216617584229, + "num_tokens": 452911880.0, + "step": 18707 + }, + { + "epoch": 1.5426733734641709, + "grad_norm": 1.9831820726394653, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7428566813468933, + "num_tokens": 452942274.0, + "step": 18708 + }, + { + "epoch": 1.5427558340892222, + "grad_norm": 2.006178140640259, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7494646310806274, + "num_tokens": 452972064.0, + "step": 18709 + }, + { + "epoch": 1.542838294714274, + "grad_norm": 1.9804178476333618, + "learning_rate": 1e-06, + "loss": 0.7712, + "mean_token_accuracy": 0.7446766495704651, + "num_tokens": 453004589.0, + "step": 18710 + }, + { + "epoch": 1.5429207553393254, + "grad_norm": 2.186394453048706, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7494518756866455, + "num_tokens": 453031426.0, + "step": 18711 + }, + { + "epoch": 1.5430032159643772, + "grad_norm": 2.0012199878692627, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7810759544372559, + "num_tokens": 453060454.0, + "step": 18712 + }, + { + "epoch": 1.5430856765894285, + "grad_norm": 2.032528877258301, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.7790110111236572, + "num_tokens": 453087951.0, + "step": 18713 + }, + { + "epoch": 1.54316813721448, + "grad_norm": 2.0592200756073, + "learning_rate": 1e-06, + "loss": 0.5779, + "mean_token_accuracy": 0.7959123253822327, + "num_tokens": 453115405.0, + "step": 18714 + }, + { + "epoch": 1.5432505978395317, + "grad_norm": 2.312248468399048, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7754185199737549, + "num_tokens": 453137658.0, + "step": 18715 + }, + { + "epoch": 1.543333058464583, + "grad_norm": 2.2943038940429688, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.7602071762084961, + "num_tokens": 453160161.0, + "step": 18716 + }, + { + "epoch": 1.5434155190896348, + "grad_norm": 2.2055399417877197, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7540366053581238, + "num_tokens": 453184090.0, + "step": 18717 + }, + { + "epoch": 1.5434979797146862, + "grad_norm": 1.8466664552688599, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7703975439071655, + "num_tokens": 453217174.0, + "step": 18718 + }, + { + "epoch": 1.5435804403397377, + "grad_norm": 1.8708120584487915, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.7778041362762451, + "num_tokens": 453250887.0, + "step": 18719 + }, + { + "epoch": 1.5436629009647893, + "grad_norm": 2.139542818069458, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.729486346244812, + "num_tokens": 453279036.0, + "step": 18720 + }, + { + "epoch": 1.5437453615898409, + "grad_norm": 2.1018905639648438, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7835004329681396, + "num_tokens": 453305122.0, + "step": 18721 + }, + { + "epoch": 1.5438278222148925, + "grad_norm": 2.032245635986328, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7729178667068481, + "num_tokens": 453333110.0, + "step": 18722 + }, + { + "epoch": 1.5439102828399438, + "grad_norm": 2.097026824951172, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.7831345796585083, + "num_tokens": 453361147.0, + "step": 18723 + }, + { + "epoch": 1.5439927434649956, + "grad_norm": 2.199784994125366, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.7904704809188843, + "num_tokens": 453384861.0, + "step": 18724 + }, + { + "epoch": 1.544075204090047, + "grad_norm": 1.988389492034912, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7791308164596558, + "num_tokens": 453415254.0, + "step": 18725 + }, + { + "epoch": 1.5441576647150985, + "grad_norm": 2.097683906555176, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.745044469833374, + "num_tokens": 453442683.0, + "step": 18726 + }, + { + "epoch": 1.54424012534015, + "grad_norm": 2.1494009494781494, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7395238280296326, + "num_tokens": 453470765.0, + "step": 18727 + }, + { + "epoch": 1.5443225859652017, + "grad_norm": 2.265712261199951, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7560919523239136, + "num_tokens": 453494105.0, + "step": 18728 + }, + { + "epoch": 1.5444050465902532, + "grad_norm": 1.9846569299697876, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7701804637908936, + "num_tokens": 453525308.0, + "step": 18729 + }, + { + "epoch": 1.5444875072153046, + "grad_norm": 2.1433920860290527, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.726391077041626, + "num_tokens": 453552455.0, + "step": 18730 + }, + { + "epoch": 1.5445699678403564, + "grad_norm": 2.057810068130493, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.753265917301178, + "num_tokens": 453580672.0, + "step": 18731 + }, + { + "epoch": 1.5446524284654077, + "grad_norm": 1.8490574359893799, + "learning_rate": 1e-06, + "loss": 0.5884, + "mean_token_accuracy": 0.8072551488876343, + "num_tokens": 453612900.0, + "step": 18732 + }, + { + "epoch": 1.5447348890904593, + "grad_norm": 2.1830458641052246, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7699663043022156, + "num_tokens": 453636473.0, + "step": 18733 + }, + { + "epoch": 1.5448173497155109, + "grad_norm": 1.9570945501327515, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7713502645492554, + "num_tokens": 453666442.0, + "step": 18734 + }, + { + "epoch": 1.5448998103405622, + "grad_norm": 1.9273420572280884, + "learning_rate": 1e-06, + "loss": 0.7746, + "mean_token_accuracy": 0.7478580474853516, + "num_tokens": 453701988.0, + "step": 18735 + }, + { + "epoch": 1.544982270965614, + "grad_norm": 2.009772539138794, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7530758380889893, + "num_tokens": 453730716.0, + "step": 18736 + }, + { + "epoch": 1.5450647315906654, + "grad_norm": 2.0549004077911377, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.772097110748291, + "num_tokens": 453760043.0, + "step": 18737 + }, + { + "epoch": 1.545147192215717, + "grad_norm": 2.023759365081787, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7699359655380249, + "num_tokens": 453788389.0, + "step": 18738 + }, + { + "epoch": 1.5452296528407685, + "grad_norm": 1.7332489490509033, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7722033858299255, + "num_tokens": 453826252.0, + "step": 18739 + }, + { + "epoch": 1.54531211346582, + "grad_norm": 2.046893358230591, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.716809093952179, + "num_tokens": 453856353.0, + "step": 18740 + }, + { + "epoch": 1.5453945740908717, + "grad_norm": 1.7454391717910767, + "learning_rate": 1e-06, + "loss": 0.6221, + "mean_token_accuracy": 0.7914062738418579, + "num_tokens": 453892358.0, + "step": 18741 + }, + { + "epoch": 1.545477034715923, + "grad_norm": 1.980418086051941, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7458634376525879, + "num_tokens": 453923199.0, + "step": 18742 + }, + { + "epoch": 1.5455594953409748, + "grad_norm": 2.203350782394409, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7094674110412598, + "num_tokens": 453952068.0, + "step": 18743 + }, + { + "epoch": 1.5456419559660262, + "grad_norm": 2.1154532432556152, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7688407301902771, + "num_tokens": 453976872.0, + "step": 18744 + }, + { + "epoch": 1.5457244165910777, + "grad_norm": 1.9142361879348755, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.782180905342102, + "num_tokens": 454008227.0, + "step": 18745 + }, + { + "epoch": 1.5458068772161293, + "grad_norm": 1.9550837278366089, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.739354133605957, + "num_tokens": 454038673.0, + "step": 18746 + }, + { + "epoch": 1.5458893378411809, + "grad_norm": 2.276185989379883, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7370361685752869, + "num_tokens": 454062957.0, + "step": 18747 + }, + { + "epoch": 1.5459717984662324, + "grad_norm": 2.119885206222534, + "learning_rate": 1e-06, + "loss": 0.7805, + "mean_token_accuracy": 0.7457404136657715, + "num_tokens": 454087855.0, + "step": 18748 + }, + { + "epoch": 1.5460542590912838, + "grad_norm": 2.182415246963501, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7352079153060913, + "num_tokens": 454114060.0, + "step": 18749 + }, + { + "epoch": 1.5461367197163356, + "grad_norm": 2.0335092544555664, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7744941711425781, + "num_tokens": 454139042.0, + "step": 18750 + }, + { + "epoch": 1.546219180341387, + "grad_norm": 1.875288724899292, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7620495557785034, + "num_tokens": 454173140.0, + "step": 18751 + }, + { + "epoch": 1.5463016409664385, + "grad_norm": 1.9526516199111938, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7590357065200806, + "num_tokens": 454207573.0, + "step": 18752 + }, + { + "epoch": 1.54638410159149, + "grad_norm": 1.9078024625778198, + "learning_rate": 1e-06, + "loss": 0.5946, + "mean_token_accuracy": 0.799301266670227, + "num_tokens": 454236795.0, + "step": 18753 + }, + { + "epoch": 1.5464665622165414, + "grad_norm": 1.8539658784866333, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7456436157226562, + "num_tokens": 454269244.0, + "step": 18754 + }, + { + "epoch": 1.5465490228415932, + "grad_norm": 2.072514772415161, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7557724714279175, + "num_tokens": 454297302.0, + "step": 18755 + }, + { + "epoch": 1.5466314834666446, + "grad_norm": 1.725543737411499, + "learning_rate": 1e-06, + "loss": 0.5855, + "mean_token_accuracy": 0.812178909778595, + "num_tokens": 454330869.0, + "step": 18756 + }, + { + "epoch": 1.5467139440916964, + "grad_norm": 1.7663459777832031, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7385187149047852, + "num_tokens": 454368787.0, + "step": 18757 + }, + { + "epoch": 1.5467964047167477, + "grad_norm": 1.8879026174545288, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.7876179218292236, + "num_tokens": 454398856.0, + "step": 18758 + }, + { + "epoch": 1.5468788653417993, + "grad_norm": 2.003931999206543, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.778769314289093, + "num_tokens": 454428999.0, + "step": 18759 + }, + { + "epoch": 1.5469613259668509, + "grad_norm": 2.240433692932129, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7444535493850708, + "num_tokens": 454454854.0, + "step": 18760 + }, + { + "epoch": 1.5470437865919022, + "grad_norm": 2.079655170440674, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7616902589797974, + "num_tokens": 454482928.0, + "step": 18761 + }, + { + "epoch": 1.547126247216954, + "grad_norm": 1.884086012840271, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7706971168518066, + "num_tokens": 454514751.0, + "step": 18762 + }, + { + "epoch": 1.5472087078420054, + "grad_norm": 2.3350467681884766, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7500874400138855, + "num_tokens": 454540531.0, + "step": 18763 + }, + { + "epoch": 1.547291168467057, + "grad_norm": 1.777287483215332, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.752515435218811, + "num_tokens": 454573663.0, + "step": 18764 + }, + { + "epoch": 1.5473736290921085, + "grad_norm": 1.9318324327468872, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7362857460975647, + "num_tokens": 454607013.0, + "step": 18765 + }, + { + "epoch": 1.54745608971716, + "grad_norm": 1.9654964208602905, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7639622688293457, + "num_tokens": 454637771.0, + "step": 18766 + }, + { + "epoch": 1.5475385503422117, + "grad_norm": 1.9940519332885742, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.7961188554763794, + "num_tokens": 454666195.0, + "step": 18767 + }, + { + "epoch": 1.547621010967263, + "grad_norm": 2.179715633392334, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.7957804799079895, + "num_tokens": 454689656.0, + "step": 18768 + }, + { + "epoch": 1.5477034715923148, + "grad_norm": 1.7173657417297363, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.7936822772026062, + "num_tokens": 454726709.0, + "step": 18769 + }, + { + "epoch": 1.5477859322173662, + "grad_norm": 1.973000407218933, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7578577995300293, + "num_tokens": 454757810.0, + "step": 18770 + }, + { + "epoch": 1.5478683928424177, + "grad_norm": 2.10652232170105, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7443889379501343, + "num_tokens": 454786956.0, + "step": 18771 + }, + { + "epoch": 1.5479508534674693, + "grad_norm": 2.156834840774536, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7467350363731384, + "num_tokens": 454815521.0, + "step": 18772 + }, + { + "epoch": 1.5480333140925209, + "grad_norm": 2.180135488510132, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7675176858901978, + "num_tokens": 454840601.0, + "step": 18773 + }, + { + "epoch": 1.5481157747175724, + "grad_norm": 2.073275327682495, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.7847567796707153, + "num_tokens": 454870582.0, + "step": 18774 + }, + { + "epoch": 1.5481982353426238, + "grad_norm": 2.162891387939453, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7800440788269043, + "num_tokens": 454894136.0, + "step": 18775 + }, + { + "epoch": 1.5482806959676756, + "grad_norm": 1.8418139219284058, + "learning_rate": 1e-06, + "loss": 0.5646, + "mean_token_accuracy": 0.815972089767456, + "num_tokens": 454924286.0, + "step": 18776 + }, + { + "epoch": 1.548363156592727, + "grad_norm": 2.442692279815674, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7323325872421265, + "num_tokens": 454946801.0, + "step": 18777 + }, + { + "epoch": 1.5484456172177785, + "grad_norm": 1.8674315214157104, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7684255838394165, + "num_tokens": 454978018.0, + "step": 18778 + }, + { + "epoch": 1.54852807784283, + "grad_norm": 1.9147673845291138, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.7806651592254639, + "num_tokens": 455010283.0, + "step": 18779 + }, + { + "epoch": 1.5486105384678814, + "grad_norm": 1.9997581243515015, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7219787836074829, + "num_tokens": 455041968.0, + "step": 18780 + }, + { + "epoch": 1.5486929990929332, + "grad_norm": 2.139249801635742, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7599844336509705, + "num_tokens": 455070051.0, + "step": 18781 + }, + { + "epoch": 1.5487754597179846, + "grad_norm": 2.162153482437134, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7461993098258972, + "num_tokens": 455097224.0, + "step": 18782 + }, + { + "epoch": 1.5488579203430362, + "grad_norm": 2.1030783653259277, + "learning_rate": 1e-06, + "loss": 0.5832, + "mean_token_accuracy": 0.7986221313476562, + "num_tokens": 455122951.0, + "step": 18783 + }, + { + "epoch": 1.5489403809680877, + "grad_norm": 2.0324645042419434, + "learning_rate": 1e-06, + "loss": 0.5884, + "mean_token_accuracy": 0.8098775744438171, + "num_tokens": 455155613.0, + "step": 18784 + }, + { + "epoch": 1.5490228415931393, + "grad_norm": 2.1823012828826904, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.7877209186553955, + "num_tokens": 455181127.0, + "step": 18785 + }, + { + "epoch": 1.5491053022181909, + "grad_norm": 2.003391742706299, + "learning_rate": 1e-06, + "loss": 0.5999, + "mean_token_accuracy": 0.796442985534668, + "num_tokens": 455207996.0, + "step": 18786 + }, + { + "epoch": 1.5491877628432422, + "grad_norm": 2.1129703521728516, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7312873601913452, + "num_tokens": 455235569.0, + "step": 18787 + }, + { + "epoch": 1.549270223468294, + "grad_norm": 1.7714098691940308, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7511827945709229, + "num_tokens": 455269922.0, + "step": 18788 + }, + { + "epoch": 1.5493526840933454, + "grad_norm": 1.9794095754623413, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7762952446937561, + "num_tokens": 455299832.0, + "step": 18789 + }, + { + "epoch": 1.549435144718397, + "grad_norm": 2.0468568801879883, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7773038148880005, + "num_tokens": 455334364.0, + "step": 18790 + }, + { + "epoch": 1.5495176053434485, + "grad_norm": 2.1758079528808594, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7188833951950073, + "num_tokens": 455364206.0, + "step": 18791 + }, + { + "epoch": 1.5496000659685, + "grad_norm": 2.1333398818969727, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.737036943435669, + "num_tokens": 455390787.0, + "step": 18792 + }, + { + "epoch": 1.5496825265935517, + "grad_norm": 2.077284336090088, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7403203845024109, + "num_tokens": 455419800.0, + "step": 18793 + }, + { + "epoch": 1.549764987218603, + "grad_norm": 2.1054160594940186, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.707394540309906, + "num_tokens": 455449714.0, + "step": 18794 + }, + { + "epoch": 1.5498474478436548, + "grad_norm": 2.2299280166625977, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7016853094100952, + "num_tokens": 455476042.0, + "step": 18795 + }, + { + "epoch": 1.5499299084687062, + "grad_norm": 2.33734130859375, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7455234527587891, + "num_tokens": 455499201.0, + "step": 18796 + }, + { + "epoch": 1.5500123690937577, + "grad_norm": 2.065411329269409, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7312225103378296, + "num_tokens": 455527403.0, + "step": 18797 + }, + { + "epoch": 1.5500948297188093, + "grad_norm": 1.97938871383667, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7786545753479004, + "num_tokens": 455557855.0, + "step": 18798 + }, + { + "epoch": 1.5501772903438606, + "grad_norm": 1.9678488969802856, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.753677487373352, + "num_tokens": 455589642.0, + "step": 18799 + }, + { + "epoch": 1.5502597509689124, + "grad_norm": 2.0049238204956055, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7868338823318481, + "num_tokens": 455617611.0, + "step": 18800 + }, + { + "epoch": 1.5503422115939638, + "grad_norm": 1.9826891422271729, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7661756873130798, + "num_tokens": 455650441.0, + "step": 18801 + }, + { + "epoch": 1.5504246722190154, + "grad_norm": 2.0602877140045166, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7681158781051636, + "num_tokens": 455678201.0, + "step": 18802 + }, + { + "epoch": 1.550507132844067, + "grad_norm": 2.0130856037139893, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7495226860046387, + "num_tokens": 455708468.0, + "step": 18803 + }, + { + "epoch": 1.5505895934691185, + "grad_norm": 2.1193490028381348, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.7452853322029114, + "num_tokens": 455736687.0, + "step": 18804 + }, + { + "epoch": 1.55067205409417, + "grad_norm": 2.189181089401245, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7446553707122803, + "num_tokens": 455762436.0, + "step": 18805 + }, + { + "epoch": 1.5507545147192214, + "grad_norm": 2.017629861831665, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7446985840797424, + "num_tokens": 455793097.0, + "step": 18806 + }, + { + "epoch": 1.5508369753442732, + "grad_norm": 1.7277501821517944, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.790622353553772, + "num_tokens": 455829767.0, + "step": 18807 + }, + { + "epoch": 1.5509194359693246, + "grad_norm": 1.8865376710891724, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7323106527328491, + "num_tokens": 455863341.0, + "step": 18808 + }, + { + "epoch": 1.5510018965943762, + "grad_norm": 2.254758358001709, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7565478086471558, + "num_tokens": 455890763.0, + "step": 18809 + }, + { + "epoch": 1.5510843572194277, + "grad_norm": 1.9870473146438599, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.7899559140205383, + "num_tokens": 455920398.0, + "step": 18810 + }, + { + "epoch": 1.5511668178444793, + "grad_norm": 2.108945846557617, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7640165686607361, + "num_tokens": 455945635.0, + "step": 18811 + }, + { + "epoch": 1.5512492784695309, + "grad_norm": 2.1091222763061523, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7420073747634888, + "num_tokens": 455975180.0, + "step": 18812 + }, + { + "epoch": 1.5513317390945822, + "grad_norm": 1.9168769121170044, + "learning_rate": 1e-06, + "loss": 0.5636, + "mean_token_accuracy": 0.8014880418777466, + "num_tokens": 456001845.0, + "step": 18813 + }, + { + "epoch": 1.551414199719634, + "grad_norm": 2.3526172637939453, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7141339778900146, + "num_tokens": 456028023.0, + "step": 18814 + }, + { + "epoch": 1.5514966603446854, + "grad_norm": 2.3765411376953125, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7259390354156494, + "num_tokens": 456051338.0, + "step": 18815 + }, + { + "epoch": 1.551579120969737, + "grad_norm": 1.8471356630325317, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.72650146484375, + "num_tokens": 456089677.0, + "step": 18816 + }, + { + "epoch": 1.5516615815947885, + "grad_norm": 1.9729995727539062, + "learning_rate": 1e-06, + "loss": 0.5382, + "mean_token_accuracy": 0.8150888681411743, + "num_tokens": 456115004.0, + "step": 18817 + }, + { + "epoch": 1.55174404221984, + "grad_norm": 2.1136579513549805, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7399377822875977, + "num_tokens": 456145026.0, + "step": 18818 + }, + { + "epoch": 1.5518265028448917, + "grad_norm": 1.870381236076355, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7448139190673828, + "num_tokens": 456180253.0, + "step": 18819 + }, + { + "epoch": 1.551908963469943, + "grad_norm": 2.120798110961914, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7432980537414551, + "num_tokens": 456209152.0, + "step": 18820 + }, + { + "epoch": 1.5519914240949948, + "grad_norm": 1.9041342735290527, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7659172415733337, + "num_tokens": 456239421.0, + "step": 18821 + }, + { + "epoch": 1.5520738847200461, + "grad_norm": 1.8826955556869507, + "learning_rate": 1e-06, + "loss": 0.6093, + "mean_token_accuracy": 0.7976244688034058, + "num_tokens": 456271004.0, + "step": 18822 + }, + { + "epoch": 1.5521563453450977, + "grad_norm": 1.9780287742614746, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7485310435295105, + "num_tokens": 456301819.0, + "step": 18823 + }, + { + "epoch": 1.5522388059701493, + "grad_norm": 2.1341845989227295, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7176113128662109, + "num_tokens": 456329892.0, + "step": 18824 + }, + { + "epoch": 1.5523212665952006, + "grad_norm": 2.166058301925659, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7307397723197937, + "num_tokens": 456356449.0, + "step": 18825 + }, + { + "epoch": 1.5524037272202524, + "grad_norm": 2.287076234817505, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.739808201789856, + "num_tokens": 456381440.0, + "step": 18826 + }, + { + "epoch": 1.5524861878453038, + "grad_norm": 1.8195834159851074, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.780204176902771, + "num_tokens": 456415192.0, + "step": 18827 + }, + { + "epoch": 1.5525686484703554, + "grad_norm": 2.1006357669830322, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7281746864318848, + "num_tokens": 456441970.0, + "step": 18828 + }, + { + "epoch": 1.552651109095407, + "grad_norm": 1.9214835166931152, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.7900307178497314, + "num_tokens": 456473521.0, + "step": 18829 + }, + { + "epoch": 1.5527335697204585, + "grad_norm": 1.94021737575531, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7635062336921692, + "num_tokens": 456505837.0, + "step": 18830 + }, + { + "epoch": 1.55281603034551, + "grad_norm": 2.0840635299682617, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7363399267196655, + "num_tokens": 456535381.0, + "step": 18831 + }, + { + "epoch": 1.5528984909705614, + "grad_norm": 2.0401220321655273, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7755813598632812, + "num_tokens": 456562893.0, + "step": 18832 + }, + { + "epoch": 1.5529809515956132, + "grad_norm": 1.962158441543579, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7371169924736023, + "num_tokens": 456591819.0, + "step": 18833 + }, + { + "epoch": 1.5530634122206646, + "grad_norm": 1.8573994636535645, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.784706175327301, + "num_tokens": 456622216.0, + "step": 18834 + }, + { + "epoch": 1.5531458728457161, + "grad_norm": 1.992105484008789, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7578058838844299, + "num_tokens": 456650217.0, + "step": 18835 + }, + { + "epoch": 1.5532283334707677, + "grad_norm": 1.9404873847961426, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.7952147722244263, + "num_tokens": 456684302.0, + "step": 18836 + }, + { + "epoch": 1.5533107940958193, + "grad_norm": 1.9364606142044067, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7478336095809937, + "num_tokens": 456716716.0, + "step": 18837 + }, + { + "epoch": 1.5533932547208709, + "grad_norm": 1.9299123287200928, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7484422922134399, + "num_tokens": 456749259.0, + "step": 18838 + }, + { + "epoch": 1.5534757153459222, + "grad_norm": 1.8289381265640259, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.771378755569458, + "num_tokens": 456781039.0, + "step": 18839 + }, + { + "epoch": 1.553558175970974, + "grad_norm": 2.088914394378662, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7789413928985596, + "num_tokens": 456809256.0, + "step": 18840 + }, + { + "epoch": 1.5536406365960254, + "grad_norm": 2.053194046020508, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7524674534797668, + "num_tokens": 456838491.0, + "step": 18841 + }, + { + "epoch": 1.553723097221077, + "grad_norm": 2.1780362129211426, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7544600963592529, + "num_tokens": 456866368.0, + "step": 18842 + }, + { + "epoch": 1.5538055578461285, + "grad_norm": 2.2764999866485596, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7643980979919434, + "num_tokens": 456894258.0, + "step": 18843 + }, + { + "epoch": 1.5538880184711799, + "grad_norm": 2.061007261276245, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7325879335403442, + "num_tokens": 456922350.0, + "step": 18844 + }, + { + "epoch": 1.5539704790962316, + "grad_norm": 1.894473910331726, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7562437057495117, + "num_tokens": 456952784.0, + "step": 18845 + }, + { + "epoch": 1.554052939721283, + "grad_norm": 2.079440116882324, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7645241022109985, + "num_tokens": 456979328.0, + "step": 18846 + }, + { + "epoch": 1.5541354003463346, + "grad_norm": 2.1348652839660645, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7599062919616699, + "num_tokens": 457003393.0, + "step": 18847 + }, + { + "epoch": 1.5542178609713861, + "grad_norm": 1.938327431678772, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7228102087974548, + "num_tokens": 457037224.0, + "step": 18848 + }, + { + "epoch": 1.5543003215964377, + "grad_norm": 1.9961774349212646, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7552346587181091, + "num_tokens": 457069390.0, + "step": 18849 + }, + { + "epoch": 1.5543827822214893, + "grad_norm": 2.0294406414031982, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7745323181152344, + "num_tokens": 457099004.0, + "step": 18850 + }, + { + "epoch": 1.5544652428465406, + "grad_norm": 1.821713924407959, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.7774859666824341, + "num_tokens": 457132785.0, + "step": 18851 + }, + { + "epoch": 1.5545477034715924, + "grad_norm": 2.0322680473327637, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7739821672439575, + "num_tokens": 457162032.0, + "step": 18852 + }, + { + "epoch": 1.5546301640966438, + "grad_norm": 1.9632062911987305, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7660664916038513, + "num_tokens": 457193239.0, + "step": 18853 + }, + { + "epoch": 1.5547126247216954, + "grad_norm": 2.113079786300659, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7545284628868103, + "num_tokens": 457222041.0, + "step": 18854 + }, + { + "epoch": 1.554795085346747, + "grad_norm": 1.9402461051940918, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.7799943685531616, + "num_tokens": 457253476.0, + "step": 18855 + }, + { + "epoch": 1.5548775459717985, + "grad_norm": 1.8863742351531982, + "learning_rate": 1e-06, + "loss": 0.6126, + "mean_token_accuracy": 0.7975193858146667, + "num_tokens": 457286949.0, + "step": 18856 + }, + { + "epoch": 1.55496000659685, + "grad_norm": 2.179225444793701, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7636349201202393, + "num_tokens": 457315918.0, + "step": 18857 + }, + { + "epoch": 1.5550424672219014, + "grad_norm": 2.252558946609497, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7371450662612915, + "num_tokens": 457342356.0, + "step": 18858 + }, + { + "epoch": 1.5551249278469532, + "grad_norm": 2.1294870376586914, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7649300694465637, + "num_tokens": 457372242.0, + "step": 18859 + }, + { + "epoch": 1.5552073884720046, + "grad_norm": 2.2296760082244873, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.770495593547821, + "num_tokens": 457396438.0, + "step": 18860 + }, + { + "epoch": 1.5552898490970561, + "grad_norm": 1.8226102590560913, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7723760008811951, + "num_tokens": 457431810.0, + "step": 18861 + }, + { + "epoch": 1.5553723097221077, + "grad_norm": 2.1735496520996094, + "learning_rate": 1e-06, + "loss": 0.6127, + "mean_token_accuracy": 0.7846760749816895, + "num_tokens": 457456590.0, + "step": 18862 + }, + { + "epoch": 1.555454770347159, + "grad_norm": 1.9605302810668945, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7632877826690674, + "num_tokens": 457486306.0, + "step": 18863 + }, + { + "epoch": 1.5555372309722109, + "grad_norm": 2.1857686042785645, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.7544033527374268, + "num_tokens": 457513048.0, + "step": 18864 + }, + { + "epoch": 1.5556196915972622, + "grad_norm": 2.0525705814361572, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7406340837478638, + "num_tokens": 457541630.0, + "step": 18865 + }, + { + "epoch": 1.555702152222314, + "grad_norm": 1.8517659902572632, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.7533442974090576, + "num_tokens": 457574618.0, + "step": 18866 + }, + { + "epoch": 1.5557846128473654, + "grad_norm": 1.9929020404815674, + "learning_rate": 1e-06, + "loss": 0.5812, + "mean_token_accuracy": 0.8238744735717773, + "num_tokens": 457601590.0, + "step": 18867 + }, + { + "epoch": 1.555867073472417, + "grad_norm": 2.071194887161255, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7678132653236389, + "num_tokens": 457630144.0, + "step": 18868 + }, + { + "epoch": 1.5559495340974685, + "grad_norm": 1.8736343383789062, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.7827084064483643, + "num_tokens": 457662508.0, + "step": 18869 + }, + { + "epoch": 1.5560319947225199, + "grad_norm": 2.1735124588012695, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7503899335861206, + "num_tokens": 457690322.0, + "step": 18870 + }, + { + "epoch": 1.5561144553475716, + "grad_norm": 2.013476610183716, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7585705518722534, + "num_tokens": 457720210.0, + "step": 18871 + }, + { + "epoch": 1.556196915972623, + "grad_norm": 1.7826822996139526, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.7882232069969177, + "num_tokens": 457752716.0, + "step": 18872 + }, + { + "epoch": 1.5562793765976746, + "grad_norm": 2.028346538543701, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7538217306137085, + "num_tokens": 457780824.0, + "step": 18873 + }, + { + "epoch": 1.5563618372227261, + "grad_norm": 2.0014395713806152, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7649072408676147, + "num_tokens": 457808847.0, + "step": 18874 + }, + { + "epoch": 1.5564442978477777, + "grad_norm": 2.120112895965576, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7321223616600037, + "num_tokens": 457836485.0, + "step": 18875 + }, + { + "epoch": 1.5565267584728293, + "grad_norm": 1.9564822912216187, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7517330050468445, + "num_tokens": 457869137.0, + "step": 18876 + }, + { + "epoch": 1.5566092190978806, + "grad_norm": 1.8989169597625732, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.7823374271392822, + "num_tokens": 457902404.0, + "step": 18877 + }, + { + "epoch": 1.5566916797229324, + "grad_norm": 1.6874094009399414, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.7564198970794678, + "num_tokens": 457941167.0, + "step": 18878 + }, + { + "epoch": 1.5567741403479838, + "grad_norm": 1.8197799921035767, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7508436441421509, + "num_tokens": 457977729.0, + "step": 18879 + }, + { + "epoch": 1.5568566009730354, + "grad_norm": 2.0987579822540283, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7700632214546204, + "num_tokens": 458006161.0, + "step": 18880 + }, + { + "epoch": 1.556939061598087, + "grad_norm": 2.5199077129364014, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7335189580917358, + "num_tokens": 458026914.0, + "step": 18881 + }, + { + "epoch": 1.5570215222231385, + "grad_norm": 2.348005533218384, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7613162994384766, + "num_tokens": 458049117.0, + "step": 18882 + }, + { + "epoch": 1.55710398284819, + "grad_norm": 1.9593480825424194, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7790484428405762, + "num_tokens": 458076980.0, + "step": 18883 + }, + { + "epoch": 1.5571864434732414, + "grad_norm": 2.1021230220794678, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7712740898132324, + "num_tokens": 458102209.0, + "step": 18884 + }, + { + "epoch": 1.5572689040982932, + "grad_norm": 1.751288652420044, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7850534915924072, + "num_tokens": 458137394.0, + "step": 18885 + }, + { + "epoch": 1.5573513647233446, + "grad_norm": 2.087395668029785, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7444442510604858, + "num_tokens": 458167047.0, + "step": 18886 + }, + { + "epoch": 1.5574338253483961, + "grad_norm": 1.7647095918655396, + "learning_rate": 1e-06, + "loss": 0.6093, + "mean_token_accuracy": 0.8008152842521667, + "num_tokens": 458198398.0, + "step": 18887 + }, + { + "epoch": 1.5575162859734477, + "grad_norm": 1.9383617639541626, + "learning_rate": 1e-06, + "loss": 0.7746, + "mean_token_accuracy": 0.7502753734588623, + "num_tokens": 458233004.0, + "step": 18888 + }, + { + "epoch": 1.557598746598499, + "grad_norm": 1.855255365371704, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7576384544372559, + "num_tokens": 458263008.0, + "step": 18889 + }, + { + "epoch": 1.5576812072235509, + "grad_norm": 2.011754274368286, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7619563937187195, + "num_tokens": 458291947.0, + "step": 18890 + }, + { + "epoch": 1.5577636678486022, + "grad_norm": 1.90836501121521, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.7916349768638611, + "num_tokens": 458322249.0, + "step": 18891 + }, + { + "epoch": 1.5578461284736538, + "grad_norm": 2.057866334915161, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7357614636421204, + "num_tokens": 458349766.0, + "step": 18892 + }, + { + "epoch": 1.5579285890987054, + "grad_norm": 1.9058558940887451, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.7855849266052246, + "num_tokens": 458380105.0, + "step": 18893 + }, + { + "epoch": 1.558011049723757, + "grad_norm": 2.2586421966552734, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7237104773521423, + "num_tokens": 458406053.0, + "step": 18894 + }, + { + "epoch": 1.5580935103488085, + "grad_norm": 2.118560552597046, + "learning_rate": 1e-06, + "loss": 0.5364, + "mean_token_accuracy": 0.816901445388794, + "num_tokens": 458428582.0, + "step": 18895 + }, + { + "epoch": 1.5581759709738598, + "grad_norm": 1.958553671836853, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7237855195999146, + "num_tokens": 458461579.0, + "step": 18896 + }, + { + "epoch": 1.5582584315989116, + "grad_norm": 1.8159579038619995, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.770797073841095, + "num_tokens": 458495141.0, + "step": 18897 + }, + { + "epoch": 1.558340892223963, + "grad_norm": 1.9663364887237549, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7461320757865906, + "num_tokens": 458529206.0, + "step": 18898 + }, + { + "epoch": 1.5584233528490146, + "grad_norm": 1.7775819301605225, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.7804132699966431, + "num_tokens": 458565860.0, + "step": 18899 + }, + { + "epoch": 1.5585058134740661, + "grad_norm": 1.726963996887207, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.7974122166633606, + "num_tokens": 458605456.0, + "step": 18900 + }, + { + "epoch": 1.5585882740991177, + "grad_norm": 2.150200605392456, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7603849768638611, + "num_tokens": 458632450.0, + "step": 18901 + }, + { + "epoch": 1.5586707347241693, + "grad_norm": 2.4490115642547607, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7282722592353821, + "num_tokens": 458654859.0, + "step": 18902 + }, + { + "epoch": 1.5587531953492206, + "grad_norm": 1.8434373140335083, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7498698234558105, + "num_tokens": 458687713.0, + "step": 18903 + }, + { + "epoch": 1.5588356559742724, + "grad_norm": 1.794308066368103, + "learning_rate": 1e-06, + "loss": 0.5887, + "mean_token_accuracy": 0.8105471134185791, + "num_tokens": 458719822.0, + "step": 18904 + }, + { + "epoch": 1.5589181165993238, + "grad_norm": 2.0325098037719727, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.7973607778549194, + "num_tokens": 458747296.0, + "step": 18905 + }, + { + "epoch": 1.5590005772243754, + "grad_norm": 1.7551627159118652, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.7951250076293945, + "num_tokens": 458780911.0, + "step": 18906 + }, + { + "epoch": 1.559083037849427, + "grad_norm": 1.8473023176193237, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.777327835559845, + "num_tokens": 458815329.0, + "step": 18907 + }, + { + "epoch": 1.5591654984744783, + "grad_norm": 1.8681658506393433, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7639201283454895, + "num_tokens": 458845612.0, + "step": 18908 + }, + { + "epoch": 1.55924795909953, + "grad_norm": 1.9141114950180054, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7632084488868713, + "num_tokens": 458879064.0, + "step": 18909 + }, + { + "epoch": 1.5593304197245814, + "grad_norm": 1.937825083732605, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.77686607837677, + "num_tokens": 458910563.0, + "step": 18910 + }, + { + "epoch": 1.5594128803496332, + "grad_norm": 2.2192320823669434, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7743216156959534, + "num_tokens": 458935187.0, + "step": 18911 + }, + { + "epoch": 1.5594953409746846, + "grad_norm": 2.136122703552246, + "learning_rate": 1e-06, + "loss": 0.7746, + "mean_token_accuracy": 0.7548868656158447, + "num_tokens": 458962330.0, + "step": 18912 + }, + { + "epoch": 1.5595778015997361, + "grad_norm": 1.867850661277771, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7493295669555664, + "num_tokens": 458993655.0, + "step": 18913 + }, + { + "epoch": 1.5596602622247877, + "grad_norm": 1.8655184507369995, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.791161060333252, + "num_tokens": 459027878.0, + "step": 18914 + }, + { + "epoch": 1.559742722849839, + "grad_norm": 1.9254188537597656, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.748347818851471, + "num_tokens": 459058392.0, + "step": 18915 + }, + { + "epoch": 1.5598251834748909, + "grad_norm": 2.1789844036102295, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.764624297618866, + "num_tokens": 459082664.0, + "step": 18916 + }, + { + "epoch": 1.5599076440999422, + "grad_norm": 2.1545846462249756, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7477155327796936, + "num_tokens": 459111458.0, + "step": 18917 + }, + { + "epoch": 1.5599901047249938, + "grad_norm": 2.296661615371704, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7634661197662354, + "num_tokens": 459136569.0, + "step": 18918 + }, + { + "epoch": 1.5600725653500453, + "grad_norm": 2.0416290760040283, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7662169933319092, + "num_tokens": 459163799.0, + "step": 18919 + }, + { + "epoch": 1.560155025975097, + "grad_norm": 2.0327916145324707, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.7882562875747681, + "num_tokens": 459192355.0, + "step": 18920 + }, + { + "epoch": 1.5602374866001485, + "grad_norm": 1.7626644372940063, + "learning_rate": 1e-06, + "loss": 0.5849, + "mean_token_accuracy": 0.8016830682754517, + "num_tokens": 459224154.0, + "step": 18921 + }, + { + "epoch": 1.5603199472251998, + "grad_norm": 2.2245748043060303, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7376916408538818, + "num_tokens": 459250851.0, + "step": 18922 + }, + { + "epoch": 1.5604024078502516, + "grad_norm": 2.1602649688720703, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.765971302986145, + "num_tokens": 459277012.0, + "step": 18923 + }, + { + "epoch": 1.560484868475303, + "grad_norm": 1.8150279521942139, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7706380486488342, + "num_tokens": 459312157.0, + "step": 18924 + }, + { + "epoch": 1.5605673291003546, + "grad_norm": 1.8255103826522827, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7580536603927612, + "num_tokens": 459347103.0, + "step": 18925 + }, + { + "epoch": 1.5606497897254061, + "grad_norm": 1.8266972303390503, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7548104524612427, + "num_tokens": 459382320.0, + "step": 18926 + }, + { + "epoch": 1.5607322503504577, + "grad_norm": 1.9992741346359253, + "learning_rate": 1e-06, + "loss": 0.6018, + "mean_token_accuracy": 0.791621744632721, + "num_tokens": 459410121.0, + "step": 18927 + }, + { + "epoch": 1.5608147109755093, + "grad_norm": 2.102144956588745, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7720130681991577, + "num_tokens": 459438815.0, + "step": 18928 + }, + { + "epoch": 1.5608971716005606, + "grad_norm": 1.9140284061431885, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7290747165679932, + "num_tokens": 459470230.0, + "step": 18929 + }, + { + "epoch": 1.5609796322256124, + "grad_norm": 1.9824353456497192, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7184268236160278, + "num_tokens": 459501242.0, + "step": 18930 + }, + { + "epoch": 1.5610620928506638, + "grad_norm": 2.1124517917633057, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7746623158454895, + "num_tokens": 459528830.0, + "step": 18931 + }, + { + "epoch": 1.5611445534757153, + "grad_norm": 1.5782212018966675, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.7923303246498108, + "num_tokens": 459566146.0, + "step": 18932 + }, + { + "epoch": 1.561227014100767, + "grad_norm": 1.9693604707717896, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7624718546867371, + "num_tokens": 459595117.0, + "step": 18933 + }, + { + "epoch": 1.5613094747258183, + "grad_norm": 1.9676485061645508, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7464065551757812, + "num_tokens": 459625729.0, + "step": 18934 + }, + { + "epoch": 1.56139193535087, + "grad_norm": 2.172161817550659, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7407698035240173, + "num_tokens": 459654401.0, + "step": 18935 + }, + { + "epoch": 1.5614743959759214, + "grad_norm": 2.1701669692993164, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.774151623249054, + "num_tokens": 459682494.0, + "step": 18936 + }, + { + "epoch": 1.561556856600973, + "grad_norm": 1.8621946573257446, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7591267824172974, + "num_tokens": 459715759.0, + "step": 18937 + }, + { + "epoch": 1.5616393172260246, + "grad_norm": 1.7156950235366821, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7750517129898071, + "num_tokens": 459756072.0, + "step": 18938 + }, + { + "epoch": 1.5617217778510761, + "grad_norm": 2.177804946899414, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7324528098106384, + "num_tokens": 459781999.0, + "step": 18939 + }, + { + "epoch": 1.5618042384761277, + "grad_norm": 2.213245391845703, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7718124985694885, + "num_tokens": 459806438.0, + "step": 18940 + }, + { + "epoch": 1.561886699101179, + "grad_norm": 2.0530714988708496, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.733048141002655, + "num_tokens": 459835377.0, + "step": 18941 + }, + { + "epoch": 1.5619691597262308, + "grad_norm": 2.0048046112060547, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7341355681419373, + "num_tokens": 459869406.0, + "step": 18942 + }, + { + "epoch": 1.5620516203512822, + "grad_norm": 1.82896089553833, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7297504544258118, + "num_tokens": 459904796.0, + "step": 18943 + }, + { + "epoch": 1.5621340809763338, + "grad_norm": 1.730352520942688, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7352796792984009, + "num_tokens": 459941675.0, + "step": 18944 + }, + { + "epoch": 1.5622165416013853, + "grad_norm": 1.7679048776626587, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.7952191829681396, + "num_tokens": 459975026.0, + "step": 18945 + }, + { + "epoch": 1.562299002226437, + "grad_norm": 1.9201229810714722, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.74789959192276, + "num_tokens": 460005347.0, + "step": 18946 + }, + { + "epoch": 1.5623814628514885, + "grad_norm": 1.710287094116211, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7818799614906311, + "num_tokens": 460043897.0, + "step": 18947 + }, + { + "epoch": 1.5624639234765398, + "grad_norm": 2.2639858722686768, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.7881799340248108, + "num_tokens": 460067482.0, + "step": 18948 + }, + { + "epoch": 1.5625463841015916, + "grad_norm": 1.9695146083831787, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.770087718963623, + "num_tokens": 460100553.0, + "step": 18949 + }, + { + "epoch": 1.562628844726643, + "grad_norm": 1.7156264781951904, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7820366621017456, + "num_tokens": 460139193.0, + "step": 18950 + }, + { + "epoch": 1.5627113053516946, + "grad_norm": 2.0606849193573, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7749685049057007, + "num_tokens": 460168318.0, + "step": 18951 + }, + { + "epoch": 1.5627937659767461, + "grad_norm": 2.02950382232666, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7934577465057373, + "num_tokens": 460197082.0, + "step": 18952 + }, + { + "epoch": 1.5628762266017975, + "grad_norm": 2.012561798095703, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.7570674419403076, + "num_tokens": 460229016.0, + "step": 18953 + }, + { + "epoch": 1.5629586872268493, + "grad_norm": 2.098428249359131, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7142837047576904, + "num_tokens": 460255832.0, + "step": 18954 + }, + { + "epoch": 1.5630411478519006, + "grad_norm": 1.859994888305664, + "learning_rate": 1e-06, + "loss": 0.6134, + "mean_token_accuracy": 0.8034899234771729, + "num_tokens": 460287022.0, + "step": 18955 + }, + { + "epoch": 1.5631236084769524, + "grad_norm": 1.8755972385406494, + "learning_rate": 1e-06, + "loss": 0.7738, + "mean_token_accuracy": 0.7471358776092529, + "num_tokens": 460320668.0, + "step": 18956 + }, + { + "epoch": 1.5632060691020038, + "grad_norm": 2.144984245300293, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7649400234222412, + "num_tokens": 460347291.0, + "step": 18957 + }, + { + "epoch": 1.5632885297270553, + "grad_norm": 2.0205917358398438, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7594431638717651, + "num_tokens": 460374678.0, + "step": 18958 + }, + { + "epoch": 1.563370990352107, + "grad_norm": 1.800768256187439, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7639089822769165, + "num_tokens": 460411555.0, + "step": 18959 + }, + { + "epoch": 1.5634534509771583, + "grad_norm": 1.921284556388855, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.750443696975708, + "num_tokens": 460442803.0, + "step": 18960 + }, + { + "epoch": 1.56353591160221, + "grad_norm": 1.8754942417144775, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7673195004463196, + "num_tokens": 460476800.0, + "step": 18961 + }, + { + "epoch": 1.5636183722272614, + "grad_norm": 2.5291733741760254, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.7892105579376221, + "num_tokens": 460501095.0, + "step": 18962 + }, + { + "epoch": 1.563700832852313, + "grad_norm": 1.7204879522323608, + "learning_rate": 1e-06, + "loss": 0.5714, + "mean_token_accuracy": 0.806311845779419, + "num_tokens": 460536311.0, + "step": 18963 + }, + { + "epoch": 1.5637832934773646, + "grad_norm": 1.9913502931594849, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7187429070472717, + "num_tokens": 460568933.0, + "step": 18964 + }, + { + "epoch": 1.5638657541024161, + "grad_norm": 2.0890395641326904, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7372026443481445, + "num_tokens": 460600550.0, + "step": 18965 + }, + { + "epoch": 1.5639482147274677, + "grad_norm": 2.0344691276550293, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.7923208475112915, + "num_tokens": 460628511.0, + "step": 18966 + }, + { + "epoch": 1.564030675352519, + "grad_norm": 1.8269389867782593, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7874444723129272, + "num_tokens": 460661611.0, + "step": 18967 + }, + { + "epoch": 1.5641131359775708, + "grad_norm": 2.2019617557525635, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7667233347892761, + "num_tokens": 460687636.0, + "step": 18968 + }, + { + "epoch": 1.5641955966026222, + "grad_norm": 2.1625144481658936, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.753404438495636, + "num_tokens": 460713523.0, + "step": 18969 + }, + { + "epoch": 1.5642780572276738, + "grad_norm": 1.943027377128601, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7508711814880371, + "num_tokens": 460745504.0, + "step": 18970 + }, + { + "epoch": 1.5643605178527253, + "grad_norm": 1.9524022340774536, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7276800274848938, + "num_tokens": 460777166.0, + "step": 18971 + }, + { + "epoch": 1.564442978477777, + "grad_norm": 1.9425573348999023, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7529651522636414, + "num_tokens": 460808307.0, + "step": 18972 + }, + { + "epoch": 1.5645254391028285, + "grad_norm": 2.07161808013916, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7614192962646484, + "num_tokens": 460835228.0, + "step": 18973 + }, + { + "epoch": 1.5646078997278798, + "grad_norm": 2.1360220909118652, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7419525384902954, + "num_tokens": 460862122.0, + "step": 18974 + }, + { + "epoch": 1.5646903603529316, + "grad_norm": 1.724881649017334, + "learning_rate": 1e-06, + "loss": 0.555, + "mean_token_accuracy": 0.7998248338699341, + "num_tokens": 460895798.0, + "step": 18975 + }, + { + "epoch": 1.564772820977983, + "grad_norm": 2.0761959552764893, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7479840517044067, + "num_tokens": 460923724.0, + "step": 18976 + }, + { + "epoch": 1.5648552816030346, + "grad_norm": 1.7617934942245483, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7725566625595093, + "num_tokens": 460959773.0, + "step": 18977 + }, + { + "epoch": 1.5649377422280861, + "grad_norm": 1.730521559715271, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.7993872761726379, + "num_tokens": 460996144.0, + "step": 18978 + }, + { + "epoch": 1.5650202028531375, + "grad_norm": 2.10217547416687, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7616209983825684, + "num_tokens": 461023204.0, + "step": 18979 + }, + { + "epoch": 1.5651026634781893, + "grad_norm": 1.7690457105636597, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.7998241782188416, + "num_tokens": 461058115.0, + "step": 18980 + }, + { + "epoch": 1.5651851241032406, + "grad_norm": 2.20955228805542, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7482407093048096, + "num_tokens": 461084748.0, + "step": 18981 + }, + { + "epoch": 1.5652675847282922, + "grad_norm": 2.2924792766571045, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.778374195098877, + "num_tokens": 461107374.0, + "step": 18982 + }, + { + "epoch": 1.5653500453533438, + "grad_norm": 2.204308032989502, + "learning_rate": 1e-06, + "loss": 0.7775, + "mean_token_accuracy": 0.7496787905693054, + "num_tokens": 461134110.0, + "step": 18983 + }, + { + "epoch": 1.5654325059783953, + "grad_norm": 1.7355574369430542, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7246254086494446, + "num_tokens": 461169945.0, + "step": 18984 + }, + { + "epoch": 1.565514966603447, + "grad_norm": 1.9050828218460083, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.7934523224830627, + "num_tokens": 461201275.0, + "step": 18985 + }, + { + "epoch": 1.5655974272284983, + "grad_norm": 1.6045256853103638, + "learning_rate": 1e-06, + "loss": 0.5807, + "mean_token_accuracy": 0.8114994764328003, + "num_tokens": 461237991.0, + "step": 18986 + }, + { + "epoch": 1.56567988785355, + "grad_norm": 2.047229528427124, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7678046226501465, + "num_tokens": 461269193.0, + "step": 18987 + }, + { + "epoch": 1.5657623484786014, + "grad_norm": 2.33919358253479, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.727666437625885, + "num_tokens": 461291981.0, + "step": 18988 + }, + { + "epoch": 1.565844809103653, + "grad_norm": 2.399566173553467, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7160154581069946, + "num_tokens": 461314419.0, + "step": 18989 + }, + { + "epoch": 1.5659272697287046, + "grad_norm": 2.0763373374938965, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.7804306149482727, + "num_tokens": 461341969.0, + "step": 18990 + }, + { + "epoch": 1.5660097303537561, + "grad_norm": 1.7915287017822266, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7784891128540039, + "num_tokens": 461376218.0, + "step": 18991 + }, + { + "epoch": 1.5660921909788077, + "grad_norm": 1.8386719226837158, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.768137514591217, + "num_tokens": 461411618.0, + "step": 18992 + }, + { + "epoch": 1.566174651603859, + "grad_norm": 1.8153259754180908, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.7996364235877991, + "num_tokens": 461443605.0, + "step": 18993 + }, + { + "epoch": 1.5662571122289108, + "grad_norm": 1.9459797143936157, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.7882003784179688, + "num_tokens": 461472711.0, + "step": 18994 + }, + { + "epoch": 1.5663395728539622, + "grad_norm": 2.270904779434204, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7707545757293701, + "num_tokens": 461496095.0, + "step": 18995 + }, + { + "epoch": 1.5664220334790138, + "grad_norm": 1.9323465824127197, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7696642875671387, + "num_tokens": 461530859.0, + "step": 18996 + }, + { + "epoch": 1.5665044941040653, + "grad_norm": 2.1775779724121094, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7359716892242432, + "num_tokens": 461559121.0, + "step": 18997 + }, + { + "epoch": 1.5665869547291167, + "grad_norm": 2.376131772994995, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7394009232521057, + "num_tokens": 461581581.0, + "step": 18998 + }, + { + "epoch": 1.5666694153541685, + "grad_norm": 2.0011260509490967, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7563795447349548, + "num_tokens": 461611133.0, + "step": 18999 + }, + { + "epoch": 1.5667518759792198, + "grad_norm": 2.0776314735412598, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7585489749908447, + "num_tokens": 461642046.0, + "step": 19000 + }, + { + "epoch": 1.5668343366042714, + "grad_norm": 1.9720687866210938, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7438519597053528, + "num_tokens": 461671763.0, + "step": 19001 + }, + { + "epoch": 1.566916797229323, + "grad_norm": 1.8942259550094604, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7528719902038574, + "num_tokens": 461705635.0, + "step": 19002 + }, + { + "epoch": 1.5669992578543745, + "grad_norm": 2.0977585315704346, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.7846798896789551, + "num_tokens": 461731955.0, + "step": 19003 + }, + { + "epoch": 1.5670817184794261, + "grad_norm": 1.9436668157577515, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7729711532592773, + "num_tokens": 461762067.0, + "step": 19004 + }, + { + "epoch": 1.5671641791044775, + "grad_norm": 1.789982795715332, + "learning_rate": 1e-06, + "loss": 0.6086, + "mean_token_accuracy": 0.7948864102363586, + "num_tokens": 461794317.0, + "step": 19005 + }, + { + "epoch": 1.5672466397295293, + "grad_norm": 1.9825838804244995, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7583146095275879, + "num_tokens": 461823593.0, + "step": 19006 + }, + { + "epoch": 1.5673291003545806, + "grad_norm": 1.8786333799362183, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7509194612503052, + "num_tokens": 461857074.0, + "step": 19007 + }, + { + "epoch": 1.5674115609796322, + "grad_norm": 1.8452180624008179, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7554530501365662, + "num_tokens": 461894705.0, + "step": 19008 + }, + { + "epoch": 1.5674940216046838, + "grad_norm": 2.24222993850708, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7241280674934387, + "num_tokens": 461921304.0, + "step": 19009 + }, + { + "epoch": 1.5675764822297353, + "grad_norm": 2.0923070907592773, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.734235942363739, + "num_tokens": 461952012.0, + "step": 19010 + }, + { + "epoch": 1.567658942854787, + "grad_norm": 2.12119460105896, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7527475357055664, + "num_tokens": 461980039.0, + "step": 19011 + }, + { + "epoch": 1.5677414034798383, + "grad_norm": 2.4078588485717773, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7513784170150757, + "num_tokens": 462003830.0, + "step": 19012 + }, + { + "epoch": 1.56782386410489, + "grad_norm": 2.1106302738189697, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7721164226531982, + "num_tokens": 462030228.0, + "step": 19013 + }, + { + "epoch": 1.5679063247299414, + "grad_norm": 1.9285838603973389, + "learning_rate": 1e-06, + "loss": 0.7887, + "mean_token_accuracy": 0.7497084140777588, + "num_tokens": 462060439.0, + "step": 19014 + }, + { + "epoch": 1.567988785354993, + "grad_norm": 1.9246238470077515, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.7936011552810669, + "num_tokens": 462091176.0, + "step": 19015 + }, + { + "epoch": 1.5680712459800445, + "grad_norm": 2.3293209075927734, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7247916460037231, + "num_tokens": 462115425.0, + "step": 19016 + }, + { + "epoch": 1.5681537066050961, + "grad_norm": 2.369950294494629, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7724488973617554, + "num_tokens": 462139929.0, + "step": 19017 + }, + { + "epoch": 1.5682361672301477, + "grad_norm": 2.378185749053955, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7745144367218018, + "num_tokens": 462162084.0, + "step": 19018 + }, + { + "epoch": 1.568318627855199, + "grad_norm": 1.9285073280334473, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.765982985496521, + "num_tokens": 462194943.0, + "step": 19019 + }, + { + "epoch": 1.5684010884802508, + "grad_norm": 2.048224925994873, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7482825517654419, + "num_tokens": 462224075.0, + "step": 19020 + }, + { + "epoch": 1.5684835491053022, + "grad_norm": 1.8489800691604614, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7636210918426514, + "num_tokens": 462258910.0, + "step": 19021 + }, + { + "epoch": 1.5685660097303538, + "grad_norm": 2.040022373199463, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7092754244804382, + "num_tokens": 462288432.0, + "step": 19022 + }, + { + "epoch": 1.5686484703554053, + "grad_norm": 2.2546064853668213, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7590833902359009, + "num_tokens": 462320298.0, + "step": 19023 + }, + { + "epoch": 1.5687309309804567, + "grad_norm": 1.8764244318008423, + "learning_rate": 1e-06, + "loss": 0.5293, + "mean_token_accuracy": 0.8197526931762695, + "num_tokens": 462351846.0, + "step": 19024 + }, + { + "epoch": 1.5688133916055085, + "grad_norm": 1.831655502319336, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.770756185054779, + "num_tokens": 462386313.0, + "step": 19025 + }, + { + "epoch": 1.5688958522305598, + "grad_norm": 2.1369922161102295, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7661522626876831, + "num_tokens": 462413040.0, + "step": 19026 + }, + { + "epoch": 1.5689783128556114, + "grad_norm": 2.3446006774902344, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7630816698074341, + "num_tokens": 462437246.0, + "step": 19027 + }, + { + "epoch": 1.569060773480663, + "grad_norm": 2.2564585208892822, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.7933242321014404, + "num_tokens": 462465848.0, + "step": 19028 + }, + { + "epoch": 1.5691432341057145, + "grad_norm": 1.7537312507629395, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7588555216789246, + "num_tokens": 462504846.0, + "step": 19029 + }, + { + "epoch": 1.5692256947307661, + "grad_norm": 1.9241234064102173, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7753551006317139, + "num_tokens": 462535974.0, + "step": 19030 + }, + { + "epoch": 1.5693081553558175, + "grad_norm": 2.433501958847046, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7813950181007385, + "num_tokens": 462557778.0, + "step": 19031 + }, + { + "epoch": 1.5693906159808693, + "grad_norm": 2.129335641860962, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7669573426246643, + "num_tokens": 462584088.0, + "step": 19032 + }, + { + "epoch": 1.5694730766059206, + "grad_norm": 1.9447846412658691, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7470629811286926, + "num_tokens": 462614372.0, + "step": 19033 + }, + { + "epoch": 1.5695555372309722, + "grad_norm": 1.861502766609192, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7481756210327148, + "num_tokens": 462650045.0, + "step": 19034 + }, + { + "epoch": 1.5696379978560238, + "grad_norm": 2.2851803302764893, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7680826783180237, + "num_tokens": 462673393.0, + "step": 19035 + }, + { + "epoch": 1.5697204584810753, + "grad_norm": 1.9624533653259277, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7752401828765869, + "num_tokens": 462705754.0, + "step": 19036 + }, + { + "epoch": 1.569802919106127, + "grad_norm": 2.0901260375976562, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7589108943939209, + "num_tokens": 462735013.0, + "step": 19037 + }, + { + "epoch": 1.5698853797311783, + "grad_norm": 1.8280857801437378, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.7913261651992798, + "num_tokens": 462769040.0, + "step": 19038 + }, + { + "epoch": 1.56996784035623, + "grad_norm": 2.093188762664795, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7406521439552307, + "num_tokens": 462796160.0, + "step": 19039 + }, + { + "epoch": 1.5700503009812814, + "grad_norm": 1.9068423509597778, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7533228397369385, + "num_tokens": 462830659.0, + "step": 19040 + }, + { + "epoch": 1.570132761606333, + "grad_norm": 1.9710838794708252, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.7923986911773682, + "num_tokens": 462859427.0, + "step": 19041 + }, + { + "epoch": 1.5702152222313845, + "grad_norm": 2.084092617034912, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7610873579978943, + "num_tokens": 462886091.0, + "step": 19042 + }, + { + "epoch": 1.570297682856436, + "grad_norm": 2.005932331085205, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7419461607933044, + "num_tokens": 462915040.0, + "step": 19043 + }, + { + "epoch": 1.5703801434814877, + "grad_norm": 2.1695003509521484, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.723209798336029, + "num_tokens": 462943188.0, + "step": 19044 + }, + { + "epoch": 1.570462604106539, + "grad_norm": 1.8003100156784058, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7662479877471924, + "num_tokens": 462977022.0, + "step": 19045 + }, + { + "epoch": 1.5705450647315906, + "grad_norm": 2.0815858840942383, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7590823769569397, + "num_tokens": 463007648.0, + "step": 19046 + }, + { + "epoch": 1.5706275253566422, + "grad_norm": 1.8816378116607666, + "learning_rate": 1e-06, + "loss": 0.5902, + "mean_token_accuracy": 0.7936356067657471, + "num_tokens": 463039794.0, + "step": 19047 + }, + { + "epoch": 1.5707099859816938, + "grad_norm": 2.0465354919433594, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7557284832000732, + "num_tokens": 463068163.0, + "step": 19048 + }, + { + "epoch": 1.5707924466067453, + "grad_norm": 2.1264073848724365, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7693056464195251, + "num_tokens": 463094550.0, + "step": 19049 + }, + { + "epoch": 1.5708749072317967, + "grad_norm": 2.098775863647461, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7277673482894897, + "num_tokens": 463124984.0, + "step": 19050 + }, + { + "epoch": 1.5709573678568485, + "grad_norm": 2.220876455307007, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.7664497494697571, + "num_tokens": 463150457.0, + "step": 19051 + }, + { + "epoch": 1.5710398284818998, + "grad_norm": 1.8701748847961426, + "learning_rate": 1e-06, + "loss": 0.4752, + "mean_token_accuracy": 0.8275163173675537, + "num_tokens": 463178577.0, + "step": 19052 + }, + { + "epoch": 1.5711222891069514, + "grad_norm": 1.9043760299682617, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7608405351638794, + "num_tokens": 463211442.0, + "step": 19053 + }, + { + "epoch": 1.571204749732003, + "grad_norm": 2.2626912593841553, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7616719007492065, + "num_tokens": 463237925.0, + "step": 19054 + }, + { + "epoch": 1.5712872103570545, + "grad_norm": 1.8291116952896118, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7275214195251465, + "num_tokens": 463274559.0, + "step": 19055 + }, + { + "epoch": 1.5713696709821061, + "grad_norm": 1.9242388010025024, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7432378530502319, + "num_tokens": 463308942.0, + "step": 19056 + }, + { + "epoch": 1.5714521316071575, + "grad_norm": 1.896760106086731, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.795055091381073, + "num_tokens": 463340816.0, + "step": 19057 + }, + { + "epoch": 1.5715345922322093, + "grad_norm": 1.7298551797866821, + "learning_rate": 1e-06, + "loss": 0.5777, + "mean_token_accuracy": 0.8066166639328003, + "num_tokens": 463374758.0, + "step": 19058 + }, + { + "epoch": 1.5716170528572606, + "grad_norm": 1.9552381038665771, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7771288752555847, + "num_tokens": 463405699.0, + "step": 19059 + }, + { + "epoch": 1.5716995134823122, + "grad_norm": 1.9513572454452515, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7801315784454346, + "num_tokens": 463434887.0, + "step": 19060 + }, + { + "epoch": 1.5717819741073638, + "grad_norm": 2.1272943019866943, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7640488147735596, + "num_tokens": 463462288.0, + "step": 19061 + }, + { + "epoch": 1.571864434732415, + "grad_norm": 1.9449533224105835, + "learning_rate": 1e-06, + "loss": 0.7999, + "mean_token_accuracy": 0.7433294057846069, + "num_tokens": 463494810.0, + "step": 19062 + }, + { + "epoch": 1.571946895357467, + "grad_norm": 2.1147515773773193, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7602652907371521, + "num_tokens": 463523459.0, + "step": 19063 + }, + { + "epoch": 1.5720293559825183, + "grad_norm": 2.2749991416931152, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.787154495716095, + "num_tokens": 463551954.0, + "step": 19064 + }, + { + "epoch": 1.57211181660757, + "grad_norm": 2.196852922439575, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7238243222236633, + "num_tokens": 463579565.0, + "step": 19065 + }, + { + "epoch": 1.5721942772326214, + "grad_norm": 1.8060797452926636, + "learning_rate": 1e-06, + "loss": 0.5332, + "mean_token_accuracy": 0.8223022222518921, + "num_tokens": 463610642.0, + "step": 19066 + }, + { + "epoch": 1.572276737857673, + "grad_norm": 1.8719196319580078, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7757980227470398, + "num_tokens": 463639642.0, + "step": 19067 + }, + { + "epoch": 1.5723591984827245, + "grad_norm": 1.9406486749649048, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7421215772628784, + "num_tokens": 463672358.0, + "step": 19068 + }, + { + "epoch": 1.572441659107776, + "grad_norm": 1.916113018989563, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7889101505279541, + "num_tokens": 463703741.0, + "step": 19069 + }, + { + "epoch": 1.5725241197328277, + "grad_norm": 2.16951060295105, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7637357711791992, + "num_tokens": 463729521.0, + "step": 19070 + }, + { + "epoch": 1.572606580357879, + "grad_norm": 1.9807586669921875, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7699676156044006, + "num_tokens": 463759949.0, + "step": 19071 + }, + { + "epoch": 1.5726890409829306, + "grad_norm": 1.876633644104004, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7499234080314636, + "num_tokens": 463792939.0, + "step": 19072 + }, + { + "epoch": 1.5727715016079822, + "grad_norm": 2.1785295009613037, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7294142246246338, + "num_tokens": 463820694.0, + "step": 19073 + }, + { + "epoch": 1.5728539622330338, + "grad_norm": 2.1693005561828613, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.739072322845459, + "num_tokens": 463846077.0, + "step": 19074 + }, + { + "epoch": 1.5729364228580853, + "grad_norm": 1.9977765083312988, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7668876051902771, + "num_tokens": 463877252.0, + "step": 19075 + }, + { + "epoch": 1.5730188834831367, + "grad_norm": 2.052673578262329, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7389254570007324, + "num_tokens": 463906049.0, + "step": 19076 + }, + { + "epoch": 1.5731013441081885, + "grad_norm": 2.1174378395080566, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7257425785064697, + "num_tokens": 463934497.0, + "step": 19077 + }, + { + "epoch": 1.5731838047332398, + "grad_norm": 2.0995662212371826, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7649228572845459, + "num_tokens": 463963309.0, + "step": 19078 + }, + { + "epoch": 1.5732662653582914, + "grad_norm": 2.0095434188842773, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.7785733342170715, + "num_tokens": 463992591.0, + "step": 19079 + }, + { + "epoch": 1.573348725983343, + "grad_norm": 1.7176954746246338, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.7924615144729614, + "num_tokens": 464031125.0, + "step": 19080 + }, + { + "epoch": 1.5734311866083945, + "grad_norm": 2.0958945751190186, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7515091896057129, + "num_tokens": 464061476.0, + "step": 19081 + }, + { + "epoch": 1.5735136472334461, + "grad_norm": 1.9484236240386963, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7717394828796387, + "num_tokens": 464092558.0, + "step": 19082 + }, + { + "epoch": 1.5735961078584975, + "grad_norm": 2.1193838119506836, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.750626266002655, + "num_tokens": 464122553.0, + "step": 19083 + }, + { + "epoch": 1.5736785684835493, + "grad_norm": 2.3695621490478516, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7515054941177368, + "num_tokens": 464145507.0, + "step": 19084 + }, + { + "epoch": 1.5737610291086006, + "grad_norm": 1.9325367212295532, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7828458547592163, + "num_tokens": 464178276.0, + "step": 19085 + }, + { + "epoch": 1.5738434897336522, + "grad_norm": 1.982845664024353, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7723766565322876, + "num_tokens": 464207469.0, + "step": 19086 + }, + { + "epoch": 1.5739259503587038, + "grad_norm": 1.9124454259872437, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7661487460136414, + "num_tokens": 464239803.0, + "step": 19087 + }, + { + "epoch": 1.574008410983755, + "grad_norm": 2.2622275352478027, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7165763974189758, + "num_tokens": 464264383.0, + "step": 19088 + }, + { + "epoch": 1.574090871608807, + "grad_norm": 2.102715492248535, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7538142204284668, + "num_tokens": 464293117.0, + "step": 19089 + }, + { + "epoch": 1.5741733322338582, + "grad_norm": 1.7208712100982666, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7759828567504883, + "num_tokens": 464328319.0, + "step": 19090 + }, + { + "epoch": 1.5742557928589098, + "grad_norm": 2.5984675884246826, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.702972412109375, + "num_tokens": 464347394.0, + "step": 19091 + }, + { + "epoch": 1.5743382534839614, + "grad_norm": 2.046668767929077, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7763449549674988, + "num_tokens": 464376232.0, + "step": 19092 + }, + { + "epoch": 1.574420714109013, + "grad_norm": 1.7620763778686523, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7875774502754211, + "num_tokens": 464411623.0, + "step": 19093 + }, + { + "epoch": 1.5745031747340645, + "grad_norm": 1.9142560958862305, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.7632884979248047, + "num_tokens": 464442742.0, + "step": 19094 + }, + { + "epoch": 1.5745856353591159, + "grad_norm": 2.1953108310699463, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.7952914237976074, + "num_tokens": 464466190.0, + "step": 19095 + }, + { + "epoch": 1.5746680959841677, + "grad_norm": 1.9903016090393066, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.752508282661438, + "num_tokens": 464495690.0, + "step": 19096 + }, + { + "epoch": 1.574750556609219, + "grad_norm": 2.0734329223632812, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.7764031291007996, + "num_tokens": 464524708.0, + "step": 19097 + }, + { + "epoch": 1.5748330172342706, + "grad_norm": 1.9067715406417847, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7661094069480896, + "num_tokens": 464556399.0, + "step": 19098 + }, + { + "epoch": 1.5749154778593222, + "grad_norm": 1.9688668251037598, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7388719320297241, + "num_tokens": 464587589.0, + "step": 19099 + }, + { + "epoch": 1.5749979384843737, + "grad_norm": 1.6614757776260376, + "learning_rate": 1e-06, + "loss": 0.5588, + "mean_token_accuracy": 0.8110620975494385, + "num_tokens": 464626310.0, + "step": 19100 + }, + { + "epoch": 1.5750803991094253, + "grad_norm": 2.051234245300293, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.762914776802063, + "num_tokens": 464652648.0, + "step": 19101 + }, + { + "epoch": 1.5751628597344767, + "grad_norm": 2.0745480060577393, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.7605429887771606, + "num_tokens": 464680033.0, + "step": 19102 + }, + { + "epoch": 1.5752453203595285, + "grad_norm": 1.9286282062530518, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7609389424324036, + "num_tokens": 464711165.0, + "step": 19103 + }, + { + "epoch": 1.5753277809845798, + "grad_norm": 2.1635100841522217, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.740497350692749, + "num_tokens": 464735587.0, + "step": 19104 + }, + { + "epoch": 1.5754102416096314, + "grad_norm": 1.9309961795806885, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7439337968826294, + "num_tokens": 464765946.0, + "step": 19105 + }, + { + "epoch": 1.575492702234683, + "grad_norm": 1.8989108800888062, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7803291082382202, + "num_tokens": 464798309.0, + "step": 19106 + }, + { + "epoch": 1.5755751628597343, + "grad_norm": 1.9699459075927734, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7267298102378845, + "num_tokens": 464827910.0, + "step": 19107 + }, + { + "epoch": 1.575657623484786, + "grad_norm": 1.866243839263916, + "learning_rate": 1e-06, + "loss": 0.5897, + "mean_token_accuracy": 0.8074714541435242, + "num_tokens": 464858063.0, + "step": 19108 + }, + { + "epoch": 1.5757400841098375, + "grad_norm": 2.032362222671509, + "learning_rate": 1e-06, + "loss": 0.783, + "mean_token_accuracy": 0.7543917298316956, + "num_tokens": 464886283.0, + "step": 19109 + }, + { + "epoch": 1.5758225447348893, + "grad_norm": 1.7380670309066772, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7905118465423584, + "num_tokens": 464921556.0, + "step": 19110 + }, + { + "epoch": 1.5759050053599406, + "grad_norm": 1.986194372177124, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.7833408117294312, + "num_tokens": 464949692.0, + "step": 19111 + }, + { + "epoch": 1.5759874659849922, + "grad_norm": 2.009982109069824, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7617146968841553, + "num_tokens": 464980828.0, + "step": 19112 + }, + { + "epoch": 1.5760699266100437, + "grad_norm": 2.106269598007202, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7636215686798096, + "num_tokens": 465009724.0, + "step": 19113 + }, + { + "epoch": 1.576152387235095, + "grad_norm": 2.0830204486846924, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7281705141067505, + "num_tokens": 465036540.0, + "step": 19114 + }, + { + "epoch": 1.576234847860147, + "grad_norm": 1.9963709115982056, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7629241943359375, + "num_tokens": 465066777.0, + "step": 19115 + }, + { + "epoch": 1.5763173084851982, + "grad_norm": 2.01773738861084, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7506834268569946, + "num_tokens": 465096605.0, + "step": 19116 + }, + { + "epoch": 1.5763997691102498, + "grad_norm": 2.098573684692383, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7245820164680481, + "num_tokens": 465124823.0, + "step": 19117 + }, + { + "epoch": 1.5764822297353014, + "grad_norm": 2.0356240272521973, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.754799485206604, + "num_tokens": 465153472.0, + "step": 19118 + }, + { + "epoch": 1.576564690360353, + "grad_norm": 2.0599849224090576, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.7745704650878906, + "num_tokens": 465182483.0, + "step": 19119 + }, + { + "epoch": 1.5766471509854045, + "grad_norm": 1.9204381704330444, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7766146063804626, + "num_tokens": 465210191.0, + "step": 19120 + }, + { + "epoch": 1.5767296116104559, + "grad_norm": 2.435511827468872, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7786978483200073, + "num_tokens": 465229892.0, + "step": 19121 + }, + { + "epoch": 1.5768120722355077, + "grad_norm": 2.1616365909576416, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7437604665756226, + "num_tokens": 465256347.0, + "step": 19122 + }, + { + "epoch": 1.576894532860559, + "grad_norm": 1.7517058849334717, + "learning_rate": 1e-06, + "loss": 0.5589, + "mean_token_accuracy": 0.8140993714332581, + "num_tokens": 465291481.0, + "step": 19123 + }, + { + "epoch": 1.5769769934856106, + "grad_norm": 2.2039742469787598, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.767989993095398, + "num_tokens": 465316442.0, + "step": 19124 + }, + { + "epoch": 1.5770594541106622, + "grad_norm": 2.294036865234375, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7159607410430908, + "num_tokens": 465340583.0, + "step": 19125 + }, + { + "epoch": 1.5771419147357137, + "grad_norm": 2.1217188835144043, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7718372344970703, + "num_tokens": 465366351.0, + "step": 19126 + }, + { + "epoch": 1.5772243753607653, + "grad_norm": 1.993444561958313, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7897791266441345, + "num_tokens": 465397276.0, + "step": 19127 + }, + { + "epoch": 1.5773068359858167, + "grad_norm": 1.7453844547271729, + "learning_rate": 1e-06, + "loss": 0.4821, + "mean_token_accuracy": 0.8383220434188843, + "num_tokens": 465431209.0, + "step": 19128 + }, + { + "epoch": 1.5773892966108685, + "grad_norm": 2.1807668209075928, + "learning_rate": 1e-06, + "loss": 0.7617, + "mean_token_accuracy": 0.7611663937568665, + "num_tokens": 465457954.0, + "step": 19129 + }, + { + "epoch": 1.5774717572359198, + "grad_norm": 2.0351061820983887, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7306296825408936, + "num_tokens": 465489948.0, + "step": 19130 + }, + { + "epoch": 1.5775542178609714, + "grad_norm": 2.36444354057312, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7487321496009827, + "num_tokens": 465513018.0, + "step": 19131 + }, + { + "epoch": 1.577636678486023, + "grad_norm": 2.007420539855957, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7459336519241333, + "num_tokens": 465543023.0, + "step": 19132 + }, + { + "epoch": 1.5777191391110743, + "grad_norm": 2.007100820541382, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.74592125415802, + "num_tokens": 465575411.0, + "step": 19133 + }, + { + "epoch": 1.577801599736126, + "grad_norm": 1.8967355489730835, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7503626346588135, + "num_tokens": 465608932.0, + "step": 19134 + }, + { + "epoch": 1.5778840603611775, + "grad_norm": 1.8809256553649902, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7790192365646362, + "num_tokens": 465642473.0, + "step": 19135 + }, + { + "epoch": 1.577966520986229, + "grad_norm": 1.9771310091018677, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.727530837059021, + "num_tokens": 465674050.0, + "step": 19136 + }, + { + "epoch": 1.5780489816112806, + "grad_norm": 2.0324888229370117, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7103562355041504, + "num_tokens": 465704658.0, + "step": 19137 + }, + { + "epoch": 1.5781314422363322, + "grad_norm": 1.962465524673462, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.8028880953788757, + "num_tokens": 465732254.0, + "step": 19138 + }, + { + "epoch": 1.5782139028613837, + "grad_norm": 2.1740384101867676, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7423690557479858, + "num_tokens": 465759281.0, + "step": 19139 + }, + { + "epoch": 1.578296363486435, + "grad_norm": 2.1024327278137207, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7653828263282776, + "num_tokens": 465788386.0, + "step": 19140 + }, + { + "epoch": 1.5783788241114869, + "grad_norm": 2.021432638168335, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7784899473190308, + "num_tokens": 465819274.0, + "step": 19141 + }, + { + "epoch": 1.5784612847365382, + "grad_norm": 2.219247341156006, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7690560221672058, + "num_tokens": 465844874.0, + "step": 19142 + }, + { + "epoch": 1.5785437453615898, + "grad_norm": 1.9987090826034546, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7186415195465088, + "num_tokens": 465876053.0, + "step": 19143 + }, + { + "epoch": 1.5786262059866414, + "grad_norm": 1.7950834035873413, + "learning_rate": 1e-06, + "loss": 0.781, + "mean_token_accuracy": 0.7544087171554565, + "num_tokens": 465910901.0, + "step": 19144 + }, + { + "epoch": 1.578708666611693, + "grad_norm": 2.117630958557129, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7727426290512085, + "num_tokens": 465935682.0, + "step": 19145 + }, + { + "epoch": 1.5787911272367445, + "grad_norm": 1.9603880643844604, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7438931465148926, + "num_tokens": 465965311.0, + "step": 19146 + }, + { + "epoch": 1.5788735878617959, + "grad_norm": 2.13604736328125, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.7869728207588196, + "num_tokens": 465990421.0, + "step": 19147 + }, + { + "epoch": 1.5789560484868477, + "grad_norm": 1.9331258535385132, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7671425342559814, + "num_tokens": 466022265.0, + "step": 19148 + }, + { + "epoch": 1.579038509111899, + "grad_norm": 1.8414735794067383, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7488546967506409, + "num_tokens": 466059443.0, + "step": 19149 + }, + { + "epoch": 1.5791209697369506, + "grad_norm": 1.797099232673645, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7527937889099121, + "num_tokens": 466093897.0, + "step": 19150 + }, + { + "epoch": 1.5792034303620022, + "grad_norm": 1.8737486600875854, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.7912495136260986, + "num_tokens": 466123070.0, + "step": 19151 + }, + { + "epoch": 1.5792858909870535, + "grad_norm": 2.075010299682617, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7711011171340942, + "num_tokens": 466151509.0, + "step": 19152 + }, + { + "epoch": 1.5793683516121053, + "grad_norm": 1.8979907035827637, + "learning_rate": 1e-06, + "loss": 0.5614, + "mean_token_accuracy": 0.813652753829956, + "num_tokens": 466182359.0, + "step": 19153 + }, + { + "epoch": 1.5794508122371567, + "grad_norm": 1.9096593856811523, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7491095066070557, + "num_tokens": 466212507.0, + "step": 19154 + }, + { + "epoch": 1.5795332728622085, + "grad_norm": 2.181506633758545, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7279901504516602, + "num_tokens": 466240296.0, + "step": 19155 + }, + { + "epoch": 1.5796157334872598, + "grad_norm": 1.8502205610275269, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7696681618690491, + "num_tokens": 466277509.0, + "step": 19156 + }, + { + "epoch": 1.5796981941123114, + "grad_norm": 2.15818452835083, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7688146829605103, + "num_tokens": 466302465.0, + "step": 19157 + }, + { + "epoch": 1.579780654737363, + "grad_norm": 2.3190035820007324, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7660108208656311, + "num_tokens": 466329188.0, + "step": 19158 + }, + { + "epoch": 1.5798631153624143, + "grad_norm": 1.8424034118652344, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7474839687347412, + "num_tokens": 466362361.0, + "step": 19159 + }, + { + "epoch": 1.579945575987466, + "grad_norm": 1.8227826356887817, + "learning_rate": 1e-06, + "loss": 0.624, + "mean_token_accuracy": 0.797139048576355, + "num_tokens": 466396745.0, + "step": 19160 + }, + { + "epoch": 1.5800280366125175, + "grad_norm": 1.943009376525879, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.7973184585571289, + "num_tokens": 466427507.0, + "step": 19161 + }, + { + "epoch": 1.580110497237569, + "grad_norm": 2.2274909019470215, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.7821760177612305, + "num_tokens": 466451781.0, + "step": 19162 + }, + { + "epoch": 1.5801929578626206, + "grad_norm": 1.7974395751953125, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7487186193466187, + "num_tokens": 466486429.0, + "step": 19163 + }, + { + "epoch": 1.5802754184876722, + "grad_norm": 2.160820484161377, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.76018226146698, + "num_tokens": 466513205.0, + "step": 19164 + }, + { + "epoch": 1.5803578791127237, + "grad_norm": 1.84684419631958, + "learning_rate": 1e-06, + "loss": 0.5924, + "mean_token_accuracy": 0.8013186454772949, + "num_tokens": 466544362.0, + "step": 19165 + }, + { + "epoch": 1.580440339737775, + "grad_norm": 1.7767916917800903, + "learning_rate": 1e-06, + "loss": 0.7303, + "mean_token_accuracy": 0.7727212905883789, + "num_tokens": 466579153.0, + "step": 19166 + }, + { + "epoch": 1.5805228003628269, + "grad_norm": 1.8046371936798096, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.7945054173469543, + "num_tokens": 466610820.0, + "step": 19167 + }, + { + "epoch": 1.5806052609878782, + "grad_norm": 2.148409605026245, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.7501716613769531, + "num_tokens": 466640775.0, + "step": 19168 + }, + { + "epoch": 1.5806877216129298, + "grad_norm": 1.6916124820709229, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.762404203414917, + "num_tokens": 466679193.0, + "step": 19169 + }, + { + "epoch": 1.5807701822379814, + "grad_norm": 2.0870213508605957, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.750469446182251, + "num_tokens": 466709743.0, + "step": 19170 + }, + { + "epoch": 1.580852642863033, + "grad_norm": 2.3544983863830566, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.7817678451538086, + "num_tokens": 466734461.0, + "step": 19171 + }, + { + "epoch": 1.5809351034880845, + "grad_norm": 2.0835702419281006, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.7563645839691162, + "num_tokens": 466764423.0, + "step": 19172 + }, + { + "epoch": 1.5810175641131359, + "grad_norm": 1.8762379884719849, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7810378074645996, + "num_tokens": 466796060.0, + "step": 19173 + }, + { + "epoch": 1.5811000247381877, + "grad_norm": 1.9000556468963623, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.7858489155769348, + "num_tokens": 466825913.0, + "step": 19174 + }, + { + "epoch": 1.581182485363239, + "grad_norm": 1.819841980934143, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7799578309059143, + "num_tokens": 466860864.0, + "step": 19175 + }, + { + "epoch": 1.5812649459882906, + "grad_norm": 1.5896438360214233, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.7906093001365662, + "num_tokens": 466901065.0, + "step": 19176 + }, + { + "epoch": 1.5813474066133422, + "grad_norm": 2.1138832569122314, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7469598650932312, + "num_tokens": 466930710.0, + "step": 19177 + }, + { + "epoch": 1.5814298672383935, + "grad_norm": 2.288134813308716, + "learning_rate": 1e-06, + "loss": 0.7722, + "mean_token_accuracy": 0.7607216835021973, + "num_tokens": 466953998.0, + "step": 19178 + }, + { + "epoch": 1.5815123278634453, + "grad_norm": 2.165395498275757, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7444383502006531, + "num_tokens": 466979218.0, + "step": 19179 + }, + { + "epoch": 1.5815947884884967, + "grad_norm": 1.7502485513687134, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7785442471504211, + "num_tokens": 467017334.0, + "step": 19180 + }, + { + "epoch": 1.5816772491135482, + "grad_norm": 1.8971941471099854, + "learning_rate": 1e-06, + "loss": 0.5841, + "mean_token_accuracy": 0.8117203712463379, + "num_tokens": 467049347.0, + "step": 19181 + }, + { + "epoch": 1.5817597097385998, + "grad_norm": 2.082960605621338, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7701693177223206, + "num_tokens": 467075935.0, + "step": 19182 + }, + { + "epoch": 1.5818421703636514, + "grad_norm": 2.100160837173462, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7612341642379761, + "num_tokens": 467106135.0, + "step": 19183 + }, + { + "epoch": 1.581924630988703, + "grad_norm": 2.2839925289154053, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.747309148311615, + "num_tokens": 467132230.0, + "step": 19184 + }, + { + "epoch": 1.5820070916137543, + "grad_norm": 2.240485191345215, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7538528442382812, + "num_tokens": 467159243.0, + "step": 19185 + }, + { + "epoch": 1.582089552238806, + "grad_norm": 2.2094693183898926, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7503848075866699, + "num_tokens": 467185988.0, + "step": 19186 + }, + { + "epoch": 1.5821720128638574, + "grad_norm": 1.9750529527664185, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7483084201812744, + "num_tokens": 467213995.0, + "step": 19187 + }, + { + "epoch": 1.582254473488909, + "grad_norm": 2.275041341781616, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7489814758300781, + "num_tokens": 467237142.0, + "step": 19188 + }, + { + "epoch": 1.5823369341139606, + "grad_norm": 1.9813125133514404, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7530953288078308, + "num_tokens": 467268399.0, + "step": 19189 + }, + { + "epoch": 1.5824193947390122, + "grad_norm": 1.9261138439178467, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7777345180511475, + "num_tokens": 467298695.0, + "step": 19190 + }, + { + "epoch": 1.5825018553640637, + "grad_norm": 1.8048394918441772, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7584234476089478, + "num_tokens": 467336023.0, + "step": 19191 + }, + { + "epoch": 1.582584315989115, + "grad_norm": 1.871872901916504, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.766057550907135, + "num_tokens": 467365350.0, + "step": 19192 + }, + { + "epoch": 1.5826667766141669, + "grad_norm": 1.9944387674331665, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7526779174804688, + "num_tokens": 467392890.0, + "step": 19193 + }, + { + "epoch": 1.5827492372392182, + "grad_norm": 1.875388264656067, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7504103183746338, + "num_tokens": 467423195.0, + "step": 19194 + }, + { + "epoch": 1.5828316978642698, + "grad_norm": 1.9161388874053955, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7678533792495728, + "num_tokens": 467460293.0, + "step": 19195 + }, + { + "epoch": 1.5829141584893214, + "grad_norm": 2.1834638118743896, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7559529542922974, + "num_tokens": 467484611.0, + "step": 19196 + }, + { + "epoch": 1.5829966191143727, + "grad_norm": 1.87860107421875, + "learning_rate": 1e-06, + "loss": 0.5535, + "mean_token_accuracy": 0.8143067359924316, + "num_tokens": 467514374.0, + "step": 19197 + }, + { + "epoch": 1.5830790797394245, + "grad_norm": 2.050715446472168, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.760834813117981, + "num_tokens": 467543607.0, + "step": 19198 + }, + { + "epoch": 1.5831615403644759, + "grad_norm": 1.9702736139297485, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.7911455631256104, + "num_tokens": 467573745.0, + "step": 19199 + }, + { + "epoch": 1.5832440009895277, + "grad_norm": 1.9076811075210571, + "learning_rate": 1e-06, + "loss": 0.7799, + "mean_token_accuracy": 0.7577211856842041, + "num_tokens": 467603355.0, + "step": 19200 + }, + { + "epoch": 1.583326461614579, + "grad_norm": 1.9177542924880981, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7213018536567688, + "num_tokens": 467637056.0, + "step": 19201 + }, + { + "epoch": 1.5834089222396306, + "grad_norm": 2.0570170879364014, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7379180192947388, + "num_tokens": 467666759.0, + "step": 19202 + }, + { + "epoch": 1.5834913828646822, + "grad_norm": 1.9523426294326782, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7570286989212036, + "num_tokens": 467696637.0, + "step": 19203 + }, + { + "epoch": 1.5835738434897335, + "grad_norm": 1.9788434505462646, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7632740139961243, + "num_tokens": 467726249.0, + "step": 19204 + }, + { + "epoch": 1.5836563041147853, + "grad_norm": 2.437441110610962, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7573704123497009, + "num_tokens": 467748277.0, + "step": 19205 + }, + { + "epoch": 1.5837387647398367, + "grad_norm": 1.9805591106414795, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7301556468009949, + "num_tokens": 467781118.0, + "step": 19206 + }, + { + "epoch": 1.5838212253648882, + "grad_norm": 2.2144417762756348, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7688408493995667, + "num_tokens": 467804043.0, + "step": 19207 + }, + { + "epoch": 1.5839036859899398, + "grad_norm": 1.8851110935211182, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7490193247795105, + "num_tokens": 467837693.0, + "step": 19208 + }, + { + "epoch": 1.5839861466149914, + "grad_norm": 2.0736353397369385, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7597702741622925, + "num_tokens": 467867657.0, + "step": 19209 + }, + { + "epoch": 1.584068607240043, + "grad_norm": 1.9910527467727661, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7652196884155273, + "num_tokens": 467895800.0, + "step": 19210 + }, + { + "epoch": 1.5841510678650943, + "grad_norm": 1.975777268409729, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7896708250045776, + "num_tokens": 467926480.0, + "step": 19211 + }, + { + "epoch": 1.584233528490146, + "grad_norm": 1.8223793506622314, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7670726180076599, + "num_tokens": 467961473.0, + "step": 19212 + }, + { + "epoch": 1.5843159891151974, + "grad_norm": 1.8278095722198486, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7717416286468506, + "num_tokens": 467996776.0, + "step": 19213 + }, + { + "epoch": 1.584398449740249, + "grad_norm": 2.047678232192993, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7481768131256104, + "num_tokens": 468027980.0, + "step": 19214 + }, + { + "epoch": 1.5844809103653006, + "grad_norm": 1.9846233129501343, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7595858573913574, + "num_tokens": 468056547.0, + "step": 19215 + }, + { + "epoch": 1.5845633709903522, + "grad_norm": 2.0918545722961426, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7427120804786682, + "num_tokens": 468084650.0, + "step": 19216 + }, + { + "epoch": 1.5846458316154037, + "grad_norm": 1.8106306791305542, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.7849693894386292, + "num_tokens": 468121368.0, + "step": 19217 + }, + { + "epoch": 1.584728292240455, + "grad_norm": 1.9273500442504883, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.8010703325271606, + "num_tokens": 468148319.0, + "step": 19218 + }, + { + "epoch": 1.5848107528655069, + "grad_norm": 2.0988521575927734, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7421910166740417, + "num_tokens": 468175900.0, + "step": 19219 + }, + { + "epoch": 1.5848932134905582, + "grad_norm": 2.121706247329712, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7188783884048462, + "num_tokens": 468204379.0, + "step": 19220 + }, + { + "epoch": 1.5849756741156098, + "grad_norm": 2.064631938934326, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.79954993724823, + "num_tokens": 468232602.0, + "step": 19221 + }, + { + "epoch": 1.5850581347406614, + "grad_norm": 2.051948070526123, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7443369626998901, + "num_tokens": 468263127.0, + "step": 19222 + }, + { + "epoch": 1.5851405953657127, + "grad_norm": 1.7084518671035767, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.7867486476898193, + "num_tokens": 468300268.0, + "step": 19223 + }, + { + "epoch": 1.5852230559907645, + "grad_norm": 2.121276617050171, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7546491026878357, + "num_tokens": 468327350.0, + "step": 19224 + }, + { + "epoch": 1.5853055166158159, + "grad_norm": 1.8965389728546143, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.7871174812316895, + "num_tokens": 468360738.0, + "step": 19225 + }, + { + "epoch": 1.5853879772408674, + "grad_norm": 1.9013547897338867, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.746248722076416, + "num_tokens": 468395000.0, + "step": 19226 + }, + { + "epoch": 1.585470437865919, + "grad_norm": 1.8838858604431152, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.756920337677002, + "num_tokens": 468431060.0, + "step": 19227 + }, + { + "epoch": 1.5855528984909706, + "grad_norm": 2.0096921920776367, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.7784281969070435, + "num_tokens": 468460115.0, + "step": 19228 + }, + { + "epoch": 1.5856353591160222, + "grad_norm": 2.3000197410583496, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7643318176269531, + "num_tokens": 468484200.0, + "step": 19229 + }, + { + "epoch": 1.5857178197410735, + "grad_norm": 1.8729724884033203, + "learning_rate": 1e-06, + "loss": 0.7621, + "mean_token_accuracy": 0.755122184753418, + "num_tokens": 468517206.0, + "step": 19230 + }, + { + "epoch": 1.5858002803661253, + "grad_norm": 1.781708002090454, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7092033624649048, + "num_tokens": 468557452.0, + "step": 19231 + }, + { + "epoch": 1.5858827409911767, + "grad_norm": 2.0404584407806396, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7869199514389038, + "num_tokens": 468583919.0, + "step": 19232 + }, + { + "epoch": 1.5859652016162282, + "grad_norm": 1.898029088973999, + "learning_rate": 1e-06, + "loss": 0.5203, + "mean_token_accuracy": 0.8218634724617004, + "num_tokens": 468613957.0, + "step": 19233 + }, + { + "epoch": 1.5860476622412798, + "grad_norm": 2.3315834999084473, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7451395988464355, + "num_tokens": 468637844.0, + "step": 19234 + }, + { + "epoch": 1.5861301228663314, + "grad_norm": 2.1145830154418945, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7513914108276367, + "num_tokens": 468664857.0, + "step": 19235 + }, + { + "epoch": 1.586212583491383, + "grad_norm": 2.1507604122161865, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.7797256112098694, + "num_tokens": 468690075.0, + "step": 19236 + }, + { + "epoch": 1.5862950441164343, + "grad_norm": 1.9449448585510254, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7711049318313599, + "num_tokens": 468718925.0, + "step": 19237 + }, + { + "epoch": 1.586377504741486, + "grad_norm": 2.403580904006958, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.7818077802658081, + "num_tokens": 468740332.0, + "step": 19238 + }, + { + "epoch": 1.5864599653665374, + "grad_norm": 1.9708693027496338, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7766895294189453, + "num_tokens": 468770254.0, + "step": 19239 + }, + { + "epoch": 1.586542425991589, + "grad_norm": 1.8495965003967285, + "learning_rate": 1e-06, + "loss": 0.5716, + "mean_token_accuracy": 0.8088077306747437, + "num_tokens": 468800785.0, + "step": 19240 + }, + { + "epoch": 1.5866248866166406, + "grad_norm": 1.9740036725997925, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7656600475311279, + "num_tokens": 468834711.0, + "step": 19241 + }, + { + "epoch": 1.586707347241692, + "grad_norm": 1.9699957370758057, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7414737343788147, + "num_tokens": 468866258.0, + "step": 19242 + }, + { + "epoch": 1.5867898078667437, + "grad_norm": 2.0597047805786133, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.7581443786621094, + "num_tokens": 468896429.0, + "step": 19243 + }, + { + "epoch": 1.586872268491795, + "grad_norm": 1.8186802864074707, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7665953636169434, + "num_tokens": 468930198.0, + "step": 19244 + }, + { + "epoch": 1.5869547291168467, + "grad_norm": 1.9736942052841187, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7618033289909363, + "num_tokens": 468961091.0, + "step": 19245 + }, + { + "epoch": 1.5870371897418982, + "grad_norm": 2.0024850368499756, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7493404150009155, + "num_tokens": 468992355.0, + "step": 19246 + }, + { + "epoch": 1.5871196503669498, + "grad_norm": 1.888327717781067, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.787419855594635, + "num_tokens": 469023890.0, + "step": 19247 + }, + { + "epoch": 1.5872021109920014, + "grad_norm": 2.0523178577423096, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7283339500427246, + "num_tokens": 469053541.0, + "step": 19248 + }, + { + "epoch": 1.5872845716170527, + "grad_norm": 2.1676626205444336, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7565401792526245, + "num_tokens": 469082670.0, + "step": 19249 + }, + { + "epoch": 1.5873670322421045, + "grad_norm": 2.013720750808716, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7859068512916565, + "num_tokens": 469108741.0, + "step": 19250 + }, + { + "epoch": 1.5874494928671559, + "grad_norm": 2.114593029022217, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7656461596488953, + "num_tokens": 469133697.0, + "step": 19251 + }, + { + "epoch": 1.5875319534922074, + "grad_norm": 2.1350467205047607, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7547725439071655, + "num_tokens": 469160963.0, + "step": 19252 + }, + { + "epoch": 1.587614414117259, + "grad_norm": 1.8110500574111938, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7732868194580078, + "num_tokens": 469195259.0, + "step": 19253 + }, + { + "epoch": 1.5876968747423106, + "grad_norm": 1.8111367225646973, + "learning_rate": 1e-06, + "loss": 0.7483, + "mean_token_accuracy": 0.7675852179527283, + "num_tokens": 469230515.0, + "step": 19254 + }, + { + "epoch": 1.5877793353673622, + "grad_norm": 2.05642032623291, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7626319527626038, + "num_tokens": 469257427.0, + "step": 19255 + }, + { + "epoch": 1.5878617959924135, + "grad_norm": 1.78477144241333, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.7911186814308167, + "num_tokens": 469289065.0, + "step": 19256 + }, + { + "epoch": 1.5879442566174653, + "grad_norm": 2.075014591217041, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7574400305747986, + "num_tokens": 469317247.0, + "step": 19257 + }, + { + "epoch": 1.5880267172425166, + "grad_norm": 2.016563892364502, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7763419151306152, + "num_tokens": 469345971.0, + "step": 19258 + }, + { + "epoch": 1.5881091778675682, + "grad_norm": 2.0537784099578857, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7863671183586121, + "num_tokens": 469375282.0, + "step": 19259 + }, + { + "epoch": 1.5881916384926198, + "grad_norm": 1.959073543548584, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7504832744598389, + "num_tokens": 469406572.0, + "step": 19260 + }, + { + "epoch": 1.5882740991176714, + "grad_norm": 2.1576051712036133, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.74273681640625, + "num_tokens": 469431792.0, + "step": 19261 + }, + { + "epoch": 1.588356559742723, + "grad_norm": 2.0698511600494385, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7268238067626953, + "num_tokens": 469459808.0, + "step": 19262 + }, + { + "epoch": 1.5884390203677743, + "grad_norm": 1.914900779724121, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7745028138160706, + "num_tokens": 469493520.0, + "step": 19263 + }, + { + "epoch": 1.588521480992826, + "grad_norm": 2.0332584381103516, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7601410150527954, + "num_tokens": 469522720.0, + "step": 19264 + }, + { + "epoch": 1.5886039416178774, + "grad_norm": 2.1015727519989014, + "learning_rate": 1e-06, + "loss": 0.7531, + "mean_token_accuracy": 0.7579365372657776, + "num_tokens": 469552961.0, + "step": 19265 + }, + { + "epoch": 1.588686402242929, + "grad_norm": 2.146380662918091, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7529867887496948, + "num_tokens": 469577845.0, + "step": 19266 + }, + { + "epoch": 1.5887688628679806, + "grad_norm": 2.0293772220611572, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.7923462986946106, + "num_tokens": 469609567.0, + "step": 19267 + }, + { + "epoch": 1.588851323493032, + "grad_norm": 2.237194061279297, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7316758632659912, + "num_tokens": 469636035.0, + "step": 19268 + }, + { + "epoch": 1.5889337841180837, + "grad_norm": 1.8005530834197998, + "learning_rate": 1e-06, + "loss": 0.5681, + "mean_token_accuracy": 0.8103978633880615, + "num_tokens": 469666297.0, + "step": 19269 + }, + { + "epoch": 1.589016244743135, + "grad_norm": 2.135485887527466, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7533539533615112, + "num_tokens": 469694785.0, + "step": 19270 + }, + { + "epoch": 1.5890987053681866, + "grad_norm": 2.322072982788086, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.74833083152771, + "num_tokens": 469716257.0, + "step": 19271 + }, + { + "epoch": 1.5891811659932382, + "grad_norm": 1.8818660974502563, + "learning_rate": 1e-06, + "loss": 0.6177, + "mean_token_accuracy": 0.7985249757766724, + "num_tokens": 469744493.0, + "step": 19272 + }, + { + "epoch": 1.5892636266182898, + "grad_norm": 1.9135504961013794, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.746360182762146, + "num_tokens": 469776842.0, + "step": 19273 + }, + { + "epoch": 1.5893460872433414, + "grad_norm": 2.0373375415802, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7634162902832031, + "num_tokens": 469808732.0, + "step": 19274 + }, + { + "epoch": 1.5894285478683927, + "grad_norm": 2.194861650466919, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.7921761870384216, + "num_tokens": 469834458.0, + "step": 19275 + }, + { + "epoch": 1.5895110084934445, + "grad_norm": 1.8494377136230469, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.7844763994216919, + "num_tokens": 469866234.0, + "step": 19276 + }, + { + "epoch": 1.5895934691184959, + "grad_norm": 1.9199572801589966, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.7810630798339844, + "num_tokens": 469897351.0, + "step": 19277 + }, + { + "epoch": 1.5896759297435474, + "grad_norm": 1.991440773010254, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7918136119842529, + "num_tokens": 469924820.0, + "step": 19278 + }, + { + "epoch": 1.589758390368599, + "grad_norm": 1.6558573246002197, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7685881853103638, + "num_tokens": 469964067.0, + "step": 19279 + }, + { + "epoch": 1.5898408509936506, + "grad_norm": 2.055431604385376, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.751562237739563, + "num_tokens": 469991201.0, + "step": 19280 + }, + { + "epoch": 1.5899233116187022, + "grad_norm": 2.1660752296447754, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.7968078851699829, + "num_tokens": 470015716.0, + "step": 19281 + }, + { + "epoch": 1.5900057722437535, + "grad_norm": 1.85335111618042, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7494121789932251, + "num_tokens": 470049866.0, + "step": 19282 + }, + { + "epoch": 1.5900882328688053, + "grad_norm": 2.1121408939361572, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7704941034317017, + "num_tokens": 470076406.0, + "step": 19283 + }, + { + "epoch": 1.5901706934938566, + "grad_norm": 1.9003010988235474, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7539098262786865, + "num_tokens": 470110202.0, + "step": 19284 + }, + { + "epoch": 1.5902531541189082, + "grad_norm": 1.9430991411209106, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.7991559505462646, + "num_tokens": 470139672.0, + "step": 19285 + }, + { + "epoch": 1.5903356147439598, + "grad_norm": 2.2506096363067627, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.749261200428009, + "num_tokens": 470162483.0, + "step": 19286 + }, + { + "epoch": 1.5904180753690111, + "grad_norm": 1.890528917312622, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7699688076972961, + "num_tokens": 470191485.0, + "step": 19287 + }, + { + "epoch": 1.590500535994063, + "grad_norm": 1.9153845310211182, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7355692982673645, + "num_tokens": 470223213.0, + "step": 19288 + }, + { + "epoch": 1.5905829966191143, + "grad_norm": 1.8904610872268677, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.770098090171814, + "num_tokens": 470254380.0, + "step": 19289 + }, + { + "epoch": 1.5906654572441659, + "grad_norm": 2.0441880226135254, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7867675423622131, + "num_tokens": 470284093.0, + "step": 19290 + }, + { + "epoch": 1.5907479178692174, + "grad_norm": 2.628460168838501, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7441918253898621, + "num_tokens": 470304217.0, + "step": 19291 + }, + { + "epoch": 1.590830378494269, + "grad_norm": 2.1209473609924316, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.78346848487854, + "num_tokens": 470329234.0, + "step": 19292 + }, + { + "epoch": 1.5909128391193206, + "grad_norm": 1.963442325592041, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.7665871381759644, + "num_tokens": 470361833.0, + "step": 19293 + }, + { + "epoch": 1.590995299744372, + "grad_norm": 1.7552523612976074, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.8021243810653687, + "num_tokens": 470395170.0, + "step": 19294 + }, + { + "epoch": 1.5910777603694237, + "grad_norm": 2.234982490539551, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7865304350852966, + "num_tokens": 470419413.0, + "step": 19295 + }, + { + "epoch": 1.591160220994475, + "grad_norm": 2.0891456604003906, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.7601470947265625, + "num_tokens": 470445837.0, + "step": 19296 + }, + { + "epoch": 1.5912426816195266, + "grad_norm": 1.9961192607879639, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7575150728225708, + "num_tokens": 470475206.0, + "step": 19297 + }, + { + "epoch": 1.5913251422445782, + "grad_norm": 1.8961560726165771, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7456781268119812, + "num_tokens": 470508259.0, + "step": 19298 + }, + { + "epoch": 1.5914076028696298, + "grad_norm": 2.0295913219451904, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7640198469161987, + "num_tokens": 470538992.0, + "step": 19299 + }, + { + "epoch": 1.5914900634946814, + "grad_norm": 2.024716377258301, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.786703884601593, + "num_tokens": 470565007.0, + "step": 19300 + }, + { + "epoch": 1.5915725241197327, + "grad_norm": 1.8900765180587769, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7672548294067383, + "num_tokens": 470597552.0, + "step": 19301 + }, + { + "epoch": 1.5916549847447845, + "grad_norm": 1.845542311668396, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.7888221740722656, + "num_tokens": 470629944.0, + "step": 19302 + }, + { + "epoch": 1.5917374453698359, + "grad_norm": 2.179067373275757, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.747602105140686, + "num_tokens": 470654045.0, + "step": 19303 + }, + { + "epoch": 1.5918199059948874, + "grad_norm": 2.039355754852295, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7729268670082092, + "num_tokens": 470678102.0, + "step": 19304 + }, + { + "epoch": 1.591902366619939, + "grad_norm": 2.028548002243042, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7356079816818237, + "num_tokens": 470707446.0, + "step": 19305 + }, + { + "epoch": 1.5919848272449904, + "grad_norm": 1.9756278991699219, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7358208894729614, + "num_tokens": 470738050.0, + "step": 19306 + }, + { + "epoch": 1.5920672878700421, + "grad_norm": 2.1377294063568115, + "learning_rate": 1e-06, + "loss": 0.5651, + "mean_token_accuracy": 0.8065004348754883, + "num_tokens": 470762654.0, + "step": 19307 + }, + { + "epoch": 1.5921497484950935, + "grad_norm": 2.2795839309692383, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7600255608558655, + "num_tokens": 470786713.0, + "step": 19308 + }, + { + "epoch": 1.5922322091201453, + "grad_norm": 2.0477614402770996, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.7695987224578857, + "num_tokens": 470813932.0, + "step": 19309 + }, + { + "epoch": 1.5923146697451966, + "grad_norm": 1.9753268957138062, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7813911437988281, + "num_tokens": 470841042.0, + "step": 19310 + }, + { + "epoch": 1.5923971303702482, + "grad_norm": 1.9976270198822021, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7077015042304993, + "num_tokens": 470872951.0, + "step": 19311 + }, + { + "epoch": 1.5924795909952998, + "grad_norm": 2.0153467655181885, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7682276964187622, + "num_tokens": 470902472.0, + "step": 19312 + }, + { + "epoch": 1.5925620516203511, + "grad_norm": 1.947379231452942, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.7762311697006226, + "num_tokens": 470932039.0, + "step": 19313 + }, + { + "epoch": 1.592644512245403, + "grad_norm": 1.79537034034729, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.77492356300354, + "num_tokens": 470967862.0, + "step": 19314 + }, + { + "epoch": 1.5927269728704543, + "grad_norm": 1.7128589153289795, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.7971361875534058, + "num_tokens": 471005249.0, + "step": 19315 + }, + { + "epoch": 1.5928094334955059, + "grad_norm": 1.86465322971344, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7269394993782043, + "num_tokens": 471043005.0, + "step": 19316 + }, + { + "epoch": 1.5928918941205574, + "grad_norm": 2.3504416942596436, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7487075924873352, + "num_tokens": 471066503.0, + "step": 19317 + }, + { + "epoch": 1.592974354745609, + "grad_norm": 2.011373519897461, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.7871414422988892, + "num_tokens": 471094251.0, + "step": 19318 + }, + { + "epoch": 1.5930568153706606, + "grad_norm": 1.8732365369796753, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7529418468475342, + "num_tokens": 471128059.0, + "step": 19319 + }, + { + "epoch": 1.593139275995712, + "grad_norm": 1.7486127614974976, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7820102572441101, + "num_tokens": 471167312.0, + "step": 19320 + }, + { + "epoch": 1.5932217366207637, + "grad_norm": 1.9990798234939575, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7480456829071045, + "num_tokens": 471197962.0, + "step": 19321 + }, + { + "epoch": 1.593304197245815, + "grad_norm": 2.0762295722961426, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7467440962791443, + "num_tokens": 471227956.0, + "step": 19322 + }, + { + "epoch": 1.5933866578708666, + "grad_norm": 1.9144765138626099, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7711015343666077, + "num_tokens": 471261895.0, + "step": 19323 + }, + { + "epoch": 1.5934691184959182, + "grad_norm": 1.6999527215957642, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7934095859527588, + "num_tokens": 471305520.0, + "step": 19324 + }, + { + "epoch": 1.5935515791209698, + "grad_norm": 2.3557889461517334, + "learning_rate": 1e-06, + "loss": 0.7988, + "mean_token_accuracy": 0.7491624355316162, + "num_tokens": 471328820.0, + "step": 19325 + }, + { + "epoch": 1.5936340397460214, + "grad_norm": 1.8792179822921753, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7571115493774414, + "num_tokens": 471362572.0, + "step": 19326 + }, + { + "epoch": 1.5937165003710727, + "grad_norm": 1.7949014902114868, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7802174687385559, + "num_tokens": 471397233.0, + "step": 19327 + }, + { + "epoch": 1.5937989609961245, + "grad_norm": 1.9612913131713867, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7592651844024658, + "num_tokens": 471428169.0, + "step": 19328 + }, + { + "epoch": 1.5938814216211759, + "grad_norm": 2.119316577911377, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.747954249382019, + "num_tokens": 471454921.0, + "step": 19329 + }, + { + "epoch": 1.5939638822462274, + "grad_norm": 2.0601842403411865, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7621164917945862, + "num_tokens": 471481369.0, + "step": 19330 + }, + { + "epoch": 1.594046342871279, + "grad_norm": 1.8701239824295044, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.8007595539093018, + "num_tokens": 471510295.0, + "step": 19331 + }, + { + "epoch": 1.5941288034963303, + "grad_norm": 1.9042723178863525, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.7956407070159912, + "num_tokens": 471540651.0, + "step": 19332 + }, + { + "epoch": 1.5942112641213821, + "grad_norm": 1.7976874113082886, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7628711462020874, + "num_tokens": 471579962.0, + "step": 19333 + }, + { + "epoch": 1.5942937247464335, + "grad_norm": 2.0776848793029785, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7593073844909668, + "num_tokens": 471606821.0, + "step": 19334 + }, + { + "epoch": 1.594376185371485, + "grad_norm": 2.172161817550659, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7720015645027161, + "num_tokens": 471633241.0, + "step": 19335 + }, + { + "epoch": 1.5944586459965366, + "grad_norm": 1.9891319274902344, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.7877086400985718, + "num_tokens": 471660775.0, + "step": 19336 + }, + { + "epoch": 1.5945411066215882, + "grad_norm": 1.8272755146026611, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7516025304794312, + "num_tokens": 471696068.0, + "step": 19337 + }, + { + "epoch": 1.5946235672466398, + "grad_norm": 2.487058162689209, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7724505662918091, + "num_tokens": 471716934.0, + "step": 19338 + }, + { + "epoch": 1.5947060278716911, + "grad_norm": 1.8526328802108765, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7504273056983948, + "num_tokens": 471749465.0, + "step": 19339 + }, + { + "epoch": 1.594788488496743, + "grad_norm": 1.9084430932998657, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.800311803817749, + "num_tokens": 471778909.0, + "step": 19340 + }, + { + "epoch": 1.5948709491217943, + "grad_norm": 2.1681902408599854, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7675495743751526, + "num_tokens": 471803845.0, + "step": 19341 + }, + { + "epoch": 1.5949534097468459, + "grad_norm": 2.052927255630493, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7779607772827148, + "num_tokens": 471832315.0, + "step": 19342 + }, + { + "epoch": 1.5950358703718974, + "grad_norm": 2.09912371635437, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7397642731666565, + "num_tokens": 471861002.0, + "step": 19343 + }, + { + "epoch": 1.595118330996949, + "grad_norm": 2.02962327003479, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7580756545066833, + "num_tokens": 471890360.0, + "step": 19344 + }, + { + "epoch": 1.5952007916220006, + "grad_norm": 2.203033447265625, + "learning_rate": 1e-06, + "loss": 0.5589, + "mean_token_accuracy": 0.814106822013855, + "num_tokens": 471916345.0, + "step": 19345 + }, + { + "epoch": 1.595283252247052, + "grad_norm": 2.319976568222046, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7511292099952698, + "num_tokens": 471939737.0, + "step": 19346 + }, + { + "epoch": 1.5953657128721037, + "grad_norm": 1.9547268152236938, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7875380516052246, + "num_tokens": 471972597.0, + "step": 19347 + }, + { + "epoch": 1.595448173497155, + "grad_norm": 1.8877836465835571, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7806423902511597, + "num_tokens": 472003010.0, + "step": 19348 + }, + { + "epoch": 1.5955306341222066, + "grad_norm": 2.335268020629883, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7084808945655823, + "num_tokens": 472028369.0, + "step": 19349 + }, + { + "epoch": 1.5956130947472582, + "grad_norm": 2.097241163253784, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7181956768035889, + "num_tokens": 472058027.0, + "step": 19350 + }, + { + "epoch": 1.5956955553723096, + "grad_norm": 1.9810696840286255, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7660290002822876, + "num_tokens": 472087849.0, + "step": 19351 + }, + { + "epoch": 1.5957780159973614, + "grad_norm": 2.3275160789489746, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7132301330566406, + "num_tokens": 472112734.0, + "step": 19352 + }, + { + "epoch": 1.5958604766224127, + "grad_norm": 2.0732011795043945, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7691587209701538, + "num_tokens": 472139156.0, + "step": 19353 + }, + { + "epoch": 1.5959429372474645, + "grad_norm": 2.0663833618164062, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.7872229218482971, + "num_tokens": 472166063.0, + "step": 19354 + }, + { + "epoch": 1.5960253978725158, + "grad_norm": 2.266834020614624, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7629212737083435, + "num_tokens": 472192617.0, + "step": 19355 + }, + { + "epoch": 1.5961078584975674, + "grad_norm": 1.8262670040130615, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7829983234405518, + "num_tokens": 472229062.0, + "step": 19356 + }, + { + "epoch": 1.596190319122619, + "grad_norm": 2.2444305419921875, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7383859157562256, + "num_tokens": 472254828.0, + "step": 19357 + }, + { + "epoch": 1.5962727797476703, + "grad_norm": 2.022517204284668, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.7860751748085022, + "num_tokens": 472285100.0, + "step": 19358 + }, + { + "epoch": 1.5963552403727221, + "grad_norm": 1.8948919773101807, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7540323734283447, + "num_tokens": 472316832.0, + "step": 19359 + }, + { + "epoch": 1.5964377009977735, + "grad_norm": 1.878664255142212, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7538049221038818, + "num_tokens": 472350440.0, + "step": 19360 + }, + { + "epoch": 1.596520161622825, + "grad_norm": 2.3906784057617188, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7143917679786682, + "num_tokens": 472375189.0, + "step": 19361 + }, + { + "epoch": 1.5966026222478766, + "grad_norm": 2.090949535369873, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.7843924760818481, + "num_tokens": 472401348.0, + "step": 19362 + }, + { + "epoch": 1.5966850828729282, + "grad_norm": 1.7976025342941284, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7896661162376404, + "num_tokens": 472435034.0, + "step": 19363 + }, + { + "epoch": 1.5967675434979798, + "grad_norm": 2.163565158843994, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7799644470214844, + "num_tokens": 472458865.0, + "step": 19364 + }, + { + "epoch": 1.5968500041230311, + "grad_norm": 1.831766128540039, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7556062340736389, + "num_tokens": 472493389.0, + "step": 19365 + }, + { + "epoch": 1.596932464748083, + "grad_norm": 2.015838623046875, + "learning_rate": 1e-06, + "loss": 0.6031, + "mean_token_accuracy": 0.7965189218521118, + "num_tokens": 472523802.0, + "step": 19366 + }, + { + "epoch": 1.5970149253731343, + "grad_norm": 2.2014265060424805, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.746403157711029, + "num_tokens": 472550783.0, + "step": 19367 + }, + { + "epoch": 1.5970973859981858, + "grad_norm": 2.4362096786499023, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7765763998031616, + "num_tokens": 472572572.0, + "step": 19368 + }, + { + "epoch": 1.5971798466232374, + "grad_norm": 2.14043927192688, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7018413543701172, + "num_tokens": 472601114.0, + "step": 19369 + }, + { + "epoch": 1.597262307248289, + "grad_norm": 2.021167278289795, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7532401084899902, + "num_tokens": 472631865.0, + "step": 19370 + }, + { + "epoch": 1.5973447678733406, + "grad_norm": 2.1026153564453125, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7440005540847778, + "num_tokens": 472661091.0, + "step": 19371 + }, + { + "epoch": 1.597427228498392, + "grad_norm": 2.696885824203491, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.710482120513916, + "num_tokens": 472681178.0, + "step": 19372 + }, + { + "epoch": 1.5975096891234437, + "grad_norm": 1.9515217542648315, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7617352604866028, + "num_tokens": 472709965.0, + "step": 19373 + }, + { + "epoch": 1.597592149748495, + "grad_norm": 2.0029053688049316, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7449035048484802, + "num_tokens": 472738932.0, + "step": 19374 + }, + { + "epoch": 1.5976746103735466, + "grad_norm": 1.9943674802780151, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7297147512435913, + "num_tokens": 472771637.0, + "step": 19375 + }, + { + "epoch": 1.5977570709985982, + "grad_norm": 2.2390763759613037, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.741506040096283, + "num_tokens": 472797287.0, + "step": 19376 + }, + { + "epoch": 1.5978395316236496, + "grad_norm": 2.06135892868042, + "learning_rate": 1e-06, + "loss": 0.6147, + "mean_token_accuracy": 0.7900557518005371, + "num_tokens": 472826029.0, + "step": 19377 + }, + { + "epoch": 1.5979219922487014, + "grad_norm": 1.7609652280807495, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.8013204336166382, + "num_tokens": 472860167.0, + "step": 19378 + }, + { + "epoch": 1.5980044528737527, + "grad_norm": 1.939041256904602, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7209669947624207, + "num_tokens": 472892047.0, + "step": 19379 + }, + { + "epoch": 1.5980869134988043, + "grad_norm": 1.7487834692001343, + "learning_rate": 1e-06, + "loss": 0.5351, + "mean_token_accuracy": 0.8153411149978638, + "num_tokens": 472925127.0, + "step": 19380 + }, + { + "epoch": 1.5981693741238558, + "grad_norm": 2.0620524883270264, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7842826247215271, + "num_tokens": 472956300.0, + "step": 19381 + }, + { + "epoch": 1.5982518347489074, + "grad_norm": 2.057961940765381, + "learning_rate": 1e-06, + "loss": 0.6001, + "mean_token_accuracy": 0.798209547996521, + "num_tokens": 472980952.0, + "step": 19382 + }, + { + "epoch": 1.598334295373959, + "grad_norm": 1.7705466747283936, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7679968476295471, + "num_tokens": 473017243.0, + "step": 19383 + }, + { + "epoch": 1.5984167559990103, + "grad_norm": 1.6269469261169434, + "learning_rate": 1e-06, + "loss": 0.4841, + "mean_token_accuracy": 0.8334478735923767, + "num_tokens": 473052113.0, + "step": 19384 + }, + { + "epoch": 1.5984992166240621, + "grad_norm": 2.1758244037628174, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.7854788303375244, + "num_tokens": 473077933.0, + "step": 19385 + }, + { + "epoch": 1.5985816772491135, + "grad_norm": 2.045438766479492, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7451104521751404, + "num_tokens": 473106542.0, + "step": 19386 + }, + { + "epoch": 1.598664137874165, + "grad_norm": 1.827949047088623, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.7870609760284424, + "num_tokens": 473141677.0, + "step": 19387 + }, + { + "epoch": 1.5987465984992166, + "grad_norm": 2.2201972007751465, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7474393248558044, + "num_tokens": 473166810.0, + "step": 19388 + }, + { + "epoch": 1.5988290591242682, + "grad_norm": 2.055197238922119, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7725676894187927, + "num_tokens": 473195229.0, + "step": 19389 + }, + { + "epoch": 1.5989115197493198, + "grad_norm": 2.125314235687256, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.7856038212776184, + "num_tokens": 473222027.0, + "step": 19390 + }, + { + "epoch": 1.5989939803743711, + "grad_norm": 2.143773317337036, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7867792844772339, + "num_tokens": 473248818.0, + "step": 19391 + }, + { + "epoch": 1.599076440999423, + "grad_norm": 2.130655288696289, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7378461360931396, + "num_tokens": 473277769.0, + "step": 19392 + }, + { + "epoch": 1.5991589016244743, + "grad_norm": 1.9141225814819336, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.7455064058303833, + "num_tokens": 473312006.0, + "step": 19393 + }, + { + "epoch": 1.5992413622495258, + "grad_norm": 1.847234845161438, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7479796409606934, + "num_tokens": 473344576.0, + "step": 19394 + }, + { + "epoch": 1.5993238228745774, + "grad_norm": 1.8768869638442993, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7812797427177429, + "num_tokens": 473376862.0, + "step": 19395 + }, + { + "epoch": 1.5994062834996288, + "grad_norm": 2.1281728744506836, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.752081036567688, + "num_tokens": 473404511.0, + "step": 19396 + }, + { + "epoch": 1.5994887441246806, + "grad_norm": 2.0080249309539795, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7422181963920593, + "num_tokens": 473432182.0, + "step": 19397 + }, + { + "epoch": 1.599571204749732, + "grad_norm": 2.094041585922241, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7464001178741455, + "num_tokens": 473461978.0, + "step": 19398 + }, + { + "epoch": 1.5996536653747837, + "grad_norm": 1.8676643371582031, + "learning_rate": 1e-06, + "loss": 0.5326, + "mean_token_accuracy": 0.8092008829116821, + "num_tokens": 473490721.0, + "step": 19399 + }, + { + "epoch": 1.599736125999835, + "grad_norm": 2.1172902584075928, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7510582804679871, + "num_tokens": 473519148.0, + "step": 19400 + }, + { + "epoch": 1.5998185866248866, + "grad_norm": 2.073399305343628, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7325187921524048, + "num_tokens": 473546435.0, + "step": 19401 + }, + { + "epoch": 1.5999010472499382, + "grad_norm": 2.126208782196045, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7593568563461304, + "num_tokens": 473571277.0, + "step": 19402 + }, + { + "epoch": 1.5999835078749896, + "grad_norm": 2.086977005004883, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7769514322280884, + "num_tokens": 473597843.0, + "step": 19403 + }, + { + "epoch": 1.6000659685000413, + "grad_norm": 1.901240587234497, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.7802571058273315, + "num_tokens": 473629216.0, + "step": 19404 + }, + { + "epoch": 1.6001484291250927, + "grad_norm": 2.3521041870117188, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7433109283447266, + "num_tokens": 473653155.0, + "step": 19405 + }, + { + "epoch": 1.6002308897501443, + "grad_norm": 2.2293078899383545, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7637975215911865, + "num_tokens": 473677634.0, + "step": 19406 + }, + { + "epoch": 1.6003133503751958, + "grad_norm": 2.1948604583740234, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7241137623786926, + "num_tokens": 473705827.0, + "step": 19407 + }, + { + "epoch": 1.6003958110002474, + "grad_norm": 1.9330863952636719, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7764055728912354, + "num_tokens": 473734594.0, + "step": 19408 + }, + { + "epoch": 1.600478271625299, + "grad_norm": 2.156332015991211, + "learning_rate": 1e-06, + "loss": 0.6066, + "mean_token_accuracy": 0.8013920187950134, + "num_tokens": 473760250.0, + "step": 19409 + }, + { + "epoch": 1.6005607322503503, + "grad_norm": 2.0644805431365967, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.768179714679718, + "num_tokens": 473789285.0, + "step": 19410 + }, + { + "epoch": 1.6006431928754021, + "grad_norm": 2.064770221710205, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7596762776374817, + "num_tokens": 473818081.0, + "step": 19411 + }, + { + "epoch": 1.6007256535004535, + "grad_norm": 2.1246790885925293, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7626554369926453, + "num_tokens": 473843832.0, + "step": 19412 + }, + { + "epoch": 1.600808114125505, + "grad_norm": 1.8864794969558716, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7480684518814087, + "num_tokens": 473878402.0, + "step": 19413 + }, + { + "epoch": 1.6008905747505566, + "grad_norm": 1.7462410926818848, + "learning_rate": 1e-06, + "loss": 0.4936, + "mean_token_accuracy": 0.8265316486358643, + "num_tokens": 473911177.0, + "step": 19414 + }, + { + "epoch": 1.6009730353756082, + "grad_norm": 2.0162506103515625, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7664194107055664, + "num_tokens": 473940801.0, + "step": 19415 + }, + { + "epoch": 1.6010554960006598, + "grad_norm": 2.064518451690674, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7819928526878357, + "num_tokens": 473967917.0, + "step": 19416 + }, + { + "epoch": 1.6011379566257111, + "grad_norm": 1.9300236701965332, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7652171850204468, + "num_tokens": 474000666.0, + "step": 19417 + }, + { + "epoch": 1.601220417250763, + "grad_norm": 1.902021884918213, + "learning_rate": 1e-06, + "loss": 0.5819, + "mean_token_accuracy": 0.8106366395950317, + "num_tokens": 474032112.0, + "step": 19418 + }, + { + "epoch": 1.6013028778758143, + "grad_norm": 1.6093356609344482, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.8041760921478271, + "num_tokens": 474070535.0, + "step": 19419 + }, + { + "epoch": 1.6013853385008658, + "grad_norm": 1.9429924488067627, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.7874656915664673, + "num_tokens": 474100221.0, + "step": 19420 + }, + { + "epoch": 1.6014677991259174, + "grad_norm": 1.896747350692749, + "learning_rate": 1e-06, + "loss": 0.5978, + "mean_token_accuracy": 0.806869626045227, + "num_tokens": 474131440.0, + "step": 19421 + }, + { + "epoch": 1.6015502597509688, + "grad_norm": 1.982561469078064, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7289063930511475, + "num_tokens": 474162948.0, + "step": 19422 + }, + { + "epoch": 1.6016327203760206, + "grad_norm": 2.1898956298828125, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7738138437271118, + "num_tokens": 474187143.0, + "step": 19423 + }, + { + "epoch": 1.601715181001072, + "grad_norm": 1.828059434890747, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7747637033462524, + "num_tokens": 474223146.0, + "step": 19424 + }, + { + "epoch": 1.6017976416261235, + "grad_norm": 1.778760552406311, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.777132511138916, + "num_tokens": 474255687.0, + "step": 19425 + }, + { + "epoch": 1.601880102251175, + "grad_norm": 2.177560806274414, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7510316371917725, + "num_tokens": 474282662.0, + "step": 19426 + }, + { + "epoch": 1.6019625628762266, + "grad_norm": 2.0596084594726562, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7399046421051025, + "num_tokens": 474310699.0, + "step": 19427 + }, + { + "epoch": 1.6020450235012782, + "grad_norm": 2.002127170562744, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7292430400848389, + "num_tokens": 474341257.0, + "step": 19428 + }, + { + "epoch": 1.6021274841263295, + "grad_norm": 1.9951204061508179, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7716806530952454, + "num_tokens": 474369361.0, + "step": 19429 + }, + { + "epoch": 1.6022099447513813, + "grad_norm": 1.8632105588912964, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.7908329963684082, + "num_tokens": 474399207.0, + "step": 19430 + }, + { + "epoch": 1.6022924053764327, + "grad_norm": 1.9974907636642456, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7751190662384033, + "num_tokens": 474426247.0, + "step": 19431 + }, + { + "epoch": 1.6023748660014843, + "grad_norm": 1.7582553625106812, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7739899158477783, + "num_tokens": 474459559.0, + "step": 19432 + }, + { + "epoch": 1.6024573266265358, + "grad_norm": 1.8912431001663208, + "learning_rate": 1e-06, + "loss": 0.5916, + "mean_token_accuracy": 0.8076414465904236, + "num_tokens": 474491491.0, + "step": 19433 + }, + { + "epoch": 1.6025397872515874, + "grad_norm": 2.1160759925842285, + "learning_rate": 1e-06, + "loss": 0.7619, + "mean_token_accuracy": 0.7554111480712891, + "num_tokens": 474517823.0, + "step": 19434 + }, + { + "epoch": 1.602622247876639, + "grad_norm": 2.074446439743042, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7753864526748657, + "num_tokens": 474546863.0, + "step": 19435 + }, + { + "epoch": 1.6027047085016903, + "grad_norm": 1.798542857170105, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.7946513891220093, + "num_tokens": 474580047.0, + "step": 19436 + }, + { + "epoch": 1.6027871691267421, + "grad_norm": 1.9987350702285767, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7299004793167114, + "num_tokens": 474608889.0, + "step": 19437 + }, + { + "epoch": 1.6028696297517935, + "grad_norm": 2.247445821762085, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7668759822845459, + "num_tokens": 474635073.0, + "step": 19438 + }, + { + "epoch": 1.602952090376845, + "grad_norm": 1.8379625082015991, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7820426225662231, + "num_tokens": 474668025.0, + "step": 19439 + }, + { + "epoch": 1.6030345510018966, + "grad_norm": 1.965775966644287, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.761271595954895, + "num_tokens": 474697698.0, + "step": 19440 + }, + { + "epoch": 1.603117011626948, + "grad_norm": 1.7269418239593506, + "learning_rate": 1e-06, + "loss": 0.6129, + "mean_token_accuracy": 0.7965413928031921, + "num_tokens": 474732966.0, + "step": 19441 + }, + { + "epoch": 1.6031994722519998, + "grad_norm": 2.177217960357666, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7704384326934814, + "num_tokens": 474759660.0, + "step": 19442 + }, + { + "epoch": 1.6032819328770511, + "grad_norm": 2.0409975051879883, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.7978373169898987, + "num_tokens": 474786902.0, + "step": 19443 + }, + { + "epoch": 1.6033643935021027, + "grad_norm": 1.806989073753357, + "learning_rate": 1e-06, + "loss": 0.6041, + "mean_token_accuracy": 0.8012084364891052, + "num_tokens": 474819116.0, + "step": 19444 + }, + { + "epoch": 1.6034468541271543, + "grad_norm": 2.040400505065918, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7535600662231445, + "num_tokens": 474847322.0, + "step": 19445 + }, + { + "epoch": 1.6035293147522058, + "grad_norm": 1.9189329147338867, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7684880495071411, + "num_tokens": 474880116.0, + "step": 19446 + }, + { + "epoch": 1.6036117753772574, + "grad_norm": 1.8326835632324219, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7729519009590149, + "num_tokens": 474913948.0, + "step": 19447 + }, + { + "epoch": 1.6036942360023088, + "grad_norm": 1.8924046754837036, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7716548442840576, + "num_tokens": 474944133.0, + "step": 19448 + }, + { + "epoch": 1.6037766966273606, + "grad_norm": 2.044055461883545, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7781755924224854, + "num_tokens": 474972098.0, + "step": 19449 + }, + { + "epoch": 1.603859157252412, + "grad_norm": 2.068030834197998, + "learning_rate": 1e-06, + "loss": 0.6074, + "mean_token_accuracy": 0.7848944664001465, + "num_tokens": 474998761.0, + "step": 19450 + }, + { + "epoch": 1.6039416178774635, + "grad_norm": 1.9316864013671875, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7284612655639648, + "num_tokens": 475031249.0, + "step": 19451 + }, + { + "epoch": 1.604024078502515, + "grad_norm": 1.9880430698394775, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7656424045562744, + "num_tokens": 475059283.0, + "step": 19452 + }, + { + "epoch": 1.6041065391275666, + "grad_norm": 2.0269558429718018, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7566134333610535, + "num_tokens": 475086873.0, + "step": 19453 + }, + { + "epoch": 1.6041889997526182, + "grad_norm": 2.0489072799682617, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6966511607170105, + "num_tokens": 475117083.0, + "step": 19454 + }, + { + "epoch": 1.6042714603776695, + "grad_norm": 2.032458543777466, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.7844523787498474, + "num_tokens": 475144374.0, + "step": 19455 + }, + { + "epoch": 1.6043539210027213, + "grad_norm": 2.1083760261535645, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7729333639144897, + "num_tokens": 475169901.0, + "step": 19456 + }, + { + "epoch": 1.6044363816277727, + "grad_norm": 2.032038450241089, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7751863598823547, + "num_tokens": 475196593.0, + "step": 19457 + }, + { + "epoch": 1.6045188422528243, + "grad_norm": 1.8823421001434326, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7805710434913635, + "num_tokens": 475228627.0, + "step": 19458 + }, + { + "epoch": 1.6046013028778758, + "grad_norm": 1.843395471572876, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.7843603491783142, + "num_tokens": 475262304.0, + "step": 19459 + }, + { + "epoch": 1.6046837635029274, + "grad_norm": 1.9192534685134888, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7680376172065735, + "num_tokens": 475293478.0, + "step": 19460 + }, + { + "epoch": 1.604766224127979, + "grad_norm": 1.8148231506347656, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7594214677810669, + "num_tokens": 475327068.0, + "step": 19461 + }, + { + "epoch": 1.6048486847530303, + "grad_norm": 2.297264814376831, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7386646270751953, + "num_tokens": 475349201.0, + "step": 19462 + }, + { + "epoch": 1.6049311453780821, + "grad_norm": 1.7571929693222046, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7793242335319519, + "num_tokens": 475382997.0, + "step": 19463 + }, + { + "epoch": 1.6050136060031335, + "grad_norm": 1.7893413305282593, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7944867014884949, + "num_tokens": 475419110.0, + "step": 19464 + }, + { + "epoch": 1.605096066628185, + "grad_norm": 1.9463461637496948, + "learning_rate": 1e-06, + "loss": 0.5937, + "mean_token_accuracy": 0.8025764226913452, + "num_tokens": 475450209.0, + "step": 19465 + }, + { + "epoch": 1.6051785272532366, + "grad_norm": 2.0219345092773438, + "learning_rate": 1e-06, + "loss": 0.7768, + "mean_token_accuracy": 0.7593858242034912, + "num_tokens": 475483101.0, + "step": 19466 + }, + { + "epoch": 1.605260987878288, + "grad_norm": 2.1569697856903076, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7811464071273804, + "num_tokens": 475510110.0, + "step": 19467 + }, + { + "epoch": 1.6053434485033398, + "grad_norm": 1.9424500465393066, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7337914705276489, + "num_tokens": 475542616.0, + "step": 19468 + }, + { + "epoch": 1.6054259091283911, + "grad_norm": 2.017552614212036, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7493096590042114, + "num_tokens": 475579468.0, + "step": 19469 + }, + { + "epoch": 1.6055083697534427, + "grad_norm": 2.0860018730163574, + "learning_rate": 1e-06, + "loss": 0.5539, + "mean_token_accuracy": 0.8144525289535522, + "num_tokens": 475604092.0, + "step": 19470 + }, + { + "epoch": 1.6055908303784943, + "grad_norm": 1.9648557901382446, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7771444320678711, + "num_tokens": 475634356.0, + "step": 19471 + }, + { + "epoch": 1.6056732910035458, + "grad_norm": 1.72615385055542, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.7975724935531616, + "num_tokens": 475670621.0, + "step": 19472 + }, + { + "epoch": 1.6057557516285974, + "grad_norm": 2.0985753536224365, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7729044556617737, + "num_tokens": 475701372.0, + "step": 19473 + }, + { + "epoch": 1.6058382122536488, + "grad_norm": 2.1172163486480713, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7274870872497559, + "num_tokens": 475729876.0, + "step": 19474 + }, + { + "epoch": 1.6059206728787006, + "grad_norm": 1.9342318773269653, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7386890649795532, + "num_tokens": 475762993.0, + "step": 19475 + }, + { + "epoch": 1.606003133503752, + "grad_norm": 2.139099359512329, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7669457793235779, + "num_tokens": 475788693.0, + "step": 19476 + }, + { + "epoch": 1.6060855941288035, + "grad_norm": 1.8055164813995361, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7499189972877502, + "num_tokens": 475825243.0, + "step": 19477 + }, + { + "epoch": 1.606168054753855, + "grad_norm": 1.7528001070022583, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.7872215509414673, + "num_tokens": 475857322.0, + "step": 19478 + }, + { + "epoch": 1.6062505153789066, + "grad_norm": 2.0257017612457275, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7446087002754211, + "num_tokens": 475891612.0, + "step": 19479 + }, + { + "epoch": 1.6063329760039582, + "grad_norm": 2.0286786556243896, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.760187029838562, + "num_tokens": 475922026.0, + "step": 19480 + }, + { + "epoch": 1.6064154366290095, + "grad_norm": 1.976879358291626, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.758738100528717, + "num_tokens": 475954053.0, + "step": 19481 + }, + { + "epoch": 1.6064978972540613, + "grad_norm": 1.8214936256408691, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7756446599960327, + "num_tokens": 475989486.0, + "step": 19482 + }, + { + "epoch": 1.6065803578791127, + "grad_norm": 2.27313494682312, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.773788571357727, + "num_tokens": 476015874.0, + "step": 19483 + }, + { + "epoch": 1.6066628185041643, + "grad_norm": 2.126528739929199, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.796099066734314, + "num_tokens": 476044225.0, + "step": 19484 + }, + { + "epoch": 1.6067452791292158, + "grad_norm": 1.894068956375122, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7541585564613342, + "num_tokens": 476077807.0, + "step": 19485 + }, + { + "epoch": 1.6068277397542672, + "grad_norm": 1.8701430559158325, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7508412599563599, + "num_tokens": 476115002.0, + "step": 19486 + }, + { + "epoch": 1.606910200379319, + "grad_norm": 2.1338651180267334, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7473511695861816, + "num_tokens": 476144360.0, + "step": 19487 + }, + { + "epoch": 1.6069926610043703, + "grad_norm": 1.9592305421829224, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7744387984275818, + "num_tokens": 476174599.0, + "step": 19488 + }, + { + "epoch": 1.607075121629422, + "grad_norm": 1.959911584854126, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7631747722625732, + "num_tokens": 476206743.0, + "step": 19489 + }, + { + "epoch": 1.6071575822544735, + "grad_norm": 2.071528434753418, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.714834988117218, + "num_tokens": 476236943.0, + "step": 19490 + }, + { + "epoch": 1.607240042879525, + "grad_norm": 1.9613213539123535, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7415357828140259, + "num_tokens": 476270463.0, + "step": 19491 + }, + { + "epoch": 1.6073225035045766, + "grad_norm": 2.15222430229187, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.747050404548645, + "num_tokens": 476297150.0, + "step": 19492 + }, + { + "epoch": 1.607404964129628, + "grad_norm": 2.3550174236297607, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.7965266108512878, + "num_tokens": 476320814.0, + "step": 19493 + }, + { + "epoch": 1.6074874247546798, + "grad_norm": 1.9328221082687378, + "learning_rate": 1e-06, + "loss": 0.5831, + "mean_token_accuracy": 0.8059875965118408, + "num_tokens": 476349982.0, + "step": 19494 + }, + { + "epoch": 1.6075698853797311, + "grad_norm": 2.0007896423339844, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7717996835708618, + "num_tokens": 476379482.0, + "step": 19495 + }, + { + "epoch": 1.6076523460047827, + "grad_norm": 1.8898048400878906, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7494736909866333, + "num_tokens": 476411568.0, + "step": 19496 + }, + { + "epoch": 1.6077348066298343, + "grad_norm": 1.771241545677185, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7681561708450317, + "num_tokens": 476445762.0, + "step": 19497 + }, + { + "epoch": 1.6078172672548858, + "grad_norm": 1.795336365699768, + "learning_rate": 1e-06, + "loss": 0.5118, + "mean_token_accuracy": 0.8291548490524292, + "num_tokens": 476477911.0, + "step": 19498 + }, + { + "epoch": 1.6078997278799374, + "grad_norm": 2.0247445106506348, + "learning_rate": 1e-06, + "loss": 0.5528, + "mean_token_accuracy": 0.809613823890686, + "num_tokens": 476506553.0, + "step": 19499 + }, + { + "epoch": 1.6079821885049888, + "grad_norm": 2.2784719467163086, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.742690920829773, + "num_tokens": 476531523.0, + "step": 19500 + }, + { + "epoch": 1.6080646491300405, + "grad_norm": 2.1930763721466064, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7571766376495361, + "num_tokens": 476557316.0, + "step": 19501 + }, + { + "epoch": 1.608147109755092, + "grad_norm": 1.8215861320495605, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7546176910400391, + "num_tokens": 476592463.0, + "step": 19502 + }, + { + "epoch": 1.6082295703801435, + "grad_norm": 1.8875702619552612, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7727895975112915, + "num_tokens": 476623245.0, + "step": 19503 + }, + { + "epoch": 1.608312031005195, + "grad_norm": 1.9130187034606934, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7526607513427734, + "num_tokens": 476656357.0, + "step": 19504 + }, + { + "epoch": 1.6083944916302464, + "grad_norm": 1.962611198425293, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7780176997184753, + "num_tokens": 476686865.0, + "step": 19505 + }, + { + "epoch": 1.6084769522552982, + "grad_norm": 1.7714680433273315, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.7875606417655945, + "num_tokens": 476725396.0, + "step": 19506 + }, + { + "epoch": 1.6085594128803495, + "grad_norm": 1.7456072568893433, + "learning_rate": 1e-06, + "loss": 0.5988, + "mean_token_accuracy": 0.8113370537757874, + "num_tokens": 476758679.0, + "step": 19507 + }, + { + "epoch": 1.6086418735054013, + "grad_norm": 1.9612879753112793, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.7859361171722412, + "num_tokens": 476789743.0, + "step": 19508 + }, + { + "epoch": 1.6087243341304527, + "grad_norm": 2.160400390625, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7669415473937988, + "num_tokens": 476816931.0, + "step": 19509 + }, + { + "epoch": 1.6088067947555043, + "grad_norm": 1.8849536180496216, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.7971541881561279, + "num_tokens": 476849319.0, + "step": 19510 + }, + { + "epoch": 1.6088892553805558, + "grad_norm": 1.9094164371490479, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.7886275053024292, + "num_tokens": 476879391.0, + "step": 19511 + }, + { + "epoch": 1.6089717160056072, + "grad_norm": 1.8202623128890991, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.8019632697105408, + "num_tokens": 476914444.0, + "step": 19512 + }, + { + "epoch": 1.609054176630659, + "grad_norm": 2.247046947479248, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.770108163356781, + "num_tokens": 476940166.0, + "step": 19513 + }, + { + "epoch": 1.6091366372557103, + "grad_norm": 1.955777883529663, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7387765645980835, + "num_tokens": 476974286.0, + "step": 19514 + }, + { + "epoch": 1.609219097880762, + "grad_norm": 1.6301653385162354, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.8054698705673218, + "num_tokens": 477012539.0, + "step": 19515 + }, + { + "epoch": 1.6093015585058135, + "grad_norm": 1.8479928970336914, + "learning_rate": 1e-06, + "loss": 0.5866, + "mean_token_accuracy": 0.7975900173187256, + "num_tokens": 477043812.0, + "step": 19516 + }, + { + "epoch": 1.609384019130865, + "grad_norm": 2.075305461883545, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7678991556167603, + "num_tokens": 477073551.0, + "step": 19517 + }, + { + "epoch": 1.6094664797559166, + "grad_norm": 1.844803810119629, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7504437565803528, + "num_tokens": 477108010.0, + "step": 19518 + }, + { + "epoch": 1.609548940380968, + "grad_norm": 1.9225242137908936, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7809321284294128, + "num_tokens": 477139656.0, + "step": 19519 + }, + { + "epoch": 1.6096314010060198, + "grad_norm": 2.1684508323669434, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7511945962905884, + "num_tokens": 477167216.0, + "step": 19520 + }, + { + "epoch": 1.609713861631071, + "grad_norm": 1.9315916299819946, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7952538728713989, + "num_tokens": 477196319.0, + "step": 19521 + }, + { + "epoch": 1.6097963222561227, + "grad_norm": 1.7650136947631836, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7675219178199768, + "num_tokens": 477232341.0, + "step": 19522 + }, + { + "epoch": 1.6098787828811743, + "grad_norm": 2.1323695182800293, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7558725476264954, + "num_tokens": 477258621.0, + "step": 19523 + }, + { + "epoch": 1.6099612435062258, + "grad_norm": 2.1388959884643555, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7387173175811768, + "num_tokens": 477285261.0, + "step": 19524 + }, + { + "epoch": 1.6100437041312774, + "grad_norm": 1.7563257217407227, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.7609924674034119, + "num_tokens": 477319693.0, + "step": 19525 + }, + { + "epoch": 1.6101261647563287, + "grad_norm": 2.203782558441162, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7488877773284912, + "num_tokens": 477345149.0, + "step": 19526 + }, + { + "epoch": 1.6102086253813805, + "grad_norm": 2.1946094036102295, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7464174032211304, + "num_tokens": 477370557.0, + "step": 19527 + }, + { + "epoch": 1.610291086006432, + "grad_norm": 1.8975000381469727, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.7727813124656677, + "num_tokens": 477402168.0, + "step": 19528 + }, + { + "epoch": 1.6103735466314835, + "grad_norm": 2.031756639480591, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.777690052986145, + "num_tokens": 477429453.0, + "step": 19529 + }, + { + "epoch": 1.610456007256535, + "grad_norm": 1.9000418186187744, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7751460075378418, + "num_tokens": 477459279.0, + "step": 19530 + }, + { + "epoch": 1.6105384678815864, + "grad_norm": 1.8706560134887695, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7644729614257812, + "num_tokens": 477493825.0, + "step": 19531 + }, + { + "epoch": 1.6106209285066382, + "grad_norm": 2.0274622440338135, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7752440571784973, + "num_tokens": 477523518.0, + "step": 19532 + }, + { + "epoch": 1.6107033891316895, + "grad_norm": 1.8996762037277222, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.7937289476394653, + "num_tokens": 477554825.0, + "step": 19533 + }, + { + "epoch": 1.610785849756741, + "grad_norm": 1.9473600387573242, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7269505262374878, + "num_tokens": 477586699.0, + "step": 19534 + }, + { + "epoch": 1.6108683103817927, + "grad_norm": 2.0264618396759033, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.7992565631866455, + "num_tokens": 477613235.0, + "step": 19535 + }, + { + "epoch": 1.6109507710068443, + "grad_norm": 2.082714319229126, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7241474390029907, + "num_tokens": 477642348.0, + "step": 19536 + }, + { + "epoch": 1.6110332316318958, + "grad_norm": 2.0235273838043213, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7330026030540466, + "num_tokens": 477673377.0, + "step": 19537 + }, + { + "epoch": 1.6111156922569472, + "grad_norm": 2.0447494983673096, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7626179456710815, + "num_tokens": 477703521.0, + "step": 19538 + }, + { + "epoch": 1.611198152881999, + "grad_norm": 1.7755025625228882, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7705586552619934, + "num_tokens": 477739577.0, + "step": 19539 + }, + { + "epoch": 1.6112806135070503, + "grad_norm": 1.9692015647888184, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7305238842964172, + "num_tokens": 477768328.0, + "step": 19540 + }, + { + "epoch": 1.611363074132102, + "grad_norm": 2.0079169273376465, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7320844531059265, + "num_tokens": 477799454.0, + "step": 19541 + }, + { + "epoch": 1.6114455347571535, + "grad_norm": 2.217588186264038, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.766735315322876, + "num_tokens": 477822829.0, + "step": 19542 + }, + { + "epoch": 1.611527995382205, + "grad_norm": 2.032130241394043, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7651289105415344, + "num_tokens": 477851247.0, + "step": 19543 + }, + { + "epoch": 1.6116104560072566, + "grad_norm": 1.8481552600860596, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7412528395652771, + "num_tokens": 477891696.0, + "step": 19544 + }, + { + "epoch": 1.611692916632308, + "grad_norm": 1.9438509941101074, + "learning_rate": 1e-06, + "loss": 0.7877, + "mean_token_accuracy": 0.7442061901092529, + "num_tokens": 477922224.0, + "step": 19545 + }, + { + "epoch": 1.6117753772573598, + "grad_norm": 2.0613372325897217, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7797276377677917, + "num_tokens": 477948727.0, + "step": 19546 + }, + { + "epoch": 1.611857837882411, + "grad_norm": 2.15854549407959, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.721251368522644, + "num_tokens": 477974972.0, + "step": 19547 + }, + { + "epoch": 1.6119402985074627, + "grad_norm": 2.2836382389068604, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7502106428146362, + "num_tokens": 478000960.0, + "step": 19548 + }, + { + "epoch": 1.6120227591325142, + "grad_norm": 1.896546483039856, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7657397985458374, + "num_tokens": 478031938.0, + "step": 19549 + }, + { + "epoch": 1.6121052197575656, + "grad_norm": 1.8712198734283447, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7410478591918945, + "num_tokens": 478066342.0, + "step": 19550 + }, + { + "epoch": 1.6121876803826174, + "grad_norm": 2.1825850009918213, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.7827081680297852, + "num_tokens": 478093316.0, + "step": 19551 + }, + { + "epoch": 1.6122701410076687, + "grad_norm": 2.3640170097351074, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7612159252166748, + "num_tokens": 478115951.0, + "step": 19552 + }, + { + "epoch": 1.6123526016327205, + "grad_norm": 1.9295650720596313, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.750309944152832, + "num_tokens": 478151110.0, + "step": 19553 + }, + { + "epoch": 1.612435062257772, + "grad_norm": 1.9022536277770996, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7700710296630859, + "num_tokens": 478185948.0, + "step": 19554 + }, + { + "epoch": 1.6125175228828235, + "grad_norm": 1.946612000465393, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7639870643615723, + "num_tokens": 478216096.0, + "step": 19555 + }, + { + "epoch": 1.612599983507875, + "grad_norm": 2.0324864387512207, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7485215067863464, + "num_tokens": 478244987.0, + "step": 19556 + }, + { + "epoch": 1.6126824441329264, + "grad_norm": 1.9366943836212158, + "learning_rate": 1e-06, + "loss": 0.606, + "mean_token_accuracy": 0.7966523766517639, + "num_tokens": 478276624.0, + "step": 19557 + }, + { + "epoch": 1.6127649047579782, + "grad_norm": 1.7736058235168457, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7643255591392517, + "num_tokens": 478311823.0, + "step": 19558 + }, + { + "epoch": 1.6128473653830295, + "grad_norm": 2.240699529647827, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.772083580493927, + "num_tokens": 478341916.0, + "step": 19559 + }, + { + "epoch": 1.612929826008081, + "grad_norm": 1.903669834136963, + "learning_rate": 1e-06, + "loss": 0.5935, + "mean_token_accuracy": 0.7977137565612793, + "num_tokens": 478370553.0, + "step": 19560 + }, + { + "epoch": 1.6130122866331327, + "grad_norm": 1.8490113019943237, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7491305470466614, + "num_tokens": 478405304.0, + "step": 19561 + }, + { + "epoch": 1.6130947472581842, + "grad_norm": 2.062241315841675, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7539197206497192, + "num_tokens": 478435123.0, + "step": 19562 + }, + { + "epoch": 1.6131772078832358, + "grad_norm": 1.8633215427398682, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7556227445602417, + "num_tokens": 478472697.0, + "step": 19563 + }, + { + "epoch": 1.6132596685082872, + "grad_norm": 2.102689266204834, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7381834387779236, + "num_tokens": 478498484.0, + "step": 19564 + }, + { + "epoch": 1.613342129133339, + "grad_norm": 1.8716561794281006, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.7877079844474792, + "num_tokens": 478529027.0, + "step": 19565 + }, + { + "epoch": 1.6134245897583903, + "grad_norm": 1.9430694580078125, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7620826363563538, + "num_tokens": 478560416.0, + "step": 19566 + }, + { + "epoch": 1.6135070503834419, + "grad_norm": 1.9336507320404053, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.7876417636871338, + "num_tokens": 478592209.0, + "step": 19567 + }, + { + "epoch": 1.6135895110084935, + "grad_norm": 2.316016912460327, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7545348405838013, + "num_tokens": 478615122.0, + "step": 19568 + }, + { + "epoch": 1.613671971633545, + "grad_norm": 2.5075855255126953, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7413005232810974, + "num_tokens": 478636115.0, + "step": 19569 + }, + { + "epoch": 1.6137544322585966, + "grad_norm": 1.9070390462875366, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7846529483795166, + "num_tokens": 478666473.0, + "step": 19570 + }, + { + "epoch": 1.613836892883648, + "grad_norm": 1.978776454925537, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7510815858840942, + "num_tokens": 478696502.0, + "step": 19571 + }, + { + "epoch": 1.6139193535086997, + "grad_norm": 1.8430721759796143, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7561824321746826, + "num_tokens": 478730972.0, + "step": 19572 + }, + { + "epoch": 1.614001814133751, + "grad_norm": 2.088754177093506, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7689120769500732, + "num_tokens": 478760514.0, + "step": 19573 + }, + { + "epoch": 1.6140842747588027, + "grad_norm": 1.993622064590454, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7285767197608948, + "num_tokens": 478791873.0, + "step": 19574 + }, + { + "epoch": 1.6141667353838542, + "grad_norm": 2.0631277561187744, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7591498494148254, + "num_tokens": 478819020.0, + "step": 19575 + }, + { + "epoch": 1.6142491960089056, + "grad_norm": 1.876073956489563, + "learning_rate": 1e-06, + "loss": 0.5726, + "mean_token_accuracy": 0.8035812377929688, + "num_tokens": 478850112.0, + "step": 19576 + }, + { + "epoch": 1.6143316566339574, + "grad_norm": 1.8758156299591064, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7538274526596069, + "num_tokens": 478884703.0, + "step": 19577 + }, + { + "epoch": 1.6144141172590087, + "grad_norm": 1.9247230291366577, + "learning_rate": 1e-06, + "loss": 0.558, + "mean_token_accuracy": 0.8109319806098938, + "num_tokens": 478914826.0, + "step": 19578 + }, + { + "epoch": 1.6144965778840603, + "grad_norm": 2.220954418182373, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7729809284210205, + "num_tokens": 478939912.0, + "step": 19579 + }, + { + "epoch": 1.6145790385091119, + "grad_norm": 2.0756754875183105, + "learning_rate": 1e-06, + "loss": 0.5886, + "mean_token_accuracy": 0.793458104133606, + "num_tokens": 478964980.0, + "step": 19580 + }, + { + "epoch": 1.6146614991341635, + "grad_norm": 1.886188268661499, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7632193565368652, + "num_tokens": 478998037.0, + "step": 19581 + }, + { + "epoch": 1.614743959759215, + "grad_norm": 1.9369803667068481, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.764270007610321, + "num_tokens": 479027963.0, + "step": 19582 + }, + { + "epoch": 1.6148264203842664, + "grad_norm": 2.122608184814453, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7754117250442505, + "num_tokens": 479053876.0, + "step": 19583 + }, + { + "epoch": 1.6149088810093182, + "grad_norm": 1.943955898284912, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7737606167793274, + "num_tokens": 479083982.0, + "step": 19584 + }, + { + "epoch": 1.6149913416343695, + "grad_norm": 1.9060999155044556, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7264658808708191, + "num_tokens": 479116091.0, + "step": 19585 + }, + { + "epoch": 1.615073802259421, + "grad_norm": 1.9673588275909424, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7725949287414551, + "num_tokens": 479145741.0, + "step": 19586 + }, + { + "epoch": 1.6151562628844727, + "grad_norm": 2.038452625274658, + "learning_rate": 1e-06, + "loss": 0.618, + "mean_token_accuracy": 0.7810356616973877, + "num_tokens": 479173532.0, + "step": 19587 + }, + { + "epoch": 1.6152387235095242, + "grad_norm": 2.008790969848633, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7446987628936768, + "num_tokens": 479204128.0, + "step": 19588 + }, + { + "epoch": 1.6153211841345758, + "grad_norm": 1.9266133308410645, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7667240500450134, + "num_tokens": 479232889.0, + "step": 19589 + }, + { + "epoch": 1.6154036447596272, + "grad_norm": 1.9102058410644531, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7678080201148987, + "num_tokens": 479262025.0, + "step": 19590 + }, + { + "epoch": 1.615486105384679, + "grad_norm": 1.7004311084747314, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.7894867062568665, + "num_tokens": 479297912.0, + "step": 19591 + }, + { + "epoch": 1.6155685660097303, + "grad_norm": 1.87669038772583, + "learning_rate": 1e-06, + "loss": 0.5828, + "mean_token_accuracy": 0.8066667914390564, + "num_tokens": 479327551.0, + "step": 19592 + }, + { + "epoch": 1.6156510266347819, + "grad_norm": 1.855536699295044, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7471973896026611, + "num_tokens": 479362872.0, + "step": 19593 + }, + { + "epoch": 1.6157334872598335, + "grad_norm": 1.9207746982574463, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7791456580162048, + "num_tokens": 479394237.0, + "step": 19594 + }, + { + "epoch": 1.6158159478848848, + "grad_norm": 2.403593063354492, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7469215393066406, + "num_tokens": 479418147.0, + "step": 19595 + }, + { + "epoch": 1.6158984085099366, + "grad_norm": 1.970353126525879, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7636784315109253, + "num_tokens": 479446812.0, + "step": 19596 + }, + { + "epoch": 1.615980869134988, + "grad_norm": 2.0700418949127197, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7705260515213013, + "num_tokens": 479472491.0, + "step": 19597 + }, + { + "epoch": 1.6160633297600397, + "grad_norm": 2.39785099029541, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7449647188186646, + "num_tokens": 479497312.0, + "step": 19598 + }, + { + "epoch": 1.616145790385091, + "grad_norm": 1.9271728992462158, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7586007714271545, + "num_tokens": 479528843.0, + "step": 19599 + }, + { + "epoch": 1.6162282510101427, + "grad_norm": 2.006319999694824, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7532693147659302, + "num_tokens": 479558242.0, + "step": 19600 + }, + { + "epoch": 1.6163107116351942, + "grad_norm": 2.045712947845459, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7784534692764282, + "num_tokens": 479584770.0, + "step": 19601 + }, + { + "epoch": 1.6163931722602456, + "grad_norm": 1.7830009460449219, + "learning_rate": 1e-06, + "loss": 0.5093, + "mean_token_accuracy": 0.8228785395622253, + "num_tokens": 479615389.0, + "step": 19602 + }, + { + "epoch": 1.6164756328852974, + "grad_norm": 2.0406508445739746, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.767630934715271, + "num_tokens": 479646562.0, + "step": 19603 + }, + { + "epoch": 1.6165580935103487, + "grad_norm": 2.0496065616607666, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.7863069772720337, + "num_tokens": 479674026.0, + "step": 19604 + }, + { + "epoch": 1.6166405541354003, + "grad_norm": 1.8540040254592896, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.7856259346008301, + "num_tokens": 479705299.0, + "step": 19605 + }, + { + "epoch": 1.6167230147604519, + "grad_norm": 2.061868190765381, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.7849687933921814, + "num_tokens": 479733047.0, + "step": 19606 + }, + { + "epoch": 1.6168054753855035, + "grad_norm": 2.0104541778564453, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.7935997247695923, + "num_tokens": 479758491.0, + "step": 19607 + }, + { + "epoch": 1.616887936010555, + "grad_norm": 1.8776509761810303, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7645915746688843, + "num_tokens": 479791541.0, + "step": 19608 + }, + { + "epoch": 1.6169703966356064, + "grad_norm": 2.0136237144470215, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7499622106552124, + "num_tokens": 479820276.0, + "step": 19609 + }, + { + "epoch": 1.6170528572606582, + "grad_norm": 2.192732095718384, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7551140785217285, + "num_tokens": 479846465.0, + "step": 19610 + }, + { + "epoch": 1.6171353178857095, + "grad_norm": 2.0185933113098145, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7678290605545044, + "num_tokens": 479877643.0, + "step": 19611 + }, + { + "epoch": 1.617217778510761, + "grad_norm": 2.4241766929626465, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7301985025405884, + "num_tokens": 479901045.0, + "step": 19612 + }, + { + "epoch": 1.6173002391358127, + "grad_norm": 2.238075017929077, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7412697076797485, + "num_tokens": 479927849.0, + "step": 19613 + }, + { + "epoch": 1.6173826997608642, + "grad_norm": 2.0525360107421875, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7753685116767883, + "num_tokens": 479956888.0, + "step": 19614 + }, + { + "epoch": 1.6174651603859158, + "grad_norm": 1.9770245552062988, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7455110549926758, + "num_tokens": 479988324.0, + "step": 19615 + }, + { + "epoch": 1.6175476210109672, + "grad_norm": 2.17193603515625, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7372288703918457, + "num_tokens": 480016469.0, + "step": 19616 + }, + { + "epoch": 1.617630081636019, + "grad_norm": 2.4037024974823, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.7526969313621521, + "num_tokens": 480038422.0, + "step": 19617 + }, + { + "epoch": 1.6177125422610703, + "grad_norm": 2.3559205532073975, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7581630945205688, + "num_tokens": 480063201.0, + "step": 19618 + }, + { + "epoch": 1.6177950028861219, + "grad_norm": 2.1195712089538574, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7510817050933838, + "num_tokens": 480091848.0, + "step": 19619 + }, + { + "epoch": 1.6178774635111735, + "grad_norm": 2.1984846591949463, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7484117746353149, + "num_tokens": 480118269.0, + "step": 19620 + }, + { + "epoch": 1.6179599241362248, + "grad_norm": 1.990342140197754, + "learning_rate": 1e-06, + "loss": 0.5557, + "mean_token_accuracy": 0.8088780045509338, + "num_tokens": 480146129.0, + "step": 19621 + }, + { + "epoch": 1.6180423847612766, + "grad_norm": 1.902010202407837, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7646836042404175, + "num_tokens": 480180403.0, + "step": 19622 + }, + { + "epoch": 1.618124845386328, + "grad_norm": 1.9616780281066895, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7680923938751221, + "num_tokens": 480208849.0, + "step": 19623 + }, + { + "epoch": 1.6182073060113795, + "grad_norm": 1.86276376247406, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.771645188331604, + "num_tokens": 480240558.0, + "step": 19624 + }, + { + "epoch": 1.618289766636431, + "grad_norm": 2.0279829502105713, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7366580367088318, + "num_tokens": 480268685.0, + "step": 19625 + }, + { + "epoch": 1.6183722272614827, + "grad_norm": 2.1424496173858643, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.732903003692627, + "num_tokens": 480296735.0, + "step": 19626 + }, + { + "epoch": 1.6184546878865342, + "grad_norm": 2.356217384338379, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7005923390388489, + "num_tokens": 480321367.0, + "step": 19627 + }, + { + "epoch": 1.6185371485115856, + "grad_norm": 1.6993026733398438, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7825384140014648, + "num_tokens": 480360397.0, + "step": 19628 + }, + { + "epoch": 1.6186196091366374, + "grad_norm": 1.9392720460891724, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7780727744102478, + "num_tokens": 480389746.0, + "step": 19629 + }, + { + "epoch": 1.6187020697616887, + "grad_norm": 2.1352219581604004, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7694241404533386, + "num_tokens": 480413933.0, + "step": 19630 + }, + { + "epoch": 1.6187845303867403, + "grad_norm": 1.8576438426971436, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7483159303665161, + "num_tokens": 480445559.0, + "step": 19631 + }, + { + "epoch": 1.6188669910117919, + "grad_norm": 1.9182723760604858, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7622888088226318, + "num_tokens": 480477094.0, + "step": 19632 + }, + { + "epoch": 1.6189494516368435, + "grad_norm": 2.1149864196777344, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7659135460853577, + "num_tokens": 480505474.0, + "step": 19633 + }, + { + "epoch": 1.619031912261895, + "grad_norm": 1.8910887241363525, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.7793406844139099, + "num_tokens": 480536680.0, + "step": 19634 + }, + { + "epoch": 1.6191143728869464, + "grad_norm": 1.830690622329712, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7855679988861084, + "num_tokens": 480569137.0, + "step": 19635 + }, + { + "epoch": 1.6191968335119982, + "grad_norm": 2.295760154724121, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7441695928573608, + "num_tokens": 480591962.0, + "step": 19636 + }, + { + "epoch": 1.6192792941370495, + "grad_norm": 1.8834718465805054, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7818326950073242, + "num_tokens": 480622778.0, + "step": 19637 + }, + { + "epoch": 1.619361754762101, + "grad_norm": 2.2232577800750732, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7331563234329224, + "num_tokens": 480648747.0, + "step": 19638 + }, + { + "epoch": 1.6194442153871527, + "grad_norm": 1.717369794845581, + "learning_rate": 1e-06, + "loss": 0.5009, + "mean_token_accuracy": 0.8208703398704529, + "num_tokens": 480681663.0, + "step": 19639 + }, + { + "epoch": 1.619526676012204, + "grad_norm": 1.9506362676620483, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.751636266708374, + "num_tokens": 480713277.0, + "step": 19640 + }, + { + "epoch": 1.6196091366372558, + "grad_norm": 2.314683437347412, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7571297883987427, + "num_tokens": 480738065.0, + "step": 19641 + }, + { + "epoch": 1.6196915972623072, + "grad_norm": 1.9346946477890015, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.784837007522583, + "num_tokens": 480764367.0, + "step": 19642 + }, + { + "epoch": 1.6197740578873587, + "grad_norm": 1.9257279634475708, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7600926756858826, + "num_tokens": 480796207.0, + "step": 19643 + }, + { + "epoch": 1.6198565185124103, + "grad_norm": 2.211052656173706, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7755978107452393, + "num_tokens": 480822465.0, + "step": 19644 + }, + { + "epoch": 1.6199389791374619, + "grad_norm": 1.9836342334747314, + "learning_rate": 1e-06, + "loss": 0.7486, + "mean_token_accuracy": 0.7627371549606323, + "num_tokens": 480852485.0, + "step": 19645 + }, + { + "epoch": 1.6200214397625134, + "grad_norm": 1.830418586730957, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7849122285842896, + "num_tokens": 480886893.0, + "step": 19646 + }, + { + "epoch": 1.6201039003875648, + "grad_norm": 1.9418962001800537, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7545754313468933, + "num_tokens": 480918596.0, + "step": 19647 + }, + { + "epoch": 1.6201863610126166, + "grad_norm": 1.7645740509033203, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.7438793778419495, + "num_tokens": 480955427.0, + "step": 19648 + }, + { + "epoch": 1.620268821637668, + "grad_norm": 2.059175968170166, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7811582088470459, + "num_tokens": 480982933.0, + "step": 19649 + }, + { + "epoch": 1.6203512822627195, + "grad_norm": 2.2119479179382324, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.7501262426376343, + "num_tokens": 481007818.0, + "step": 19650 + }, + { + "epoch": 1.620433742887771, + "grad_norm": 1.8555681705474854, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7642308473587036, + "num_tokens": 481045215.0, + "step": 19651 + }, + { + "epoch": 1.6205162035128227, + "grad_norm": 2.04378342628479, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7570531368255615, + "num_tokens": 481073183.0, + "step": 19652 + }, + { + "epoch": 1.6205986641378742, + "grad_norm": 1.8971843719482422, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7707840800285339, + "num_tokens": 481105124.0, + "step": 19653 + }, + { + "epoch": 1.6206811247629256, + "grad_norm": 2.170146942138672, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7373050451278687, + "num_tokens": 481131714.0, + "step": 19654 + }, + { + "epoch": 1.6207635853879774, + "grad_norm": 1.8827942609786987, + "learning_rate": 1e-06, + "loss": 0.4943, + "mean_token_accuracy": 0.828193187713623, + "num_tokens": 481163605.0, + "step": 19655 + }, + { + "epoch": 1.6208460460130287, + "grad_norm": 1.826582908630371, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7798017263412476, + "num_tokens": 481200384.0, + "step": 19656 + }, + { + "epoch": 1.6209285066380803, + "grad_norm": 2.2277579307556152, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7460339069366455, + "num_tokens": 481225665.0, + "step": 19657 + }, + { + "epoch": 1.6210109672631319, + "grad_norm": 2.1607766151428223, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7639756798744202, + "num_tokens": 481253020.0, + "step": 19658 + }, + { + "epoch": 1.6210934278881834, + "grad_norm": 1.9730730056762695, + "learning_rate": 1e-06, + "loss": 0.579, + "mean_token_accuracy": 0.801524817943573, + "num_tokens": 481280732.0, + "step": 19659 + }, + { + "epoch": 1.621175888513235, + "grad_norm": 1.9963936805725098, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.787128210067749, + "num_tokens": 481309535.0, + "step": 19660 + }, + { + "epoch": 1.6212583491382864, + "grad_norm": 1.6793928146362305, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.7567151188850403, + "num_tokens": 481349702.0, + "step": 19661 + }, + { + "epoch": 1.6213408097633382, + "grad_norm": 1.995191216468811, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.739741861820221, + "num_tokens": 481378490.0, + "step": 19662 + }, + { + "epoch": 1.6214232703883895, + "grad_norm": 1.7452938556671143, + "learning_rate": 1e-06, + "loss": 0.5707, + "mean_token_accuracy": 0.8157908916473389, + "num_tokens": 481412531.0, + "step": 19663 + }, + { + "epoch": 1.621505731013441, + "grad_norm": 1.933989405632019, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7629776000976562, + "num_tokens": 481443631.0, + "step": 19664 + }, + { + "epoch": 1.6215881916384927, + "grad_norm": 1.696955680847168, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7521750926971436, + "num_tokens": 481486831.0, + "step": 19665 + }, + { + "epoch": 1.621670652263544, + "grad_norm": 1.6869800090789795, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7738425731658936, + "num_tokens": 481526274.0, + "step": 19666 + }, + { + "epoch": 1.6217531128885958, + "grad_norm": 1.7070232629776, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.768489420413971, + "num_tokens": 481563884.0, + "step": 19667 + }, + { + "epoch": 1.6218355735136472, + "grad_norm": 1.769612193107605, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7800166606903076, + "num_tokens": 481600507.0, + "step": 19668 + }, + { + "epoch": 1.6219180341386987, + "grad_norm": 2.127446413040161, + "learning_rate": 1e-06, + "loss": 0.6047, + "mean_token_accuracy": 0.7987862825393677, + "num_tokens": 481626172.0, + "step": 19669 + }, + { + "epoch": 1.6220004947637503, + "grad_norm": 2.0283076763153076, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7598865032196045, + "num_tokens": 481655688.0, + "step": 19670 + }, + { + "epoch": 1.6220829553888019, + "grad_norm": 1.779466152191162, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7874070405960083, + "num_tokens": 481689919.0, + "step": 19671 + }, + { + "epoch": 1.6221654160138534, + "grad_norm": 1.9129940271377563, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7679826617240906, + "num_tokens": 481720611.0, + "step": 19672 + }, + { + "epoch": 1.6222478766389048, + "grad_norm": 2.2586498260498047, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7490088939666748, + "num_tokens": 481747940.0, + "step": 19673 + }, + { + "epoch": 1.6223303372639566, + "grad_norm": 1.7865790128707886, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.7959991693496704, + "num_tokens": 481782169.0, + "step": 19674 + }, + { + "epoch": 1.622412797889008, + "grad_norm": 2.1728737354278564, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7588964104652405, + "num_tokens": 481805810.0, + "step": 19675 + }, + { + "epoch": 1.6224952585140595, + "grad_norm": 1.8644165992736816, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7388927340507507, + "num_tokens": 481835367.0, + "step": 19676 + }, + { + "epoch": 1.622577719139111, + "grad_norm": 1.8824357986450195, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7819620370864868, + "num_tokens": 481864600.0, + "step": 19677 + }, + { + "epoch": 1.6226601797641627, + "grad_norm": 1.9268388748168945, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7348137497901917, + "num_tokens": 481893016.0, + "step": 19678 + }, + { + "epoch": 1.6227426403892142, + "grad_norm": 2.0278677940368652, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7520644664764404, + "num_tokens": 481921064.0, + "step": 19679 + }, + { + "epoch": 1.6228251010142656, + "grad_norm": 1.7773607969284058, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7723400592803955, + "num_tokens": 481958065.0, + "step": 19680 + }, + { + "epoch": 1.6229075616393174, + "grad_norm": 1.8986397981643677, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7695109248161316, + "num_tokens": 481992776.0, + "step": 19681 + }, + { + "epoch": 1.6229900222643687, + "grad_norm": 1.9552987813949585, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7525634765625, + "num_tokens": 482022283.0, + "step": 19682 + }, + { + "epoch": 1.6230724828894203, + "grad_norm": 2.1655497550964355, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7242914438247681, + "num_tokens": 482049923.0, + "step": 19683 + }, + { + "epoch": 1.6231549435144719, + "grad_norm": 1.9746155738830566, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.752914547920227, + "num_tokens": 482080510.0, + "step": 19684 + }, + { + "epoch": 1.6232374041395232, + "grad_norm": 1.7499579191207886, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7745718955993652, + "num_tokens": 482118351.0, + "step": 19685 + }, + { + "epoch": 1.623319864764575, + "grad_norm": 2.434809684753418, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7719684839248657, + "num_tokens": 482143642.0, + "step": 19686 + }, + { + "epoch": 1.6234023253896264, + "grad_norm": 1.9459538459777832, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7732219099998474, + "num_tokens": 482175731.0, + "step": 19687 + }, + { + "epoch": 1.623484786014678, + "grad_norm": 1.934409499168396, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.7946028113365173, + "num_tokens": 482205141.0, + "step": 19688 + }, + { + "epoch": 1.6235672466397295, + "grad_norm": 2.000264883041382, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7514107823371887, + "num_tokens": 482233477.0, + "step": 19689 + }, + { + "epoch": 1.623649707264781, + "grad_norm": 2.119819164276123, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7723817825317383, + "num_tokens": 482261282.0, + "step": 19690 + }, + { + "epoch": 1.6237321678898327, + "grad_norm": 1.975158452987671, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7648905515670776, + "num_tokens": 482293189.0, + "step": 19691 + }, + { + "epoch": 1.623814628514884, + "grad_norm": 2.048274278640747, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7452364563941956, + "num_tokens": 482321825.0, + "step": 19692 + }, + { + "epoch": 1.6238970891399358, + "grad_norm": 1.8883056640625, + "learning_rate": 1e-06, + "loss": 0.5848, + "mean_token_accuracy": 0.8072243928909302, + "num_tokens": 482352562.0, + "step": 19693 + }, + { + "epoch": 1.6239795497649872, + "grad_norm": 1.6893686056137085, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7752823233604431, + "num_tokens": 482393655.0, + "step": 19694 + }, + { + "epoch": 1.6240620103900387, + "grad_norm": 2.3599023818969727, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.716762900352478, + "num_tokens": 482418079.0, + "step": 19695 + }, + { + "epoch": 1.6241444710150903, + "grad_norm": 2.0766844749450684, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7792371511459351, + "num_tokens": 482443009.0, + "step": 19696 + }, + { + "epoch": 1.6242269316401419, + "grad_norm": 2.3039865493774414, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7639466524124146, + "num_tokens": 482465772.0, + "step": 19697 + }, + { + "epoch": 1.6243093922651934, + "grad_norm": 1.8713140487670898, + "learning_rate": 1e-06, + "loss": 0.5898, + "mean_token_accuracy": 0.7942473888397217, + "num_tokens": 482496516.0, + "step": 19698 + }, + { + "epoch": 1.6243918528902448, + "grad_norm": 2.2040669918060303, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.753491222858429, + "num_tokens": 482523006.0, + "step": 19699 + }, + { + "epoch": 1.6244743135152966, + "grad_norm": 1.9813615083694458, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7227940559387207, + "num_tokens": 482556264.0, + "step": 19700 + }, + { + "epoch": 1.624556774140348, + "grad_norm": 1.901649832725525, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7614713907241821, + "num_tokens": 482588680.0, + "step": 19701 + }, + { + "epoch": 1.6246392347653995, + "grad_norm": 2.0292210578918457, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7489619851112366, + "num_tokens": 482619479.0, + "step": 19702 + }, + { + "epoch": 1.624721695390451, + "grad_norm": 2.1068484783172607, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7415607571601868, + "num_tokens": 482646466.0, + "step": 19703 + }, + { + "epoch": 1.6248041560155024, + "grad_norm": 2.18389630317688, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7199190855026245, + "num_tokens": 482674705.0, + "step": 19704 + }, + { + "epoch": 1.6248866166405542, + "grad_norm": 2.0739874839782715, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7325628995895386, + "num_tokens": 482703601.0, + "step": 19705 + }, + { + "epoch": 1.6249690772656056, + "grad_norm": 1.999711513519287, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7846012711524963, + "num_tokens": 482731726.0, + "step": 19706 + }, + { + "epoch": 1.6250515378906574, + "grad_norm": 1.9969861507415771, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.7804689407348633, + "num_tokens": 482759604.0, + "step": 19707 + }, + { + "epoch": 1.6251339985157087, + "grad_norm": 2.480715274810791, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7393254637718201, + "num_tokens": 482783100.0, + "step": 19708 + }, + { + "epoch": 1.6252164591407603, + "grad_norm": 1.7875711917877197, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7401456832885742, + "num_tokens": 482819864.0, + "step": 19709 + }, + { + "epoch": 1.6252989197658119, + "grad_norm": 2.038414239883423, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.773924708366394, + "num_tokens": 482847498.0, + "step": 19710 + }, + { + "epoch": 1.6253813803908632, + "grad_norm": 1.9435395002365112, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.770682692527771, + "num_tokens": 482877808.0, + "step": 19711 + }, + { + "epoch": 1.625463841015915, + "grad_norm": 2.04522967338562, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7602689266204834, + "num_tokens": 482906340.0, + "step": 19712 + }, + { + "epoch": 1.6255463016409664, + "grad_norm": 2.059353828430176, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.77326500415802, + "num_tokens": 482936776.0, + "step": 19713 + }, + { + "epoch": 1.625628762266018, + "grad_norm": 2.041088581085205, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.758985698223114, + "num_tokens": 482966396.0, + "step": 19714 + }, + { + "epoch": 1.6257112228910695, + "grad_norm": 1.8266427516937256, + "learning_rate": 1e-06, + "loss": 0.5511, + "mean_token_accuracy": 0.8053708076477051, + "num_tokens": 482999797.0, + "step": 19715 + }, + { + "epoch": 1.625793683516121, + "grad_norm": 2.138930082321167, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7868822813034058, + "num_tokens": 483026074.0, + "step": 19716 + }, + { + "epoch": 1.6258761441411727, + "grad_norm": 2.097709894180298, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.744070827960968, + "num_tokens": 483056274.0, + "step": 19717 + }, + { + "epoch": 1.625958604766224, + "grad_norm": 2.0704641342163086, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7515391111373901, + "num_tokens": 483087717.0, + "step": 19718 + }, + { + "epoch": 1.6260410653912758, + "grad_norm": 1.6435693502426147, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.792705774307251, + "num_tokens": 483128331.0, + "step": 19719 + }, + { + "epoch": 1.6261235260163271, + "grad_norm": 2.0974254608154297, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7626514434814453, + "num_tokens": 483157036.0, + "step": 19720 + }, + { + "epoch": 1.6262059866413787, + "grad_norm": 1.8896491527557373, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7596219778060913, + "num_tokens": 483188809.0, + "step": 19721 + }, + { + "epoch": 1.6262884472664303, + "grad_norm": 1.9531948566436768, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7496941089630127, + "num_tokens": 483221087.0, + "step": 19722 + }, + { + "epoch": 1.6263709078914819, + "grad_norm": 1.9439562559127808, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.7841875553131104, + "num_tokens": 483250192.0, + "step": 19723 + }, + { + "epoch": 1.6264533685165334, + "grad_norm": 1.964009165763855, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.762950599193573, + "num_tokens": 483279499.0, + "step": 19724 + }, + { + "epoch": 1.6265358291415848, + "grad_norm": 1.9495161771774292, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7412298917770386, + "num_tokens": 483313189.0, + "step": 19725 + }, + { + "epoch": 1.6266182897666366, + "grad_norm": 1.7576850652694702, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7744219303131104, + "num_tokens": 483348545.0, + "step": 19726 + }, + { + "epoch": 1.626700750391688, + "grad_norm": 1.940126895904541, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7333810329437256, + "num_tokens": 483381613.0, + "step": 19727 + }, + { + "epoch": 1.6267832110167395, + "grad_norm": 2.0161168575286865, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7424390316009521, + "num_tokens": 483412168.0, + "step": 19728 + }, + { + "epoch": 1.626865671641791, + "grad_norm": 2.0437848567962646, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7608644962310791, + "num_tokens": 483439173.0, + "step": 19729 + }, + { + "epoch": 1.6269481322668424, + "grad_norm": 2.1405510902404785, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7677079439163208, + "num_tokens": 483464927.0, + "step": 19730 + }, + { + "epoch": 1.6270305928918942, + "grad_norm": 2.157971143722534, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.764199435710907, + "num_tokens": 483490347.0, + "step": 19731 + }, + { + "epoch": 1.6271130535169456, + "grad_norm": 1.91659414768219, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7834690809249878, + "num_tokens": 483521996.0, + "step": 19732 + }, + { + "epoch": 1.6271955141419971, + "grad_norm": 2.0101401805877686, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7601537108421326, + "num_tokens": 483552319.0, + "step": 19733 + }, + { + "epoch": 1.6272779747670487, + "grad_norm": 1.8667467832565308, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.7894627451896667, + "num_tokens": 483584995.0, + "step": 19734 + }, + { + "epoch": 1.6273604353921003, + "grad_norm": 2.265526056289673, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7363603115081787, + "num_tokens": 483608557.0, + "step": 19735 + }, + { + "epoch": 1.6274428960171519, + "grad_norm": 1.8655774593353271, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.7943446636199951, + "num_tokens": 483642306.0, + "step": 19736 + }, + { + "epoch": 1.6275253566422032, + "grad_norm": 2.0206103324890137, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.795304536819458, + "num_tokens": 483668893.0, + "step": 19737 + }, + { + "epoch": 1.627607817267255, + "grad_norm": 1.8540444374084473, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7849796414375305, + "num_tokens": 483698840.0, + "step": 19738 + }, + { + "epoch": 1.6276902778923064, + "grad_norm": 2.1071152687072754, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7439709901809692, + "num_tokens": 483726396.0, + "step": 19739 + }, + { + "epoch": 1.627772738517358, + "grad_norm": 1.821175217628479, + "learning_rate": 1e-06, + "loss": 0.7642, + "mean_token_accuracy": 0.761386513710022, + "num_tokens": 483761018.0, + "step": 19740 + }, + { + "epoch": 1.6278551991424095, + "grad_norm": 2.0903210639953613, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7327890396118164, + "num_tokens": 483792874.0, + "step": 19741 + }, + { + "epoch": 1.627937659767461, + "grad_norm": 1.855956792831421, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.7857776880264282, + "num_tokens": 483830479.0, + "step": 19742 + }, + { + "epoch": 1.6280201203925126, + "grad_norm": 1.9171260595321655, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7708789110183716, + "num_tokens": 483860938.0, + "step": 19743 + }, + { + "epoch": 1.628102581017564, + "grad_norm": 1.6842408180236816, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.7923709154129028, + "num_tokens": 483898432.0, + "step": 19744 + }, + { + "epoch": 1.6281850416426158, + "grad_norm": 1.9098728895187378, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7338451147079468, + "num_tokens": 483929187.0, + "step": 19745 + }, + { + "epoch": 1.6282675022676671, + "grad_norm": 2.1253836154937744, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.7973512411117554, + "num_tokens": 483953508.0, + "step": 19746 + }, + { + "epoch": 1.6283499628927187, + "grad_norm": 1.8415387868881226, + "learning_rate": 1e-06, + "loss": 0.6, + "mean_token_accuracy": 0.8038020133972168, + "num_tokens": 483982015.0, + "step": 19747 + }, + { + "epoch": 1.6284324235177703, + "grad_norm": 1.853190302848816, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7817494869232178, + "num_tokens": 484013153.0, + "step": 19748 + }, + { + "epoch": 1.6285148841428216, + "grad_norm": 1.759891152381897, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7642514109611511, + "num_tokens": 484052102.0, + "step": 19749 + }, + { + "epoch": 1.6285973447678734, + "grad_norm": 1.989689826965332, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.73324054479599, + "num_tokens": 484082473.0, + "step": 19750 + }, + { + "epoch": 1.6286798053929248, + "grad_norm": 2.174429416656494, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7542839050292969, + "num_tokens": 484109862.0, + "step": 19751 + }, + { + "epoch": 1.6287622660179766, + "grad_norm": 2.1511764526367188, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7619667053222656, + "num_tokens": 484136862.0, + "step": 19752 + }, + { + "epoch": 1.628844726643028, + "grad_norm": 1.9366414546966553, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7526302337646484, + "num_tokens": 484168060.0, + "step": 19753 + }, + { + "epoch": 1.6289271872680795, + "grad_norm": 2.4279732704162598, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6904354691505432, + "num_tokens": 484191117.0, + "step": 19754 + }, + { + "epoch": 1.629009647893131, + "grad_norm": 1.9421746730804443, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7503374814987183, + "num_tokens": 484221290.0, + "step": 19755 + }, + { + "epoch": 1.6290921085181824, + "grad_norm": 2.3408093452453613, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7259145975112915, + "num_tokens": 484245252.0, + "step": 19756 + }, + { + "epoch": 1.6291745691432342, + "grad_norm": 1.962746024131775, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7710380554199219, + "num_tokens": 484274514.0, + "step": 19757 + }, + { + "epoch": 1.6292570297682856, + "grad_norm": 1.9050618410110474, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7784223556518555, + "num_tokens": 484305284.0, + "step": 19758 + }, + { + "epoch": 1.6293394903933371, + "grad_norm": 1.8990492820739746, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7708501219749451, + "num_tokens": 484337502.0, + "step": 19759 + }, + { + "epoch": 1.6294219510183887, + "grad_norm": 2.2617950439453125, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7696695327758789, + "num_tokens": 484358590.0, + "step": 19760 + }, + { + "epoch": 1.6295044116434403, + "grad_norm": 1.8731801509857178, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7781611084938049, + "num_tokens": 484389205.0, + "step": 19761 + }, + { + "epoch": 1.6295868722684919, + "grad_norm": 2.2655653953552246, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7317878007888794, + "num_tokens": 484416387.0, + "step": 19762 + }, + { + "epoch": 1.6296693328935432, + "grad_norm": 1.9607150554656982, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.762489914894104, + "num_tokens": 484446633.0, + "step": 19763 + }, + { + "epoch": 1.629751793518595, + "grad_norm": 2.008364200592041, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.7854254245758057, + "num_tokens": 484477819.0, + "step": 19764 + }, + { + "epoch": 1.6298342541436464, + "grad_norm": 2.0239086151123047, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7352030873298645, + "num_tokens": 484509763.0, + "step": 19765 + }, + { + "epoch": 1.629916714768698, + "grad_norm": 1.8419828414916992, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7165995836257935, + "num_tokens": 484547587.0, + "step": 19766 + }, + { + "epoch": 1.6299991753937495, + "grad_norm": 2.055894374847412, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7469422817230225, + "num_tokens": 484578838.0, + "step": 19767 + }, + { + "epoch": 1.630081636018801, + "grad_norm": 2.4413833618164062, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7515795230865479, + "num_tokens": 484599543.0, + "step": 19768 + }, + { + "epoch": 1.6301640966438526, + "grad_norm": 2.207444190979004, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7571859359741211, + "num_tokens": 484627369.0, + "step": 19769 + }, + { + "epoch": 1.630246557268904, + "grad_norm": 1.6926805973052979, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7789779901504517, + "num_tokens": 484666243.0, + "step": 19770 + }, + { + "epoch": 1.6303290178939558, + "grad_norm": 2.0161643028259277, + "learning_rate": 1e-06, + "loss": 0.7452, + "mean_token_accuracy": 0.7695907950401306, + "num_tokens": 484695000.0, + "step": 19771 + }, + { + "epoch": 1.6304114785190071, + "grad_norm": 1.9141786098480225, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7451019287109375, + "num_tokens": 484728059.0, + "step": 19772 + }, + { + "epoch": 1.6304939391440587, + "grad_norm": 1.8231712579727173, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7379346489906311, + "num_tokens": 484762536.0, + "step": 19773 + }, + { + "epoch": 1.6305763997691103, + "grad_norm": 1.8708752393722534, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.7814228534698486, + "num_tokens": 484793982.0, + "step": 19774 + }, + { + "epoch": 1.6306588603941616, + "grad_norm": 1.7782628536224365, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7737637162208557, + "num_tokens": 484831296.0, + "step": 19775 + }, + { + "epoch": 1.6307413210192134, + "grad_norm": 2.3112547397613525, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7480793595314026, + "num_tokens": 484855207.0, + "step": 19776 + }, + { + "epoch": 1.6308237816442648, + "grad_norm": 2.081425666809082, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.741317629814148, + "num_tokens": 484884726.0, + "step": 19777 + }, + { + "epoch": 1.6309062422693164, + "grad_norm": 2.1149258613586426, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7505759000778198, + "num_tokens": 484915108.0, + "step": 19778 + }, + { + "epoch": 1.630988702894368, + "grad_norm": 2.09454607963562, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7255119681358337, + "num_tokens": 484944555.0, + "step": 19779 + }, + { + "epoch": 1.6310711635194195, + "grad_norm": 2.080336332321167, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7533299922943115, + "num_tokens": 484970950.0, + "step": 19780 + }, + { + "epoch": 1.631153624144471, + "grad_norm": 1.957359790802002, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.793036699295044, + "num_tokens": 484998865.0, + "step": 19781 + }, + { + "epoch": 1.6312360847695224, + "grad_norm": 1.7369977235794067, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7515411972999573, + "num_tokens": 485034656.0, + "step": 19782 + }, + { + "epoch": 1.6313185453945742, + "grad_norm": 1.9926289319992065, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7749019861221313, + "num_tokens": 485062228.0, + "step": 19783 + }, + { + "epoch": 1.6314010060196256, + "grad_norm": 1.7904528379440308, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7758427262306213, + "num_tokens": 485096693.0, + "step": 19784 + }, + { + "epoch": 1.6314834666446771, + "grad_norm": 2.0155246257781982, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7409440875053406, + "num_tokens": 485125736.0, + "step": 19785 + }, + { + "epoch": 1.6315659272697287, + "grad_norm": 2.1930201053619385, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7658339738845825, + "num_tokens": 485150702.0, + "step": 19786 + }, + { + "epoch": 1.6316483878947803, + "grad_norm": 2.409808397293091, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.704393744468689, + "num_tokens": 485176288.0, + "step": 19787 + }, + { + "epoch": 1.6317308485198319, + "grad_norm": 1.7183808088302612, + "learning_rate": 1e-06, + "loss": 0.5586, + "mean_token_accuracy": 0.8240368962287903, + "num_tokens": 485210489.0, + "step": 19788 + }, + { + "epoch": 1.6318133091448832, + "grad_norm": 2.1646950244903564, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.7814517617225647, + "num_tokens": 485236071.0, + "step": 19789 + }, + { + "epoch": 1.631895769769935, + "grad_norm": 2.1811227798461914, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.737258791923523, + "num_tokens": 485262491.0, + "step": 19790 + }, + { + "epoch": 1.6319782303949864, + "grad_norm": 1.998288631439209, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.789433479309082, + "num_tokens": 485292433.0, + "step": 19791 + }, + { + "epoch": 1.632060691020038, + "grad_norm": 2.0057265758514404, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7534019351005554, + "num_tokens": 485323130.0, + "step": 19792 + }, + { + "epoch": 1.6321431516450895, + "grad_norm": 2.1379544734954834, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7660292387008667, + "num_tokens": 485350078.0, + "step": 19793 + }, + { + "epoch": 1.6322256122701408, + "grad_norm": 1.9938745498657227, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7703015804290771, + "num_tokens": 485379635.0, + "step": 19794 + }, + { + "epoch": 1.6323080728951926, + "grad_norm": 2.0493390560150146, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7280833721160889, + "num_tokens": 485408576.0, + "step": 19795 + }, + { + "epoch": 1.632390533520244, + "grad_norm": 2.1391782760620117, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7505238056182861, + "num_tokens": 485437445.0, + "step": 19796 + }, + { + "epoch": 1.6324729941452958, + "grad_norm": 2.0134572982788086, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7620683908462524, + "num_tokens": 485465893.0, + "step": 19797 + }, + { + "epoch": 1.6325554547703471, + "grad_norm": 2.001887321472168, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.792717456817627, + "num_tokens": 485495351.0, + "step": 19798 + }, + { + "epoch": 1.6326379153953987, + "grad_norm": 2.0060980319976807, + "learning_rate": 1e-06, + "loss": 0.5905, + "mean_token_accuracy": 0.7979705929756165, + "num_tokens": 485521480.0, + "step": 19799 + }, + { + "epoch": 1.6327203760204503, + "grad_norm": 2.061410665512085, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7737011909484863, + "num_tokens": 485550867.0, + "step": 19800 + }, + { + "epoch": 1.6328028366455016, + "grad_norm": 1.8698275089263916, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7642381191253662, + "num_tokens": 485584624.0, + "step": 19801 + }, + { + "epoch": 1.6328852972705534, + "grad_norm": 1.9700109958648682, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7680855989456177, + "num_tokens": 485618279.0, + "step": 19802 + }, + { + "epoch": 1.6329677578956048, + "grad_norm": 2.310826063156128, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7263832092285156, + "num_tokens": 485641758.0, + "step": 19803 + }, + { + "epoch": 1.6330502185206563, + "grad_norm": 1.9873745441436768, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7310924530029297, + "num_tokens": 485673968.0, + "step": 19804 + }, + { + "epoch": 1.633132679145708, + "grad_norm": 2.0640923976898193, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7478413581848145, + "num_tokens": 485702135.0, + "step": 19805 + }, + { + "epoch": 1.6332151397707595, + "grad_norm": 1.8561805486679077, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7352423667907715, + "num_tokens": 485737154.0, + "step": 19806 + }, + { + "epoch": 1.633297600395811, + "grad_norm": 2.035928249359131, + "learning_rate": 1e-06, + "loss": 0.5811, + "mean_token_accuracy": 0.7985527515411377, + "num_tokens": 485764890.0, + "step": 19807 + }, + { + "epoch": 1.6333800610208624, + "grad_norm": 1.9070433378219604, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7531793117523193, + "num_tokens": 485798235.0, + "step": 19808 + }, + { + "epoch": 1.6334625216459142, + "grad_norm": 1.9744558334350586, + "learning_rate": 1e-06, + "loss": 0.6076, + "mean_token_accuracy": 0.796726644039154, + "num_tokens": 485824864.0, + "step": 19809 + }, + { + "epoch": 1.6335449822709656, + "grad_norm": 1.969671607017517, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.734609842300415, + "num_tokens": 485855117.0, + "step": 19810 + }, + { + "epoch": 1.6336274428960171, + "grad_norm": 1.73162841796875, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7751790285110474, + "num_tokens": 485895796.0, + "step": 19811 + }, + { + "epoch": 1.6337099035210687, + "grad_norm": 1.715529441833496, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.8020075559616089, + "num_tokens": 485934023.0, + "step": 19812 + }, + { + "epoch": 1.6337923641461203, + "grad_norm": 1.893838882446289, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7630887031555176, + "num_tokens": 485965407.0, + "step": 19813 + }, + { + "epoch": 1.6338748247711719, + "grad_norm": 1.9895623922348022, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7764946818351746, + "num_tokens": 485996077.0, + "step": 19814 + }, + { + "epoch": 1.6339572853962232, + "grad_norm": 1.8991137742996216, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7772437930107117, + "num_tokens": 486029498.0, + "step": 19815 + }, + { + "epoch": 1.634039746021275, + "grad_norm": 2.1072802543640137, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7570523023605347, + "num_tokens": 486056188.0, + "step": 19816 + }, + { + "epoch": 1.6341222066463263, + "grad_norm": 1.8744086027145386, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7260080575942993, + "num_tokens": 486091827.0, + "step": 19817 + }, + { + "epoch": 1.634204667271378, + "grad_norm": 2.3010244369506836, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7529115080833435, + "num_tokens": 486115797.0, + "step": 19818 + }, + { + "epoch": 1.6342871278964295, + "grad_norm": 2.221190929412842, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7564914226531982, + "num_tokens": 486142530.0, + "step": 19819 + }, + { + "epoch": 1.6343695885214808, + "grad_norm": 1.9812523126602173, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7690520286560059, + "num_tokens": 486174389.0, + "step": 19820 + }, + { + "epoch": 1.6344520491465326, + "grad_norm": 2.1067841053009033, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.783391535282135, + "num_tokens": 486200235.0, + "step": 19821 + }, + { + "epoch": 1.634534509771584, + "grad_norm": 1.880496859550476, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7518397569656372, + "num_tokens": 486230443.0, + "step": 19822 + }, + { + "epoch": 1.6346169703966356, + "grad_norm": 1.8980274200439453, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7605918645858765, + "num_tokens": 486262717.0, + "step": 19823 + }, + { + "epoch": 1.6346994310216871, + "grad_norm": 1.7600772380828857, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.80021071434021, + "num_tokens": 486296053.0, + "step": 19824 + }, + { + "epoch": 1.6347818916467387, + "grad_norm": 2.0154237747192383, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7579218745231628, + "num_tokens": 486324532.0, + "step": 19825 + }, + { + "epoch": 1.6348643522717903, + "grad_norm": 1.8125633001327515, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7818526029586792, + "num_tokens": 486357124.0, + "step": 19826 + }, + { + "epoch": 1.6349468128968416, + "grad_norm": 2.0960731506347656, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7620970010757446, + "num_tokens": 486385229.0, + "step": 19827 + }, + { + "epoch": 1.6350292735218934, + "grad_norm": 2.0627975463867188, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7120732665061951, + "num_tokens": 486415939.0, + "step": 19828 + }, + { + "epoch": 1.6351117341469448, + "grad_norm": 2.2450973987579346, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7460726499557495, + "num_tokens": 486440093.0, + "step": 19829 + }, + { + "epoch": 1.6351941947719963, + "grad_norm": 1.68952476978302, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7716970443725586, + "num_tokens": 486476246.0, + "step": 19830 + }, + { + "epoch": 1.635276655397048, + "grad_norm": 1.9069098234176636, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7710903882980347, + "num_tokens": 486509234.0, + "step": 19831 + }, + { + "epoch": 1.6353591160220995, + "grad_norm": 1.8567150831222534, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7471733093261719, + "num_tokens": 486544124.0, + "step": 19832 + }, + { + "epoch": 1.635441576647151, + "grad_norm": 1.7705687284469604, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7816694378852844, + "num_tokens": 486583920.0, + "step": 19833 + }, + { + "epoch": 1.6355240372722024, + "grad_norm": 1.851367712020874, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7763512134552002, + "num_tokens": 486617641.0, + "step": 19834 + }, + { + "epoch": 1.6356064978972542, + "grad_norm": 2.167093276977539, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7784786224365234, + "num_tokens": 486640996.0, + "step": 19835 + }, + { + "epoch": 1.6356889585223056, + "grad_norm": 1.7839099168777466, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7497987747192383, + "num_tokens": 486677965.0, + "step": 19836 + }, + { + "epoch": 1.6357714191473571, + "grad_norm": 2.0548222064971924, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7537769079208374, + "num_tokens": 486706638.0, + "step": 19837 + }, + { + "epoch": 1.6358538797724087, + "grad_norm": 1.7230547666549683, + "learning_rate": 1e-06, + "loss": 0.5102, + "mean_token_accuracy": 0.8265707492828369, + "num_tokens": 486744161.0, + "step": 19838 + }, + { + "epoch": 1.63593634039746, + "grad_norm": 1.7151209115982056, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.761896014213562, + "num_tokens": 486783759.0, + "step": 19839 + }, + { + "epoch": 1.6360188010225118, + "grad_norm": 2.150979518890381, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7270375490188599, + "num_tokens": 486810961.0, + "step": 19840 + }, + { + "epoch": 1.6361012616475632, + "grad_norm": 1.9152175188064575, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7688683867454529, + "num_tokens": 486846342.0, + "step": 19841 + }, + { + "epoch": 1.6361837222726148, + "grad_norm": 1.9732224941253662, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7249078750610352, + "num_tokens": 486877782.0, + "step": 19842 + }, + { + "epoch": 1.6362661828976663, + "grad_norm": 2.2604756355285645, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7651844620704651, + "num_tokens": 486903271.0, + "step": 19843 + }, + { + "epoch": 1.636348643522718, + "grad_norm": 1.9866571426391602, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7576638460159302, + "num_tokens": 486934727.0, + "step": 19844 + }, + { + "epoch": 1.6364311041477695, + "grad_norm": 1.984301209449768, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.7985867261886597, + "num_tokens": 486963731.0, + "step": 19845 + }, + { + "epoch": 1.6365135647728208, + "grad_norm": 1.9524729251861572, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7902165651321411, + "num_tokens": 486993098.0, + "step": 19846 + }, + { + "epoch": 1.6365960253978726, + "grad_norm": 1.8618298768997192, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7646791934967041, + "num_tokens": 487024182.0, + "step": 19847 + }, + { + "epoch": 1.636678486022924, + "grad_norm": 1.9810031652450562, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.713913083076477, + "num_tokens": 487057416.0, + "step": 19848 + }, + { + "epoch": 1.6367609466479756, + "grad_norm": 1.7596426010131836, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.7997243404388428, + "num_tokens": 487092411.0, + "step": 19849 + }, + { + "epoch": 1.6368434072730271, + "grad_norm": 1.8141682147979736, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7561296820640564, + "num_tokens": 487126768.0, + "step": 19850 + }, + { + "epoch": 1.6369258678980787, + "grad_norm": 1.9945091009140015, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7418426871299744, + "num_tokens": 487157563.0, + "step": 19851 + }, + { + "epoch": 1.6370083285231303, + "grad_norm": 1.9841243028640747, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.759433388710022, + "num_tokens": 487186104.0, + "step": 19852 + }, + { + "epoch": 1.6370907891481816, + "grad_norm": 1.8749393224716187, + "learning_rate": 1e-06, + "loss": 0.6158, + "mean_token_accuracy": 0.7872329950332642, + "num_tokens": 487215990.0, + "step": 19853 + }, + { + "epoch": 1.6371732497732334, + "grad_norm": 1.8105344772338867, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7594364285469055, + "num_tokens": 487250321.0, + "step": 19854 + }, + { + "epoch": 1.6372557103982848, + "grad_norm": 2.2248072624206543, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.740131676197052, + "num_tokens": 487276469.0, + "step": 19855 + }, + { + "epoch": 1.6373381710233363, + "grad_norm": 2.2637765407562256, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7200672626495361, + "num_tokens": 487302197.0, + "step": 19856 + }, + { + "epoch": 1.637420631648388, + "grad_norm": 1.8146288394927979, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.7851715087890625, + "num_tokens": 487337537.0, + "step": 19857 + }, + { + "epoch": 1.6375030922734395, + "grad_norm": 1.854540467262268, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7421582937240601, + "num_tokens": 487371961.0, + "step": 19858 + }, + { + "epoch": 1.637585552898491, + "grad_norm": 1.771945834159851, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7716035842895508, + "num_tokens": 487409383.0, + "step": 19859 + }, + { + "epoch": 1.6376680135235424, + "grad_norm": 1.9278508424758911, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7636675834655762, + "num_tokens": 487439030.0, + "step": 19860 + }, + { + "epoch": 1.6377504741485942, + "grad_norm": 1.7096576690673828, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7578725218772888, + "num_tokens": 487478469.0, + "step": 19861 + }, + { + "epoch": 1.6378329347736456, + "grad_norm": 1.856549859046936, + "learning_rate": 1e-06, + "loss": 0.6059, + "mean_token_accuracy": 0.7951287031173706, + "num_tokens": 487510352.0, + "step": 19862 + }, + { + "epoch": 1.6379153953986971, + "grad_norm": 2.1499922275543213, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7307931184768677, + "num_tokens": 487537297.0, + "step": 19863 + }, + { + "epoch": 1.6379978560237487, + "grad_norm": 2.1327834129333496, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7706900835037231, + "num_tokens": 487564651.0, + "step": 19864 + }, + { + "epoch": 1.6380803166488, + "grad_norm": 2.11727237701416, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7749488949775696, + "num_tokens": 487591603.0, + "step": 19865 + }, + { + "epoch": 1.6381627772738518, + "grad_norm": 2.028434991836548, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7368955016136169, + "num_tokens": 487622457.0, + "step": 19866 + }, + { + "epoch": 1.6382452378989032, + "grad_norm": 1.763881802558899, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.7972750663757324, + "num_tokens": 487657015.0, + "step": 19867 + }, + { + "epoch": 1.6383276985239548, + "grad_norm": 2.0188663005828857, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.7858270406723022, + "num_tokens": 487685978.0, + "step": 19868 + }, + { + "epoch": 1.6384101591490063, + "grad_norm": 2.0057220458984375, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7442412376403809, + "num_tokens": 487716545.0, + "step": 19869 + }, + { + "epoch": 1.638492619774058, + "grad_norm": 1.7816606760025024, + "learning_rate": 1e-06, + "loss": 0.595, + "mean_token_accuracy": 0.8068239688873291, + "num_tokens": 487749627.0, + "step": 19870 + }, + { + "epoch": 1.6385750803991095, + "grad_norm": 1.7799948453903198, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.7795256972312927, + "num_tokens": 487782651.0, + "step": 19871 + }, + { + "epoch": 1.6386575410241608, + "grad_norm": 2.0184953212738037, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.7507753372192383, + "num_tokens": 487812798.0, + "step": 19872 + }, + { + "epoch": 1.6387400016492126, + "grad_norm": 2.0041277408599854, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7599695920944214, + "num_tokens": 487842389.0, + "step": 19873 + }, + { + "epoch": 1.638822462274264, + "grad_norm": 1.9567641019821167, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7354235649108887, + "num_tokens": 487873283.0, + "step": 19874 + }, + { + "epoch": 1.6389049228993156, + "grad_norm": 1.8118784427642822, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7657029628753662, + "num_tokens": 487906008.0, + "step": 19875 + }, + { + "epoch": 1.6389873835243671, + "grad_norm": 2.0244994163513184, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7678161263465881, + "num_tokens": 487934423.0, + "step": 19876 + }, + { + "epoch": 1.6390698441494187, + "grad_norm": 2.029444932937622, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7259302139282227, + "num_tokens": 487963565.0, + "step": 19877 + }, + { + "epoch": 1.6391523047744703, + "grad_norm": 2.2384891510009766, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7709604501724243, + "num_tokens": 487986860.0, + "step": 19878 + }, + { + "epoch": 1.6392347653995216, + "grad_norm": 2.145969867706299, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7342747449874878, + "num_tokens": 488014932.0, + "step": 19879 + }, + { + "epoch": 1.6393172260245734, + "grad_norm": 1.7220007181167603, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7631343603134155, + "num_tokens": 488053731.0, + "step": 19880 + }, + { + "epoch": 1.6393996866496248, + "grad_norm": 1.9712634086608887, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7158393859863281, + "num_tokens": 488085586.0, + "step": 19881 + }, + { + "epoch": 1.6394821472746763, + "grad_norm": 1.8407833576202393, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7724736928939819, + "num_tokens": 488119119.0, + "step": 19882 + }, + { + "epoch": 1.639564607899728, + "grad_norm": 2.1031787395477295, + "learning_rate": 1e-06, + "loss": 0.5978, + "mean_token_accuracy": 0.7924560308456421, + "num_tokens": 488142461.0, + "step": 19883 + }, + { + "epoch": 1.6396470685247793, + "grad_norm": 1.9923211336135864, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7626480460166931, + "num_tokens": 488172714.0, + "step": 19884 + }, + { + "epoch": 1.639729529149831, + "grad_norm": 2.0556821823120117, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7426348328590393, + "num_tokens": 488204503.0, + "step": 19885 + }, + { + "epoch": 1.6398119897748824, + "grad_norm": 2.0692715644836426, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.7857143878936768, + "num_tokens": 488231276.0, + "step": 19886 + }, + { + "epoch": 1.639894450399934, + "grad_norm": 2.1587400436401367, + "learning_rate": 1e-06, + "loss": 0.6004, + "mean_token_accuracy": 0.7950689792633057, + "num_tokens": 488255717.0, + "step": 19887 + }, + { + "epoch": 1.6399769110249856, + "grad_norm": 1.742624044418335, + "learning_rate": 1e-06, + "loss": 0.5978, + "mean_token_accuracy": 0.8004391193389893, + "num_tokens": 488289846.0, + "step": 19888 + }, + { + "epoch": 1.6400593716500371, + "grad_norm": 1.9591083526611328, + "learning_rate": 1e-06, + "loss": 0.7396, + "mean_token_accuracy": 0.7641744613647461, + "num_tokens": 488322087.0, + "step": 19889 + }, + { + "epoch": 1.6401418322750887, + "grad_norm": 2.2693381309509277, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7480850219726562, + "num_tokens": 488347963.0, + "step": 19890 + }, + { + "epoch": 1.64022429290014, + "grad_norm": 1.7903130054473877, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7822176814079285, + "num_tokens": 488382501.0, + "step": 19891 + }, + { + "epoch": 1.6403067535251918, + "grad_norm": 2.0226471424102783, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7528319358825684, + "num_tokens": 488410490.0, + "step": 19892 + }, + { + "epoch": 1.6403892141502432, + "grad_norm": 1.9931857585906982, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7655289173126221, + "num_tokens": 488441789.0, + "step": 19893 + }, + { + "epoch": 1.6404716747752948, + "grad_norm": 2.2250213623046875, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7538527846336365, + "num_tokens": 488466525.0, + "step": 19894 + }, + { + "epoch": 1.6405541354003463, + "grad_norm": 2.3273873329162598, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7513570189476013, + "num_tokens": 488491264.0, + "step": 19895 + }, + { + "epoch": 1.640636596025398, + "grad_norm": 1.9543834924697876, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7525826692581177, + "num_tokens": 488526187.0, + "step": 19896 + }, + { + "epoch": 1.6407190566504495, + "grad_norm": 1.8872498273849487, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7686959505081177, + "num_tokens": 488557608.0, + "step": 19897 + }, + { + "epoch": 1.6408015172755008, + "grad_norm": 1.8418684005737305, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7603569626808167, + "num_tokens": 488591052.0, + "step": 19898 + }, + { + "epoch": 1.6408839779005526, + "grad_norm": 1.9936846494674683, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7387741208076477, + "num_tokens": 488621892.0, + "step": 19899 + }, + { + "epoch": 1.640966438525604, + "grad_norm": 2.3151934146881104, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7627020478248596, + "num_tokens": 488644860.0, + "step": 19900 + }, + { + "epoch": 1.6410488991506555, + "grad_norm": 2.156555414199829, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6967658996582031, + "num_tokens": 488672617.0, + "step": 19901 + }, + { + "epoch": 1.6411313597757071, + "grad_norm": 1.7867465019226074, + "learning_rate": 1e-06, + "loss": 0.5744, + "mean_token_accuracy": 0.8098165988922119, + "num_tokens": 488704399.0, + "step": 19902 + }, + { + "epoch": 1.6412138204007585, + "grad_norm": 2.147928476333618, + "learning_rate": 1e-06, + "loss": 0.5824, + "mean_token_accuracy": 0.8058220148086548, + "num_tokens": 488730205.0, + "step": 19903 + }, + { + "epoch": 1.6412962810258103, + "grad_norm": 1.9261130094528198, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7549842000007629, + "num_tokens": 488763482.0, + "step": 19904 + }, + { + "epoch": 1.6413787416508616, + "grad_norm": 1.955653429031372, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7551231980323792, + "num_tokens": 488795152.0, + "step": 19905 + }, + { + "epoch": 1.6414612022759134, + "grad_norm": 1.7203127145767212, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7800484895706177, + "num_tokens": 488831608.0, + "step": 19906 + }, + { + "epoch": 1.6415436629009648, + "grad_norm": 2.038294553756714, + "learning_rate": 1e-06, + "loss": 0.52, + "mean_token_accuracy": 0.8239744901657104, + "num_tokens": 488856038.0, + "step": 19907 + }, + { + "epoch": 1.6416261235260163, + "grad_norm": 1.9213179349899292, + "learning_rate": 1e-06, + "loss": 0.5799, + "mean_token_accuracy": 0.8013935089111328, + "num_tokens": 488884880.0, + "step": 19908 + }, + { + "epoch": 1.641708584151068, + "grad_norm": 1.9739922285079956, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7502970695495605, + "num_tokens": 488914320.0, + "step": 19909 + }, + { + "epoch": 1.6417910447761193, + "grad_norm": 2.326394557952881, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.743057370185852, + "num_tokens": 488935836.0, + "step": 19910 + }, + { + "epoch": 1.641873505401171, + "grad_norm": 2.2318308353424072, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7138974070549011, + "num_tokens": 488960985.0, + "step": 19911 + }, + { + "epoch": 1.6419559660262224, + "grad_norm": 1.9532792568206787, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.7887678742408752, + "num_tokens": 488990158.0, + "step": 19912 + }, + { + "epoch": 1.642038426651274, + "grad_norm": 1.909368872642517, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7478387355804443, + "num_tokens": 489023507.0, + "step": 19913 + }, + { + "epoch": 1.6421208872763255, + "grad_norm": 2.3817737102508545, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7317770719528198, + "num_tokens": 489049276.0, + "step": 19914 + }, + { + "epoch": 1.6422033479013771, + "grad_norm": 2.1718733310699463, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7255125045776367, + "num_tokens": 489078692.0, + "step": 19915 + }, + { + "epoch": 1.6422858085264287, + "grad_norm": 2.058896064758301, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7513468265533447, + "num_tokens": 489106122.0, + "step": 19916 + }, + { + "epoch": 1.64236826915148, + "grad_norm": 2.1171305179595947, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7582786083221436, + "num_tokens": 489133915.0, + "step": 19917 + }, + { + "epoch": 1.6424507297765318, + "grad_norm": 1.996297836303711, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7253698706626892, + "num_tokens": 489168021.0, + "step": 19918 + }, + { + "epoch": 1.6425331904015832, + "grad_norm": 2.0731191635131836, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7468858361244202, + "num_tokens": 489195855.0, + "step": 19919 + }, + { + "epoch": 1.6426156510266348, + "grad_norm": 1.7241100072860718, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7789714336395264, + "num_tokens": 489232821.0, + "step": 19920 + }, + { + "epoch": 1.6426981116516863, + "grad_norm": 2.0114495754241943, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7150914072990417, + "num_tokens": 489265046.0, + "step": 19921 + }, + { + "epoch": 1.642780572276738, + "grad_norm": 1.86654794216156, + "learning_rate": 1e-06, + "loss": 0.6194, + "mean_token_accuracy": 0.8000038862228394, + "num_tokens": 489295651.0, + "step": 19922 + }, + { + "epoch": 1.6428630329017895, + "grad_norm": 1.9093161821365356, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7664502263069153, + "num_tokens": 489329213.0, + "step": 19923 + }, + { + "epoch": 1.6429454935268408, + "grad_norm": 1.7557494640350342, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.7836552262306213, + "num_tokens": 489367648.0, + "step": 19924 + }, + { + "epoch": 1.6430279541518926, + "grad_norm": 1.6869040727615356, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7721255421638489, + "num_tokens": 489404941.0, + "step": 19925 + }, + { + "epoch": 1.643110414776944, + "grad_norm": 1.8047640323638916, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7500183582305908, + "num_tokens": 489440419.0, + "step": 19926 + }, + { + "epoch": 1.6431928754019955, + "grad_norm": 1.9990475177764893, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7586956024169922, + "num_tokens": 489472350.0, + "step": 19927 + }, + { + "epoch": 1.6432753360270471, + "grad_norm": 2.193837881088257, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7340494394302368, + "num_tokens": 489500098.0, + "step": 19928 + }, + { + "epoch": 1.6433577966520985, + "grad_norm": 1.9369486570358276, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.7585172653198242, + "num_tokens": 489531425.0, + "step": 19929 + }, + { + "epoch": 1.6434402572771503, + "grad_norm": 1.8650062084197998, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7384979128837585, + "num_tokens": 489563057.0, + "step": 19930 + }, + { + "epoch": 1.6435227179022016, + "grad_norm": 2.3298838138580322, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7318831086158752, + "num_tokens": 489587962.0, + "step": 19931 + }, + { + "epoch": 1.6436051785272532, + "grad_norm": 1.9628950357437134, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7681212425231934, + "num_tokens": 489619620.0, + "step": 19932 + }, + { + "epoch": 1.6436876391523048, + "grad_norm": 1.6807811260223389, + "learning_rate": 1e-06, + "loss": 0.7722, + "mean_token_accuracy": 0.7603558301925659, + "num_tokens": 489660613.0, + "step": 19933 + }, + { + "epoch": 1.6437700997773563, + "grad_norm": 1.7221826314926147, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.7838117480278015, + "num_tokens": 489694158.0, + "step": 19934 + }, + { + "epoch": 1.643852560402408, + "grad_norm": 1.844860553741455, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.7911310791969299, + "num_tokens": 489727484.0, + "step": 19935 + }, + { + "epoch": 1.6439350210274593, + "grad_norm": 1.8389239311218262, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.742387056350708, + "num_tokens": 489761354.0, + "step": 19936 + }, + { + "epoch": 1.644017481652511, + "grad_norm": 1.99729323387146, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7122264504432678, + "num_tokens": 489792635.0, + "step": 19937 + }, + { + "epoch": 1.6440999422775624, + "grad_norm": 1.8152523040771484, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.7595789432525635, + "num_tokens": 489829290.0, + "step": 19938 + }, + { + "epoch": 1.644182402902614, + "grad_norm": 2.105489492416382, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7609024047851562, + "num_tokens": 489856281.0, + "step": 19939 + }, + { + "epoch": 1.6442648635276655, + "grad_norm": 1.565210223197937, + "learning_rate": 1e-06, + "loss": 0.5132, + "mean_token_accuracy": 0.8267245888710022, + "num_tokens": 489895190.0, + "step": 19940 + }, + { + "epoch": 1.6443473241527171, + "grad_norm": 2.293185234069824, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7347702980041504, + "num_tokens": 489923209.0, + "step": 19941 + }, + { + "epoch": 1.6444297847777687, + "grad_norm": 1.7969557046890259, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7671798467636108, + "num_tokens": 489958906.0, + "step": 19942 + }, + { + "epoch": 1.64451224540282, + "grad_norm": 1.8799022436141968, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7739725708961487, + "num_tokens": 489991019.0, + "step": 19943 + }, + { + "epoch": 1.6445947060278718, + "grad_norm": 2.1153719425201416, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7774498462677002, + "num_tokens": 490017807.0, + "step": 19944 + }, + { + "epoch": 1.6446771666529232, + "grad_norm": 2.0061817169189453, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7200363874435425, + "num_tokens": 490048692.0, + "step": 19945 + }, + { + "epoch": 1.6447596272779748, + "grad_norm": 1.8177090883255005, + "learning_rate": 1e-06, + "loss": 0.6017, + "mean_token_accuracy": 0.8033589720726013, + "num_tokens": 490080380.0, + "step": 19946 + }, + { + "epoch": 1.6448420879030263, + "grad_norm": 2.111220598220825, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.781877338886261, + "num_tokens": 490109323.0, + "step": 19947 + }, + { + "epoch": 1.6449245485280777, + "grad_norm": 2.046417474746704, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7301021814346313, + "num_tokens": 490139008.0, + "step": 19948 + }, + { + "epoch": 1.6450070091531295, + "grad_norm": 1.76168954372406, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7246273756027222, + "num_tokens": 490179004.0, + "step": 19949 + }, + { + "epoch": 1.6450894697781808, + "grad_norm": 2.1726975440979004, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7354851961135864, + "num_tokens": 490204357.0, + "step": 19950 + }, + { + "epoch": 1.6451719304032326, + "grad_norm": 1.9590325355529785, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7401885390281677, + "num_tokens": 490235201.0, + "step": 19951 + }, + { + "epoch": 1.645254391028284, + "grad_norm": 2.1346077919006348, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7815972566604614, + "num_tokens": 490262662.0, + "step": 19952 + }, + { + "epoch": 1.6453368516533355, + "grad_norm": 2.0748462677001953, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7785277962684631, + "num_tokens": 490291007.0, + "step": 19953 + }, + { + "epoch": 1.6454193122783871, + "grad_norm": 1.890358328819275, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.7575116157531738, + "num_tokens": 490324026.0, + "step": 19954 + }, + { + "epoch": 1.6455017729034385, + "grad_norm": 1.884716510772705, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7807213664054871, + "num_tokens": 490357477.0, + "step": 19955 + }, + { + "epoch": 1.6455842335284903, + "grad_norm": 1.861393928527832, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7528735399246216, + "num_tokens": 490393701.0, + "step": 19956 + }, + { + "epoch": 1.6456666941535416, + "grad_norm": 1.925717830657959, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7508139610290527, + "num_tokens": 490425818.0, + "step": 19957 + }, + { + "epoch": 1.6457491547785932, + "grad_norm": 2.0529890060424805, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7754313945770264, + "num_tokens": 490454282.0, + "step": 19958 + }, + { + "epoch": 1.6458316154036448, + "grad_norm": 2.0380945205688477, + "learning_rate": 1e-06, + "loss": 0.6194, + "mean_token_accuracy": 0.7979851961135864, + "num_tokens": 490481637.0, + "step": 19959 + }, + { + "epoch": 1.6459140760286963, + "grad_norm": 1.9568060636520386, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7676278352737427, + "num_tokens": 490513761.0, + "step": 19960 + }, + { + "epoch": 1.645996536653748, + "grad_norm": 2.0687828063964844, + "learning_rate": 1e-06, + "loss": 0.7852, + "mean_token_accuracy": 0.7486578226089478, + "num_tokens": 490544439.0, + "step": 19961 + }, + { + "epoch": 1.6460789972787992, + "grad_norm": 1.991776943206787, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7592322826385498, + "num_tokens": 490573146.0, + "step": 19962 + }, + { + "epoch": 1.646161457903851, + "grad_norm": 1.9236394166946411, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.7813039422035217, + "num_tokens": 490605108.0, + "step": 19963 + }, + { + "epoch": 1.6462439185289024, + "grad_norm": 1.934207558631897, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7506447434425354, + "num_tokens": 490639113.0, + "step": 19964 + }, + { + "epoch": 1.646326379153954, + "grad_norm": 2.2625339031219482, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7307630777359009, + "num_tokens": 490665088.0, + "step": 19965 + }, + { + "epoch": 1.6464088397790055, + "grad_norm": 1.8929389715194702, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7697098255157471, + "num_tokens": 490700481.0, + "step": 19966 + }, + { + "epoch": 1.6464913004040571, + "grad_norm": 1.9120938777923584, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7384096384048462, + "num_tokens": 490736277.0, + "step": 19967 + }, + { + "epoch": 1.6465737610291087, + "grad_norm": 1.9223246574401855, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7588567733764648, + "num_tokens": 490770746.0, + "step": 19968 + }, + { + "epoch": 1.64665622165416, + "grad_norm": 2.0719382762908936, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7458741068840027, + "num_tokens": 490800926.0, + "step": 19969 + }, + { + "epoch": 1.6467386822792118, + "grad_norm": 2.230743885040283, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7515537738800049, + "num_tokens": 490828051.0, + "step": 19970 + }, + { + "epoch": 1.6468211429042632, + "grad_norm": 1.9310646057128906, + "learning_rate": 1e-06, + "loss": 0.6123, + "mean_token_accuracy": 0.7981650829315186, + "num_tokens": 490856489.0, + "step": 19971 + }, + { + "epoch": 1.6469036035293148, + "grad_norm": 2.041591167449951, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.7548630237579346, + "num_tokens": 490885478.0, + "step": 19972 + }, + { + "epoch": 1.6469860641543663, + "grad_norm": 1.851433277130127, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7197632193565369, + "num_tokens": 490925515.0, + "step": 19973 + }, + { + "epoch": 1.6470685247794177, + "grad_norm": 1.9016894102096558, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.764923632144928, + "num_tokens": 490956274.0, + "step": 19974 + }, + { + "epoch": 1.6471509854044695, + "grad_norm": 1.9876776933670044, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7362009286880493, + "num_tokens": 490987879.0, + "step": 19975 + }, + { + "epoch": 1.6472334460295208, + "grad_norm": 2.0175557136535645, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7693572044372559, + "num_tokens": 491016351.0, + "step": 19976 + }, + { + "epoch": 1.6473159066545724, + "grad_norm": 1.9727740287780762, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7124195098876953, + "num_tokens": 491047191.0, + "step": 19977 + }, + { + "epoch": 1.647398367279624, + "grad_norm": 1.946584939956665, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7604540586471558, + "num_tokens": 491076644.0, + "step": 19978 + }, + { + "epoch": 1.6474808279046755, + "grad_norm": 1.9588181972503662, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7618356943130493, + "num_tokens": 491109306.0, + "step": 19979 + }, + { + "epoch": 1.647563288529727, + "grad_norm": 2.089088201522827, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.7855279445648193, + "num_tokens": 491138303.0, + "step": 19980 + }, + { + "epoch": 1.6476457491547785, + "grad_norm": 2.0579967498779297, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7468646764755249, + "num_tokens": 491167372.0, + "step": 19981 + }, + { + "epoch": 1.6477282097798303, + "grad_norm": 2.1997528076171875, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.7963216304779053, + "num_tokens": 491192134.0, + "step": 19982 + }, + { + "epoch": 1.6478106704048816, + "grad_norm": 2.012601137161255, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7817593812942505, + "num_tokens": 491217495.0, + "step": 19983 + }, + { + "epoch": 1.6478931310299332, + "grad_norm": 2.4291656017303467, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.7793647050857544, + "num_tokens": 491237707.0, + "step": 19984 + }, + { + "epoch": 1.6479755916549848, + "grad_norm": 1.8501558303833008, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.7863014936447144, + "num_tokens": 491270037.0, + "step": 19985 + }, + { + "epoch": 1.6480580522800363, + "grad_norm": 1.7304548025131226, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.7915975451469421, + "num_tokens": 491304722.0, + "step": 19986 + }, + { + "epoch": 1.648140512905088, + "grad_norm": 2.007523536682129, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7604184150695801, + "num_tokens": 491334070.0, + "step": 19987 + }, + { + "epoch": 1.6482229735301392, + "grad_norm": 1.8427481651306152, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7656734585762024, + "num_tokens": 491369371.0, + "step": 19988 + }, + { + "epoch": 1.648305434155191, + "grad_norm": 1.8697106838226318, + "learning_rate": 1e-06, + "loss": 0.5404, + "mean_token_accuracy": 0.8134607672691345, + "num_tokens": 491401922.0, + "step": 19989 + }, + { + "epoch": 1.6483878947802424, + "grad_norm": 2.197488307952881, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7498438954353333, + "num_tokens": 491428281.0, + "step": 19990 + }, + { + "epoch": 1.648470355405294, + "grad_norm": 1.9743720293045044, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7554951310157776, + "num_tokens": 491457423.0, + "step": 19991 + }, + { + "epoch": 1.6485528160303455, + "grad_norm": 2.147449254989624, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7746226787567139, + "num_tokens": 491484626.0, + "step": 19992 + }, + { + "epoch": 1.6486352766553969, + "grad_norm": 1.9820810556411743, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7769209742546082, + "num_tokens": 491514234.0, + "step": 19993 + }, + { + "epoch": 1.6487177372804487, + "grad_norm": 1.9107096195220947, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7176006436347961, + "num_tokens": 491547556.0, + "step": 19994 + }, + { + "epoch": 1.6488001979055, + "grad_norm": 2.105422258377075, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7468475103378296, + "num_tokens": 491574547.0, + "step": 19995 + }, + { + "epoch": 1.6488826585305518, + "grad_norm": 2.0928401947021484, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7539193034172058, + "num_tokens": 491599848.0, + "step": 19996 + }, + { + "epoch": 1.6489651191556032, + "grad_norm": 2.164041757583618, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7375524640083313, + "num_tokens": 491626838.0, + "step": 19997 + }, + { + "epoch": 1.6490475797806547, + "grad_norm": 1.7565945386886597, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7370994091033936, + "num_tokens": 491662404.0, + "step": 19998 + }, + { + "epoch": 1.6491300404057063, + "grad_norm": 2.0078227519989014, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7415570020675659, + "num_tokens": 491691069.0, + "step": 19999 + }, + { + "epoch": 1.6492125010307577, + "grad_norm": 2.1883010864257812, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7265183925628662, + "num_tokens": 491717983.0, + "step": 20000 + }, + { + "epoch": 1.6492949616558095, + "grad_norm": 2.306291341781616, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.719298243522644, + "num_tokens": 491743069.0, + "step": 20001 + }, + { + "epoch": 1.6493774222808608, + "grad_norm": 1.905449390411377, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7183772325515747, + "num_tokens": 491778767.0, + "step": 20002 + }, + { + "epoch": 1.6494598829059124, + "grad_norm": 2.059265375137329, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7697064876556396, + "num_tokens": 491807137.0, + "step": 20003 + }, + { + "epoch": 1.649542343530964, + "grad_norm": 1.961924433708191, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.7865310311317444, + "num_tokens": 491835611.0, + "step": 20004 + }, + { + "epoch": 1.6496248041560155, + "grad_norm": 2.3872313499450684, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7494193315505981, + "num_tokens": 491857380.0, + "step": 20005 + }, + { + "epoch": 1.649707264781067, + "grad_norm": 2.0055763721466064, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7548714280128479, + "num_tokens": 491887409.0, + "step": 20006 + }, + { + "epoch": 1.6497897254061185, + "grad_norm": 2.1222753524780273, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.748186469078064, + "num_tokens": 491913737.0, + "step": 20007 + }, + { + "epoch": 1.6498721860311703, + "grad_norm": 1.919907569885254, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7660790681838989, + "num_tokens": 491947044.0, + "step": 20008 + }, + { + "epoch": 1.6499546466562216, + "grad_norm": 1.8379271030426025, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7724580764770508, + "num_tokens": 491978738.0, + "step": 20009 + }, + { + "epoch": 1.6500371072812732, + "grad_norm": 1.9752223491668701, + "learning_rate": 1e-06, + "loss": 0.585, + "mean_token_accuracy": 0.7969461679458618, + "num_tokens": 492008734.0, + "step": 20010 + }, + { + "epoch": 1.6501195679063247, + "grad_norm": 2.478478193283081, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7267286777496338, + "num_tokens": 492029002.0, + "step": 20011 + }, + { + "epoch": 1.6502020285313763, + "grad_norm": 2.404454231262207, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7705349326133728, + "num_tokens": 492052228.0, + "step": 20012 + }, + { + "epoch": 1.650284489156428, + "grad_norm": 1.8224116563796997, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7668256163597107, + "num_tokens": 492088763.0, + "step": 20013 + }, + { + "epoch": 1.6503669497814792, + "grad_norm": 2.0668113231658936, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.7761918902397156, + "num_tokens": 492117994.0, + "step": 20014 + }, + { + "epoch": 1.650449410406531, + "grad_norm": 2.0733256340026855, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7538580894470215, + "num_tokens": 492149489.0, + "step": 20015 + }, + { + "epoch": 1.6505318710315824, + "grad_norm": 1.9111748933792114, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7763437628746033, + "num_tokens": 492183452.0, + "step": 20016 + }, + { + "epoch": 1.650614331656634, + "grad_norm": 1.9622466564178467, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7624256610870361, + "num_tokens": 492215520.0, + "step": 20017 + }, + { + "epoch": 1.6506967922816855, + "grad_norm": 1.8698246479034424, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7181053161621094, + "num_tokens": 492249918.0, + "step": 20018 + }, + { + "epoch": 1.6507792529067369, + "grad_norm": 2.0819931030273438, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7607786655426025, + "num_tokens": 492278983.0, + "step": 20019 + }, + { + "epoch": 1.6508617135317887, + "grad_norm": 2.0964574813842773, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7817465662956238, + "num_tokens": 492306977.0, + "step": 20020 + }, + { + "epoch": 1.65094417415684, + "grad_norm": 1.9291390180587769, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7280217409133911, + "num_tokens": 492342822.0, + "step": 20021 + }, + { + "epoch": 1.6510266347818916, + "grad_norm": 1.8775240182876587, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.7782837152481079, + "num_tokens": 492373691.0, + "step": 20022 + }, + { + "epoch": 1.6511090954069432, + "grad_norm": 2.0201938152313232, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.770484209060669, + "num_tokens": 492399335.0, + "step": 20023 + }, + { + "epoch": 1.6511915560319947, + "grad_norm": 2.167880058288574, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7651808261871338, + "num_tokens": 492426249.0, + "step": 20024 + }, + { + "epoch": 1.6512740166570463, + "grad_norm": 1.7822961807250977, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7761552333831787, + "num_tokens": 492459625.0, + "step": 20025 + }, + { + "epoch": 1.6513564772820977, + "grad_norm": 1.9272602796554565, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.7817769646644592, + "num_tokens": 492492159.0, + "step": 20026 + }, + { + "epoch": 1.6514389379071495, + "grad_norm": 1.9358841180801392, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7741293907165527, + "num_tokens": 492523932.0, + "step": 20027 + }, + { + "epoch": 1.6515213985322008, + "grad_norm": 2.027536392211914, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7331615686416626, + "num_tokens": 492552128.0, + "step": 20028 + }, + { + "epoch": 1.6516038591572524, + "grad_norm": 1.894002079963684, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7599484324455261, + "num_tokens": 492583645.0, + "step": 20029 + }, + { + "epoch": 1.651686319782304, + "grad_norm": 1.9890819787979126, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7641317844390869, + "num_tokens": 492613479.0, + "step": 20030 + }, + { + "epoch": 1.6517687804073555, + "grad_norm": 2.289263963699341, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7610198855400085, + "num_tokens": 492637734.0, + "step": 20031 + }, + { + "epoch": 1.651851241032407, + "grad_norm": 2.415097236633301, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7452572584152222, + "num_tokens": 492658761.0, + "step": 20032 + }, + { + "epoch": 1.6519337016574585, + "grad_norm": 2.218569278717041, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7280511260032654, + "num_tokens": 492686453.0, + "step": 20033 + }, + { + "epoch": 1.6520161622825102, + "grad_norm": 2.0973358154296875, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7392776012420654, + "num_tokens": 492717109.0, + "step": 20034 + }, + { + "epoch": 1.6520986229075616, + "grad_norm": 2.1420581340789795, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7541905641555786, + "num_tokens": 492741791.0, + "step": 20035 + }, + { + "epoch": 1.6521810835326132, + "grad_norm": 1.9118070602416992, + "learning_rate": 1e-06, + "loss": 0.5936, + "mean_token_accuracy": 0.7985070943832397, + "num_tokens": 492770428.0, + "step": 20036 + }, + { + "epoch": 1.6522635441576647, + "grad_norm": 2.3694841861724854, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7412633299827576, + "num_tokens": 492793591.0, + "step": 20037 + }, + { + "epoch": 1.652346004782716, + "grad_norm": 2.325483798980713, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.74078768491745, + "num_tokens": 492818905.0, + "step": 20038 + }, + { + "epoch": 1.6524284654077679, + "grad_norm": 2.2330806255340576, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7512009143829346, + "num_tokens": 492846957.0, + "step": 20039 + }, + { + "epoch": 1.6525109260328192, + "grad_norm": 2.100804328918457, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7615727186203003, + "num_tokens": 492875736.0, + "step": 20040 + }, + { + "epoch": 1.652593386657871, + "grad_norm": 2.0837950706481934, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7517995834350586, + "num_tokens": 492903936.0, + "step": 20041 + }, + { + "epoch": 1.6526758472829224, + "grad_norm": 1.924089789390564, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7229297161102295, + "num_tokens": 492938779.0, + "step": 20042 + }, + { + "epoch": 1.652758307907974, + "grad_norm": 1.9352787733078003, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.8045184016227722, + "num_tokens": 492969260.0, + "step": 20043 + }, + { + "epoch": 1.6528407685330255, + "grad_norm": 2.5412302017211914, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7227160930633545, + "num_tokens": 492990145.0, + "step": 20044 + }, + { + "epoch": 1.6529232291580769, + "grad_norm": 2.262683868408203, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7080805897712708, + "num_tokens": 493019858.0, + "step": 20045 + }, + { + "epoch": 1.6530056897831287, + "grad_norm": 1.786635160446167, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7312157154083252, + "num_tokens": 493056441.0, + "step": 20046 + }, + { + "epoch": 1.65308815040818, + "grad_norm": 2.1027815341949463, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7502164244651794, + "num_tokens": 493085553.0, + "step": 20047 + }, + { + "epoch": 1.6531706110332316, + "grad_norm": 1.7072926759719849, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.7846688032150269, + "num_tokens": 493123968.0, + "step": 20048 + }, + { + "epoch": 1.6532530716582832, + "grad_norm": 1.9528203010559082, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7496738433837891, + "num_tokens": 493156242.0, + "step": 20049 + }, + { + "epoch": 1.6533355322833347, + "grad_norm": 1.9796020984649658, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7771285772323608, + "num_tokens": 493188628.0, + "step": 20050 + }, + { + "epoch": 1.6534179929083863, + "grad_norm": 1.8872710466384888, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7322725653648376, + "num_tokens": 493221741.0, + "step": 20051 + }, + { + "epoch": 1.6535004535334377, + "grad_norm": 1.798022985458374, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7665610909461975, + "num_tokens": 493255472.0, + "step": 20052 + }, + { + "epoch": 1.6535829141584895, + "grad_norm": 1.8168631792068481, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7574965953826904, + "num_tokens": 493292427.0, + "step": 20053 + }, + { + "epoch": 1.6536653747835408, + "grad_norm": 2.035703420639038, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.745742917060852, + "num_tokens": 493324830.0, + "step": 20054 + }, + { + "epoch": 1.6537478354085924, + "grad_norm": 1.9916211366653442, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7606319189071655, + "num_tokens": 493354386.0, + "step": 20055 + }, + { + "epoch": 1.653830296033644, + "grad_norm": 1.9649492502212524, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7694904804229736, + "num_tokens": 493383817.0, + "step": 20056 + }, + { + "epoch": 1.6539127566586955, + "grad_norm": 2.0222251415252686, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7516891956329346, + "num_tokens": 493412809.0, + "step": 20057 + }, + { + "epoch": 1.653995217283747, + "grad_norm": 1.7580235004425049, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7787188291549683, + "num_tokens": 493447455.0, + "step": 20058 + }, + { + "epoch": 1.6540776779087984, + "grad_norm": 1.7208553552627563, + "learning_rate": 1e-06, + "loss": 0.5071, + "mean_token_accuracy": 0.8271299600601196, + "num_tokens": 493480530.0, + "step": 20059 + }, + { + "epoch": 1.6541601385338502, + "grad_norm": 1.974374771118164, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7514121532440186, + "num_tokens": 493511377.0, + "step": 20060 + }, + { + "epoch": 1.6542425991589016, + "grad_norm": 2.209604024887085, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7605326771736145, + "num_tokens": 493540204.0, + "step": 20061 + }, + { + "epoch": 1.6543250597839532, + "grad_norm": 2.1153647899627686, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.764473557472229, + "num_tokens": 493565556.0, + "step": 20062 + }, + { + "epoch": 1.6544075204090047, + "grad_norm": 2.243640899658203, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7594825029373169, + "num_tokens": 493589812.0, + "step": 20063 + }, + { + "epoch": 1.654489981034056, + "grad_norm": 1.948424220085144, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7461142539978027, + "num_tokens": 493621479.0, + "step": 20064 + }, + { + "epoch": 1.6545724416591079, + "grad_norm": 1.972428560256958, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7856152057647705, + "num_tokens": 493651304.0, + "step": 20065 + }, + { + "epoch": 1.6546549022841592, + "grad_norm": 2.2653985023498535, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7539242506027222, + "num_tokens": 493673576.0, + "step": 20066 + }, + { + "epoch": 1.6547373629092108, + "grad_norm": 1.633669376373291, + "learning_rate": 1e-06, + "loss": 0.6315, + "mean_token_accuracy": 0.798404335975647, + "num_tokens": 493712322.0, + "step": 20067 + }, + { + "epoch": 1.6548198235342624, + "grad_norm": 1.767237663269043, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7754203677177429, + "num_tokens": 493747637.0, + "step": 20068 + }, + { + "epoch": 1.654902284159314, + "grad_norm": 2.2242186069488525, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7683530449867249, + "num_tokens": 493770596.0, + "step": 20069 + }, + { + "epoch": 1.6549847447843655, + "grad_norm": 2.0507826805114746, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7423045635223389, + "num_tokens": 493800234.0, + "step": 20070 + }, + { + "epoch": 1.6550672054094169, + "grad_norm": 1.9418444633483887, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7490689754486084, + "num_tokens": 493829327.0, + "step": 20071 + }, + { + "epoch": 1.6551496660344687, + "grad_norm": 1.7931767702102661, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7720003128051758, + "num_tokens": 493866368.0, + "step": 20072 + }, + { + "epoch": 1.65523212665952, + "grad_norm": 1.9515409469604492, + "learning_rate": 1e-06, + "loss": 0.596, + "mean_token_accuracy": 0.8004448413848877, + "num_tokens": 493895248.0, + "step": 20073 + }, + { + "epoch": 1.6553145872845716, + "grad_norm": 1.8444952964782715, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7494092583656311, + "num_tokens": 493929861.0, + "step": 20074 + }, + { + "epoch": 1.6553970479096232, + "grad_norm": 2.0791327953338623, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7739205360412598, + "num_tokens": 493954970.0, + "step": 20075 + }, + { + "epoch": 1.6554795085346747, + "grad_norm": 2.0635995864868164, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7397078275680542, + "num_tokens": 493986725.0, + "step": 20076 + }, + { + "epoch": 1.6555619691597263, + "grad_norm": 1.8311995267868042, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7585276365280151, + "num_tokens": 494023937.0, + "step": 20077 + }, + { + "epoch": 1.6556444297847777, + "grad_norm": 1.9466042518615723, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.7883345484733582, + "num_tokens": 494053096.0, + "step": 20078 + }, + { + "epoch": 1.6557268904098295, + "grad_norm": 2.010077953338623, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.720992386341095, + "num_tokens": 494087084.0, + "step": 20079 + }, + { + "epoch": 1.6558093510348808, + "grad_norm": 1.9955412149429321, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7673878073692322, + "num_tokens": 494117574.0, + "step": 20080 + }, + { + "epoch": 1.6558918116599324, + "grad_norm": 1.8842849731445312, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7660644054412842, + "num_tokens": 494149342.0, + "step": 20081 + }, + { + "epoch": 1.655974272284984, + "grad_norm": 2.0897607803344727, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7451327443122864, + "num_tokens": 494179054.0, + "step": 20082 + }, + { + "epoch": 1.6560567329100353, + "grad_norm": 2.2622079849243164, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7344779372215271, + "num_tokens": 494203117.0, + "step": 20083 + }, + { + "epoch": 1.656139193535087, + "grad_norm": 2.0324230194091797, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7804437875747681, + "num_tokens": 494229349.0, + "step": 20084 + }, + { + "epoch": 1.6562216541601384, + "grad_norm": 1.924681305885315, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7614154815673828, + "num_tokens": 494259638.0, + "step": 20085 + }, + { + "epoch": 1.65630411478519, + "grad_norm": 2.1353721618652344, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7710962295532227, + "num_tokens": 494287043.0, + "step": 20086 + }, + { + "epoch": 1.6563865754102416, + "grad_norm": 1.854442834854126, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7592390775680542, + "num_tokens": 494322495.0, + "step": 20087 + }, + { + "epoch": 1.6564690360352932, + "grad_norm": 1.8428622484207153, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.7863516211509705, + "num_tokens": 494353701.0, + "step": 20088 + }, + { + "epoch": 1.6565514966603447, + "grad_norm": 2.2154719829559326, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7449284791946411, + "num_tokens": 494379591.0, + "step": 20089 + }, + { + "epoch": 1.656633957285396, + "grad_norm": 1.8521422147750854, + "learning_rate": 1e-06, + "loss": 0.7486, + "mean_token_accuracy": 0.7597243785858154, + "num_tokens": 494411247.0, + "step": 20090 + }, + { + "epoch": 1.6567164179104479, + "grad_norm": 1.9196444749832153, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7452073097229004, + "num_tokens": 494441621.0, + "step": 20091 + }, + { + "epoch": 1.6567988785354992, + "grad_norm": 1.985592007637024, + "learning_rate": 1e-06, + "loss": 0.5643, + "mean_token_accuracy": 0.8095593452453613, + "num_tokens": 494469398.0, + "step": 20092 + }, + { + "epoch": 1.6568813391605508, + "grad_norm": 1.9213383197784424, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7389183044433594, + "num_tokens": 494503072.0, + "step": 20093 + }, + { + "epoch": 1.6569637997856024, + "grad_norm": 1.921215295791626, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7623066902160645, + "num_tokens": 494535041.0, + "step": 20094 + }, + { + "epoch": 1.657046260410654, + "grad_norm": 2.08842396736145, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7169088125228882, + "num_tokens": 494563653.0, + "step": 20095 + }, + { + "epoch": 1.6571287210357055, + "grad_norm": 1.9217435121536255, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.745387077331543, + "num_tokens": 494597740.0, + "step": 20096 + }, + { + "epoch": 1.6572111816607569, + "grad_norm": 2.079054355621338, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7843245267868042, + "num_tokens": 494627205.0, + "step": 20097 + }, + { + "epoch": 1.6572936422858087, + "grad_norm": 1.8681411743164062, + "learning_rate": 1e-06, + "loss": 0.5794, + "mean_token_accuracy": 0.8175477981567383, + "num_tokens": 494660286.0, + "step": 20098 + }, + { + "epoch": 1.65737610291086, + "grad_norm": 2.039661169052124, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7622725963592529, + "num_tokens": 494693371.0, + "step": 20099 + }, + { + "epoch": 1.6574585635359116, + "grad_norm": 1.9474889039993286, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7809830904006958, + "num_tokens": 494724758.0, + "step": 20100 + }, + { + "epoch": 1.6575410241609632, + "grad_norm": 1.8339694738388062, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7749489545822144, + "num_tokens": 494758912.0, + "step": 20101 + }, + { + "epoch": 1.6576234847860145, + "grad_norm": 1.8264338970184326, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.7862030267715454, + "num_tokens": 494794842.0, + "step": 20102 + }, + { + "epoch": 1.6577059454110663, + "grad_norm": 2.0819294452667236, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7590094804763794, + "num_tokens": 494823565.0, + "step": 20103 + }, + { + "epoch": 1.6577884060361177, + "grad_norm": 1.8796086311340332, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7878141403198242, + "num_tokens": 494860180.0, + "step": 20104 + }, + { + "epoch": 1.6578708666611695, + "grad_norm": 1.8571945428848267, + "learning_rate": 1e-06, + "loss": 0.5011, + "mean_token_accuracy": 0.8375270366668701, + "num_tokens": 494894282.0, + "step": 20105 + }, + { + "epoch": 1.6579533272862208, + "grad_norm": 2.192551374435425, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7163116931915283, + "num_tokens": 494924357.0, + "step": 20106 + }, + { + "epoch": 1.6580357879112724, + "grad_norm": 2.072387933731079, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7558996677398682, + "num_tokens": 494953990.0, + "step": 20107 + }, + { + "epoch": 1.658118248536324, + "grad_norm": 1.8416625261306763, + "learning_rate": 1e-06, + "loss": 0.5961, + "mean_token_accuracy": 0.8040905594825745, + "num_tokens": 494984328.0, + "step": 20108 + }, + { + "epoch": 1.6582007091613753, + "grad_norm": 2.2825396060943604, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7550829648971558, + "num_tokens": 495009719.0, + "step": 20109 + }, + { + "epoch": 1.658283169786427, + "grad_norm": 2.0444211959838867, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7675977349281311, + "num_tokens": 495039102.0, + "step": 20110 + }, + { + "epoch": 1.6583656304114784, + "grad_norm": 1.903862476348877, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7659047842025757, + "num_tokens": 495071419.0, + "step": 20111 + }, + { + "epoch": 1.65844809103653, + "grad_norm": 1.739189863204956, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7779788970947266, + "num_tokens": 495103237.0, + "step": 20112 + }, + { + "epoch": 1.6585305516615816, + "grad_norm": 1.7768391370773315, + "learning_rate": 1e-06, + "loss": 0.5722, + "mean_token_accuracy": 0.8109796047210693, + "num_tokens": 495135665.0, + "step": 20113 + }, + { + "epoch": 1.6586130122866332, + "grad_norm": 2.0708093643188477, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7797406911849976, + "num_tokens": 495159520.0, + "step": 20114 + }, + { + "epoch": 1.6586954729116847, + "grad_norm": 2.1218061447143555, + "learning_rate": 1e-06, + "loss": 0.5904, + "mean_token_accuracy": 0.7814942002296448, + "num_tokens": 495185007.0, + "step": 20115 + }, + { + "epoch": 1.658777933536736, + "grad_norm": 1.896432876586914, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7768738269805908, + "num_tokens": 495215179.0, + "step": 20116 + }, + { + "epoch": 1.6588603941617879, + "grad_norm": 2.2711105346679688, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.729278564453125, + "num_tokens": 495240115.0, + "step": 20117 + }, + { + "epoch": 1.6589428547868392, + "grad_norm": 1.9990880489349365, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7753444910049438, + "num_tokens": 495269262.0, + "step": 20118 + }, + { + "epoch": 1.6590253154118908, + "grad_norm": 2.161432981491089, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7361180782318115, + "num_tokens": 495299095.0, + "step": 20119 + }, + { + "epoch": 1.6591077760369424, + "grad_norm": 2.0819318294525146, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7731386423110962, + "num_tokens": 495326844.0, + "step": 20120 + }, + { + "epoch": 1.659190236661994, + "grad_norm": 2.0081887245178223, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7583795189857483, + "num_tokens": 495355882.0, + "step": 20121 + }, + { + "epoch": 1.6592726972870455, + "grad_norm": 2.013791561126709, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7590333223342896, + "num_tokens": 495382015.0, + "step": 20122 + }, + { + "epoch": 1.6593551579120969, + "grad_norm": 1.8362032175064087, + "learning_rate": 1e-06, + "loss": 0.5554, + "mean_token_accuracy": 0.8197758793830872, + "num_tokens": 495412790.0, + "step": 20123 + }, + { + "epoch": 1.6594376185371487, + "grad_norm": 1.9095993041992188, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7779549956321716, + "num_tokens": 495443567.0, + "step": 20124 + }, + { + "epoch": 1.6595200791622, + "grad_norm": 1.9653773307800293, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7571247816085815, + "num_tokens": 495473110.0, + "step": 20125 + }, + { + "epoch": 1.6596025397872516, + "grad_norm": 2.1442718505859375, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7409509420394897, + "num_tokens": 495498799.0, + "step": 20126 + }, + { + "epoch": 1.6596850004123032, + "grad_norm": 1.9575636386871338, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.7857984304428101, + "num_tokens": 495526201.0, + "step": 20127 + }, + { + "epoch": 1.6597674610373545, + "grad_norm": 1.7114596366882324, + "learning_rate": 1e-06, + "loss": 0.7669, + "mean_token_accuracy": 0.7567977905273438, + "num_tokens": 495564672.0, + "step": 20128 + }, + { + "epoch": 1.6598499216624063, + "grad_norm": 1.9374579191207886, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.7902494668960571, + "num_tokens": 495595603.0, + "step": 20129 + }, + { + "epoch": 1.6599323822874577, + "grad_norm": 2.0278735160827637, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7687605619430542, + "num_tokens": 495623754.0, + "step": 20130 + }, + { + "epoch": 1.6600148429125092, + "grad_norm": 1.7438150644302368, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.761581540107727, + "num_tokens": 495659394.0, + "step": 20131 + }, + { + "epoch": 1.6600973035375608, + "grad_norm": 2.1323912143707275, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7742416262626648, + "num_tokens": 495686170.0, + "step": 20132 + }, + { + "epoch": 1.6601797641626124, + "grad_norm": 2.1958820819854736, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7641550302505493, + "num_tokens": 495713546.0, + "step": 20133 + }, + { + "epoch": 1.660262224787664, + "grad_norm": 1.8399462699890137, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7527269124984741, + "num_tokens": 495748991.0, + "step": 20134 + }, + { + "epoch": 1.6603446854127153, + "grad_norm": 1.9812772274017334, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7791301012039185, + "num_tokens": 495778203.0, + "step": 20135 + }, + { + "epoch": 1.660427146037767, + "grad_norm": 2.1658596992492676, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7552524209022522, + "num_tokens": 495802934.0, + "step": 20136 + }, + { + "epoch": 1.6605096066628184, + "grad_norm": 1.9974298477172852, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7456403374671936, + "num_tokens": 495831953.0, + "step": 20137 + }, + { + "epoch": 1.66059206728787, + "grad_norm": 1.6730949878692627, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.7942301034927368, + "num_tokens": 495872121.0, + "step": 20138 + }, + { + "epoch": 1.6606745279129216, + "grad_norm": 2.104313373565674, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7493570446968079, + "num_tokens": 495901298.0, + "step": 20139 + }, + { + "epoch": 1.6607569885379732, + "grad_norm": 1.683009147644043, + "learning_rate": 1e-06, + "loss": 0.6145, + "mean_token_accuracy": 0.7908657789230347, + "num_tokens": 495937418.0, + "step": 20140 + }, + { + "epoch": 1.6608394491630247, + "grad_norm": 2.080599308013916, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7391899824142456, + "num_tokens": 495966809.0, + "step": 20141 + }, + { + "epoch": 1.660921909788076, + "grad_norm": 2.2444963455200195, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7421784996986389, + "num_tokens": 495992373.0, + "step": 20142 + }, + { + "epoch": 1.6610043704131279, + "grad_norm": 1.8250433206558228, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7623046636581421, + "num_tokens": 496026528.0, + "step": 20143 + }, + { + "epoch": 1.6610868310381792, + "grad_norm": 2.0957014560699463, + "learning_rate": 1e-06, + "loss": 0.6072, + "mean_token_accuracy": 0.8029975295066833, + "num_tokens": 496054706.0, + "step": 20144 + }, + { + "epoch": 1.6611692916632308, + "grad_norm": 2.317927122116089, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7547976970672607, + "num_tokens": 496077760.0, + "step": 20145 + }, + { + "epoch": 1.6612517522882824, + "grad_norm": 2.0697848796844482, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7578582167625427, + "num_tokens": 496106986.0, + "step": 20146 + }, + { + "epoch": 1.6613342129133337, + "grad_norm": 1.887220025062561, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.7942181825637817, + "num_tokens": 496137887.0, + "step": 20147 + }, + { + "epoch": 1.6614166735383855, + "grad_norm": 1.8601402044296265, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7303774356842041, + "num_tokens": 496172008.0, + "step": 20148 + }, + { + "epoch": 1.6614991341634369, + "grad_norm": 2.045186996459961, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7609512805938721, + "num_tokens": 496198031.0, + "step": 20149 + }, + { + "epoch": 1.6615815947884887, + "grad_norm": 2.2931900024414062, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7459463477134705, + "num_tokens": 496223198.0, + "step": 20150 + }, + { + "epoch": 1.66166405541354, + "grad_norm": 2.031468391418457, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7486340999603271, + "num_tokens": 496253863.0, + "step": 20151 + }, + { + "epoch": 1.6617465160385916, + "grad_norm": 2.1017630100250244, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.773460328578949, + "num_tokens": 496279381.0, + "step": 20152 + }, + { + "epoch": 1.6618289766636432, + "grad_norm": 1.6049625873565674, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7650638222694397, + "num_tokens": 496319461.0, + "step": 20153 + }, + { + "epoch": 1.6619114372886945, + "grad_norm": 1.9240845441818237, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7716946005821228, + "num_tokens": 496353085.0, + "step": 20154 + }, + { + "epoch": 1.6619938979137463, + "grad_norm": 1.7897182703018188, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7875372767448425, + "num_tokens": 496385485.0, + "step": 20155 + }, + { + "epoch": 1.6620763585387976, + "grad_norm": 2.164722442626953, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7809261083602905, + "num_tokens": 496411079.0, + "step": 20156 + }, + { + "epoch": 1.6621588191638492, + "grad_norm": 2.02242112159729, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.7860808968544006, + "num_tokens": 496439724.0, + "step": 20157 + }, + { + "epoch": 1.6622412797889008, + "grad_norm": 1.9623430967330933, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7631970047950745, + "num_tokens": 496469466.0, + "step": 20158 + }, + { + "epoch": 1.6623237404139524, + "grad_norm": 1.8816661834716797, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7876007556915283, + "num_tokens": 496498917.0, + "step": 20159 + }, + { + "epoch": 1.662406201039004, + "grad_norm": 1.9998606443405151, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.6948733925819397, + "num_tokens": 496533738.0, + "step": 20160 + }, + { + "epoch": 1.6624886616640553, + "grad_norm": 2.087176561355591, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7037996053695679, + "num_tokens": 496564391.0, + "step": 20161 + }, + { + "epoch": 1.662571122289107, + "grad_norm": 1.9550023078918457, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7548965811729431, + "num_tokens": 496595313.0, + "step": 20162 + }, + { + "epoch": 1.6626535829141584, + "grad_norm": 1.9257745742797852, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7715433835983276, + "num_tokens": 496627261.0, + "step": 20163 + }, + { + "epoch": 1.66273604353921, + "grad_norm": 2.149170160293579, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7733571529388428, + "num_tokens": 496653200.0, + "step": 20164 + }, + { + "epoch": 1.6628185041642616, + "grad_norm": 2.2189693450927734, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7672426104545593, + "num_tokens": 496677681.0, + "step": 20165 + }, + { + "epoch": 1.6629009647893132, + "grad_norm": 1.9152768850326538, + "learning_rate": 1e-06, + "loss": 0.5349, + "mean_token_accuracy": 0.8159211874008179, + "num_tokens": 496704658.0, + "step": 20166 + }, + { + "epoch": 1.6629834254143647, + "grad_norm": 1.937652587890625, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7582792043685913, + "num_tokens": 496736168.0, + "step": 20167 + }, + { + "epoch": 1.663065886039416, + "grad_norm": 2.0743112564086914, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.778913140296936, + "num_tokens": 496764134.0, + "step": 20168 + }, + { + "epoch": 1.6631483466644679, + "grad_norm": 2.26822829246521, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7510371208190918, + "num_tokens": 496788487.0, + "step": 20169 + }, + { + "epoch": 1.6632308072895192, + "grad_norm": 2.03558349609375, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7625130414962769, + "num_tokens": 496817655.0, + "step": 20170 + }, + { + "epoch": 1.6633132679145708, + "grad_norm": 1.8738465309143066, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7728731632232666, + "num_tokens": 496848923.0, + "step": 20171 + }, + { + "epoch": 1.6633957285396224, + "grad_norm": 2.2047581672668457, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7485023736953735, + "num_tokens": 496874751.0, + "step": 20172 + }, + { + "epoch": 1.6634781891646737, + "grad_norm": 1.9383249282836914, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7694580554962158, + "num_tokens": 496905031.0, + "step": 20173 + }, + { + "epoch": 1.6635606497897255, + "grad_norm": 2.0908756256103516, + "learning_rate": 1e-06, + "loss": 0.6021, + "mean_token_accuracy": 0.7924391031265259, + "num_tokens": 496930956.0, + "step": 20174 + }, + { + "epoch": 1.6636431104147769, + "grad_norm": 2.219179391860962, + "learning_rate": 1e-06, + "loss": 0.5841, + "mean_token_accuracy": 0.8050800561904907, + "num_tokens": 496955477.0, + "step": 20175 + }, + { + "epoch": 1.6637255710398284, + "grad_norm": 2.1282451152801514, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7260043025016785, + "num_tokens": 496985122.0, + "step": 20176 + }, + { + "epoch": 1.66380803166488, + "grad_norm": 1.8970993757247925, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7037623524665833, + "num_tokens": 497018473.0, + "step": 20177 + }, + { + "epoch": 1.6638904922899316, + "grad_norm": 2.1425089836120605, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7602875232696533, + "num_tokens": 497046412.0, + "step": 20178 + }, + { + "epoch": 1.6639729529149831, + "grad_norm": 2.0409655570983887, + "learning_rate": 1e-06, + "loss": 0.5871, + "mean_token_accuracy": 0.8018304109573364, + "num_tokens": 497073756.0, + "step": 20179 + }, + { + "epoch": 1.6640554135400345, + "grad_norm": 2.206237554550171, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7512632608413696, + "num_tokens": 497099232.0, + "step": 20180 + }, + { + "epoch": 1.6641378741650863, + "grad_norm": 2.1400458812713623, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.746029257774353, + "num_tokens": 497127932.0, + "step": 20181 + }, + { + "epoch": 1.6642203347901376, + "grad_norm": 1.8928152322769165, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.7523206472396851, + "num_tokens": 497161918.0, + "step": 20182 + }, + { + "epoch": 1.6643027954151892, + "grad_norm": 2.3361196517944336, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7525489926338196, + "num_tokens": 497186083.0, + "step": 20183 + }, + { + "epoch": 1.6643852560402408, + "grad_norm": 1.891596794128418, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7437191009521484, + "num_tokens": 497221329.0, + "step": 20184 + }, + { + "epoch": 1.6644677166652924, + "grad_norm": 1.8148579597473145, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.789208173751831, + "num_tokens": 497252716.0, + "step": 20185 + }, + { + "epoch": 1.664550177290344, + "grad_norm": 2.1137354373931885, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7131104469299316, + "num_tokens": 497281896.0, + "step": 20186 + }, + { + "epoch": 1.6646326379153953, + "grad_norm": 1.9379023313522339, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7280550599098206, + "num_tokens": 497316595.0, + "step": 20187 + }, + { + "epoch": 1.664715098540447, + "grad_norm": 1.8334366083145142, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.7792559862136841, + "num_tokens": 497348042.0, + "step": 20188 + }, + { + "epoch": 1.6647975591654984, + "grad_norm": 2.2580931186676025, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7094253897666931, + "num_tokens": 497372804.0, + "step": 20189 + }, + { + "epoch": 1.66488001979055, + "grad_norm": 1.9819072484970093, + "learning_rate": 1e-06, + "loss": 0.5943, + "mean_token_accuracy": 0.7949755787849426, + "num_tokens": 497399741.0, + "step": 20190 + }, + { + "epoch": 1.6649624804156016, + "grad_norm": 2.0153157711029053, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7431962490081787, + "num_tokens": 497431748.0, + "step": 20191 + }, + { + "epoch": 1.665044941040653, + "grad_norm": 2.0343079566955566, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7709789872169495, + "num_tokens": 497461031.0, + "step": 20192 + }, + { + "epoch": 1.6651274016657047, + "grad_norm": 1.7393602132797241, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7559280395507812, + "num_tokens": 497501188.0, + "step": 20193 + }, + { + "epoch": 1.665209862290756, + "grad_norm": 2.1552650928497314, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7855873107910156, + "num_tokens": 497524903.0, + "step": 20194 + }, + { + "epoch": 1.6652923229158079, + "grad_norm": 1.9618641138076782, + "learning_rate": 1e-06, + "loss": 0.6074, + "mean_token_accuracy": 0.7953299880027771, + "num_tokens": 497552999.0, + "step": 20195 + }, + { + "epoch": 1.6653747835408592, + "grad_norm": 2.1824939250946045, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7656697630882263, + "num_tokens": 497576616.0, + "step": 20196 + }, + { + "epoch": 1.6654572441659108, + "grad_norm": 1.9983742237091064, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7135730981826782, + "num_tokens": 497607792.0, + "step": 20197 + }, + { + "epoch": 1.6655397047909624, + "grad_norm": 2.312469959259033, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7261382341384888, + "num_tokens": 497630297.0, + "step": 20198 + }, + { + "epoch": 1.6656221654160137, + "grad_norm": 1.9700322151184082, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.7880820631980896, + "num_tokens": 497659316.0, + "step": 20199 + }, + { + "epoch": 1.6657046260410655, + "grad_norm": 1.9131907224655151, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.787735104560852, + "num_tokens": 497691571.0, + "step": 20200 + }, + { + "epoch": 1.6657870866661169, + "grad_norm": 1.888805627822876, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7660961747169495, + "num_tokens": 497723263.0, + "step": 20201 + }, + { + "epoch": 1.6658695472911684, + "grad_norm": 2.2187609672546387, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7789945602416992, + "num_tokens": 497748313.0, + "step": 20202 + }, + { + "epoch": 1.66595200791622, + "grad_norm": 1.8222692012786865, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7734120488166809, + "num_tokens": 497784268.0, + "step": 20203 + }, + { + "epoch": 1.6660344685412716, + "grad_norm": 1.6328473091125488, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7594971060752869, + "num_tokens": 497825239.0, + "step": 20204 + }, + { + "epoch": 1.6661169291663231, + "grad_norm": 2.150541067123413, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7680307626724243, + "num_tokens": 497850265.0, + "step": 20205 + }, + { + "epoch": 1.6661993897913745, + "grad_norm": 2.1750361919403076, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7646142244338989, + "num_tokens": 497874063.0, + "step": 20206 + }, + { + "epoch": 1.6662818504164263, + "grad_norm": 1.7863836288452148, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.7929457426071167, + "num_tokens": 497907861.0, + "step": 20207 + }, + { + "epoch": 1.6663643110414776, + "grad_norm": 2.33583927154541, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7524304389953613, + "num_tokens": 497928557.0, + "step": 20208 + }, + { + "epoch": 1.6664467716665292, + "grad_norm": 2.5377509593963623, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7261675000190735, + "num_tokens": 497949581.0, + "step": 20209 + }, + { + "epoch": 1.6665292322915808, + "grad_norm": 1.8986259698867798, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7326091527938843, + "num_tokens": 497987070.0, + "step": 20210 + }, + { + "epoch": 1.6666116929166324, + "grad_norm": 2.1019723415374756, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7520018815994263, + "num_tokens": 498015858.0, + "step": 20211 + }, + { + "epoch": 1.666694153541684, + "grad_norm": 2.093306303024292, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7426108717918396, + "num_tokens": 498043099.0, + "step": 20212 + }, + { + "epoch": 1.6667766141667353, + "grad_norm": 1.8475826978683472, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7519737482070923, + "num_tokens": 498079982.0, + "step": 20213 + }, + { + "epoch": 1.666859074791787, + "grad_norm": 2.194549322128296, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7443199753761292, + "num_tokens": 498104991.0, + "step": 20214 + }, + { + "epoch": 1.6669415354168384, + "grad_norm": 2.1403591632843018, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7185850739479065, + "num_tokens": 498132621.0, + "step": 20215 + }, + { + "epoch": 1.66702399604189, + "grad_norm": 1.8910170793533325, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7507615685462952, + "num_tokens": 498169899.0, + "step": 20216 + }, + { + "epoch": 1.6671064566669416, + "grad_norm": 1.8109328746795654, + "learning_rate": 1e-06, + "loss": 0.5424, + "mean_token_accuracy": 0.8216259479522705, + "num_tokens": 498205053.0, + "step": 20217 + }, + { + "epoch": 1.667188917291993, + "grad_norm": 1.8846784830093384, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7535848617553711, + "num_tokens": 498237650.0, + "step": 20218 + }, + { + "epoch": 1.6672713779170447, + "grad_norm": 1.8354723453521729, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7734987735748291, + "num_tokens": 498271006.0, + "step": 20219 + }, + { + "epoch": 1.667353838542096, + "grad_norm": 2.024071455001831, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.7844178676605225, + "num_tokens": 498300710.0, + "step": 20220 + }, + { + "epoch": 1.6674362991671476, + "grad_norm": 1.8628431558609009, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7465190887451172, + "num_tokens": 498335346.0, + "step": 20221 + }, + { + "epoch": 1.6675187597921992, + "grad_norm": 1.7733769416809082, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7714989185333252, + "num_tokens": 498369356.0, + "step": 20222 + }, + { + "epoch": 1.6676012204172508, + "grad_norm": 2.1431634426116943, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7359592914581299, + "num_tokens": 498397387.0, + "step": 20223 + }, + { + "epoch": 1.6676836810423024, + "grad_norm": 1.9208672046661377, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7535208463668823, + "num_tokens": 498431461.0, + "step": 20224 + }, + { + "epoch": 1.6677661416673537, + "grad_norm": 2.132014751434326, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7506260871887207, + "num_tokens": 498458011.0, + "step": 20225 + }, + { + "epoch": 1.6678486022924055, + "grad_norm": 2.083672523498535, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.758607029914856, + "num_tokens": 498485947.0, + "step": 20226 + }, + { + "epoch": 1.6679310629174569, + "grad_norm": 1.6643705368041992, + "learning_rate": 1e-06, + "loss": 0.5961, + "mean_token_accuracy": 0.8040866851806641, + "num_tokens": 498529147.0, + "step": 20227 + }, + { + "epoch": 1.6680135235425084, + "grad_norm": 1.9738057851791382, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7604965567588806, + "num_tokens": 498558050.0, + "step": 20228 + }, + { + "epoch": 1.66809598416756, + "grad_norm": 1.7675411701202393, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7476290464401245, + "num_tokens": 498593971.0, + "step": 20229 + }, + { + "epoch": 1.6681784447926116, + "grad_norm": 2.1825978755950928, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.762585461139679, + "num_tokens": 498618597.0, + "step": 20230 + }, + { + "epoch": 1.6682609054176631, + "grad_norm": 1.801363229751587, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.8059341907501221, + "num_tokens": 498648043.0, + "step": 20231 + }, + { + "epoch": 1.6683433660427145, + "grad_norm": 1.876627802848816, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7026435136795044, + "num_tokens": 498684707.0, + "step": 20232 + }, + { + "epoch": 1.6684258266677663, + "grad_norm": 2.041146755218506, + "learning_rate": 1e-06, + "loss": 0.6066, + "mean_token_accuracy": 0.7959550619125366, + "num_tokens": 498712716.0, + "step": 20233 + }, + { + "epoch": 1.6685082872928176, + "grad_norm": 1.924613356590271, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7653332352638245, + "num_tokens": 498746726.0, + "step": 20234 + }, + { + "epoch": 1.6685907479178692, + "grad_norm": 2.1801390647888184, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7240834832191467, + "num_tokens": 498773671.0, + "step": 20235 + }, + { + "epoch": 1.6686732085429208, + "grad_norm": 1.9719624519348145, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7666245698928833, + "num_tokens": 498802733.0, + "step": 20236 + }, + { + "epoch": 1.6687556691679721, + "grad_norm": 1.952783465385437, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7434489727020264, + "num_tokens": 498835339.0, + "step": 20237 + }, + { + "epoch": 1.668838129793024, + "grad_norm": 1.995608925819397, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7814409732818604, + "num_tokens": 498866518.0, + "step": 20238 + }, + { + "epoch": 1.6689205904180753, + "grad_norm": 2.189758777618408, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7247190475463867, + "num_tokens": 498891851.0, + "step": 20239 + }, + { + "epoch": 1.669003051043127, + "grad_norm": 1.7161802053451538, + "learning_rate": 1e-06, + "loss": 0.5952, + "mean_token_accuracy": 0.7969670295715332, + "num_tokens": 498927360.0, + "step": 20240 + }, + { + "epoch": 1.6690855116681784, + "grad_norm": 1.9943205118179321, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7684301137924194, + "num_tokens": 498959400.0, + "step": 20241 + }, + { + "epoch": 1.66916797229323, + "grad_norm": 1.9446909427642822, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.741617739200592, + "num_tokens": 498991238.0, + "step": 20242 + }, + { + "epoch": 1.6692504329182816, + "grad_norm": 1.8418210744857788, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.758234977722168, + "num_tokens": 499023752.0, + "step": 20243 + }, + { + "epoch": 1.669332893543333, + "grad_norm": 2.0742406845092773, + "learning_rate": 1e-06, + "loss": 0.5561, + "mean_token_accuracy": 0.807905912399292, + "num_tokens": 499049592.0, + "step": 20244 + }, + { + "epoch": 1.6694153541683847, + "grad_norm": 2.0590901374816895, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7408660650253296, + "num_tokens": 499079808.0, + "step": 20245 + }, + { + "epoch": 1.669497814793436, + "grad_norm": 1.984982967376709, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7209050059318542, + "num_tokens": 499109530.0, + "step": 20246 + }, + { + "epoch": 1.6695802754184876, + "grad_norm": 2.4383223056793213, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7340273857116699, + "num_tokens": 499132133.0, + "step": 20247 + }, + { + "epoch": 1.6696627360435392, + "grad_norm": 1.9204636812210083, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7551124691963196, + "num_tokens": 499163129.0, + "step": 20248 + }, + { + "epoch": 1.6697451966685908, + "grad_norm": 1.887734293937683, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7410592436790466, + "num_tokens": 499197934.0, + "step": 20249 + }, + { + "epoch": 1.6698276572936424, + "grad_norm": 2.148639440536499, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7425492405891418, + "num_tokens": 499224015.0, + "step": 20250 + }, + { + "epoch": 1.6699101179186937, + "grad_norm": 2.232708692550659, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7646639943122864, + "num_tokens": 499247826.0, + "step": 20251 + }, + { + "epoch": 1.6699925785437455, + "grad_norm": 2.1552679538726807, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7486522197723389, + "num_tokens": 499273481.0, + "step": 20252 + }, + { + "epoch": 1.6700750391687968, + "grad_norm": 1.9936169385910034, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7612519264221191, + "num_tokens": 499302339.0, + "step": 20253 + }, + { + "epoch": 1.6701574997938484, + "grad_norm": 2.0587375164031982, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7641027569770813, + "num_tokens": 499331687.0, + "step": 20254 + }, + { + "epoch": 1.6702399604189, + "grad_norm": 2.093106746673584, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7315337657928467, + "num_tokens": 499358164.0, + "step": 20255 + }, + { + "epoch": 1.6703224210439516, + "grad_norm": 1.9380645751953125, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.74000084400177, + "num_tokens": 499393174.0, + "step": 20256 + }, + { + "epoch": 1.6704048816690031, + "grad_norm": 1.9572035074234009, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7305619716644287, + "num_tokens": 499425679.0, + "step": 20257 + }, + { + "epoch": 1.6704873422940545, + "grad_norm": 2.372830390930176, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7744169235229492, + "num_tokens": 499449576.0, + "step": 20258 + }, + { + "epoch": 1.6705698029191063, + "grad_norm": 1.943629503250122, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7750084400177002, + "num_tokens": 499480885.0, + "step": 20259 + }, + { + "epoch": 1.6706522635441576, + "grad_norm": 1.6862674951553345, + "learning_rate": 1e-06, + "loss": 0.5779, + "mean_token_accuracy": 0.8151732087135315, + "num_tokens": 499517571.0, + "step": 20260 + }, + { + "epoch": 1.6707347241692092, + "grad_norm": 2.249774694442749, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7348347902297974, + "num_tokens": 499541704.0, + "step": 20261 + }, + { + "epoch": 1.6708171847942608, + "grad_norm": 2.2728374004364014, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.754822313785553, + "num_tokens": 499564835.0, + "step": 20262 + }, + { + "epoch": 1.6708996454193121, + "grad_norm": 1.860129714012146, + "learning_rate": 1e-06, + "loss": 0.5544, + "mean_token_accuracy": 0.8056765794754028, + "num_tokens": 499595724.0, + "step": 20263 + }, + { + "epoch": 1.670982106044364, + "grad_norm": 2.149329423904419, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7593590021133423, + "num_tokens": 499620363.0, + "step": 20264 + }, + { + "epoch": 1.6710645666694153, + "grad_norm": 2.014256477355957, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7513213157653809, + "num_tokens": 499649458.0, + "step": 20265 + }, + { + "epoch": 1.6711470272944668, + "grad_norm": 1.9037338495254517, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7682766914367676, + "num_tokens": 499681735.0, + "step": 20266 + }, + { + "epoch": 1.6712294879195184, + "grad_norm": 1.953650951385498, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.7881963849067688, + "num_tokens": 499714191.0, + "step": 20267 + }, + { + "epoch": 1.67131194854457, + "grad_norm": 2.1772327423095703, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7309988737106323, + "num_tokens": 499740390.0, + "step": 20268 + }, + { + "epoch": 1.6713944091696216, + "grad_norm": 1.9541621208190918, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7756589651107788, + "num_tokens": 499772230.0, + "step": 20269 + }, + { + "epoch": 1.671476869794673, + "grad_norm": 2.0445125102996826, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7857077121734619, + "num_tokens": 499800205.0, + "step": 20270 + }, + { + "epoch": 1.6715593304197247, + "grad_norm": 1.7745229005813599, + "learning_rate": 1e-06, + "loss": 0.6013, + "mean_token_accuracy": 0.7975956201553345, + "num_tokens": 499833022.0, + "step": 20271 + }, + { + "epoch": 1.671641791044776, + "grad_norm": 1.9498000144958496, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7586824893951416, + "num_tokens": 499867194.0, + "step": 20272 + }, + { + "epoch": 1.6717242516698276, + "grad_norm": 1.992207646369934, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7628085613250732, + "num_tokens": 499895059.0, + "step": 20273 + }, + { + "epoch": 1.6718067122948792, + "grad_norm": 1.767769455909729, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7528568506240845, + "num_tokens": 499935605.0, + "step": 20274 + }, + { + "epoch": 1.6718891729199308, + "grad_norm": 1.995069980621338, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7550033926963806, + "num_tokens": 499965706.0, + "step": 20275 + }, + { + "epoch": 1.6719716335449823, + "grad_norm": 2.0243756771087646, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.749427855014801, + "num_tokens": 499995394.0, + "step": 20276 + }, + { + "epoch": 1.6720540941700337, + "grad_norm": 2.1910786628723145, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7422381639480591, + "num_tokens": 500021431.0, + "step": 20277 + }, + { + "epoch": 1.6721365547950855, + "grad_norm": 2.1041886806488037, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7767309546470642, + "num_tokens": 500046781.0, + "step": 20278 + }, + { + "epoch": 1.6722190154201368, + "grad_norm": 1.9231468439102173, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.782089114189148, + "num_tokens": 500079595.0, + "step": 20279 + }, + { + "epoch": 1.6723014760451884, + "grad_norm": 1.7492053508758545, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7748631834983826, + "num_tokens": 500116107.0, + "step": 20280 + }, + { + "epoch": 1.67238393667024, + "grad_norm": 1.9939919710159302, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.74997478723526, + "num_tokens": 500145350.0, + "step": 20281 + }, + { + "epoch": 1.6724663972952913, + "grad_norm": 1.945348858833313, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7472431659698486, + "num_tokens": 500174534.0, + "step": 20282 + }, + { + "epoch": 1.6725488579203431, + "grad_norm": 1.9188265800476074, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7381868362426758, + "num_tokens": 500207401.0, + "step": 20283 + }, + { + "epoch": 1.6726313185453945, + "grad_norm": 2.1836695671081543, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7439810037612915, + "num_tokens": 500235479.0, + "step": 20284 + }, + { + "epoch": 1.672713779170446, + "grad_norm": 2.0223567485809326, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7734535932540894, + "num_tokens": 500261345.0, + "step": 20285 + }, + { + "epoch": 1.6727962397954976, + "grad_norm": 1.9414026737213135, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7439913749694824, + "num_tokens": 500294964.0, + "step": 20286 + }, + { + "epoch": 1.6728787004205492, + "grad_norm": 1.9354712963104248, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7549120783805847, + "num_tokens": 500326993.0, + "step": 20287 + }, + { + "epoch": 1.6729611610456008, + "grad_norm": 1.907064437866211, + "learning_rate": 1e-06, + "loss": 0.7704, + "mean_token_accuracy": 0.7505509853363037, + "num_tokens": 500361800.0, + "step": 20288 + }, + { + "epoch": 1.6730436216706521, + "grad_norm": 1.7113194465637207, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7563794851303101, + "num_tokens": 500400328.0, + "step": 20289 + }, + { + "epoch": 1.673126082295704, + "grad_norm": 1.9946595430374146, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7199752330780029, + "num_tokens": 500431860.0, + "step": 20290 + }, + { + "epoch": 1.6732085429207553, + "grad_norm": 1.7153977155685425, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7853128910064697, + "num_tokens": 500467725.0, + "step": 20291 + }, + { + "epoch": 1.6732910035458068, + "grad_norm": 1.8822122812271118, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7653535604476929, + "num_tokens": 500500128.0, + "step": 20292 + }, + { + "epoch": 1.6733734641708584, + "grad_norm": 1.886852502822876, + "learning_rate": 1e-06, + "loss": 0.5588, + "mean_token_accuracy": 0.8057633638381958, + "num_tokens": 500528928.0, + "step": 20293 + }, + { + "epoch": 1.67345592479591, + "grad_norm": 1.925316333770752, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7837048172950745, + "num_tokens": 500560354.0, + "step": 20294 + }, + { + "epoch": 1.6735383854209616, + "grad_norm": 1.7958964109420776, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7898153066635132, + "num_tokens": 500594338.0, + "step": 20295 + }, + { + "epoch": 1.673620846046013, + "grad_norm": 2.050658702850342, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7830836772918701, + "num_tokens": 500624178.0, + "step": 20296 + }, + { + "epoch": 1.6737033066710647, + "grad_norm": 1.8737812042236328, + "learning_rate": 1e-06, + "loss": 0.5685, + "mean_token_accuracy": 0.81221604347229, + "num_tokens": 500652982.0, + "step": 20297 + }, + { + "epoch": 1.673785767296116, + "grad_norm": 2.0451855659484863, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.7877733707427979, + "num_tokens": 500680756.0, + "step": 20298 + }, + { + "epoch": 1.6738682279211676, + "grad_norm": 1.967761754989624, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.760016143321991, + "num_tokens": 500709059.0, + "step": 20299 + }, + { + "epoch": 1.6739506885462192, + "grad_norm": 2.262192964553833, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.744434118270874, + "num_tokens": 500732615.0, + "step": 20300 + }, + { + "epoch": 1.6740331491712708, + "grad_norm": 1.9640552997589111, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7779476642608643, + "num_tokens": 500762072.0, + "step": 20301 + }, + { + "epoch": 1.6741156097963223, + "grad_norm": 1.9293497800827026, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.7811028957366943, + "num_tokens": 500792503.0, + "step": 20302 + }, + { + "epoch": 1.6741980704213737, + "grad_norm": 1.769431233406067, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7834948301315308, + "num_tokens": 500829629.0, + "step": 20303 + }, + { + "epoch": 1.6742805310464255, + "grad_norm": 1.9274686574935913, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7842519283294678, + "num_tokens": 500858987.0, + "step": 20304 + }, + { + "epoch": 1.6743629916714768, + "grad_norm": 2.020491123199463, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7325981855392456, + "num_tokens": 500891057.0, + "step": 20305 + }, + { + "epoch": 1.6744454522965284, + "grad_norm": 1.8591487407684326, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.783842921257019, + "num_tokens": 500923418.0, + "step": 20306 + }, + { + "epoch": 1.67452791292158, + "grad_norm": 2.3253321647644043, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7491454482078552, + "num_tokens": 500946380.0, + "step": 20307 + }, + { + "epoch": 1.6746103735466313, + "grad_norm": 2.0977237224578857, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.7827816009521484, + "num_tokens": 500973643.0, + "step": 20308 + }, + { + "epoch": 1.6746928341716831, + "grad_norm": 1.9089069366455078, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7573628425598145, + "num_tokens": 501004860.0, + "step": 20309 + }, + { + "epoch": 1.6747752947967345, + "grad_norm": 1.774370551109314, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.7916910648345947, + "num_tokens": 501035356.0, + "step": 20310 + }, + { + "epoch": 1.674857755421786, + "grad_norm": 1.7151727676391602, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.7833684682846069, + "num_tokens": 501070526.0, + "step": 20311 + }, + { + "epoch": 1.6749402160468376, + "grad_norm": 2.332906723022461, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7680261731147766, + "num_tokens": 501092624.0, + "step": 20312 + }, + { + "epoch": 1.6750226766718892, + "grad_norm": 1.8843709230422974, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7352243065834045, + "num_tokens": 501125831.0, + "step": 20313 + }, + { + "epoch": 1.6751051372969408, + "grad_norm": 2.223961591720581, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7592891454696655, + "num_tokens": 501150940.0, + "step": 20314 + }, + { + "epoch": 1.6751875979219921, + "grad_norm": 2.066577911376953, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7566110491752625, + "num_tokens": 501176654.0, + "step": 20315 + }, + { + "epoch": 1.675270058547044, + "grad_norm": 2.0801358222961426, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7675503492355347, + "num_tokens": 501205119.0, + "step": 20316 + }, + { + "epoch": 1.6753525191720953, + "grad_norm": 2.091356039047241, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.784125566482544, + "num_tokens": 501230673.0, + "step": 20317 + }, + { + "epoch": 1.6754349797971468, + "grad_norm": 2.0044138431549072, + "learning_rate": 1e-06, + "loss": 0.7988, + "mean_token_accuracy": 0.744972288608551, + "num_tokens": 501259909.0, + "step": 20318 + }, + { + "epoch": 1.6755174404221984, + "grad_norm": 1.799687385559082, + "learning_rate": 1e-06, + "loss": 0.5532, + "mean_token_accuracy": 0.8067353963851929, + "num_tokens": 501292584.0, + "step": 20319 + }, + { + "epoch": 1.67559990104725, + "grad_norm": 1.7773525714874268, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7697755098342896, + "num_tokens": 501326616.0, + "step": 20320 + }, + { + "epoch": 1.6756823616723016, + "grad_norm": 2.00059175491333, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7500976324081421, + "num_tokens": 501357952.0, + "step": 20321 + }, + { + "epoch": 1.675764822297353, + "grad_norm": 1.7065684795379639, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.761283278465271, + "num_tokens": 501394154.0, + "step": 20322 + }, + { + "epoch": 1.6758472829224047, + "grad_norm": 2.2088353633880615, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7582215070724487, + "num_tokens": 501419919.0, + "step": 20323 + }, + { + "epoch": 1.675929743547456, + "grad_norm": 1.9861661195755005, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7509642243385315, + "num_tokens": 501451733.0, + "step": 20324 + }, + { + "epoch": 1.6760122041725076, + "grad_norm": 2.178985834121704, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.745253324508667, + "num_tokens": 501477749.0, + "step": 20325 + }, + { + "epoch": 1.6760946647975592, + "grad_norm": 1.713311791419983, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7461751699447632, + "num_tokens": 501512005.0, + "step": 20326 + }, + { + "epoch": 1.6761771254226105, + "grad_norm": 2.018998384475708, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7640235424041748, + "num_tokens": 501538950.0, + "step": 20327 + }, + { + "epoch": 1.6762595860476623, + "grad_norm": 2.055624008178711, + "learning_rate": 1e-06, + "loss": 0.5591, + "mean_token_accuracy": 0.8093535304069519, + "num_tokens": 501567830.0, + "step": 20328 + }, + { + "epoch": 1.6763420466727137, + "grad_norm": 2.260392904281616, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7616649270057678, + "num_tokens": 501591229.0, + "step": 20329 + }, + { + "epoch": 1.6764245072977653, + "grad_norm": 1.7227058410644531, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7693178057670593, + "num_tokens": 501627055.0, + "step": 20330 + }, + { + "epoch": 1.6765069679228168, + "grad_norm": 2.177870273590088, + "learning_rate": 1e-06, + "loss": 0.6059, + "mean_token_accuracy": 0.7992767095565796, + "num_tokens": 501652637.0, + "step": 20331 + }, + { + "epoch": 1.6765894285478684, + "grad_norm": 1.9197807312011719, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7468522787094116, + "num_tokens": 501684243.0, + "step": 20332 + }, + { + "epoch": 1.67667188917292, + "grad_norm": 1.936841368675232, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7612717151641846, + "num_tokens": 501714053.0, + "step": 20333 + }, + { + "epoch": 1.6767543497979713, + "grad_norm": 1.9577924013137817, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7589889764785767, + "num_tokens": 501743104.0, + "step": 20334 + }, + { + "epoch": 1.6768368104230231, + "grad_norm": 2.169023275375366, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7646899819374084, + "num_tokens": 501767347.0, + "step": 20335 + }, + { + "epoch": 1.6769192710480745, + "grad_norm": 2.039179801940918, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7382633090019226, + "num_tokens": 501799660.0, + "step": 20336 + }, + { + "epoch": 1.677001731673126, + "grad_norm": 1.816577434539795, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7653040885925293, + "num_tokens": 501834894.0, + "step": 20337 + }, + { + "epoch": 1.6770841922981776, + "grad_norm": 2.3705058097839355, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7499678134918213, + "num_tokens": 501857084.0, + "step": 20338 + }, + { + "epoch": 1.6771666529232292, + "grad_norm": 1.927661657333374, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.781671404838562, + "num_tokens": 501886167.0, + "step": 20339 + }, + { + "epoch": 1.6772491135482808, + "grad_norm": 2.058393716812134, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7562915086746216, + "num_tokens": 501913464.0, + "step": 20340 + }, + { + "epoch": 1.6773315741733321, + "grad_norm": 2.0267255306243896, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7468437552452087, + "num_tokens": 501942757.0, + "step": 20341 + }, + { + "epoch": 1.677414034798384, + "grad_norm": 2.4744017124176025, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7594484090805054, + "num_tokens": 501964460.0, + "step": 20342 + }, + { + "epoch": 1.6774964954234353, + "grad_norm": 1.8433361053466797, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7441756725311279, + "num_tokens": 501999402.0, + "step": 20343 + }, + { + "epoch": 1.6775789560484868, + "grad_norm": 1.7257506847381592, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7354938983917236, + "num_tokens": 502037610.0, + "step": 20344 + }, + { + "epoch": 1.6776614166735384, + "grad_norm": 2.1224191188812256, + "learning_rate": 1e-06, + "loss": 0.5991, + "mean_token_accuracy": 0.7860089540481567, + "num_tokens": 502063429.0, + "step": 20345 + }, + { + "epoch": 1.6777438772985898, + "grad_norm": 1.8201879262924194, + "learning_rate": 1e-06, + "loss": 0.6044, + "mean_token_accuracy": 0.7911062240600586, + "num_tokens": 502095986.0, + "step": 20346 + }, + { + "epoch": 1.6778263379236416, + "grad_norm": 2.4197611808776855, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.756929337978363, + "num_tokens": 502116290.0, + "step": 20347 + }, + { + "epoch": 1.677908798548693, + "grad_norm": 2.2825584411621094, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7278357744216919, + "num_tokens": 502140584.0, + "step": 20348 + }, + { + "epoch": 1.6779912591737447, + "grad_norm": 2.1099326610565186, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.764843761920929, + "num_tokens": 502168580.0, + "step": 20349 + }, + { + "epoch": 1.678073719798796, + "grad_norm": 2.4699792861938477, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7414562106132507, + "num_tokens": 502190166.0, + "step": 20350 + }, + { + "epoch": 1.6781561804238476, + "grad_norm": 1.7761951684951782, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7621676921844482, + "num_tokens": 502229541.0, + "step": 20351 + }, + { + "epoch": 1.6782386410488992, + "grad_norm": 1.7812176942825317, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7719734907150269, + "num_tokens": 502266930.0, + "step": 20352 + }, + { + "epoch": 1.6783211016739505, + "grad_norm": 2.16326642036438, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.709797203540802, + "num_tokens": 502296661.0, + "step": 20353 + }, + { + "epoch": 1.6784035622990023, + "grad_norm": 1.9040969610214233, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.77013099193573, + "num_tokens": 502328635.0, + "step": 20354 + }, + { + "epoch": 1.6784860229240537, + "grad_norm": 2.203033924102783, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.753974437713623, + "num_tokens": 502352450.0, + "step": 20355 + }, + { + "epoch": 1.6785684835491053, + "grad_norm": 1.9898148775100708, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7175253033638, + "num_tokens": 502383951.0, + "step": 20356 + }, + { + "epoch": 1.6786509441741568, + "grad_norm": 1.8751832246780396, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7102732062339783, + "num_tokens": 502419556.0, + "step": 20357 + }, + { + "epoch": 1.6787334047992084, + "grad_norm": 1.9204051494598389, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7378113269805908, + "num_tokens": 502452425.0, + "step": 20358 + }, + { + "epoch": 1.67881586542426, + "grad_norm": 1.8340319395065308, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7588297724723816, + "num_tokens": 502488643.0, + "step": 20359 + }, + { + "epoch": 1.6788983260493113, + "grad_norm": 2.3448164463043213, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7496367692947388, + "num_tokens": 502513253.0, + "step": 20360 + }, + { + "epoch": 1.6789807866743631, + "grad_norm": 1.7213993072509766, + "learning_rate": 1e-06, + "loss": 0.6005, + "mean_token_accuracy": 0.8029725551605225, + "num_tokens": 502548955.0, + "step": 20361 + }, + { + "epoch": 1.6790632472994145, + "grad_norm": 1.8329675197601318, + "learning_rate": 1e-06, + "loss": 0.7821, + "mean_token_accuracy": 0.7607502937316895, + "num_tokens": 502583723.0, + "step": 20362 + }, + { + "epoch": 1.679145707924466, + "grad_norm": 1.84669828414917, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.773758053779602, + "num_tokens": 502615482.0, + "step": 20363 + }, + { + "epoch": 1.6792281685495176, + "grad_norm": 2.0508205890655518, + "learning_rate": 1e-06, + "loss": 0.6219, + "mean_token_accuracy": 0.8019843101501465, + "num_tokens": 502642653.0, + "step": 20364 + }, + { + "epoch": 1.6793106291745692, + "grad_norm": 1.9116268157958984, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7876899242401123, + "num_tokens": 502670940.0, + "step": 20365 + }, + { + "epoch": 1.6793930897996208, + "grad_norm": 1.9247664213180542, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7386494278907776, + "num_tokens": 502703895.0, + "step": 20366 + }, + { + "epoch": 1.6794755504246721, + "grad_norm": 2.1868484020233154, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7738474607467651, + "num_tokens": 502729337.0, + "step": 20367 + }, + { + "epoch": 1.679558011049724, + "grad_norm": 1.8541291952133179, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.783868670463562, + "num_tokens": 502763525.0, + "step": 20368 + }, + { + "epoch": 1.6796404716747753, + "grad_norm": 2.0557069778442383, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7491002082824707, + "num_tokens": 502790025.0, + "step": 20369 + }, + { + "epoch": 1.6797229322998268, + "grad_norm": 1.6939575672149658, + "learning_rate": 1e-06, + "loss": 0.5772, + "mean_token_accuracy": 0.8077217936515808, + "num_tokens": 502826528.0, + "step": 20370 + }, + { + "epoch": 1.6798053929248784, + "grad_norm": 2.117731809616089, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7681387066841125, + "num_tokens": 502854698.0, + "step": 20371 + }, + { + "epoch": 1.6798878535499298, + "grad_norm": 1.9339886903762817, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.7880816459655762, + "num_tokens": 502885014.0, + "step": 20372 + }, + { + "epoch": 1.6799703141749815, + "grad_norm": 1.7428817749023438, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7575435638427734, + "num_tokens": 502924300.0, + "step": 20373 + }, + { + "epoch": 1.680052774800033, + "grad_norm": 2.0614662170410156, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7325460910797119, + "num_tokens": 502952446.0, + "step": 20374 + }, + { + "epoch": 1.6801352354250845, + "grad_norm": 1.9897780418395996, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7774596214294434, + "num_tokens": 502978856.0, + "step": 20375 + }, + { + "epoch": 1.680217696050136, + "grad_norm": 1.901645302772522, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7682961225509644, + "num_tokens": 503010490.0, + "step": 20376 + }, + { + "epoch": 1.6803001566751876, + "grad_norm": 1.7601243257522583, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7694677710533142, + "num_tokens": 503048014.0, + "step": 20377 + }, + { + "epoch": 1.6803826173002392, + "grad_norm": 1.955435872077942, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7759899497032166, + "num_tokens": 503078776.0, + "step": 20378 + }, + { + "epoch": 1.6804650779252905, + "grad_norm": 1.9646438360214233, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.7452765703201294, + "num_tokens": 503109727.0, + "step": 20379 + }, + { + "epoch": 1.6805475385503423, + "grad_norm": 1.959983229637146, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7616285085678101, + "num_tokens": 503139717.0, + "step": 20380 + }, + { + "epoch": 1.6806299991753937, + "grad_norm": 1.898136854171753, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7571039795875549, + "num_tokens": 503172214.0, + "step": 20381 + }, + { + "epoch": 1.6807124598004453, + "grad_norm": 1.9584144353866577, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7541273832321167, + "num_tokens": 503203141.0, + "step": 20382 + }, + { + "epoch": 1.6807949204254968, + "grad_norm": 2.201612949371338, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7399982213973999, + "num_tokens": 503230233.0, + "step": 20383 + }, + { + "epoch": 1.6808773810505484, + "grad_norm": 1.8581128120422363, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7510726451873779, + "num_tokens": 503262500.0, + "step": 20384 + }, + { + "epoch": 1.6809598416756, + "grad_norm": 2.6002423763275146, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7394810914993286, + "num_tokens": 503282888.0, + "step": 20385 + }, + { + "epoch": 1.6810423023006513, + "grad_norm": 2.074836254119873, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7678726315498352, + "num_tokens": 503310104.0, + "step": 20386 + }, + { + "epoch": 1.6811247629257031, + "grad_norm": 2.189605951309204, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7229576706886292, + "num_tokens": 503341313.0, + "step": 20387 + }, + { + "epoch": 1.6812072235507545, + "grad_norm": 1.9949665069580078, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7778512835502625, + "num_tokens": 503369370.0, + "step": 20388 + }, + { + "epoch": 1.681289684175806, + "grad_norm": 2.0390686988830566, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7540568709373474, + "num_tokens": 503399635.0, + "step": 20389 + }, + { + "epoch": 1.6813721448008576, + "grad_norm": 1.8125792741775513, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.7551114559173584, + "num_tokens": 503435798.0, + "step": 20390 + }, + { + "epoch": 1.681454605425909, + "grad_norm": 2.048391103744507, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7524269819259644, + "num_tokens": 503464123.0, + "step": 20391 + }, + { + "epoch": 1.6815370660509608, + "grad_norm": 1.8463616371154785, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7597285509109497, + "num_tokens": 503499105.0, + "step": 20392 + }, + { + "epoch": 1.681619526676012, + "grad_norm": 2.0098843574523926, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.7950469255447388, + "num_tokens": 503529248.0, + "step": 20393 + }, + { + "epoch": 1.681701987301064, + "grad_norm": 1.8315702676773071, + "learning_rate": 1e-06, + "loss": 0.5903, + "mean_token_accuracy": 0.7950013875961304, + "num_tokens": 503563104.0, + "step": 20394 + }, + { + "epoch": 1.6817844479261153, + "grad_norm": 2.0056350231170654, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7168447971343994, + "num_tokens": 503597525.0, + "step": 20395 + }, + { + "epoch": 1.6818669085511668, + "grad_norm": 2.1565968990325928, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7616514563560486, + "num_tokens": 503625386.0, + "step": 20396 + }, + { + "epoch": 1.6819493691762184, + "grad_norm": 2.0066239833831787, + "learning_rate": 1e-06, + "loss": 0.5751, + "mean_token_accuracy": 0.8102592825889587, + "num_tokens": 503652931.0, + "step": 20397 + }, + { + "epoch": 1.6820318298012698, + "grad_norm": 2.148409366607666, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7504758238792419, + "num_tokens": 503678836.0, + "step": 20398 + }, + { + "epoch": 1.6821142904263215, + "grad_norm": 1.8241608142852783, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.790461540222168, + "num_tokens": 503710368.0, + "step": 20399 + }, + { + "epoch": 1.682196751051373, + "grad_norm": 1.7657428979873657, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.8080188035964966, + "num_tokens": 503744085.0, + "step": 20400 + }, + { + "epoch": 1.6822792116764245, + "grad_norm": 1.9458000659942627, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.7875492572784424, + "num_tokens": 503776242.0, + "step": 20401 + }, + { + "epoch": 1.682361672301476, + "grad_norm": 2.0585756301879883, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7760942578315735, + "num_tokens": 503802182.0, + "step": 20402 + }, + { + "epoch": 1.6824441329265276, + "grad_norm": 1.826927661895752, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7397953271865845, + "num_tokens": 503834547.0, + "step": 20403 + }, + { + "epoch": 1.6825265935515792, + "grad_norm": 1.8554048538208008, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7403229475021362, + "num_tokens": 503868976.0, + "step": 20404 + }, + { + "epoch": 1.6826090541766305, + "grad_norm": 1.6556811332702637, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7703607678413391, + "num_tokens": 503906520.0, + "step": 20405 + }, + { + "epoch": 1.6826915148016823, + "grad_norm": 1.6601544618606567, + "learning_rate": 1e-06, + "loss": 0.5372, + "mean_token_accuracy": 0.8230902552604675, + "num_tokens": 503943538.0, + "step": 20406 + }, + { + "epoch": 1.6827739754267337, + "grad_norm": 2.114246129989624, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7515950798988342, + "num_tokens": 503973297.0, + "step": 20407 + }, + { + "epoch": 1.6828564360517853, + "grad_norm": 1.796517014503479, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.7893353700637817, + "num_tokens": 504002867.0, + "step": 20408 + }, + { + "epoch": 1.6829388966768368, + "grad_norm": 1.8076192140579224, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.776638925075531, + "num_tokens": 504033936.0, + "step": 20409 + }, + { + "epoch": 1.6830213573018884, + "grad_norm": 1.9106923341751099, + "learning_rate": 1e-06, + "loss": 0.6039, + "mean_token_accuracy": 0.799441397190094, + "num_tokens": 504064972.0, + "step": 20410 + }, + { + "epoch": 1.68310381792694, + "grad_norm": 1.8014665842056274, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7590673565864563, + "num_tokens": 504102987.0, + "step": 20411 + }, + { + "epoch": 1.6831862785519913, + "grad_norm": 1.8969335556030273, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.789869487285614, + "num_tokens": 504130905.0, + "step": 20412 + }, + { + "epoch": 1.6832687391770431, + "grad_norm": 1.8156237602233887, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7837144136428833, + "num_tokens": 504164939.0, + "step": 20413 + }, + { + "epoch": 1.6833511998020945, + "grad_norm": 2.2704055309295654, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7425510883331299, + "num_tokens": 504188790.0, + "step": 20414 + }, + { + "epoch": 1.683433660427146, + "grad_norm": 2.1926801204681396, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7551185488700867, + "num_tokens": 504213011.0, + "step": 20415 + }, + { + "epoch": 1.6835161210521976, + "grad_norm": 2.0942554473876953, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7470322847366333, + "num_tokens": 504241313.0, + "step": 20416 + }, + { + "epoch": 1.683598581677249, + "grad_norm": 2.148388624191284, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7414143681526184, + "num_tokens": 504269997.0, + "step": 20417 + }, + { + "epoch": 1.6836810423023008, + "grad_norm": 1.8682702779769897, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7793900966644287, + "num_tokens": 504302454.0, + "step": 20418 + }, + { + "epoch": 1.683763502927352, + "grad_norm": 2.0686535835266113, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7760359644889832, + "num_tokens": 504327699.0, + "step": 20419 + }, + { + "epoch": 1.6838459635524037, + "grad_norm": 1.8778979778289795, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7862815856933594, + "num_tokens": 504358578.0, + "step": 20420 + }, + { + "epoch": 1.6839284241774553, + "grad_norm": 2.2076878547668457, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7709949612617493, + "num_tokens": 504385598.0, + "step": 20421 + }, + { + "epoch": 1.6840108848025068, + "grad_norm": 2.0206239223480225, + "learning_rate": 1e-06, + "loss": 0.6086, + "mean_token_accuracy": 0.7893630266189575, + "num_tokens": 504413003.0, + "step": 20422 + }, + { + "epoch": 1.6840933454275584, + "grad_norm": 2.324789524078369, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7602264285087585, + "num_tokens": 504437043.0, + "step": 20423 + }, + { + "epoch": 1.6841758060526097, + "grad_norm": 2.265489339828491, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7509499788284302, + "num_tokens": 504461577.0, + "step": 20424 + }, + { + "epoch": 1.6842582666776615, + "grad_norm": 2.0480833053588867, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7304556965827942, + "num_tokens": 504489465.0, + "step": 20425 + }, + { + "epoch": 1.684340727302713, + "grad_norm": 1.8416755199432373, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7718532085418701, + "num_tokens": 504521862.0, + "step": 20426 + }, + { + "epoch": 1.6844231879277645, + "grad_norm": 2.1345202922821045, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7429084777832031, + "num_tokens": 504552007.0, + "step": 20427 + }, + { + "epoch": 1.684505648552816, + "grad_norm": 1.8961166143417358, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7526659369468689, + "num_tokens": 504583716.0, + "step": 20428 + }, + { + "epoch": 1.6845881091778676, + "grad_norm": 1.9662290811538696, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.7862344980239868, + "num_tokens": 504614319.0, + "step": 20429 + }, + { + "epoch": 1.6846705698029192, + "grad_norm": 1.8910486698150635, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7808434963226318, + "num_tokens": 504647841.0, + "step": 20430 + }, + { + "epoch": 1.6847530304279705, + "grad_norm": 1.9270119667053223, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7781389951705933, + "num_tokens": 504677910.0, + "step": 20431 + }, + { + "epoch": 1.6848354910530223, + "grad_norm": 2.1627063751220703, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.7504943609237671, + "num_tokens": 504707861.0, + "step": 20432 + }, + { + "epoch": 1.6849179516780737, + "grad_norm": 2.0110833644866943, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7620944976806641, + "num_tokens": 504736686.0, + "step": 20433 + }, + { + "epoch": 1.6850004123031253, + "grad_norm": 2.0699868202209473, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7540349364280701, + "num_tokens": 504764451.0, + "step": 20434 + }, + { + "epoch": 1.6850828729281768, + "grad_norm": 2.1610326766967773, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.7951676845550537, + "num_tokens": 504791240.0, + "step": 20435 + }, + { + "epoch": 1.6851653335532282, + "grad_norm": 1.8313653469085693, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7837169766426086, + "num_tokens": 504823134.0, + "step": 20436 + }, + { + "epoch": 1.68524779417828, + "grad_norm": 1.9603252410888672, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7648935317993164, + "num_tokens": 504852883.0, + "step": 20437 + }, + { + "epoch": 1.6853302548033313, + "grad_norm": 2.0438952445983887, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.7855161428451538, + "num_tokens": 504878166.0, + "step": 20438 + }, + { + "epoch": 1.6854127154283831, + "grad_norm": 1.7936571836471558, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.7836344242095947, + "num_tokens": 504911752.0, + "step": 20439 + }, + { + "epoch": 1.6854951760534345, + "grad_norm": 1.775645136833191, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7503427863121033, + "num_tokens": 504949837.0, + "step": 20440 + }, + { + "epoch": 1.685577636678486, + "grad_norm": 1.9598935842514038, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.7855911254882812, + "num_tokens": 504978698.0, + "step": 20441 + }, + { + "epoch": 1.6856600973035376, + "grad_norm": 1.7535314559936523, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.779301106929779, + "num_tokens": 505012064.0, + "step": 20442 + }, + { + "epoch": 1.685742557928589, + "grad_norm": 2.2039105892181396, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.7935777902603149, + "num_tokens": 505034778.0, + "step": 20443 + }, + { + "epoch": 1.6858250185536408, + "grad_norm": 1.8648158311843872, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7220432758331299, + "num_tokens": 505067971.0, + "step": 20444 + }, + { + "epoch": 1.685907479178692, + "grad_norm": 1.9175294637680054, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7547528147697449, + "num_tokens": 505100353.0, + "step": 20445 + }, + { + "epoch": 1.6859899398037437, + "grad_norm": 1.9148842096328735, + "learning_rate": 1e-06, + "loss": 0.6214, + "mean_token_accuracy": 0.79280024766922, + "num_tokens": 505130082.0, + "step": 20446 + }, + { + "epoch": 1.6860724004287952, + "grad_norm": 2.0727930068969727, + "learning_rate": 1e-06, + "loss": 0.5815, + "mean_token_accuracy": 0.8012539148330688, + "num_tokens": 505156280.0, + "step": 20447 + }, + { + "epoch": 1.6861548610538468, + "grad_norm": 1.6743438243865967, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.7847328186035156, + "num_tokens": 505197468.0, + "step": 20448 + }, + { + "epoch": 1.6862373216788984, + "grad_norm": 2.0121572017669678, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7439730167388916, + "num_tokens": 505227091.0, + "step": 20449 + }, + { + "epoch": 1.6863197823039497, + "grad_norm": 1.8464797735214233, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7804788947105408, + "num_tokens": 505260181.0, + "step": 20450 + }, + { + "epoch": 1.6864022429290015, + "grad_norm": 1.8432962894439697, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7713732719421387, + "num_tokens": 505292333.0, + "step": 20451 + }, + { + "epoch": 1.6864847035540529, + "grad_norm": 1.9327797889709473, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7755297422409058, + "num_tokens": 505320983.0, + "step": 20452 + }, + { + "epoch": 1.6865671641791045, + "grad_norm": 2.0687522888183594, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.766320526599884, + "num_tokens": 505349677.0, + "step": 20453 + }, + { + "epoch": 1.686649624804156, + "grad_norm": 1.9364619255065918, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7570139169692993, + "num_tokens": 505381626.0, + "step": 20454 + }, + { + "epoch": 1.6867320854292076, + "grad_norm": 1.937638282775879, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.7711108922958374, + "num_tokens": 505410565.0, + "step": 20455 + }, + { + "epoch": 1.6868145460542592, + "grad_norm": 2.0132553577423096, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7491698265075684, + "num_tokens": 505439312.0, + "step": 20456 + }, + { + "epoch": 1.6868970066793105, + "grad_norm": 2.189718008041382, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.734288215637207, + "num_tokens": 505466365.0, + "step": 20457 + }, + { + "epoch": 1.6869794673043623, + "grad_norm": 1.9653023481369019, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.7930119037628174, + "num_tokens": 505493937.0, + "step": 20458 + }, + { + "epoch": 1.6870619279294137, + "grad_norm": 1.7241182327270508, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.7942044734954834, + "num_tokens": 505529742.0, + "step": 20459 + }, + { + "epoch": 1.6871443885544652, + "grad_norm": 2.0926320552825928, + "learning_rate": 1e-06, + "loss": 0.569, + "mean_token_accuracy": 0.7949328422546387, + "num_tokens": 505554895.0, + "step": 20460 + }, + { + "epoch": 1.6872268491795168, + "grad_norm": 1.8059189319610596, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.763589084148407, + "num_tokens": 505591114.0, + "step": 20461 + }, + { + "epoch": 1.6873093098045682, + "grad_norm": 1.8593887090682983, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7791152000427246, + "num_tokens": 505622419.0, + "step": 20462 + }, + { + "epoch": 1.68739177042962, + "grad_norm": 1.9199364185333252, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7914080619812012, + "num_tokens": 505655889.0, + "step": 20463 + }, + { + "epoch": 1.6874742310546713, + "grad_norm": 2.0546717643737793, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7539498805999756, + "num_tokens": 505687279.0, + "step": 20464 + }, + { + "epoch": 1.6875566916797229, + "grad_norm": 2.0781772136688232, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.768276035785675, + "num_tokens": 505719014.0, + "step": 20465 + }, + { + "epoch": 1.6876391523047745, + "grad_norm": 1.8387795686721802, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7759366631507874, + "num_tokens": 505750384.0, + "step": 20466 + }, + { + "epoch": 1.687721612929826, + "grad_norm": 1.917744517326355, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7775697112083435, + "num_tokens": 505781745.0, + "step": 20467 + }, + { + "epoch": 1.6878040735548776, + "grad_norm": 2.2005481719970703, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7343735694885254, + "num_tokens": 505808471.0, + "step": 20468 + }, + { + "epoch": 1.687886534179929, + "grad_norm": 1.925348162651062, + "learning_rate": 1e-06, + "loss": 0.5567, + "mean_token_accuracy": 0.8012962341308594, + "num_tokens": 505838304.0, + "step": 20469 + }, + { + "epoch": 1.6879689948049807, + "grad_norm": 2.182454824447632, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7721871137619019, + "num_tokens": 505862980.0, + "step": 20470 + }, + { + "epoch": 1.688051455430032, + "grad_norm": 2.1017024517059326, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.777137815952301, + "num_tokens": 505890931.0, + "step": 20471 + }, + { + "epoch": 1.6881339160550837, + "grad_norm": 1.9710593223571777, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7728243470191956, + "num_tokens": 505918615.0, + "step": 20472 + }, + { + "epoch": 1.6882163766801352, + "grad_norm": 1.9928406476974487, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7499262690544128, + "num_tokens": 505947615.0, + "step": 20473 + }, + { + "epoch": 1.6882988373051868, + "grad_norm": 1.7553730010986328, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.7929338812828064, + "num_tokens": 505984277.0, + "step": 20474 + }, + { + "epoch": 1.6883812979302384, + "grad_norm": 1.917696475982666, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7646558880805969, + "num_tokens": 506014952.0, + "step": 20475 + }, + { + "epoch": 1.6884637585552897, + "grad_norm": 1.9585661888122559, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7660624384880066, + "num_tokens": 506046206.0, + "step": 20476 + }, + { + "epoch": 1.6885462191803415, + "grad_norm": 2.0265519618988037, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7325477600097656, + "num_tokens": 506076171.0, + "step": 20477 + }, + { + "epoch": 1.6886286798053929, + "grad_norm": 2.0479440689086914, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7779532074928284, + "num_tokens": 506103669.0, + "step": 20478 + }, + { + "epoch": 1.6887111404304445, + "grad_norm": 1.9945323467254639, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7261966466903687, + "num_tokens": 506134104.0, + "step": 20479 + }, + { + "epoch": 1.688793601055496, + "grad_norm": 2.0454697608947754, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7539047002792358, + "num_tokens": 506161452.0, + "step": 20480 + }, + { + "epoch": 1.6888760616805474, + "grad_norm": 2.114583969116211, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7514040470123291, + "num_tokens": 506190845.0, + "step": 20481 + }, + { + "epoch": 1.6889585223055992, + "grad_norm": 1.9846891164779663, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7604526281356812, + "num_tokens": 506220173.0, + "step": 20482 + }, + { + "epoch": 1.6890409829306505, + "grad_norm": 1.7571959495544434, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7588827610015869, + "num_tokens": 506258129.0, + "step": 20483 + }, + { + "epoch": 1.689123443555702, + "grad_norm": 1.8058897256851196, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.78233802318573, + "num_tokens": 506291946.0, + "step": 20484 + }, + { + "epoch": 1.6892059041807537, + "grad_norm": 2.184386968612671, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.7844370603561401, + "num_tokens": 506319295.0, + "step": 20485 + }, + { + "epoch": 1.6892883648058052, + "grad_norm": 1.9357010126113892, + "learning_rate": 1e-06, + "loss": 0.5392, + "mean_token_accuracy": 0.8177320957183838, + "num_tokens": 506351276.0, + "step": 20486 + }, + { + "epoch": 1.6893708254308568, + "grad_norm": 2.1686370372772217, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7544392347335815, + "num_tokens": 506376865.0, + "step": 20487 + }, + { + "epoch": 1.6894532860559082, + "grad_norm": 2.1693735122680664, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7462837100028992, + "num_tokens": 506405241.0, + "step": 20488 + }, + { + "epoch": 1.68953574668096, + "grad_norm": 1.8568555116653442, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.766581654548645, + "num_tokens": 506437363.0, + "step": 20489 + }, + { + "epoch": 1.6896182073060113, + "grad_norm": 2.18135404586792, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7223004698753357, + "num_tokens": 506463870.0, + "step": 20490 + }, + { + "epoch": 1.6897006679310629, + "grad_norm": 2.1485393047332764, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7649168968200684, + "num_tokens": 506491583.0, + "step": 20491 + }, + { + "epoch": 1.6897831285561145, + "grad_norm": 2.109677314758301, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7599821090698242, + "num_tokens": 506517249.0, + "step": 20492 + }, + { + "epoch": 1.689865589181166, + "grad_norm": 1.8906699419021606, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.7896265387535095, + "num_tokens": 506547848.0, + "step": 20493 + }, + { + "epoch": 1.6899480498062176, + "grad_norm": 1.939739465713501, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.795080304145813, + "num_tokens": 506579595.0, + "step": 20494 + }, + { + "epoch": 1.690030510431269, + "grad_norm": 2.0883686542510986, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7315152287483215, + "num_tokens": 506610308.0, + "step": 20495 + }, + { + "epoch": 1.6901129710563207, + "grad_norm": 1.9129139184951782, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7838809490203857, + "num_tokens": 506640393.0, + "step": 20496 + }, + { + "epoch": 1.690195431681372, + "grad_norm": 2.2868733406066895, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7736361026763916, + "num_tokens": 506665118.0, + "step": 20497 + }, + { + "epoch": 1.6902778923064237, + "grad_norm": 2.146449327468872, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7494858503341675, + "num_tokens": 506692599.0, + "step": 20498 + }, + { + "epoch": 1.6903603529314752, + "grad_norm": 2.05774188041687, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7858822345733643, + "num_tokens": 506720002.0, + "step": 20499 + }, + { + "epoch": 1.6904428135565268, + "grad_norm": 1.9534025192260742, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7546721696853638, + "num_tokens": 506751769.0, + "step": 20500 + }, + { + "epoch": 1.6905252741815784, + "grad_norm": 1.9407029151916504, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7691934704780579, + "num_tokens": 506784535.0, + "step": 20501 + }, + { + "epoch": 1.6906077348066297, + "grad_norm": 1.9322054386138916, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7542421817779541, + "num_tokens": 506815862.0, + "step": 20502 + }, + { + "epoch": 1.6906901954316815, + "grad_norm": 1.8316353559494019, + "learning_rate": 1e-06, + "loss": 0.5715, + "mean_token_accuracy": 0.8092327117919922, + "num_tokens": 506848031.0, + "step": 20503 + }, + { + "epoch": 1.6907726560567329, + "grad_norm": 1.925200343132019, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.7988444566726685, + "num_tokens": 506875583.0, + "step": 20504 + }, + { + "epoch": 1.6908551166817845, + "grad_norm": 2.389604330062866, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7405171394348145, + "num_tokens": 506896783.0, + "step": 20505 + }, + { + "epoch": 1.690937577306836, + "grad_norm": 1.9099507331848145, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7866367101669312, + "num_tokens": 506930158.0, + "step": 20506 + }, + { + "epoch": 1.6910200379318874, + "grad_norm": 1.9047292470932007, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7785918712615967, + "num_tokens": 506962547.0, + "step": 20507 + }, + { + "epoch": 1.6911024985569392, + "grad_norm": 2.128929615020752, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.7016138434410095, + "num_tokens": 506990417.0, + "step": 20508 + }, + { + "epoch": 1.6911849591819905, + "grad_norm": 2.236074686050415, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.769244372844696, + "num_tokens": 507018684.0, + "step": 20509 + }, + { + "epoch": 1.691267419807042, + "grad_norm": 2.3557326793670654, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7460134625434875, + "num_tokens": 507042616.0, + "step": 20510 + }, + { + "epoch": 1.6913498804320937, + "grad_norm": 1.9575586318969727, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7506055235862732, + "num_tokens": 507072014.0, + "step": 20511 + }, + { + "epoch": 1.6914323410571452, + "grad_norm": 1.8224104642868042, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7475441098213196, + "num_tokens": 507109360.0, + "step": 20512 + }, + { + "epoch": 1.6915148016821968, + "grad_norm": 2.019693374633789, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7633414268493652, + "num_tokens": 507135124.0, + "step": 20513 + }, + { + "epoch": 1.6915972623072482, + "grad_norm": 2.0859673023223877, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7592769265174866, + "num_tokens": 507162651.0, + "step": 20514 + }, + { + "epoch": 1.6916797229323, + "grad_norm": 2.1511456966400146, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7714895606040955, + "num_tokens": 507187732.0, + "step": 20515 + }, + { + "epoch": 1.6917621835573513, + "grad_norm": 1.817969560623169, + "learning_rate": 1e-06, + "loss": 0.5305, + "mean_token_accuracy": 0.80698561668396, + "num_tokens": 507219757.0, + "step": 20516 + }, + { + "epoch": 1.6918446441824029, + "grad_norm": 1.8294473886489868, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.784674882888794, + "num_tokens": 507254070.0, + "step": 20517 + }, + { + "epoch": 1.6919271048074545, + "grad_norm": 2.043826103210449, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7220966815948486, + "num_tokens": 507283324.0, + "step": 20518 + }, + { + "epoch": 1.692009565432506, + "grad_norm": 1.8494479656219482, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7881131768226624, + "num_tokens": 507314506.0, + "step": 20519 + }, + { + "epoch": 1.6920920260575576, + "grad_norm": 1.8678855895996094, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.786948025226593, + "num_tokens": 507344121.0, + "step": 20520 + }, + { + "epoch": 1.692174486682609, + "grad_norm": 1.856341004371643, + "learning_rate": 1e-06, + "loss": 0.5667, + "mean_token_accuracy": 0.8117928504943848, + "num_tokens": 507375468.0, + "step": 20521 + }, + { + "epoch": 1.6922569473076607, + "grad_norm": 1.92206871509552, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7604939937591553, + "num_tokens": 507405561.0, + "step": 20522 + }, + { + "epoch": 1.692339407932712, + "grad_norm": 2.090116500854492, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.786052405834198, + "num_tokens": 507430259.0, + "step": 20523 + }, + { + "epoch": 1.6924218685577637, + "grad_norm": 2.2136340141296387, + "learning_rate": 1e-06, + "loss": 0.5662, + "mean_token_accuracy": 0.8092204332351685, + "num_tokens": 507460483.0, + "step": 20524 + }, + { + "epoch": 1.6925043291828152, + "grad_norm": 2.062612771987915, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7566357851028442, + "num_tokens": 507487573.0, + "step": 20525 + }, + { + "epoch": 1.6925867898078666, + "grad_norm": 1.9803276062011719, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.7484421730041504, + "num_tokens": 507518390.0, + "step": 20526 + }, + { + "epoch": 1.6926692504329184, + "grad_norm": 2.0201029777526855, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7650855779647827, + "num_tokens": 507547644.0, + "step": 20527 + }, + { + "epoch": 1.6927517110579697, + "grad_norm": 2.0916824340820312, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.719037652015686, + "num_tokens": 507576690.0, + "step": 20528 + }, + { + "epoch": 1.6928341716830213, + "grad_norm": 1.7651381492614746, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7772226333618164, + "num_tokens": 507614682.0, + "step": 20529 + }, + { + "epoch": 1.6929166323080729, + "grad_norm": 2.0118019580841064, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7326605319976807, + "num_tokens": 507644235.0, + "step": 20530 + }, + { + "epoch": 1.6929990929331244, + "grad_norm": 1.8556358814239502, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.745787501335144, + "num_tokens": 507680441.0, + "step": 20531 + }, + { + "epoch": 1.693081553558176, + "grad_norm": 1.7244373559951782, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7708821296691895, + "num_tokens": 507719740.0, + "step": 20532 + }, + { + "epoch": 1.6931640141832274, + "grad_norm": 1.8620153665542603, + "learning_rate": 1e-06, + "loss": 0.6034, + "mean_token_accuracy": 0.7928661704063416, + "num_tokens": 507750536.0, + "step": 20533 + }, + { + "epoch": 1.6932464748082792, + "grad_norm": 2.159762382507324, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7598017454147339, + "num_tokens": 507776077.0, + "step": 20534 + }, + { + "epoch": 1.6933289354333305, + "grad_norm": 1.9894027709960938, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7478294372558594, + "num_tokens": 507808063.0, + "step": 20535 + }, + { + "epoch": 1.693411396058382, + "grad_norm": 2.024042844772339, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.7819654941558838, + "num_tokens": 507838605.0, + "step": 20536 + }, + { + "epoch": 1.6934938566834337, + "grad_norm": 1.9942957162857056, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7768778800964355, + "num_tokens": 507866884.0, + "step": 20537 + }, + { + "epoch": 1.6935763173084852, + "grad_norm": 1.8668935298919678, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7567002773284912, + "num_tokens": 507901994.0, + "step": 20538 + }, + { + "epoch": 1.6936587779335368, + "grad_norm": 2.3848040103912354, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7351350784301758, + "num_tokens": 507925747.0, + "step": 20539 + }, + { + "epoch": 1.6937412385585882, + "grad_norm": 1.9767169952392578, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7563650608062744, + "num_tokens": 507956014.0, + "step": 20540 + }, + { + "epoch": 1.69382369918364, + "grad_norm": 2.2750916481018066, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.752126932144165, + "num_tokens": 507981322.0, + "step": 20541 + }, + { + "epoch": 1.6939061598086913, + "grad_norm": 2.0855648517608643, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7364798188209534, + "num_tokens": 508009503.0, + "step": 20542 + }, + { + "epoch": 1.6939886204337429, + "grad_norm": 2.1395883560180664, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7499807476997375, + "num_tokens": 508037916.0, + "step": 20543 + }, + { + "epoch": 1.6940710810587944, + "grad_norm": 1.9777532815933228, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7798462510108948, + "num_tokens": 508066640.0, + "step": 20544 + }, + { + "epoch": 1.6941535416838458, + "grad_norm": 2.6761488914489746, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.766812801361084, + "num_tokens": 508095099.0, + "step": 20545 + }, + { + "epoch": 1.6942360023088976, + "grad_norm": 1.965903639793396, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.7883385419845581, + "num_tokens": 508123981.0, + "step": 20546 + }, + { + "epoch": 1.694318462933949, + "grad_norm": 2.146878480911255, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7502475380897522, + "num_tokens": 508153389.0, + "step": 20547 + }, + { + "epoch": 1.6944009235590007, + "grad_norm": 2.1025290489196777, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.7697490453720093, + "num_tokens": 508179730.0, + "step": 20548 + }, + { + "epoch": 1.694483384184052, + "grad_norm": 1.8476381301879883, + "learning_rate": 1e-06, + "loss": 0.5897, + "mean_token_accuracy": 0.7994462251663208, + "num_tokens": 508211862.0, + "step": 20549 + }, + { + "epoch": 1.6945658448091037, + "grad_norm": 1.9501502513885498, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.7917987108230591, + "num_tokens": 508240018.0, + "step": 20550 + }, + { + "epoch": 1.6946483054341552, + "grad_norm": 1.634652853012085, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.8040425777435303, + "num_tokens": 508277649.0, + "step": 20551 + }, + { + "epoch": 1.6947307660592066, + "grad_norm": 2.274596929550171, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7395682334899902, + "num_tokens": 508302112.0, + "step": 20552 + }, + { + "epoch": 1.6948132266842584, + "grad_norm": 1.980826735496521, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7774901986122131, + "num_tokens": 508331775.0, + "step": 20553 + }, + { + "epoch": 1.6948956873093097, + "grad_norm": 2.381840229034424, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.7833409309387207, + "num_tokens": 508356308.0, + "step": 20554 + }, + { + "epoch": 1.6949781479343613, + "grad_norm": 1.8573631048202515, + "learning_rate": 1e-06, + "loss": 0.4809, + "mean_token_accuracy": 0.8348443508148193, + "num_tokens": 508388260.0, + "step": 20555 + }, + { + "epoch": 1.6950606085594129, + "grad_norm": 1.9243085384368896, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7560793161392212, + "num_tokens": 508421675.0, + "step": 20556 + }, + { + "epoch": 1.6951430691844644, + "grad_norm": 2.0231990814208984, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7414575219154358, + "num_tokens": 508451440.0, + "step": 20557 + }, + { + "epoch": 1.695225529809516, + "grad_norm": 1.7499817609786987, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7651541233062744, + "num_tokens": 508488829.0, + "step": 20558 + }, + { + "epoch": 1.6953079904345674, + "grad_norm": 2.1614129543304443, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7677519917488098, + "num_tokens": 508515681.0, + "step": 20559 + }, + { + "epoch": 1.6953904510596192, + "grad_norm": 1.7379555702209473, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7828578948974609, + "num_tokens": 508555067.0, + "step": 20560 + }, + { + "epoch": 1.6954729116846705, + "grad_norm": 2.0039336681365967, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7738906145095825, + "num_tokens": 508585503.0, + "step": 20561 + }, + { + "epoch": 1.695555372309722, + "grad_norm": 2.068269729614258, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.731156587600708, + "num_tokens": 508613810.0, + "step": 20562 + }, + { + "epoch": 1.6956378329347737, + "grad_norm": 2.2225587368011475, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7592200040817261, + "num_tokens": 508639163.0, + "step": 20563 + }, + { + "epoch": 1.6957202935598252, + "grad_norm": 1.8883503675460815, + "learning_rate": 1e-06, + "loss": 0.5529, + "mean_token_accuracy": 0.803839385509491, + "num_tokens": 508669897.0, + "step": 20564 + }, + { + "epoch": 1.6958027541848768, + "grad_norm": 2.231020212173462, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7509735822677612, + "num_tokens": 508695644.0, + "step": 20565 + }, + { + "epoch": 1.6958852148099282, + "grad_norm": 2.0114550590515137, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7487125992774963, + "num_tokens": 508724039.0, + "step": 20566 + }, + { + "epoch": 1.69596767543498, + "grad_norm": 2.3314294815063477, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7493396997451782, + "num_tokens": 508748322.0, + "step": 20567 + }, + { + "epoch": 1.6960501360600313, + "grad_norm": 1.9634186029434204, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7251043319702148, + "num_tokens": 508780099.0, + "step": 20568 + }, + { + "epoch": 1.6961325966850829, + "grad_norm": 1.9021828174591064, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7424583435058594, + "num_tokens": 508812650.0, + "step": 20569 + }, + { + "epoch": 1.6962150573101344, + "grad_norm": 1.7758654356002808, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7580896019935608, + "num_tokens": 508849591.0, + "step": 20570 + }, + { + "epoch": 1.6962975179351858, + "grad_norm": 2.172987937927246, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7644765973091125, + "num_tokens": 508874971.0, + "step": 20571 + }, + { + "epoch": 1.6963799785602376, + "grad_norm": 1.8624272346496582, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7415813207626343, + "num_tokens": 508910439.0, + "step": 20572 + }, + { + "epoch": 1.696462439185289, + "grad_norm": 1.8024617433547974, + "learning_rate": 1e-06, + "loss": 0.5932, + "mean_token_accuracy": 0.790242075920105, + "num_tokens": 508940061.0, + "step": 20573 + }, + { + "epoch": 1.6965448998103405, + "grad_norm": 2.1299917697906494, + "learning_rate": 1e-06, + "loss": 0.5617, + "mean_token_accuracy": 0.8081209063529968, + "num_tokens": 508964650.0, + "step": 20574 + }, + { + "epoch": 1.696627360435392, + "grad_norm": 1.9388575553894043, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7799760103225708, + "num_tokens": 508994958.0, + "step": 20575 + }, + { + "epoch": 1.6967098210604437, + "grad_norm": 1.9778996706008911, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7659034729003906, + "num_tokens": 509025562.0, + "step": 20576 + }, + { + "epoch": 1.6967922816854952, + "grad_norm": 1.9835841655731201, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7816891670227051, + "num_tokens": 509056432.0, + "step": 20577 + }, + { + "epoch": 1.6968747423105466, + "grad_norm": 1.8780250549316406, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.786878764629364, + "num_tokens": 509086451.0, + "step": 20578 + }, + { + "epoch": 1.6969572029355984, + "grad_norm": 2.279768228530884, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7307100296020508, + "num_tokens": 509111450.0, + "step": 20579 + }, + { + "epoch": 1.6970396635606497, + "grad_norm": 2.0124154090881348, + "learning_rate": 1e-06, + "loss": 0.5798, + "mean_token_accuracy": 0.8039895296096802, + "num_tokens": 509138650.0, + "step": 20580 + }, + { + "epoch": 1.6971221241857013, + "grad_norm": 1.9160434007644653, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.758202075958252, + "num_tokens": 509170147.0, + "step": 20581 + }, + { + "epoch": 1.6972045848107529, + "grad_norm": 2.2931435108184814, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7417521476745605, + "num_tokens": 509194683.0, + "step": 20582 + }, + { + "epoch": 1.6972870454358044, + "grad_norm": 1.9685090780258179, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7397179007530212, + "num_tokens": 509226390.0, + "step": 20583 + }, + { + "epoch": 1.697369506060856, + "grad_norm": 1.909585952758789, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7286086678504944, + "num_tokens": 509262738.0, + "step": 20584 + }, + { + "epoch": 1.6974519666859074, + "grad_norm": 2.010901927947998, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7447249889373779, + "num_tokens": 509292128.0, + "step": 20585 + }, + { + "epoch": 1.6975344273109592, + "grad_norm": 1.9031926393508911, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.7539961338043213, + "num_tokens": 509325465.0, + "step": 20586 + }, + { + "epoch": 1.6976168879360105, + "grad_norm": 1.6838279962539673, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7638247609138489, + "num_tokens": 509364278.0, + "step": 20587 + }, + { + "epoch": 1.697699348561062, + "grad_norm": 2.096072196960449, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7697858810424805, + "num_tokens": 509394869.0, + "step": 20588 + }, + { + "epoch": 1.6977818091861137, + "grad_norm": 1.9045273065567017, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7630267143249512, + "num_tokens": 509429240.0, + "step": 20589 + }, + { + "epoch": 1.697864269811165, + "grad_norm": 1.8763008117675781, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.7966948747634888, + "num_tokens": 509461674.0, + "step": 20590 + }, + { + "epoch": 1.6979467304362168, + "grad_norm": 2.2178139686584473, + "learning_rate": 1e-06, + "loss": 0.589, + "mean_token_accuracy": 0.8089538216590881, + "num_tokens": 509486327.0, + "step": 20591 + }, + { + "epoch": 1.6980291910612682, + "grad_norm": 2.049187660217285, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7519444823265076, + "num_tokens": 509516014.0, + "step": 20592 + }, + { + "epoch": 1.69811165168632, + "grad_norm": 1.8297805786132812, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7836271524429321, + "num_tokens": 509550162.0, + "step": 20593 + }, + { + "epoch": 1.6981941123113713, + "grad_norm": 2.061488628387451, + "learning_rate": 1e-06, + "loss": 0.8014, + "mean_token_accuracy": 0.7423075437545776, + "num_tokens": 509579921.0, + "step": 20594 + }, + { + "epoch": 1.6982765729364229, + "grad_norm": 2.4553165435791016, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.735152006149292, + "num_tokens": 509603909.0, + "step": 20595 + }, + { + "epoch": 1.6983590335614744, + "grad_norm": 1.9047324657440186, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7624400854110718, + "num_tokens": 509635430.0, + "step": 20596 + }, + { + "epoch": 1.6984414941865258, + "grad_norm": 1.7641228437423706, + "learning_rate": 1e-06, + "loss": 0.5763, + "mean_token_accuracy": 0.8039329051971436, + "num_tokens": 509668313.0, + "step": 20597 + }, + { + "epoch": 1.6985239548115776, + "grad_norm": 2.0068070888519287, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7642239332199097, + "num_tokens": 509697118.0, + "step": 20598 + }, + { + "epoch": 1.698606415436629, + "grad_norm": 1.8094301223754883, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7668288350105286, + "num_tokens": 509735542.0, + "step": 20599 + }, + { + "epoch": 1.6986888760616805, + "grad_norm": 2.2087881565093994, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7448068857192993, + "num_tokens": 509762064.0, + "step": 20600 + }, + { + "epoch": 1.698771336686732, + "grad_norm": 1.735702395439148, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7711288928985596, + "num_tokens": 509800450.0, + "step": 20601 + }, + { + "epoch": 1.6988537973117837, + "grad_norm": 1.8908582925796509, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.783832848072052, + "num_tokens": 509830502.0, + "step": 20602 + }, + { + "epoch": 1.6989362579368352, + "grad_norm": 2.2043423652648926, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7586195468902588, + "num_tokens": 509857684.0, + "step": 20603 + }, + { + "epoch": 1.6990187185618866, + "grad_norm": 2.330554485321045, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7540072798728943, + "num_tokens": 509879937.0, + "step": 20604 + }, + { + "epoch": 1.6991011791869384, + "grad_norm": 2.3342201709747314, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.7718090415000916, + "num_tokens": 509902662.0, + "step": 20605 + }, + { + "epoch": 1.6991836398119897, + "grad_norm": 1.992961049079895, + "learning_rate": 1e-06, + "loss": 0.8014, + "mean_token_accuracy": 0.7583742141723633, + "num_tokens": 509934195.0, + "step": 20606 + }, + { + "epoch": 1.6992661004370413, + "grad_norm": 1.9481161832809448, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7830625772476196, + "num_tokens": 509966266.0, + "step": 20607 + }, + { + "epoch": 1.6993485610620929, + "grad_norm": 2.0258030891418457, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7225636839866638, + "num_tokens": 509998301.0, + "step": 20608 + }, + { + "epoch": 1.6994310216871444, + "grad_norm": 2.0025060176849365, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.7972924113273621, + "num_tokens": 510024292.0, + "step": 20609 + }, + { + "epoch": 1.699513482312196, + "grad_norm": 1.9017161130905151, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7808458209037781, + "num_tokens": 510057439.0, + "step": 20610 + }, + { + "epoch": 1.6995959429372474, + "grad_norm": 1.923708438873291, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7645021677017212, + "num_tokens": 510090362.0, + "step": 20611 + }, + { + "epoch": 1.6996784035622992, + "grad_norm": 2.010003089904785, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7671180367469788, + "num_tokens": 510116883.0, + "step": 20612 + }, + { + "epoch": 1.6997608641873505, + "grad_norm": 1.836836814880371, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7918158769607544, + "num_tokens": 510151048.0, + "step": 20613 + }, + { + "epoch": 1.699843324812402, + "grad_norm": 1.798588514328003, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7504485845565796, + "num_tokens": 510184913.0, + "step": 20614 + }, + { + "epoch": 1.6999257854374537, + "grad_norm": 1.9319121837615967, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7780848741531372, + "num_tokens": 510212628.0, + "step": 20615 + }, + { + "epoch": 1.700008246062505, + "grad_norm": 2.036189079284668, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7584594488143921, + "num_tokens": 510242420.0, + "step": 20616 + }, + { + "epoch": 1.7000907066875568, + "grad_norm": 1.9069983959197998, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7092919945716858, + "num_tokens": 510275905.0, + "step": 20617 + }, + { + "epoch": 1.7001731673126081, + "grad_norm": 2.2290689945220947, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7641931772232056, + "num_tokens": 510299978.0, + "step": 20618 + }, + { + "epoch": 1.7002556279376597, + "grad_norm": 2.3545620441436768, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7621819376945496, + "num_tokens": 510322120.0, + "step": 20619 + }, + { + "epoch": 1.7003380885627113, + "grad_norm": 1.9512560367584229, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7113627195358276, + "num_tokens": 510355643.0, + "step": 20620 + }, + { + "epoch": 1.7004205491877629, + "grad_norm": 2.230135440826416, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.7644942998886108, + "num_tokens": 510382344.0, + "step": 20621 + }, + { + "epoch": 1.7005030098128144, + "grad_norm": 2.180366277694702, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7452739477157593, + "num_tokens": 510410193.0, + "step": 20622 + }, + { + "epoch": 1.7005854704378658, + "grad_norm": 1.9198946952819824, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7684919834136963, + "num_tokens": 510441252.0, + "step": 20623 + }, + { + "epoch": 1.7006679310629176, + "grad_norm": 1.9222179651260376, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7721285820007324, + "num_tokens": 510471813.0, + "step": 20624 + }, + { + "epoch": 1.700750391687969, + "grad_norm": 2.3280117511749268, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7415648698806763, + "num_tokens": 510494987.0, + "step": 20625 + }, + { + "epoch": 1.7008328523130205, + "grad_norm": 1.8125561475753784, + "learning_rate": 1e-06, + "loss": 0.5992, + "mean_token_accuracy": 0.8061875104904175, + "num_tokens": 510525821.0, + "step": 20626 + }, + { + "epoch": 1.700915312938072, + "grad_norm": 2.0208740234375, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7756908535957336, + "num_tokens": 510555170.0, + "step": 20627 + }, + { + "epoch": 1.7009977735631236, + "grad_norm": 2.0898373126983643, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.754416823387146, + "num_tokens": 510584785.0, + "step": 20628 + }, + { + "epoch": 1.7010802341881752, + "grad_norm": 2.06946063041687, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7764400839805603, + "num_tokens": 510612536.0, + "step": 20629 + }, + { + "epoch": 1.7011626948132266, + "grad_norm": 2.040987491607666, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7501221895217896, + "num_tokens": 510641665.0, + "step": 20630 + }, + { + "epoch": 1.7012451554382784, + "grad_norm": 1.94027578830719, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7784495949745178, + "num_tokens": 510672526.0, + "step": 20631 + }, + { + "epoch": 1.7013276160633297, + "grad_norm": 2.0214695930480957, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7666772603988647, + "num_tokens": 510699457.0, + "step": 20632 + }, + { + "epoch": 1.7014100766883813, + "grad_norm": 1.9614107608795166, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7346900105476379, + "num_tokens": 510731523.0, + "step": 20633 + }, + { + "epoch": 1.7014925373134329, + "grad_norm": 1.7176668643951416, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.7996540069580078, + "num_tokens": 510766893.0, + "step": 20634 + }, + { + "epoch": 1.7015749979384842, + "grad_norm": 1.9944840669631958, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7414284944534302, + "num_tokens": 510796725.0, + "step": 20635 + }, + { + "epoch": 1.701657458563536, + "grad_norm": 2.036989212036133, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.7834595441818237, + "num_tokens": 510825934.0, + "step": 20636 + }, + { + "epoch": 1.7017399191885874, + "grad_norm": 2.536404848098755, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7550376057624817, + "num_tokens": 510845074.0, + "step": 20637 + }, + { + "epoch": 1.7018223798136392, + "grad_norm": 2.0464320182800293, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.8073914051055908, + "num_tokens": 510872795.0, + "step": 20638 + }, + { + "epoch": 1.7019048404386905, + "grad_norm": 2.209571361541748, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.7559068202972412, + "num_tokens": 510897427.0, + "step": 20639 + }, + { + "epoch": 1.701987301063742, + "grad_norm": 1.954771637916565, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7604244947433472, + "num_tokens": 510927266.0, + "step": 20640 + }, + { + "epoch": 1.7020697616887936, + "grad_norm": 1.8994778394699097, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7631224393844604, + "num_tokens": 510957795.0, + "step": 20641 + }, + { + "epoch": 1.702152222313845, + "grad_norm": 2.094320297241211, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7321639060974121, + "num_tokens": 510986723.0, + "step": 20642 + }, + { + "epoch": 1.7022346829388968, + "grad_norm": 1.8898197412490845, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.7886901497840881, + "num_tokens": 511017015.0, + "step": 20643 + }, + { + "epoch": 1.7023171435639481, + "grad_norm": 1.901275396347046, + "learning_rate": 1e-06, + "loss": 0.5922, + "mean_token_accuracy": 0.7988526821136475, + "num_tokens": 511046701.0, + "step": 20644 + }, + { + "epoch": 1.7023996041889997, + "grad_norm": 1.9603450298309326, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7504853010177612, + "num_tokens": 511080286.0, + "step": 20645 + }, + { + "epoch": 1.7024820648140513, + "grad_norm": 1.8615460395812988, + "learning_rate": 1e-06, + "loss": 0.5773, + "mean_token_accuracy": 0.8025591969490051, + "num_tokens": 511110316.0, + "step": 20646 + }, + { + "epoch": 1.7025645254391029, + "grad_norm": 2.1711013317108154, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.722897469997406, + "num_tokens": 511135945.0, + "step": 20647 + }, + { + "epoch": 1.7026469860641544, + "grad_norm": 1.8200567960739136, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7680187225341797, + "num_tokens": 511168948.0, + "step": 20648 + }, + { + "epoch": 1.7027294466892058, + "grad_norm": 2.135615348815918, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7607907056808472, + "num_tokens": 511194547.0, + "step": 20649 + }, + { + "epoch": 1.7028119073142576, + "grad_norm": 1.8899354934692383, + "learning_rate": 1e-06, + "loss": 0.6046, + "mean_token_accuracy": 0.7986186742782593, + "num_tokens": 511228982.0, + "step": 20650 + }, + { + "epoch": 1.702894367939309, + "grad_norm": 2.410351514816284, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7716126441955566, + "num_tokens": 511250351.0, + "step": 20651 + }, + { + "epoch": 1.7029768285643605, + "grad_norm": 1.8146332502365112, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.7894155979156494, + "num_tokens": 511285358.0, + "step": 20652 + }, + { + "epoch": 1.703059289189412, + "grad_norm": 2.065009593963623, + "learning_rate": 1e-06, + "loss": 0.6263, + "mean_token_accuracy": 0.7914303541183472, + "num_tokens": 511314462.0, + "step": 20653 + }, + { + "epoch": 1.7031417498144636, + "grad_norm": 1.9740386009216309, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.77684485912323, + "num_tokens": 511347131.0, + "step": 20654 + }, + { + "epoch": 1.7032242104395152, + "grad_norm": 1.8769713640213013, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7623084783554077, + "num_tokens": 511381689.0, + "step": 20655 + }, + { + "epoch": 1.7033066710645666, + "grad_norm": 1.9296036958694458, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7548475861549377, + "num_tokens": 511411869.0, + "step": 20656 + }, + { + "epoch": 1.7033891316896184, + "grad_norm": 1.8619616031646729, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7630575895309448, + "num_tokens": 511442692.0, + "step": 20657 + }, + { + "epoch": 1.7034715923146697, + "grad_norm": 1.8649779558181763, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7380945682525635, + "num_tokens": 511477888.0, + "step": 20658 + }, + { + "epoch": 1.7035540529397213, + "grad_norm": 1.984818935394287, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7528940439224243, + "num_tokens": 511507494.0, + "step": 20659 + }, + { + "epoch": 1.7036365135647729, + "grad_norm": 1.981770396232605, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7349942922592163, + "num_tokens": 511539752.0, + "step": 20660 + }, + { + "epoch": 1.7037189741898242, + "grad_norm": 1.8865848779678345, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7119171023368835, + "num_tokens": 511573987.0, + "step": 20661 + }, + { + "epoch": 1.703801434814876, + "grad_norm": 1.9281556606292725, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7452526092529297, + "num_tokens": 511604922.0, + "step": 20662 + }, + { + "epoch": 1.7038838954399274, + "grad_norm": 2.075439691543579, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7518914937973022, + "num_tokens": 511634360.0, + "step": 20663 + }, + { + "epoch": 1.703966356064979, + "grad_norm": 1.717179298400879, + "learning_rate": 1e-06, + "loss": 0.5566, + "mean_token_accuracy": 0.8163610100746155, + "num_tokens": 511672171.0, + "step": 20664 + }, + { + "epoch": 1.7040488166900305, + "grad_norm": 1.8710808753967285, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7292511463165283, + "num_tokens": 511705779.0, + "step": 20665 + }, + { + "epoch": 1.704131277315082, + "grad_norm": 1.8725149631500244, + "learning_rate": 1e-06, + "loss": 0.5555, + "mean_token_accuracy": 0.8144099712371826, + "num_tokens": 511737174.0, + "step": 20666 + }, + { + "epoch": 1.7042137379401336, + "grad_norm": 1.9224814176559448, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7379307746887207, + "num_tokens": 511770885.0, + "step": 20667 + }, + { + "epoch": 1.704296198565185, + "grad_norm": 1.9586762189865112, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7803864479064941, + "num_tokens": 511799065.0, + "step": 20668 + }, + { + "epoch": 1.7043786591902368, + "grad_norm": 1.973575234413147, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7584470510482788, + "num_tokens": 511831498.0, + "step": 20669 + }, + { + "epoch": 1.7044611198152881, + "grad_norm": 2.1512162685394287, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.787848949432373, + "num_tokens": 511856542.0, + "step": 20670 + }, + { + "epoch": 1.7045435804403397, + "grad_norm": 2.3992743492126465, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7320228815078735, + "num_tokens": 511878953.0, + "step": 20671 + }, + { + "epoch": 1.7046260410653913, + "grad_norm": 1.9522678852081299, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.7999963760375977, + "num_tokens": 511910754.0, + "step": 20672 + }, + { + "epoch": 1.7047085016904429, + "grad_norm": 1.8711670637130737, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.7706444263458252, + "num_tokens": 511942746.0, + "step": 20673 + }, + { + "epoch": 1.7047909623154944, + "grad_norm": 2.0549705028533936, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7679494619369507, + "num_tokens": 511970071.0, + "step": 20674 + }, + { + "epoch": 1.7048734229405458, + "grad_norm": 1.7100619077682495, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7692051529884338, + "num_tokens": 512008002.0, + "step": 20675 + }, + { + "epoch": 1.7049558835655976, + "grad_norm": 2.124051809310913, + "learning_rate": 1e-06, + "loss": 0.7642, + "mean_token_accuracy": 0.7536914348602295, + "num_tokens": 512033732.0, + "step": 20676 + }, + { + "epoch": 1.705038344190649, + "grad_norm": 2.0123493671417236, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.7859542369842529, + "num_tokens": 512061180.0, + "step": 20677 + }, + { + "epoch": 1.7051208048157005, + "grad_norm": 2.4664759635925293, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7351371049880981, + "num_tokens": 512083250.0, + "step": 20678 + }, + { + "epoch": 1.705203265440752, + "grad_norm": 1.8534820079803467, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7689481973648071, + "num_tokens": 512115444.0, + "step": 20679 + }, + { + "epoch": 1.7052857260658034, + "grad_norm": 1.8706402778625488, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7818241715431213, + "num_tokens": 512147161.0, + "step": 20680 + }, + { + "epoch": 1.7053681866908552, + "grad_norm": 2.4093854427337646, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7474165558815002, + "num_tokens": 512170349.0, + "step": 20681 + }, + { + "epoch": 1.7054506473159066, + "grad_norm": 2.1613738536834717, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7533911466598511, + "num_tokens": 512197921.0, + "step": 20682 + }, + { + "epoch": 1.7055331079409581, + "grad_norm": 1.9387686252593994, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7584620714187622, + "num_tokens": 512227862.0, + "step": 20683 + }, + { + "epoch": 1.7056155685660097, + "grad_norm": 2.020385503768921, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7690795063972473, + "num_tokens": 512256104.0, + "step": 20684 + }, + { + "epoch": 1.7056980291910613, + "grad_norm": 1.8879454135894775, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7721352577209473, + "num_tokens": 512287479.0, + "step": 20685 + }, + { + "epoch": 1.7057804898161129, + "grad_norm": 2.2736895084381104, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.759994387626648, + "num_tokens": 512313561.0, + "step": 20686 + }, + { + "epoch": 1.7058629504411642, + "grad_norm": 1.9459666013717651, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.773607611656189, + "num_tokens": 512346192.0, + "step": 20687 + }, + { + "epoch": 1.705945411066216, + "grad_norm": 1.9637726545333862, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7508436441421509, + "num_tokens": 512378113.0, + "step": 20688 + }, + { + "epoch": 1.7060278716912674, + "grad_norm": 2.0355496406555176, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7532280683517456, + "num_tokens": 512408270.0, + "step": 20689 + }, + { + "epoch": 1.706110332316319, + "grad_norm": 1.76943838596344, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.7801271677017212, + "num_tokens": 512440696.0, + "step": 20690 + }, + { + "epoch": 1.7061927929413705, + "grad_norm": 2.117302656173706, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7816717028617859, + "num_tokens": 512464409.0, + "step": 20691 + }, + { + "epoch": 1.706275253566422, + "grad_norm": 1.8467612266540527, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7709249258041382, + "num_tokens": 512497850.0, + "step": 20692 + }, + { + "epoch": 1.7063577141914736, + "grad_norm": 1.8879722356796265, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7775961756706238, + "num_tokens": 512529562.0, + "step": 20693 + }, + { + "epoch": 1.706440174816525, + "grad_norm": 1.7501096725463867, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7747915983200073, + "num_tokens": 512564039.0, + "step": 20694 + }, + { + "epoch": 1.7065226354415768, + "grad_norm": 2.011505126953125, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7768334746360779, + "num_tokens": 512591763.0, + "step": 20695 + }, + { + "epoch": 1.7066050960666281, + "grad_norm": 1.8788108825683594, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7540604472160339, + "num_tokens": 512622914.0, + "step": 20696 + }, + { + "epoch": 1.7066875566916797, + "grad_norm": 1.9976754188537598, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7215030193328857, + "num_tokens": 512653656.0, + "step": 20697 + }, + { + "epoch": 1.7067700173167313, + "grad_norm": 1.9095379114151, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7594954967498779, + "num_tokens": 512685606.0, + "step": 20698 + }, + { + "epoch": 1.7068524779417829, + "grad_norm": 1.71111261844635, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7719786763191223, + "num_tokens": 512726049.0, + "step": 20699 + }, + { + "epoch": 1.7069349385668344, + "grad_norm": 2.677060604095459, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.7819200754165649, + "num_tokens": 512744271.0, + "step": 20700 + }, + { + "epoch": 1.7070173991918858, + "grad_norm": 1.8908748626708984, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7666068077087402, + "num_tokens": 512777601.0, + "step": 20701 + }, + { + "epoch": 1.7070998598169376, + "grad_norm": 2.0179877281188965, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7544137835502625, + "num_tokens": 512805744.0, + "step": 20702 + }, + { + "epoch": 1.707182320441989, + "grad_norm": 2.0992066860198975, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7686179876327515, + "num_tokens": 512830557.0, + "step": 20703 + }, + { + "epoch": 1.7072647810670405, + "grad_norm": 2.147751569747925, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.7835211753845215, + "num_tokens": 512856431.0, + "step": 20704 + }, + { + "epoch": 1.707347241692092, + "grad_norm": 1.8738058805465698, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.7826434969902039, + "num_tokens": 512890183.0, + "step": 20705 + }, + { + "epoch": 1.7074297023171434, + "grad_norm": 2.062007188796997, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7698585987091064, + "num_tokens": 512916243.0, + "step": 20706 + }, + { + "epoch": 1.7075121629421952, + "grad_norm": 2.3327035903930664, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7226043343544006, + "num_tokens": 512939459.0, + "step": 20707 + }, + { + "epoch": 1.7075946235672466, + "grad_norm": 1.9699583053588867, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7755320072174072, + "num_tokens": 512968545.0, + "step": 20708 + }, + { + "epoch": 1.7076770841922981, + "grad_norm": 2.0004234313964844, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.784570574760437, + "num_tokens": 512996511.0, + "step": 20709 + }, + { + "epoch": 1.7077595448173497, + "grad_norm": 1.9187463521957397, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7551409602165222, + "num_tokens": 513028614.0, + "step": 20710 + }, + { + "epoch": 1.7078420054424013, + "grad_norm": 1.7800707817077637, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7774779796600342, + "num_tokens": 513059889.0, + "step": 20711 + }, + { + "epoch": 1.7079244660674529, + "grad_norm": 1.9408427476882935, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7389110922813416, + "num_tokens": 513090489.0, + "step": 20712 + }, + { + "epoch": 1.7080069266925042, + "grad_norm": 1.7200522422790527, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.777715802192688, + "num_tokens": 513126569.0, + "step": 20713 + }, + { + "epoch": 1.708089387317556, + "grad_norm": 2.151728391647339, + "learning_rate": 1e-06, + "loss": 0.5693, + "mean_token_accuracy": 0.8059749007225037, + "num_tokens": 513150211.0, + "step": 20714 + }, + { + "epoch": 1.7081718479426073, + "grad_norm": 2.197277784347534, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7702761292457581, + "num_tokens": 513173341.0, + "step": 20715 + }, + { + "epoch": 1.708254308567659, + "grad_norm": 1.732326626777649, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7360760569572449, + "num_tokens": 513213581.0, + "step": 20716 + }, + { + "epoch": 1.7083367691927105, + "grad_norm": 1.9857314825057983, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.763826310634613, + "num_tokens": 513242063.0, + "step": 20717 + }, + { + "epoch": 1.708419229817762, + "grad_norm": 2.0521323680877686, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7922368049621582, + "num_tokens": 513269401.0, + "step": 20718 + }, + { + "epoch": 1.7085016904428136, + "grad_norm": 2.2304675579071045, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.772283673286438, + "num_tokens": 513293990.0, + "step": 20719 + }, + { + "epoch": 1.708584151067865, + "grad_norm": 2.2173690795898438, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7402619123458862, + "num_tokens": 513319999.0, + "step": 20720 + }, + { + "epoch": 1.7086666116929168, + "grad_norm": 2.250995635986328, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7632023096084595, + "num_tokens": 513342618.0, + "step": 20721 + }, + { + "epoch": 1.7087490723179681, + "grad_norm": 2.0649771690368652, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7343748807907104, + "num_tokens": 513373162.0, + "step": 20722 + }, + { + "epoch": 1.7088315329430197, + "grad_norm": 2.0210483074188232, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.721994161605835, + "num_tokens": 513402848.0, + "step": 20723 + }, + { + "epoch": 1.7089139935680713, + "grad_norm": 2.033799171447754, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7526793479919434, + "num_tokens": 513431475.0, + "step": 20724 + }, + { + "epoch": 1.7089964541931226, + "grad_norm": 1.963721752166748, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7202035188674927, + "num_tokens": 513459698.0, + "step": 20725 + }, + { + "epoch": 1.7090789148181744, + "grad_norm": 1.8630561828613281, + "learning_rate": 1e-06, + "loss": 0.7887, + "mean_token_accuracy": 0.750384509563446, + "num_tokens": 513491421.0, + "step": 20726 + }, + { + "epoch": 1.7091613754432258, + "grad_norm": 1.87619149684906, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.7523040175437927, + "num_tokens": 513526077.0, + "step": 20727 + }, + { + "epoch": 1.7092438360682773, + "grad_norm": 2.0750529766082764, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7530933022499084, + "num_tokens": 513554367.0, + "step": 20728 + }, + { + "epoch": 1.709326296693329, + "grad_norm": 2.122204303741455, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7657336592674255, + "num_tokens": 513580956.0, + "step": 20729 + }, + { + "epoch": 1.7094087573183805, + "grad_norm": 1.9764516353607178, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7754555344581604, + "num_tokens": 513609638.0, + "step": 20730 + }, + { + "epoch": 1.709491217943432, + "grad_norm": 2.041473150253296, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.7896528840065002, + "num_tokens": 513636213.0, + "step": 20731 + }, + { + "epoch": 1.7095736785684834, + "grad_norm": 1.9013431072235107, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.737632155418396, + "num_tokens": 513670562.0, + "step": 20732 + }, + { + "epoch": 1.7096561391935352, + "grad_norm": 1.9814033508300781, + "learning_rate": 1e-06, + "loss": 0.5682, + "mean_token_accuracy": 0.815711498260498, + "num_tokens": 513699740.0, + "step": 20733 + }, + { + "epoch": 1.7097385998185866, + "grad_norm": 1.9841586351394653, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.7880204916000366, + "num_tokens": 513727870.0, + "step": 20734 + }, + { + "epoch": 1.7098210604436381, + "grad_norm": 1.7897651195526123, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.7828032970428467, + "num_tokens": 513761514.0, + "step": 20735 + }, + { + "epoch": 1.7099035210686897, + "grad_norm": 2.0547871589660645, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7318898439407349, + "num_tokens": 513791333.0, + "step": 20736 + }, + { + "epoch": 1.7099859816937413, + "grad_norm": 1.9690680503845215, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7704190015792847, + "num_tokens": 513822401.0, + "step": 20737 + }, + { + "epoch": 1.7100684423187928, + "grad_norm": 2.139279842376709, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7168681621551514, + "num_tokens": 513852284.0, + "step": 20738 + }, + { + "epoch": 1.7101509029438442, + "grad_norm": 1.8645992279052734, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.753321647644043, + "num_tokens": 513887691.0, + "step": 20739 + }, + { + "epoch": 1.710233363568896, + "grad_norm": 1.9562900066375732, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7604527473449707, + "num_tokens": 513916145.0, + "step": 20740 + }, + { + "epoch": 1.7103158241939473, + "grad_norm": 2.04022216796875, + "learning_rate": 1e-06, + "loss": 0.5925, + "mean_token_accuracy": 0.7997422814369202, + "num_tokens": 513945236.0, + "step": 20741 + }, + { + "epoch": 1.710398284818999, + "grad_norm": 2.2139604091644287, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7515865564346313, + "num_tokens": 513974694.0, + "step": 20742 + }, + { + "epoch": 1.7104807454440505, + "grad_norm": 2.1180710792541504, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7728033065795898, + "num_tokens": 514001562.0, + "step": 20743 + }, + { + "epoch": 1.7105632060691018, + "grad_norm": 2.2271132469177246, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.7900971174240112, + "num_tokens": 514025148.0, + "step": 20744 + }, + { + "epoch": 1.7106456666941536, + "grad_norm": 1.732622742652893, + "learning_rate": 1e-06, + "loss": 0.5824, + "mean_token_accuracy": 0.8076568841934204, + "num_tokens": 514057072.0, + "step": 20745 + }, + { + "epoch": 1.710728127319205, + "grad_norm": 1.7694092988967896, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7799646258354187, + "num_tokens": 514093528.0, + "step": 20746 + }, + { + "epoch": 1.7108105879442568, + "grad_norm": 1.9558411836624146, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7337052822113037, + "num_tokens": 514130854.0, + "step": 20747 + }, + { + "epoch": 1.7108930485693081, + "grad_norm": 2.135867118835449, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7500089406967163, + "num_tokens": 514161050.0, + "step": 20748 + }, + { + "epoch": 1.7109755091943597, + "grad_norm": 1.7871685028076172, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7630940079689026, + "num_tokens": 514198799.0, + "step": 20749 + }, + { + "epoch": 1.7110579698194113, + "grad_norm": 1.791023850440979, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7810521125793457, + "num_tokens": 514236959.0, + "step": 20750 + }, + { + "epoch": 1.7111404304444626, + "grad_norm": 1.957902193069458, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.7595767974853516, + "num_tokens": 514268921.0, + "step": 20751 + }, + { + "epoch": 1.7112228910695144, + "grad_norm": 2.2492828369140625, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7683281898498535, + "num_tokens": 514294067.0, + "step": 20752 + }, + { + "epoch": 1.7113053516945658, + "grad_norm": 1.961406946182251, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7688382863998413, + "num_tokens": 514323412.0, + "step": 20753 + }, + { + "epoch": 1.7113878123196173, + "grad_norm": 1.8590446710586548, + "learning_rate": 1e-06, + "loss": 0.5699, + "mean_token_accuracy": 0.8116552829742432, + "num_tokens": 514355384.0, + "step": 20754 + }, + { + "epoch": 1.711470272944669, + "grad_norm": 1.9178414344787598, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.7873514890670776, + "num_tokens": 514388207.0, + "step": 20755 + }, + { + "epoch": 1.7115527335697205, + "grad_norm": 1.7926244735717773, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7794277667999268, + "num_tokens": 514426296.0, + "step": 20756 + }, + { + "epoch": 1.711635194194772, + "grad_norm": 1.9313082695007324, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7540735602378845, + "num_tokens": 514457796.0, + "step": 20757 + }, + { + "epoch": 1.7117176548198234, + "grad_norm": 1.932189702987671, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.7791701555252075, + "num_tokens": 514491871.0, + "step": 20758 + }, + { + "epoch": 1.7118001154448752, + "grad_norm": 1.9867475032806396, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7688442468643188, + "num_tokens": 514519883.0, + "step": 20759 + }, + { + "epoch": 1.7118825760699266, + "grad_norm": 2.051825523376465, + "learning_rate": 1e-06, + "loss": 0.7722, + "mean_token_accuracy": 0.7675663232803345, + "num_tokens": 514547131.0, + "step": 20760 + }, + { + "epoch": 1.7119650366949781, + "grad_norm": 2.080390691757202, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7335466742515564, + "num_tokens": 514576911.0, + "step": 20761 + }, + { + "epoch": 1.7120474973200297, + "grad_norm": 2.0092568397521973, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7809397578239441, + "num_tokens": 514604893.0, + "step": 20762 + }, + { + "epoch": 1.7121299579450813, + "grad_norm": 2.0376620292663574, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7774361371994019, + "num_tokens": 514630424.0, + "step": 20763 + }, + { + "epoch": 1.7122124185701328, + "grad_norm": 2.5542259216308594, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7331349849700928, + "num_tokens": 514649520.0, + "step": 20764 + }, + { + "epoch": 1.7122948791951842, + "grad_norm": 2.218994379043579, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7495154142379761, + "num_tokens": 514675536.0, + "step": 20765 + }, + { + "epoch": 1.712377339820236, + "grad_norm": 1.7287496328353882, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7709310054779053, + "num_tokens": 514714982.0, + "step": 20766 + }, + { + "epoch": 1.7124598004452873, + "grad_norm": 1.9392439126968384, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.7856911420822144, + "num_tokens": 514742322.0, + "step": 20767 + }, + { + "epoch": 1.712542261070339, + "grad_norm": 1.9457776546478271, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.7835944890975952, + "num_tokens": 514775603.0, + "step": 20768 + }, + { + "epoch": 1.7126247216953905, + "grad_norm": 1.8352957963943481, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7016618251800537, + "num_tokens": 514812198.0, + "step": 20769 + }, + { + "epoch": 1.7127071823204418, + "grad_norm": 1.9328033924102783, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7375457882881165, + "num_tokens": 514844170.0, + "step": 20770 + }, + { + "epoch": 1.7127896429454936, + "grad_norm": 2.04005765914917, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7552691698074341, + "num_tokens": 514870886.0, + "step": 20771 + }, + { + "epoch": 1.712872103570545, + "grad_norm": 1.850579023361206, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7780578136444092, + "num_tokens": 514904155.0, + "step": 20772 + }, + { + "epoch": 1.7129545641955966, + "grad_norm": 2.196394681930542, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7718549370765686, + "num_tokens": 514931292.0, + "step": 20773 + }, + { + "epoch": 1.7130370248206481, + "grad_norm": 2.3188984394073486, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7402011156082153, + "num_tokens": 514954528.0, + "step": 20774 + }, + { + "epoch": 1.7131194854456997, + "grad_norm": 1.705040693283081, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.7894244194030762, + "num_tokens": 514992396.0, + "step": 20775 + }, + { + "epoch": 1.7132019460707513, + "grad_norm": 1.7902508974075317, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7757593393325806, + "num_tokens": 515026817.0, + "step": 20776 + }, + { + "epoch": 1.7132844066958026, + "grad_norm": 1.8489505052566528, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7510803937911987, + "num_tokens": 515060857.0, + "step": 20777 + }, + { + "epoch": 1.7133668673208544, + "grad_norm": 2.1787261962890625, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7223601341247559, + "num_tokens": 515087159.0, + "step": 20778 + }, + { + "epoch": 1.7134493279459058, + "grad_norm": 2.004159688949585, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.7924286127090454, + "num_tokens": 515116802.0, + "step": 20779 + }, + { + "epoch": 1.7135317885709573, + "grad_norm": 2.1173675060272217, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.722902774810791, + "num_tokens": 515146000.0, + "step": 20780 + }, + { + "epoch": 1.713614249196009, + "grad_norm": 1.8480541706085205, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7270517349243164, + "num_tokens": 515181952.0, + "step": 20781 + }, + { + "epoch": 1.7136967098210605, + "grad_norm": 2.084077835083008, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7698558568954468, + "num_tokens": 515210179.0, + "step": 20782 + }, + { + "epoch": 1.713779170446112, + "grad_norm": 1.8723505735397339, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7358022928237915, + "num_tokens": 515243130.0, + "step": 20783 + }, + { + "epoch": 1.7138616310711634, + "grad_norm": 2.1105427742004395, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7338153123855591, + "num_tokens": 515271300.0, + "step": 20784 + }, + { + "epoch": 1.7139440916962152, + "grad_norm": 1.9409490823745728, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.7906444668769836, + "num_tokens": 515302366.0, + "step": 20785 + }, + { + "epoch": 1.7140265523212665, + "grad_norm": 1.9089176654815674, + "learning_rate": 1e-06, + "loss": 0.6063, + "mean_token_accuracy": 0.7953051328659058, + "num_tokens": 515333025.0, + "step": 20786 + }, + { + "epoch": 1.7141090129463181, + "grad_norm": 1.7287981510162354, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7769221663475037, + "num_tokens": 515370378.0, + "step": 20787 + }, + { + "epoch": 1.7141914735713697, + "grad_norm": 2.066936731338501, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.7564488053321838, + "num_tokens": 515398766.0, + "step": 20788 + }, + { + "epoch": 1.714273934196421, + "grad_norm": 1.9390957355499268, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7716149687767029, + "num_tokens": 515429470.0, + "step": 20789 + }, + { + "epoch": 1.7143563948214728, + "grad_norm": 2.05647349357605, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7381781935691833, + "num_tokens": 515459486.0, + "step": 20790 + }, + { + "epoch": 1.7144388554465242, + "grad_norm": 2.2561309337615967, + "learning_rate": 1e-06, + "loss": 0.5436, + "mean_token_accuracy": 0.8004577159881592, + "num_tokens": 515481784.0, + "step": 20791 + }, + { + "epoch": 1.714521316071576, + "grad_norm": 2.013917922973633, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7660029530525208, + "num_tokens": 515513502.0, + "step": 20792 + }, + { + "epoch": 1.7146037766966273, + "grad_norm": 1.9854109287261963, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.7766023874282837, + "num_tokens": 515539689.0, + "step": 20793 + }, + { + "epoch": 1.714686237321679, + "grad_norm": 1.9935154914855957, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7637979984283447, + "num_tokens": 515569363.0, + "step": 20794 + }, + { + "epoch": 1.7147686979467305, + "grad_norm": 2.273491621017456, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7065776586532593, + "num_tokens": 515594828.0, + "step": 20795 + }, + { + "epoch": 1.7148511585717818, + "grad_norm": 1.9461545944213867, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.78291916847229, + "num_tokens": 515626337.0, + "step": 20796 + }, + { + "epoch": 1.7149336191968336, + "grad_norm": 1.9921579360961914, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7866389751434326, + "num_tokens": 515655032.0, + "step": 20797 + }, + { + "epoch": 1.715016079821885, + "grad_norm": 2.263887882232666, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7496381998062134, + "num_tokens": 515679897.0, + "step": 20798 + }, + { + "epoch": 1.7150985404469365, + "grad_norm": 2.3438408374786377, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7468823194503784, + "num_tokens": 515706697.0, + "step": 20799 + }, + { + "epoch": 1.7151810010719881, + "grad_norm": 1.9896621704101562, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7693289518356323, + "num_tokens": 515739595.0, + "step": 20800 + }, + { + "epoch": 1.7152634616970397, + "grad_norm": 2.32442307472229, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7748591303825378, + "num_tokens": 515761960.0, + "step": 20801 + }, + { + "epoch": 1.7153459223220913, + "grad_norm": 2.2830796241760254, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7534014582633972, + "num_tokens": 515785857.0, + "step": 20802 + }, + { + "epoch": 1.7154283829471426, + "grad_norm": 1.795810341835022, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.745927095413208, + "num_tokens": 515821886.0, + "step": 20803 + }, + { + "epoch": 1.7155108435721944, + "grad_norm": 1.7158924341201782, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7519428133964539, + "num_tokens": 515858758.0, + "step": 20804 + }, + { + "epoch": 1.7155933041972458, + "grad_norm": 1.8674758672714233, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7482763528823853, + "num_tokens": 515894131.0, + "step": 20805 + }, + { + "epoch": 1.7156757648222973, + "grad_norm": 1.8204056024551392, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7512695789337158, + "num_tokens": 515929386.0, + "step": 20806 + }, + { + "epoch": 1.715758225447349, + "grad_norm": 1.8105993270874023, + "learning_rate": 1e-06, + "loss": 0.5887, + "mean_token_accuracy": 0.8051201105117798, + "num_tokens": 515960840.0, + "step": 20807 + }, + { + "epoch": 1.7158406860724005, + "grad_norm": 2.2226123809814453, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.774194061756134, + "num_tokens": 515985818.0, + "step": 20808 + }, + { + "epoch": 1.715923146697452, + "grad_norm": 2.138627052307129, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.7891423106193542, + "num_tokens": 516010754.0, + "step": 20809 + }, + { + "epoch": 1.7160056073225034, + "grad_norm": 1.9813138246536255, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.794140100479126, + "num_tokens": 516036356.0, + "step": 20810 + }, + { + "epoch": 1.7160880679475552, + "grad_norm": 1.7703477144241333, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7300873398780823, + "num_tokens": 516072611.0, + "step": 20811 + }, + { + "epoch": 1.7161705285726065, + "grad_norm": 1.9046074151992798, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7625768184661865, + "num_tokens": 516105989.0, + "step": 20812 + }, + { + "epoch": 1.7162529891976581, + "grad_norm": 1.98228919506073, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.77045738697052, + "num_tokens": 516137386.0, + "step": 20813 + }, + { + "epoch": 1.7163354498227097, + "grad_norm": 2.1943929195404053, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7597242593765259, + "num_tokens": 516163346.0, + "step": 20814 + }, + { + "epoch": 1.716417910447761, + "grad_norm": 2.0616822242736816, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7438512444496155, + "num_tokens": 516191355.0, + "step": 20815 + }, + { + "epoch": 1.7165003710728128, + "grad_norm": 1.9504551887512207, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.783627450466156, + "num_tokens": 516223245.0, + "step": 20816 + }, + { + "epoch": 1.7165828316978642, + "grad_norm": 1.999248743057251, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7507917284965515, + "num_tokens": 516252753.0, + "step": 20817 + }, + { + "epoch": 1.7166652923229158, + "grad_norm": 1.8303402662277222, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.7819125652313232, + "num_tokens": 516286674.0, + "step": 20818 + }, + { + "epoch": 1.7167477529479673, + "grad_norm": 1.894680142402649, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7096825838088989, + "num_tokens": 516320043.0, + "step": 20819 + }, + { + "epoch": 1.716830213573019, + "grad_norm": 1.8583812713623047, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7506457567214966, + "num_tokens": 516355771.0, + "step": 20820 + }, + { + "epoch": 1.7169126741980705, + "grad_norm": 1.6543138027191162, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.7888299226760864, + "num_tokens": 516397470.0, + "step": 20821 + }, + { + "epoch": 1.7169951348231218, + "grad_norm": 2.093944549560547, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.787386417388916, + "num_tokens": 516426817.0, + "step": 20822 + }, + { + "epoch": 1.7170775954481736, + "grad_norm": 2.8121719360351562, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7104902267456055, + "num_tokens": 516446079.0, + "step": 20823 + }, + { + "epoch": 1.717160056073225, + "grad_norm": 2.047239065170288, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7720310688018799, + "num_tokens": 516475670.0, + "step": 20824 + }, + { + "epoch": 1.7172425166982765, + "grad_norm": 1.699710488319397, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.7943984866142273, + "num_tokens": 516512814.0, + "step": 20825 + }, + { + "epoch": 1.7173249773233281, + "grad_norm": 2.2148191928863525, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7246356010437012, + "num_tokens": 516537127.0, + "step": 20826 + }, + { + "epoch": 1.7174074379483797, + "grad_norm": 1.6410537958145142, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7728619575500488, + "num_tokens": 516577540.0, + "step": 20827 + }, + { + "epoch": 1.7174898985734313, + "grad_norm": 1.8696166276931763, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.787931501865387, + "num_tokens": 516610788.0, + "step": 20828 + }, + { + "epoch": 1.7175723591984826, + "grad_norm": 2.1552228927612305, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7607320547103882, + "num_tokens": 516636900.0, + "step": 20829 + }, + { + "epoch": 1.7176548198235344, + "grad_norm": 2.1162219047546387, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7566995024681091, + "num_tokens": 516665948.0, + "step": 20830 + }, + { + "epoch": 1.7177372804485858, + "grad_norm": 1.8510433435440063, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.7938887476921082, + "num_tokens": 516697695.0, + "step": 20831 + }, + { + "epoch": 1.7178197410736373, + "grad_norm": 1.6783744096755981, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.7792381644248962, + "num_tokens": 516735829.0, + "step": 20832 + }, + { + "epoch": 1.717902201698689, + "grad_norm": 2.143826484680176, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7433583736419678, + "num_tokens": 516762180.0, + "step": 20833 + }, + { + "epoch": 1.7179846623237403, + "grad_norm": 2.0009477138519287, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7400025725364685, + "num_tokens": 516792741.0, + "step": 20834 + }, + { + "epoch": 1.718067122948792, + "grad_norm": 1.8778616189956665, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.738262414932251, + "num_tokens": 516827354.0, + "step": 20835 + }, + { + "epoch": 1.7181495835738434, + "grad_norm": 2.022955894470215, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7279720306396484, + "num_tokens": 516856957.0, + "step": 20836 + }, + { + "epoch": 1.7182320441988952, + "grad_norm": 1.666653037071228, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.7983941435813904, + "num_tokens": 516894661.0, + "step": 20837 + }, + { + "epoch": 1.7183145048239465, + "grad_norm": 2.0347182750701904, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.751408576965332, + "num_tokens": 516923944.0, + "step": 20838 + }, + { + "epoch": 1.7183969654489981, + "grad_norm": 1.6659537553787231, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.7960216999053955, + "num_tokens": 516959855.0, + "step": 20839 + }, + { + "epoch": 1.7184794260740497, + "grad_norm": 2.1226186752319336, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7712622880935669, + "num_tokens": 516985334.0, + "step": 20840 + }, + { + "epoch": 1.718561886699101, + "grad_norm": 1.907781720161438, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7581773996353149, + "num_tokens": 517018375.0, + "step": 20841 + }, + { + "epoch": 1.7186443473241528, + "grad_norm": 2.259381055831909, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7701238393783569, + "num_tokens": 517042449.0, + "step": 20842 + }, + { + "epoch": 1.7187268079492042, + "grad_norm": 2.0071864128112793, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7696949243545532, + "num_tokens": 517074324.0, + "step": 20843 + }, + { + "epoch": 1.7188092685742558, + "grad_norm": 1.9315849542617798, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.7909791469573975, + "num_tokens": 517104137.0, + "step": 20844 + }, + { + "epoch": 1.7188917291993073, + "grad_norm": 2.157336473464966, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7796522378921509, + "num_tokens": 517127902.0, + "step": 20845 + }, + { + "epoch": 1.718974189824359, + "grad_norm": 1.8853302001953125, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7743405103683472, + "num_tokens": 517158866.0, + "step": 20846 + }, + { + "epoch": 1.7190566504494105, + "grad_norm": 2.0329315662384033, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7306699752807617, + "num_tokens": 517189974.0, + "step": 20847 + }, + { + "epoch": 1.7191391110744618, + "grad_norm": 2.009194850921631, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7542853355407715, + "num_tokens": 517222240.0, + "step": 20848 + }, + { + "epoch": 1.7192215716995136, + "grad_norm": 2.119011878967285, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7620721459388733, + "num_tokens": 517249644.0, + "step": 20849 + }, + { + "epoch": 1.719304032324565, + "grad_norm": 2.5577855110168457, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7694475650787354, + "num_tokens": 517269587.0, + "step": 20850 + }, + { + "epoch": 1.7193864929496165, + "grad_norm": 2.1534223556518555, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7467526793479919, + "num_tokens": 517295189.0, + "step": 20851 + }, + { + "epoch": 1.7194689535746681, + "grad_norm": 1.8277698755264282, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7895315885543823, + "num_tokens": 517327973.0, + "step": 20852 + }, + { + "epoch": 1.7195514141997197, + "grad_norm": 1.8906452655792236, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7708289623260498, + "num_tokens": 517360385.0, + "step": 20853 + }, + { + "epoch": 1.7196338748247713, + "grad_norm": 2.0193259716033936, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7742651700973511, + "num_tokens": 517389045.0, + "step": 20854 + }, + { + "epoch": 1.7197163354498226, + "grad_norm": 2.0294244289398193, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7337685823440552, + "num_tokens": 517419820.0, + "step": 20855 + }, + { + "epoch": 1.7197987960748744, + "grad_norm": 1.9745553731918335, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7449468374252319, + "num_tokens": 517448649.0, + "step": 20856 + }, + { + "epoch": 1.7198812566999258, + "grad_norm": 2.010702133178711, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7452898621559143, + "num_tokens": 517477597.0, + "step": 20857 + }, + { + "epoch": 1.7199637173249773, + "grad_norm": 2.4395604133605957, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7235006093978882, + "num_tokens": 517499590.0, + "step": 20858 + }, + { + "epoch": 1.720046177950029, + "grad_norm": 1.938131332397461, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7757931351661682, + "num_tokens": 517530875.0, + "step": 20859 + }, + { + "epoch": 1.7201286385750802, + "grad_norm": 2.0344350337982178, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7462344765663147, + "num_tokens": 517559889.0, + "step": 20860 + }, + { + "epoch": 1.720211099200132, + "grad_norm": 2.238358497619629, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7754303216934204, + "num_tokens": 517584210.0, + "step": 20861 + }, + { + "epoch": 1.7202935598251834, + "grad_norm": 2.026088237762451, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7395569682121277, + "num_tokens": 517613035.0, + "step": 20862 + }, + { + "epoch": 1.720376020450235, + "grad_norm": 1.8752738237380981, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7718719244003296, + "num_tokens": 517644800.0, + "step": 20863 + }, + { + "epoch": 1.7204584810752865, + "grad_norm": 1.89553964138031, + "learning_rate": 1e-06, + "loss": 0.5744, + "mean_token_accuracy": 0.8116264939308167, + "num_tokens": 517675468.0, + "step": 20864 + }, + { + "epoch": 1.7205409417003381, + "grad_norm": 1.9553431272506714, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7775965332984924, + "num_tokens": 517707465.0, + "step": 20865 + }, + { + "epoch": 1.7206234023253897, + "grad_norm": 1.794553518295288, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7682827115058899, + "num_tokens": 517742122.0, + "step": 20866 + }, + { + "epoch": 1.720705862950441, + "grad_norm": 2.0951695442199707, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7798649668693542, + "num_tokens": 517770652.0, + "step": 20867 + }, + { + "epoch": 1.7207883235754928, + "grad_norm": 2.192092180252075, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7688099145889282, + "num_tokens": 517797284.0, + "step": 20868 + }, + { + "epoch": 1.7208707842005442, + "grad_norm": 1.9620245695114136, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7512311339378357, + "num_tokens": 517827617.0, + "step": 20869 + }, + { + "epoch": 1.7209532448255958, + "grad_norm": 2.089716672897339, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7837092280387878, + "num_tokens": 517853531.0, + "step": 20870 + }, + { + "epoch": 1.7210357054506473, + "grad_norm": 2.1580772399902344, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7459954619407654, + "num_tokens": 517882080.0, + "step": 20871 + }, + { + "epoch": 1.721118166075699, + "grad_norm": 2.3208818435668945, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7745862603187561, + "num_tokens": 517903966.0, + "step": 20872 + }, + { + "epoch": 1.7212006267007505, + "grad_norm": 1.8495131731033325, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7400835752487183, + "num_tokens": 517940847.0, + "step": 20873 + }, + { + "epoch": 1.7212830873258018, + "grad_norm": 1.9117701053619385, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.7809867858886719, + "num_tokens": 517972787.0, + "step": 20874 + }, + { + "epoch": 1.7213655479508536, + "grad_norm": 1.8285648822784424, + "learning_rate": 1e-06, + "loss": 0.5898, + "mean_token_accuracy": 0.8073722124099731, + "num_tokens": 518003921.0, + "step": 20875 + }, + { + "epoch": 1.721448008575905, + "grad_norm": 1.891211986541748, + "learning_rate": 1e-06, + "loss": 0.5722, + "mean_token_accuracy": 0.8044067621231079, + "num_tokens": 518034902.0, + "step": 20876 + }, + { + "epoch": 1.7215304692009565, + "grad_norm": 2.0285704135894775, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.7876378893852234, + "num_tokens": 518063471.0, + "step": 20877 + }, + { + "epoch": 1.721612929826008, + "grad_norm": 1.951458215713501, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7388312220573425, + "num_tokens": 518095173.0, + "step": 20878 + }, + { + "epoch": 1.7216953904510595, + "grad_norm": 2.0581300258636475, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7783401608467102, + "num_tokens": 518123897.0, + "step": 20879 + }, + { + "epoch": 1.7217778510761113, + "grad_norm": 2.0144131183624268, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7512925863265991, + "num_tokens": 518154113.0, + "step": 20880 + }, + { + "epoch": 1.7218603117011626, + "grad_norm": 1.9711462259292603, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.7816430330276489, + "num_tokens": 518183169.0, + "step": 20881 + }, + { + "epoch": 1.7219427723262144, + "grad_norm": 2.210529088973999, + "learning_rate": 1e-06, + "loss": 0.5992, + "mean_token_accuracy": 0.7972419261932373, + "num_tokens": 518206604.0, + "step": 20882 + }, + { + "epoch": 1.7220252329512657, + "grad_norm": 1.820452332496643, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7930213809013367, + "num_tokens": 518241417.0, + "step": 20883 + }, + { + "epoch": 1.7221076935763173, + "grad_norm": 2.156048536300659, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.744268000125885, + "num_tokens": 518269491.0, + "step": 20884 + }, + { + "epoch": 1.722190154201369, + "grad_norm": 1.8722784519195557, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7743256092071533, + "num_tokens": 518304916.0, + "step": 20885 + }, + { + "epoch": 1.7222726148264202, + "grad_norm": 1.8972245454788208, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7448744773864746, + "num_tokens": 518337496.0, + "step": 20886 + }, + { + "epoch": 1.722355075451472, + "grad_norm": 2.2595293521881104, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7803776264190674, + "num_tokens": 518361964.0, + "step": 20887 + }, + { + "epoch": 1.7224375360765234, + "grad_norm": 1.745718002319336, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7552403211593628, + "num_tokens": 518400614.0, + "step": 20888 + }, + { + "epoch": 1.722519996701575, + "grad_norm": 1.6915631294250488, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7360382080078125, + "num_tokens": 518438674.0, + "step": 20889 + }, + { + "epoch": 1.7226024573266265, + "grad_norm": 2.0712456703186035, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7639176845550537, + "num_tokens": 518465173.0, + "step": 20890 + }, + { + "epoch": 1.722684917951678, + "grad_norm": 1.9802565574645996, + "learning_rate": 1e-06, + "loss": 0.7683, + "mean_token_accuracy": 0.7562114000320435, + "num_tokens": 518495588.0, + "step": 20891 + }, + { + "epoch": 1.7227673785767297, + "grad_norm": 2.1537036895751953, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7574533224105835, + "num_tokens": 518524000.0, + "step": 20892 + }, + { + "epoch": 1.722849839201781, + "grad_norm": 2.1800413131713867, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7368956208229065, + "num_tokens": 518550424.0, + "step": 20893 + }, + { + "epoch": 1.7229322998268328, + "grad_norm": 2.125812530517578, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7760370969772339, + "num_tokens": 518576966.0, + "step": 20894 + }, + { + "epoch": 1.7230147604518842, + "grad_norm": 2.004173755645752, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7631843090057373, + "num_tokens": 518608626.0, + "step": 20895 + }, + { + "epoch": 1.7230972210769357, + "grad_norm": 1.8860281705856323, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7801668643951416, + "num_tokens": 518640619.0, + "step": 20896 + }, + { + "epoch": 1.7231796817019873, + "grad_norm": 1.9120184183120728, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.7844010591506958, + "num_tokens": 518673120.0, + "step": 20897 + }, + { + "epoch": 1.723262142327039, + "grad_norm": 2.018660545349121, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.7870978116989136, + "num_tokens": 518699599.0, + "step": 20898 + }, + { + "epoch": 1.7233446029520905, + "grad_norm": 2.0548789501190186, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.7815157175064087, + "num_tokens": 518724611.0, + "step": 20899 + }, + { + "epoch": 1.7234270635771418, + "grad_norm": 1.916515588760376, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7553749084472656, + "num_tokens": 518756145.0, + "step": 20900 + }, + { + "epoch": 1.7235095242021936, + "grad_norm": 2.077773332595825, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7674539089202881, + "num_tokens": 518782515.0, + "step": 20901 + }, + { + "epoch": 1.723591984827245, + "grad_norm": 1.8077588081359863, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.7858538627624512, + "num_tokens": 518816856.0, + "step": 20902 + }, + { + "epoch": 1.7236744454522965, + "grad_norm": 1.8844335079193115, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7730860710144043, + "num_tokens": 518849807.0, + "step": 20903 + }, + { + "epoch": 1.723756906077348, + "grad_norm": 2.0802595615386963, + "learning_rate": 1e-06, + "loss": 0.6066, + "mean_token_accuracy": 0.7898856401443481, + "num_tokens": 518877282.0, + "step": 20904 + }, + { + "epoch": 1.7238393667023995, + "grad_norm": 2.071960210800171, + "learning_rate": 1e-06, + "loss": 0.6031, + "mean_token_accuracy": 0.8038797378540039, + "num_tokens": 518903431.0, + "step": 20905 + }, + { + "epoch": 1.7239218273274513, + "grad_norm": 2.221808910369873, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7477973699569702, + "num_tokens": 518928185.0, + "step": 20906 + }, + { + "epoch": 1.7240042879525026, + "grad_norm": 1.9262990951538086, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.7874838709831238, + "num_tokens": 518957973.0, + "step": 20907 + }, + { + "epoch": 1.7240867485775542, + "grad_norm": 1.8205058574676514, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7542080879211426, + "num_tokens": 518993418.0, + "step": 20908 + }, + { + "epoch": 1.7241692092026057, + "grad_norm": 1.8028842210769653, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.7950975894927979, + "num_tokens": 519028477.0, + "step": 20909 + }, + { + "epoch": 1.7242516698276573, + "grad_norm": 2.414975166320801, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7430533766746521, + "num_tokens": 519051369.0, + "step": 20910 + }, + { + "epoch": 1.724334130452709, + "grad_norm": 1.9374505281448364, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7564031481742859, + "num_tokens": 519082219.0, + "step": 20911 + }, + { + "epoch": 1.7244165910777602, + "grad_norm": 1.9425544738769531, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7773276567459106, + "num_tokens": 519114859.0, + "step": 20912 + }, + { + "epoch": 1.724499051702812, + "grad_norm": 1.8763911724090576, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7827834486961365, + "num_tokens": 519148081.0, + "step": 20913 + }, + { + "epoch": 1.7245815123278634, + "grad_norm": 1.751115322113037, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7506319880485535, + "num_tokens": 519186168.0, + "step": 20914 + }, + { + "epoch": 1.724663972952915, + "grad_norm": 1.8747750520706177, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.762793779373169, + "num_tokens": 519217995.0, + "step": 20915 + }, + { + "epoch": 1.7247464335779665, + "grad_norm": 2.0360324382781982, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7632690668106079, + "num_tokens": 519249029.0, + "step": 20916 + }, + { + "epoch": 1.724828894203018, + "grad_norm": 2.1360230445861816, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7743504047393799, + "num_tokens": 519274724.0, + "step": 20917 + }, + { + "epoch": 1.7249113548280697, + "grad_norm": 1.9178274869918823, + "learning_rate": 1e-06, + "loss": 0.6101, + "mean_token_accuracy": 0.7919384241104126, + "num_tokens": 519305370.0, + "step": 20918 + }, + { + "epoch": 1.724993815453121, + "grad_norm": 2.030221939086914, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7333043813705444, + "num_tokens": 519333817.0, + "step": 20919 + }, + { + "epoch": 1.7250762760781728, + "grad_norm": 2.168039321899414, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7111481428146362, + "num_tokens": 519361932.0, + "step": 20920 + }, + { + "epoch": 1.7251587367032242, + "grad_norm": 1.9403821229934692, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7523238062858582, + "num_tokens": 519392425.0, + "step": 20921 + }, + { + "epoch": 1.7252411973282757, + "grad_norm": 2.017587661743164, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7826207876205444, + "num_tokens": 519426302.0, + "step": 20922 + }, + { + "epoch": 1.7253236579533273, + "grad_norm": 2.109217643737793, + "learning_rate": 1e-06, + "loss": 0.5976, + "mean_token_accuracy": 0.8099225759506226, + "num_tokens": 519450185.0, + "step": 20923 + }, + { + "epoch": 1.7254061185783787, + "grad_norm": 2.3232967853546143, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7375774383544922, + "num_tokens": 519473728.0, + "step": 20924 + }, + { + "epoch": 1.7254885792034305, + "grad_norm": 1.8822994232177734, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.7873333692550659, + "num_tokens": 519505140.0, + "step": 20925 + }, + { + "epoch": 1.7255710398284818, + "grad_norm": 2.1204140186309814, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7488093376159668, + "num_tokens": 519533348.0, + "step": 20926 + }, + { + "epoch": 1.7256535004535334, + "grad_norm": 2.13503360748291, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7565655708312988, + "num_tokens": 519559252.0, + "step": 20927 + }, + { + "epoch": 1.725735961078585, + "grad_norm": 2.1403589248657227, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7475481033325195, + "num_tokens": 519587467.0, + "step": 20928 + }, + { + "epoch": 1.7258184217036365, + "grad_norm": 2.030881404876709, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.7885242104530334, + "num_tokens": 519617732.0, + "step": 20929 + }, + { + "epoch": 1.725900882328688, + "grad_norm": 1.9094792604446411, + "learning_rate": 1e-06, + "loss": 0.7775, + "mean_token_accuracy": 0.7560280561447144, + "num_tokens": 519651966.0, + "step": 20930 + }, + { + "epoch": 1.7259833429537395, + "grad_norm": 1.990607500076294, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7524756193161011, + "num_tokens": 519682430.0, + "step": 20931 + }, + { + "epoch": 1.7260658035787912, + "grad_norm": 1.9914765357971191, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7458418607711792, + "num_tokens": 519712049.0, + "step": 20932 + }, + { + "epoch": 1.7261482642038426, + "grad_norm": 1.9630892276763916, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.7869099974632263, + "num_tokens": 519741210.0, + "step": 20933 + }, + { + "epoch": 1.7262307248288942, + "grad_norm": 2.085300922393799, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.735034704208374, + "num_tokens": 519768964.0, + "step": 20934 + }, + { + "epoch": 1.7263131854539457, + "grad_norm": 2.0779032707214355, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7339937686920166, + "num_tokens": 519799195.0, + "step": 20935 + }, + { + "epoch": 1.7263956460789973, + "grad_norm": 1.9730204343795776, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7619067430496216, + "num_tokens": 519827998.0, + "step": 20936 + }, + { + "epoch": 1.7264781067040489, + "grad_norm": 2.148303747177124, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7268174886703491, + "num_tokens": 519857246.0, + "step": 20937 + }, + { + "epoch": 1.7265605673291002, + "grad_norm": 1.9182277917861938, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7792915105819702, + "num_tokens": 519888090.0, + "step": 20938 + }, + { + "epoch": 1.726643027954152, + "grad_norm": 2.1335949897766113, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7769870758056641, + "num_tokens": 519914159.0, + "step": 20939 + }, + { + "epoch": 1.7267254885792034, + "grad_norm": 1.9012088775634766, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.7911103367805481, + "num_tokens": 519942275.0, + "step": 20940 + }, + { + "epoch": 1.726807949204255, + "grad_norm": 2.131514072418213, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7345072031021118, + "num_tokens": 519969075.0, + "step": 20941 + }, + { + "epoch": 1.7268904098293065, + "grad_norm": 1.9134151935577393, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7581028342247009, + "num_tokens": 520001254.0, + "step": 20942 + }, + { + "epoch": 1.7269728704543579, + "grad_norm": 2.141340732574463, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7408053874969482, + "num_tokens": 520027544.0, + "step": 20943 + }, + { + "epoch": 1.7270553310794097, + "grad_norm": 1.738112449645996, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.7841891050338745, + "num_tokens": 520064451.0, + "step": 20944 + }, + { + "epoch": 1.727137791704461, + "grad_norm": 2.2515738010406494, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7343195080757141, + "num_tokens": 520088495.0, + "step": 20945 + }, + { + "epoch": 1.7272202523295128, + "grad_norm": 1.9648381471633911, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7369598150253296, + "num_tokens": 520121033.0, + "step": 20946 + }, + { + "epoch": 1.7273027129545642, + "grad_norm": 1.9296350479125977, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7350577116012573, + "num_tokens": 520155118.0, + "step": 20947 + }, + { + "epoch": 1.7273851735796157, + "grad_norm": 1.852647066116333, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.7793173789978027, + "num_tokens": 520187360.0, + "step": 20948 + }, + { + "epoch": 1.7274676342046673, + "grad_norm": 1.8164838552474976, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7190417051315308, + "num_tokens": 520224128.0, + "step": 20949 + }, + { + "epoch": 1.7275500948297187, + "grad_norm": 2.0227723121643066, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7386350631713867, + "num_tokens": 520252462.0, + "step": 20950 + }, + { + "epoch": 1.7276325554547705, + "grad_norm": 1.8437381982803345, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7447112798690796, + "num_tokens": 520287178.0, + "step": 20951 + }, + { + "epoch": 1.7277150160798218, + "grad_norm": 2.160006046295166, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.741519033908844, + "num_tokens": 520315073.0, + "step": 20952 + }, + { + "epoch": 1.7277974767048734, + "grad_norm": 2.0394582748413086, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7451802492141724, + "num_tokens": 520343744.0, + "step": 20953 + }, + { + "epoch": 1.727879937329925, + "grad_norm": 2.1862847805023193, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7758142948150635, + "num_tokens": 520368427.0, + "step": 20954 + }, + { + "epoch": 1.7279623979549765, + "grad_norm": 1.8959804773330688, + "learning_rate": 1e-06, + "loss": 0.5528, + "mean_token_accuracy": 0.8147941827774048, + "num_tokens": 520397470.0, + "step": 20955 + }, + { + "epoch": 1.728044858580028, + "grad_norm": 1.8882027864456177, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7758187651634216, + "num_tokens": 520429851.0, + "step": 20956 + }, + { + "epoch": 1.7281273192050794, + "grad_norm": 2.0799930095672607, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7572904825210571, + "num_tokens": 520457606.0, + "step": 20957 + }, + { + "epoch": 1.7282097798301312, + "grad_norm": 1.9693982601165771, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.788833737373352, + "num_tokens": 520487816.0, + "step": 20958 + }, + { + "epoch": 1.7282922404551826, + "grad_norm": 2.1133315563201904, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7467703819274902, + "num_tokens": 520516691.0, + "step": 20959 + }, + { + "epoch": 1.7283747010802342, + "grad_norm": 1.856291651725769, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7747374773025513, + "num_tokens": 520547551.0, + "step": 20960 + }, + { + "epoch": 1.7284571617052857, + "grad_norm": 2.1041476726531982, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7817708253860474, + "num_tokens": 520575110.0, + "step": 20961 + }, + { + "epoch": 1.7285396223303373, + "grad_norm": 2.200822353363037, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7792811393737793, + "num_tokens": 520600899.0, + "step": 20962 + }, + { + "epoch": 1.7286220829553889, + "grad_norm": 2.21185564994812, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7175570726394653, + "num_tokens": 520626411.0, + "step": 20963 + }, + { + "epoch": 1.7287045435804402, + "grad_norm": 1.9812631607055664, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.769507646560669, + "num_tokens": 520654198.0, + "step": 20964 + }, + { + "epoch": 1.728787004205492, + "grad_norm": 1.8034076690673828, + "learning_rate": 1e-06, + "loss": 0.6024, + "mean_token_accuracy": 0.8008050322532654, + "num_tokens": 520691141.0, + "step": 20965 + }, + { + "epoch": 1.7288694648305434, + "grad_norm": 1.8141086101531982, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.7960825562477112, + "num_tokens": 520725914.0, + "step": 20966 + }, + { + "epoch": 1.728951925455595, + "grad_norm": 2.273157835006714, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7369089126586914, + "num_tokens": 520751445.0, + "step": 20967 + }, + { + "epoch": 1.7290343860806465, + "grad_norm": 2.2871499061584473, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7622174024581909, + "num_tokens": 520776378.0, + "step": 20968 + }, + { + "epoch": 1.7291168467056979, + "grad_norm": 1.9365721940994263, + "learning_rate": 1e-06, + "loss": 0.5556, + "mean_token_accuracy": 0.8042916059494019, + "num_tokens": 520804557.0, + "step": 20969 + }, + { + "epoch": 1.7291993073307497, + "grad_norm": 1.9297187328338623, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7526082396507263, + "num_tokens": 520836008.0, + "step": 20970 + }, + { + "epoch": 1.729281767955801, + "grad_norm": 1.9628510475158691, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7442915439605713, + "num_tokens": 520868662.0, + "step": 20971 + }, + { + "epoch": 1.7293642285808526, + "grad_norm": 1.8933593034744263, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.7919701337814331, + "num_tokens": 520901603.0, + "step": 20972 + }, + { + "epoch": 1.7294466892059042, + "grad_norm": 1.9654347896575928, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7827948331832886, + "num_tokens": 520928540.0, + "step": 20973 + }, + { + "epoch": 1.7295291498309557, + "grad_norm": 2.219858169555664, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.7829084396362305, + "num_tokens": 520952268.0, + "step": 20974 + }, + { + "epoch": 1.7296116104560073, + "grad_norm": 2.252574920654297, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7441089749336243, + "num_tokens": 520978643.0, + "step": 20975 + }, + { + "epoch": 1.7296940710810587, + "grad_norm": 1.9518479108810425, + "learning_rate": 1e-06, + "loss": 0.5289, + "mean_token_accuracy": 0.8250546455383301, + "num_tokens": 521007595.0, + "step": 20976 + }, + { + "epoch": 1.7297765317061105, + "grad_norm": 1.7758675813674927, + "learning_rate": 1e-06, + "loss": 0.6061, + "mean_token_accuracy": 0.8075952529907227, + "num_tokens": 521040173.0, + "step": 20977 + }, + { + "epoch": 1.7298589923311618, + "grad_norm": 1.8524906635284424, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7695585489273071, + "num_tokens": 521073060.0, + "step": 20978 + }, + { + "epoch": 1.7299414529562134, + "grad_norm": 2.177474021911621, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7321611642837524, + "num_tokens": 521100296.0, + "step": 20979 + }, + { + "epoch": 1.730023913581265, + "grad_norm": 2.1461164951324463, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.76579749584198, + "num_tokens": 521126141.0, + "step": 20980 + }, + { + "epoch": 1.7301063742063165, + "grad_norm": 2.0084710121154785, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7332525849342346, + "num_tokens": 521158710.0, + "step": 20981 + }, + { + "epoch": 1.730188834831368, + "grad_norm": 1.785789132118225, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7802348136901855, + "num_tokens": 521198651.0, + "step": 20982 + }, + { + "epoch": 1.7302712954564194, + "grad_norm": 2.001523494720459, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7628422975540161, + "num_tokens": 521229284.0, + "step": 20983 + }, + { + "epoch": 1.7303537560814712, + "grad_norm": 1.6325445175170898, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.7843614816665649, + "num_tokens": 521265971.0, + "step": 20984 + }, + { + "epoch": 1.7304362167065226, + "grad_norm": 1.9265657663345337, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.7921361923217773, + "num_tokens": 521295766.0, + "step": 20985 + }, + { + "epoch": 1.7305186773315742, + "grad_norm": 2.241703748703003, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7605085372924805, + "num_tokens": 521320579.0, + "step": 20986 + }, + { + "epoch": 1.7306011379566257, + "grad_norm": 2.2410285472869873, + "learning_rate": 1e-06, + "loss": 0.5792, + "mean_token_accuracy": 0.8016650080680847, + "num_tokens": 521345406.0, + "step": 20987 + }, + { + "epoch": 1.730683598581677, + "grad_norm": 1.7803914546966553, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.7853167057037354, + "num_tokens": 521381189.0, + "step": 20988 + }, + { + "epoch": 1.7307660592067289, + "grad_norm": 1.6118505001068115, + "learning_rate": 1e-06, + "loss": 0.5221, + "mean_token_accuracy": 0.8206979632377625, + "num_tokens": 521420835.0, + "step": 20989 + }, + { + "epoch": 1.7308485198317802, + "grad_norm": 1.7864711284637451, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.720861554145813, + "num_tokens": 521457749.0, + "step": 20990 + }, + { + "epoch": 1.730930980456832, + "grad_norm": 1.7671706676483154, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7451196312904358, + "num_tokens": 521496846.0, + "step": 20991 + }, + { + "epoch": 1.7310134410818834, + "grad_norm": 1.7963171005249023, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.7945210337638855, + "num_tokens": 521531062.0, + "step": 20992 + }, + { + "epoch": 1.731095901706935, + "grad_norm": 2.101503372192383, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.785763144493103, + "num_tokens": 521558577.0, + "step": 20993 + }, + { + "epoch": 1.7311783623319865, + "grad_norm": 2.1888444423675537, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7768633365631104, + "num_tokens": 521583525.0, + "step": 20994 + }, + { + "epoch": 1.7312608229570379, + "grad_norm": 1.935829520225525, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7686942219734192, + "num_tokens": 521613326.0, + "step": 20995 + }, + { + "epoch": 1.7313432835820897, + "grad_norm": 1.8371772766113281, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7582031488418579, + "num_tokens": 521645553.0, + "step": 20996 + }, + { + "epoch": 1.731425744207141, + "grad_norm": 2.5878243446350098, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.782217264175415, + "num_tokens": 521665075.0, + "step": 20997 + }, + { + "epoch": 1.7315082048321926, + "grad_norm": 1.9102511405944824, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7343333959579468, + "num_tokens": 521698428.0, + "step": 20998 + }, + { + "epoch": 1.7315906654572442, + "grad_norm": 1.8636181354522705, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7644459009170532, + "num_tokens": 521731991.0, + "step": 20999 + }, + { + "epoch": 1.7316731260822957, + "grad_norm": 2.3092241287231445, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7817831039428711, + "num_tokens": 521753453.0, + "step": 21000 + }, + { + "epoch": 1.7317555867073473, + "grad_norm": 1.9928048849105835, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.776237428188324, + "num_tokens": 521781910.0, + "step": 21001 + }, + { + "epoch": 1.7318380473323987, + "grad_norm": 2.1382687091827393, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7196968793869019, + "num_tokens": 521809442.0, + "step": 21002 + }, + { + "epoch": 1.7319205079574505, + "grad_norm": 1.991498589515686, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7691432237625122, + "num_tokens": 521843314.0, + "step": 21003 + }, + { + "epoch": 1.7320029685825018, + "grad_norm": 2.161726951599121, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7649831771850586, + "num_tokens": 521871123.0, + "step": 21004 + }, + { + "epoch": 1.7320854292075534, + "grad_norm": 1.9094682931900024, + "learning_rate": 1e-06, + "loss": 0.5575, + "mean_token_accuracy": 0.8184033632278442, + "num_tokens": 521899314.0, + "step": 21005 + }, + { + "epoch": 1.732167889832605, + "grad_norm": 2.0770041942596436, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7481383085250854, + "num_tokens": 521929134.0, + "step": 21006 + }, + { + "epoch": 1.7322503504576565, + "grad_norm": 1.923537015914917, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7805346846580505, + "num_tokens": 521958631.0, + "step": 21007 + }, + { + "epoch": 1.732332811082708, + "grad_norm": 1.942801833152771, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7750619649887085, + "num_tokens": 521990385.0, + "step": 21008 + }, + { + "epoch": 1.7324152717077594, + "grad_norm": 2.0110087394714355, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7858068943023682, + "num_tokens": 522018466.0, + "step": 21009 + }, + { + "epoch": 1.7324977323328112, + "grad_norm": 2.0522401332855225, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7691650390625, + "num_tokens": 522046225.0, + "step": 21010 + }, + { + "epoch": 1.7325801929578626, + "grad_norm": 2.2804343700408936, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7632337808609009, + "num_tokens": 522069946.0, + "step": 21011 + }, + { + "epoch": 1.7326626535829142, + "grad_norm": 1.8241984844207764, + "learning_rate": 1e-06, + "loss": 0.5426, + "mean_token_accuracy": 0.8083173632621765, + "num_tokens": 522101046.0, + "step": 21012 + }, + { + "epoch": 1.7327451142079657, + "grad_norm": 1.8115569353103638, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.735314130783081, + "num_tokens": 522136610.0, + "step": 21013 + }, + { + "epoch": 1.732827574833017, + "grad_norm": 2.1147544384002686, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7740516662597656, + "num_tokens": 522165555.0, + "step": 21014 + }, + { + "epoch": 1.7329100354580689, + "grad_norm": 1.8915849924087524, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.7921750545501709, + "num_tokens": 522200373.0, + "step": 21015 + }, + { + "epoch": 1.7329924960831202, + "grad_norm": 2.1494154930114746, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.7585379481315613, + "num_tokens": 522226867.0, + "step": 21016 + }, + { + "epoch": 1.7330749567081718, + "grad_norm": 2.2501161098480225, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7405406832695007, + "num_tokens": 522251357.0, + "step": 21017 + }, + { + "epoch": 1.7331574173332234, + "grad_norm": 1.880629301071167, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.7935878038406372, + "num_tokens": 522281448.0, + "step": 21018 + }, + { + "epoch": 1.733239877958275, + "grad_norm": 1.7632163763046265, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.7894225120544434, + "num_tokens": 522319223.0, + "step": 21019 + }, + { + "epoch": 1.7333223385833265, + "grad_norm": 2.132969379425049, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7705286741256714, + "num_tokens": 522345743.0, + "step": 21020 + }, + { + "epoch": 1.7334047992083779, + "grad_norm": 2.17220139503479, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7853445410728455, + "num_tokens": 522374244.0, + "step": 21021 + }, + { + "epoch": 1.7334872598334297, + "grad_norm": 1.7698590755462646, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7712929844856262, + "num_tokens": 522411038.0, + "step": 21022 + }, + { + "epoch": 1.733569720458481, + "grad_norm": 2.040640115737915, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7274163365364075, + "num_tokens": 522440888.0, + "step": 21023 + }, + { + "epoch": 1.7336521810835326, + "grad_norm": 2.1425118446350098, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7539322376251221, + "num_tokens": 522468261.0, + "step": 21024 + }, + { + "epoch": 1.7337346417085842, + "grad_norm": 1.997092366218567, + "learning_rate": 1e-06, + "loss": 0.783, + "mean_token_accuracy": 0.749987781047821, + "num_tokens": 522497753.0, + "step": 21025 + }, + { + "epoch": 1.7338171023336357, + "grad_norm": 1.8313812017440796, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7410950660705566, + "num_tokens": 522533941.0, + "step": 21026 + }, + { + "epoch": 1.7338995629586873, + "grad_norm": 1.8605234622955322, + "learning_rate": 1e-06, + "loss": 0.5865, + "mean_token_accuracy": 0.7967528104782104, + "num_tokens": 522564116.0, + "step": 21027 + }, + { + "epoch": 1.7339820235837387, + "grad_norm": 1.760756254196167, + "learning_rate": 1e-06, + "loss": 0.7705, + "mean_token_accuracy": 0.7528482675552368, + "num_tokens": 522602010.0, + "step": 21028 + }, + { + "epoch": 1.7340644842087904, + "grad_norm": 1.7301905155181885, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7827305197715759, + "num_tokens": 522635193.0, + "step": 21029 + }, + { + "epoch": 1.7341469448338418, + "grad_norm": 2.090026378631592, + "learning_rate": 1e-06, + "loss": 0.7926, + "mean_token_accuracy": 0.7339562177658081, + "num_tokens": 522664089.0, + "step": 21030 + }, + { + "epoch": 1.7342294054588934, + "grad_norm": 2.1063029766082764, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.753042459487915, + "num_tokens": 522692191.0, + "step": 21031 + }, + { + "epoch": 1.734311866083945, + "grad_norm": 2.2978148460388184, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7591069936752319, + "num_tokens": 522717112.0, + "step": 21032 + }, + { + "epoch": 1.7343943267089963, + "grad_norm": 1.9015636444091797, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7517919540405273, + "num_tokens": 522749935.0, + "step": 21033 + }, + { + "epoch": 1.734476787334048, + "grad_norm": 2.1331913471221924, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7374792098999023, + "num_tokens": 522777022.0, + "step": 21034 + }, + { + "epoch": 1.7345592479590994, + "grad_norm": 1.9436326026916504, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7783390283584595, + "num_tokens": 522803997.0, + "step": 21035 + }, + { + "epoch": 1.7346417085841512, + "grad_norm": 1.9323539733886719, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.747654139995575, + "num_tokens": 522837284.0, + "step": 21036 + }, + { + "epoch": 1.7347241692092026, + "grad_norm": 1.9254107475280762, + "learning_rate": 1e-06, + "loss": 0.5867, + "mean_token_accuracy": 0.8099465370178223, + "num_tokens": 522868221.0, + "step": 21037 + }, + { + "epoch": 1.7348066298342542, + "grad_norm": 2.440441846847534, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7294255495071411, + "num_tokens": 522892786.0, + "step": 21038 + }, + { + "epoch": 1.7348890904593057, + "grad_norm": 2.093292474746704, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7513617277145386, + "num_tokens": 522919881.0, + "step": 21039 + }, + { + "epoch": 1.734971551084357, + "grad_norm": 2.0603039264678955, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7633928060531616, + "num_tokens": 522949465.0, + "step": 21040 + }, + { + "epoch": 1.7350540117094089, + "grad_norm": 2.2268059253692627, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7324161529541016, + "num_tokens": 522976121.0, + "step": 21041 + }, + { + "epoch": 1.7351364723344602, + "grad_norm": 2.0832669734954834, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.779579222202301, + "num_tokens": 523002804.0, + "step": 21042 + }, + { + "epoch": 1.7352189329595118, + "grad_norm": 2.0044949054718018, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7363196611404419, + "num_tokens": 523035854.0, + "step": 21043 + }, + { + "epoch": 1.7353013935845634, + "grad_norm": 1.9561896324157715, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.740656852722168, + "num_tokens": 523066677.0, + "step": 21044 + }, + { + "epoch": 1.735383854209615, + "grad_norm": 1.9589617252349854, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.781096339225769, + "num_tokens": 523096989.0, + "step": 21045 + }, + { + "epoch": 1.7354663148346665, + "grad_norm": 2.153670072555542, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7455323934555054, + "num_tokens": 523124606.0, + "step": 21046 + }, + { + "epoch": 1.7355487754597179, + "grad_norm": 1.9755468368530273, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7296315431594849, + "num_tokens": 523155586.0, + "step": 21047 + }, + { + "epoch": 1.7356312360847697, + "grad_norm": 2.1717617511749268, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7326818704605103, + "num_tokens": 523179842.0, + "step": 21048 + }, + { + "epoch": 1.735713696709821, + "grad_norm": 2.3691182136535645, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7781885266304016, + "num_tokens": 523203935.0, + "step": 21049 + }, + { + "epoch": 1.7357961573348726, + "grad_norm": 1.9044075012207031, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.8014818429946899, + "num_tokens": 523233088.0, + "step": 21050 + }, + { + "epoch": 1.7358786179599242, + "grad_norm": 1.8364856243133545, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.7801585793495178, + "num_tokens": 523266228.0, + "step": 21051 + }, + { + "epoch": 1.7359610785849757, + "grad_norm": 1.8634417057037354, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7719168066978455, + "num_tokens": 523302052.0, + "step": 21052 + }, + { + "epoch": 1.7360435392100273, + "grad_norm": 1.795758843421936, + "learning_rate": 1e-06, + "loss": 0.574, + "mean_token_accuracy": 0.806525707244873, + "num_tokens": 523337435.0, + "step": 21053 + }, + { + "epoch": 1.7361259998350786, + "grad_norm": 2.0891292095184326, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7491745948791504, + "num_tokens": 523364939.0, + "step": 21054 + }, + { + "epoch": 1.7362084604601304, + "grad_norm": 2.399181842803955, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7289633750915527, + "num_tokens": 523388940.0, + "step": 21055 + }, + { + "epoch": 1.7362909210851818, + "grad_norm": 2.1373050212860107, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7482788562774658, + "num_tokens": 523416354.0, + "step": 21056 + }, + { + "epoch": 1.7363733817102334, + "grad_norm": 1.92549729347229, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7667529582977295, + "num_tokens": 523447815.0, + "step": 21057 + }, + { + "epoch": 1.736455842335285, + "grad_norm": 1.957534909248352, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7342666387557983, + "num_tokens": 523479360.0, + "step": 21058 + }, + { + "epoch": 1.7365383029603363, + "grad_norm": 1.9250036478042603, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7602033615112305, + "num_tokens": 523511627.0, + "step": 21059 + }, + { + "epoch": 1.736620763585388, + "grad_norm": 2.2179505825042725, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7583661079406738, + "num_tokens": 523534805.0, + "step": 21060 + }, + { + "epoch": 1.7367032242104394, + "grad_norm": 1.9129769802093506, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7704715728759766, + "num_tokens": 523565159.0, + "step": 21061 + }, + { + "epoch": 1.736785684835491, + "grad_norm": 2.1244916915893555, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7868814468383789, + "num_tokens": 523592374.0, + "step": 21062 + }, + { + "epoch": 1.7368681454605426, + "grad_norm": 2.592881202697754, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.709579348564148, + "num_tokens": 523615664.0, + "step": 21063 + }, + { + "epoch": 1.7369506060855942, + "grad_norm": 1.7196592092514038, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7423017024993896, + "num_tokens": 523655815.0, + "step": 21064 + }, + { + "epoch": 1.7370330667106457, + "grad_norm": 1.844826102256775, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.7975552082061768, + "num_tokens": 523688194.0, + "step": 21065 + }, + { + "epoch": 1.737115527335697, + "grad_norm": 2.0085060596466064, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7452789545059204, + "num_tokens": 523719329.0, + "step": 21066 + }, + { + "epoch": 1.7371979879607489, + "grad_norm": 2.1414966583251953, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7613873481750488, + "num_tokens": 523748087.0, + "step": 21067 + }, + { + "epoch": 1.7372804485858002, + "grad_norm": 1.9335557222366333, + "learning_rate": 1e-06, + "loss": 0.6001, + "mean_token_accuracy": 0.785163402557373, + "num_tokens": 523780444.0, + "step": 21068 + }, + { + "epoch": 1.7373629092108518, + "grad_norm": 2.1607367992401123, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7509459257125854, + "num_tokens": 523806712.0, + "step": 21069 + }, + { + "epoch": 1.7374453698359034, + "grad_norm": 2.149982213973999, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7649670839309692, + "num_tokens": 523832917.0, + "step": 21070 + }, + { + "epoch": 1.737527830460955, + "grad_norm": 2.1198134422302246, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7716139554977417, + "num_tokens": 523859871.0, + "step": 21071 + }, + { + "epoch": 1.7376102910860065, + "grad_norm": 2.044088125228882, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.7502101063728333, + "num_tokens": 523885446.0, + "step": 21072 + }, + { + "epoch": 1.7376927517110579, + "grad_norm": 1.6870523691177368, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7783846855163574, + "num_tokens": 523921506.0, + "step": 21073 + }, + { + "epoch": 1.7377752123361097, + "grad_norm": 2.0911154747009277, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.791590690612793, + "num_tokens": 523950788.0, + "step": 21074 + }, + { + "epoch": 1.737857672961161, + "grad_norm": 2.179051160812378, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.747680127620697, + "num_tokens": 523977900.0, + "step": 21075 + }, + { + "epoch": 1.7379401335862126, + "grad_norm": 1.7586740255355835, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.753788411617279, + "num_tokens": 524015598.0, + "step": 21076 + }, + { + "epoch": 1.7380225942112641, + "grad_norm": 2.1071016788482666, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.7849817276000977, + "num_tokens": 524041160.0, + "step": 21077 + }, + { + "epoch": 1.7381050548363155, + "grad_norm": 1.8262920379638672, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7232621908187866, + "num_tokens": 524077563.0, + "step": 21078 + }, + { + "epoch": 1.7381875154613673, + "grad_norm": 1.9986635446548462, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.759458601474762, + "num_tokens": 524107755.0, + "step": 21079 + }, + { + "epoch": 1.7382699760864186, + "grad_norm": 1.9185442924499512, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.754490852355957, + "num_tokens": 524141969.0, + "step": 21080 + }, + { + "epoch": 1.7383524367114704, + "grad_norm": 1.7681992053985596, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7669925689697266, + "num_tokens": 524177249.0, + "step": 21081 + }, + { + "epoch": 1.7384348973365218, + "grad_norm": 2.0626142024993896, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7560958862304688, + "num_tokens": 524204324.0, + "step": 21082 + }, + { + "epoch": 1.7385173579615734, + "grad_norm": 1.8101263046264648, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7662962675094604, + "num_tokens": 524238595.0, + "step": 21083 + }, + { + "epoch": 1.738599818586625, + "grad_norm": 2.212529182434082, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7646706700325012, + "num_tokens": 524262761.0, + "step": 21084 + }, + { + "epoch": 1.7386822792116763, + "grad_norm": 1.9586671590805054, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.7832726836204529, + "num_tokens": 524294213.0, + "step": 21085 + }, + { + "epoch": 1.738764739836728, + "grad_norm": 2.063994884490967, + "learning_rate": 1e-06, + "loss": 0.6097, + "mean_token_accuracy": 0.7905412316322327, + "num_tokens": 524319309.0, + "step": 21086 + }, + { + "epoch": 1.7388472004617794, + "grad_norm": 2.1491377353668213, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7353503108024597, + "num_tokens": 524348487.0, + "step": 21087 + }, + { + "epoch": 1.738929661086831, + "grad_norm": 1.903038740158081, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7705591320991516, + "num_tokens": 524380671.0, + "step": 21088 + }, + { + "epoch": 1.7390121217118826, + "grad_norm": 2.008242130279541, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7589735388755798, + "num_tokens": 524407627.0, + "step": 21089 + }, + { + "epoch": 1.7390945823369341, + "grad_norm": 2.0269978046417236, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7545415759086609, + "num_tokens": 524438118.0, + "step": 21090 + }, + { + "epoch": 1.7391770429619857, + "grad_norm": 1.8426772356033325, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7474783062934875, + "num_tokens": 524474103.0, + "step": 21091 + }, + { + "epoch": 1.739259503587037, + "grad_norm": 2.05307936668396, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7497274875640869, + "num_tokens": 524502313.0, + "step": 21092 + }, + { + "epoch": 1.7393419642120889, + "grad_norm": 2.0969858169555664, + "learning_rate": 1e-06, + "loss": 0.7768, + "mean_token_accuracy": 0.7511637210845947, + "num_tokens": 524528437.0, + "step": 21093 + }, + { + "epoch": 1.7394244248371402, + "grad_norm": 2.275242567062378, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7289546728134155, + "num_tokens": 524552489.0, + "step": 21094 + }, + { + "epoch": 1.7395068854621918, + "grad_norm": 2.168461799621582, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7226475477218628, + "num_tokens": 524579158.0, + "step": 21095 + }, + { + "epoch": 1.7395893460872434, + "grad_norm": 1.9190279245376587, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7690210342407227, + "num_tokens": 524613258.0, + "step": 21096 + }, + { + "epoch": 1.739671806712295, + "grad_norm": 1.8218879699707031, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7854756116867065, + "num_tokens": 524646271.0, + "step": 21097 + }, + { + "epoch": 1.7397542673373465, + "grad_norm": 2.0006349086761475, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7531572580337524, + "num_tokens": 524675352.0, + "step": 21098 + }, + { + "epoch": 1.7398367279623979, + "grad_norm": 1.9924556016921997, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7751452922821045, + "num_tokens": 524705045.0, + "step": 21099 + }, + { + "epoch": 1.7399191885874496, + "grad_norm": 1.9129551649093628, + "learning_rate": 1e-06, + "loss": 0.7768, + "mean_token_accuracy": 0.7535520792007446, + "num_tokens": 524737449.0, + "step": 21100 + }, + { + "epoch": 1.740001649212501, + "grad_norm": 2.1925368309020996, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7515464425086975, + "num_tokens": 524763280.0, + "step": 21101 + }, + { + "epoch": 1.7400841098375526, + "grad_norm": 2.29327130317688, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7147008180618286, + "num_tokens": 524789655.0, + "step": 21102 + }, + { + "epoch": 1.7401665704626041, + "grad_norm": 1.9716953039169312, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7887481451034546, + "num_tokens": 524818442.0, + "step": 21103 + }, + { + "epoch": 1.7402490310876555, + "grad_norm": 2.0220537185668945, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.7747883796691895, + "num_tokens": 524848004.0, + "step": 21104 + }, + { + "epoch": 1.7403314917127073, + "grad_norm": 2.0371880531311035, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7429611086845398, + "num_tokens": 524878447.0, + "step": 21105 + }, + { + "epoch": 1.7404139523377586, + "grad_norm": 2.0716371536254883, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.7967014908790588, + "num_tokens": 524906344.0, + "step": 21106 + }, + { + "epoch": 1.7404964129628102, + "grad_norm": 1.9234371185302734, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.7882363796234131, + "num_tokens": 524935673.0, + "step": 21107 + }, + { + "epoch": 1.7405788735878618, + "grad_norm": 2.4974143505096436, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7433165907859802, + "num_tokens": 524958346.0, + "step": 21108 + }, + { + "epoch": 1.7406613342129134, + "grad_norm": 1.951193928718567, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7717854976654053, + "num_tokens": 524987018.0, + "step": 21109 + }, + { + "epoch": 1.740743794837965, + "grad_norm": 1.9113752841949463, + "learning_rate": 1e-06, + "loss": 0.5055, + "mean_token_accuracy": 0.8242424726486206, + "num_tokens": 525014454.0, + "step": 21110 + }, + { + "epoch": 1.7408262554630163, + "grad_norm": 2.0728707313537598, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7292590737342834, + "num_tokens": 525043913.0, + "step": 21111 + }, + { + "epoch": 1.740908716088068, + "grad_norm": 1.8205194473266602, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.7956891655921936, + "num_tokens": 525077550.0, + "step": 21112 + }, + { + "epoch": 1.7409911767131194, + "grad_norm": 1.8847540616989136, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7813158631324768, + "num_tokens": 525108175.0, + "step": 21113 + }, + { + "epoch": 1.741073637338171, + "grad_norm": 1.8242666721343994, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7418213486671448, + "num_tokens": 525144244.0, + "step": 21114 + }, + { + "epoch": 1.7411560979632226, + "grad_norm": 1.932581901550293, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7557472586631775, + "num_tokens": 525176044.0, + "step": 21115 + }, + { + "epoch": 1.7412385585882741, + "grad_norm": 2.0582833290100098, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.757202684879303, + "num_tokens": 525204608.0, + "step": 21116 + }, + { + "epoch": 1.7413210192133257, + "grad_norm": 2.413346290588379, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7769978046417236, + "num_tokens": 525227197.0, + "step": 21117 + }, + { + "epoch": 1.741403479838377, + "grad_norm": 2.171058416366577, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7374842166900635, + "num_tokens": 525254173.0, + "step": 21118 + }, + { + "epoch": 1.7414859404634289, + "grad_norm": 2.161081075668335, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7413111925125122, + "num_tokens": 525282017.0, + "step": 21119 + }, + { + "epoch": 1.7415684010884802, + "grad_norm": 1.7989548444747925, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.7892950773239136, + "num_tokens": 525314838.0, + "step": 21120 + }, + { + "epoch": 1.7416508617135318, + "grad_norm": 1.7869423627853394, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7764018774032593, + "num_tokens": 525350945.0, + "step": 21121 + }, + { + "epoch": 1.7417333223385834, + "grad_norm": 1.8730740547180176, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7369194626808167, + "num_tokens": 525385486.0, + "step": 21122 + }, + { + "epoch": 1.7418157829636347, + "grad_norm": 1.8542722463607788, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7497139573097229, + "num_tokens": 525421228.0, + "step": 21123 + }, + { + "epoch": 1.7418982435886865, + "grad_norm": 1.981652855873108, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7780890464782715, + "num_tokens": 525450223.0, + "step": 21124 + }, + { + "epoch": 1.7419807042137379, + "grad_norm": 1.9877699613571167, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7658202648162842, + "num_tokens": 525483378.0, + "step": 21125 + }, + { + "epoch": 1.7420631648387894, + "grad_norm": 1.9928950071334839, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.754665195941925, + "num_tokens": 525514437.0, + "step": 21126 + }, + { + "epoch": 1.742145625463841, + "grad_norm": 2.0325639247894287, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7279702425003052, + "num_tokens": 525543059.0, + "step": 21127 + }, + { + "epoch": 1.7422280860888926, + "grad_norm": 2.116490125656128, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7792750597000122, + "num_tokens": 525568584.0, + "step": 21128 + }, + { + "epoch": 1.7423105467139441, + "grad_norm": 1.9544763565063477, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.792630136013031, + "num_tokens": 525597068.0, + "step": 21129 + }, + { + "epoch": 1.7423930073389955, + "grad_norm": 1.93297278881073, + "learning_rate": 1e-06, + "loss": 0.7683, + "mean_token_accuracy": 0.7498667240142822, + "num_tokens": 525629322.0, + "step": 21130 + }, + { + "epoch": 1.7424754679640473, + "grad_norm": 2.392456293106079, + "learning_rate": 1e-06, + "loss": 0.7563, + "mean_token_accuracy": 0.7539953589439392, + "num_tokens": 525650117.0, + "step": 21131 + }, + { + "epoch": 1.7425579285890986, + "grad_norm": 2.004765510559082, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7827714085578918, + "num_tokens": 525679012.0, + "step": 21132 + }, + { + "epoch": 1.7426403892141502, + "grad_norm": 1.988224983215332, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7811700701713562, + "num_tokens": 525708507.0, + "step": 21133 + }, + { + "epoch": 1.7427228498392018, + "grad_norm": 1.9189108610153198, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.7996613383293152, + "num_tokens": 525737830.0, + "step": 21134 + }, + { + "epoch": 1.7428053104642534, + "grad_norm": 2.176738977432251, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7387866973876953, + "num_tokens": 525766331.0, + "step": 21135 + }, + { + "epoch": 1.742887771089305, + "grad_norm": 1.9871253967285156, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7735621929168701, + "num_tokens": 525799051.0, + "step": 21136 + }, + { + "epoch": 1.7429702317143563, + "grad_norm": 1.800752878189087, + "learning_rate": 1e-06, + "loss": 0.5244, + "mean_token_accuracy": 0.7981759905815125, + "num_tokens": 525829552.0, + "step": 21137 + }, + { + "epoch": 1.743052692339408, + "grad_norm": 2.0497729778289795, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.742394208908081, + "num_tokens": 525857798.0, + "step": 21138 + }, + { + "epoch": 1.7431351529644594, + "grad_norm": 1.9971191883087158, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7560685873031616, + "num_tokens": 525887364.0, + "step": 21139 + }, + { + "epoch": 1.743217613589511, + "grad_norm": 1.8337348699569702, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7646452784538269, + "num_tokens": 525924243.0, + "step": 21140 + }, + { + "epoch": 1.7433000742145626, + "grad_norm": 2.312870979309082, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7435570955276489, + "num_tokens": 525948904.0, + "step": 21141 + }, + { + "epoch": 1.7433825348396141, + "grad_norm": 2.181927442550659, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.767507791519165, + "num_tokens": 525978179.0, + "step": 21142 + }, + { + "epoch": 1.7434649954646657, + "grad_norm": 1.8362576961517334, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.7956690788269043, + "num_tokens": 526012195.0, + "step": 21143 + }, + { + "epoch": 1.743547456089717, + "grad_norm": 1.9001613855361938, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.7942721843719482, + "num_tokens": 526043866.0, + "step": 21144 + }, + { + "epoch": 1.7436299167147689, + "grad_norm": 1.974501132965088, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7427945733070374, + "num_tokens": 526075261.0, + "step": 21145 + }, + { + "epoch": 1.7437123773398202, + "grad_norm": 1.9008467197418213, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.7928372621536255, + "num_tokens": 526105495.0, + "step": 21146 + }, + { + "epoch": 1.7437948379648718, + "grad_norm": 1.9949132204055786, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7511599063873291, + "num_tokens": 526132003.0, + "step": 21147 + }, + { + "epoch": 1.7438772985899234, + "grad_norm": 2.371790885925293, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7182601094245911, + "num_tokens": 526158122.0, + "step": 21148 + }, + { + "epoch": 1.7439597592149747, + "grad_norm": 2.1935229301452637, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7451331615447998, + "num_tokens": 526185251.0, + "step": 21149 + }, + { + "epoch": 1.7440422198400265, + "grad_norm": 2.164539337158203, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7384264469146729, + "num_tokens": 526210865.0, + "step": 21150 + }, + { + "epoch": 1.7441246804650778, + "grad_norm": 2.0387117862701416, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7218942046165466, + "num_tokens": 526241751.0, + "step": 21151 + }, + { + "epoch": 1.7442071410901294, + "grad_norm": 2.1321089267730713, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7645598649978638, + "num_tokens": 526266096.0, + "step": 21152 + }, + { + "epoch": 1.744289601715181, + "grad_norm": 1.8783423900604248, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.7812032699584961, + "num_tokens": 526295571.0, + "step": 21153 + }, + { + "epoch": 1.7443720623402326, + "grad_norm": 2.2784008979797363, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7464641332626343, + "num_tokens": 526321539.0, + "step": 21154 + }, + { + "epoch": 1.7444545229652841, + "grad_norm": 2.0832693576812744, + "learning_rate": 1e-06, + "loss": 0.5498, + "mean_token_accuracy": 0.8140480518341064, + "num_tokens": 526347202.0, + "step": 21155 + }, + { + "epoch": 1.7445369835903355, + "grad_norm": 2.0835089683532715, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7439693212509155, + "num_tokens": 526375356.0, + "step": 21156 + }, + { + "epoch": 1.7446194442153873, + "grad_norm": 1.990737795829773, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7710572481155396, + "num_tokens": 526406864.0, + "step": 21157 + }, + { + "epoch": 1.7447019048404386, + "grad_norm": 2.253429889678955, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7344071269035339, + "num_tokens": 526434472.0, + "step": 21158 + }, + { + "epoch": 1.7447843654654902, + "grad_norm": 1.9785305261611938, + "learning_rate": 1e-06, + "loss": 0.6251, + "mean_token_accuracy": 0.7971922159194946, + "num_tokens": 526465787.0, + "step": 21159 + }, + { + "epoch": 1.7448668260905418, + "grad_norm": 1.9210437536239624, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7357874512672424, + "num_tokens": 526499670.0, + "step": 21160 + }, + { + "epoch": 1.7449492867155934, + "grad_norm": 1.914058804512024, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7327736616134644, + "num_tokens": 526531560.0, + "step": 21161 + }, + { + "epoch": 1.745031747340645, + "grad_norm": 1.9457619190216064, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.7993922233581543, + "num_tokens": 526560875.0, + "step": 21162 + }, + { + "epoch": 1.7451142079656963, + "grad_norm": 2.200584650039673, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7782479524612427, + "num_tokens": 526585924.0, + "step": 21163 + }, + { + "epoch": 1.745196668590748, + "grad_norm": 2.0180561542510986, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7468628883361816, + "num_tokens": 526616419.0, + "step": 21164 + }, + { + "epoch": 1.7452791292157994, + "grad_norm": 1.970513105392456, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7185636162757874, + "num_tokens": 526649218.0, + "step": 21165 + }, + { + "epoch": 1.745361589840851, + "grad_norm": 2.200479507446289, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7545153498649597, + "num_tokens": 526673816.0, + "step": 21166 + }, + { + "epoch": 1.7454440504659026, + "grad_norm": 2.1470084190368652, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.7149412035942078, + "num_tokens": 526702608.0, + "step": 21167 + }, + { + "epoch": 1.745526511090954, + "grad_norm": 1.8558380603790283, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7657897472381592, + "num_tokens": 526732356.0, + "step": 21168 + }, + { + "epoch": 1.7456089717160057, + "grad_norm": 1.95926833152771, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7765414714813232, + "num_tokens": 526762132.0, + "step": 21169 + }, + { + "epoch": 1.745691432341057, + "grad_norm": 1.9554219245910645, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7367450594902039, + "num_tokens": 526792502.0, + "step": 21170 + }, + { + "epoch": 1.7457738929661086, + "grad_norm": 1.9032831192016602, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7622118592262268, + "num_tokens": 526825053.0, + "step": 21171 + }, + { + "epoch": 1.7458563535911602, + "grad_norm": 1.7597253322601318, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7594515085220337, + "num_tokens": 526862470.0, + "step": 21172 + }, + { + "epoch": 1.7459388142162118, + "grad_norm": 1.8769707679748535, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7109300494194031, + "num_tokens": 526896940.0, + "step": 21173 + }, + { + "epoch": 1.7460212748412633, + "grad_norm": 1.9132746458053589, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7558184266090393, + "num_tokens": 526926714.0, + "step": 21174 + }, + { + "epoch": 1.7461037354663147, + "grad_norm": 1.905706763267517, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.7878296971321106, + "num_tokens": 526956470.0, + "step": 21175 + }, + { + "epoch": 1.7461861960913665, + "grad_norm": 1.9686733484268188, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.7683599591255188, + "num_tokens": 526985055.0, + "step": 21176 + }, + { + "epoch": 1.7462686567164178, + "grad_norm": 1.9298527240753174, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7381864190101624, + "num_tokens": 527015663.0, + "step": 21177 + }, + { + "epoch": 1.7463511173414694, + "grad_norm": 1.9713432788848877, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7702318429946899, + "num_tokens": 527047625.0, + "step": 21178 + }, + { + "epoch": 1.746433577966521, + "grad_norm": 1.8791512250900269, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7684648036956787, + "num_tokens": 527079332.0, + "step": 21179 + }, + { + "epoch": 1.7465160385915726, + "grad_norm": 1.976833462715149, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.760754406452179, + "num_tokens": 527111715.0, + "step": 21180 + }, + { + "epoch": 1.7465984992166241, + "grad_norm": 2.007417678833008, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7401249408721924, + "num_tokens": 527141682.0, + "step": 21181 + }, + { + "epoch": 1.7466809598416755, + "grad_norm": 1.9993592500686646, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7427948713302612, + "num_tokens": 527170298.0, + "step": 21182 + }, + { + "epoch": 1.7467634204667273, + "grad_norm": 2.5754308700561523, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.759853184223175, + "num_tokens": 527189062.0, + "step": 21183 + }, + { + "epoch": 1.7468458810917786, + "grad_norm": 2.0996129512786865, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.77281653881073, + "num_tokens": 527214409.0, + "step": 21184 + }, + { + "epoch": 1.7469283417168302, + "grad_norm": 1.8982717990875244, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7512789964675903, + "num_tokens": 527249624.0, + "step": 21185 + }, + { + "epoch": 1.7470108023418818, + "grad_norm": 2.257357597351074, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.726459264755249, + "num_tokens": 527276191.0, + "step": 21186 + }, + { + "epoch": 1.7470932629669331, + "grad_norm": 1.9442170858383179, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7709441184997559, + "num_tokens": 527307527.0, + "step": 21187 + }, + { + "epoch": 1.747175723591985, + "grad_norm": 2.3513429164886475, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.7942253947257996, + "num_tokens": 527329577.0, + "step": 21188 + }, + { + "epoch": 1.7472581842170363, + "grad_norm": 2.0552616119384766, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7389279007911682, + "num_tokens": 527357771.0, + "step": 21189 + }, + { + "epoch": 1.747340644842088, + "grad_norm": 1.8169888257980347, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7454547882080078, + "num_tokens": 527394102.0, + "step": 21190 + }, + { + "epoch": 1.7474231054671394, + "grad_norm": 1.8569214344024658, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.7865818738937378, + "num_tokens": 527427286.0, + "step": 21191 + }, + { + "epoch": 1.747505566092191, + "grad_norm": 1.908454418182373, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7678595781326294, + "num_tokens": 527458515.0, + "step": 21192 + }, + { + "epoch": 1.7475880267172426, + "grad_norm": 2.068542003631592, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.761067271232605, + "num_tokens": 527488495.0, + "step": 21193 + }, + { + "epoch": 1.747670487342294, + "grad_norm": 2.0461878776550293, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7600671648979187, + "num_tokens": 527517777.0, + "step": 21194 + }, + { + "epoch": 1.7477529479673457, + "grad_norm": 1.999743103981018, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.7873170375823975, + "num_tokens": 527549303.0, + "step": 21195 + }, + { + "epoch": 1.747835408592397, + "grad_norm": 1.9799835681915283, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.790343701839447, + "num_tokens": 527578859.0, + "step": 21196 + }, + { + "epoch": 1.7479178692174486, + "grad_norm": 2.0751712322235107, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.7867639064788818, + "num_tokens": 527607847.0, + "step": 21197 + }, + { + "epoch": 1.7480003298425002, + "grad_norm": 2.1944053173065186, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7514495849609375, + "num_tokens": 527635871.0, + "step": 21198 + }, + { + "epoch": 1.7480827904675518, + "grad_norm": 1.804247498512268, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7442792654037476, + "num_tokens": 527671967.0, + "step": 21199 + }, + { + "epoch": 1.7481652510926033, + "grad_norm": 1.9954183101654053, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7674341201782227, + "num_tokens": 527703940.0, + "step": 21200 + }, + { + "epoch": 1.7482477117176547, + "grad_norm": 2.269312620162964, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7540450096130371, + "num_tokens": 527729682.0, + "step": 21201 + }, + { + "epoch": 1.7483301723427065, + "grad_norm": 2.3765921592712402, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7765135765075684, + "num_tokens": 527754916.0, + "step": 21202 + }, + { + "epoch": 1.7484126329677578, + "grad_norm": 2.0079424381256104, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7652971148490906, + "num_tokens": 527786690.0, + "step": 21203 + }, + { + "epoch": 1.7484950935928094, + "grad_norm": 1.9283027648925781, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7751355171203613, + "num_tokens": 527818797.0, + "step": 21204 + }, + { + "epoch": 1.748577554217861, + "grad_norm": 1.9970625638961792, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7882034182548523, + "num_tokens": 527845319.0, + "step": 21205 + }, + { + "epoch": 1.7486600148429126, + "grad_norm": 1.7805006504058838, + "learning_rate": 1e-06, + "loss": 0.5175, + "mean_token_accuracy": 0.8279706239700317, + "num_tokens": 527880349.0, + "step": 21206 + }, + { + "epoch": 1.7487424754679641, + "grad_norm": 1.9361426830291748, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7530902028083801, + "num_tokens": 527913537.0, + "step": 21207 + }, + { + "epoch": 1.7488249360930155, + "grad_norm": 1.9659903049468994, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7759031057357788, + "num_tokens": 527943903.0, + "step": 21208 + }, + { + "epoch": 1.7489073967180673, + "grad_norm": 2.0466411113739014, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7573646306991577, + "num_tokens": 527973484.0, + "step": 21209 + }, + { + "epoch": 1.7489898573431186, + "grad_norm": 1.7772188186645508, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7725590467453003, + "num_tokens": 528006627.0, + "step": 21210 + }, + { + "epoch": 1.7490723179681702, + "grad_norm": 1.7062877416610718, + "learning_rate": 1e-06, + "loss": 0.615, + "mean_token_accuracy": 0.8051851987838745, + "num_tokens": 528045966.0, + "step": 21211 + }, + { + "epoch": 1.7491547785932218, + "grad_norm": 2.0212972164154053, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7392153739929199, + "num_tokens": 528078623.0, + "step": 21212 + }, + { + "epoch": 1.7492372392182731, + "grad_norm": 1.9306268692016602, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.7824095487594604, + "num_tokens": 528106659.0, + "step": 21213 + }, + { + "epoch": 1.749319699843325, + "grad_norm": 1.8126341104507446, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.7935896515846252, + "num_tokens": 528140081.0, + "step": 21214 + }, + { + "epoch": 1.7494021604683763, + "grad_norm": 1.798366904258728, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7581155300140381, + "num_tokens": 528173316.0, + "step": 21215 + }, + { + "epoch": 1.7494846210934278, + "grad_norm": 2.088505983352661, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7602981925010681, + "num_tokens": 528198796.0, + "step": 21216 + }, + { + "epoch": 1.7495670817184794, + "grad_norm": 2.079157590866089, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7469329833984375, + "num_tokens": 528225357.0, + "step": 21217 + }, + { + "epoch": 1.749649542343531, + "grad_norm": 2.17117977142334, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7303860783576965, + "num_tokens": 528251209.0, + "step": 21218 + }, + { + "epoch": 1.7497320029685826, + "grad_norm": 1.8158634901046753, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7650212049484253, + "num_tokens": 528285477.0, + "step": 21219 + }, + { + "epoch": 1.749814463593634, + "grad_norm": 2.195794105529785, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7566765546798706, + "num_tokens": 528310539.0, + "step": 21220 + }, + { + "epoch": 1.7498969242186857, + "grad_norm": 2.245946168899536, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7702703475952148, + "num_tokens": 528335274.0, + "step": 21221 + }, + { + "epoch": 1.749979384843737, + "grad_norm": 1.9015963077545166, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7396433353424072, + "num_tokens": 528367925.0, + "step": 21222 + }, + { + "epoch": 1.7500618454687886, + "grad_norm": 1.810101866722107, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7713897228240967, + "num_tokens": 528404370.0, + "step": 21223 + }, + { + "epoch": 1.7501443060938402, + "grad_norm": 2.4732658863067627, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7633155584335327, + "num_tokens": 528425985.0, + "step": 21224 + }, + { + "epoch": 1.7502267667188918, + "grad_norm": 2.0594420433044434, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7471777200698853, + "num_tokens": 528455346.0, + "step": 21225 + }, + { + "epoch": 1.7503092273439433, + "grad_norm": 1.8742096424102783, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7795950174331665, + "num_tokens": 528486581.0, + "step": 21226 + }, + { + "epoch": 1.7503916879689947, + "grad_norm": 2.0615739822387695, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7416505813598633, + "num_tokens": 528514384.0, + "step": 21227 + }, + { + "epoch": 1.7504741485940465, + "grad_norm": 1.9469186067581177, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7532902956008911, + "num_tokens": 528548581.0, + "step": 21228 + }, + { + "epoch": 1.7505566092190978, + "grad_norm": 1.9065481424331665, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7462519407272339, + "num_tokens": 528578415.0, + "step": 21229 + }, + { + "epoch": 1.7506390698441494, + "grad_norm": 1.8625421524047852, + "learning_rate": 1e-06, + "loss": 0.6022, + "mean_token_accuracy": 0.8006507158279419, + "num_tokens": 528611336.0, + "step": 21230 + }, + { + "epoch": 1.750721530469201, + "grad_norm": 2.219533681869507, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7537610530853271, + "num_tokens": 528637536.0, + "step": 21231 + }, + { + "epoch": 1.7508039910942523, + "grad_norm": 1.9599485397338867, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.7852960824966431, + "num_tokens": 528668054.0, + "step": 21232 + }, + { + "epoch": 1.7508864517193041, + "grad_norm": 2.1995644569396973, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7237076163291931, + "num_tokens": 528696310.0, + "step": 21233 + }, + { + "epoch": 1.7509689123443555, + "grad_norm": 2.083360433578491, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7478184700012207, + "num_tokens": 528724787.0, + "step": 21234 + }, + { + "epoch": 1.7510513729694073, + "grad_norm": 1.6892237663269043, + "learning_rate": 1e-06, + "loss": 0.5971, + "mean_token_accuracy": 0.8030184507369995, + "num_tokens": 528757510.0, + "step": 21235 + }, + { + "epoch": 1.7511338335944586, + "grad_norm": 2.027137517929077, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7867372035980225, + "num_tokens": 528785770.0, + "step": 21236 + }, + { + "epoch": 1.7512162942195102, + "grad_norm": 1.8610315322875977, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7405601739883423, + "num_tokens": 528820382.0, + "step": 21237 + }, + { + "epoch": 1.7512987548445618, + "grad_norm": 1.8001341819763184, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7386754155158997, + "num_tokens": 528855812.0, + "step": 21238 + }, + { + "epoch": 1.7513812154696131, + "grad_norm": 2.263768434524536, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7581827640533447, + "num_tokens": 528880835.0, + "step": 21239 + }, + { + "epoch": 1.751463676094665, + "grad_norm": 2.1476967334747314, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7199798226356506, + "num_tokens": 528909635.0, + "step": 21240 + }, + { + "epoch": 1.7515461367197163, + "grad_norm": 1.7089966535568237, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.7911235094070435, + "num_tokens": 528950340.0, + "step": 21241 + }, + { + "epoch": 1.7516285973447678, + "grad_norm": 1.6938339471817017, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7538633346557617, + "num_tokens": 528992086.0, + "step": 21242 + }, + { + "epoch": 1.7517110579698194, + "grad_norm": 2.0361194610595703, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7517147064208984, + "num_tokens": 529019787.0, + "step": 21243 + }, + { + "epoch": 1.751793518594871, + "grad_norm": 1.7870128154754639, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7587223052978516, + "num_tokens": 529056134.0, + "step": 21244 + }, + { + "epoch": 1.7518759792199226, + "grad_norm": 2.042250871658325, + "learning_rate": 1e-06, + "loss": 0.6089, + "mean_token_accuracy": 0.7853305339813232, + "num_tokens": 529083272.0, + "step": 21245 + }, + { + "epoch": 1.751958439844974, + "grad_norm": 1.9619462490081787, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7649016976356506, + "num_tokens": 529113158.0, + "step": 21246 + }, + { + "epoch": 1.7520409004700257, + "grad_norm": 2.131381034851074, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7777812480926514, + "num_tokens": 529142408.0, + "step": 21247 + }, + { + "epoch": 1.752123361095077, + "grad_norm": 1.8282197713851929, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7693129777908325, + "num_tokens": 529173819.0, + "step": 21248 + }, + { + "epoch": 1.7522058217201286, + "grad_norm": 2.172488212585449, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7581455111503601, + "num_tokens": 529200144.0, + "step": 21249 + }, + { + "epoch": 1.7522882823451802, + "grad_norm": 1.9016236066818237, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.7827724814414978, + "num_tokens": 529231709.0, + "step": 21250 + }, + { + "epoch": 1.7523707429702318, + "grad_norm": 2.177046775817871, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7365132570266724, + "num_tokens": 529256870.0, + "step": 21251 + }, + { + "epoch": 1.7524532035952833, + "grad_norm": 2.0315968990325928, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7765673398971558, + "num_tokens": 529285680.0, + "step": 21252 + }, + { + "epoch": 1.7525356642203347, + "grad_norm": 1.6488384008407593, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7684112787246704, + "num_tokens": 529325240.0, + "step": 21253 + }, + { + "epoch": 1.7526181248453865, + "grad_norm": 2.1034529209136963, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7312345504760742, + "num_tokens": 529352905.0, + "step": 21254 + }, + { + "epoch": 1.7527005854704378, + "grad_norm": 2.1030852794647217, + "learning_rate": 1e-06, + "loss": 0.5712, + "mean_token_accuracy": 0.8122770190238953, + "num_tokens": 529379547.0, + "step": 21255 + }, + { + "epoch": 1.7527830460954894, + "grad_norm": 1.8927319049835205, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7791004180908203, + "num_tokens": 529409735.0, + "step": 21256 + }, + { + "epoch": 1.752865506720541, + "grad_norm": 1.8768945932388306, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7374414801597595, + "num_tokens": 529443249.0, + "step": 21257 + }, + { + "epoch": 1.7529479673455923, + "grad_norm": 1.8939229249954224, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7517617344856262, + "num_tokens": 529475032.0, + "step": 21258 + }, + { + "epoch": 1.7530304279706441, + "grad_norm": 2.0001533031463623, + "learning_rate": 1e-06, + "loss": 0.5755, + "mean_token_accuracy": 0.8051211833953857, + "num_tokens": 529502723.0, + "step": 21259 + }, + { + "epoch": 1.7531128885956955, + "grad_norm": 1.8531497716903687, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7326812148094177, + "num_tokens": 529536116.0, + "step": 21260 + }, + { + "epoch": 1.753195349220747, + "grad_norm": 1.9360491037368774, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7153671383857727, + "num_tokens": 529570842.0, + "step": 21261 + }, + { + "epoch": 1.7532778098457986, + "grad_norm": 1.9105918407440186, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7575260400772095, + "num_tokens": 529603197.0, + "step": 21262 + }, + { + "epoch": 1.7533602704708502, + "grad_norm": 2.2430169582366943, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7590150237083435, + "num_tokens": 529626936.0, + "step": 21263 + }, + { + "epoch": 1.7534427310959018, + "grad_norm": 2.1305198669433594, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7743834257125854, + "num_tokens": 529655959.0, + "step": 21264 + }, + { + "epoch": 1.7535251917209531, + "grad_norm": 1.9203547239303589, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7553635835647583, + "num_tokens": 529686700.0, + "step": 21265 + }, + { + "epoch": 1.753607652346005, + "grad_norm": 1.9665307998657227, + "learning_rate": 1e-06, + "loss": 0.5851, + "mean_token_accuracy": 0.8057565093040466, + "num_tokens": 529714516.0, + "step": 21266 + }, + { + "epoch": 1.7536901129710563, + "grad_norm": 1.7930439710617065, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7480912208557129, + "num_tokens": 529753897.0, + "step": 21267 + }, + { + "epoch": 1.7537725735961078, + "grad_norm": 1.9469528198242188, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7575564980506897, + "num_tokens": 529785109.0, + "step": 21268 + }, + { + "epoch": 1.7538550342211594, + "grad_norm": 2.1423745155334473, + "learning_rate": 1e-06, + "loss": 0.7936, + "mean_token_accuracy": 0.7400493621826172, + "num_tokens": 529813082.0, + "step": 21269 + }, + { + "epoch": 1.753937494846211, + "grad_norm": 1.9970362186431885, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.7911326885223389, + "num_tokens": 529842788.0, + "step": 21270 + }, + { + "epoch": 1.7540199554712625, + "grad_norm": 2.1898880004882812, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7361963987350464, + "num_tokens": 529869082.0, + "step": 21271 + }, + { + "epoch": 1.754102416096314, + "grad_norm": 1.860636591911316, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7453112006187439, + "num_tokens": 529902675.0, + "step": 21272 + }, + { + "epoch": 1.7541848767213657, + "grad_norm": 1.9866290092468262, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.7750422358512878, + "num_tokens": 529931272.0, + "step": 21273 + }, + { + "epoch": 1.754267337346417, + "grad_norm": 2.0277857780456543, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7656622529029846, + "num_tokens": 529959686.0, + "step": 21274 + }, + { + "epoch": 1.7543497979714686, + "grad_norm": 2.207082748413086, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7378343343734741, + "num_tokens": 529985789.0, + "step": 21275 + }, + { + "epoch": 1.7544322585965202, + "grad_norm": 2.085530996322632, + "learning_rate": 1e-06, + "loss": 0.588, + "mean_token_accuracy": 0.8120037317276001, + "num_tokens": 530012145.0, + "step": 21276 + }, + { + "epoch": 1.7545147192215715, + "grad_norm": 1.6944429874420166, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7800813317298889, + "num_tokens": 530047299.0, + "step": 21277 + }, + { + "epoch": 1.7545971798466233, + "grad_norm": 1.9275269508361816, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7520741820335388, + "num_tokens": 530078311.0, + "step": 21278 + }, + { + "epoch": 1.7546796404716747, + "grad_norm": 1.9659608602523804, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7586864829063416, + "num_tokens": 530107725.0, + "step": 21279 + }, + { + "epoch": 1.7547621010967265, + "grad_norm": 1.7966277599334717, + "learning_rate": 1e-06, + "loss": 0.582, + "mean_token_accuracy": 0.7959159016609192, + "num_tokens": 530142063.0, + "step": 21280 + }, + { + "epoch": 1.7548445617217778, + "grad_norm": 1.8249825239181519, + "learning_rate": 1e-06, + "loss": 0.5563, + "mean_token_accuracy": 0.8053656816482544, + "num_tokens": 530174172.0, + "step": 21281 + }, + { + "epoch": 1.7549270223468294, + "grad_norm": 1.8962265253067017, + "learning_rate": 1e-06, + "loss": 0.6147, + "mean_token_accuracy": 0.7889572978019714, + "num_tokens": 530202863.0, + "step": 21282 + }, + { + "epoch": 1.755009482971881, + "grad_norm": 2.1142935752868652, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7652993202209473, + "num_tokens": 530231103.0, + "step": 21283 + }, + { + "epoch": 1.7550919435969323, + "grad_norm": 2.110220432281494, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.73665452003479, + "num_tokens": 530260985.0, + "step": 21284 + }, + { + "epoch": 1.7551744042219841, + "grad_norm": 2.1465795040130615, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7127010822296143, + "num_tokens": 530286943.0, + "step": 21285 + }, + { + "epoch": 1.7552568648470355, + "grad_norm": 2.1954288482666016, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7617378234863281, + "num_tokens": 530312631.0, + "step": 21286 + }, + { + "epoch": 1.755339325472087, + "grad_norm": 1.7869778871536255, + "learning_rate": 1e-06, + "loss": 0.4938, + "mean_token_accuracy": 0.8146999478340149, + "num_tokens": 530345131.0, + "step": 21287 + }, + { + "epoch": 1.7554217860971386, + "grad_norm": 2.248227119445801, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7583092451095581, + "num_tokens": 530370237.0, + "step": 21288 + }, + { + "epoch": 1.7555042467221902, + "grad_norm": 2.0608417987823486, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7452177405357361, + "num_tokens": 530399960.0, + "step": 21289 + }, + { + "epoch": 1.7555867073472418, + "grad_norm": 2.1132137775421143, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7515887022018433, + "num_tokens": 530426875.0, + "step": 21290 + }, + { + "epoch": 1.755669167972293, + "grad_norm": 2.001363754272461, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7519774436950684, + "num_tokens": 530458479.0, + "step": 21291 + }, + { + "epoch": 1.755751628597345, + "grad_norm": 1.8395099639892578, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7642427682876587, + "num_tokens": 530494963.0, + "step": 21292 + }, + { + "epoch": 1.7558340892223963, + "grad_norm": 2.3899412155151367, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7170529961585999, + "num_tokens": 530519462.0, + "step": 21293 + }, + { + "epoch": 1.7559165498474478, + "grad_norm": 2.246429920196533, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7660123109817505, + "num_tokens": 530543397.0, + "step": 21294 + }, + { + "epoch": 1.7559990104724994, + "grad_norm": 1.9925225973129272, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7749349474906921, + "num_tokens": 530572841.0, + "step": 21295 + }, + { + "epoch": 1.756081471097551, + "grad_norm": 1.7752344608306885, + "learning_rate": 1e-06, + "loss": 0.5505, + "mean_token_accuracy": 0.81605464220047, + "num_tokens": 530604952.0, + "step": 21296 + }, + { + "epoch": 1.7561639317226025, + "grad_norm": 1.6560664176940918, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7848194241523743, + "num_tokens": 530644597.0, + "step": 21297 + }, + { + "epoch": 1.756246392347654, + "grad_norm": 2.0530765056610107, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.7504123449325562, + "num_tokens": 530674421.0, + "step": 21298 + }, + { + "epoch": 1.7563288529727057, + "grad_norm": 1.9521864652633667, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7142797708511353, + "num_tokens": 530708342.0, + "step": 21299 + }, + { + "epoch": 1.756411313597757, + "grad_norm": 2.0234904289245605, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7448436617851257, + "num_tokens": 530738064.0, + "step": 21300 + }, + { + "epoch": 1.7564937742228086, + "grad_norm": 1.9434279203414917, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7601429224014282, + "num_tokens": 530766593.0, + "step": 21301 + }, + { + "epoch": 1.7565762348478602, + "grad_norm": 2.114057779312134, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7269521951675415, + "num_tokens": 530793100.0, + "step": 21302 + }, + { + "epoch": 1.7566586954729115, + "grad_norm": 1.8981560468673706, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7425376176834106, + "num_tokens": 530825860.0, + "step": 21303 + }, + { + "epoch": 1.7567411560979633, + "grad_norm": 1.9795082807540894, + "learning_rate": 1e-06, + "loss": 0.7614, + "mean_token_accuracy": 0.7534445524215698, + "num_tokens": 530854502.0, + "step": 21304 + }, + { + "epoch": 1.7568236167230147, + "grad_norm": 1.8624287843704224, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7791682481765747, + "num_tokens": 530888294.0, + "step": 21305 + }, + { + "epoch": 1.7569060773480663, + "grad_norm": 2.006938934326172, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7571449279785156, + "num_tokens": 530918343.0, + "step": 21306 + }, + { + "epoch": 1.7569885379731178, + "grad_norm": 1.8725993633270264, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7723429203033447, + "num_tokens": 530949546.0, + "step": 21307 + }, + { + "epoch": 1.7570709985981694, + "grad_norm": 1.753998875617981, + "learning_rate": 1e-06, + "loss": 0.5224, + "mean_token_accuracy": 0.8290092945098877, + "num_tokens": 530982424.0, + "step": 21308 + }, + { + "epoch": 1.757153459223221, + "grad_norm": 2.4077227115631104, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7529590129852295, + "num_tokens": 531003839.0, + "step": 21309 + }, + { + "epoch": 1.7572359198482723, + "grad_norm": 2.043149948120117, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7613883018493652, + "num_tokens": 531031341.0, + "step": 21310 + }, + { + "epoch": 1.7573183804733241, + "grad_norm": 2.212205171585083, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7613422274589539, + "num_tokens": 531056792.0, + "step": 21311 + }, + { + "epoch": 1.7574008410983755, + "grad_norm": 1.8897814750671387, + "learning_rate": 1e-06, + "loss": 0.5698, + "mean_token_accuracy": 0.8123927116394043, + "num_tokens": 531084961.0, + "step": 21312 + }, + { + "epoch": 1.757483301723427, + "grad_norm": 1.8963441848754883, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.750046968460083, + "num_tokens": 531121399.0, + "step": 21313 + }, + { + "epoch": 1.7575657623484786, + "grad_norm": 1.8919705152511597, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7769932746887207, + "num_tokens": 531153380.0, + "step": 21314 + }, + { + "epoch": 1.7576482229735302, + "grad_norm": 2.1226537227630615, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.7899811267852783, + "num_tokens": 531179512.0, + "step": 21315 + }, + { + "epoch": 1.7577306835985818, + "grad_norm": 2.0130820274353027, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.747050940990448, + "num_tokens": 531206426.0, + "step": 21316 + }, + { + "epoch": 1.757813144223633, + "grad_norm": 1.7198377847671509, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.7954140901565552, + "num_tokens": 531240308.0, + "step": 21317 + }, + { + "epoch": 1.757895604848685, + "grad_norm": 1.9299010038375854, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7685439586639404, + "num_tokens": 531271310.0, + "step": 21318 + }, + { + "epoch": 1.7579780654737363, + "grad_norm": 2.1620960235595703, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7264696955680847, + "num_tokens": 531299197.0, + "step": 21319 + }, + { + "epoch": 1.7580605260987878, + "grad_norm": 2.0399959087371826, + "learning_rate": 1e-06, + "loss": 0.5907, + "mean_token_accuracy": 0.7965360879898071, + "num_tokens": 531326090.0, + "step": 21320 + }, + { + "epoch": 1.7581429867238394, + "grad_norm": 2.0687668323516846, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7702429294586182, + "num_tokens": 531352598.0, + "step": 21321 + }, + { + "epoch": 1.7582254473488907, + "grad_norm": 2.2431323528289795, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7291796207427979, + "num_tokens": 531375487.0, + "step": 21322 + }, + { + "epoch": 1.7583079079739425, + "grad_norm": 1.7521438598632812, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7678985595703125, + "num_tokens": 531410243.0, + "step": 21323 + }, + { + "epoch": 1.758390368598994, + "grad_norm": 2.134361743927002, + "learning_rate": 1e-06, + "loss": 0.5455, + "mean_token_accuracy": 0.8083466291427612, + "num_tokens": 531433409.0, + "step": 21324 + }, + { + "epoch": 1.7584728292240455, + "grad_norm": 2.4111108779907227, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7300037145614624, + "num_tokens": 531457213.0, + "step": 21325 + }, + { + "epoch": 1.758555289849097, + "grad_norm": 1.8290311098098755, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.7968514561653137, + "num_tokens": 531487697.0, + "step": 21326 + }, + { + "epoch": 1.7586377504741486, + "grad_norm": 1.8202837705612183, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7440624237060547, + "num_tokens": 531522702.0, + "step": 21327 + }, + { + "epoch": 1.7587202110992002, + "grad_norm": 1.9604789018630981, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.796952486038208, + "num_tokens": 531550518.0, + "step": 21328 + }, + { + "epoch": 1.7588026717242515, + "grad_norm": 2.423276662826538, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7543460130691528, + "num_tokens": 531573138.0, + "step": 21329 + }, + { + "epoch": 1.7588851323493033, + "grad_norm": 2.2436673641204834, + "learning_rate": 1e-06, + "loss": 0.7239, + "mean_token_accuracy": 0.7750685811042786, + "num_tokens": 531597375.0, + "step": 21330 + }, + { + "epoch": 1.7589675929743547, + "grad_norm": 1.9383949041366577, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7147802710533142, + "num_tokens": 531631989.0, + "step": 21331 + }, + { + "epoch": 1.7590500535994062, + "grad_norm": 1.6809049844741821, + "learning_rate": 1e-06, + "loss": 0.5474, + "mean_token_accuracy": 0.822258472442627, + "num_tokens": 531667459.0, + "step": 21332 + }, + { + "epoch": 1.7591325142244578, + "grad_norm": 1.8271368741989136, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.7929123640060425, + "num_tokens": 531699867.0, + "step": 21333 + }, + { + "epoch": 1.7592149748495094, + "grad_norm": 1.9797585010528564, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7719931602478027, + "num_tokens": 531729654.0, + "step": 21334 + }, + { + "epoch": 1.759297435474561, + "grad_norm": 2.0201423168182373, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7484053373336792, + "num_tokens": 531761183.0, + "step": 21335 + }, + { + "epoch": 1.7593798960996123, + "grad_norm": 1.9934616088867188, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7287471890449524, + "num_tokens": 531795992.0, + "step": 21336 + }, + { + "epoch": 1.7594623567246641, + "grad_norm": 2.1939518451690674, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7263464331626892, + "num_tokens": 531824120.0, + "step": 21337 + }, + { + "epoch": 1.7595448173497155, + "grad_norm": 2.2036116123199463, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7468711137771606, + "num_tokens": 531849739.0, + "step": 21338 + }, + { + "epoch": 1.759627277974767, + "grad_norm": 2.180259943008423, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7656186819076538, + "num_tokens": 531874483.0, + "step": 21339 + }, + { + "epoch": 1.7597097385998186, + "grad_norm": 2.277621269226074, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.745459258556366, + "num_tokens": 531898777.0, + "step": 21340 + }, + { + "epoch": 1.7597921992248702, + "grad_norm": 1.9697659015655518, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.7794692516326904, + "num_tokens": 531924822.0, + "step": 21341 + }, + { + "epoch": 1.7598746598499218, + "grad_norm": 1.932417631149292, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7666035294532776, + "num_tokens": 531955033.0, + "step": 21342 + }, + { + "epoch": 1.759957120474973, + "grad_norm": 1.975523591041565, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7539863586425781, + "num_tokens": 531983922.0, + "step": 21343 + }, + { + "epoch": 1.760039581100025, + "grad_norm": 2.0790836811065674, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7402685880661011, + "num_tokens": 532012189.0, + "step": 21344 + }, + { + "epoch": 1.7601220417250762, + "grad_norm": 1.8874285221099854, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7526184320449829, + "num_tokens": 532044086.0, + "step": 21345 + }, + { + "epoch": 1.7602045023501278, + "grad_norm": 2.3478341102600098, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6992310285568237, + "num_tokens": 532068355.0, + "step": 21346 + }, + { + "epoch": 1.7602869629751794, + "grad_norm": 1.8975257873535156, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7892321348190308, + "num_tokens": 532099390.0, + "step": 21347 + }, + { + "epoch": 1.7603694236002307, + "grad_norm": 2.216297149658203, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7289164662361145, + "num_tokens": 532127045.0, + "step": 21348 + }, + { + "epoch": 1.7604518842252825, + "grad_norm": 1.909538745880127, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.752705454826355, + "num_tokens": 532158717.0, + "step": 21349 + }, + { + "epoch": 1.7605343448503339, + "grad_norm": 2.0088160037994385, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7446051239967346, + "num_tokens": 532189302.0, + "step": 21350 + }, + { + "epoch": 1.7606168054753855, + "grad_norm": 2.0244226455688477, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7495012283325195, + "num_tokens": 532218577.0, + "step": 21351 + }, + { + "epoch": 1.760699266100437, + "grad_norm": 1.6793909072875977, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.792417585849762, + "num_tokens": 532257051.0, + "step": 21352 + }, + { + "epoch": 1.7607817267254886, + "grad_norm": 2.241607189178467, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7272853851318359, + "num_tokens": 532281829.0, + "step": 21353 + }, + { + "epoch": 1.7608641873505402, + "grad_norm": 2.147062301635742, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7297441959381104, + "num_tokens": 532309686.0, + "step": 21354 + }, + { + "epoch": 1.7609466479755915, + "grad_norm": 1.9498472213745117, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.742613673210144, + "num_tokens": 532339661.0, + "step": 21355 + }, + { + "epoch": 1.7610291086006433, + "grad_norm": 1.935662865638733, + "learning_rate": 1e-06, + "loss": 0.5881, + "mean_token_accuracy": 0.8058664798736572, + "num_tokens": 532367724.0, + "step": 21356 + }, + { + "epoch": 1.7611115692256947, + "grad_norm": 1.9765902757644653, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7603011727333069, + "num_tokens": 532398036.0, + "step": 21357 + }, + { + "epoch": 1.7611940298507462, + "grad_norm": 1.9714550971984863, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.7954529523849487, + "num_tokens": 532431711.0, + "step": 21358 + }, + { + "epoch": 1.7612764904757978, + "grad_norm": 1.8400145769119263, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7747060656547546, + "num_tokens": 532466897.0, + "step": 21359 + }, + { + "epoch": 1.7613589511008494, + "grad_norm": 1.8879082202911377, + "learning_rate": 1e-06, + "loss": 0.5669, + "mean_token_accuracy": 0.7991673946380615, + "num_tokens": 532495214.0, + "step": 21360 + }, + { + "epoch": 1.761441411725901, + "grad_norm": 1.8989652395248413, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7754276990890503, + "num_tokens": 532525889.0, + "step": 21361 + }, + { + "epoch": 1.7615238723509523, + "grad_norm": 2.0402841567993164, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7283902764320374, + "num_tokens": 532555306.0, + "step": 21362 + }, + { + "epoch": 1.761606332976004, + "grad_norm": 1.9762529134750366, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7797427773475647, + "num_tokens": 532587335.0, + "step": 21363 + }, + { + "epoch": 1.7616887936010555, + "grad_norm": 1.9073978662490845, + "learning_rate": 1e-06, + "loss": 0.5669, + "mean_token_accuracy": 0.8138741254806519, + "num_tokens": 532618178.0, + "step": 21364 + }, + { + "epoch": 1.761771254226107, + "grad_norm": 1.9923455715179443, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7701377272605896, + "num_tokens": 532651393.0, + "step": 21365 + }, + { + "epoch": 1.7618537148511586, + "grad_norm": 2.0004661083221436, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7765769958496094, + "num_tokens": 532680895.0, + "step": 21366 + }, + { + "epoch": 1.76193617547621, + "grad_norm": 1.8592188358306885, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7659848928451538, + "num_tokens": 532714844.0, + "step": 21367 + }, + { + "epoch": 1.7620186361012617, + "grad_norm": 2.217299699783325, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7401940822601318, + "num_tokens": 532741366.0, + "step": 21368 + }, + { + "epoch": 1.762101096726313, + "grad_norm": 2.0874228477478027, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7890207767486572, + "num_tokens": 532769405.0, + "step": 21369 + }, + { + "epoch": 1.7621835573513647, + "grad_norm": 1.793649673461914, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7650068998336792, + "num_tokens": 532806585.0, + "step": 21370 + }, + { + "epoch": 1.7622660179764162, + "grad_norm": 2.1685893535614014, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7204761505126953, + "num_tokens": 532835674.0, + "step": 21371 + }, + { + "epoch": 1.7623484786014678, + "grad_norm": 1.9924522638320923, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7677018642425537, + "num_tokens": 532864482.0, + "step": 21372 + }, + { + "epoch": 1.7624309392265194, + "grad_norm": 1.8281391859054565, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.7443387508392334, + "num_tokens": 532898943.0, + "step": 21373 + }, + { + "epoch": 1.7625133998515707, + "grad_norm": 1.8613334894180298, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7355831265449524, + "num_tokens": 532935049.0, + "step": 21374 + }, + { + "epoch": 1.7625958604766225, + "grad_norm": 2.210827112197876, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7452439665794373, + "num_tokens": 532960614.0, + "step": 21375 + }, + { + "epoch": 1.7626783211016739, + "grad_norm": 1.838422417640686, + "learning_rate": 1e-06, + "loss": 0.7483, + "mean_token_accuracy": 0.7694091200828552, + "num_tokens": 532992791.0, + "step": 21376 + }, + { + "epoch": 1.7627607817267255, + "grad_norm": 2.0978009700775146, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7664998769760132, + "num_tokens": 533018756.0, + "step": 21377 + }, + { + "epoch": 1.762843242351777, + "grad_norm": 2.0214107036590576, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.7822552919387817, + "num_tokens": 533044826.0, + "step": 21378 + }, + { + "epoch": 1.7629257029768286, + "grad_norm": 1.8748048543930054, + "learning_rate": 1e-06, + "loss": 0.6003, + "mean_token_accuracy": 0.7934216260910034, + "num_tokens": 533075301.0, + "step": 21379 + }, + { + "epoch": 1.7630081636018802, + "grad_norm": 2.102316379547119, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7806253433227539, + "num_tokens": 533104251.0, + "step": 21380 + }, + { + "epoch": 1.7630906242269315, + "grad_norm": 2.031722068786621, + "learning_rate": 1e-06, + "loss": 0.5852, + "mean_token_accuracy": 0.8081607818603516, + "num_tokens": 533132855.0, + "step": 21381 + }, + { + "epoch": 1.7631730848519833, + "grad_norm": 2.30696964263916, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7533875703811646, + "num_tokens": 533160171.0, + "step": 21382 + }, + { + "epoch": 1.7632555454770347, + "grad_norm": 1.6832534074783325, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.7893933653831482, + "num_tokens": 533200784.0, + "step": 21383 + }, + { + "epoch": 1.7633380061020862, + "grad_norm": 2.1361324787139893, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7388596534729004, + "num_tokens": 533228108.0, + "step": 21384 + }, + { + "epoch": 1.7634204667271378, + "grad_norm": 1.966334581375122, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.7922487258911133, + "num_tokens": 533259788.0, + "step": 21385 + }, + { + "epoch": 1.7635029273521892, + "grad_norm": 1.8791025876998901, + "learning_rate": 1e-06, + "loss": 0.7621, + "mean_token_accuracy": 0.7586156129837036, + "num_tokens": 533292823.0, + "step": 21386 + }, + { + "epoch": 1.763585387977241, + "grad_norm": 2.0849626064300537, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7940562963485718, + "num_tokens": 533320992.0, + "step": 21387 + }, + { + "epoch": 1.7636678486022923, + "grad_norm": 2.2168025970458984, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7567522525787354, + "num_tokens": 533344536.0, + "step": 21388 + }, + { + "epoch": 1.763750309227344, + "grad_norm": 1.7913870811462402, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7866711616516113, + "num_tokens": 533377940.0, + "step": 21389 + }, + { + "epoch": 1.7638327698523955, + "grad_norm": 1.8319571018218994, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.773232102394104, + "num_tokens": 533408451.0, + "step": 21390 + }, + { + "epoch": 1.763915230477447, + "grad_norm": 2.0595011711120605, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7763558626174927, + "num_tokens": 533435394.0, + "step": 21391 + }, + { + "epoch": 1.7639976911024986, + "grad_norm": 2.1107115745544434, + "learning_rate": 1e-06, + "loss": 0.6023, + "mean_token_accuracy": 0.7984272241592407, + "num_tokens": 533460270.0, + "step": 21392 + }, + { + "epoch": 1.76408015172755, + "grad_norm": 1.9941496849060059, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7837920188903809, + "num_tokens": 533488945.0, + "step": 21393 + }, + { + "epoch": 1.7641626123526017, + "grad_norm": 2.0056633949279785, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7647411227226257, + "num_tokens": 533515791.0, + "step": 21394 + }, + { + "epoch": 1.764245072977653, + "grad_norm": 1.8684412240982056, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7557356357574463, + "num_tokens": 533550242.0, + "step": 21395 + }, + { + "epoch": 1.7643275336027047, + "grad_norm": 1.9705226421356201, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7658594250679016, + "num_tokens": 533581868.0, + "step": 21396 + }, + { + "epoch": 1.7644099942277562, + "grad_norm": 1.8730558156967163, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7694236040115356, + "num_tokens": 533612822.0, + "step": 21397 + }, + { + "epoch": 1.7644924548528078, + "grad_norm": 2.0291571617126465, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7758347988128662, + "num_tokens": 533642585.0, + "step": 21398 + }, + { + "epoch": 1.7645749154778594, + "grad_norm": 2.2545437812805176, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7493828535079956, + "num_tokens": 533667725.0, + "step": 21399 + }, + { + "epoch": 1.7646573761029107, + "grad_norm": 1.8085507154464722, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.7875260710716248, + "num_tokens": 533700357.0, + "step": 21400 + }, + { + "epoch": 1.7647398367279625, + "grad_norm": 1.9560667276382446, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7585458159446716, + "num_tokens": 533730994.0, + "step": 21401 + }, + { + "epoch": 1.7648222973530139, + "grad_norm": 1.742244839668274, + "learning_rate": 1e-06, + "loss": 0.5673, + "mean_token_accuracy": 0.8115352392196655, + "num_tokens": 533765381.0, + "step": 21402 + }, + { + "epoch": 1.7649047579780655, + "grad_norm": 1.887410044670105, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7714815139770508, + "num_tokens": 533795108.0, + "step": 21403 + }, + { + "epoch": 1.764987218603117, + "grad_norm": 1.7478675842285156, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7369811534881592, + "num_tokens": 533831867.0, + "step": 21404 + }, + { + "epoch": 1.7650696792281686, + "grad_norm": 2.1356701850891113, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.7754400968551636, + "num_tokens": 533860649.0, + "step": 21405 + }, + { + "epoch": 1.7651521398532202, + "grad_norm": 1.8160263299942017, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7626435160636902, + "num_tokens": 533892799.0, + "step": 21406 + }, + { + "epoch": 1.7652346004782715, + "grad_norm": 1.868070363998413, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7527657747268677, + "num_tokens": 533925583.0, + "step": 21407 + }, + { + "epoch": 1.7653170611033233, + "grad_norm": 1.7713215351104736, + "learning_rate": 1e-06, + "loss": 0.5712, + "mean_token_accuracy": 0.8034824132919312, + "num_tokens": 533961509.0, + "step": 21408 + }, + { + "epoch": 1.7653995217283747, + "grad_norm": 1.8122478723526, + "learning_rate": 1e-06, + "loss": 0.5772, + "mean_token_accuracy": 0.8019595146179199, + "num_tokens": 533994265.0, + "step": 21409 + }, + { + "epoch": 1.7654819823534262, + "grad_norm": 2.0115208625793457, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7268016338348389, + "num_tokens": 534029293.0, + "step": 21410 + }, + { + "epoch": 1.7655644429784778, + "grad_norm": 1.8265836238861084, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7475508451461792, + "num_tokens": 534063953.0, + "step": 21411 + }, + { + "epoch": 1.7656469036035292, + "grad_norm": 2.0443758964538574, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7520160675048828, + "num_tokens": 534095297.0, + "step": 21412 + }, + { + "epoch": 1.765729364228581, + "grad_norm": 2.230196237564087, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.727368950843811, + "num_tokens": 534121215.0, + "step": 21413 + }, + { + "epoch": 1.7658118248536323, + "grad_norm": 2.0737459659576416, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7479175925254822, + "num_tokens": 534148436.0, + "step": 21414 + }, + { + "epoch": 1.7658942854786839, + "grad_norm": 2.2061593532562256, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.777823269367218, + "num_tokens": 534171177.0, + "step": 21415 + }, + { + "epoch": 1.7659767461037355, + "grad_norm": 1.8573981523513794, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.773575484752655, + "num_tokens": 534203138.0, + "step": 21416 + }, + { + "epoch": 1.766059206728787, + "grad_norm": 1.9263415336608887, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7585965394973755, + "num_tokens": 534234260.0, + "step": 21417 + }, + { + "epoch": 1.7661416673538386, + "grad_norm": 2.3798742294311523, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7506533861160278, + "num_tokens": 534255009.0, + "step": 21418 + }, + { + "epoch": 1.76622412797889, + "grad_norm": 2.0864908695220947, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7337559461593628, + "num_tokens": 534283209.0, + "step": 21419 + }, + { + "epoch": 1.7663065886039417, + "grad_norm": 2.06850004196167, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7548272609710693, + "num_tokens": 534314629.0, + "step": 21420 + }, + { + "epoch": 1.766389049228993, + "grad_norm": 2.0287835597991943, + "learning_rate": 1e-06, + "loss": 0.5812, + "mean_token_accuracy": 0.8054873943328857, + "num_tokens": 534342039.0, + "step": 21421 + }, + { + "epoch": 1.7664715098540447, + "grad_norm": 1.9305909872055054, + "learning_rate": 1e-06, + "loss": 0.5679, + "mean_token_accuracy": 0.8057811856269836, + "num_tokens": 534370855.0, + "step": 21422 + }, + { + "epoch": 1.7665539704790962, + "grad_norm": 2.1006710529327393, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7618292570114136, + "num_tokens": 534398588.0, + "step": 21423 + }, + { + "epoch": 1.7666364311041478, + "grad_norm": 2.234792709350586, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7793178558349609, + "num_tokens": 534422966.0, + "step": 21424 + }, + { + "epoch": 1.7667188917291994, + "grad_norm": 1.8510487079620361, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.7927399277687073, + "num_tokens": 534454840.0, + "step": 21425 + }, + { + "epoch": 1.7668013523542507, + "grad_norm": 2.063929319381714, + "learning_rate": 1e-06, + "loss": 0.5955, + "mean_token_accuracy": 0.8038408756256104, + "num_tokens": 534480577.0, + "step": 21426 + }, + { + "epoch": 1.7668838129793025, + "grad_norm": 2.260009765625, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7462506890296936, + "num_tokens": 534504160.0, + "step": 21427 + }, + { + "epoch": 1.7669662736043539, + "grad_norm": 2.091918468475342, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7685551643371582, + "num_tokens": 534532989.0, + "step": 21428 + }, + { + "epoch": 1.7670487342294054, + "grad_norm": 1.894030213356018, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7719120979309082, + "num_tokens": 534563345.0, + "step": 21429 + }, + { + "epoch": 1.767131194854457, + "grad_norm": 2.0709171295166016, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7815179228782654, + "num_tokens": 534593010.0, + "step": 21430 + }, + { + "epoch": 1.7672136554795084, + "grad_norm": 1.8911329507827759, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7646251916885376, + "num_tokens": 534626614.0, + "step": 21431 + }, + { + "epoch": 1.7672961161045602, + "grad_norm": 2.1152493953704834, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7796436548233032, + "num_tokens": 534654798.0, + "step": 21432 + }, + { + "epoch": 1.7673785767296115, + "grad_norm": 2.108405113220215, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7584542036056519, + "num_tokens": 534680846.0, + "step": 21433 + }, + { + "epoch": 1.7674610373546633, + "grad_norm": 2.0180466175079346, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7758640646934509, + "num_tokens": 534710180.0, + "step": 21434 + }, + { + "epoch": 1.7675434979797147, + "grad_norm": 2.1738827228546143, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7361846566200256, + "num_tokens": 534735871.0, + "step": 21435 + }, + { + "epoch": 1.7676259586047662, + "grad_norm": 1.929377794265747, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7637869119644165, + "num_tokens": 534764898.0, + "step": 21436 + }, + { + "epoch": 1.7677084192298178, + "grad_norm": 2.2190518379211426, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7512319087982178, + "num_tokens": 534789361.0, + "step": 21437 + }, + { + "epoch": 1.7677908798548692, + "grad_norm": 1.8526971340179443, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.7872884273529053, + "num_tokens": 534820049.0, + "step": 21438 + }, + { + "epoch": 1.767873340479921, + "grad_norm": 2.5898149013519287, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7543042898178101, + "num_tokens": 534840245.0, + "step": 21439 + }, + { + "epoch": 1.7679558011049723, + "grad_norm": 1.9794301986694336, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7432159185409546, + "num_tokens": 534868720.0, + "step": 21440 + }, + { + "epoch": 1.7680382617300239, + "grad_norm": 1.9750545024871826, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.7938297986984253, + "num_tokens": 534894484.0, + "step": 21441 + }, + { + "epoch": 1.7681207223550754, + "grad_norm": 1.8683769702911377, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.7885466814041138, + "num_tokens": 534925857.0, + "step": 21442 + }, + { + "epoch": 1.768203182980127, + "grad_norm": 1.7027031183242798, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.737779438495636, + "num_tokens": 534962977.0, + "step": 21443 + }, + { + "epoch": 1.7682856436051786, + "grad_norm": 1.879951000213623, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.7922639846801758, + "num_tokens": 534992451.0, + "step": 21444 + }, + { + "epoch": 1.76836810423023, + "grad_norm": 2.3801467418670654, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7586299777030945, + "num_tokens": 535015696.0, + "step": 21445 + }, + { + "epoch": 1.7684505648552817, + "grad_norm": 2.109781503677368, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7644362449645996, + "num_tokens": 535042712.0, + "step": 21446 + }, + { + "epoch": 1.768533025480333, + "grad_norm": 2.3357317447662354, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7657778263092041, + "num_tokens": 535065851.0, + "step": 21447 + }, + { + "epoch": 1.7686154861053847, + "grad_norm": 1.987923502922058, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.760343611240387, + "num_tokens": 535096014.0, + "step": 21448 + }, + { + "epoch": 1.7686979467304362, + "grad_norm": 2.0083799362182617, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7895684242248535, + "num_tokens": 535125151.0, + "step": 21449 + }, + { + "epoch": 1.7687804073554878, + "grad_norm": 1.9243700504302979, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7791309356689453, + "num_tokens": 535159822.0, + "step": 21450 + }, + { + "epoch": 1.7688628679805394, + "grad_norm": 1.8891774415969849, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7620099782943726, + "num_tokens": 535193157.0, + "step": 21451 + }, + { + "epoch": 1.7689453286055907, + "grad_norm": 2.055354118347168, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7767665386199951, + "num_tokens": 535219545.0, + "step": 21452 + }, + { + "epoch": 1.7690277892306425, + "grad_norm": 2.3422608375549316, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7793437242507935, + "num_tokens": 535240806.0, + "step": 21453 + }, + { + "epoch": 1.7691102498556939, + "grad_norm": 1.9708201885223389, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7115461826324463, + "num_tokens": 535275484.0, + "step": 21454 + }, + { + "epoch": 1.7691927104807454, + "grad_norm": 1.9737516641616821, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7096598148345947, + "num_tokens": 535307184.0, + "step": 21455 + }, + { + "epoch": 1.769275171105797, + "grad_norm": 2.2432861328125, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7735607624053955, + "num_tokens": 535328656.0, + "step": 21456 + }, + { + "epoch": 1.7693576317308484, + "grad_norm": 2.1255249977111816, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7624434232711792, + "num_tokens": 535356289.0, + "step": 21457 + }, + { + "epoch": 1.7694400923559002, + "grad_norm": 2.1378183364868164, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7525191307067871, + "num_tokens": 535384120.0, + "step": 21458 + }, + { + "epoch": 1.7695225529809515, + "grad_norm": 2.190981149673462, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7736307978630066, + "num_tokens": 535408317.0, + "step": 21459 + }, + { + "epoch": 1.769605013606003, + "grad_norm": 1.9773976802825928, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7696830034255981, + "num_tokens": 535439854.0, + "step": 21460 + }, + { + "epoch": 1.7696874742310547, + "grad_norm": 1.925323724746704, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7597581744194031, + "num_tokens": 535473794.0, + "step": 21461 + }, + { + "epoch": 1.7697699348561062, + "grad_norm": 1.870584487915039, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.7532501220703125, + "num_tokens": 535506578.0, + "step": 21462 + }, + { + "epoch": 1.7698523954811578, + "grad_norm": 1.9649728536605835, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7575644254684448, + "num_tokens": 535535890.0, + "step": 21463 + }, + { + "epoch": 1.7699348561062092, + "grad_norm": 1.9534603357315063, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7146825790405273, + "num_tokens": 535569079.0, + "step": 21464 + }, + { + "epoch": 1.770017316731261, + "grad_norm": 1.7317450046539307, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7634615898132324, + "num_tokens": 535603199.0, + "step": 21465 + }, + { + "epoch": 1.7700997773563123, + "grad_norm": 1.8128786087036133, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7778924107551575, + "num_tokens": 535635664.0, + "step": 21466 + }, + { + "epoch": 1.7701822379813639, + "grad_norm": 1.841481328010559, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.773464560508728, + "num_tokens": 535668803.0, + "step": 21467 + }, + { + "epoch": 1.7702646986064154, + "grad_norm": 1.8176652193069458, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.8009825348854065, + "num_tokens": 535699592.0, + "step": 21468 + }, + { + "epoch": 1.770347159231467, + "grad_norm": 1.8974838256835938, + "learning_rate": 1e-06, + "loss": 0.6177, + "mean_token_accuracy": 0.7958365678787231, + "num_tokens": 535731031.0, + "step": 21469 + }, + { + "epoch": 1.7704296198565186, + "grad_norm": 1.923232078552246, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7773308157920837, + "num_tokens": 535761552.0, + "step": 21470 + }, + { + "epoch": 1.77051208048157, + "grad_norm": 2.3532161712646484, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.7479645609855652, + "num_tokens": 535785368.0, + "step": 21471 + }, + { + "epoch": 1.7705945411066217, + "grad_norm": 1.8891552686691284, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.8006088137626648, + "num_tokens": 535815735.0, + "step": 21472 + }, + { + "epoch": 1.770677001731673, + "grad_norm": 1.9583401679992676, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7594631314277649, + "num_tokens": 535845460.0, + "step": 21473 + }, + { + "epoch": 1.7707594623567247, + "grad_norm": 2.1565351486206055, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.784131646156311, + "num_tokens": 535868435.0, + "step": 21474 + }, + { + "epoch": 1.7708419229817762, + "grad_norm": 2.205355644226074, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7625506520271301, + "num_tokens": 535895875.0, + "step": 21475 + }, + { + "epoch": 1.7709243836068276, + "grad_norm": 1.987212896347046, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7722867727279663, + "num_tokens": 535924536.0, + "step": 21476 + }, + { + "epoch": 1.7710068442318794, + "grad_norm": 2.2904510498046875, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7387802600860596, + "num_tokens": 535948091.0, + "step": 21477 + }, + { + "epoch": 1.7710893048569307, + "grad_norm": 2.1938021183013916, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.7860269546508789, + "num_tokens": 535971660.0, + "step": 21478 + }, + { + "epoch": 1.7711717654819825, + "grad_norm": 2.118137836456299, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7540784478187561, + "num_tokens": 535998704.0, + "step": 21479 + }, + { + "epoch": 1.7712542261070339, + "grad_norm": 1.8840699195861816, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7291698455810547, + "num_tokens": 536032348.0, + "step": 21480 + }, + { + "epoch": 1.7713366867320854, + "grad_norm": 2.2123162746429443, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.7494052648544312, + "num_tokens": 536057206.0, + "step": 21481 + }, + { + "epoch": 1.771419147357137, + "grad_norm": 1.9334139823913574, + "learning_rate": 1e-06, + "loss": 0.5782, + "mean_token_accuracy": 0.8019940853118896, + "num_tokens": 536084254.0, + "step": 21482 + }, + { + "epoch": 1.7715016079821884, + "grad_norm": 1.9820417165756226, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7640126347541809, + "num_tokens": 536113614.0, + "step": 21483 + }, + { + "epoch": 1.7715840686072402, + "grad_norm": 1.8790472745895386, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7670438289642334, + "num_tokens": 536145917.0, + "step": 21484 + }, + { + "epoch": 1.7716665292322915, + "grad_norm": 2.019519567489624, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7299401760101318, + "num_tokens": 536175157.0, + "step": 21485 + }, + { + "epoch": 1.771748989857343, + "grad_norm": 1.9051655530929565, + "learning_rate": 1e-06, + "loss": 0.6233, + "mean_token_accuracy": 0.7973215579986572, + "num_tokens": 536204957.0, + "step": 21486 + }, + { + "epoch": 1.7718314504823947, + "grad_norm": 1.8695476055145264, + "learning_rate": 1e-06, + "loss": 0.6068, + "mean_token_accuracy": 0.7996668815612793, + "num_tokens": 536234247.0, + "step": 21487 + }, + { + "epoch": 1.7719139111074462, + "grad_norm": 1.9240572452545166, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7358194589614868, + "num_tokens": 536267979.0, + "step": 21488 + }, + { + "epoch": 1.7719963717324978, + "grad_norm": 1.68741774559021, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.7865846157073975, + "num_tokens": 536302991.0, + "step": 21489 + }, + { + "epoch": 1.7720788323575491, + "grad_norm": 2.0524392127990723, + "learning_rate": 1e-06, + "loss": 0.5784, + "mean_token_accuracy": 0.798685610294342, + "num_tokens": 536330411.0, + "step": 21490 + }, + { + "epoch": 1.772161292982601, + "grad_norm": 1.9703761339187622, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7822369337081909, + "num_tokens": 536362170.0, + "step": 21491 + }, + { + "epoch": 1.7722437536076523, + "grad_norm": 2.129716157913208, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.726803719997406, + "num_tokens": 536387741.0, + "step": 21492 + }, + { + "epoch": 1.7723262142327039, + "grad_norm": 2.023155689239502, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7624866962432861, + "num_tokens": 536417566.0, + "step": 21493 + }, + { + "epoch": 1.7724086748577554, + "grad_norm": 2.0012989044189453, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7869541645050049, + "num_tokens": 536447723.0, + "step": 21494 + }, + { + "epoch": 1.772491135482807, + "grad_norm": 2.1887454986572266, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7422496676445007, + "num_tokens": 536474572.0, + "step": 21495 + }, + { + "epoch": 1.7725735961078586, + "grad_norm": 1.9644041061401367, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7349367141723633, + "num_tokens": 536507264.0, + "step": 21496 + }, + { + "epoch": 1.77265605673291, + "grad_norm": 2.0527560710906982, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.7630665898323059, + "num_tokens": 536534899.0, + "step": 21497 + }, + { + "epoch": 1.7727385173579617, + "grad_norm": 2.0976970195770264, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7424885630607605, + "num_tokens": 536563223.0, + "step": 21498 + }, + { + "epoch": 1.772820977983013, + "grad_norm": 1.9484317302703857, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7337294816970825, + "num_tokens": 536596354.0, + "step": 21499 + }, + { + "epoch": 1.7729034386080647, + "grad_norm": 1.9446966648101807, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7685962915420532, + "num_tokens": 536624793.0, + "step": 21500 + }, + { + "epoch": 1.7729858992331162, + "grad_norm": 2.00223445892334, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7626603245735168, + "num_tokens": 536653080.0, + "step": 21501 + }, + { + "epoch": 1.7730683598581676, + "grad_norm": 1.847586750984192, + "learning_rate": 1e-06, + "loss": 0.5994, + "mean_token_accuracy": 0.8063626885414124, + "num_tokens": 536685209.0, + "step": 21502 + }, + { + "epoch": 1.7731508204832194, + "grad_norm": 2.1318273544311523, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7742624282836914, + "num_tokens": 536712955.0, + "step": 21503 + }, + { + "epoch": 1.7732332811082707, + "grad_norm": 2.0437889099121094, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7235281467437744, + "num_tokens": 536741287.0, + "step": 21504 + }, + { + "epoch": 1.7733157417333223, + "grad_norm": 1.9561653137207031, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7595686316490173, + "num_tokens": 536776769.0, + "step": 21505 + }, + { + "epoch": 1.7733982023583739, + "grad_norm": 2.2612252235412598, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7519683837890625, + "num_tokens": 536801715.0, + "step": 21506 + }, + { + "epoch": 1.7734806629834254, + "grad_norm": 1.8893711566925049, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7502709627151489, + "num_tokens": 536835037.0, + "step": 21507 + }, + { + "epoch": 1.773563123608477, + "grad_norm": 1.8566584587097168, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.7969827651977539, + "num_tokens": 536868962.0, + "step": 21508 + }, + { + "epoch": 1.7736455842335284, + "grad_norm": 1.8493993282318115, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7180843353271484, + "num_tokens": 536905985.0, + "step": 21509 + }, + { + "epoch": 1.7737280448585802, + "grad_norm": 1.8610141277313232, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7767577767372131, + "num_tokens": 536939948.0, + "step": 21510 + }, + { + "epoch": 1.7738105054836315, + "grad_norm": 1.8973976373672485, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7364327907562256, + "num_tokens": 536975580.0, + "step": 21511 + }, + { + "epoch": 1.773892966108683, + "grad_norm": 2.0490646362304688, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7501410245895386, + "num_tokens": 537003514.0, + "step": 21512 + }, + { + "epoch": 1.7739754267337347, + "grad_norm": 2.1034185886383057, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7324744462966919, + "num_tokens": 537030592.0, + "step": 21513 + }, + { + "epoch": 1.7740578873587862, + "grad_norm": 1.9833242893218994, + "learning_rate": 1e-06, + "loss": 0.5493, + "mean_token_accuracy": 0.8108528852462769, + "num_tokens": 537056063.0, + "step": 21514 + }, + { + "epoch": 1.7741403479838378, + "grad_norm": 2.0398311614990234, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.737583339214325, + "num_tokens": 537085469.0, + "step": 21515 + }, + { + "epoch": 1.7742228086088891, + "grad_norm": 2.2702174186706543, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7670084834098816, + "num_tokens": 537110083.0, + "step": 21516 + }, + { + "epoch": 1.774305269233941, + "grad_norm": 1.973191738128662, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7599924802780151, + "num_tokens": 537141137.0, + "step": 21517 + }, + { + "epoch": 1.7743877298589923, + "grad_norm": 1.9998294115066528, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7567331790924072, + "num_tokens": 537173496.0, + "step": 21518 + }, + { + "epoch": 1.7744701904840439, + "grad_norm": 2.1689932346343994, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.7967038154602051, + "num_tokens": 537197185.0, + "step": 21519 + }, + { + "epoch": 1.7745526511090954, + "grad_norm": 1.9619853496551514, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7376453876495361, + "num_tokens": 537231267.0, + "step": 21520 + }, + { + "epoch": 1.7746351117341468, + "grad_norm": 1.8588194847106934, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7330788373947144, + "num_tokens": 537265405.0, + "step": 21521 + }, + { + "epoch": 1.7747175723591986, + "grad_norm": 2.0266871452331543, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7850887775421143, + "num_tokens": 537297044.0, + "step": 21522 + }, + { + "epoch": 1.77480003298425, + "grad_norm": 1.8867188692092896, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7747751474380493, + "num_tokens": 537331160.0, + "step": 21523 + }, + { + "epoch": 1.7748824936093015, + "grad_norm": 1.9907798767089844, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7571573257446289, + "num_tokens": 537361190.0, + "step": 21524 + }, + { + "epoch": 1.774964954234353, + "grad_norm": 2.136754274368286, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7236368656158447, + "num_tokens": 537388155.0, + "step": 21525 + }, + { + "epoch": 1.7750474148594046, + "grad_norm": 2.133512020111084, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.728793740272522, + "num_tokens": 537414301.0, + "step": 21526 + }, + { + "epoch": 1.7751298754844562, + "grad_norm": 2.1273763179779053, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7696913480758667, + "num_tokens": 537438094.0, + "step": 21527 + }, + { + "epoch": 1.7752123361095076, + "grad_norm": 1.9857996702194214, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7588563561439514, + "num_tokens": 537469345.0, + "step": 21528 + }, + { + "epoch": 1.7752947967345594, + "grad_norm": 1.767221450805664, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7316772937774658, + "num_tokens": 537505968.0, + "step": 21529 + }, + { + "epoch": 1.7753772573596107, + "grad_norm": 2.0214414596557617, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7723038196563721, + "num_tokens": 537531874.0, + "step": 21530 + }, + { + "epoch": 1.7754597179846623, + "grad_norm": 2.052633762359619, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.723477840423584, + "num_tokens": 537562401.0, + "step": 21531 + }, + { + "epoch": 1.7755421786097139, + "grad_norm": 1.927095890045166, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7407636642456055, + "num_tokens": 537596789.0, + "step": 21532 + }, + { + "epoch": 1.7756246392347654, + "grad_norm": 1.9179326295852661, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7271692752838135, + "num_tokens": 537629492.0, + "step": 21533 + }, + { + "epoch": 1.775707099859817, + "grad_norm": 1.9559707641601562, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.7934249639511108, + "num_tokens": 537660021.0, + "step": 21534 + }, + { + "epoch": 1.7757895604848684, + "grad_norm": 2.1479122638702393, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7382873296737671, + "num_tokens": 537686565.0, + "step": 21535 + }, + { + "epoch": 1.7758720211099202, + "grad_norm": 2.0737016201019287, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7660222053527832, + "num_tokens": 537713868.0, + "step": 21536 + }, + { + "epoch": 1.7759544817349715, + "grad_norm": 2.100909471511841, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7520373463630676, + "num_tokens": 537739811.0, + "step": 21537 + }, + { + "epoch": 1.776036942360023, + "grad_norm": 2.0352728366851807, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7620664834976196, + "num_tokens": 537771075.0, + "step": 21538 + }, + { + "epoch": 1.7761194029850746, + "grad_norm": 2.159667730331421, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7794713377952576, + "num_tokens": 537795462.0, + "step": 21539 + }, + { + "epoch": 1.7762018636101262, + "grad_norm": 2.343742609024048, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7776377201080322, + "num_tokens": 537815376.0, + "step": 21540 + }, + { + "epoch": 1.7762843242351778, + "grad_norm": 2.315531015396118, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7457906007766724, + "num_tokens": 537839053.0, + "step": 21541 + }, + { + "epoch": 1.7763667848602291, + "grad_norm": 1.7838656902313232, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7421160340309143, + "num_tokens": 537875069.0, + "step": 21542 + }, + { + "epoch": 1.776449245485281, + "grad_norm": 2.040170431137085, + "learning_rate": 1e-06, + "loss": 0.6001, + "mean_token_accuracy": 0.7972055673599243, + "num_tokens": 537905234.0, + "step": 21543 + }, + { + "epoch": 1.7765317061103323, + "grad_norm": 1.896462321281433, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7710321545600891, + "num_tokens": 537936414.0, + "step": 21544 + }, + { + "epoch": 1.7766141667353839, + "grad_norm": 1.8267433643341064, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7585773468017578, + "num_tokens": 537971003.0, + "step": 21545 + }, + { + "epoch": 1.7766966273604354, + "grad_norm": 1.7424697875976562, + "learning_rate": 1e-06, + "loss": 0.6115, + "mean_token_accuracy": 0.7977591753005981, + "num_tokens": 538002773.0, + "step": 21546 + }, + { + "epoch": 1.7767790879854868, + "grad_norm": 2.305493116378784, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7680083513259888, + "num_tokens": 538024771.0, + "step": 21547 + }, + { + "epoch": 1.7768615486105386, + "grad_norm": 1.9912748336791992, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.7811485528945923, + "num_tokens": 538051809.0, + "step": 21548 + }, + { + "epoch": 1.77694400923559, + "grad_norm": 2.263277769088745, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7058840990066528, + "num_tokens": 538077924.0, + "step": 21549 + }, + { + "epoch": 1.7770264698606415, + "grad_norm": 1.8572824001312256, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7919015884399414, + "num_tokens": 538108602.0, + "step": 21550 + }, + { + "epoch": 1.777108930485693, + "grad_norm": 1.94331955909729, + "learning_rate": 1e-06, + "loss": 0.6132, + "mean_token_accuracy": 0.7840445041656494, + "num_tokens": 538136754.0, + "step": 21551 + }, + { + "epoch": 1.7771913911107446, + "grad_norm": 2.2158308029174805, + "learning_rate": 1e-06, + "loss": 0.6087, + "mean_token_accuracy": 0.7890805006027222, + "num_tokens": 538160035.0, + "step": 21552 + }, + { + "epoch": 1.7772738517357962, + "grad_norm": 2.029414176940918, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7842050194740295, + "num_tokens": 538187778.0, + "step": 21553 + }, + { + "epoch": 1.7773563123608476, + "grad_norm": 1.8533638715744019, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7663016319274902, + "num_tokens": 538224675.0, + "step": 21554 + }, + { + "epoch": 1.7774387729858994, + "grad_norm": 1.9985774755477905, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7812067270278931, + "num_tokens": 538252850.0, + "step": 21555 + }, + { + "epoch": 1.7775212336109507, + "grad_norm": 1.7233099937438965, + "learning_rate": 1e-06, + "loss": 0.5795, + "mean_token_accuracy": 0.7992231845855713, + "num_tokens": 538288259.0, + "step": 21556 + }, + { + "epoch": 1.7776036942360023, + "grad_norm": 2.1591098308563232, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7535739541053772, + "num_tokens": 538313363.0, + "step": 21557 + }, + { + "epoch": 1.7776861548610539, + "grad_norm": 1.8761317729949951, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.7818546295166016, + "num_tokens": 538346207.0, + "step": 21558 + }, + { + "epoch": 1.7777686154861054, + "grad_norm": 2.1028950214385986, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7490724921226501, + "num_tokens": 538372844.0, + "step": 21559 + }, + { + "epoch": 1.777851076111157, + "grad_norm": 1.9815233945846558, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7456731200218201, + "num_tokens": 538403674.0, + "step": 21560 + }, + { + "epoch": 1.7779335367362084, + "grad_norm": 2.321502447128296, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7609691023826599, + "num_tokens": 538427289.0, + "step": 21561 + }, + { + "epoch": 1.7780159973612601, + "grad_norm": 2.232630491256714, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7601006031036377, + "num_tokens": 538452098.0, + "step": 21562 + }, + { + "epoch": 1.7780984579863115, + "grad_norm": 2.0643725395202637, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7679450511932373, + "num_tokens": 538478689.0, + "step": 21563 + }, + { + "epoch": 1.778180918611363, + "grad_norm": 2.0057284832000732, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.7560566663742065, + "num_tokens": 538509076.0, + "step": 21564 + }, + { + "epoch": 1.7782633792364146, + "grad_norm": 2.019749879837036, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7307743430137634, + "num_tokens": 538538882.0, + "step": 21565 + }, + { + "epoch": 1.778345839861466, + "grad_norm": 1.8617368936538696, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.7813354134559631, + "num_tokens": 538571412.0, + "step": 21566 + }, + { + "epoch": 1.7784283004865178, + "grad_norm": 2.329721689224243, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7227599620819092, + "num_tokens": 538596750.0, + "step": 21567 + }, + { + "epoch": 1.7785107611115691, + "grad_norm": 1.8827412128448486, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7673870921134949, + "num_tokens": 538626647.0, + "step": 21568 + }, + { + "epoch": 1.7785932217366207, + "grad_norm": 1.8957009315490723, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7464786767959595, + "num_tokens": 538657597.0, + "step": 21569 + }, + { + "epoch": 1.7786756823616723, + "grad_norm": 1.9191540479660034, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7710319757461548, + "num_tokens": 538688055.0, + "step": 21570 + }, + { + "epoch": 1.7787581429867239, + "grad_norm": 2.1970152854919434, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7745648622512817, + "num_tokens": 538711910.0, + "step": 21571 + }, + { + "epoch": 1.7788406036117754, + "grad_norm": 1.964407205581665, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7547969222068787, + "num_tokens": 538742080.0, + "step": 21572 + }, + { + "epoch": 1.7789230642368268, + "grad_norm": 1.8493539094924927, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7566331624984741, + "num_tokens": 538776198.0, + "step": 21573 + }, + { + "epoch": 1.7790055248618786, + "grad_norm": 2.2267963886260986, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.786819577217102, + "num_tokens": 538799054.0, + "step": 21574 + }, + { + "epoch": 1.77908798548693, + "grad_norm": 2.032304525375366, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7634742259979248, + "num_tokens": 538826701.0, + "step": 21575 + }, + { + "epoch": 1.7791704461119815, + "grad_norm": 1.8229490518569946, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.756745457649231, + "num_tokens": 538862127.0, + "step": 21576 + }, + { + "epoch": 1.779252906737033, + "grad_norm": 1.9793040752410889, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7486972212791443, + "num_tokens": 538890654.0, + "step": 21577 + }, + { + "epoch": 1.7793353673620846, + "grad_norm": 1.9921423196792603, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7621946930885315, + "num_tokens": 538920501.0, + "step": 21578 + }, + { + "epoch": 1.7794178279871362, + "grad_norm": 2.0453264713287354, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.7859054207801819, + "num_tokens": 538949146.0, + "step": 21579 + }, + { + "epoch": 1.7795002886121876, + "grad_norm": 2.036882162094116, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7433892488479614, + "num_tokens": 538979083.0, + "step": 21580 + }, + { + "epoch": 1.7795827492372394, + "grad_norm": 1.6901124715805054, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7927140593528748, + "num_tokens": 539014854.0, + "step": 21581 + }, + { + "epoch": 1.7796652098622907, + "grad_norm": 1.8311741352081299, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7532979249954224, + "num_tokens": 539046706.0, + "step": 21582 + }, + { + "epoch": 1.7797476704873423, + "grad_norm": 1.903114914894104, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7695761919021606, + "num_tokens": 539079601.0, + "step": 21583 + }, + { + "epoch": 1.7798301311123939, + "grad_norm": 2.1984219551086426, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7416877746582031, + "num_tokens": 539105503.0, + "step": 21584 + }, + { + "epoch": 1.7799125917374452, + "grad_norm": 1.9063595533370972, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7488960027694702, + "num_tokens": 539136833.0, + "step": 21585 + }, + { + "epoch": 1.779995052362497, + "grad_norm": 2.0672659873962402, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7687669396400452, + "num_tokens": 539163444.0, + "step": 21586 + }, + { + "epoch": 1.7800775129875483, + "grad_norm": 1.9654616117477417, + "learning_rate": 1e-06, + "loss": 0.5232, + "mean_token_accuracy": 0.8073773384094238, + "num_tokens": 539188159.0, + "step": 21587 + }, + { + "epoch": 1.7801599736126001, + "grad_norm": 2.0480916500091553, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7209402918815613, + "num_tokens": 539216504.0, + "step": 21588 + }, + { + "epoch": 1.7802424342376515, + "grad_norm": 2.0872018337249756, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7715688347816467, + "num_tokens": 539244705.0, + "step": 21589 + }, + { + "epoch": 1.780324894862703, + "grad_norm": 2.0681521892547607, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7597560882568359, + "num_tokens": 539273038.0, + "step": 21590 + }, + { + "epoch": 1.7804073554877546, + "grad_norm": 1.822567105293274, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7471306324005127, + "num_tokens": 539312511.0, + "step": 21591 + }, + { + "epoch": 1.780489816112806, + "grad_norm": 1.8241480588912964, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7692145705223083, + "num_tokens": 539347671.0, + "step": 21592 + }, + { + "epoch": 1.7805722767378578, + "grad_norm": 2.112520933151245, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7284212112426758, + "num_tokens": 539376253.0, + "step": 21593 + }, + { + "epoch": 1.7806547373629091, + "grad_norm": 1.9467430114746094, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.768095850944519, + "num_tokens": 539408394.0, + "step": 21594 + }, + { + "epoch": 1.7807371979879607, + "grad_norm": 2.32664155960083, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7619942426681519, + "num_tokens": 539431904.0, + "step": 21595 + }, + { + "epoch": 1.7808196586130123, + "grad_norm": 2.061286211013794, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7165462970733643, + "num_tokens": 539459362.0, + "step": 21596 + }, + { + "epoch": 1.7809021192380639, + "grad_norm": 1.9178543090820312, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.7909877300262451, + "num_tokens": 539489230.0, + "step": 21597 + }, + { + "epoch": 1.7809845798631154, + "grad_norm": 2.1352241039276123, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7378140687942505, + "num_tokens": 539518904.0, + "step": 21598 + }, + { + "epoch": 1.7810670404881668, + "grad_norm": 1.9587205648422241, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7623530626296997, + "num_tokens": 539551710.0, + "step": 21599 + }, + { + "epoch": 1.7811495011132186, + "grad_norm": 1.928297519683838, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7755992412567139, + "num_tokens": 539582957.0, + "step": 21600 + }, + { + "epoch": 1.78123196173827, + "grad_norm": 1.7309027910232544, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7746166586875916, + "num_tokens": 539619640.0, + "step": 21601 + }, + { + "epoch": 1.7813144223633215, + "grad_norm": 2.025559186935425, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7315012812614441, + "num_tokens": 539648422.0, + "step": 21602 + }, + { + "epoch": 1.781396882988373, + "grad_norm": 1.9511843919754028, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7635396718978882, + "num_tokens": 539681358.0, + "step": 21603 + }, + { + "epoch": 1.7814793436134246, + "grad_norm": 2.045165538787842, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7483769655227661, + "num_tokens": 539709315.0, + "step": 21604 + }, + { + "epoch": 1.7815618042384762, + "grad_norm": 2.209921360015869, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7418162822723389, + "num_tokens": 539733343.0, + "step": 21605 + }, + { + "epoch": 1.7816442648635276, + "grad_norm": 1.8577507734298706, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7518938183784485, + "num_tokens": 539764286.0, + "step": 21606 + }, + { + "epoch": 1.7817267254885794, + "grad_norm": 1.797163724899292, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.7989535331726074, + "num_tokens": 539795237.0, + "step": 21607 + }, + { + "epoch": 1.7818091861136307, + "grad_norm": 1.7451941967010498, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7744922637939453, + "num_tokens": 539831148.0, + "step": 21608 + }, + { + "epoch": 1.7818916467386823, + "grad_norm": 1.8465194702148438, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.7866593599319458, + "num_tokens": 539864663.0, + "step": 21609 + }, + { + "epoch": 1.7819741073637339, + "grad_norm": 1.9414582252502441, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.7933903932571411, + "num_tokens": 539893343.0, + "step": 21610 + }, + { + "epoch": 1.7820565679887852, + "grad_norm": 1.9644300937652588, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.7900209426879883, + "num_tokens": 539922708.0, + "step": 21611 + }, + { + "epoch": 1.782139028613837, + "grad_norm": 1.9197471141815186, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7502596378326416, + "num_tokens": 539953455.0, + "step": 21612 + }, + { + "epoch": 1.7822214892388883, + "grad_norm": 1.9701426029205322, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7546045780181885, + "num_tokens": 539983813.0, + "step": 21613 + }, + { + "epoch": 1.78230394986394, + "grad_norm": 1.8323559761047363, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7670865058898926, + "num_tokens": 540016857.0, + "step": 21614 + }, + { + "epoch": 1.7823864104889915, + "grad_norm": 2.0589261054992676, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7535625696182251, + "num_tokens": 540045531.0, + "step": 21615 + }, + { + "epoch": 1.782468871114043, + "grad_norm": 2.0516631603240967, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.7771974205970764, + "num_tokens": 540070158.0, + "step": 21616 + }, + { + "epoch": 1.7825513317390946, + "grad_norm": 1.7541918754577637, + "learning_rate": 1e-06, + "loss": 0.7452, + "mean_token_accuracy": 0.759682297706604, + "num_tokens": 540108360.0, + "step": 21617 + }, + { + "epoch": 1.782633792364146, + "grad_norm": 1.9668021202087402, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7818127870559692, + "num_tokens": 540139635.0, + "step": 21618 + }, + { + "epoch": 1.7827162529891978, + "grad_norm": 2.2225728034973145, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.7121196389198303, + "num_tokens": 540166810.0, + "step": 21619 + }, + { + "epoch": 1.7827987136142491, + "grad_norm": 2.2615432739257812, + "learning_rate": 1e-06, + "loss": 0.6129, + "mean_token_accuracy": 0.7916883230209351, + "num_tokens": 540189402.0, + "step": 21620 + }, + { + "epoch": 1.7828811742393007, + "grad_norm": 1.8741552829742432, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7776443958282471, + "num_tokens": 540221284.0, + "step": 21621 + }, + { + "epoch": 1.7829636348643523, + "grad_norm": 1.894506573677063, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.7885004281997681, + "num_tokens": 540250066.0, + "step": 21622 + }, + { + "epoch": 1.7830460954894038, + "grad_norm": 1.94381844997406, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.7984112501144409, + "num_tokens": 540282086.0, + "step": 21623 + }, + { + "epoch": 1.7831285561144554, + "grad_norm": 1.8539735078811646, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.7825310230255127, + "num_tokens": 540314372.0, + "step": 21624 + }, + { + "epoch": 1.7832110167395068, + "grad_norm": 2.208021640777588, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7405623197555542, + "num_tokens": 540340785.0, + "step": 21625 + }, + { + "epoch": 1.7832934773645586, + "grad_norm": 2.046501874923706, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7789191007614136, + "num_tokens": 540368688.0, + "step": 21626 + }, + { + "epoch": 1.78337593798961, + "grad_norm": 2.083101749420166, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7762150764465332, + "num_tokens": 540396361.0, + "step": 21627 + }, + { + "epoch": 1.7834583986146615, + "grad_norm": 1.9551206827163696, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.7977885603904724, + "num_tokens": 540425746.0, + "step": 21628 + }, + { + "epoch": 1.783540859239713, + "grad_norm": 1.8674348592758179, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.774379551410675, + "num_tokens": 540459263.0, + "step": 21629 + }, + { + "epoch": 1.7836233198647644, + "grad_norm": 1.9320603609085083, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7664122581481934, + "num_tokens": 540490200.0, + "step": 21630 + }, + { + "epoch": 1.7837057804898162, + "grad_norm": 1.820741057395935, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7800165414810181, + "num_tokens": 540523659.0, + "step": 21631 + }, + { + "epoch": 1.7837882411148676, + "grad_norm": 2.104357957839966, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7256242036819458, + "num_tokens": 540552101.0, + "step": 21632 + }, + { + "epoch": 1.7838707017399194, + "grad_norm": 2.0067741870880127, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.7661009430885315, + "num_tokens": 540586186.0, + "step": 21633 + }, + { + "epoch": 1.7839531623649707, + "grad_norm": 2.0015664100646973, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7781320810317993, + "num_tokens": 540614920.0, + "step": 21634 + }, + { + "epoch": 1.7840356229900223, + "grad_norm": 1.9770739078521729, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7889922857284546, + "num_tokens": 540645750.0, + "step": 21635 + }, + { + "epoch": 1.7841180836150738, + "grad_norm": 1.8746378421783447, + "learning_rate": 1e-06, + "loss": 0.5847, + "mean_token_accuracy": 0.7971506118774414, + "num_tokens": 540676345.0, + "step": 21636 + }, + { + "epoch": 1.7842005442401252, + "grad_norm": 1.8774389028549194, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.768812894821167, + "num_tokens": 540708968.0, + "step": 21637 + }, + { + "epoch": 1.784283004865177, + "grad_norm": 1.988509178161621, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7545403242111206, + "num_tokens": 540738149.0, + "step": 21638 + }, + { + "epoch": 1.7843654654902283, + "grad_norm": 1.959581971168518, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7758917212486267, + "num_tokens": 540768419.0, + "step": 21639 + }, + { + "epoch": 1.78444792611528, + "grad_norm": 1.9849900007247925, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7783900499343872, + "num_tokens": 540798838.0, + "step": 21640 + }, + { + "epoch": 1.7845303867403315, + "grad_norm": 1.9569544792175293, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7640471458435059, + "num_tokens": 540828628.0, + "step": 21641 + }, + { + "epoch": 1.784612847365383, + "grad_norm": 1.98201322555542, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7724079489707947, + "num_tokens": 540857439.0, + "step": 21642 + }, + { + "epoch": 1.7846953079904346, + "grad_norm": 1.9094818830490112, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.7901933193206787, + "num_tokens": 540886732.0, + "step": 21643 + }, + { + "epoch": 1.784777768615486, + "grad_norm": 1.8712496757507324, + "learning_rate": 1e-06, + "loss": 0.5834, + "mean_token_accuracy": 0.8028842210769653, + "num_tokens": 540915977.0, + "step": 21644 + }, + { + "epoch": 1.7848602292405378, + "grad_norm": 2.248218536376953, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7437472939491272, + "num_tokens": 540940762.0, + "step": 21645 + }, + { + "epoch": 1.7849426898655891, + "grad_norm": 1.8032817840576172, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7558016777038574, + "num_tokens": 540977982.0, + "step": 21646 + }, + { + "epoch": 1.7850251504906407, + "grad_norm": 2.156949758529663, + "learning_rate": 1e-06, + "loss": 0.571, + "mean_token_accuracy": 0.8042920231819153, + "num_tokens": 541002734.0, + "step": 21647 + }, + { + "epoch": 1.7851076111156923, + "grad_norm": 2.020036458969116, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7798979878425598, + "num_tokens": 541032500.0, + "step": 21648 + }, + { + "epoch": 1.7851900717407438, + "grad_norm": 1.8115062713623047, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7825180292129517, + "num_tokens": 541067759.0, + "step": 21649 + }, + { + "epoch": 1.7852725323657954, + "grad_norm": 1.8433282375335693, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7693784236907959, + "num_tokens": 541103428.0, + "step": 21650 + }, + { + "epoch": 1.7853549929908468, + "grad_norm": 1.829433798789978, + "learning_rate": 1e-06, + "loss": 0.5691, + "mean_token_accuracy": 0.8082445859909058, + "num_tokens": 541133679.0, + "step": 21651 + }, + { + "epoch": 1.7854374536158986, + "grad_norm": 1.9142301082611084, + "learning_rate": 1e-06, + "loss": 0.6127, + "mean_token_accuracy": 0.79436856508255, + "num_tokens": 541163850.0, + "step": 21652 + }, + { + "epoch": 1.78551991424095, + "grad_norm": 2.0605030059814453, + "learning_rate": 1e-06, + "loss": 0.5592, + "mean_token_accuracy": 0.8081506490707397, + "num_tokens": 541191892.0, + "step": 21653 + }, + { + "epoch": 1.7856023748660015, + "grad_norm": 1.9576042890548706, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7650508284568787, + "num_tokens": 541220501.0, + "step": 21654 + }, + { + "epoch": 1.785684835491053, + "grad_norm": 1.917683720588684, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.7903650999069214, + "num_tokens": 541250235.0, + "step": 21655 + }, + { + "epoch": 1.7857672961161044, + "grad_norm": 2.2018184661865234, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7762368321418762, + "num_tokens": 541273415.0, + "step": 21656 + }, + { + "epoch": 1.7858497567411562, + "grad_norm": 2.606945276260376, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7180328965187073, + "num_tokens": 541294636.0, + "step": 21657 + }, + { + "epoch": 1.7859322173662076, + "grad_norm": 2.0670454502105713, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7887760996818542, + "num_tokens": 541320566.0, + "step": 21658 + }, + { + "epoch": 1.7860146779912591, + "grad_norm": 2.1810386180877686, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7724008560180664, + "num_tokens": 541345173.0, + "step": 21659 + }, + { + "epoch": 1.7860971386163107, + "grad_norm": 1.99552321434021, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7416226863861084, + "num_tokens": 541374878.0, + "step": 21660 + }, + { + "epoch": 1.7861795992413623, + "grad_norm": 2.246419906616211, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7372854351997375, + "num_tokens": 541400699.0, + "step": 21661 + }, + { + "epoch": 1.7862620598664138, + "grad_norm": 1.869468092918396, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7344903349876404, + "num_tokens": 541433190.0, + "step": 21662 + }, + { + "epoch": 1.7863445204914652, + "grad_norm": 2.0341129302978516, + "learning_rate": 1e-06, + "loss": 0.5585, + "mean_token_accuracy": 0.8138139843940735, + "num_tokens": 541459822.0, + "step": 21663 + }, + { + "epoch": 1.786426981116517, + "grad_norm": 2.0070972442626953, + "learning_rate": 1e-06, + "loss": 0.6086, + "mean_token_accuracy": 0.7983372211456299, + "num_tokens": 541488271.0, + "step": 21664 + }, + { + "epoch": 1.7865094417415683, + "grad_norm": 1.8681730031967163, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7479926347732544, + "num_tokens": 541522997.0, + "step": 21665 + }, + { + "epoch": 1.78659190236662, + "grad_norm": 1.9978529214859009, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7593878507614136, + "num_tokens": 541552540.0, + "step": 21666 + }, + { + "epoch": 1.7866743629916715, + "grad_norm": 2.01081919670105, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7870165109634399, + "num_tokens": 541579775.0, + "step": 21667 + }, + { + "epoch": 1.786756823616723, + "grad_norm": 1.9309580326080322, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7598878145217896, + "num_tokens": 541613094.0, + "step": 21668 + }, + { + "epoch": 1.7868392842417746, + "grad_norm": 1.9787020683288574, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7671819925308228, + "num_tokens": 541641680.0, + "step": 21669 + }, + { + "epoch": 1.786921744866826, + "grad_norm": 1.8357963562011719, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7794373631477356, + "num_tokens": 541676851.0, + "step": 21670 + }, + { + "epoch": 1.7870042054918778, + "grad_norm": 1.8615435361862183, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7755432724952698, + "num_tokens": 541714680.0, + "step": 21671 + }, + { + "epoch": 1.7870866661169291, + "grad_norm": 1.951350212097168, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7683800458908081, + "num_tokens": 541748102.0, + "step": 21672 + }, + { + "epoch": 1.7871691267419807, + "grad_norm": 2.181318759918213, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6913566589355469, + "num_tokens": 541774880.0, + "step": 21673 + }, + { + "epoch": 1.7872515873670323, + "grad_norm": 1.93149995803833, + "learning_rate": 1e-06, + "loss": 0.7452, + "mean_token_accuracy": 0.7739688754081726, + "num_tokens": 541804454.0, + "step": 21674 + }, + { + "epoch": 1.7873340479920836, + "grad_norm": 1.9377843141555786, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7301805019378662, + "num_tokens": 541837461.0, + "step": 21675 + }, + { + "epoch": 1.7874165086171354, + "grad_norm": 2.0211050510406494, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.759670615196228, + "num_tokens": 541866708.0, + "step": 21676 + }, + { + "epoch": 1.7874989692421868, + "grad_norm": 1.9254649877548218, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7480489015579224, + "num_tokens": 541899735.0, + "step": 21677 + }, + { + "epoch": 1.7875814298672386, + "grad_norm": 2.019352912902832, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.739449143409729, + "num_tokens": 541928880.0, + "step": 21678 + }, + { + "epoch": 1.78766389049229, + "grad_norm": 1.83574640750885, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.769297182559967, + "num_tokens": 541961483.0, + "step": 21679 + }, + { + "epoch": 1.7877463511173415, + "grad_norm": 2.373530149459839, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7684017419815063, + "num_tokens": 541986613.0, + "step": 21680 + }, + { + "epoch": 1.787828811742393, + "grad_norm": 2.2104194164276123, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7441897392272949, + "num_tokens": 542012322.0, + "step": 21681 + }, + { + "epoch": 1.7879112723674444, + "grad_norm": 2.286884307861328, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7485719919204712, + "num_tokens": 542036342.0, + "step": 21682 + }, + { + "epoch": 1.7879937329924962, + "grad_norm": 1.6891263723373413, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7622075080871582, + "num_tokens": 542075530.0, + "step": 21683 + }, + { + "epoch": 1.7880761936175475, + "grad_norm": 1.8494014739990234, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7610109448432922, + "num_tokens": 542108667.0, + "step": 21684 + }, + { + "epoch": 1.7881586542425991, + "grad_norm": 1.9003007411956787, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.7566441893577576, + "num_tokens": 542141090.0, + "step": 21685 + }, + { + "epoch": 1.7882411148676507, + "grad_norm": 1.938607096672058, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.795143723487854, + "num_tokens": 542170835.0, + "step": 21686 + }, + { + "epoch": 1.7883235754927023, + "grad_norm": 1.8990944623947144, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7747966051101685, + "num_tokens": 542200653.0, + "step": 21687 + }, + { + "epoch": 1.7884060361177538, + "grad_norm": 2.047420024871826, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7190411686897278, + "num_tokens": 542231534.0, + "step": 21688 + }, + { + "epoch": 1.7884884967428052, + "grad_norm": 1.7883957624435425, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7634541392326355, + "num_tokens": 542266019.0, + "step": 21689 + }, + { + "epoch": 1.788570957367857, + "grad_norm": 2.001544237136841, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7406851053237915, + "num_tokens": 542297914.0, + "step": 21690 + }, + { + "epoch": 1.7886534179929083, + "grad_norm": 2.0764949321746826, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7668513059616089, + "num_tokens": 542325784.0, + "step": 21691 + }, + { + "epoch": 1.78873587861796, + "grad_norm": 1.9714879989624023, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7707590460777283, + "num_tokens": 542358253.0, + "step": 21692 + }, + { + "epoch": 1.7888183392430115, + "grad_norm": 2.4614152908325195, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7580119371414185, + "num_tokens": 542381950.0, + "step": 21693 + }, + { + "epoch": 1.788900799868063, + "grad_norm": 1.9224215745925903, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7356997728347778, + "num_tokens": 542416131.0, + "step": 21694 + }, + { + "epoch": 1.7889832604931146, + "grad_norm": 1.967646598815918, + "learning_rate": 1e-06, + "loss": 0.7722, + "mean_token_accuracy": 0.7516182065010071, + "num_tokens": 542447965.0, + "step": 21695 + }, + { + "epoch": 1.789065721118166, + "grad_norm": 2.02630352973938, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7556563019752502, + "num_tokens": 542476034.0, + "step": 21696 + }, + { + "epoch": 1.7891481817432178, + "grad_norm": 1.9354321956634521, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7429873943328857, + "num_tokens": 542507499.0, + "step": 21697 + }, + { + "epoch": 1.7892306423682691, + "grad_norm": 2.0328216552734375, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7766447067260742, + "num_tokens": 542536120.0, + "step": 21698 + }, + { + "epoch": 1.7893131029933207, + "grad_norm": 2.0167458057403564, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7676997184753418, + "num_tokens": 542564942.0, + "step": 21699 + }, + { + "epoch": 1.7893955636183723, + "grad_norm": 2.0717012882232666, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7702256441116333, + "num_tokens": 542592487.0, + "step": 21700 + }, + { + "epoch": 1.7894780242434236, + "grad_norm": 2.031710386276245, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7701076865196228, + "num_tokens": 542619144.0, + "step": 21701 + }, + { + "epoch": 1.7895604848684754, + "grad_norm": 2.1177027225494385, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7363441586494446, + "num_tokens": 542646780.0, + "step": 21702 + }, + { + "epoch": 1.7896429454935268, + "grad_norm": 1.918784499168396, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.7547022104263306, + "num_tokens": 542679780.0, + "step": 21703 + }, + { + "epoch": 1.7897254061185783, + "grad_norm": 2.159061908721924, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7605341672897339, + "num_tokens": 542706734.0, + "step": 21704 + }, + { + "epoch": 1.78980786674363, + "grad_norm": 2.0321288108825684, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.7848660945892334, + "num_tokens": 542734405.0, + "step": 21705 + }, + { + "epoch": 1.7898903273686815, + "grad_norm": 1.8934417963027954, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7698860168457031, + "num_tokens": 542764506.0, + "step": 21706 + }, + { + "epoch": 1.789972787993733, + "grad_norm": 2.1356091499328613, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7008198499679565, + "num_tokens": 542791908.0, + "step": 21707 + }, + { + "epoch": 1.7900552486187844, + "grad_norm": 1.9768646955490112, + "learning_rate": 1e-06, + "loss": 0.7886, + "mean_token_accuracy": 0.7473568916320801, + "num_tokens": 542821907.0, + "step": 21708 + }, + { + "epoch": 1.7901377092438362, + "grad_norm": 1.8242018222808838, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7740786671638489, + "num_tokens": 542853479.0, + "step": 21709 + }, + { + "epoch": 1.7902201698688875, + "grad_norm": 2.1886918544769287, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7729873657226562, + "num_tokens": 542880068.0, + "step": 21710 + }, + { + "epoch": 1.7903026304939391, + "grad_norm": 2.0922203063964844, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7729532718658447, + "num_tokens": 542908465.0, + "step": 21711 + }, + { + "epoch": 1.7903850911189907, + "grad_norm": 1.996827483177185, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7776893377304077, + "num_tokens": 542935168.0, + "step": 21712 + }, + { + "epoch": 1.7904675517440423, + "grad_norm": 2.3719871044158936, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.755050539970398, + "num_tokens": 542958282.0, + "step": 21713 + }, + { + "epoch": 1.7905500123690938, + "grad_norm": 2.0142531394958496, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.750579833984375, + "num_tokens": 542986254.0, + "step": 21714 + }, + { + "epoch": 1.7906324729941452, + "grad_norm": 1.7503918409347534, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.798589289188385, + "num_tokens": 543019083.0, + "step": 21715 + }, + { + "epoch": 1.790714933619197, + "grad_norm": 1.6291682720184326, + "learning_rate": 1e-06, + "loss": 0.5769, + "mean_token_accuracy": 0.817406415939331, + "num_tokens": 543054627.0, + "step": 21716 + }, + { + "epoch": 1.7907973942442483, + "grad_norm": 2.53737735748291, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7752487659454346, + "num_tokens": 543074637.0, + "step": 21717 + }, + { + "epoch": 1.7908798548693, + "grad_norm": 1.783048391342163, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7638879418373108, + "num_tokens": 543112303.0, + "step": 21718 + }, + { + "epoch": 1.7909623154943515, + "grad_norm": 2.001093626022339, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7872000932693481, + "num_tokens": 543143097.0, + "step": 21719 + }, + { + "epoch": 1.7910447761194028, + "grad_norm": 1.9079506397247314, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.796414852142334, + "num_tokens": 543173609.0, + "step": 21720 + }, + { + "epoch": 1.7911272367444546, + "grad_norm": 2.03543758392334, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7291507720947266, + "num_tokens": 543204060.0, + "step": 21721 + }, + { + "epoch": 1.791209697369506, + "grad_norm": 2.3258228302001953, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7486067414283752, + "num_tokens": 543226995.0, + "step": 21722 + }, + { + "epoch": 1.7912921579945575, + "grad_norm": 1.8393211364746094, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.738629937171936, + "num_tokens": 543261192.0, + "step": 21723 + }, + { + "epoch": 1.7913746186196091, + "grad_norm": 2.0331814289093018, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.736436128616333, + "num_tokens": 543291672.0, + "step": 21724 + }, + { + "epoch": 1.7914570792446607, + "grad_norm": 1.964076280593872, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7618430256843567, + "num_tokens": 543324165.0, + "step": 21725 + }, + { + "epoch": 1.7915395398697123, + "grad_norm": 2.092590570449829, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7731226086616516, + "num_tokens": 543352482.0, + "step": 21726 + }, + { + "epoch": 1.7916220004947636, + "grad_norm": 2.0773582458496094, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7223842144012451, + "num_tokens": 543382051.0, + "step": 21727 + }, + { + "epoch": 1.7917044611198154, + "grad_norm": 1.8741849660873413, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7571172118186951, + "num_tokens": 543413196.0, + "step": 21728 + }, + { + "epoch": 1.7917869217448668, + "grad_norm": 1.7803772687911987, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7685422897338867, + "num_tokens": 543454813.0, + "step": 21729 + }, + { + "epoch": 1.7918693823699183, + "grad_norm": 1.9885603189468384, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7453044652938843, + "num_tokens": 543487006.0, + "step": 21730 + }, + { + "epoch": 1.79195184299497, + "grad_norm": 2.051039934158325, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.7976018190383911, + "num_tokens": 543515764.0, + "step": 21731 + }, + { + "epoch": 1.7920343036200215, + "grad_norm": 2.201009750366211, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.769466757774353, + "num_tokens": 543542809.0, + "step": 21732 + }, + { + "epoch": 1.792116764245073, + "grad_norm": 2.3085179328918457, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7490787506103516, + "num_tokens": 543565087.0, + "step": 21733 + }, + { + "epoch": 1.7921992248701244, + "grad_norm": 2.000366687774658, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7484373450279236, + "num_tokens": 543597613.0, + "step": 21734 + }, + { + "epoch": 1.7922816854951762, + "grad_norm": 1.9188872575759888, + "learning_rate": 1e-06, + "loss": 0.4928, + "mean_token_accuracy": 0.833163321018219, + "num_tokens": 543625223.0, + "step": 21735 + }, + { + "epoch": 1.7923641461202275, + "grad_norm": 1.8763455152511597, + "learning_rate": 1e-06, + "loss": 0.5079, + "mean_token_accuracy": 0.8248386383056641, + "num_tokens": 543654542.0, + "step": 21736 + }, + { + "epoch": 1.7924466067452791, + "grad_norm": 1.89073646068573, + "learning_rate": 1e-06, + "loss": 0.6062, + "mean_token_accuracy": 0.792803943157196, + "num_tokens": 543684379.0, + "step": 21737 + }, + { + "epoch": 1.7925290673703307, + "grad_norm": 1.8581857681274414, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7794033288955688, + "num_tokens": 543714837.0, + "step": 21738 + }, + { + "epoch": 1.7926115279953823, + "grad_norm": 1.83732271194458, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.7908291816711426, + "num_tokens": 543744088.0, + "step": 21739 + }, + { + "epoch": 1.7926939886204338, + "grad_norm": 1.7363996505737305, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7663658261299133, + "num_tokens": 543781046.0, + "step": 21740 + }, + { + "epoch": 1.7927764492454852, + "grad_norm": 2.089999198913574, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.7702890634536743, + "num_tokens": 543809755.0, + "step": 21741 + }, + { + "epoch": 1.792858909870537, + "grad_norm": 1.9307008981704712, + "learning_rate": 1e-06, + "loss": 0.7452, + "mean_token_accuracy": 0.7644237875938416, + "num_tokens": 543840646.0, + "step": 21742 + }, + { + "epoch": 1.7929413704955883, + "grad_norm": 1.902390718460083, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7855872511863708, + "num_tokens": 543872975.0, + "step": 21743 + }, + { + "epoch": 1.79302383112064, + "grad_norm": 2.184373378753662, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.744147777557373, + "num_tokens": 543897203.0, + "step": 21744 + }, + { + "epoch": 1.7931062917456915, + "grad_norm": 2.04858660697937, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7114746570587158, + "num_tokens": 543927177.0, + "step": 21745 + }, + { + "epoch": 1.7931887523707428, + "grad_norm": 2.0870559215545654, + "learning_rate": 1e-06, + "loss": 0.553, + "mean_token_accuracy": 0.8118685483932495, + "num_tokens": 543951565.0, + "step": 21746 + }, + { + "epoch": 1.7932712129957946, + "grad_norm": 2.1707117557525635, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7596408128738403, + "num_tokens": 543977064.0, + "step": 21747 + }, + { + "epoch": 1.793353673620846, + "grad_norm": 1.8780890703201294, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7451931238174438, + "num_tokens": 544010952.0, + "step": 21748 + }, + { + "epoch": 1.7934361342458975, + "grad_norm": 1.9237260818481445, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.7924673557281494, + "num_tokens": 544040796.0, + "step": 21749 + }, + { + "epoch": 1.7935185948709491, + "grad_norm": 2.364531993865967, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7306602001190186, + "num_tokens": 544062323.0, + "step": 21750 + }, + { + "epoch": 1.7936010554960007, + "grad_norm": 1.9660834074020386, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7492454648017883, + "num_tokens": 544091211.0, + "step": 21751 + }, + { + "epoch": 1.7936835161210523, + "grad_norm": 2.030029535293579, + "learning_rate": 1e-06, + "loss": 0.59, + "mean_token_accuracy": 0.8039268255233765, + "num_tokens": 544118708.0, + "step": 21752 + }, + { + "epoch": 1.7937659767461036, + "grad_norm": 1.716937780380249, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.75791996717453, + "num_tokens": 544157067.0, + "step": 21753 + }, + { + "epoch": 1.7938484373711554, + "grad_norm": 2.1507577896118164, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7665194272994995, + "num_tokens": 544182775.0, + "step": 21754 + }, + { + "epoch": 1.7939308979962068, + "grad_norm": 1.8818124532699585, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7722695469856262, + "num_tokens": 544214293.0, + "step": 21755 + }, + { + "epoch": 1.7940133586212583, + "grad_norm": 1.8560009002685547, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7358387112617493, + "num_tokens": 544251928.0, + "step": 21756 + }, + { + "epoch": 1.79409581924631, + "grad_norm": 2.033562421798706, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.6951472759246826, + "num_tokens": 544284026.0, + "step": 21757 + }, + { + "epoch": 1.7941782798713615, + "grad_norm": 1.816189169883728, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7840004563331604, + "num_tokens": 544318552.0, + "step": 21758 + }, + { + "epoch": 1.794260740496413, + "grad_norm": 2.374051570892334, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.714480996131897, + "num_tokens": 544346559.0, + "step": 21759 + }, + { + "epoch": 1.7943432011214644, + "grad_norm": 1.608119010925293, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7655107975006104, + "num_tokens": 544390959.0, + "step": 21760 + }, + { + "epoch": 1.7944256617465162, + "grad_norm": 1.8177908658981323, + "learning_rate": 1e-06, + "loss": 0.5772, + "mean_token_accuracy": 0.8046272397041321, + "num_tokens": 544421696.0, + "step": 21761 + }, + { + "epoch": 1.7945081223715675, + "grad_norm": 2.5715231895446777, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7245795726776123, + "num_tokens": 544442497.0, + "step": 21762 + }, + { + "epoch": 1.794590582996619, + "grad_norm": 1.8341156244277954, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7608391046524048, + "num_tokens": 544477165.0, + "step": 21763 + }, + { + "epoch": 1.7946730436216707, + "grad_norm": 1.9581139087677002, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7670146822929382, + "num_tokens": 544506801.0, + "step": 21764 + }, + { + "epoch": 1.794755504246722, + "grad_norm": 2.247248411178589, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.75120609998703, + "num_tokens": 544529655.0, + "step": 21765 + }, + { + "epoch": 1.7948379648717738, + "grad_norm": 2.055945634841919, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7734267711639404, + "num_tokens": 544559236.0, + "step": 21766 + }, + { + "epoch": 1.7949204254968252, + "grad_norm": 2.080819845199585, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7759024500846863, + "num_tokens": 544585608.0, + "step": 21767 + }, + { + "epoch": 1.7950028861218768, + "grad_norm": 2.1642894744873047, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7744243144989014, + "num_tokens": 544612320.0, + "step": 21768 + }, + { + "epoch": 1.7950853467469283, + "grad_norm": 1.9689815044403076, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7644162178039551, + "num_tokens": 544642860.0, + "step": 21769 + }, + { + "epoch": 1.79516780737198, + "grad_norm": 1.8435307741165161, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7671648859977722, + "num_tokens": 544676792.0, + "step": 21770 + }, + { + "epoch": 1.7952502679970315, + "grad_norm": 1.9304883480072021, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7369515895843506, + "num_tokens": 544708209.0, + "step": 21771 + }, + { + "epoch": 1.7953327286220828, + "grad_norm": 2.0607852935791016, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7396732568740845, + "num_tokens": 544735307.0, + "step": 21772 + }, + { + "epoch": 1.7954151892471346, + "grad_norm": 2.022556781768799, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7373785972595215, + "num_tokens": 544766612.0, + "step": 21773 + }, + { + "epoch": 1.795497649872186, + "grad_norm": 2.2206931114196777, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7370706796646118, + "num_tokens": 544793554.0, + "step": 21774 + }, + { + "epoch": 1.7955801104972375, + "grad_norm": 2.018038511276245, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7795037627220154, + "num_tokens": 544823399.0, + "step": 21775 + }, + { + "epoch": 1.795662571122289, + "grad_norm": 2.200352668762207, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7509253621101379, + "num_tokens": 544848531.0, + "step": 21776 + }, + { + "epoch": 1.7957450317473407, + "grad_norm": 2.0836260318756104, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7532594799995422, + "num_tokens": 544875659.0, + "step": 21777 + }, + { + "epoch": 1.7958274923723923, + "grad_norm": 2.155791997909546, + "learning_rate": 1e-06, + "loss": 1.0282, + "mean_token_accuracy": 0.6979997754096985, + "num_tokens": 544904387.0, + "step": 21778 + }, + { + "epoch": 1.7959099529974436, + "grad_norm": 1.8903571367263794, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7535372972488403, + "num_tokens": 544938769.0, + "step": 21779 + }, + { + "epoch": 1.7959924136224954, + "grad_norm": 2.0574586391448975, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7503299713134766, + "num_tokens": 544966840.0, + "step": 21780 + }, + { + "epoch": 1.7960748742475467, + "grad_norm": 2.119178056716919, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7848540544509888, + "num_tokens": 544995997.0, + "step": 21781 + }, + { + "epoch": 1.7961573348725983, + "grad_norm": 1.9481873512268066, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.782001256942749, + "num_tokens": 545026130.0, + "step": 21782 + }, + { + "epoch": 1.79623979549765, + "grad_norm": 1.8456569910049438, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.7965894937515259, + "num_tokens": 545057102.0, + "step": 21783 + }, + { + "epoch": 1.7963222561227012, + "grad_norm": 1.9988924264907837, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.7952394485473633, + "num_tokens": 545086846.0, + "step": 21784 + }, + { + "epoch": 1.796404716747753, + "grad_norm": 1.7612738609313965, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.765512228012085, + "num_tokens": 545123638.0, + "step": 21785 + }, + { + "epoch": 1.7964871773728044, + "grad_norm": 1.959949254989624, + "learning_rate": 1e-06, + "loss": 0.54, + "mean_token_accuracy": 0.8176068067550659, + "num_tokens": 545153209.0, + "step": 21786 + }, + { + "epoch": 1.7965696379978562, + "grad_norm": 2.055757999420166, + "learning_rate": 1e-06, + "loss": 0.6103, + "mean_token_accuracy": 0.7856688499450684, + "num_tokens": 545182674.0, + "step": 21787 + }, + { + "epoch": 1.7966520986229075, + "grad_norm": 1.8628846406936646, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7728890180587769, + "num_tokens": 545216369.0, + "step": 21788 + }, + { + "epoch": 1.796734559247959, + "grad_norm": 1.9094120264053345, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7662234902381897, + "num_tokens": 545246383.0, + "step": 21789 + }, + { + "epoch": 1.7968170198730107, + "grad_norm": 1.897923231124878, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7566648721694946, + "num_tokens": 545277336.0, + "step": 21790 + }, + { + "epoch": 1.796899480498062, + "grad_norm": 2.0701613426208496, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7427146434783936, + "num_tokens": 545303808.0, + "step": 21791 + }, + { + "epoch": 1.7969819411231138, + "grad_norm": 2.0120253562927246, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7782604694366455, + "num_tokens": 545331429.0, + "step": 21792 + }, + { + "epoch": 1.7970644017481652, + "grad_norm": 2.1680996417999268, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.7920722961425781, + "num_tokens": 545357825.0, + "step": 21793 + }, + { + "epoch": 1.7971468623732167, + "grad_norm": 2.0600039958953857, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7696297764778137, + "num_tokens": 545388157.0, + "step": 21794 + }, + { + "epoch": 1.7972293229982683, + "grad_norm": 2.1459567546844482, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7594131231307983, + "num_tokens": 545414398.0, + "step": 21795 + }, + { + "epoch": 1.79731178362332, + "grad_norm": 1.9319918155670166, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7716516256332397, + "num_tokens": 545443851.0, + "step": 21796 + }, + { + "epoch": 1.7973942442483715, + "grad_norm": 1.8949873447418213, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7627310156822205, + "num_tokens": 545476760.0, + "step": 21797 + }, + { + "epoch": 1.7974767048734228, + "grad_norm": 2.1266250610351562, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7619411945343018, + "num_tokens": 545501686.0, + "step": 21798 + }, + { + "epoch": 1.7975591654984746, + "grad_norm": 2.0581307411193848, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.7939528226852417, + "num_tokens": 545526702.0, + "step": 21799 + }, + { + "epoch": 1.797641626123526, + "grad_norm": 1.7967373132705688, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7740809321403503, + "num_tokens": 545565310.0, + "step": 21800 + }, + { + "epoch": 1.7977240867485775, + "grad_norm": 1.8618135452270508, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.7895815372467041, + "num_tokens": 545597610.0, + "step": 21801 + }, + { + "epoch": 1.797806547373629, + "grad_norm": 2.126024007797241, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7630467414855957, + "num_tokens": 545622784.0, + "step": 21802 + }, + { + "epoch": 1.7978890079986807, + "grad_norm": 1.7694000005722046, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.766616702079773, + "num_tokens": 545658025.0, + "step": 21803 + }, + { + "epoch": 1.7979714686237322, + "grad_norm": 1.6877905130386353, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7639333009719849, + "num_tokens": 545698349.0, + "step": 21804 + }, + { + "epoch": 1.7980539292487836, + "grad_norm": 1.7957614660263062, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7598495483398438, + "num_tokens": 545735597.0, + "step": 21805 + }, + { + "epoch": 1.7981363898738354, + "grad_norm": 2.1897006034851074, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7390392422676086, + "num_tokens": 545762363.0, + "step": 21806 + }, + { + "epoch": 1.7982188504988867, + "grad_norm": 2.0543482303619385, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7624008655548096, + "num_tokens": 545791648.0, + "step": 21807 + }, + { + "epoch": 1.7983013111239383, + "grad_norm": 2.6062428951263428, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7311903238296509, + "num_tokens": 545812003.0, + "step": 21808 + }, + { + "epoch": 1.79838377174899, + "grad_norm": 2.0495595932006836, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7597966194152832, + "num_tokens": 545840615.0, + "step": 21809 + }, + { + "epoch": 1.7984662323740412, + "grad_norm": 2.2724251747131348, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7837798595428467, + "num_tokens": 545864092.0, + "step": 21810 + }, + { + "epoch": 1.798548692999093, + "grad_norm": 2.2905197143554688, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7403673529624939, + "num_tokens": 545889606.0, + "step": 21811 + }, + { + "epoch": 1.7986311536241444, + "grad_norm": 1.9140621423721313, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.769996166229248, + "num_tokens": 545923442.0, + "step": 21812 + }, + { + "epoch": 1.798713614249196, + "grad_norm": 1.8840962648391724, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7384686470031738, + "num_tokens": 545954418.0, + "step": 21813 + }, + { + "epoch": 1.7987960748742475, + "grad_norm": 1.952468752861023, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7783691883087158, + "num_tokens": 545985818.0, + "step": 21814 + }, + { + "epoch": 1.798878535499299, + "grad_norm": 2.0972702503204346, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7688276171684265, + "num_tokens": 546010402.0, + "step": 21815 + }, + { + "epoch": 1.7989609961243507, + "grad_norm": 1.7840497493743896, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.7888188362121582, + "num_tokens": 546041615.0, + "step": 21816 + }, + { + "epoch": 1.799043456749402, + "grad_norm": 1.821628212928772, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7859296798706055, + "num_tokens": 546074836.0, + "step": 21817 + }, + { + "epoch": 1.7991259173744538, + "grad_norm": 2.1652114391326904, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7386178970336914, + "num_tokens": 546101355.0, + "step": 21818 + }, + { + "epoch": 1.7992083779995052, + "grad_norm": 2.1298704147338867, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7440540790557861, + "num_tokens": 546129243.0, + "step": 21819 + }, + { + "epoch": 1.7992908386245567, + "grad_norm": 1.8769487142562866, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7733443975448608, + "num_tokens": 546162188.0, + "step": 21820 + }, + { + "epoch": 1.7993732992496083, + "grad_norm": 2.077493906021118, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7294872999191284, + "num_tokens": 546191660.0, + "step": 21821 + }, + { + "epoch": 1.7994557598746599, + "grad_norm": 1.6665523052215576, + "learning_rate": 1e-06, + "loss": 0.5249, + "mean_token_accuracy": 0.8167359828948975, + "num_tokens": 546226048.0, + "step": 21822 + }, + { + "epoch": 1.7995382204997115, + "grad_norm": 1.9232250452041626, + "learning_rate": 1e-06, + "loss": 0.7798, + "mean_token_accuracy": 0.7508659362792969, + "num_tokens": 546258930.0, + "step": 21823 + }, + { + "epoch": 1.7996206811247628, + "grad_norm": 2.061300277709961, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.7633412480354309, + "num_tokens": 546285921.0, + "step": 21824 + }, + { + "epoch": 1.7997031417498146, + "grad_norm": 1.7674797773361206, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7705258131027222, + "num_tokens": 546323205.0, + "step": 21825 + }, + { + "epoch": 1.799785602374866, + "grad_norm": 1.6637907028198242, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.7897055149078369, + "num_tokens": 546360503.0, + "step": 21826 + }, + { + "epoch": 1.7998680629999175, + "grad_norm": 1.9304754734039307, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7603353261947632, + "num_tokens": 546394192.0, + "step": 21827 + }, + { + "epoch": 1.799950523624969, + "grad_norm": 1.863247036933899, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7840227484703064, + "num_tokens": 546426126.0, + "step": 21828 + }, + { + "epoch": 1.8000329842500205, + "grad_norm": 1.99937903881073, + "learning_rate": 1e-06, + "loss": 0.5489, + "mean_token_accuracy": 0.815313458442688, + "num_tokens": 546452576.0, + "step": 21829 + }, + { + "epoch": 1.8001154448750722, + "grad_norm": 2.0308001041412354, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7810109853744507, + "num_tokens": 546479978.0, + "step": 21830 + }, + { + "epoch": 1.8001979055001236, + "grad_norm": 1.9518870115280151, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7343966960906982, + "num_tokens": 546511521.0, + "step": 21831 + }, + { + "epoch": 1.8002803661251754, + "grad_norm": 2.0877299308776855, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7604494094848633, + "num_tokens": 546540515.0, + "step": 21832 + }, + { + "epoch": 1.8003628267502267, + "grad_norm": 1.9515918493270874, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7451930046081543, + "num_tokens": 546573437.0, + "step": 21833 + }, + { + "epoch": 1.8004452873752783, + "grad_norm": 1.8257088661193848, + "learning_rate": 1e-06, + "loss": 0.6229, + "mean_token_accuracy": 0.7978899478912354, + "num_tokens": 546606500.0, + "step": 21834 + }, + { + "epoch": 1.8005277480003299, + "grad_norm": 2.3102505207061768, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7428858280181885, + "num_tokens": 546629331.0, + "step": 21835 + }, + { + "epoch": 1.8006102086253812, + "grad_norm": 2.1320676803588867, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7777173519134521, + "num_tokens": 546655435.0, + "step": 21836 + }, + { + "epoch": 1.800692669250433, + "grad_norm": 1.9672513008117676, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7554315328598022, + "num_tokens": 546687468.0, + "step": 21837 + }, + { + "epoch": 1.8007751298754844, + "grad_norm": 1.9216750860214233, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7651739120483398, + "num_tokens": 546720247.0, + "step": 21838 + }, + { + "epoch": 1.800857590500536, + "grad_norm": 1.9920344352722168, + "learning_rate": 1e-06, + "loss": 0.5455, + "mean_token_accuracy": 0.8170880079269409, + "num_tokens": 546749105.0, + "step": 21839 + }, + { + "epoch": 1.8009400511255875, + "grad_norm": 2.1177895069122314, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.7424253821372986, + "num_tokens": 546778935.0, + "step": 21840 + }, + { + "epoch": 1.801022511750639, + "grad_norm": 1.9230308532714844, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.7833395600318909, + "num_tokens": 546807522.0, + "step": 21841 + }, + { + "epoch": 1.8011049723756907, + "grad_norm": 1.9480748176574707, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.7974730730056763, + "num_tokens": 546836319.0, + "step": 21842 + }, + { + "epoch": 1.801187433000742, + "grad_norm": 1.9517223834991455, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7820248007774353, + "num_tokens": 546864839.0, + "step": 21843 + }, + { + "epoch": 1.8012698936257938, + "grad_norm": 1.8081238269805908, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7536115050315857, + "num_tokens": 546897610.0, + "step": 21844 + }, + { + "epoch": 1.8013523542508452, + "grad_norm": 1.8976263999938965, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7815228700637817, + "num_tokens": 546927710.0, + "step": 21845 + }, + { + "epoch": 1.8014348148758967, + "grad_norm": 1.910887598991394, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7786704897880554, + "num_tokens": 546959865.0, + "step": 21846 + }, + { + "epoch": 1.8015172755009483, + "grad_norm": 2.1080691814422607, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7808641195297241, + "num_tokens": 546987970.0, + "step": 21847 + }, + { + "epoch": 1.8015997361259999, + "grad_norm": 1.887418508529663, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7546168565750122, + "num_tokens": 547021806.0, + "step": 21848 + }, + { + "epoch": 1.8016821967510515, + "grad_norm": 2.383636713027954, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7716574668884277, + "num_tokens": 547043553.0, + "step": 21849 + }, + { + "epoch": 1.8017646573761028, + "grad_norm": 2.040501832962036, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7444190979003906, + "num_tokens": 547071917.0, + "step": 21850 + }, + { + "epoch": 1.8018471180011546, + "grad_norm": 1.7313650846481323, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.7912721037864685, + "num_tokens": 547109297.0, + "step": 21851 + }, + { + "epoch": 1.801929578626206, + "grad_norm": 1.8206559419631958, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7873117923736572, + "num_tokens": 547144246.0, + "step": 21852 + }, + { + "epoch": 1.8020120392512575, + "grad_norm": 2.0647130012512207, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7617509961128235, + "num_tokens": 547172264.0, + "step": 21853 + }, + { + "epoch": 1.802094499876309, + "grad_norm": 1.7277674674987793, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7556405067443848, + "num_tokens": 547206628.0, + "step": 21854 + }, + { + "epoch": 1.8021769605013604, + "grad_norm": 1.8237875699996948, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7544910907745361, + "num_tokens": 547241329.0, + "step": 21855 + }, + { + "epoch": 1.8022594211264122, + "grad_norm": 1.8479163646697998, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7398179769515991, + "num_tokens": 547274289.0, + "step": 21856 + }, + { + "epoch": 1.8023418817514636, + "grad_norm": 1.9708362817764282, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7761345505714417, + "num_tokens": 547305074.0, + "step": 21857 + }, + { + "epoch": 1.8024243423765152, + "grad_norm": 2.0345253944396973, + "learning_rate": 1e-06, + "loss": 0.7688, + "mean_token_accuracy": 0.7614769339561462, + "num_tokens": 547335895.0, + "step": 21858 + }, + { + "epoch": 1.8025068030015667, + "grad_norm": 2.546194553375244, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7257121801376343, + "num_tokens": 547355640.0, + "step": 21859 + }, + { + "epoch": 1.8025892636266183, + "grad_norm": 2.0077030658721924, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7531750202178955, + "num_tokens": 547385309.0, + "step": 21860 + }, + { + "epoch": 1.8026717242516699, + "grad_norm": 1.9830948114395142, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7791397571563721, + "num_tokens": 547414305.0, + "step": 21861 + }, + { + "epoch": 1.8027541848767212, + "grad_norm": 1.802196741104126, + "learning_rate": 1e-06, + "loss": 0.508, + "mean_token_accuracy": 0.82798832654953, + "num_tokens": 547448230.0, + "step": 21862 + }, + { + "epoch": 1.802836645501773, + "grad_norm": 1.969983458518982, + "learning_rate": 1e-06, + "loss": 0.5897, + "mean_token_accuracy": 0.8008362650871277, + "num_tokens": 547474549.0, + "step": 21863 + }, + { + "epoch": 1.8029191061268244, + "grad_norm": 2.029780149459839, + "learning_rate": 1e-06, + "loss": 0.6062, + "mean_token_accuracy": 0.7905498147010803, + "num_tokens": 547500280.0, + "step": 21864 + }, + { + "epoch": 1.803001566751876, + "grad_norm": 1.995881199836731, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7728368043899536, + "num_tokens": 547527900.0, + "step": 21865 + }, + { + "epoch": 1.8030840273769275, + "grad_norm": 2.010855197906494, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7821919918060303, + "num_tokens": 547557476.0, + "step": 21866 + }, + { + "epoch": 1.803166488001979, + "grad_norm": 2.082171678543091, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7667561769485474, + "num_tokens": 547585639.0, + "step": 21867 + }, + { + "epoch": 1.8032489486270307, + "grad_norm": 2.2897634506225586, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7160098552703857, + "num_tokens": 547614214.0, + "step": 21868 + }, + { + "epoch": 1.803331409252082, + "grad_norm": 2.0218117237091064, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7582879066467285, + "num_tokens": 547644262.0, + "step": 21869 + }, + { + "epoch": 1.8034138698771338, + "grad_norm": 2.3751637935638428, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.70295649766922, + "num_tokens": 547668412.0, + "step": 21870 + }, + { + "epoch": 1.8034963305021852, + "grad_norm": 1.882431983947754, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.7882895469665527, + "num_tokens": 547699698.0, + "step": 21871 + }, + { + "epoch": 1.8035787911272367, + "grad_norm": 2.2515995502471924, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7237760424613953, + "num_tokens": 547729790.0, + "step": 21872 + }, + { + "epoch": 1.8036612517522883, + "grad_norm": 1.93098783493042, + "learning_rate": 1e-06, + "loss": 0.5762, + "mean_token_accuracy": 0.8090054392814636, + "num_tokens": 547757558.0, + "step": 21873 + }, + { + "epoch": 1.8037437123773397, + "grad_norm": 1.95345139503479, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7754248380661011, + "num_tokens": 547786830.0, + "step": 21874 + }, + { + "epoch": 1.8038261730023915, + "grad_norm": 1.9779689311981201, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7635942101478577, + "num_tokens": 547814742.0, + "step": 21875 + }, + { + "epoch": 1.8039086336274428, + "grad_norm": 1.828343152999878, + "learning_rate": 1e-06, + "loss": 0.5849, + "mean_token_accuracy": 0.802505612373352, + "num_tokens": 547848704.0, + "step": 21876 + }, + { + "epoch": 1.8039910942524946, + "grad_norm": 1.9281117916107178, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7492682933807373, + "num_tokens": 547882165.0, + "step": 21877 + }, + { + "epoch": 1.804073554877546, + "grad_norm": 1.8861453533172607, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7352252006530762, + "num_tokens": 547915532.0, + "step": 21878 + }, + { + "epoch": 1.8041560155025975, + "grad_norm": 1.8846749067306519, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7286417484283447, + "num_tokens": 547949714.0, + "step": 21879 + }, + { + "epoch": 1.804238476127649, + "grad_norm": 2.0582289695739746, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7702350616455078, + "num_tokens": 547976920.0, + "step": 21880 + }, + { + "epoch": 1.8043209367527004, + "grad_norm": 1.9774750471115112, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7714605331420898, + "num_tokens": 548006019.0, + "step": 21881 + }, + { + "epoch": 1.8044033973777522, + "grad_norm": 2.0701205730438232, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.7747132182121277, + "num_tokens": 548032830.0, + "step": 21882 + }, + { + "epoch": 1.8044858580028036, + "grad_norm": 2.0195465087890625, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7684313058853149, + "num_tokens": 548064660.0, + "step": 21883 + }, + { + "epoch": 1.8045683186278552, + "grad_norm": 1.9559268951416016, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7381312847137451, + "num_tokens": 548095548.0, + "step": 21884 + }, + { + "epoch": 1.8046507792529067, + "grad_norm": 2.173466205596924, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.769038200378418, + "num_tokens": 548121314.0, + "step": 21885 + }, + { + "epoch": 1.8047332398779583, + "grad_norm": 2.0101025104522705, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7591682076454163, + "num_tokens": 548151718.0, + "step": 21886 + }, + { + "epoch": 1.8048157005030099, + "grad_norm": 1.9492303133010864, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7475780248641968, + "num_tokens": 548183134.0, + "step": 21887 + }, + { + "epoch": 1.8048981611280612, + "grad_norm": 1.9048645496368408, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7495375871658325, + "num_tokens": 548213724.0, + "step": 21888 + }, + { + "epoch": 1.804980621753113, + "grad_norm": 2.023561954498291, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7472668886184692, + "num_tokens": 548243076.0, + "step": 21889 + }, + { + "epoch": 1.8050630823781644, + "grad_norm": 1.9642484188079834, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.7956059575080872, + "num_tokens": 548271329.0, + "step": 21890 + }, + { + "epoch": 1.805145543003216, + "grad_norm": 1.8782881498336792, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7778298854827881, + "num_tokens": 548301775.0, + "step": 21891 + }, + { + "epoch": 1.8052280036282675, + "grad_norm": 2.0173568725585938, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7358361482620239, + "num_tokens": 548332928.0, + "step": 21892 + }, + { + "epoch": 1.805310464253319, + "grad_norm": 2.0681405067443848, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.715883731842041, + "num_tokens": 548361458.0, + "step": 21893 + }, + { + "epoch": 1.8053929248783707, + "grad_norm": 1.7765178680419922, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7533215880393982, + "num_tokens": 548397014.0, + "step": 21894 + }, + { + "epoch": 1.805475385503422, + "grad_norm": 1.9686020612716675, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7538706064224243, + "num_tokens": 548426367.0, + "step": 21895 + }, + { + "epoch": 1.8055578461284738, + "grad_norm": 2.198775053024292, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7190865874290466, + "num_tokens": 548453262.0, + "step": 21896 + }, + { + "epoch": 1.8056403067535252, + "grad_norm": 2.281203031539917, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.7908681631088257, + "num_tokens": 548479152.0, + "step": 21897 + }, + { + "epoch": 1.8057227673785767, + "grad_norm": 2.2246837615966797, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.7907419204711914, + "num_tokens": 548506426.0, + "step": 21898 + }, + { + "epoch": 1.8058052280036283, + "grad_norm": 2.021016836166382, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.751782238483429, + "num_tokens": 548536108.0, + "step": 21899 + }, + { + "epoch": 1.8058876886286797, + "grad_norm": 1.694251298904419, + "learning_rate": 1e-06, + "loss": 0.5934, + "mean_token_accuracy": 0.7976809144020081, + "num_tokens": 548571782.0, + "step": 21900 + }, + { + "epoch": 1.8059701492537314, + "grad_norm": 1.795597791671753, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7850771546363831, + "num_tokens": 548603207.0, + "step": 21901 + }, + { + "epoch": 1.8060526098787828, + "grad_norm": 1.9277637004852295, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7735075950622559, + "num_tokens": 548632078.0, + "step": 21902 + }, + { + "epoch": 1.8061350705038344, + "grad_norm": 2.0892088413238525, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7742689847946167, + "num_tokens": 548656869.0, + "step": 21903 + }, + { + "epoch": 1.806217531128886, + "grad_norm": 1.8837025165557861, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7552427649497986, + "num_tokens": 548688122.0, + "step": 21904 + }, + { + "epoch": 1.8062999917539375, + "grad_norm": 1.9442083835601807, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7570359706878662, + "num_tokens": 548718939.0, + "step": 21905 + }, + { + "epoch": 1.806382452378989, + "grad_norm": 2.02956223487854, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7640773057937622, + "num_tokens": 548747789.0, + "step": 21906 + }, + { + "epoch": 1.8064649130040404, + "grad_norm": 1.8462247848510742, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7672499418258667, + "num_tokens": 548778560.0, + "step": 21907 + }, + { + "epoch": 1.8065473736290922, + "grad_norm": 2.1226227283477783, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7699370980262756, + "num_tokens": 548804399.0, + "step": 21908 + }, + { + "epoch": 1.8066298342541436, + "grad_norm": 2.1329360008239746, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7549079656600952, + "num_tokens": 548830129.0, + "step": 21909 + }, + { + "epoch": 1.8067122948791952, + "grad_norm": 1.8885538578033447, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7502256631851196, + "num_tokens": 548865610.0, + "step": 21910 + }, + { + "epoch": 1.8067947555042467, + "grad_norm": 1.874562382698059, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7560152411460876, + "num_tokens": 548896600.0, + "step": 21911 + }, + { + "epoch": 1.8068772161292983, + "grad_norm": 2.0203986167907715, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.757561445236206, + "num_tokens": 548923424.0, + "step": 21912 + }, + { + "epoch": 1.8069596767543499, + "grad_norm": 1.6820571422576904, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7487058639526367, + "num_tokens": 548964431.0, + "step": 21913 + }, + { + "epoch": 1.8070421373794012, + "grad_norm": 2.012803554534912, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7356206178665161, + "num_tokens": 548994048.0, + "step": 21914 + }, + { + "epoch": 1.807124598004453, + "grad_norm": 2.091031551361084, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7411614060401917, + "num_tokens": 549022776.0, + "step": 21915 + }, + { + "epoch": 1.8072070586295044, + "grad_norm": 1.9983162879943848, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7337217330932617, + "num_tokens": 549051991.0, + "step": 21916 + }, + { + "epoch": 1.807289519254556, + "grad_norm": 2.0663628578186035, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7583432793617249, + "num_tokens": 549080351.0, + "step": 21917 + }, + { + "epoch": 1.8073719798796075, + "grad_norm": 1.8363442420959473, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.7924183011054993, + "num_tokens": 549115471.0, + "step": 21918 + }, + { + "epoch": 1.8074544405046589, + "grad_norm": 1.935351014137268, + "learning_rate": 1e-06, + "loss": 0.7954, + "mean_token_accuracy": 0.7524944543838501, + "num_tokens": 549149367.0, + "step": 21919 + }, + { + "epoch": 1.8075369011297107, + "grad_norm": 1.900131344795227, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7517764568328857, + "num_tokens": 549181485.0, + "step": 21920 + }, + { + "epoch": 1.807619361754762, + "grad_norm": 1.9484919309616089, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.789094090461731, + "num_tokens": 549211091.0, + "step": 21921 + }, + { + "epoch": 1.8077018223798138, + "grad_norm": 1.9655845165252686, + "learning_rate": 1e-06, + "loss": 0.6144, + "mean_token_accuracy": 0.796890377998352, + "num_tokens": 549237286.0, + "step": 21922 + }, + { + "epoch": 1.8077842830048652, + "grad_norm": 1.820375919342041, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7729008793830872, + "num_tokens": 549271456.0, + "step": 21923 + }, + { + "epoch": 1.8078667436299167, + "grad_norm": 1.909541130065918, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.7972750663757324, + "num_tokens": 549305568.0, + "step": 21924 + }, + { + "epoch": 1.8079492042549683, + "grad_norm": 1.9564629793167114, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7752105593681335, + "num_tokens": 549335525.0, + "step": 21925 + }, + { + "epoch": 1.8080316648800197, + "grad_norm": 2.0142323970794678, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7755014896392822, + "num_tokens": 549367242.0, + "step": 21926 + }, + { + "epoch": 1.8081141255050714, + "grad_norm": 2.269690990447998, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7369289398193359, + "num_tokens": 549390803.0, + "step": 21927 + }, + { + "epoch": 1.8081965861301228, + "grad_norm": 2.01951265335083, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7629364728927612, + "num_tokens": 549422638.0, + "step": 21928 + }, + { + "epoch": 1.8082790467551744, + "grad_norm": 2.130448579788208, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7215961813926697, + "num_tokens": 549450819.0, + "step": 21929 + }, + { + "epoch": 1.808361507380226, + "grad_norm": 1.9102109670639038, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7669312953948975, + "num_tokens": 549482014.0, + "step": 21930 + }, + { + "epoch": 1.8084439680052775, + "grad_norm": 2.0836117267608643, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7177326679229736, + "num_tokens": 549511439.0, + "step": 21931 + }, + { + "epoch": 1.808526428630329, + "grad_norm": 1.957869529724121, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7290400266647339, + "num_tokens": 549545635.0, + "step": 21932 + }, + { + "epoch": 1.8086088892553804, + "grad_norm": 2.1906239986419678, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7737401723861694, + "num_tokens": 549570127.0, + "step": 21933 + }, + { + "epoch": 1.8086913498804322, + "grad_norm": 2.3905670642852783, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.750774621963501, + "num_tokens": 549594654.0, + "step": 21934 + }, + { + "epoch": 1.8087738105054836, + "grad_norm": 1.9767577648162842, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7683391571044922, + "num_tokens": 549626707.0, + "step": 21935 + }, + { + "epoch": 1.8088562711305352, + "grad_norm": 2.091057062149048, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7283772230148315, + "num_tokens": 549655947.0, + "step": 21936 + }, + { + "epoch": 1.8089387317555867, + "grad_norm": 1.9051222801208496, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7716677188873291, + "num_tokens": 549689232.0, + "step": 21937 + }, + { + "epoch": 1.8090211923806383, + "grad_norm": 1.9597529172897339, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7357596158981323, + "num_tokens": 549719894.0, + "step": 21938 + }, + { + "epoch": 1.8091036530056899, + "grad_norm": 1.8568994998931885, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7815404534339905, + "num_tokens": 549752449.0, + "step": 21939 + }, + { + "epoch": 1.8091861136307412, + "grad_norm": 2.2238545417785645, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7748458385467529, + "num_tokens": 549776021.0, + "step": 21940 + }, + { + "epoch": 1.809268574255793, + "grad_norm": 1.7748457193374634, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7650373578071594, + "num_tokens": 549812162.0, + "step": 21941 + }, + { + "epoch": 1.8093510348808444, + "grad_norm": 1.9230302572250366, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7803633213043213, + "num_tokens": 549842358.0, + "step": 21942 + }, + { + "epoch": 1.809433495505896, + "grad_norm": 1.9291938543319702, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.7791624665260315, + "num_tokens": 549870907.0, + "step": 21943 + }, + { + "epoch": 1.8095159561309475, + "grad_norm": 1.9952154159545898, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.784160315990448, + "num_tokens": 549895884.0, + "step": 21944 + }, + { + "epoch": 1.8095984167559989, + "grad_norm": 1.9004913568496704, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7737220525741577, + "num_tokens": 549927149.0, + "step": 21945 + }, + { + "epoch": 1.8096808773810507, + "grad_norm": 2.0407938957214355, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7765958905220032, + "num_tokens": 549955482.0, + "step": 21946 + }, + { + "epoch": 1.809763338006102, + "grad_norm": 2.327594757080078, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7344549894332886, + "num_tokens": 549978909.0, + "step": 21947 + }, + { + "epoch": 1.8098457986311536, + "grad_norm": 1.9547407627105713, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7381829023361206, + "num_tokens": 550009906.0, + "step": 21948 + }, + { + "epoch": 1.8099282592562052, + "grad_norm": 2.010298013687134, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7584462761878967, + "num_tokens": 550041510.0, + "step": 21949 + }, + { + "epoch": 1.8100107198812567, + "grad_norm": 2.2263951301574707, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7695567607879639, + "num_tokens": 550069198.0, + "step": 21950 + }, + { + "epoch": 1.8100931805063083, + "grad_norm": 1.763596773147583, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7678200006484985, + "num_tokens": 550106746.0, + "step": 21951 + }, + { + "epoch": 1.8101756411313596, + "grad_norm": 1.855719804763794, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.770968496799469, + "num_tokens": 550139934.0, + "step": 21952 + }, + { + "epoch": 1.8102581017564114, + "grad_norm": 1.8815516233444214, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7710440158843994, + "num_tokens": 550169409.0, + "step": 21953 + }, + { + "epoch": 1.8103405623814628, + "grad_norm": 1.8626667261123657, + "learning_rate": 1e-06, + "loss": 0.6044, + "mean_token_accuracy": 0.8021407127380371, + "num_tokens": 550204996.0, + "step": 21954 + }, + { + "epoch": 1.8104230230065144, + "grad_norm": 1.8348362445831299, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7730119824409485, + "num_tokens": 550239493.0, + "step": 21955 + }, + { + "epoch": 1.810505483631566, + "grad_norm": 1.9327902793884277, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7692062854766846, + "num_tokens": 550268043.0, + "step": 21956 + }, + { + "epoch": 1.8105879442566175, + "grad_norm": 2.1239562034606934, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7389694452285767, + "num_tokens": 550293361.0, + "step": 21957 + }, + { + "epoch": 1.810670404881669, + "grad_norm": 2.036478042602539, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7574878334999084, + "num_tokens": 550324867.0, + "step": 21958 + }, + { + "epoch": 1.8107528655067204, + "grad_norm": 1.8938895463943481, + "learning_rate": 1e-06, + "loss": 0.572, + "mean_token_accuracy": 0.8033573627471924, + "num_tokens": 550354457.0, + "step": 21959 + }, + { + "epoch": 1.8108353261317722, + "grad_norm": 1.9926819801330566, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7697225213050842, + "num_tokens": 550386118.0, + "step": 21960 + }, + { + "epoch": 1.8109177867568236, + "grad_norm": 2.1441335678100586, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7597396373748779, + "num_tokens": 550412541.0, + "step": 21961 + }, + { + "epoch": 1.8110002473818752, + "grad_norm": 1.920412302017212, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.734380304813385, + "num_tokens": 550442617.0, + "step": 21962 + }, + { + "epoch": 1.8110827080069267, + "grad_norm": 2.047652244567871, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7574987411499023, + "num_tokens": 550471752.0, + "step": 21963 + }, + { + "epoch": 1.811165168631978, + "grad_norm": 2.0852468013763428, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7536094188690186, + "num_tokens": 550499240.0, + "step": 21964 + }, + { + "epoch": 1.8112476292570299, + "grad_norm": 2.3175477981567383, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7637472748756409, + "num_tokens": 550521659.0, + "step": 21965 + }, + { + "epoch": 1.8113300898820812, + "grad_norm": 2.132547378540039, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7185168266296387, + "num_tokens": 550549321.0, + "step": 21966 + }, + { + "epoch": 1.8114125505071328, + "grad_norm": 1.9718799591064453, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7591918706893921, + "num_tokens": 550578075.0, + "step": 21967 + }, + { + "epoch": 1.8114950111321844, + "grad_norm": 2.229086399078369, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7531853318214417, + "num_tokens": 550606635.0, + "step": 21968 + }, + { + "epoch": 1.811577471757236, + "grad_norm": 2.0994482040405273, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7256531715393066, + "num_tokens": 550636507.0, + "step": 21969 + }, + { + "epoch": 1.8116599323822875, + "grad_norm": 2.10732102394104, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7628859877586365, + "num_tokens": 550663429.0, + "step": 21970 + }, + { + "epoch": 1.8117423930073389, + "grad_norm": 1.7914677858352661, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.7802081108093262, + "num_tokens": 550696559.0, + "step": 21971 + }, + { + "epoch": 1.8118248536323907, + "grad_norm": 2.072916030883789, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7471588253974915, + "num_tokens": 550723391.0, + "step": 21972 + }, + { + "epoch": 1.811907314257442, + "grad_norm": 2.110647201538086, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7521666288375854, + "num_tokens": 550747293.0, + "step": 21973 + }, + { + "epoch": 1.8119897748824936, + "grad_norm": 1.8629295825958252, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7554410696029663, + "num_tokens": 550781635.0, + "step": 21974 + }, + { + "epoch": 1.8120722355075451, + "grad_norm": 1.7523733377456665, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7322795391082764, + "num_tokens": 550818029.0, + "step": 21975 + }, + { + "epoch": 1.8121546961325967, + "grad_norm": 2.018610715866089, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7744799852371216, + "num_tokens": 550850699.0, + "step": 21976 + }, + { + "epoch": 1.8122371567576483, + "grad_norm": 1.9585866928100586, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7396613955497742, + "num_tokens": 550881649.0, + "step": 21977 + }, + { + "epoch": 1.8123196173826996, + "grad_norm": 1.7611161470413208, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.7959632277488708, + "num_tokens": 550915711.0, + "step": 21978 + }, + { + "epoch": 1.8124020780077514, + "grad_norm": 1.9860732555389404, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.768903911113739, + "num_tokens": 550945712.0, + "step": 21979 + }, + { + "epoch": 1.8124845386328028, + "grad_norm": 2.1619465351104736, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7564995288848877, + "num_tokens": 550970420.0, + "step": 21980 + }, + { + "epoch": 1.8125669992578544, + "grad_norm": 1.8918689489364624, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.7815776467323303, + "num_tokens": 551000880.0, + "step": 21981 + }, + { + "epoch": 1.812649459882906, + "grad_norm": 2.422586679458618, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7780060768127441, + "num_tokens": 551021266.0, + "step": 21982 + }, + { + "epoch": 1.8127319205079573, + "grad_norm": 2.0354604721069336, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7374803423881531, + "num_tokens": 551050242.0, + "step": 21983 + }, + { + "epoch": 1.812814381133009, + "grad_norm": 2.048009157180786, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7661175727844238, + "num_tokens": 551078539.0, + "step": 21984 + }, + { + "epoch": 1.8128968417580604, + "grad_norm": 2.0719940662384033, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7244721055030823, + "num_tokens": 551107879.0, + "step": 21985 + }, + { + "epoch": 1.8129793023831122, + "grad_norm": 2.134920358657837, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7494189739227295, + "num_tokens": 551133740.0, + "step": 21986 + }, + { + "epoch": 1.8130617630081636, + "grad_norm": 2.1448771953582764, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7542030215263367, + "num_tokens": 551160844.0, + "step": 21987 + }, + { + "epoch": 1.8131442236332151, + "grad_norm": 2.3610754013061523, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7490319013595581, + "num_tokens": 551184900.0, + "step": 21988 + }, + { + "epoch": 1.8132266842582667, + "grad_norm": 1.9992280006408691, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.776869535446167, + "num_tokens": 551217356.0, + "step": 21989 + }, + { + "epoch": 1.813309144883318, + "grad_norm": 1.9495576620101929, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7814921140670776, + "num_tokens": 551248532.0, + "step": 21990 + }, + { + "epoch": 1.8133916055083699, + "grad_norm": 1.9174549579620361, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.7783123254776001, + "num_tokens": 551277476.0, + "step": 21991 + }, + { + "epoch": 1.8134740661334212, + "grad_norm": 2.125430107116699, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7507508397102356, + "num_tokens": 551304369.0, + "step": 21992 + }, + { + "epoch": 1.8135565267584728, + "grad_norm": 1.9462881088256836, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7697758674621582, + "num_tokens": 551335109.0, + "step": 21993 + }, + { + "epoch": 1.8136389873835244, + "grad_norm": 2.0176737308502197, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7459456920623779, + "num_tokens": 551364805.0, + "step": 21994 + }, + { + "epoch": 1.813721448008576, + "grad_norm": 1.995489478111267, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.7832533121109009, + "num_tokens": 551393834.0, + "step": 21995 + }, + { + "epoch": 1.8138039086336275, + "grad_norm": 1.9514000415802002, + "learning_rate": 1e-06, + "loss": 0.5948, + "mean_token_accuracy": 0.8023841381072998, + "num_tokens": 551422018.0, + "step": 21996 + }, + { + "epoch": 1.8138863692586789, + "grad_norm": 2.1827051639556885, + "learning_rate": 1e-06, + "loss": 0.5921, + "mean_token_accuracy": 0.7919672727584839, + "num_tokens": 551445229.0, + "step": 21997 + }, + { + "epoch": 1.8139688298837306, + "grad_norm": 2.108370065689087, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.739392101764679, + "num_tokens": 551472136.0, + "step": 21998 + }, + { + "epoch": 1.814051290508782, + "grad_norm": 2.101811408996582, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7522574663162231, + "num_tokens": 551498928.0, + "step": 21999 + }, + { + "epoch": 1.8141337511338336, + "grad_norm": 2.0807480812072754, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7657819390296936, + "num_tokens": 551526819.0, + "step": 22000 + }, + { + "epoch": 1.8142162117588851, + "grad_norm": 1.7817186117172241, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.770399272441864, + "num_tokens": 551562461.0, + "step": 22001 + }, + { + "epoch": 1.8142986723839367, + "grad_norm": 2.147702217102051, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7325278520584106, + "num_tokens": 551587949.0, + "step": 22002 + }, + { + "epoch": 1.8143811330089883, + "grad_norm": 1.8093647956848145, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7679156064987183, + "num_tokens": 551622497.0, + "step": 22003 + }, + { + "epoch": 1.8144635936340396, + "grad_norm": 1.7308686971664429, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.744082510471344, + "num_tokens": 551660481.0, + "step": 22004 + }, + { + "epoch": 1.8145460542590914, + "grad_norm": 2.145857572555542, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7211255431175232, + "num_tokens": 551688973.0, + "step": 22005 + }, + { + "epoch": 1.8146285148841428, + "grad_norm": 1.9619382619857788, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.7597715258598328, + "num_tokens": 551717869.0, + "step": 22006 + }, + { + "epoch": 1.8147109755091944, + "grad_norm": 1.9913816452026367, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7762077450752258, + "num_tokens": 551745536.0, + "step": 22007 + }, + { + "epoch": 1.814793436134246, + "grad_norm": 1.898141860961914, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7726173400878906, + "num_tokens": 551776443.0, + "step": 22008 + }, + { + "epoch": 1.8148758967592973, + "grad_norm": 1.9627208709716797, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7673692107200623, + "num_tokens": 551811462.0, + "step": 22009 + }, + { + "epoch": 1.814958357384349, + "grad_norm": 2.0530993938446045, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7549795508384705, + "num_tokens": 551839728.0, + "step": 22010 + }, + { + "epoch": 1.8150408180094004, + "grad_norm": 2.0283584594726562, + "learning_rate": 1e-06, + "loss": 0.6229, + "mean_token_accuracy": 0.8014296293258667, + "num_tokens": 551865876.0, + "step": 22011 + }, + { + "epoch": 1.815123278634452, + "grad_norm": 1.8867685794830322, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.78875732421875, + "num_tokens": 551896254.0, + "step": 22012 + }, + { + "epoch": 1.8152057392595036, + "grad_norm": 1.968253254890442, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7179750204086304, + "num_tokens": 551930184.0, + "step": 22013 + }, + { + "epoch": 1.8152881998845551, + "grad_norm": 1.77894926071167, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7848175168037415, + "num_tokens": 551963347.0, + "step": 22014 + }, + { + "epoch": 1.8153706605096067, + "grad_norm": 2.2811508178710938, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7325477600097656, + "num_tokens": 551987148.0, + "step": 22015 + }, + { + "epoch": 1.815453121134658, + "grad_norm": 1.769274115562439, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7807081937789917, + "num_tokens": 552020758.0, + "step": 22016 + }, + { + "epoch": 1.8155355817597099, + "grad_norm": 1.8646831512451172, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.798500657081604, + "num_tokens": 552052794.0, + "step": 22017 + }, + { + "epoch": 1.8156180423847612, + "grad_norm": 2.0952727794647217, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7643038034439087, + "num_tokens": 552080920.0, + "step": 22018 + }, + { + "epoch": 1.8157005030098128, + "grad_norm": 2.1243441104888916, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.7843127250671387, + "num_tokens": 552107356.0, + "step": 22019 + }, + { + "epoch": 1.8157829636348644, + "grad_norm": 1.9204343557357788, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7442172765731812, + "num_tokens": 552137822.0, + "step": 22020 + }, + { + "epoch": 1.815865424259916, + "grad_norm": 1.9837007522583008, + "learning_rate": 1e-06, + "loss": 0.5298, + "mean_token_accuracy": 0.809325635433197, + "num_tokens": 552166348.0, + "step": 22021 + }, + { + "epoch": 1.8159478848849675, + "grad_norm": 2.154386043548584, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7617193460464478, + "num_tokens": 552193541.0, + "step": 22022 + }, + { + "epoch": 1.8160303455100189, + "grad_norm": 2.2089178562164307, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7496201992034912, + "num_tokens": 552219836.0, + "step": 22023 + }, + { + "epoch": 1.8161128061350706, + "grad_norm": 2.344848155975342, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7555572986602783, + "num_tokens": 552242921.0, + "step": 22024 + }, + { + "epoch": 1.816195266760122, + "grad_norm": 1.982586145401001, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7671006321907043, + "num_tokens": 552274020.0, + "step": 22025 + }, + { + "epoch": 1.8162777273851736, + "grad_norm": 1.9436721801757812, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7193100452423096, + "num_tokens": 552306226.0, + "step": 22026 + }, + { + "epoch": 1.8163601880102251, + "grad_norm": 1.8815741539001465, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.782503604888916, + "num_tokens": 552336977.0, + "step": 22027 + }, + { + "epoch": 1.8164426486352765, + "grad_norm": 1.9796233177185059, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7932780981063843, + "num_tokens": 552368029.0, + "step": 22028 + }, + { + "epoch": 1.8165251092603283, + "grad_norm": 1.891563057899475, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7525360584259033, + "num_tokens": 552400615.0, + "step": 22029 + }, + { + "epoch": 1.8166075698853796, + "grad_norm": 2.041315793991089, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7531076669692993, + "num_tokens": 552428415.0, + "step": 22030 + }, + { + "epoch": 1.8166900305104314, + "grad_norm": 1.8802156448364258, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7662359476089478, + "num_tokens": 552460095.0, + "step": 22031 + }, + { + "epoch": 1.8167724911354828, + "grad_norm": 1.8322516679763794, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7714177370071411, + "num_tokens": 552495467.0, + "step": 22032 + }, + { + "epoch": 1.8168549517605344, + "grad_norm": 1.9686249494552612, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.7788373231887817, + "num_tokens": 552526350.0, + "step": 22033 + }, + { + "epoch": 1.816937412385586, + "grad_norm": 2.451348304748535, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.764519453048706, + "num_tokens": 552545509.0, + "step": 22034 + }, + { + "epoch": 1.8170198730106373, + "grad_norm": 1.983218789100647, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7706952095031738, + "num_tokens": 552573545.0, + "step": 22035 + }, + { + "epoch": 1.817102333635689, + "grad_norm": 1.884766936302185, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.7908185720443726, + "num_tokens": 552606050.0, + "step": 22036 + }, + { + "epoch": 1.8171847942607404, + "grad_norm": 2.0078210830688477, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7205389738082886, + "num_tokens": 552635199.0, + "step": 22037 + }, + { + "epoch": 1.817267254885792, + "grad_norm": 1.902124047279358, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.7918401956558228, + "num_tokens": 552666947.0, + "step": 22038 + }, + { + "epoch": 1.8173497155108436, + "grad_norm": 2.0103461742401123, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7593453526496887, + "num_tokens": 552696818.0, + "step": 22039 + }, + { + "epoch": 1.8174321761358951, + "grad_norm": 2.288602828979492, + "learning_rate": 1e-06, + "loss": 0.5841, + "mean_token_accuracy": 0.800764799118042, + "num_tokens": 552720739.0, + "step": 22040 + }, + { + "epoch": 1.8175146367609467, + "grad_norm": 2.2044308185577393, + "learning_rate": 1e-06, + "loss": 0.7883, + "mean_token_accuracy": 0.7478299140930176, + "num_tokens": 552744226.0, + "step": 22041 + }, + { + "epoch": 1.817597097385998, + "grad_norm": 2.0174801349639893, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7632889747619629, + "num_tokens": 552773931.0, + "step": 22042 + }, + { + "epoch": 1.8176795580110499, + "grad_norm": 1.9066357612609863, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7764490842819214, + "num_tokens": 552803385.0, + "step": 22043 + }, + { + "epoch": 1.8177620186361012, + "grad_norm": 1.9510071277618408, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.7591341733932495, + "num_tokens": 552833010.0, + "step": 22044 + }, + { + "epoch": 1.8178444792611528, + "grad_norm": 1.834524154663086, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.7606363892555237, + "num_tokens": 552869181.0, + "step": 22045 + }, + { + "epoch": 1.8179269398862044, + "grad_norm": 1.8836884498596191, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7642462253570557, + "num_tokens": 552903151.0, + "step": 22046 + }, + { + "epoch": 1.818009400511256, + "grad_norm": 2.019456386566162, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7431058883666992, + "num_tokens": 552935157.0, + "step": 22047 + }, + { + "epoch": 1.8180918611363075, + "grad_norm": 2.1088130474090576, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7423622012138367, + "num_tokens": 552961173.0, + "step": 22048 + }, + { + "epoch": 1.8181743217613588, + "grad_norm": 2.504976987838745, + "learning_rate": 1e-06, + "loss": 0.5835, + "mean_token_accuracy": 0.798208475112915, + "num_tokens": 552980467.0, + "step": 22049 + }, + { + "epoch": 1.8182567823864106, + "grad_norm": 1.9301140308380127, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7507147789001465, + "num_tokens": 553014851.0, + "step": 22050 + }, + { + "epoch": 1.818339243011462, + "grad_norm": 1.8601946830749512, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7878035306930542, + "num_tokens": 553044120.0, + "step": 22051 + }, + { + "epoch": 1.8184217036365136, + "grad_norm": 2.0467031002044678, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7081977128982544, + "num_tokens": 553077099.0, + "step": 22052 + }, + { + "epoch": 1.8185041642615651, + "grad_norm": 1.9667936563491821, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.7752482295036316, + "num_tokens": 553105397.0, + "step": 22053 + }, + { + "epoch": 1.8185866248866165, + "grad_norm": 2.011359453201294, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7559055089950562, + "num_tokens": 553135170.0, + "step": 22054 + }, + { + "epoch": 1.8186690855116683, + "grad_norm": 2.0823163986206055, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7731336355209351, + "num_tokens": 553161679.0, + "step": 22055 + }, + { + "epoch": 1.8187515461367196, + "grad_norm": 1.7366700172424316, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.7944375276565552, + "num_tokens": 553194643.0, + "step": 22056 + }, + { + "epoch": 1.8188340067617712, + "grad_norm": 2.090515613555908, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7720352411270142, + "num_tokens": 553219912.0, + "step": 22057 + }, + { + "epoch": 1.8189164673868228, + "grad_norm": 1.7632256746292114, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7721656560897827, + "num_tokens": 553256921.0, + "step": 22058 + }, + { + "epoch": 1.8189989280118743, + "grad_norm": 1.7241554260253906, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7840919494628906, + "num_tokens": 553293988.0, + "step": 22059 + }, + { + "epoch": 1.819081388636926, + "grad_norm": 1.9202579259872437, + "learning_rate": 1e-06, + "loss": 0.5917, + "mean_token_accuracy": 0.8042783737182617, + "num_tokens": 553321725.0, + "step": 22060 + }, + { + "epoch": 1.8191638492619773, + "grad_norm": 2.238204002380371, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7446136474609375, + "num_tokens": 553348266.0, + "step": 22061 + }, + { + "epoch": 1.819246309887029, + "grad_norm": 2.0812954902648926, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7701411247253418, + "num_tokens": 553374158.0, + "step": 22062 + }, + { + "epoch": 1.8193287705120804, + "grad_norm": 1.9105794429779053, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7273796796798706, + "num_tokens": 553407407.0, + "step": 22063 + }, + { + "epoch": 1.819411231137132, + "grad_norm": 1.991521954536438, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.7428617477416992, + "num_tokens": 553436809.0, + "step": 22064 + }, + { + "epoch": 1.8194936917621836, + "grad_norm": 1.8639743328094482, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.7638911008834839, + "num_tokens": 553471214.0, + "step": 22065 + }, + { + "epoch": 1.8195761523872351, + "grad_norm": 1.7613884210586548, + "learning_rate": 1e-06, + "loss": 0.5947, + "mean_token_accuracy": 0.80372154712677, + "num_tokens": 553503338.0, + "step": 22066 + }, + { + "epoch": 1.8196586130122867, + "grad_norm": 2.1892011165618896, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7781435251235962, + "num_tokens": 553528672.0, + "step": 22067 + }, + { + "epoch": 1.819741073637338, + "grad_norm": 1.9963239431381226, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7418633103370667, + "num_tokens": 553556819.0, + "step": 22068 + }, + { + "epoch": 1.8198235342623899, + "grad_norm": 2.150817394256592, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7578046321868896, + "num_tokens": 553582597.0, + "step": 22069 + }, + { + "epoch": 1.8199059948874412, + "grad_norm": 1.946982979774475, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7508797645568848, + "num_tokens": 553613918.0, + "step": 22070 + }, + { + "epoch": 1.8199884555124928, + "grad_norm": 2.071645736694336, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.7793862819671631, + "num_tokens": 553644295.0, + "step": 22071 + }, + { + "epoch": 1.8200709161375443, + "grad_norm": 2.3610241413116455, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7337680459022522, + "num_tokens": 553669559.0, + "step": 22072 + }, + { + "epoch": 1.8201533767625957, + "grad_norm": 1.8359969854354858, + "learning_rate": 1e-06, + "loss": 0.5164, + "mean_token_accuracy": 0.8283678293228149, + "num_tokens": 553698376.0, + "step": 22073 + }, + { + "epoch": 1.8202358373876475, + "grad_norm": 1.914918065071106, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7758709788322449, + "num_tokens": 553730567.0, + "step": 22074 + }, + { + "epoch": 1.8203182980126988, + "grad_norm": 2.2039794921875, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7461676001548767, + "num_tokens": 553756967.0, + "step": 22075 + }, + { + "epoch": 1.8204007586377506, + "grad_norm": 2.301351547241211, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7149832844734192, + "num_tokens": 553785652.0, + "step": 22076 + }, + { + "epoch": 1.820483219262802, + "grad_norm": 1.9553046226501465, + "learning_rate": 1e-06, + "loss": 0.7842, + "mean_token_accuracy": 0.7474650144577026, + "num_tokens": 553818647.0, + "step": 22077 + }, + { + "epoch": 1.8205656798878536, + "grad_norm": 1.7431888580322266, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7474945783615112, + "num_tokens": 553857951.0, + "step": 22078 + }, + { + "epoch": 1.8206481405129051, + "grad_norm": 2.015448570251465, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7580916285514832, + "num_tokens": 553887923.0, + "step": 22079 + }, + { + "epoch": 1.8207306011379565, + "grad_norm": 2.1072397232055664, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.7717515230178833, + "num_tokens": 553914267.0, + "step": 22080 + }, + { + "epoch": 1.8208130617630083, + "grad_norm": 2.048696756362915, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.7846351265907288, + "num_tokens": 553942415.0, + "step": 22081 + }, + { + "epoch": 1.8208955223880596, + "grad_norm": 1.9055476188659668, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7610549330711365, + "num_tokens": 553973152.0, + "step": 22082 + }, + { + "epoch": 1.8209779830131112, + "grad_norm": 1.873721718788147, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.741098165512085, + "num_tokens": 554007235.0, + "step": 22083 + }, + { + "epoch": 1.8210604436381628, + "grad_norm": 2.0216243267059326, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7782629132270813, + "num_tokens": 554033938.0, + "step": 22084 + }, + { + "epoch": 1.8211429042632143, + "grad_norm": 1.8676387071609497, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.782207727432251, + "num_tokens": 554064043.0, + "step": 22085 + }, + { + "epoch": 1.821225364888266, + "grad_norm": 2.212777614593506, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6929442882537842, + "num_tokens": 554092510.0, + "step": 22086 + }, + { + "epoch": 1.8213078255133173, + "grad_norm": 2.174706220626831, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7394623756408691, + "num_tokens": 554119456.0, + "step": 22087 + }, + { + "epoch": 1.821390286138369, + "grad_norm": 1.966706395149231, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7542132139205933, + "num_tokens": 554151740.0, + "step": 22088 + }, + { + "epoch": 1.8214727467634204, + "grad_norm": 2.002418041229248, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7450345754623413, + "num_tokens": 554184875.0, + "step": 22089 + }, + { + "epoch": 1.821555207388472, + "grad_norm": 1.8420264720916748, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.738455057144165, + "num_tokens": 554217978.0, + "step": 22090 + }, + { + "epoch": 1.8216376680135236, + "grad_norm": 2.0646073818206787, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7560226917266846, + "num_tokens": 554246786.0, + "step": 22091 + }, + { + "epoch": 1.8217201286385751, + "grad_norm": 2.0976345539093018, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7403305768966675, + "num_tokens": 554276591.0, + "step": 22092 + }, + { + "epoch": 1.8218025892636267, + "grad_norm": 1.7629793882369995, + "learning_rate": 1e-06, + "loss": 0.5468, + "mean_token_accuracy": 0.8152339458465576, + "num_tokens": 554306777.0, + "step": 22093 + }, + { + "epoch": 1.821885049888678, + "grad_norm": 1.9671337604522705, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.7900777459144592, + "num_tokens": 554335566.0, + "step": 22094 + }, + { + "epoch": 1.8219675105137298, + "grad_norm": 2.2170605659484863, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7399095296859741, + "num_tokens": 554361308.0, + "step": 22095 + }, + { + "epoch": 1.8220499711387812, + "grad_norm": 2.112673759460449, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7407447695732117, + "num_tokens": 554391550.0, + "step": 22096 + }, + { + "epoch": 1.8221324317638328, + "grad_norm": 1.9661651849746704, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7422764301300049, + "num_tokens": 554422301.0, + "step": 22097 + }, + { + "epoch": 1.8222148923888843, + "grad_norm": 2.187166690826416, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7353644967079163, + "num_tokens": 554448150.0, + "step": 22098 + }, + { + "epoch": 1.8222973530139357, + "grad_norm": 1.8741062879562378, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7387441396713257, + "num_tokens": 554480095.0, + "step": 22099 + }, + { + "epoch": 1.8223798136389875, + "grad_norm": 1.8615916967391968, + "learning_rate": 1e-06, + "loss": 0.5429, + "mean_token_accuracy": 0.8156837224960327, + "num_tokens": 554509395.0, + "step": 22100 + }, + { + "epoch": 1.8224622742640388, + "grad_norm": 1.8838083744049072, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.785475492477417, + "num_tokens": 554538996.0, + "step": 22101 + }, + { + "epoch": 1.8225447348890904, + "grad_norm": 1.8448576927185059, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7911893129348755, + "num_tokens": 554569259.0, + "step": 22102 + }, + { + "epoch": 1.822627195514142, + "grad_norm": 1.9382562637329102, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7052541971206665, + "num_tokens": 554604004.0, + "step": 22103 + }, + { + "epoch": 1.8227096561391936, + "grad_norm": 2.1778416633605957, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.7869567275047302, + "num_tokens": 554631965.0, + "step": 22104 + }, + { + "epoch": 1.8227921167642451, + "grad_norm": 2.061450958251953, + "learning_rate": 1e-06, + "loss": 0.7614, + "mean_token_accuracy": 0.7546837329864502, + "num_tokens": 554661743.0, + "step": 22105 + }, + { + "epoch": 1.8228745773892965, + "grad_norm": 1.7171828746795654, + "learning_rate": 1e-06, + "loss": 0.5157, + "mean_token_accuracy": 0.8261380195617676, + "num_tokens": 554699671.0, + "step": 22106 + }, + { + "epoch": 1.8229570380143483, + "grad_norm": 2.0456535816192627, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7548081874847412, + "num_tokens": 554729876.0, + "step": 22107 + }, + { + "epoch": 1.8230394986393996, + "grad_norm": 1.8535408973693848, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.781549334526062, + "num_tokens": 554760057.0, + "step": 22108 + }, + { + "epoch": 1.8231219592644512, + "grad_norm": 1.8278242349624634, + "learning_rate": 1e-06, + "loss": 0.582, + "mean_token_accuracy": 0.8057173490524292, + "num_tokens": 554790797.0, + "step": 22109 + }, + { + "epoch": 1.8232044198895028, + "grad_norm": 2.1236441135406494, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7490556240081787, + "num_tokens": 554817382.0, + "step": 22110 + }, + { + "epoch": 1.8232868805145543, + "grad_norm": 2.0343472957611084, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7435901165008545, + "num_tokens": 554845269.0, + "step": 22111 + }, + { + "epoch": 1.823369341139606, + "grad_norm": 1.6620248556137085, + "learning_rate": 1e-06, + "loss": 0.5554, + "mean_token_accuracy": 0.8112263679504395, + "num_tokens": 554882624.0, + "step": 22112 + }, + { + "epoch": 1.8234518017646573, + "grad_norm": 2.0049290657043457, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7732638120651245, + "num_tokens": 554910421.0, + "step": 22113 + }, + { + "epoch": 1.823534262389709, + "grad_norm": 1.8082836866378784, + "learning_rate": 1e-06, + "loss": 0.6041, + "mean_token_accuracy": 0.8047205209732056, + "num_tokens": 554942135.0, + "step": 22114 + }, + { + "epoch": 1.8236167230147604, + "grad_norm": 1.8864434957504272, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7368271350860596, + "num_tokens": 554976423.0, + "step": 22115 + }, + { + "epoch": 1.823699183639812, + "grad_norm": 2.031738519668579, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.756118893623352, + "num_tokens": 555004530.0, + "step": 22116 + }, + { + "epoch": 1.8237816442648636, + "grad_norm": 1.9137214422225952, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7185975313186646, + "num_tokens": 555037036.0, + "step": 22117 + }, + { + "epoch": 1.823864104889915, + "grad_norm": 2.1124422550201416, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7385138869285583, + "num_tokens": 555066578.0, + "step": 22118 + }, + { + "epoch": 1.8239465655149667, + "grad_norm": 2.1110644340515137, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7569724321365356, + "num_tokens": 555094147.0, + "step": 22119 + }, + { + "epoch": 1.824029026140018, + "grad_norm": 5.089408874511719, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7738125324249268, + "num_tokens": 555118833.0, + "step": 22120 + }, + { + "epoch": 1.8241114867650698, + "grad_norm": 1.9362999200820923, + "learning_rate": 1e-06, + "loss": 0.5945, + "mean_token_accuracy": 0.8066705465316772, + "num_tokens": 555149438.0, + "step": 22121 + }, + { + "epoch": 1.8241939473901212, + "grad_norm": 1.7931393384933472, + "learning_rate": 1e-06, + "loss": 0.7673, + "mean_token_accuracy": 0.7543094158172607, + "num_tokens": 555184872.0, + "step": 22122 + }, + { + "epoch": 1.8242764080151728, + "grad_norm": 1.7285206317901611, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.7865073680877686, + "num_tokens": 555222982.0, + "step": 22123 + }, + { + "epoch": 1.8243588686402243, + "grad_norm": 1.9045205116271973, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7399851083755493, + "num_tokens": 555257937.0, + "step": 22124 + }, + { + "epoch": 1.8244413292652757, + "grad_norm": 1.9675869941711426, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7557862997055054, + "num_tokens": 555291512.0, + "step": 22125 + }, + { + "epoch": 1.8245237898903275, + "grad_norm": 1.9982830286026, + "learning_rate": 1e-06, + "loss": 0.771, + "mean_token_accuracy": 0.7564125657081604, + "num_tokens": 555319660.0, + "step": 22126 + }, + { + "epoch": 1.8246062505153788, + "grad_norm": 1.9817209243774414, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7818920612335205, + "num_tokens": 555350721.0, + "step": 22127 + }, + { + "epoch": 1.8246887111404304, + "grad_norm": 2.0184426307678223, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7697030305862427, + "num_tokens": 555381387.0, + "step": 22128 + }, + { + "epoch": 1.824771171765482, + "grad_norm": 2.4573895931243896, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7233601808547974, + "num_tokens": 555404151.0, + "step": 22129 + }, + { + "epoch": 1.8248536323905336, + "grad_norm": 2.165987014770508, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7578827738761902, + "num_tokens": 555430686.0, + "step": 22130 + }, + { + "epoch": 1.8249360930155851, + "grad_norm": 2.0258936882019043, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.7908793687820435, + "num_tokens": 555458696.0, + "step": 22131 + }, + { + "epoch": 1.8250185536406365, + "grad_norm": 1.9118925333023071, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7710471749305725, + "num_tokens": 555492122.0, + "step": 22132 + }, + { + "epoch": 1.8251010142656883, + "grad_norm": 1.803265929222107, + "learning_rate": 1e-06, + "loss": 0.6115, + "mean_token_accuracy": 0.7986045479774475, + "num_tokens": 555524040.0, + "step": 22133 + }, + { + "epoch": 1.8251834748907396, + "grad_norm": 1.9674476385116577, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7681503295898438, + "num_tokens": 555555026.0, + "step": 22134 + }, + { + "epoch": 1.8252659355157912, + "grad_norm": 1.8766775131225586, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7707509994506836, + "num_tokens": 555585690.0, + "step": 22135 + }, + { + "epoch": 1.8253483961408428, + "grad_norm": 2.2126657962799072, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7775586843490601, + "num_tokens": 555609990.0, + "step": 22136 + }, + { + "epoch": 1.8254308567658943, + "grad_norm": 1.8998029232025146, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7662447690963745, + "num_tokens": 555641195.0, + "step": 22137 + }, + { + "epoch": 1.825513317390946, + "grad_norm": 1.8251190185546875, + "learning_rate": 1e-06, + "loss": 0.6036, + "mean_token_accuracy": 0.7927997708320618, + "num_tokens": 555673003.0, + "step": 22138 + }, + { + "epoch": 1.8255957780159973, + "grad_norm": 2.1849844455718994, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7593597769737244, + "num_tokens": 555699060.0, + "step": 22139 + }, + { + "epoch": 1.825678238641049, + "grad_norm": 1.8700965642929077, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7659822702407837, + "num_tokens": 555733975.0, + "step": 22140 + }, + { + "epoch": 1.8257606992661004, + "grad_norm": 2.330197811126709, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7564035654067993, + "num_tokens": 555758975.0, + "step": 22141 + }, + { + "epoch": 1.825843159891152, + "grad_norm": 1.9445677995681763, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.7955418825149536, + "num_tokens": 555786939.0, + "step": 22142 + }, + { + "epoch": 1.8259256205162036, + "grad_norm": 1.725305438041687, + "learning_rate": 1e-06, + "loss": 0.572, + "mean_token_accuracy": 0.8120115995407104, + "num_tokens": 555822683.0, + "step": 22143 + }, + { + "epoch": 1.826008081141255, + "grad_norm": 1.994094729423523, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7710062861442566, + "num_tokens": 555856214.0, + "step": 22144 + }, + { + "epoch": 1.8260905417663067, + "grad_norm": 2.0044045448303223, + "learning_rate": 1e-06, + "loss": 0.563, + "mean_token_accuracy": 0.8093913793563843, + "num_tokens": 555881976.0, + "step": 22145 + }, + { + "epoch": 1.826173002391358, + "grad_norm": 1.8770853281021118, + "learning_rate": 1e-06, + "loss": 0.6039, + "mean_token_accuracy": 0.7957342863082886, + "num_tokens": 555909506.0, + "step": 22146 + }, + { + "epoch": 1.8262554630164096, + "grad_norm": 2.1454849243164062, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7451348304748535, + "num_tokens": 555934476.0, + "step": 22147 + }, + { + "epoch": 1.8263379236414612, + "grad_norm": 2.039996385574341, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.733103334903717, + "num_tokens": 555963449.0, + "step": 22148 + }, + { + "epoch": 1.8264203842665128, + "grad_norm": 1.7524943351745605, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7473735809326172, + "num_tokens": 555998469.0, + "step": 22149 + }, + { + "epoch": 1.8265028448915643, + "grad_norm": 2.1239216327667236, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7570828199386597, + "num_tokens": 556026438.0, + "step": 22150 + }, + { + "epoch": 1.8265853055166157, + "grad_norm": 1.8053842782974243, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7546957731246948, + "num_tokens": 556062441.0, + "step": 22151 + }, + { + "epoch": 1.8266677661416675, + "grad_norm": 1.8992788791656494, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7635940313339233, + "num_tokens": 556095507.0, + "step": 22152 + }, + { + "epoch": 1.8267502267667188, + "grad_norm": 2.2720766067504883, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7564271688461304, + "num_tokens": 556118448.0, + "step": 22153 + }, + { + "epoch": 1.8268326873917704, + "grad_norm": 2.180386543273926, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7806482315063477, + "num_tokens": 556145281.0, + "step": 22154 + }, + { + "epoch": 1.826915148016822, + "grad_norm": 1.8031041622161865, + "learning_rate": 1e-06, + "loss": 0.5952, + "mean_token_accuracy": 0.8046495914459229, + "num_tokens": 556176018.0, + "step": 22155 + }, + { + "epoch": 1.8269976086418735, + "grad_norm": 2.3016796112060547, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7662696838378906, + "num_tokens": 556200200.0, + "step": 22156 + }, + { + "epoch": 1.8270800692669251, + "grad_norm": 1.9744703769683838, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.7434844970703125, + "num_tokens": 556230813.0, + "step": 22157 + }, + { + "epoch": 1.8271625298919765, + "grad_norm": 1.7512831687927246, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7372044920921326, + "num_tokens": 556267564.0, + "step": 22158 + }, + { + "epoch": 1.8272449905170283, + "grad_norm": 1.8686878681182861, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7290679216384888, + "num_tokens": 556302045.0, + "step": 22159 + }, + { + "epoch": 1.8273274511420796, + "grad_norm": 1.856508731842041, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.7964017987251282, + "num_tokens": 556333632.0, + "step": 22160 + }, + { + "epoch": 1.8274099117671312, + "grad_norm": 2.1312029361724854, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7332690954208374, + "num_tokens": 556359840.0, + "step": 22161 + }, + { + "epoch": 1.8274923723921828, + "grad_norm": 1.7661879062652588, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7786632180213928, + "num_tokens": 556394277.0, + "step": 22162 + }, + { + "epoch": 1.8275748330172341, + "grad_norm": 1.8571624755859375, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7401385307312012, + "num_tokens": 556428109.0, + "step": 22163 + }, + { + "epoch": 1.827657293642286, + "grad_norm": 2.0518641471862793, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7541130185127258, + "num_tokens": 556455083.0, + "step": 22164 + }, + { + "epoch": 1.8277397542673373, + "grad_norm": 1.8066630363464355, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.7823381423950195, + "num_tokens": 556488827.0, + "step": 22165 + }, + { + "epoch": 1.8278222148923888, + "grad_norm": 1.878713846206665, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7350425124168396, + "num_tokens": 556523288.0, + "step": 22166 + }, + { + "epoch": 1.8279046755174404, + "grad_norm": 1.8460874557495117, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.8033141493797302, + "num_tokens": 556552372.0, + "step": 22167 + }, + { + "epoch": 1.827987136142492, + "grad_norm": 1.8413735628128052, + "learning_rate": 1e-06, + "loss": 0.7486, + "mean_token_accuracy": 0.7569307684898376, + "num_tokens": 556588326.0, + "step": 22168 + }, + { + "epoch": 1.8280695967675435, + "grad_norm": 1.677520751953125, + "learning_rate": 1e-06, + "loss": 0.7656, + "mean_token_accuracy": 0.7505075931549072, + "num_tokens": 556628003.0, + "step": 22169 + }, + { + "epoch": 1.828152057392595, + "grad_norm": 1.8995304107666016, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.74598628282547, + "num_tokens": 556661956.0, + "step": 22170 + }, + { + "epoch": 1.8282345180176467, + "grad_norm": 1.9830588102340698, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7280737161636353, + "num_tokens": 556692596.0, + "step": 22171 + }, + { + "epoch": 1.828316978642698, + "grad_norm": 1.9204280376434326, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.7461210489273071, + "num_tokens": 556722716.0, + "step": 22172 + }, + { + "epoch": 1.8283994392677496, + "grad_norm": 1.9241418838500977, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7472504377365112, + "num_tokens": 556754966.0, + "step": 22173 + }, + { + "epoch": 1.8284818998928012, + "grad_norm": 1.9231624603271484, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7898232340812683, + "num_tokens": 556786445.0, + "step": 22174 + }, + { + "epoch": 1.8285643605178528, + "grad_norm": 2.122013807296753, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7483445405960083, + "num_tokens": 556812359.0, + "step": 22175 + }, + { + "epoch": 1.8286468211429043, + "grad_norm": 1.813706874847412, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7750585675239563, + "num_tokens": 556848800.0, + "step": 22176 + }, + { + "epoch": 1.8287292817679557, + "grad_norm": 1.8277029991149902, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7679923176765442, + "num_tokens": 556881522.0, + "step": 22177 + }, + { + "epoch": 1.8288117423930075, + "grad_norm": 1.7850933074951172, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7414072155952454, + "num_tokens": 556918196.0, + "step": 22178 + }, + { + "epoch": 1.8288942030180588, + "grad_norm": 1.8709263801574707, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7798435091972351, + "num_tokens": 556948607.0, + "step": 22179 + }, + { + "epoch": 1.8289766636431104, + "grad_norm": 2.1538634300231934, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.72404944896698, + "num_tokens": 556975846.0, + "step": 22180 + }, + { + "epoch": 1.829059124268162, + "grad_norm": 2.090808868408203, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7648696899414062, + "num_tokens": 557001697.0, + "step": 22181 + }, + { + "epoch": 1.8291415848932135, + "grad_norm": 1.824364185333252, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.7982856631278992, + "num_tokens": 557032535.0, + "step": 22182 + }, + { + "epoch": 1.8292240455182651, + "grad_norm": 2.078376293182373, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7701668739318848, + "num_tokens": 557059791.0, + "step": 22183 + }, + { + "epoch": 1.8293065061433165, + "grad_norm": 1.6452158689498901, + "learning_rate": 1e-06, + "loss": 0.5865, + "mean_token_accuracy": 0.8020035028457642, + "num_tokens": 557094569.0, + "step": 22184 + }, + { + "epoch": 1.8293889667683683, + "grad_norm": 2.0750856399536133, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7581323385238647, + "num_tokens": 557122408.0, + "step": 22185 + }, + { + "epoch": 1.8294714273934196, + "grad_norm": 1.8740977048873901, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7708745002746582, + "num_tokens": 557152796.0, + "step": 22186 + }, + { + "epoch": 1.8295538880184712, + "grad_norm": 1.905051350593567, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7573884725570679, + "num_tokens": 557185547.0, + "step": 22187 + }, + { + "epoch": 1.8296363486435228, + "grad_norm": 2.0571846961975098, + "learning_rate": 1e-06, + "loss": 0.6315, + "mean_token_accuracy": 0.7957026958465576, + "num_tokens": 557213879.0, + "step": 22188 + }, + { + "epoch": 1.829718809268574, + "grad_norm": 1.876279354095459, + "learning_rate": 1e-06, + "loss": 0.5907, + "mean_token_accuracy": 0.8020315170288086, + "num_tokens": 557243768.0, + "step": 22189 + }, + { + "epoch": 1.829801269893626, + "grad_norm": 1.8097493648529053, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7643042802810669, + "num_tokens": 557278941.0, + "step": 22190 + }, + { + "epoch": 1.8298837305186773, + "grad_norm": 1.8245666027069092, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7761439085006714, + "num_tokens": 557313071.0, + "step": 22191 + }, + { + "epoch": 1.8299661911437288, + "grad_norm": 2.4392826557159424, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7472777366638184, + "num_tokens": 557333662.0, + "step": 22192 + }, + { + "epoch": 1.8300486517687804, + "grad_norm": 1.9715543985366821, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.7920814752578735, + "num_tokens": 557366728.0, + "step": 22193 + }, + { + "epoch": 1.830131112393832, + "grad_norm": 2.1495420932769775, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7318980097770691, + "num_tokens": 557393110.0, + "step": 22194 + }, + { + "epoch": 1.8302135730188835, + "grad_norm": 1.9966706037521362, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7531891465187073, + "num_tokens": 557422049.0, + "step": 22195 + }, + { + "epoch": 1.830296033643935, + "grad_norm": 1.8349453210830688, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7871427536010742, + "num_tokens": 557453437.0, + "step": 22196 + }, + { + "epoch": 1.8303784942689867, + "grad_norm": 1.8046568632125854, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7719870805740356, + "num_tokens": 557485283.0, + "step": 22197 + }, + { + "epoch": 1.830460954894038, + "grad_norm": 2.0055763721466064, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7586650252342224, + "num_tokens": 557514745.0, + "step": 22198 + }, + { + "epoch": 1.8305434155190896, + "grad_norm": 1.8074520826339722, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7646148204803467, + "num_tokens": 557550199.0, + "step": 22199 + }, + { + "epoch": 1.8306258761441412, + "grad_norm": 2.0128817558288574, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7201793193817139, + "num_tokens": 557582682.0, + "step": 22200 + }, + { + "epoch": 1.8307083367691928, + "grad_norm": 1.929970383644104, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.773648738861084, + "num_tokens": 557613574.0, + "step": 22201 + }, + { + "epoch": 1.8307907973942443, + "grad_norm": 1.9918078184127808, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.7250226736068726, + "num_tokens": 557645466.0, + "step": 22202 + }, + { + "epoch": 1.8308732580192957, + "grad_norm": 1.9756954908370972, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7370346784591675, + "num_tokens": 557677789.0, + "step": 22203 + }, + { + "epoch": 1.8309557186443475, + "grad_norm": 2.0480902194976807, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7308880090713501, + "num_tokens": 557707954.0, + "step": 22204 + }, + { + "epoch": 1.8310381792693988, + "grad_norm": 2.387040376663208, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7565090656280518, + "num_tokens": 557729166.0, + "step": 22205 + }, + { + "epoch": 1.8311206398944504, + "grad_norm": 1.7947322130203247, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.764276385307312, + "num_tokens": 557762968.0, + "step": 22206 + }, + { + "epoch": 1.831203100519502, + "grad_norm": 1.6487016677856445, + "learning_rate": 1e-06, + "loss": 0.5791, + "mean_token_accuracy": 0.8070487976074219, + "num_tokens": 557800556.0, + "step": 22207 + }, + { + "epoch": 1.8312855611445533, + "grad_norm": 1.7345280647277832, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7657256126403809, + "num_tokens": 557834479.0, + "step": 22208 + }, + { + "epoch": 1.8313680217696051, + "grad_norm": 2.085832118988037, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7775381803512573, + "num_tokens": 557859558.0, + "step": 22209 + }, + { + "epoch": 1.8314504823946565, + "grad_norm": 2.2041332721710205, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7419018149375916, + "num_tokens": 557885517.0, + "step": 22210 + }, + { + "epoch": 1.831532943019708, + "grad_norm": 2.1372604370117188, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7239651679992676, + "num_tokens": 557912540.0, + "step": 22211 + }, + { + "epoch": 1.8316154036447596, + "grad_norm": 1.90404212474823, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7407901287078857, + "num_tokens": 557945768.0, + "step": 22212 + }, + { + "epoch": 1.8316978642698112, + "grad_norm": 1.81574285030365, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.784313440322876, + "num_tokens": 557979881.0, + "step": 22213 + }, + { + "epoch": 1.8317803248948628, + "grad_norm": 1.9013303518295288, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7697969675064087, + "num_tokens": 558014769.0, + "step": 22214 + }, + { + "epoch": 1.831862785519914, + "grad_norm": 2.1934750080108643, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.724703311920166, + "num_tokens": 558039709.0, + "step": 22215 + }, + { + "epoch": 1.831945246144966, + "grad_norm": 1.8550266027450562, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7488594651222229, + "num_tokens": 558075416.0, + "step": 22216 + }, + { + "epoch": 1.8320277067700173, + "grad_norm": 1.9496698379516602, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.76565021276474, + "num_tokens": 558106623.0, + "step": 22217 + }, + { + "epoch": 1.8321101673950688, + "grad_norm": 2.0217416286468506, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7760218381881714, + "num_tokens": 558135954.0, + "step": 22218 + }, + { + "epoch": 1.8321926280201204, + "grad_norm": 2.172718048095703, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7439459562301636, + "num_tokens": 558161919.0, + "step": 22219 + }, + { + "epoch": 1.832275088645172, + "grad_norm": 2.1726107597351074, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.7950133085250854, + "num_tokens": 558187511.0, + "step": 22220 + }, + { + "epoch": 1.8323575492702235, + "grad_norm": 2.2210428714752197, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7341707944869995, + "num_tokens": 558212826.0, + "step": 22221 + }, + { + "epoch": 1.832440009895275, + "grad_norm": 1.6531455516815186, + "learning_rate": 1e-06, + "loss": 0.5302, + "mean_token_accuracy": 0.8181577324867249, + "num_tokens": 558245993.0, + "step": 22222 + }, + { + "epoch": 1.8325224705203267, + "grad_norm": 2.1385765075683594, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.7923986911773682, + "num_tokens": 558271954.0, + "step": 22223 + }, + { + "epoch": 1.832604931145378, + "grad_norm": 1.747422456741333, + "learning_rate": 1e-06, + "loss": 0.5437, + "mean_token_accuracy": 0.8165315389633179, + "num_tokens": 558304039.0, + "step": 22224 + }, + { + "epoch": 1.8326873917704296, + "grad_norm": 2.0392251014709473, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7918474674224854, + "num_tokens": 558333785.0, + "step": 22225 + }, + { + "epoch": 1.8327698523954812, + "grad_norm": 2.2527947425842285, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7731695175170898, + "num_tokens": 558361724.0, + "step": 22226 + }, + { + "epoch": 1.8328523130205325, + "grad_norm": 1.9000053405761719, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7449443340301514, + "num_tokens": 558396751.0, + "step": 22227 + }, + { + "epoch": 1.8329347736455843, + "grad_norm": 2.192770004272461, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.764453649520874, + "num_tokens": 558422017.0, + "step": 22228 + }, + { + "epoch": 1.8330172342706357, + "grad_norm": 1.7939624786376953, + "learning_rate": 1e-06, + "loss": 0.5731, + "mean_token_accuracy": 0.8081737160682678, + "num_tokens": 558453812.0, + "step": 22229 + }, + { + "epoch": 1.8330996948956875, + "grad_norm": 2.3750526905059814, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.74828040599823, + "num_tokens": 558476734.0, + "step": 22230 + }, + { + "epoch": 1.8331821555207388, + "grad_norm": 2.03692364692688, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7456947565078735, + "num_tokens": 558505541.0, + "step": 22231 + }, + { + "epoch": 1.8332646161457904, + "grad_norm": 2.093859910964966, + "learning_rate": 1e-06, + "loss": 0.7775, + "mean_token_accuracy": 0.7576706409454346, + "num_tokens": 558534619.0, + "step": 22232 + }, + { + "epoch": 1.833347076770842, + "grad_norm": 2.1009016036987305, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7332267761230469, + "num_tokens": 558562086.0, + "step": 22233 + }, + { + "epoch": 1.8334295373958933, + "grad_norm": 2.071364402770996, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7665901184082031, + "num_tokens": 558589374.0, + "step": 22234 + }, + { + "epoch": 1.833511998020945, + "grad_norm": 2.0602340698242188, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7394025325775146, + "num_tokens": 558618155.0, + "step": 22235 + }, + { + "epoch": 1.8335944586459965, + "grad_norm": 2.120338201522827, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.7536143064498901, + "num_tokens": 558647150.0, + "step": 22236 + }, + { + "epoch": 1.833676919271048, + "grad_norm": 1.9518368244171143, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7889301180839539, + "num_tokens": 558676795.0, + "step": 22237 + }, + { + "epoch": 1.8337593798960996, + "grad_norm": 2.1368818283081055, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7709726691246033, + "num_tokens": 558701526.0, + "step": 22238 + }, + { + "epoch": 1.8338418405211512, + "grad_norm": 1.806994915008545, + "learning_rate": 1e-06, + "loss": 0.6131, + "mean_token_accuracy": 0.7978848218917847, + "num_tokens": 558734435.0, + "step": 22239 + }, + { + "epoch": 1.8339243011462028, + "grad_norm": 1.9221271276474, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.748609185218811, + "num_tokens": 558765805.0, + "step": 22240 + }, + { + "epoch": 1.834006761771254, + "grad_norm": 2.271543264389038, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7403597235679626, + "num_tokens": 558790217.0, + "step": 22241 + }, + { + "epoch": 1.834089222396306, + "grad_norm": 1.9337799549102783, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7185770273208618, + "num_tokens": 558823196.0, + "step": 22242 + }, + { + "epoch": 1.8341716830213572, + "grad_norm": 1.975981593132019, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7641966342926025, + "num_tokens": 558855938.0, + "step": 22243 + }, + { + "epoch": 1.8342541436464088, + "grad_norm": 2.0851340293884277, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7652454376220703, + "num_tokens": 558885378.0, + "step": 22244 + }, + { + "epoch": 1.8343366042714604, + "grad_norm": 1.9646332263946533, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7514200806617737, + "num_tokens": 558914797.0, + "step": 22245 + }, + { + "epoch": 1.834419064896512, + "grad_norm": 1.882264256477356, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7627483606338501, + "num_tokens": 558949375.0, + "step": 22246 + }, + { + "epoch": 1.8345015255215635, + "grad_norm": 2.4036197662353516, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7666777968406677, + "num_tokens": 558972074.0, + "step": 22247 + }, + { + "epoch": 1.8345839861466149, + "grad_norm": 1.9669477939605713, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7432833909988403, + "num_tokens": 559001225.0, + "step": 22248 + }, + { + "epoch": 1.8346664467716667, + "grad_norm": 1.9106858968734741, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.733121931552887, + "num_tokens": 559036398.0, + "step": 22249 + }, + { + "epoch": 1.834748907396718, + "grad_norm": 1.9797800779342651, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7390598058700562, + "num_tokens": 559067299.0, + "step": 22250 + }, + { + "epoch": 1.8348313680217696, + "grad_norm": 2.0876946449279785, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.729954183101654, + "num_tokens": 559096080.0, + "step": 22251 + }, + { + "epoch": 1.8349138286468212, + "grad_norm": 1.9047163724899292, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7625851631164551, + "num_tokens": 559128254.0, + "step": 22252 + }, + { + "epoch": 1.8349962892718725, + "grad_norm": 1.8225994110107422, + "learning_rate": 1e-06, + "loss": 0.7673, + "mean_token_accuracy": 0.7600926160812378, + "num_tokens": 559163395.0, + "step": 22253 + }, + { + "epoch": 1.8350787498969243, + "grad_norm": 2.0656845569610596, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.746768593788147, + "num_tokens": 559191059.0, + "step": 22254 + }, + { + "epoch": 1.8351612105219757, + "grad_norm": 2.125324010848999, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.7871825695037842, + "num_tokens": 559216653.0, + "step": 22255 + }, + { + "epoch": 1.8352436711470272, + "grad_norm": 2.1303746700286865, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7510253190994263, + "num_tokens": 559243930.0, + "step": 22256 + }, + { + "epoch": 1.8353261317720788, + "grad_norm": 2.0044102668762207, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7502130270004272, + "num_tokens": 559274553.0, + "step": 22257 + }, + { + "epoch": 1.8354085923971304, + "grad_norm": 2.1591272354125977, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7727987766265869, + "num_tokens": 559298136.0, + "step": 22258 + }, + { + "epoch": 1.835491053022182, + "grad_norm": 2.176583766937256, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7708693146705627, + "num_tokens": 559325483.0, + "step": 22259 + }, + { + "epoch": 1.8355735136472333, + "grad_norm": 1.908104658126831, + "learning_rate": 1e-06, + "loss": 0.5325, + "mean_token_accuracy": 0.802046537399292, + "num_tokens": 559353204.0, + "step": 22260 + }, + { + "epoch": 1.835655974272285, + "grad_norm": 2.030447244644165, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7358592748641968, + "num_tokens": 559384013.0, + "step": 22261 + }, + { + "epoch": 1.8357384348973365, + "grad_norm": 1.8627967834472656, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7824016809463501, + "num_tokens": 559416065.0, + "step": 22262 + }, + { + "epoch": 1.835820895522388, + "grad_norm": 1.7890597581863403, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7664034366607666, + "num_tokens": 559452413.0, + "step": 22263 + }, + { + "epoch": 1.8359033561474396, + "grad_norm": 1.9333158731460571, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7534658908843994, + "num_tokens": 559482949.0, + "step": 22264 + }, + { + "epoch": 1.8359858167724912, + "grad_norm": 1.9220367670059204, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7489792108535767, + "num_tokens": 559515290.0, + "step": 22265 + }, + { + "epoch": 1.8360682773975427, + "grad_norm": 1.7150651216506958, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.782229483127594, + "num_tokens": 559552482.0, + "step": 22266 + }, + { + "epoch": 1.836150738022594, + "grad_norm": 2.162107229232788, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7330155968666077, + "num_tokens": 559580464.0, + "step": 22267 + }, + { + "epoch": 1.836233198647646, + "grad_norm": 1.838865041732788, + "learning_rate": 1e-06, + "loss": 0.5785, + "mean_token_accuracy": 0.8074967861175537, + "num_tokens": 559615503.0, + "step": 22268 + }, + { + "epoch": 1.8363156592726972, + "grad_norm": 2.2845921516418457, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7436890006065369, + "num_tokens": 559640845.0, + "step": 22269 + }, + { + "epoch": 1.8363981198977488, + "grad_norm": 2.123599052429199, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7750442028045654, + "num_tokens": 559666208.0, + "step": 22270 + }, + { + "epoch": 1.8364805805228004, + "grad_norm": 2.014514923095703, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7489097118377686, + "num_tokens": 559699253.0, + "step": 22271 + }, + { + "epoch": 1.8365630411478517, + "grad_norm": 2.18672251701355, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7306689023971558, + "num_tokens": 559725264.0, + "step": 22272 + }, + { + "epoch": 1.8366455017729035, + "grad_norm": 1.779998540878296, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.789923369884491, + "num_tokens": 559760040.0, + "step": 22273 + }, + { + "epoch": 1.8367279623979549, + "grad_norm": 1.8907290697097778, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.8023201823234558, + "num_tokens": 559790038.0, + "step": 22274 + }, + { + "epoch": 1.8368104230230067, + "grad_norm": 2.23776912689209, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.75594162940979, + "num_tokens": 559816705.0, + "step": 22275 + }, + { + "epoch": 1.836892883648058, + "grad_norm": 1.7796602249145508, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7407413721084595, + "num_tokens": 559850172.0, + "step": 22276 + }, + { + "epoch": 1.8369753442731096, + "grad_norm": 1.8932723999023438, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7751022577285767, + "num_tokens": 559883064.0, + "step": 22277 + }, + { + "epoch": 1.8370578048981612, + "grad_norm": 2.091200113296509, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.7900760173797607, + "num_tokens": 559909164.0, + "step": 22278 + }, + { + "epoch": 1.8371402655232125, + "grad_norm": 2.116779327392578, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7143480777740479, + "num_tokens": 559936794.0, + "step": 22279 + }, + { + "epoch": 1.8372227261482643, + "grad_norm": 1.9640579223632812, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.7845498323440552, + "num_tokens": 559966959.0, + "step": 22280 + }, + { + "epoch": 1.8373051867733157, + "grad_norm": 2.4242634773254395, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7521628737449646, + "num_tokens": 559989775.0, + "step": 22281 + }, + { + "epoch": 1.8373876473983672, + "grad_norm": 2.1404571533203125, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.7828283309936523, + "num_tokens": 560013068.0, + "step": 22282 + }, + { + "epoch": 1.8374701080234188, + "grad_norm": 1.9860732555389404, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.751171886920929, + "num_tokens": 560048347.0, + "step": 22283 + }, + { + "epoch": 1.8375525686484704, + "grad_norm": 1.9441232681274414, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.7586204409599304, + "num_tokens": 560080778.0, + "step": 22284 + }, + { + "epoch": 1.837635029273522, + "grad_norm": 1.8287146091461182, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7835297584533691, + "num_tokens": 560115752.0, + "step": 22285 + }, + { + "epoch": 1.8377174898985733, + "grad_norm": 2.1683707237243652, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7639355063438416, + "num_tokens": 560142096.0, + "step": 22286 + }, + { + "epoch": 1.837799950523625, + "grad_norm": 1.8177131414413452, + "learning_rate": 1e-06, + "loss": 0.4975, + "mean_token_accuracy": 0.8341646194458008, + "num_tokens": 560174790.0, + "step": 22287 + }, + { + "epoch": 1.8378824111486765, + "grad_norm": 1.9616543054580688, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.7849332094192505, + "num_tokens": 560203293.0, + "step": 22288 + }, + { + "epoch": 1.837964871773728, + "grad_norm": 2.2280373573303223, + "learning_rate": 1e-06, + "loss": 0.7585, + "mean_token_accuracy": 0.745335578918457, + "num_tokens": 560228752.0, + "step": 22289 + }, + { + "epoch": 1.8380473323987796, + "grad_norm": 1.7740849256515503, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.759522020816803, + "num_tokens": 560264513.0, + "step": 22290 + }, + { + "epoch": 1.8381297930238312, + "grad_norm": 1.91200590133667, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7629657983779907, + "num_tokens": 560294779.0, + "step": 22291 + }, + { + "epoch": 1.8382122536488827, + "grad_norm": 1.9832940101623535, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7629152536392212, + "num_tokens": 560324309.0, + "step": 22292 + }, + { + "epoch": 1.838294714273934, + "grad_norm": 1.8721494674682617, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7641668915748596, + "num_tokens": 560355848.0, + "step": 22293 + }, + { + "epoch": 1.8383771748989859, + "grad_norm": 2.0443711280822754, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.7911955714225769, + "num_tokens": 560381677.0, + "step": 22294 + }, + { + "epoch": 1.8384596355240372, + "grad_norm": 1.99737548828125, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.7828739285469055, + "num_tokens": 560411145.0, + "step": 22295 + }, + { + "epoch": 1.8385420961490888, + "grad_norm": 2.0032730102539062, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7573256492614746, + "num_tokens": 560442690.0, + "step": 22296 + }, + { + "epoch": 1.8386245567741404, + "grad_norm": 2.1278560161590576, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7742444276809692, + "num_tokens": 560468789.0, + "step": 22297 + }, + { + "epoch": 1.8387070173991917, + "grad_norm": 2.0588161945343018, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7346839904785156, + "num_tokens": 560497543.0, + "step": 22298 + }, + { + "epoch": 1.8387894780242435, + "grad_norm": 1.7155156135559082, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.7982563972473145, + "num_tokens": 560535014.0, + "step": 22299 + }, + { + "epoch": 1.8388719386492949, + "grad_norm": 1.9664204120635986, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7407665252685547, + "num_tokens": 560567187.0, + "step": 22300 + }, + { + "epoch": 1.8389543992743465, + "grad_norm": 2.0543556213378906, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7386479377746582, + "num_tokens": 560593281.0, + "step": 22301 + }, + { + "epoch": 1.839036859899398, + "grad_norm": 1.8556269407272339, + "learning_rate": 1e-06, + "loss": 0.5656, + "mean_token_accuracy": 0.8079990148544312, + "num_tokens": 560629188.0, + "step": 22302 + }, + { + "epoch": 1.8391193205244496, + "grad_norm": 1.8779410123825073, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7545502185821533, + "num_tokens": 560659953.0, + "step": 22303 + }, + { + "epoch": 1.8392017811495012, + "grad_norm": 2.158046007156372, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7409970164299011, + "num_tokens": 560687655.0, + "step": 22304 + }, + { + "epoch": 1.8392842417745525, + "grad_norm": 2.196774482727051, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7274237871170044, + "num_tokens": 560716544.0, + "step": 22305 + }, + { + "epoch": 1.8393667023996043, + "grad_norm": 2.1748993396759033, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7633158564567566, + "num_tokens": 560742165.0, + "step": 22306 + }, + { + "epoch": 1.8394491630246557, + "grad_norm": 2.0426361560821533, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.7672795057296753, + "num_tokens": 560767755.0, + "step": 22307 + }, + { + "epoch": 1.8395316236497072, + "grad_norm": 2.1767094135284424, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7484925985336304, + "num_tokens": 560796843.0, + "step": 22308 + }, + { + "epoch": 1.8396140842747588, + "grad_norm": 2.004207134246826, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7440809011459351, + "num_tokens": 560825697.0, + "step": 22309 + }, + { + "epoch": 1.8396965448998104, + "grad_norm": 2.1656408309936523, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7274426221847534, + "num_tokens": 560855193.0, + "step": 22310 + }, + { + "epoch": 1.839779005524862, + "grad_norm": 2.064365863800049, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.779180645942688, + "num_tokens": 560881391.0, + "step": 22311 + }, + { + "epoch": 1.8398614661499133, + "grad_norm": 2.0077245235443115, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7831577062606812, + "num_tokens": 560908397.0, + "step": 22312 + }, + { + "epoch": 1.839943926774965, + "grad_norm": 1.8992698192596436, + "learning_rate": 1e-06, + "loss": 0.5701, + "mean_token_accuracy": 0.8057848215103149, + "num_tokens": 560935624.0, + "step": 22313 + }, + { + "epoch": 1.8400263874000164, + "grad_norm": 1.918882966041565, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7768157720565796, + "num_tokens": 560968873.0, + "step": 22314 + }, + { + "epoch": 1.840108848025068, + "grad_norm": 2.0830435752868652, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7412506937980652, + "num_tokens": 560996607.0, + "step": 22315 + }, + { + "epoch": 1.8401913086501196, + "grad_norm": 1.9873582124710083, + "learning_rate": 1e-06, + "loss": 0.5649, + "mean_token_accuracy": 0.8064514398574829, + "num_tokens": 561024579.0, + "step": 22316 + }, + { + "epoch": 1.840273769275171, + "grad_norm": 1.8998488187789917, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7751145362854004, + "num_tokens": 561055043.0, + "step": 22317 + }, + { + "epoch": 1.8403562299002227, + "grad_norm": 2.0553243160247803, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.7887619733810425, + "num_tokens": 561081805.0, + "step": 22318 + }, + { + "epoch": 1.840438690525274, + "grad_norm": 1.737260103225708, + "learning_rate": 1e-06, + "loss": 0.5929, + "mean_token_accuracy": 0.8126130700111389, + "num_tokens": 561115213.0, + "step": 22319 + }, + { + "epoch": 1.8405211511503259, + "grad_norm": 1.8933048248291016, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7531899213790894, + "num_tokens": 561147516.0, + "step": 22320 + }, + { + "epoch": 1.8406036117753772, + "grad_norm": 2.478660821914673, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7295643091201782, + "num_tokens": 561170686.0, + "step": 22321 + }, + { + "epoch": 1.8406860724004288, + "grad_norm": 1.900787115097046, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7602920532226562, + "num_tokens": 561200074.0, + "step": 22322 + }, + { + "epoch": 1.8407685330254804, + "grad_norm": 2.0852739810943604, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7707536220550537, + "num_tokens": 561225028.0, + "step": 22323 + }, + { + "epoch": 1.8408509936505317, + "grad_norm": 1.8758083581924438, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.7931671142578125, + "num_tokens": 561255922.0, + "step": 22324 + }, + { + "epoch": 1.8409334542755835, + "grad_norm": 2.3804068565368652, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7797998785972595, + "num_tokens": 561277797.0, + "step": 22325 + }, + { + "epoch": 1.8410159149006349, + "grad_norm": 1.9959839582443237, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7799309492111206, + "num_tokens": 561307321.0, + "step": 22326 + }, + { + "epoch": 1.8410983755256864, + "grad_norm": 1.9912996292114258, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7315529584884644, + "num_tokens": 561337406.0, + "step": 22327 + }, + { + "epoch": 1.841180836150738, + "grad_norm": 1.7728972434997559, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7808113098144531, + "num_tokens": 561372012.0, + "step": 22328 + }, + { + "epoch": 1.8412632967757896, + "grad_norm": 2.1518354415893555, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7553536891937256, + "num_tokens": 561398289.0, + "step": 22329 + }, + { + "epoch": 1.8413457574008412, + "grad_norm": 2.2947065830230713, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7553226351737976, + "num_tokens": 561422558.0, + "step": 22330 + }, + { + "epoch": 1.8414282180258925, + "grad_norm": 2.3579373359680176, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7624150514602661, + "num_tokens": 561443624.0, + "step": 22331 + }, + { + "epoch": 1.8415106786509443, + "grad_norm": 1.9740866422653198, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7671068906784058, + "num_tokens": 561474558.0, + "step": 22332 + }, + { + "epoch": 1.8415931392759957, + "grad_norm": 1.750463604927063, + "learning_rate": 1e-06, + "loss": 0.5145, + "mean_token_accuracy": 0.8286988735198975, + "num_tokens": 561508798.0, + "step": 22333 + }, + { + "epoch": 1.8416755999010472, + "grad_norm": 1.7569580078125, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.7849335670471191, + "num_tokens": 561544147.0, + "step": 22334 + }, + { + "epoch": 1.8417580605260988, + "grad_norm": 2.20794415473938, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7546197175979614, + "num_tokens": 561570433.0, + "step": 22335 + }, + { + "epoch": 1.8418405211511504, + "grad_norm": 2.0250022411346436, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7653779983520508, + "num_tokens": 561602412.0, + "step": 22336 + }, + { + "epoch": 1.841922981776202, + "grad_norm": 2.1384470462799072, + "learning_rate": 1e-06, + "loss": 0.5979, + "mean_token_accuracy": 0.7900901436805725, + "num_tokens": 561628405.0, + "step": 22337 + }, + { + "epoch": 1.8420054424012533, + "grad_norm": 1.9973852634429932, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7489922642707825, + "num_tokens": 561658495.0, + "step": 22338 + }, + { + "epoch": 1.842087903026305, + "grad_norm": 2.0638530254364014, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.733278751373291, + "num_tokens": 561686505.0, + "step": 22339 + }, + { + "epoch": 1.8421703636513564, + "grad_norm": 1.9181917905807495, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7652941346168518, + "num_tokens": 561718788.0, + "step": 22340 + }, + { + "epoch": 1.842252824276408, + "grad_norm": 1.9487708806991577, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7671741843223572, + "num_tokens": 561747949.0, + "step": 22341 + }, + { + "epoch": 1.8423352849014596, + "grad_norm": 1.8331276178359985, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7390925884246826, + "num_tokens": 561783591.0, + "step": 22342 + }, + { + "epoch": 1.842417745526511, + "grad_norm": 1.93648362159729, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7706956267356873, + "num_tokens": 561817250.0, + "step": 22343 + }, + { + "epoch": 1.8425002061515627, + "grad_norm": 1.8799619674682617, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7599796056747437, + "num_tokens": 561849114.0, + "step": 22344 + }, + { + "epoch": 1.842582666776614, + "grad_norm": 1.9372671842575073, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7536949515342712, + "num_tokens": 561880550.0, + "step": 22345 + }, + { + "epoch": 1.8426651274016657, + "grad_norm": 1.987178087234497, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.7883340120315552, + "num_tokens": 561908006.0, + "step": 22346 + }, + { + "epoch": 1.8427475880267172, + "grad_norm": 2.0224015712738037, + "learning_rate": 1e-06, + "loss": 0.5762, + "mean_token_accuracy": 0.7946810722351074, + "num_tokens": 561934899.0, + "step": 22347 + }, + { + "epoch": 1.8428300486517688, + "grad_norm": 2.0598621368408203, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7529882788658142, + "num_tokens": 561965778.0, + "step": 22348 + }, + { + "epoch": 1.8429125092768204, + "grad_norm": 2.0798251628875732, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7252659201622009, + "num_tokens": 561995739.0, + "step": 22349 + }, + { + "epoch": 1.8429949699018717, + "grad_norm": 2.166323184967041, + "learning_rate": 1e-06, + "loss": 0.5782, + "mean_token_accuracy": 0.8024106621742249, + "num_tokens": 562021375.0, + "step": 22350 + }, + { + "epoch": 1.8430774305269235, + "grad_norm": 2.2750086784362793, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.7854313254356384, + "num_tokens": 562044586.0, + "step": 22351 + }, + { + "epoch": 1.8431598911519749, + "grad_norm": 2.4369826316833496, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7672475576400757, + "num_tokens": 562065453.0, + "step": 22352 + }, + { + "epoch": 1.8432423517770264, + "grad_norm": 2.3587515354156494, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7401174306869507, + "num_tokens": 562087513.0, + "step": 22353 + }, + { + "epoch": 1.843324812402078, + "grad_norm": 2.2193691730499268, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7305077314376831, + "num_tokens": 562113737.0, + "step": 22354 + }, + { + "epoch": 1.8434072730271296, + "grad_norm": 1.993158221244812, + "learning_rate": 1e-06, + "loss": 0.5959, + "mean_token_accuracy": 0.8015658259391785, + "num_tokens": 562140294.0, + "step": 22355 + }, + { + "epoch": 1.8434897336521812, + "grad_norm": 1.7967956066131592, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7475447058677673, + "num_tokens": 562176478.0, + "step": 22356 + }, + { + "epoch": 1.8435721942772325, + "grad_norm": 2.000119924545288, + "learning_rate": 1e-06, + "loss": 0.7677, + "mean_token_accuracy": 0.752261757850647, + "num_tokens": 562204746.0, + "step": 22357 + }, + { + "epoch": 1.8436546549022843, + "grad_norm": 2.208035707473755, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7430284023284912, + "num_tokens": 562230256.0, + "step": 22358 + }, + { + "epoch": 1.8437371155273357, + "grad_norm": 2.0020155906677246, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7187641859054565, + "num_tokens": 562263016.0, + "step": 22359 + }, + { + "epoch": 1.8438195761523872, + "grad_norm": 2.026674747467041, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7467741966247559, + "num_tokens": 562291423.0, + "step": 22360 + }, + { + "epoch": 1.8439020367774388, + "grad_norm": 1.9248740673065186, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7739901542663574, + "num_tokens": 562325301.0, + "step": 22361 + }, + { + "epoch": 1.8439844974024902, + "grad_norm": 2.034186601638794, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7477380633354187, + "num_tokens": 562355353.0, + "step": 22362 + }, + { + "epoch": 1.844066958027542, + "grad_norm": 1.9094903469085693, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7783215045928955, + "num_tokens": 562385553.0, + "step": 22363 + }, + { + "epoch": 1.8441494186525933, + "grad_norm": 2.1163551807403564, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7262072563171387, + "num_tokens": 562412514.0, + "step": 22364 + }, + { + "epoch": 1.8442318792776449, + "grad_norm": 2.0260722637176514, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7387300729751587, + "num_tokens": 562444314.0, + "step": 22365 + }, + { + "epoch": 1.8443143399026964, + "grad_norm": 1.8500239849090576, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7742599248886108, + "num_tokens": 562478338.0, + "step": 22366 + }, + { + "epoch": 1.844396800527748, + "grad_norm": 1.9002301692962646, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7603992223739624, + "num_tokens": 562509851.0, + "step": 22367 + }, + { + "epoch": 1.8444792611527996, + "grad_norm": 2.2521581649780273, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7330076694488525, + "num_tokens": 562535910.0, + "step": 22368 + }, + { + "epoch": 1.844561721777851, + "grad_norm": 2.156850576400757, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.7785258293151855, + "num_tokens": 562562147.0, + "step": 22369 + }, + { + "epoch": 1.8446441824029027, + "grad_norm": 1.7862472534179688, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7763245105743408, + "num_tokens": 562594639.0, + "step": 22370 + }, + { + "epoch": 1.844726643027954, + "grad_norm": 1.7321224212646484, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7704975008964539, + "num_tokens": 562631432.0, + "step": 22371 + }, + { + "epoch": 1.8448091036530057, + "grad_norm": 1.8439487218856812, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7908006906509399, + "num_tokens": 562664325.0, + "step": 22372 + }, + { + "epoch": 1.8448915642780572, + "grad_norm": 1.791520595550537, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7658969163894653, + "num_tokens": 562697002.0, + "step": 22373 + }, + { + "epoch": 1.8449740249031088, + "grad_norm": 2.042604684829712, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7318145036697388, + "num_tokens": 562728193.0, + "step": 22374 + }, + { + "epoch": 1.8450564855281604, + "grad_norm": 2.0777862071990967, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7583423852920532, + "num_tokens": 562757064.0, + "step": 22375 + }, + { + "epoch": 1.8451389461532117, + "grad_norm": 1.7558817863464355, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7651875019073486, + "num_tokens": 562792509.0, + "step": 22376 + }, + { + "epoch": 1.8452214067782635, + "grad_norm": 2.05389142036438, + "learning_rate": 1e-06, + "loss": 0.5429, + "mean_token_accuracy": 0.8165265321731567, + "num_tokens": 562817402.0, + "step": 22377 + }, + { + "epoch": 1.8453038674033149, + "grad_norm": 2.582947015762329, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7526666522026062, + "num_tokens": 562839032.0, + "step": 22378 + }, + { + "epoch": 1.8453863280283664, + "grad_norm": 2.1549720764160156, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.7906550168991089, + "num_tokens": 562864320.0, + "step": 22379 + }, + { + "epoch": 1.845468788653418, + "grad_norm": 2.1289851665496826, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.7798261046409607, + "num_tokens": 562889845.0, + "step": 22380 + }, + { + "epoch": 1.8455512492784696, + "grad_norm": 1.9161380529403687, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7086005210876465, + "num_tokens": 562921108.0, + "step": 22381 + }, + { + "epoch": 1.8456337099035212, + "grad_norm": 2.279082775115967, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7219476103782654, + "num_tokens": 562945649.0, + "step": 22382 + }, + { + "epoch": 1.8457161705285725, + "grad_norm": 1.9217357635498047, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.783748984336853, + "num_tokens": 562978782.0, + "step": 22383 + }, + { + "epoch": 1.8457986311536243, + "grad_norm": 2.149233341217041, + "learning_rate": 1e-06, + "loss": 0.6096, + "mean_token_accuracy": 0.7772445678710938, + "num_tokens": 563004135.0, + "step": 22384 + }, + { + "epoch": 1.8458810917786757, + "grad_norm": 1.9471666812896729, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.787434458732605, + "num_tokens": 563036288.0, + "step": 22385 + }, + { + "epoch": 1.8459635524037272, + "grad_norm": 1.939184546470642, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7829960584640503, + "num_tokens": 563064713.0, + "step": 22386 + }, + { + "epoch": 1.8460460130287788, + "grad_norm": 1.9783165454864502, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7461934089660645, + "num_tokens": 563098514.0, + "step": 22387 + }, + { + "epoch": 1.8461284736538301, + "grad_norm": 2.1087026596069336, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7581372261047363, + "num_tokens": 563125692.0, + "step": 22388 + }, + { + "epoch": 1.846210934278882, + "grad_norm": 1.8933160305023193, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7834689617156982, + "num_tokens": 563157490.0, + "step": 22389 + }, + { + "epoch": 1.8462933949039333, + "grad_norm": 1.9393197298049927, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7553107738494873, + "num_tokens": 563190524.0, + "step": 22390 + }, + { + "epoch": 1.8463758555289849, + "grad_norm": 1.8607326745986938, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7458226084709167, + "num_tokens": 563226264.0, + "step": 22391 + }, + { + "epoch": 1.8464583161540364, + "grad_norm": 2.0210046768188477, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7360915541648865, + "num_tokens": 563255780.0, + "step": 22392 + }, + { + "epoch": 1.846540776779088, + "grad_norm": 2.057976245880127, + "learning_rate": 1e-06, + "loss": 0.615, + "mean_token_accuracy": 0.789974570274353, + "num_tokens": 563285935.0, + "step": 22393 + }, + { + "epoch": 1.8466232374041396, + "grad_norm": 2.0452404022216797, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7806760668754578, + "num_tokens": 563312517.0, + "step": 22394 + }, + { + "epoch": 1.846705698029191, + "grad_norm": 1.7737411260604858, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.7940407991409302, + "num_tokens": 563346171.0, + "step": 22395 + }, + { + "epoch": 1.8467881586542427, + "grad_norm": 2.0459887981414795, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7531299591064453, + "num_tokens": 563374188.0, + "step": 22396 + }, + { + "epoch": 1.846870619279294, + "grad_norm": 1.905752420425415, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7320482134819031, + "num_tokens": 563407993.0, + "step": 22397 + }, + { + "epoch": 1.8469530799043457, + "grad_norm": 1.7408901453018188, + "learning_rate": 1e-06, + "loss": 0.6103, + "mean_token_accuracy": 0.7965105175971985, + "num_tokens": 563443403.0, + "step": 22398 + }, + { + "epoch": 1.8470355405293972, + "grad_norm": 1.9640069007873535, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.7945099472999573, + "num_tokens": 563473542.0, + "step": 22399 + }, + { + "epoch": 1.8471180011544488, + "grad_norm": 2.0849435329437256, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7736867666244507, + "num_tokens": 563501326.0, + "step": 22400 + }, + { + "epoch": 1.8472004617795004, + "grad_norm": 1.9293841123580933, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7494592070579529, + "num_tokens": 563533388.0, + "step": 22401 + }, + { + "epoch": 1.8472829224045517, + "grad_norm": 1.9423327445983887, + "learning_rate": 1e-06, + "loss": 0.6022, + "mean_token_accuracy": 0.793269157409668, + "num_tokens": 563565739.0, + "step": 22402 + }, + { + "epoch": 1.8473653830296035, + "grad_norm": 1.7506988048553467, + "learning_rate": 1e-06, + "loss": 0.5802, + "mean_token_accuracy": 0.8213492631912231, + "num_tokens": 563601284.0, + "step": 22403 + }, + { + "epoch": 1.8474478436546549, + "grad_norm": 1.950643539428711, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7670102715492249, + "num_tokens": 563633125.0, + "step": 22404 + }, + { + "epoch": 1.8475303042797064, + "grad_norm": 2.015019655227661, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.729156494140625, + "num_tokens": 563663426.0, + "step": 22405 + }, + { + "epoch": 1.847612764904758, + "grad_norm": 1.726098656654358, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.774376630783081, + "num_tokens": 563699160.0, + "step": 22406 + }, + { + "epoch": 1.8476952255298094, + "grad_norm": 1.878139853477478, + "learning_rate": 1e-06, + "loss": 0.5284, + "mean_token_accuracy": 0.8092482686042786, + "num_tokens": 563729195.0, + "step": 22407 + }, + { + "epoch": 1.8477776861548612, + "grad_norm": 1.8758583068847656, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7692224979400635, + "num_tokens": 563762588.0, + "step": 22408 + }, + { + "epoch": 1.8478601467799125, + "grad_norm": 2.3904378414154053, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7296106815338135, + "num_tokens": 563784365.0, + "step": 22409 + }, + { + "epoch": 1.847942607404964, + "grad_norm": 2.533696174621582, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7315633893013, + "num_tokens": 563805958.0, + "step": 22410 + }, + { + "epoch": 1.8480250680300156, + "grad_norm": 1.771012306213379, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7741707563400269, + "num_tokens": 563843968.0, + "step": 22411 + }, + { + "epoch": 1.8481075286550672, + "grad_norm": 1.9746021032333374, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7563356757164001, + "num_tokens": 563874791.0, + "step": 22412 + }, + { + "epoch": 1.8481899892801188, + "grad_norm": 2.0861401557922363, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7291394472122192, + "num_tokens": 563903588.0, + "step": 22413 + }, + { + "epoch": 1.8482724499051701, + "grad_norm": 1.7631891965866089, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7395651340484619, + "num_tokens": 563937701.0, + "step": 22414 + }, + { + "epoch": 1.848354910530222, + "grad_norm": 1.801371693611145, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.7761801481246948, + "num_tokens": 563971018.0, + "step": 22415 + }, + { + "epoch": 1.8484373711552733, + "grad_norm": 2.427398443222046, + "learning_rate": 1e-06, + "loss": 0.8474, + "mean_token_accuracy": 0.7269605994224548, + "num_tokens": 563993665.0, + "step": 22416 + }, + { + "epoch": 1.8485198317803249, + "grad_norm": 2.121457815170288, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7588889002799988, + "num_tokens": 564022320.0, + "step": 22417 + }, + { + "epoch": 1.8486022924053764, + "grad_norm": 2.343360424041748, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7377383708953857, + "num_tokens": 564048226.0, + "step": 22418 + }, + { + "epoch": 1.848684753030428, + "grad_norm": 1.8320547342300415, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.7826445698738098, + "num_tokens": 564082281.0, + "step": 22419 + }, + { + "epoch": 1.8487672136554796, + "grad_norm": 2.2861275672912598, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7391153573989868, + "num_tokens": 564107499.0, + "step": 22420 + }, + { + "epoch": 1.848849674280531, + "grad_norm": 1.8918483257293701, + "learning_rate": 1e-06, + "loss": 0.5551, + "mean_token_accuracy": 0.8135380744934082, + "num_tokens": 564137509.0, + "step": 22421 + }, + { + "epoch": 1.8489321349055827, + "grad_norm": 1.8701667785644531, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7687958478927612, + "num_tokens": 564170242.0, + "step": 22422 + }, + { + "epoch": 1.849014595530634, + "grad_norm": 1.9796448945999146, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7663413286209106, + "num_tokens": 564200848.0, + "step": 22423 + }, + { + "epoch": 1.8490970561556856, + "grad_norm": 2.1054277420043945, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7734053730964661, + "num_tokens": 564226717.0, + "step": 22424 + }, + { + "epoch": 1.8491795167807372, + "grad_norm": 1.7954609394073486, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7565445303916931, + "num_tokens": 564261049.0, + "step": 22425 + }, + { + "epoch": 1.8492619774057886, + "grad_norm": 2.082134485244751, + "learning_rate": 1e-06, + "loss": 0.7775, + "mean_token_accuracy": 0.7584450244903564, + "num_tokens": 564290510.0, + "step": 22426 + }, + { + "epoch": 1.8493444380308404, + "grad_norm": 2.016310691833496, + "learning_rate": 1e-06, + "loss": 0.5769, + "mean_token_accuracy": 0.8052538633346558, + "num_tokens": 564316295.0, + "step": 22427 + }, + { + "epoch": 1.8494268986558917, + "grad_norm": 2.073042154312134, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7749202251434326, + "num_tokens": 564346271.0, + "step": 22428 + }, + { + "epoch": 1.8495093592809435, + "grad_norm": 1.8840172290802002, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.7977941632270813, + "num_tokens": 564375185.0, + "step": 22429 + }, + { + "epoch": 1.8495918199059949, + "grad_norm": 2.050607442855835, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7306494116783142, + "num_tokens": 564408070.0, + "step": 22430 + }, + { + "epoch": 1.8496742805310464, + "grad_norm": 2.115476131439209, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7150938510894775, + "num_tokens": 564436475.0, + "step": 22431 + }, + { + "epoch": 1.849756741156098, + "grad_norm": 1.9368863105773926, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7376944422721863, + "num_tokens": 564468023.0, + "step": 22432 + }, + { + "epoch": 1.8498392017811494, + "grad_norm": 2.400562286376953, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7386533617973328, + "num_tokens": 564492714.0, + "step": 22433 + }, + { + "epoch": 1.8499216624062012, + "grad_norm": 2.0811541080474854, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7678553462028503, + "num_tokens": 564522146.0, + "step": 22434 + }, + { + "epoch": 1.8500041230312525, + "grad_norm": 1.9697455167770386, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7687293291091919, + "num_tokens": 564552011.0, + "step": 22435 + }, + { + "epoch": 1.850086583656304, + "grad_norm": 2.001190185546875, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7739953994750977, + "num_tokens": 564585031.0, + "step": 22436 + }, + { + "epoch": 1.8501690442813556, + "grad_norm": 2.189483642578125, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7611756324768066, + "num_tokens": 564610217.0, + "step": 22437 + }, + { + "epoch": 1.8502515049064072, + "grad_norm": 2.172968626022339, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7390131950378418, + "num_tokens": 564637462.0, + "step": 22438 + }, + { + "epoch": 1.8503339655314588, + "grad_norm": 2.1348788738250732, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7363758087158203, + "num_tokens": 564663687.0, + "step": 22439 + }, + { + "epoch": 1.8504164261565101, + "grad_norm": 1.9608465433120728, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.7897915840148926, + "num_tokens": 564692736.0, + "step": 22440 + }, + { + "epoch": 1.850498886781562, + "grad_norm": 2.1008808612823486, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7446873188018799, + "num_tokens": 564721006.0, + "step": 22441 + }, + { + "epoch": 1.8505813474066133, + "grad_norm": 1.9423506259918213, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7756521105766296, + "num_tokens": 564752407.0, + "step": 22442 + }, + { + "epoch": 1.8506638080316649, + "grad_norm": 2.1868724822998047, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.737082839012146, + "num_tokens": 564776595.0, + "step": 22443 + }, + { + "epoch": 1.8507462686567164, + "grad_norm": 2.2102386951446533, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7024095058441162, + "num_tokens": 564800742.0, + "step": 22444 + }, + { + "epoch": 1.850828729281768, + "grad_norm": 1.800397276878357, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7638359069824219, + "num_tokens": 564834124.0, + "step": 22445 + }, + { + "epoch": 1.8509111899068196, + "grad_norm": 1.8472968339920044, + "learning_rate": 1e-06, + "loss": 0.524, + "mean_token_accuracy": 0.8014307022094727, + "num_tokens": 564865764.0, + "step": 22446 + }, + { + "epoch": 1.850993650531871, + "grad_norm": 2.059382438659668, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7690610885620117, + "num_tokens": 564891826.0, + "step": 22447 + }, + { + "epoch": 1.8510761111569227, + "grad_norm": 1.8925772905349731, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.739645779132843, + "num_tokens": 564925081.0, + "step": 22448 + }, + { + "epoch": 1.851158571781974, + "grad_norm": 1.9858919382095337, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7468578815460205, + "num_tokens": 564957476.0, + "step": 22449 + }, + { + "epoch": 1.8512410324070256, + "grad_norm": 2.008389711380005, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7646976113319397, + "num_tokens": 564985787.0, + "step": 22450 + }, + { + "epoch": 1.8513234930320772, + "grad_norm": 1.7759276628494263, + "learning_rate": 1e-06, + "loss": 0.6077, + "mean_token_accuracy": 0.8015583753585815, + "num_tokens": 565024386.0, + "step": 22451 + }, + { + "epoch": 1.8514059536571286, + "grad_norm": 2.0742011070251465, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.7856767177581787, + "num_tokens": 565049165.0, + "step": 22452 + }, + { + "epoch": 1.8514884142821804, + "grad_norm": 2.1092426776885986, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.744550347328186, + "num_tokens": 565077875.0, + "step": 22453 + }, + { + "epoch": 1.8515708749072317, + "grad_norm": 2.184927225112915, + "learning_rate": 1e-06, + "loss": 0.6096, + "mean_token_accuracy": 0.7968301177024841, + "num_tokens": 565101414.0, + "step": 22454 + }, + { + "epoch": 1.8516533355322833, + "grad_norm": 1.968523621559143, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7764323949813843, + "num_tokens": 565133822.0, + "step": 22455 + }, + { + "epoch": 1.8517357961573349, + "grad_norm": 1.7383427619934082, + "learning_rate": 1e-06, + "loss": 0.5881, + "mean_token_accuracy": 0.8031498193740845, + "num_tokens": 565167004.0, + "step": 22456 + }, + { + "epoch": 1.8518182567823864, + "grad_norm": 2.03633713722229, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7435382604598999, + "num_tokens": 565196376.0, + "step": 22457 + }, + { + "epoch": 1.851900717407438, + "grad_norm": 1.764358401298523, + "learning_rate": 1e-06, + "loss": 0.5955, + "mean_token_accuracy": 0.8003413677215576, + "num_tokens": 565229311.0, + "step": 22458 + }, + { + "epoch": 1.8519831780324894, + "grad_norm": 1.8916330337524414, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7490664720535278, + "num_tokens": 565264017.0, + "step": 22459 + }, + { + "epoch": 1.8520656386575411, + "grad_norm": 2.044917583465576, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.7813585996627808, + "num_tokens": 565289552.0, + "step": 22460 + }, + { + "epoch": 1.8521480992825925, + "grad_norm": 1.9345977306365967, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7393602132797241, + "num_tokens": 565318770.0, + "step": 22461 + }, + { + "epoch": 1.852230559907644, + "grad_norm": 2.4817099571228027, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7286181449890137, + "num_tokens": 565340086.0, + "step": 22462 + }, + { + "epoch": 1.8523130205326956, + "grad_norm": 1.769498586654663, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7580616474151611, + "num_tokens": 565378393.0, + "step": 22463 + }, + { + "epoch": 1.8523954811577472, + "grad_norm": 1.9115248918533325, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.7772582173347473, + "num_tokens": 565408172.0, + "step": 22464 + }, + { + "epoch": 1.8524779417827988, + "grad_norm": 1.9146397113800049, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.7874232530593872, + "num_tokens": 565436788.0, + "step": 22465 + }, + { + "epoch": 1.8525604024078501, + "grad_norm": 1.8954627513885498, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7716378569602966, + "num_tokens": 565466464.0, + "step": 22466 + }, + { + "epoch": 1.852642863032902, + "grad_norm": 1.9647538661956787, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7684003114700317, + "num_tokens": 565496342.0, + "step": 22467 + }, + { + "epoch": 1.8527253236579533, + "grad_norm": 1.8733611106872559, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.7793967127799988, + "num_tokens": 565526168.0, + "step": 22468 + }, + { + "epoch": 1.8528077842830049, + "grad_norm": 2.3682641983032227, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7806170582771301, + "num_tokens": 565546761.0, + "step": 22469 + }, + { + "epoch": 1.8528902449080564, + "grad_norm": 1.99190354347229, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7591668367385864, + "num_tokens": 565575175.0, + "step": 22470 + }, + { + "epoch": 1.8529727055331078, + "grad_norm": 2.048238515853882, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7668420076370239, + "num_tokens": 565600819.0, + "step": 22471 + }, + { + "epoch": 1.8530551661581596, + "grad_norm": 2.083423614501953, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7801798582077026, + "num_tokens": 565628142.0, + "step": 22472 + }, + { + "epoch": 1.853137626783211, + "grad_norm": 2.100276470184326, + "learning_rate": 1e-06, + "loss": 0.6221, + "mean_token_accuracy": 0.7986146807670593, + "num_tokens": 565654739.0, + "step": 22473 + }, + { + "epoch": 1.8532200874082627, + "grad_norm": 2.1665213108062744, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.722230076789856, + "num_tokens": 565682835.0, + "step": 22474 + }, + { + "epoch": 1.853302548033314, + "grad_norm": 2.029874563217163, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7706165313720703, + "num_tokens": 565710046.0, + "step": 22475 + }, + { + "epoch": 1.8533850086583656, + "grad_norm": 2.039250135421753, + "learning_rate": 1e-06, + "loss": 0.615, + "mean_token_accuracy": 0.7826148271560669, + "num_tokens": 565738041.0, + "step": 22476 + }, + { + "epoch": 1.8534674692834172, + "grad_norm": 1.8286949396133423, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7478213310241699, + "num_tokens": 565775746.0, + "step": 22477 + }, + { + "epoch": 1.8535499299084686, + "grad_norm": 1.8185482025146484, + "learning_rate": 1e-06, + "loss": 0.6089, + "mean_token_accuracy": 0.8015832901000977, + "num_tokens": 565807487.0, + "step": 22478 + }, + { + "epoch": 1.8536323905335204, + "grad_norm": 2.0806829929351807, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7194645404815674, + "num_tokens": 565836662.0, + "step": 22479 + }, + { + "epoch": 1.8537148511585717, + "grad_norm": 1.9531902074813843, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.786193311214447, + "num_tokens": 565865624.0, + "step": 22480 + }, + { + "epoch": 1.8537973117836233, + "grad_norm": 2.214076042175293, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7672914266586304, + "num_tokens": 565890048.0, + "step": 22481 + }, + { + "epoch": 1.8538797724086749, + "grad_norm": 2.1535792350769043, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7558603286743164, + "num_tokens": 565915914.0, + "step": 22482 + }, + { + "epoch": 1.8539622330337264, + "grad_norm": 1.9294754266738892, + "learning_rate": 1e-06, + "loss": 0.7799, + "mean_token_accuracy": 0.7594863176345825, + "num_tokens": 565948407.0, + "step": 22483 + }, + { + "epoch": 1.854044693658778, + "grad_norm": 1.8085674047470093, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.7884592413902283, + "num_tokens": 565979999.0, + "step": 22484 + }, + { + "epoch": 1.8541271542838293, + "grad_norm": 1.8823121786117554, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.7866291403770447, + "num_tokens": 566012940.0, + "step": 22485 + }, + { + "epoch": 1.8542096149088811, + "grad_norm": 1.9716525077819824, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7790138721466064, + "num_tokens": 566042210.0, + "step": 22486 + }, + { + "epoch": 1.8542920755339325, + "grad_norm": 2.1644341945648193, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7246070504188538, + "num_tokens": 566071985.0, + "step": 22487 + }, + { + "epoch": 1.854374536158984, + "grad_norm": 1.879120111465454, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7130653262138367, + "num_tokens": 566104658.0, + "step": 22488 + }, + { + "epoch": 1.8544569967840356, + "grad_norm": 1.7391197681427002, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7604526281356812, + "num_tokens": 566140220.0, + "step": 22489 + }, + { + "epoch": 1.8545394574090872, + "grad_norm": 2.234865427017212, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7584925889968872, + "num_tokens": 566162219.0, + "step": 22490 + }, + { + "epoch": 1.8546219180341388, + "grad_norm": 1.980895757675171, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.7931921482086182, + "num_tokens": 566192232.0, + "step": 22491 + }, + { + "epoch": 1.8547043786591901, + "grad_norm": 1.9021679162979126, + "learning_rate": 1e-06, + "loss": 0.6165, + "mean_token_accuracy": 0.7976300716400146, + "num_tokens": 566221265.0, + "step": 22492 + }, + { + "epoch": 1.854786839284242, + "grad_norm": 1.7955626249313354, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8018492460250854, + "num_tokens": 566253202.0, + "step": 22493 + }, + { + "epoch": 1.8548692999092933, + "grad_norm": 2.1096012592315674, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.7776356339454651, + "num_tokens": 566279096.0, + "step": 22494 + }, + { + "epoch": 1.8549517605343449, + "grad_norm": 1.794035792350769, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7330070734024048, + "num_tokens": 566312775.0, + "step": 22495 + }, + { + "epoch": 1.8550342211593964, + "grad_norm": 1.9771482944488525, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.7854695320129395, + "num_tokens": 566343537.0, + "step": 22496 + }, + { + "epoch": 1.8551166817844478, + "grad_norm": 2.190531015396118, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7647290825843811, + "num_tokens": 566370845.0, + "step": 22497 + }, + { + "epoch": 1.8551991424094996, + "grad_norm": 2.027496337890625, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.751266360282898, + "num_tokens": 566400477.0, + "step": 22498 + }, + { + "epoch": 1.855281603034551, + "grad_norm": 2.1810994148254395, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7645797729492188, + "num_tokens": 566426312.0, + "step": 22499 + }, + { + "epoch": 1.8553640636596025, + "grad_norm": 1.8570830821990967, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7372447848320007, + "num_tokens": 566463405.0, + "step": 22500 + }, + { + "epoch": 1.855446524284654, + "grad_norm": 2.3927128314971924, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7044087648391724, + "num_tokens": 566487196.0, + "step": 22501 + }, + { + "epoch": 1.8555289849097056, + "grad_norm": 1.8076246976852417, + "learning_rate": 1e-06, + "loss": 0.5507, + "mean_token_accuracy": 0.8102825284004211, + "num_tokens": 566519109.0, + "step": 22502 + }, + { + "epoch": 1.8556114455347572, + "grad_norm": 1.7131520509719849, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7831701636314392, + "num_tokens": 566557268.0, + "step": 22503 + }, + { + "epoch": 1.8556939061598086, + "grad_norm": 2.2195699214935303, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7737054824829102, + "num_tokens": 566581822.0, + "step": 22504 + }, + { + "epoch": 1.8557763667848604, + "grad_norm": 1.9683221578598022, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7776983976364136, + "num_tokens": 566610241.0, + "step": 22505 + }, + { + "epoch": 1.8558588274099117, + "grad_norm": 1.7170097827911377, + "learning_rate": 1e-06, + "loss": 0.5724, + "mean_token_accuracy": 0.8060193061828613, + "num_tokens": 566646839.0, + "step": 22506 + }, + { + "epoch": 1.8559412880349633, + "grad_norm": 1.974191427230835, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7397286295890808, + "num_tokens": 566676795.0, + "step": 22507 + }, + { + "epoch": 1.8560237486600148, + "grad_norm": 1.756186842918396, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7640681266784668, + "num_tokens": 566713882.0, + "step": 22508 + }, + { + "epoch": 1.8561062092850664, + "grad_norm": 1.9986777305603027, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7721849679946899, + "num_tokens": 566742178.0, + "step": 22509 + }, + { + "epoch": 1.856188669910118, + "grad_norm": 1.7677501440048218, + "learning_rate": 1e-06, + "loss": 0.602, + "mean_token_accuracy": 0.8068921566009521, + "num_tokens": 566776241.0, + "step": 22510 + }, + { + "epoch": 1.8562711305351693, + "grad_norm": 1.6919971704483032, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7548553347587585, + "num_tokens": 566816336.0, + "step": 22511 + }, + { + "epoch": 1.8563535911602211, + "grad_norm": 1.9560189247131348, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7564114928245544, + "num_tokens": 566844927.0, + "step": 22512 + }, + { + "epoch": 1.8564360517852725, + "grad_norm": 1.894029974937439, + "learning_rate": 1e-06, + "loss": 0.5864, + "mean_token_accuracy": 0.8045364618301392, + "num_tokens": 566877070.0, + "step": 22513 + }, + { + "epoch": 1.856518512410324, + "grad_norm": 1.8302557468414307, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7453486323356628, + "num_tokens": 566913132.0, + "step": 22514 + }, + { + "epoch": 1.8566009730353756, + "grad_norm": 1.9753819704055786, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.7509520053863525, + "num_tokens": 566946311.0, + "step": 22515 + }, + { + "epoch": 1.856683433660427, + "grad_norm": 1.8733035326004028, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7551646828651428, + "num_tokens": 566979972.0, + "step": 22516 + }, + { + "epoch": 1.8567658942854788, + "grad_norm": 2.1225502490997314, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7777438163757324, + "num_tokens": 567008624.0, + "step": 22517 + }, + { + "epoch": 1.8568483549105301, + "grad_norm": 2.195211410522461, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7586580514907837, + "num_tokens": 567034370.0, + "step": 22518 + }, + { + "epoch": 1.856930815535582, + "grad_norm": 1.9761011600494385, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7452256679534912, + "num_tokens": 567065053.0, + "step": 22519 + }, + { + "epoch": 1.8570132761606333, + "grad_norm": 2.144526481628418, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.7497107982635498, + "num_tokens": 567090962.0, + "step": 22520 + }, + { + "epoch": 1.8570957367856848, + "grad_norm": 2.0212562084198, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7477781176567078, + "num_tokens": 567122575.0, + "step": 22521 + }, + { + "epoch": 1.8571781974107364, + "grad_norm": 2.1720142364501953, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7191492915153503, + "num_tokens": 567148682.0, + "step": 22522 + }, + { + "epoch": 1.8572606580357878, + "grad_norm": 1.879187822341919, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7691652774810791, + "num_tokens": 567179406.0, + "step": 22523 + }, + { + "epoch": 1.8573431186608396, + "grad_norm": 2.442089080810547, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7492831945419312, + "num_tokens": 567201124.0, + "step": 22524 + }, + { + "epoch": 1.857425579285891, + "grad_norm": 2.117551565170288, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7636460065841675, + "num_tokens": 567227116.0, + "step": 22525 + }, + { + "epoch": 1.8575080399109425, + "grad_norm": 2.026782989501953, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7755530476570129, + "num_tokens": 567254219.0, + "step": 22526 + }, + { + "epoch": 1.857590500535994, + "grad_norm": 2.3989388942718506, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.6977083683013916, + "num_tokens": 567279244.0, + "step": 22527 + }, + { + "epoch": 1.8576729611610456, + "grad_norm": 1.9287947416305542, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7728661894798279, + "num_tokens": 567311099.0, + "step": 22528 + }, + { + "epoch": 1.8577554217860972, + "grad_norm": 1.918573021888733, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7615221738815308, + "num_tokens": 567344768.0, + "step": 22529 + }, + { + "epoch": 1.8578378824111486, + "grad_norm": 2.071894645690918, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7687753438949585, + "num_tokens": 567373181.0, + "step": 22530 + }, + { + "epoch": 1.8579203430362004, + "grad_norm": 1.8986940383911133, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7418556213378906, + "num_tokens": 567404708.0, + "step": 22531 + }, + { + "epoch": 1.8580028036612517, + "grad_norm": 1.8447273969650269, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7564520835876465, + "num_tokens": 567438564.0, + "step": 22532 + }, + { + "epoch": 1.8580852642863033, + "grad_norm": 1.9446743726730347, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7472583055496216, + "num_tokens": 567468578.0, + "step": 22533 + }, + { + "epoch": 1.8581677249113548, + "grad_norm": 2.0441105365753174, + "learning_rate": 1e-06, + "loss": 0.5896, + "mean_token_accuracy": 0.7871421575546265, + "num_tokens": 567494099.0, + "step": 22534 + }, + { + "epoch": 1.8582501855364064, + "grad_norm": 1.9779233932495117, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.7853440642356873, + "num_tokens": 567521416.0, + "step": 22535 + }, + { + "epoch": 1.858332646161458, + "grad_norm": 2.2477822303771973, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.762489378452301, + "num_tokens": 567544319.0, + "step": 22536 + }, + { + "epoch": 1.8584151067865093, + "grad_norm": 1.9391136169433594, + "learning_rate": 1e-06, + "loss": 0.5773, + "mean_token_accuracy": 0.8116455078125, + "num_tokens": 567572243.0, + "step": 22537 + }, + { + "epoch": 1.8584975674115611, + "grad_norm": 1.9735499620437622, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7727256417274475, + "num_tokens": 567602892.0, + "step": 22538 + }, + { + "epoch": 1.8585800280366125, + "grad_norm": 1.852355718612671, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7813248634338379, + "num_tokens": 567632345.0, + "step": 22539 + }, + { + "epoch": 1.858662488661664, + "grad_norm": 2.0327351093292236, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.7869367599487305, + "num_tokens": 567658739.0, + "step": 22540 + }, + { + "epoch": 1.8587449492867156, + "grad_norm": 1.9397205114364624, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7705879807472229, + "num_tokens": 567687439.0, + "step": 22541 + }, + { + "epoch": 1.858827409911767, + "grad_norm": 2.00783109664917, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.7827192544937134, + "num_tokens": 567714352.0, + "step": 22542 + }, + { + "epoch": 1.8589098705368188, + "grad_norm": 1.93148672580719, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7673149108886719, + "num_tokens": 567745707.0, + "step": 22543 + }, + { + "epoch": 1.8589923311618701, + "grad_norm": 1.8396739959716797, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7449228763580322, + "num_tokens": 567780647.0, + "step": 22544 + }, + { + "epoch": 1.8590747917869217, + "grad_norm": 1.892497181892395, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.7894424796104431, + "num_tokens": 567814443.0, + "step": 22545 + }, + { + "epoch": 1.8591572524119733, + "grad_norm": 2.013331651687622, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.7943946719169617, + "num_tokens": 567841054.0, + "step": 22546 + }, + { + "epoch": 1.8592397130370248, + "grad_norm": 1.7967885732650757, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7684663534164429, + "num_tokens": 567874883.0, + "step": 22547 + }, + { + "epoch": 1.8593221736620764, + "grad_norm": 1.932979941368103, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7144500017166138, + "num_tokens": 567909213.0, + "step": 22548 + }, + { + "epoch": 1.8594046342871278, + "grad_norm": 2.102830410003662, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7383438348770142, + "num_tokens": 567937717.0, + "step": 22549 + }, + { + "epoch": 1.8594870949121796, + "grad_norm": 1.8957847356796265, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.748031735420227, + "num_tokens": 567971398.0, + "step": 22550 + }, + { + "epoch": 1.859569555537231, + "grad_norm": 2.0950682163238525, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7639890313148499, + "num_tokens": 567997451.0, + "step": 22551 + }, + { + "epoch": 1.8596520161622825, + "grad_norm": 1.8266782760620117, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7527899742126465, + "num_tokens": 568033348.0, + "step": 22552 + }, + { + "epoch": 1.859734476787334, + "grad_norm": 2.042477607727051, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7518345713615417, + "num_tokens": 568060778.0, + "step": 22553 + }, + { + "epoch": 1.8598169374123856, + "grad_norm": 2.3491811752319336, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7699204087257385, + "num_tokens": 568080515.0, + "step": 22554 + }, + { + "epoch": 1.8598993980374372, + "grad_norm": 2.157954454421997, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7738981246948242, + "num_tokens": 568107654.0, + "step": 22555 + }, + { + "epoch": 1.8599818586624886, + "grad_norm": 1.8028086423873901, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7633415460586548, + "num_tokens": 568144493.0, + "step": 22556 + }, + { + "epoch": 1.8600643192875403, + "grad_norm": 1.8997671604156494, + "learning_rate": 1e-06, + "loss": 0.5888, + "mean_token_accuracy": 0.8014386892318726, + "num_tokens": 568174276.0, + "step": 22557 + }, + { + "epoch": 1.8601467799125917, + "grad_norm": 1.8539860248565674, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7840495705604553, + "num_tokens": 568202915.0, + "step": 22558 + }, + { + "epoch": 1.8602292405376433, + "grad_norm": 1.7801482677459717, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7710399627685547, + "num_tokens": 568238095.0, + "step": 22559 + }, + { + "epoch": 1.8603117011626948, + "grad_norm": 1.9036833047866821, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7638142108917236, + "num_tokens": 568271423.0, + "step": 22560 + }, + { + "epoch": 1.8603941617877462, + "grad_norm": 1.961652398109436, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7308179140090942, + "num_tokens": 568303255.0, + "step": 22561 + }, + { + "epoch": 1.860476622412798, + "grad_norm": 1.9015321731567383, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7719284296035767, + "num_tokens": 568336433.0, + "step": 22562 + }, + { + "epoch": 1.8605590830378493, + "grad_norm": 2.35435152053833, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7893532514572144, + "num_tokens": 568357133.0, + "step": 22563 + }, + { + "epoch": 1.860641543662901, + "grad_norm": 1.9519058465957642, + "learning_rate": 1e-06, + "loss": 0.594, + "mean_token_accuracy": 0.800602912902832, + "num_tokens": 568385252.0, + "step": 22564 + }, + { + "epoch": 1.8607240042879525, + "grad_norm": 1.8853299617767334, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.749031662940979, + "num_tokens": 568417509.0, + "step": 22565 + }, + { + "epoch": 1.860806464913004, + "grad_norm": 1.8778276443481445, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7500129342079163, + "num_tokens": 568451879.0, + "step": 22566 + }, + { + "epoch": 1.8608889255380556, + "grad_norm": 2.006531238555908, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7730711102485657, + "num_tokens": 568483044.0, + "step": 22567 + }, + { + "epoch": 1.860971386163107, + "grad_norm": 1.7680166959762573, + "learning_rate": 1e-06, + "loss": 0.5298, + "mean_token_accuracy": 0.8165653347969055, + "num_tokens": 568519634.0, + "step": 22568 + }, + { + "epoch": 1.8610538467881588, + "grad_norm": 2.210531234741211, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7803336381912231, + "num_tokens": 568543095.0, + "step": 22569 + }, + { + "epoch": 1.8611363074132101, + "grad_norm": 1.9063870906829834, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7609113454818726, + "num_tokens": 568577894.0, + "step": 22570 + }, + { + "epoch": 1.8612187680382617, + "grad_norm": 1.9377920627593994, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7419607043266296, + "num_tokens": 568609868.0, + "step": 22571 + }, + { + "epoch": 1.8613012286633133, + "grad_norm": 1.9279108047485352, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7687502503395081, + "num_tokens": 568642207.0, + "step": 22572 + }, + { + "epoch": 1.8613836892883648, + "grad_norm": 2.0122921466827393, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7666811347007751, + "num_tokens": 568671690.0, + "step": 22573 + }, + { + "epoch": 1.8614661499134164, + "grad_norm": 1.758199691772461, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7740690112113953, + "num_tokens": 568712760.0, + "step": 22574 + }, + { + "epoch": 1.8615486105384678, + "grad_norm": 2.036973476409912, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7779921293258667, + "num_tokens": 568740558.0, + "step": 22575 + }, + { + "epoch": 1.8616310711635196, + "grad_norm": 2.062587261199951, + "learning_rate": 1e-06, + "loss": 0.5961, + "mean_token_accuracy": 0.7911731004714966, + "num_tokens": 568768257.0, + "step": 22576 + }, + { + "epoch": 1.861713531788571, + "grad_norm": 2.0096535682678223, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7806879878044128, + "num_tokens": 568795634.0, + "step": 22577 + }, + { + "epoch": 1.8617959924136225, + "grad_norm": 1.9235942363739014, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7371024489402771, + "num_tokens": 568833362.0, + "step": 22578 + }, + { + "epoch": 1.861878453038674, + "grad_norm": 1.9305485486984253, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.7896009087562561, + "num_tokens": 568864025.0, + "step": 22579 + }, + { + "epoch": 1.8619609136637256, + "grad_norm": 1.8217031955718994, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7673063278198242, + "num_tokens": 568896120.0, + "step": 22580 + }, + { + "epoch": 1.8620433742887772, + "grad_norm": 2.3216474056243896, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7303533554077148, + "num_tokens": 568918897.0, + "step": 22581 + }, + { + "epoch": 1.8621258349138285, + "grad_norm": 1.8450603485107422, + "learning_rate": 1e-06, + "loss": 0.5935, + "mean_token_accuracy": 0.8052438497543335, + "num_tokens": 568948434.0, + "step": 22582 + }, + { + "epoch": 1.8622082955388803, + "grad_norm": 1.8021516799926758, + "learning_rate": 1e-06, + "loss": 0.5463, + "mean_token_accuracy": 0.8111196756362915, + "num_tokens": 568979815.0, + "step": 22583 + }, + { + "epoch": 1.8622907561639317, + "grad_norm": 1.8330533504486084, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.7817185521125793, + "num_tokens": 569015606.0, + "step": 22584 + }, + { + "epoch": 1.8623732167889833, + "grad_norm": 2.085275888442993, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7733936309814453, + "num_tokens": 569043134.0, + "step": 22585 + }, + { + "epoch": 1.8624556774140348, + "grad_norm": 1.949927806854248, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7882365584373474, + "num_tokens": 569070469.0, + "step": 22586 + }, + { + "epoch": 1.8625381380390862, + "grad_norm": 1.8627146482467651, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7358859181404114, + "num_tokens": 569103727.0, + "step": 22587 + }, + { + "epoch": 1.862620598664138, + "grad_norm": 1.6735119819641113, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.8059819936752319, + "num_tokens": 569142118.0, + "step": 22588 + }, + { + "epoch": 1.8627030592891893, + "grad_norm": 1.9501818418502808, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7617635130882263, + "num_tokens": 569176742.0, + "step": 22589 + }, + { + "epoch": 1.862785519914241, + "grad_norm": 1.8991295099258423, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7699408531188965, + "num_tokens": 569209061.0, + "step": 22590 + }, + { + "epoch": 1.8628679805392925, + "grad_norm": 1.7848074436187744, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7544284462928772, + "num_tokens": 569247956.0, + "step": 22591 + }, + { + "epoch": 1.862950441164344, + "grad_norm": 2.0147504806518555, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.787399172782898, + "num_tokens": 569275605.0, + "step": 22592 + }, + { + "epoch": 1.8630329017893956, + "grad_norm": 2.185896873474121, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.713686466217041, + "num_tokens": 569304975.0, + "step": 22593 + }, + { + "epoch": 1.863115362414447, + "grad_norm": 1.961721420288086, + "learning_rate": 1e-06, + "loss": 0.5936, + "mean_token_accuracy": 0.8026320934295654, + "num_tokens": 569338150.0, + "step": 22594 + }, + { + "epoch": 1.8631978230394988, + "grad_norm": 1.7591317892074585, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7888886332511902, + "num_tokens": 569374426.0, + "step": 22595 + }, + { + "epoch": 1.8632802836645501, + "grad_norm": 2.2895431518554688, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7422083616256714, + "num_tokens": 569399255.0, + "step": 22596 + }, + { + "epoch": 1.8633627442896017, + "grad_norm": 1.750291109085083, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.7866911888122559, + "num_tokens": 569434677.0, + "step": 22597 + }, + { + "epoch": 1.8634452049146533, + "grad_norm": 1.8850029706954956, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7732758522033691, + "num_tokens": 569467960.0, + "step": 22598 + }, + { + "epoch": 1.8635276655397048, + "grad_norm": 2.019178867340088, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7354437708854675, + "num_tokens": 569499783.0, + "step": 22599 + }, + { + "epoch": 1.8636101261647564, + "grad_norm": 2.0073111057281494, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7610545754432678, + "num_tokens": 569534701.0, + "step": 22600 + }, + { + "epoch": 1.8636925867898078, + "grad_norm": 2.656726837158203, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7650126218795776, + "num_tokens": 569554248.0, + "step": 22601 + }, + { + "epoch": 1.8637750474148596, + "grad_norm": 2.363917112350464, + "learning_rate": 1e-06, + "loss": 0.7767, + "mean_token_accuracy": 0.752609133720398, + "num_tokens": 569578461.0, + "step": 22602 + }, + { + "epoch": 1.863857508039911, + "grad_norm": 2.0721781253814697, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.6953748464584351, + "num_tokens": 569612592.0, + "step": 22603 + }, + { + "epoch": 1.8639399686649625, + "grad_norm": 2.1147842407226562, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7625012397766113, + "num_tokens": 569638411.0, + "step": 22604 + }, + { + "epoch": 1.864022429290014, + "grad_norm": 1.9116077423095703, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7643696069717407, + "num_tokens": 569673401.0, + "step": 22605 + }, + { + "epoch": 1.8641048899150654, + "grad_norm": 2.2418365478515625, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7473433613777161, + "num_tokens": 569700816.0, + "step": 22606 + }, + { + "epoch": 1.8641873505401172, + "grad_norm": 2.0641791820526123, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7555711269378662, + "num_tokens": 569727895.0, + "step": 22607 + }, + { + "epoch": 1.8642698111651685, + "grad_norm": 2.109064817428589, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7426934838294983, + "num_tokens": 569757444.0, + "step": 22608 + }, + { + "epoch": 1.8643522717902201, + "grad_norm": 2.15946102142334, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7352452874183655, + "num_tokens": 569784860.0, + "step": 22609 + }, + { + "epoch": 1.8644347324152717, + "grad_norm": 1.9132678508758545, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7328285574913025, + "num_tokens": 569820265.0, + "step": 22610 + }, + { + "epoch": 1.8645171930403233, + "grad_norm": 1.892107367515564, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.7926898002624512, + "num_tokens": 569849200.0, + "step": 22611 + }, + { + "epoch": 1.8645996536653748, + "grad_norm": 2.236985921859741, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.7821986675262451, + "num_tokens": 569873258.0, + "step": 22612 + }, + { + "epoch": 1.8646821142904262, + "grad_norm": 2.1667027473449707, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7235358953475952, + "num_tokens": 569902650.0, + "step": 22613 + }, + { + "epoch": 1.864764574915478, + "grad_norm": 1.8397897481918335, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7637096047401428, + "num_tokens": 569933496.0, + "step": 22614 + }, + { + "epoch": 1.8648470355405293, + "grad_norm": 1.8504081964492798, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7422928214073181, + "num_tokens": 569969626.0, + "step": 22615 + }, + { + "epoch": 1.864929496165581, + "grad_norm": 1.8197697401046753, + "learning_rate": 1e-06, + "loss": 0.5723, + "mean_token_accuracy": 0.8081743717193604, + "num_tokens": 570002059.0, + "step": 22616 + }, + { + "epoch": 1.8650119567906325, + "grad_norm": 1.8434215784072876, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7310086488723755, + "num_tokens": 570037136.0, + "step": 22617 + }, + { + "epoch": 1.865094417415684, + "grad_norm": 2.042170524597168, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7651168704032898, + "num_tokens": 570065080.0, + "step": 22618 + }, + { + "epoch": 1.8651768780407356, + "grad_norm": 1.8306844234466553, + "learning_rate": 1e-06, + "loss": 0.5865, + "mean_token_accuracy": 0.8049079179763794, + "num_tokens": 570097762.0, + "step": 22619 + }, + { + "epoch": 1.865259338665787, + "grad_norm": 1.7992299795150757, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7691069841384888, + "num_tokens": 570132693.0, + "step": 22620 + }, + { + "epoch": 1.8653417992908388, + "grad_norm": 1.8798518180847168, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7394415140151978, + "num_tokens": 570165370.0, + "step": 22621 + }, + { + "epoch": 1.8654242599158901, + "grad_norm": 1.858744502067566, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7705371975898743, + "num_tokens": 570197243.0, + "step": 22622 + }, + { + "epoch": 1.8655067205409417, + "grad_norm": 1.958511233329773, + "learning_rate": 1e-06, + "loss": 0.6155, + "mean_token_accuracy": 0.7994327545166016, + "num_tokens": 570228176.0, + "step": 22623 + }, + { + "epoch": 1.8655891811659933, + "grad_norm": 1.8209656476974487, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.7920299768447876, + "num_tokens": 570259906.0, + "step": 22624 + }, + { + "epoch": 1.8656716417910446, + "grad_norm": 2.2301554679870605, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7362147569656372, + "num_tokens": 570285511.0, + "step": 22625 + }, + { + "epoch": 1.8657541024160964, + "grad_norm": 1.7387752532958984, + "learning_rate": 1e-06, + "loss": 0.6054, + "mean_token_accuracy": 0.806734025478363, + "num_tokens": 570321124.0, + "step": 22626 + }, + { + "epoch": 1.8658365630411478, + "grad_norm": 2.004863977432251, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7490699291229248, + "num_tokens": 570351663.0, + "step": 22627 + }, + { + "epoch": 1.8659190236661995, + "grad_norm": 1.8380943536758423, + "learning_rate": 1e-06, + "loss": 0.5809, + "mean_token_accuracy": 0.8079965114593506, + "num_tokens": 570385554.0, + "step": 22628 + }, + { + "epoch": 1.866001484291251, + "grad_norm": 2.1675078868865967, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7060059905052185, + "num_tokens": 570413333.0, + "step": 22629 + }, + { + "epoch": 1.8660839449163025, + "grad_norm": 1.8857589960098267, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7410480976104736, + "num_tokens": 570447264.0, + "step": 22630 + }, + { + "epoch": 1.866166405541354, + "grad_norm": 1.662649393081665, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.805760383605957, + "num_tokens": 570483538.0, + "step": 22631 + }, + { + "epoch": 1.8662488661664054, + "grad_norm": 2.069751024246216, + "learning_rate": 1e-06, + "loss": 0.5957, + "mean_token_accuracy": 0.8056228160858154, + "num_tokens": 570511143.0, + "step": 22632 + }, + { + "epoch": 1.8663313267914572, + "grad_norm": 2.0318543910980225, + "learning_rate": 1e-06, + "loss": 0.7746, + "mean_token_accuracy": 0.7497169375419617, + "num_tokens": 570542358.0, + "step": 22633 + }, + { + "epoch": 1.8664137874165085, + "grad_norm": 2.220182180404663, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7473671436309814, + "num_tokens": 570567767.0, + "step": 22634 + }, + { + "epoch": 1.8664962480415601, + "grad_norm": 2.1082077026367188, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7553560733795166, + "num_tokens": 570594441.0, + "step": 22635 + }, + { + "epoch": 1.8665787086666117, + "grad_norm": 1.9148037433624268, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7378686666488647, + "num_tokens": 570629173.0, + "step": 22636 + }, + { + "epoch": 1.8666611692916633, + "grad_norm": 2.254399538040161, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7414791584014893, + "num_tokens": 570657627.0, + "step": 22637 + }, + { + "epoch": 1.8667436299167148, + "grad_norm": 2.048116683959961, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7689022421836853, + "num_tokens": 570686365.0, + "step": 22638 + }, + { + "epoch": 1.8668260905417662, + "grad_norm": 1.915435552597046, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7487695217132568, + "num_tokens": 570719025.0, + "step": 22639 + }, + { + "epoch": 1.866908551166818, + "grad_norm": 2.068617820739746, + "learning_rate": 1e-06, + "loss": 0.5763, + "mean_token_accuracy": 0.8045293092727661, + "num_tokens": 570745402.0, + "step": 22640 + }, + { + "epoch": 1.8669910117918693, + "grad_norm": 1.933640480041504, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7731815576553345, + "num_tokens": 570776697.0, + "step": 22641 + }, + { + "epoch": 1.867073472416921, + "grad_norm": 1.8416438102722168, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7622016668319702, + "num_tokens": 570811092.0, + "step": 22642 + }, + { + "epoch": 1.8671559330419725, + "grad_norm": 1.959782361984253, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7700157165527344, + "num_tokens": 570839664.0, + "step": 22643 + }, + { + "epoch": 1.867238393667024, + "grad_norm": 2.426506280899048, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7546236515045166, + "num_tokens": 570861386.0, + "step": 22644 + }, + { + "epoch": 1.8673208542920756, + "grad_norm": 1.9729712009429932, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7640020251274109, + "num_tokens": 570893051.0, + "step": 22645 + }, + { + "epoch": 1.867403314917127, + "grad_norm": 1.9684168100357056, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.7774195075035095, + "num_tokens": 570923980.0, + "step": 22646 + }, + { + "epoch": 1.8674857755421788, + "grad_norm": 1.9378496408462524, + "learning_rate": 1e-06, + "loss": 0.5241, + "mean_token_accuracy": 0.8169223070144653, + "num_tokens": 570952136.0, + "step": 22647 + }, + { + "epoch": 1.8675682361672301, + "grad_norm": 1.8342171907424927, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7586565017700195, + "num_tokens": 570986714.0, + "step": 22648 + }, + { + "epoch": 1.8676506967922817, + "grad_norm": 1.9319744110107422, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7620795369148254, + "num_tokens": 571020806.0, + "step": 22649 + }, + { + "epoch": 1.8677331574173333, + "grad_norm": 1.8329648971557617, + "learning_rate": 1e-06, + "loss": 0.5858, + "mean_token_accuracy": 0.8006008267402649, + "num_tokens": 571053015.0, + "step": 22650 + }, + { + "epoch": 1.8678156180423846, + "grad_norm": 2.1087229251861572, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7571992874145508, + "num_tokens": 571084586.0, + "step": 22651 + }, + { + "epoch": 1.8678980786674364, + "grad_norm": 1.9456758499145508, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7738118171691895, + "num_tokens": 571113731.0, + "step": 22652 + }, + { + "epoch": 1.8679805392924878, + "grad_norm": 1.9152462482452393, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7298352718353271, + "num_tokens": 571146018.0, + "step": 22653 + }, + { + "epoch": 1.8680629999175393, + "grad_norm": 2.0439255237579346, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7295657396316528, + "num_tokens": 571176760.0, + "step": 22654 + }, + { + "epoch": 1.868145460542591, + "grad_norm": 2.0863263607025146, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7469539642333984, + "num_tokens": 571204276.0, + "step": 22655 + }, + { + "epoch": 1.8682279211676425, + "grad_norm": 2.088491916656494, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7545592188835144, + "num_tokens": 571232256.0, + "step": 22656 + }, + { + "epoch": 1.868310381792694, + "grad_norm": 2.0151126384735107, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7336280345916748, + "num_tokens": 571267500.0, + "step": 22657 + }, + { + "epoch": 1.8683928424177454, + "grad_norm": 1.9917762279510498, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7353500127792358, + "num_tokens": 571297055.0, + "step": 22658 + }, + { + "epoch": 1.8684753030427972, + "grad_norm": 2.08528995513916, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7729108929634094, + "num_tokens": 571323765.0, + "step": 22659 + }, + { + "epoch": 1.8685577636678485, + "grad_norm": 2.0493550300598145, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7565932869911194, + "num_tokens": 571352853.0, + "step": 22660 + }, + { + "epoch": 1.8686402242929, + "grad_norm": 1.766088604927063, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7148755192756653, + "num_tokens": 571390805.0, + "step": 22661 + }, + { + "epoch": 1.8687226849179517, + "grad_norm": 2.0667660236358643, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7547423243522644, + "num_tokens": 571417311.0, + "step": 22662 + }, + { + "epoch": 1.8688051455430033, + "grad_norm": 1.8541537523269653, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7612528800964355, + "num_tokens": 571450166.0, + "step": 22663 + }, + { + "epoch": 1.8688876061680548, + "grad_norm": 2.0375559329986572, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7308698892593384, + "num_tokens": 571478142.0, + "step": 22664 + }, + { + "epoch": 1.8689700667931062, + "grad_norm": 1.7087602615356445, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.7744256258010864, + "num_tokens": 571514564.0, + "step": 22665 + }, + { + "epoch": 1.869052527418158, + "grad_norm": 2.202303886413574, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.7521083354949951, + "num_tokens": 571543038.0, + "step": 22666 + }, + { + "epoch": 1.8691349880432093, + "grad_norm": 1.9356721639633179, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7875919342041016, + "num_tokens": 571574282.0, + "step": 22667 + }, + { + "epoch": 1.869217448668261, + "grad_norm": 1.7297199964523315, + "learning_rate": 1e-06, + "loss": 0.589, + "mean_token_accuracy": 0.8068150281906128, + "num_tokens": 571608683.0, + "step": 22668 + }, + { + "epoch": 1.8692999092933125, + "grad_norm": 2.100625514984131, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.750044584274292, + "num_tokens": 571634932.0, + "step": 22669 + }, + { + "epoch": 1.8693823699183638, + "grad_norm": 2.100205183029175, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.7803658843040466, + "num_tokens": 571660741.0, + "step": 22670 + }, + { + "epoch": 1.8694648305434156, + "grad_norm": 2.0932066440582275, + "learning_rate": 1e-06, + "loss": 0.6251, + "mean_token_accuracy": 0.7829804420471191, + "num_tokens": 571690178.0, + "step": 22671 + }, + { + "epoch": 1.869547291168467, + "grad_norm": 1.9948009252548218, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7469569444656372, + "num_tokens": 571722591.0, + "step": 22672 + }, + { + "epoch": 1.8696297517935188, + "grad_norm": 2.1425721645355225, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7753893136978149, + "num_tokens": 571747851.0, + "step": 22673 + }, + { + "epoch": 1.86971221241857, + "grad_norm": 1.8223668336868286, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7618745565414429, + "num_tokens": 571780818.0, + "step": 22674 + }, + { + "epoch": 1.8697946730436217, + "grad_norm": 1.748520851135254, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7688732743263245, + "num_tokens": 571819031.0, + "step": 22675 + }, + { + "epoch": 1.8698771336686733, + "grad_norm": 2.0304598808288574, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.7864611148834229, + "num_tokens": 571844564.0, + "step": 22676 + }, + { + "epoch": 1.8699595942937246, + "grad_norm": 2.0705528259277344, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7391080260276794, + "num_tokens": 571874447.0, + "step": 22677 + }, + { + "epoch": 1.8700420549187764, + "grad_norm": 1.8996363878250122, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7573885321617126, + "num_tokens": 571909435.0, + "step": 22678 + }, + { + "epoch": 1.8701245155438277, + "grad_norm": 1.8045217990875244, + "learning_rate": 1e-06, + "loss": 0.4812, + "mean_token_accuracy": 0.8374694585800171, + "num_tokens": 571941832.0, + "step": 22679 + }, + { + "epoch": 1.8702069761688793, + "grad_norm": 1.8274685144424438, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7745109796524048, + "num_tokens": 571974839.0, + "step": 22680 + }, + { + "epoch": 1.870289436793931, + "grad_norm": 2.2912557125091553, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7656925320625305, + "num_tokens": 571998143.0, + "step": 22681 + }, + { + "epoch": 1.8703718974189825, + "grad_norm": 1.8738231658935547, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.79027259349823, + "num_tokens": 572029933.0, + "step": 22682 + }, + { + "epoch": 1.870454358044034, + "grad_norm": 1.8531266450881958, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7508230805397034, + "num_tokens": 572064046.0, + "step": 22683 + }, + { + "epoch": 1.8705368186690854, + "grad_norm": 1.7294865846633911, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.77159583568573, + "num_tokens": 572100589.0, + "step": 22684 + }, + { + "epoch": 1.8706192792941372, + "grad_norm": 2.107609987258911, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7653791308403015, + "num_tokens": 572128086.0, + "step": 22685 + }, + { + "epoch": 1.8707017399191885, + "grad_norm": 1.9804145097732544, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.778316855430603, + "num_tokens": 572155885.0, + "step": 22686 + }, + { + "epoch": 1.87078420054424, + "grad_norm": 1.9608296155929565, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7480942010879517, + "num_tokens": 572185650.0, + "step": 22687 + }, + { + "epoch": 1.8708666611692917, + "grad_norm": 1.792799711227417, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.773101270198822, + "num_tokens": 572221442.0, + "step": 22688 + }, + { + "epoch": 1.8709491217943433, + "grad_norm": 2.1177520751953125, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7473759651184082, + "num_tokens": 572250061.0, + "step": 22689 + }, + { + "epoch": 1.8710315824193948, + "grad_norm": 2.06276273727417, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7520368099212646, + "num_tokens": 572278830.0, + "step": 22690 + }, + { + "epoch": 1.8711140430444462, + "grad_norm": 1.8272627592086792, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.717609167098999, + "num_tokens": 572313763.0, + "step": 22691 + }, + { + "epoch": 1.871196503669498, + "grad_norm": 2.399775266647339, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7730498313903809, + "num_tokens": 572334235.0, + "step": 22692 + }, + { + "epoch": 1.8712789642945493, + "grad_norm": 1.792174220085144, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7655252814292908, + "num_tokens": 572368844.0, + "step": 22693 + }, + { + "epoch": 1.871361424919601, + "grad_norm": 2.168958902359009, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7652097940444946, + "num_tokens": 572394157.0, + "step": 22694 + }, + { + "epoch": 1.8714438855446525, + "grad_norm": 1.882864236831665, + "learning_rate": 1e-06, + "loss": 0.5888, + "mean_token_accuracy": 0.8030494451522827, + "num_tokens": 572424495.0, + "step": 22695 + }, + { + "epoch": 1.8715263461697038, + "grad_norm": 2.120060682296753, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.770447313785553, + "num_tokens": 572452435.0, + "step": 22696 + }, + { + "epoch": 1.8716088067947556, + "grad_norm": 2.0301945209503174, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7733097672462463, + "num_tokens": 572481525.0, + "step": 22697 + }, + { + "epoch": 1.871691267419807, + "grad_norm": 2.0432674884796143, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7659916877746582, + "num_tokens": 572508848.0, + "step": 22698 + }, + { + "epoch": 1.8717737280448585, + "grad_norm": 2.0552093982696533, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.7892258167266846, + "num_tokens": 572533480.0, + "step": 22699 + }, + { + "epoch": 1.87185618866991, + "grad_norm": 2.0824155807495117, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7803469300270081, + "num_tokens": 572559240.0, + "step": 22700 + }, + { + "epoch": 1.8719386492949617, + "grad_norm": 2.1594908237457275, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.7582908868789673, + "num_tokens": 572583432.0, + "step": 22701 + }, + { + "epoch": 1.8720211099200132, + "grad_norm": 1.877739667892456, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7754955291748047, + "num_tokens": 572616525.0, + "step": 22702 + }, + { + "epoch": 1.8721035705450646, + "grad_norm": 2.122889757156372, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7413496971130371, + "num_tokens": 572646385.0, + "step": 22703 + }, + { + "epoch": 1.8721860311701164, + "grad_norm": 1.9729588031768799, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7481417655944824, + "num_tokens": 572679602.0, + "step": 22704 + }, + { + "epoch": 1.8722684917951677, + "grad_norm": 1.8757375478744507, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7872156500816345, + "num_tokens": 572711104.0, + "step": 22705 + }, + { + "epoch": 1.8723509524202193, + "grad_norm": 2.355782985687256, + "learning_rate": 1e-06, + "loss": 0.7782, + "mean_token_accuracy": 0.7528235912322998, + "num_tokens": 572734388.0, + "step": 22706 + }, + { + "epoch": 1.872433413045271, + "grad_norm": 1.8154045343399048, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.7890794277191162, + "num_tokens": 572767415.0, + "step": 22707 + }, + { + "epoch": 1.8725158736703225, + "grad_norm": 1.9185292720794678, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.7948867082595825, + "num_tokens": 572798658.0, + "step": 22708 + }, + { + "epoch": 1.872598334295374, + "grad_norm": 2.3918139934539795, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7450217604637146, + "num_tokens": 572821098.0, + "step": 22709 + }, + { + "epoch": 1.8726807949204254, + "grad_norm": 2.167736768722534, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7434296011924744, + "num_tokens": 572848907.0, + "step": 22710 + }, + { + "epoch": 1.8727632555454772, + "grad_norm": 2.1926536560058594, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7740396857261658, + "num_tokens": 572874005.0, + "step": 22711 + }, + { + "epoch": 1.8728457161705285, + "grad_norm": 1.8179572820663452, + "learning_rate": 1e-06, + "loss": 0.5977, + "mean_token_accuracy": 0.8001531362533569, + "num_tokens": 572906038.0, + "step": 22712 + }, + { + "epoch": 1.87292817679558, + "grad_norm": 2.210475444793701, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7076441645622253, + "num_tokens": 572933318.0, + "step": 22713 + }, + { + "epoch": 1.8730106374206317, + "grad_norm": 1.9872138500213623, + "learning_rate": 1e-06, + "loss": 0.5967, + "mean_token_accuracy": 0.8020712733268738, + "num_tokens": 572962131.0, + "step": 22714 + }, + { + "epoch": 1.873093098045683, + "grad_norm": 2.0972108840942383, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7359498143196106, + "num_tokens": 572992191.0, + "step": 22715 + }, + { + "epoch": 1.8731755586707348, + "grad_norm": 1.9411526918411255, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7584595084190369, + "num_tokens": 573024303.0, + "step": 22716 + }, + { + "epoch": 1.8732580192957862, + "grad_norm": 2.110898494720459, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7385300397872925, + "num_tokens": 573052801.0, + "step": 22717 + }, + { + "epoch": 1.873340479920838, + "grad_norm": 2.053227663040161, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7511249780654907, + "num_tokens": 573080703.0, + "step": 22718 + }, + { + "epoch": 1.8734229405458893, + "grad_norm": 2.2287678718566895, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7313525080680847, + "num_tokens": 573106457.0, + "step": 22719 + }, + { + "epoch": 1.8735054011709409, + "grad_norm": 2.2393083572387695, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7570857405662537, + "num_tokens": 573130674.0, + "step": 22720 + }, + { + "epoch": 1.8735878617959925, + "grad_norm": 1.953452706336975, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7687162160873413, + "num_tokens": 573161222.0, + "step": 22721 + }, + { + "epoch": 1.8736703224210438, + "grad_norm": 2.04640793800354, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7570894956588745, + "num_tokens": 573187402.0, + "step": 22722 + }, + { + "epoch": 1.8737527830460956, + "grad_norm": 1.8509677648544312, + "learning_rate": 1e-06, + "loss": 0.5491, + "mean_token_accuracy": 0.8176237344741821, + "num_tokens": 573218189.0, + "step": 22723 + }, + { + "epoch": 1.873835243671147, + "grad_norm": 2.106462240219116, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7387462854385376, + "num_tokens": 573247859.0, + "step": 22724 + }, + { + "epoch": 1.8739177042961985, + "grad_norm": 2.2576732635498047, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7917022705078125, + "num_tokens": 573271895.0, + "step": 22725 + }, + { + "epoch": 1.87400016492125, + "grad_norm": 2.028231620788574, + "learning_rate": 1e-06, + "loss": 0.5128, + "mean_token_accuracy": 0.8140445947647095, + "num_tokens": 573299729.0, + "step": 22726 + }, + { + "epoch": 1.8740826255463017, + "grad_norm": 2.172255277633667, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7368824481964111, + "num_tokens": 573328071.0, + "step": 22727 + }, + { + "epoch": 1.8741650861713532, + "grad_norm": 2.0383079051971436, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7728078961372375, + "num_tokens": 573355491.0, + "step": 22728 + }, + { + "epoch": 1.8742475467964046, + "grad_norm": 1.9763768911361694, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7199686765670776, + "num_tokens": 573388994.0, + "step": 22729 + }, + { + "epoch": 1.8743300074214564, + "grad_norm": 2.0885868072509766, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7530782222747803, + "num_tokens": 573417183.0, + "step": 22730 + }, + { + "epoch": 1.8744124680465077, + "grad_norm": 2.1701390743255615, + "learning_rate": 1e-06, + "loss": 0.7706, + "mean_token_accuracy": 0.7618919610977173, + "num_tokens": 573441292.0, + "step": 22731 + }, + { + "epoch": 1.8744949286715593, + "grad_norm": 2.0186116695404053, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7740510702133179, + "num_tokens": 573470245.0, + "step": 22732 + }, + { + "epoch": 1.8745773892966109, + "grad_norm": 1.6723415851593018, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7573403120040894, + "num_tokens": 573508098.0, + "step": 22733 + }, + { + "epoch": 1.8746598499216625, + "grad_norm": 2.0275719165802, + "learning_rate": 1e-06, + "loss": 0.5761, + "mean_token_accuracy": 0.8000215888023376, + "num_tokens": 573534790.0, + "step": 22734 + }, + { + "epoch": 1.874742310546714, + "grad_norm": 1.8340096473693848, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.7878797054290771, + "num_tokens": 573565435.0, + "step": 22735 + }, + { + "epoch": 1.8748247711717654, + "grad_norm": 1.9777743816375732, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7821764349937439, + "num_tokens": 573594614.0, + "step": 22736 + }, + { + "epoch": 1.8749072317968172, + "grad_norm": 2.2524120807647705, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7479609251022339, + "num_tokens": 573618186.0, + "step": 22737 + }, + { + "epoch": 1.8749896924218685, + "grad_norm": 1.8022899627685547, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.7852518558502197, + "num_tokens": 573656122.0, + "step": 22738 + }, + { + "epoch": 1.87507215304692, + "grad_norm": 1.8128995895385742, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.734734296798706, + "num_tokens": 573690162.0, + "step": 22739 + }, + { + "epoch": 1.8751546136719717, + "grad_norm": 2.1887714862823486, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7303701639175415, + "num_tokens": 573715555.0, + "step": 22740 + }, + { + "epoch": 1.875237074297023, + "grad_norm": 1.9835383892059326, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.766679048538208, + "num_tokens": 573744996.0, + "step": 22741 + }, + { + "epoch": 1.8753195349220748, + "grad_norm": 2.0386712551116943, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7650450468063354, + "num_tokens": 573775853.0, + "step": 22742 + }, + { + "epoch": 1.8754019955471262, + "grad_norm": 1.8023446798324585, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.747886598110199, + "num_tokens": 573815661.0, + "step": 22743 + }, + { + "epoch": 1.8754844561721777, + "grad_norm": 1.9458988904953003, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7776415348052979, + "num_tokens": 573844518.0, + "step": 22744 + }, + { + "epoch": 1.8755669167972293, + "grad_norm": 2.3329899311065674, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.784008800983429, + "num_tokens": 573869127.0, + "step": 22745 + }, + { + "epoch": 1.8756493774222809, + "grad_norm": 2.047797918319702, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.749656081199646, + "num_tokens": 573899281.0, + "step": 22746 + }, + { + "epoch": 1.8757318380473325, + "grad_norm": 1.9854936599731445, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7497676014900208, + "num_tokens": 573928737.0, + "step": 22747 + }, + { + "epoch": 1.8758142986723838, + "grad_norm": 2.390958070755005, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7340933084487915, + "num_tokens": 573951527.0, + "step": 22748 + }, + { + "epoch": 1.8758967592974356, + "grad_norm": 2.040038585662842, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.7791095972061157, + "num_tokens": 573979778.0, + "step": 22749 + }, + { + "epoch": 1.875979219922487, + "grad_norm": 1.906935691833496, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7577778100967407, + "num_tokens": 574014947.0, + "step": 22750 + }, + { + "epoch": 1.8760616805475385, + "grad_norm": 1.8484669923782349, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7596132755279541, + "num_tokens": 574049759.0, + "step": 22751 + }, + { + "epoch": 1.87614414117259, + "grad_norm": 2.108893632888794, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7154856324195862, + "num_tokens": 574080949.0, + "step": 22752 + }, + { + "epoch": 1.8762266017976417, + "grad_norm": 1.880478024482727, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.7735531330108643, + "num_tokens": 574113670.0, + "step": 22753 + }, + { + "epoch": 1.8763090624226932, + "grad_norm": 2.237556219100952, + "learning_rate": 1e-06, + "loss": 0.7746, + "mean_token_accuracy": 0.7492133378982544, + "num_tokens": 574138764.0, + "step": 22754 + }, + { + "epoch": 1.8763915230477446, + "grad_norm": 1.8612548112869263, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7638915777206421, + "num_tokens": 574171199.0, + "step": 22755 + }, + { + "epoch": 1.8764739836727964, + "grad_norm": 1.8578481674194336, + "learning_rate": 1e-06, + "loss": 0.6106, + "mean_token_accuracy": 0.7979452013969421, + "num_tokens": 574202166.0, + "step": 22756 + }, + { + "epoch": 1.8765564442978477, + "grad_norm": 2.1242856979370117, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.755902111530304, + "num_tokens": 574231235.0, + "step": 22757 + }, + { + "epoch": 1.8766389049228993, + "grad_norm": 1.9675008058547974, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7663245797157288, + "num_tokens": 574264624.0, + "step": 22758 + }, + { + "epoch": 1.8767213655479509, + "grad_norm": 2.2160427570343018, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7753751873970032, + "num_tokens": 574292792.0, + "step": 22759 + }, + { + "epoch": 1.8768038261730022, + "grad_norm": 2.1454620361328125, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7574982643127441, + "num_tokens": 574318447.0, + "step": 22760 + }, + { + "epoch": 1.876886286798054, + "grad_norm": 2.08192777633667, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7445670366287231, + "num_tokens": 574347623.0, + "step": 22761 + }, + { + "epoch": 1.8769687474231054, + "grad_norm": 1.9526280164718628, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7615807056427002, + "num_tokens": 574379359.0, + "step": 22762 + }, + { + "epoch": 1.8770512080481572, + "grad_norm": 2.0030431747436523, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7501327991485596, + "num_tokens": 574409665.0, + "step": 22763 + }, + { + "epoch": 1.8771336686732085, + "grad_norm": 2.159722089767456, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7454378604888916, + "num_tokens": 574436190.0, + "step": 22764 + }, + { + "epoch": 1.87721612929826, + "grad_norm": 2.1218421459198, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.7484011054039001, + "num_tokens": 574464435.0, + "step": 22765 + }, + { + "epoch": 1.8772985899233117, + "grad_norm": 1.6998969316482544, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.802297830581665, + "num_tokens": 574500548.0, + "step": 22766 + }, + { + "epoch": 1.877381050548363, + "grad_norm": 1.8704214096069336, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7180746793746948, + "num_tokens": 574532595.0, + "step": 22767 + }, + { + "epoch": 1.8774635111734148, + "grad_norm": 2.150871992111206, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7105830311775208, + "num_tokens": 574560957.0, + "step": 22768 + }, + { + "epoch": 1.8775459717984662, + "grad_norm": 2.352565288543701, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7545220851898193, + "num_tokens": 574586901.0, + "step": 22769 + }, + { + "epoch": 1.8776284324235177, + "grad_norm": 1.9144691228866577, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7481728792190552, + "num_tokens": 574618245.0, + "step": 22770 + }, + { + "epoch": 1.8777108930485693, + "grad_norm": 2.1585817337036133, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7796915769577026, + "num_tokens": 574644364.0, + "step": 22771 + }, + { + "epoch": 1.8777933536736209, + "grad_norm": 2.075408935546875, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.757002592086792, + "num_tokens": 574671335.0, + "step": 22772 + }, + { + "epoch": 1.8778758142986725, + "grad_norm": 2.2755286693573, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7196469306945801, + "num_tokens": 574701578.0, + "step": 22773 + }, + { + "epoch": 1.8779582749237238, + "grad_norm": 1.8491078615188599, + "learning_rate": 1e-06, + "loss": 0.5855, + "mean_token_accuracy": 0.8076890110969543, + "num_tokens": 574732631.0, + "step": 22774 + }, + { + "epoch": 1.8780407355487756, + "grad_norm": 2.1370863914489746, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.77197265625, + "num_tokens": 574756562.0, + "step": 22775 + }, + { + "epoch": 1.878123196173827, + "grad_norm": 2.05360746383667, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7499634027481079, + "num_tokens": 574784902.0, + "step": 22776 + }, + { + "epoch": 1.8782056567988785, + "grad_norm": 2.004300117492676, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7793135643005371, + "num_tokens": 574811735.0, + "step": 22777 + }, + { + "epoch": 1.87828811742393, + "grad_norm": 2.1118764877319336, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.7505164742469788, + "num_tokens": 574842109.0, + "step": 22778 + }, + { + "epoch": 1.8783705780489817, + "grad_norm": 1.8556969165802002, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7234122157096863, + "num_tokens": 574877834.0, + "step": 22779 + }, + { + "epoch": 1.8784530386740332, + "grad_norm": 2.147219657897949, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7397865653038025, + "num_tokens": 574903829.0, + "step": 22780 + }, + { + "epoch": 1.8785354992990846, + "grad_norm": 2.1179440021514893, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.7822026014328003, + "num_tokens": 574931137.0, + "step": 22781 + }, + { + "epoch": 1.8786179599241364, + "grad_norm": 1.8277066946029663, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.7892948389053345, + "num_tokens": 574965322.0, + "step": 22782 + }, + { + "epoch": 1.8787004205491877, + "grad_norm": 2.0434718132019043, + "learning_rate": 1e-06, + "loss": 0.5536, + "mean_token_accuracy": 0.8175754547119141, + "num_tokens": 574992546.0, + "step": 22783 + }, + { + "epoch": 1.8787828811742393, + "grad_norm": 1.7586264610290527, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.76493239402771, + "num_tokens": 575028277.0, + "step": 22784 + }, + { + "epoch": 1.8788653417992909, + "grad_norm": 2.006096839904785, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7541072368621826, + "num_tokens": 575055965.0, + "step": 22785 + }, + { + "epoch": 1.8789478024243422, + "grad_norm": 1.8400721549987793, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7777560949325562, + "num_tokens": 575089568.0, + "step": 22786 + }, + { + "epoch": 1.879030263049394, + "grad_norm": 1.989863634109497, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7576848864555359, + "num_tokens": 575118325.0, + "step": 22787 + }, + { + "epoch": 1.8791127236744454, + "grad_norm": 1.8687310218811035, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7383146286010742, + "num_tokens": 575151702.0, + "step": 22788 + }, + { + "epoch": 1.879195184299497, + "grad_norm": 1.8084415197372437, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7593008875846863, + "num_tokens": 575188367.0, + "step": 22789 + }, + { + "epoch": 1.8792776449245485, + "grad_norm": 2.126035213470459, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7483583688735962, + "num_tokens": 575215483.0, + "step": 22790 + }, + { + "epoch": 1.8793601055496, + "grad_norm": 2.193510055541992, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7331751585006714, + "num_tokens": 575240400.0, + "step": 22791 + }, + { + "epoch": 1.8794425661746517, + "grad_norm": 1.8710832595825195, + "learning_rate": 1e-06, + "loss": 0.5253, + "mean_token_accuracy": 0.8143510818481445, + "num_tokens": 575272417.0, + "step": 22792 + }, + { + "epoch": 1.879525026799703, + "grad_norm": 2.0173304080963135, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.752701997756958, + "num_tokens": 575303338.0, + "step": 22793 + }, + { + "epoch": 1.8796074874247548, + "grad_norm": 2.3080856800079346, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7742518186569214, + "num_tokens": 575329776.0, + "step": 22794 + }, + { + "epoch": 1.8796899480498062, + "grad_norm": 2.143629312515259, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7452276945114136, + "num_tokens": 575356965.0, + "step": 22795 + }, + { + "epoch": 1.8797724086748577, + "grad_norm": 2.2715916633605957, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7667425870895386, + "num_tokens": 575381522.0, + "step": 22796 + }, + { + "epoch": 1.8798548692999093, + "grad_norm": 2.2534518241882324, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7505115866661072, + "num_tokens": 575405819.0, + "step": 22797 + }, + { + "epoch": 1.8799373299249609, + "grad_norm": 1.882943034172058, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7657520174980164, + "num_tokens": 575437476.0, + "step": 22798 + }, + { + "epoch": 1.8800197905500124, + "grad_norm": 2.117537260055542, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7691748142242432, + "num_tokens": 575465013.0, + "step": 22799 + }, + { + "epoch": 1.8801022511750638, + "grad_norm": 2.2013673782348633, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7762290239334106, + "num_tokens": 575494455.0, + "step": 22800 + }, + { + "epoch": 1.8801847118001156, + "grad_norm": 2.3083865642547607, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.7584257125854492, + "num_tokens": 575518479.0, + "step": 22801 + }, + { + "epoch": 1.880267172425167, + "grad_norm": 2.0177133083343506, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7735779285430908, + "num_tokens": 575550113.0, + "step": 22802 + }, + { + "epoch": 1.8803496330502185, + "grad_norm": 1.916818618774414, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.7829217910766602, + "num_tokens": 575580104.0, + "step": 22803 + }, + { + "epoch": 1.88043209367527, + "grad_norm": 1.9631388187408447, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7742891907691956, + "num_tokens": 575612639.0, + "step": 22804 + }, + { + "epoch": 1.8805145543003214, + "grad_norm": 1.7465869188308716, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.7921746373176575, + "num_tokens": 575649284.0, + "step": 22805 + }, + { + "epoch": 1.8805970149253732, + "grad_norm": 2.1116716861724854, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7616537809371948, + "num_tokens": 575676802.0, + "step": 22806 + }, + { + "epoch": 1.8806794755504246, + "grad_norm": 1.9437695741653442, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7259289622306824, + "num_tokens": 575707400.0, + "step": 22807 + }, + { + "epoch": 1.8807619361754762, + "grad_norm": 2.0287132263183594, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7678778171539307, + "num_tokens": 575737243.0, + "step": 22808 + }, + { + "epoch": 1.8808443968005277, + "grad_norm": 1.9392696619033813, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7628119587898254, + "num_tokens": 575767143.0, + "step": 22809 + }, + { + "epoch": 1.8809268574255793, + "grad_norm": 1.9224112033843994, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.7903130650520325, + "num_tokens": 575799982.0, + "step": 22810 + }, + { + "epoch": 1.8810093180506309, + "grad_norm": 1.8478212356567383, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7763393521308899, + "num_tokens": 575833535.0, + "step": 22811 + }, + { + "epoch": 1.8810917786756822, + "grad_norm": 1.9600763320922852, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.7846695184707642, + "num_tokens": 575863701.0, + "step": 22812 + }, + { + "epoch": 1.881174239300734, + "grad_norm": 1.889394760131836, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7640044689178467, + "num_tokens": 575894171.0, + "step": 22813 + }, + { + "epoch": 1.8812566999257854, + "grad_norm": 2.1157774925231934, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.773460865020752, + "num_tokens": 575921422.0, + "step": 22814 + }, + { + "epoch": 1.881339160550837, + "grad_norm": 2.0246376991271973, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.7823930978775024, + "num_tokens": 575952282.0, + "step": 22815 + }, + { + "epoch": 1.8814216211758885, + "grad_norm": 2.199871778488159, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7396581172943115, + "num_tokens": 575979091.0, + "step": 22816 + }, + { + "epoch": 1.88150408180094, + "grad_norm": 1.8449569940567017, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.771354079246521, + "num_tokens": 576013071.0, + "step": 22817 + }, + { + "epoch": 1.8815865424259917, + "grad_norm": 1.8140627145767212, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7311112880706787, + "num_tokens": 576049027.0, + "step": 22818 + }, + { + "epoch": 1.881669003051043, + "grad_norm": 1.951716661453247, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7563241720199585, + "num_tokens": 576081275.0, + "step": 22819 + }, + { + "epoch": 1.8817514636760948, + "grad_norm": 1.7870328426361084, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.7881720066070557, + "num_tokens": 576118004.0, + "step": 22820 + }, + { + "epoch": 1.8818339243011462, + "grad_norm": 1.9582080841064453, + "learning_rate": 1e-06, + "loss": 0.7704, + "mean_token_accuracy": 0.7608683109283447, + "num_tokens": 576146033.0, + "step": 22821 + }, + { + "epoch": 1.8819163849261977, + "grad_norm": 2.133242130279541, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7608674168586731, + "num_tokens": 576172362.0, + "step": 22822 + }, + { + "epoch": 1.8819988455512493, + "grad_norm": 1.945210576057434, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7305548191070557, + "num_tokens": 576202755.0, + "step": 22823 + }, + { + "epoch": 1.8820813061763007, + "grad_norm": 1.938309907913208, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.7853143215179443, + "num_tokens": 576235152.0, + "step": 22824 + }, + { + "epoch": 1.8821637668013524, + "grad_norm": 2.0063745975494385, + "learning_rate": 1e-06, + "loss": 0.5722, + "mean_token_accuracy": 0.8094898462295532, + "num_tokens": 576263408.0, + "step": 22825 + }, + { + "epoch": 1.8822462274264038, + "grad_norm": 1.9371662139892578, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7636696100234985, + "num_tokens": 576293252.0, + "step": 22826 + }, + { + "epoch": 1.8823286880514556, + "grad_norm": 2.0245585441589355, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7513788938522339, + "num_tokens": 576325569.0, + "step": 22827 + }, + { + "epoch": 1.882411148676507, + "grad_norm": 1.8516446352005005, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7430617809295654, + "num_tokens": 576359590.0, + "step": 22828 + }, + { + "epoch": 1.8824936093015585, + "grad_norm": 2.088510274887085, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.735050618648529, + "num_tokens": 576389557.0, + "step": 22829 + }, + { + "epoch": 1.88257606992661, + "grad_norm": 1.8140684366226196, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.754397988319397, + "num_tokens": 576424113.0, + "step": 22830 + }, + { + "epoch": 1.8826585305516614, + "grad_norm": 1.8479666709899902, + "learning_rate": 1e-06, + "loss": 0.7799, + "mean_token_accuracy": 0.7443250417709351, + "num_tokens": 576457447.0, + "step": 22831 + }, + { + "epoch": 1.8827409911767132, + "grad_norm": 2.1927943229675293, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7673972249031067, + "num_tokens": 576481857.0, + "step": 22832 + }, + { + "epoch": 1.8828234518017646, + "grad_norm": 2.0263001918792725, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7429368495941162, + "num_tokens": 576512946.0, + "step": 22833 + }, + { + "epoch": 1.8829059124268162, + "grad_norm": 2.218451976776123, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7721043825149536, + "num_tokens": 576537994.0, + "step": 22834 + }, + { + "epoch": 1.8829883730518677, + "grad_norm": 1.8111449480056763, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7500640749931335, + "num_tokens": 576573665.0, + "step": 22835 + }, + { + "epoch": 1.8830708336769193, + "grad_norm": 1.8926290273666382, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.7886930704116821, + "num_tokens": 576602599.0, + "step": 22836 + }, + { + "epoch": 1.8831532943019709, + "grad_norm": 2.0724098682403564, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7323797941207886, + "num_tokens": 576632393.0, + "step": 22837 + }, + { + "epoch": 1.8832357549270222, + "grad_norm": 1.9559264183044434, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.7840098142623901, + "num_tokens": 576663730.0, + "step": 22838 + }, + { + "epoch": 1.883318215552074, + "grad_norm": 2.1035971641540527, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7546240091323853, + "num_tokens": 576689695.0, + "step": 22839 + }, + { + "epoch": 1.8834006761771254, + "grad_norm": 1.7886102199554443, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7337911128997803, + "num_tokens": 576725638.0, + "step": 22840 + }, + { + "epoch": 1.883483136802177, + "grad_norm": 2.1530659198760986, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7614282965660095, + "num_tokens": 576754649.0, + "step": 22841 + }, + { + "epoch": 1.8835655974272285, + "grad_norm": 1.8156063556671143, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.773187518119812, + "num_tokens": 576791698.0, + "step": 22842 + }, + { + "epoch": 1.88364805805228, + "grad_norm": 1.9896001815795898, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7464549541473389, + "num_tokens": 576820979.0, + "step": 22843 + }, + { + "epoch": 1.8837305186773317, + "grad_norm": 1.9183111190795898, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.763653576374054, + "num_tokens": 576850129.0, + "step": 22844 + }, + { + "epoch": 1.883812979302383, + "grad_norm": 1.6899269819259644, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.7975020408630371, + "num_tokens": 576887527.0, + "step": 22845 + }, + { + "epoch": 1.8838954399274348, + "grad_norm": 2.083108425140381, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7556807398796082, + "num_tokens": 576914506.0, + "step": 22846 + }, + { + "epoch": 1.8839779005524862, + "grad_norm": 1.875800371170044, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7875473499298096, + "num_tokens": 576949972.0, + "step": 22847 + }, + { + "epoch": 1.8840603611775377, + "grad_norm": 2.280726432800293, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7705970406532288, + "num_tokens": 576974605.0, + "step": 22848 + }, + { + "epoch": 1.8841428218025893, + "grad_norm": 1.903749704360962, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.746695876121521, + "num_tokens": 577009842.0, + "step": 22849 + }, + { + "epoch": 1.8842252824276406, + "grad_norm": 1.9298133850097656, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7436584234237671, + "num_tokens": 577042514.0, + "step": 22850 + }, + { + "epoch": 1.8843077430526924, + "grad_norm": 2.525840997695923, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7018651366233826, + "num_tokens": 577065036.0, + "step": 22851 + }, + { + "epoch": 1.8843902036777438, + "grad_norm": 1.9514247179031372, + "learning_rate": 1e-06, + "loss": 0.7852, + "mean_token_accuracy": 0.7624369859695435, + "num_tokens": 577098680.0, + "step": 22852 + }, + { + "epoch": 1.8844726643027954, + "grad_norm": 1.9712823629379272, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7711032032966614, + "num_tokens": 577128574.0, + "step": 22853 + }, + { + "epoch": 1.884555124927847, + "grad_norm": 1.922363042831421, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7673543095588684, + "num_tokens": 577162540.0, + "step": 22854 + }, + { + "epoch": 1.8846375855528985, + "grad_norm": 1.9326598644256592, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.7948905229568481, + "num_tokens": 577191499.0, + "step": 22855 + }, + { + "epoch": 1.88472004617795, + "grad_norm": 1.7800748348236084, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.76407790184021, + "num_tokens": 577227926.0, + "step": 22856 + }, + { + "epoch": 1.8848025068030014, + "grad_norm": 1.8817216157913208, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7302653789520264, + "num_tokens": 577263610.0, + "step": 22857 + }, + { + "epoch": 1.8848849674280532, + "grad_norm": 2.3188607692718506, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7051719427108765, + "num_tokens": 577287777.0, + "step": 22858 + }, + { + "epoch": 1.8849674280531046, + "grad_norm": 2.0578150749206543, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7485570907592773, + "num_tokens": 577317496.0, + "step": 22859 + }, + { + "epoch": 1.8850498886781561, + "grad_norm": 2.004520893096924, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.769437313079834, + "num_tokens": 577347172.0, + "step": 22860 + }, + { + "epoch": 1.8851323493032077, + "grad_norm": 2.133753776550293, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7575587630271912, + "num_tokens": 577371797.0, + "step": 22861 + }, + { + "epoch": 1.8852148099282593, + "grad_norm": 2.0437352657318115, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7220804691314697, + "num_tokens": 577399861.0, + "step": 22862 + }, + { + "epoch": 1.8852972705533109, + "grad_norm": 1.9190176725387573, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7345916032791138, + "num_tokens": 577435098.0, + "step": 22863 + }, + { + "epoch": 1.8853797311783622, + "grad_norm": 1.9740824699401855, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.733677327632904, + "num_tokens": 577464622.0, + "step": 22864 + }, + { + "epoch": 1.885462191803414, + "grad_norm": 1.8304768800735474, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.7931249141693115, + "num_tokens": 577494973.0, + "step": 22865 + }, + { + "epoch": 1.8855446524284654, + "grad_norm": 1.957844853401184, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7811015248298645, + "num_tokens": 577524302.0, + "step": 22866 + }, + { + "epoch": 1.885627113053517, + "grad_norm": 1.86530601978302, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7527685165405273, + "num_tokens": 577558880.0, + "step": 22867 + }, + { + "epoch": 1.8857095736785685, + "grad_norm": 2.0305728912353516, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.78343665599823, + "num_tokens": 577585789.0, + "step": 22868 + }, + { + "epoch": 1.8857920343036199, + "grad_norm": 2.1114730834960938, + "learning_rate": 1e-06, + "loss": 0.5869, + "mean_token_accuracy": 0.7906507253646851, + "num_tokens": 577611727.0, + "step": 22869 + }, + { + "epoch": 1.8858744949286717, + "grad_norm": 2.00929856300354, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7771185636520386, + "num_tokens": 577640169.0, + "step": 22870 + }, + { + "epoch": 1.885956955553723, + "grad_norm": 1.9050861597061157, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7287670373916626, + "num_tokens": 577676460.0, + "step": 22871 + }, + { + "epoch": 1.8860394161787748, + "grad_norm": 1.9166358709335327, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7431535720825195, + "num_tokens": 577707399.0, + "step": 22872 + }, + { + "epoch": 1.8861218768038261, + "grad_norm": 2.0409538745880127, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7547364830970764, + "num_tokens": 577736884.0, + "step": 22873 + }, + { + "epoch": 1.8862043374288777, + "grad_norm": 2.1001625061035156, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7432043552398682, + "num_tokens": 577765608.0, + "step": 22874 + }, + { + "epoch": 1.8862867980539293, + "grad_norm": 1.7204298973083496, + "learning_rate": 1e-06, + "loss": 0.454, + "mean_token_accuracy": 0.847904622554779, + "num_tokens": 577797309.0, + "step": 22875 + }, + { + "epoch": 1.8863692586789806, + "grad_norm": 2.0450637340545654, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.7675649523735046, + "num_tokens": 577827229.0, + "step": 22876 + }, + { + "epoch": 1.8864517193040324, + "grad_norm": 1.9532777070999146, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7724810838699341, + "num_tokens": 577858990.0, + "step": 22877 + }, + { + "epoch": 1.8865341799290838, + "grad_norm": 1.9831000566482544, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7706475257873535, + "num_tokens": 577886580.0, + "step": 22878 + }, + { + "epoch": 1.8866166405541354, + "grad_norm": 2.1906750202178955, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7451378107070923, + "num_tokens": 577913425.0, + "step": 22879 + }, + { + "epoch": 1.886699101179187, + "grad_norm": 1.8115317821502686, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.7957512140274048, + "num_tokens": 577944583.0, + "step": 22880 + }, + { + "epoch": 1.8867815618042385, + "grad_norm": 2.09423565864563, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7578866481781006, + "num_tokens": 577976405.0, + "step": 22881 + }, + { + "epoch": 1.88686402242929, + "grad_norm": 2.201793909072876, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7114822864532471, + "num_tokens": 578002788.0, + "step": 22882 + }, + { + "epoch": 1.8869464830543414, + "grad_norm": 1.9057424068450928, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7440283298492432, + "num_tokens": 578035778.0, + "step": 22883 + }, + { + "epoch": 1.8870289436793932, + "grad_norm": 2.0734450817108154, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.749117910861969, + "num_tokens": 578063443.0, + "step": 22884 + }, + { + "epoch": 1.8871114043044446, + "grad_norm": 1.917427659034729, + "learning_rate": 1e-06, + "loss": 0.4781, + "mean_token_accuracy": 0.8356277942657471, + "num_tokens": 578090076.0, + "step": 22885 + }, + { + "epoch": 1.8871938649294961, + "grad_norm": 1.9966444969177246, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.7825304269790649, + "num_tokens": 578118061.0, + "step": 22886 + }, + { + "epoch": 1.8872763255545477, + "grad_norm": 2.3250646591186523, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7042480707168579, + "num_tokens": 578143644.0, + "step": 22887 + }, + { + "epoch": 1.8873587861795993, + "grad_norm": 2.0479252338409424, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7267444133758545, + "num_tokens": 578174655.0, + "step": 22888 + }, + { + "epoch": 1.8874412468046509, + "grad_norm": 2.0949387550354004, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7611851692199707, + "num_tokens": 578200934.0, + "step": 22889 + }, + { + "epoch": 1.8875237074297022, + "grad_norm": 1.8696753978729248, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.788963794708252, + "num_tokens": 578231002.0, + "step": 22890 + }, + { + "epoch": 1.887606168054754, + "grad_norm": 2.1455631256103516, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7755739688873291, + "num_tokens": 578256215.0, + "step": 22891 + }, + { + "epoch": 1.8876886286798054, + "grad_norm": 2.132904052734375, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7627201080322266, + "num_tokens": 578284388.0, + "step": 22892 + }, + { + "epoch": 1.887771089304857, + "grad_norm": 2.1770331859588623, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7382027506828308, + "num_tokens": 578309255.0, + "step": 22893 + }, + { + "epoch": 1.8878535499299085, + "grad_norm": 1.8382117748260498, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.7876687049865723, + "num_tokens": 578341769.0, + "step": 22894 + }, + { + "epoch": 1.8879360105549599, + "grad_norm": 1.8491789102554321, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7694922089576721, + "num_tokens": 578372882.0, + "step": 22895 + }, + { + "epoch": 1.8880184711800116, + "grad_norm": 1.8976134061813354, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7578192353248596, + "num_tokens": 578407165.0, + "step": 22896 + }, + { + "epoch": 1.888100931805063, + "grad_norm": 2.1140682697296143, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.7897380590438843, + "num_tokens": 578436306.0, + "step": 22897 + }, + { + "epoch": 1.8881833924301146, + "grad_norm": 2.343679428100586, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7238516807556152, + "num_tokens": 578462277.0, + "step": 22898 + }, + { + "epoch": 1.8882658530551661, + "grad_norm": 1.8281285762786865, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7839295268058777, + "num_tokens": 578496950.0, + "step": 22899 + }, + { + "epoch": 1.8883483136802177, + "grad_norm": 2.07887864112854, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7583940029144287, + "num_tokens": 578524452.0, + "step": 22900 + }, + { + "epoch": 1.8884307743052693, + "grad_norm": 2.0108330249786377, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7773011922836304, + "num_tokens": 578553453.0, + "step": 22901 + }, + { + "epoch": 1.8885132349303206, + "grad_norm": 2.1424925327301025, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7438096404075623, + "num_tokens": 578580448.0, + "step": 22902 + }, + { + "epoch": 1.8885956955553724, + "grad_norm": 1.7479195594787598, + "learning_rate": 1e-06, + "loss": 0.5014, + "mean_token_accuracy": 0.8208553791046143, + "num_tokens": 578614732.0, + "step": 22903 + }, + { + "epoch": 1.8886781561804238, + "grad_norm": 2.114098072052002, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7701596021652222, + "num_tokens": 578640839.0, + "step": 22904 + }, + { + "epoch": 1.8887606168054754, + "grad_norm": 2.0879640579223633, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.743464469909668, + "num_tokens": 578667664.0, + "step": 22905 + }, + { + "epoch": 1.888843077430527, + "grad_norm": 1.785986304283142, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7733365893363953, + "num_tokens": 578702273.0, + "step": 22906 + }, + { + "epoch": 1.8889255380555785, + "grad_norm": 1.8243786096572876, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.7838451862335205, + "num_tokens": 578735232.0, + "step": 22907 + }, + { + "epoch": 1.88900799868063, + "grad_norm": 2.1842682361602783, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.7888350486755371, + "num_tokens": 578764921.0, + "step": 22908 + }, + { + "epoch": 1.8890904593056814, + "grad_norm": 2.11946964263916, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7428880929946899, + "num_tokens": 578792695.0, + "step": 22909 + }, + { + "epoch": 1.8891729199307332, + "grad_norm": 1.9963462352752686, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.7746924161911011, + "num_tokens": 578821085.0, + "step": 22910 + }, + { + "epoch": 1.8892553805557846, + "grad_norm": 2.1581616401672363, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7595888376235962, + "num_tokens": 578850509.0, + "step": 22911 + }, + { + "epoch": 1.8893378411808361, + "grad_norm": 2.259946584701538, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7610243558883667, + "num_tokens": 578874548.0, + "step": 22912 + }, + { + "epoch": 1.8894203018058877, + "grad_norm": 2.0514516830444336, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7335744500160217, + "num_tokens": 578903737.0, + "step": 22913 + }, + { + "epoch": 1.889502762430939, + "grad_norm": 1.9419751167297363, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7412213087081909, + "num_tokens": 578934623.0, + "step": 22914 + }, + { + "epoch": 1.8895852230559909, + "grad_norm": 2.0757346153259277, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7473726272583008, + "num_tokens": 578961341.0, + "step": 22915 + }, + { + "epoch": 1.8896676836810422, + "grad_norm": 2.47342586517334, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7462326288223267, + "num_tokens": 578982148.0, + "step": 22916 + }, + { + "epoch": 1.889750144306094, + "grad_norm": 1.9288623332977295, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.766484797000885, + "num_tokens": 579011434.0, + "step": 22917 + }, + { + "epoch": 1.8898326049311454, + "grad_norm": 1.9878534078598022, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.763403594493866, + "num_tokens": 579044108.0, + "step": 22918 + }, + { + "epoch": 1.889915065556197, + "grad_norm": 1.9770451784133911, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.7852266430854797, + "num_tokens": 579074042.0, + "step": 22919 + }, + { + "epoch": 1.8899975261812485, + "grad_norm": 2.6457467079162598, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7299402356147766, + "num_tokens": 579092821.0, + "step": 22920 + }, + { + "epoch": 1.8900799868062998, + "grad_norm": 2.0237271785736084, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.7921162843704224, + "num_tokens": 579120906.0, + "step": 22921 + }, + { + "epoch": 1.8901624474313516, + "grad_norm": 1.9593725204467773, + "learning_rate": 1e-06, + "loss": 0.5671, + "mean_token_accuracy": 0.8038973808288574, + "num_tokens": 579146521.0, + "step": 22922 + }, + { + "epoch": 1.890244908056403, + "grad_norm": 1.8663848638534546, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7707717418670654, + "num_tokens": 579180013.0, + "step": 22923 + }, + { + "epoch": 1.8903273686814546, + "grad_norm": 2.01407790184021, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7397012710571289, + "num_tokens": 579212466.0, + "step": 22924 + }, + { + "epoch": 1.8904098293065061, + "grad_norm": 2.0219788551330566, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.7610282897949219, + "num_tokens": 579241033.0, + "step": 22925 + }, + { + "epoch": 1.8904922899315577, + "grad_norm": 1.8677979707717896, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7519248127937317, + "num_tokens": 579277698.0, + "step": 22926 + }, + { + "epoch": 1.8905747505566093, + "grad_norm": 2.0982556343078613, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7782538533210754, + "num_tokens": 579306726.0, + "step": 22927 + }, + { + "epoch": 1.8906572111816606, + "grad_norm": 1.7624430656433105, + "learning_rate": 1e-06, + "loss": 0.5772, + "mean_token_accuracy": 0.8116987943649292, + "num_tokens": 579341469.0, + "step": 22928 + }, + { + "epoch": 1.8907396718067124, + "grad_norm": 2.220724582672119, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7733724117279053, + "num_tokens": 579366664.0, + "step": 22929 + }, + { + "epoch": 1.8908221324317638, + "grad_norm": 2.1971333026885986, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7714808583259583, + "num_tokens": 579390893.0, + "step": 22930 + }, + { + "epoch": 1.8909045930568154, + "grad_norm": 2.152230978012085, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7717063426971436, + "num_tokens": 579416135.0, + "step": 22931 + }, + { + "epoch": 1.890987053681867, + "grad_norm": 2.0490775108337402, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7387893199920654, + "num_tokens": 579444066.0, + "step": 22932 + }, + { + "epoch": 1.8910695143069185, + "grad_norm": 2.1901695728302, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7241356372833252, + "num_tokens": 579470258.0, + "step": 22933 + }, + { + "epoch": 1.89115197493197, + "grad_norm": 1.9956039190292358, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7556934952735901, + "num_tokens": 579501285.0, + "step": 22934 + }, + { + "epoch": 1.8912344355570214, + "grad_norm": 2.1684210300445557, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7294481992721558, + "num_tokens": 579528174.0, + "step": 22935 + }, + { + "epoch": 1.8913168961820732, + "grad_norm": 2.320138931274414, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7673958539962769, + "num_tokens": 579553375.0, + "step": 22936 + }, + { + "epoch": 1.8913993568071246, + "grad_norm": 2.0754811763763428, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7548710107803345, + "num_tokens": 579582423.0, + "step": 22937 + }, + { + "epoch": 1.8914818174321761, + "grad_norm": 1.8365696668624878, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.761811375617981, + "num_tokens": 579617813.0, + "step": 22938 + }, + { + "epoch": 1.8915642780572277, + "grad_norm": 2.198425769805908, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.772663950920105, + "num_tokens": 579643496.0, + "step": 22939 + }, + { + "epoch": 1.891646738682279, + "grad_norm": 1.9689191579818726, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7538522481918335, + "num_tokens": 579675176.0, + "step": 22940 + }, + { + "epoch": 1.8917291993073309, + "grad_norm": 2.1677117347717285, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7726836204528809, + "num_tokens": 579701973.0, + "step": 22941 + }, + { + "epoch": 1.8918116599323822, + "grad_norm": 1.7428897619247437, + "learning_rate": 1e-06, + "loss": 0.4991, + "mean_token_accuracy": 0.8165277242660522, + "num_tokens": 579731823.0, + "step": 22942 + }, + { + "epoch": 1.8918941205574338, + "grad_norm": 2.139777660369873, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7345826625823975, + "num_tokens": 579761215.0, + "step": 22943 + }, + { + "epoch": 1.8919765811824854, + "grad_norm": 2.1089682579040527, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7269595861434937, + "num_tokens": 579788059.0, + "step": 22944 + }, + { + "epoch": 1.892059041807537, + "grad_norm": 2.195647716522217, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7524402737617493, + "num_tokens": 579814687.0, + "step": 22945 + }, + { + "epoch": 1.8921415024325885, + "grad_norm": 2.103386402130127, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7814664840698242, + "num_tokens": 579841162.0, + "step": 22946 + }, + { + "epoch": 1.8922239630576398, + "grad_norm": 2.025308609008789, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.731127142906189, + "num_tokens": 579870736.0, + "step": 22947 + }, + { + "epoch": 1.8923064236826916, + "grad_norm": 1.9557123184204102, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7731908559799194, + "num_tokens": 579900228.0, + "step": 22948 + }, + { + "epoch": 1.892388884307743, + "grad_norm": 1.8602226972579956, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.778922438621521, + "num_tokens": 579931965.0, + "step": 22949 + }, + { + "epoch": 1.8924713449327946, + "grad_norm": 2.0704345703125, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7544534206390381, + "num_tokens": 579962456.0, + "step": 22950 + }, + { + "epoch": 1.8925538055578461, + "grad_norm": 2.0032601356506348, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7583622336387634, + "num_tokens": 579992289.0, + "step": 22951 + }, + { + "epoch": 1.8926362661828977, + "grad_norm": 2.109790325164795, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.717586874961853, + "num_tokens": 580019527.0, + "step": 22952 + }, + { + "epoch": 1.8927187268079493, + "grad_norm": 2.0746073722839355, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7264584302902222, + "num_tokens": 580050239.0, + "step": 22953 + }, + { + "epoch": 1.8928011874330006, + "grad_norm": 1.934728741645813, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7562078833580017, + "num_tokens": 580084653.0, + "step": 22954 + }, + { + "epoch": 1.8928836480580524, + "grad_norm": 1.7742853164672852, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.787436842918396, + "num_tokens": 580116685.0, + "step": 22955 + }, + { + "epoch": 1.8929661086831038, + "grad_norm": 1.894830346107483, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7314276099205017, + "num_tokens": 580148832.0, + "step": 22956 + }, + { + "epoch": 1.8930485693081553, + "grad_norm": 1.9997310638427734, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7328512072563171, + "num_tokens": 580181119.0, + "step": 22957 + }, + { + "epoch": 1.893131029933207, + "grad_norm": 1.939920425415039, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7204673886299133, + "num_tokens": 580214967.0, + "step": 22958 + }, + { + "epoch": 1.8932134905582583, + "grad_norm": 1.7470322847366333, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.784711480140686, + "num_tokens": 580247718.0, + "step": 22959 + }, + { + "epoch": 1.89329595118331, + "grad_norm": 1.9003359079360962, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.8010412454605103, + "num_tokens": 580277647.0, + "step": 22960 + }, + { + "epoch": 1.8933784118083614, + "grad_norm": 2.1945505142211914, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7387130856513977, + "num_tokens": 580302888.0, + "step": 22961 + }, + { + "epoch": 1.8934608724334132, + "grad_norm": 2.0863938331604004, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.761336088180542, + "num_tokens": 580333722.0, + "step": 22962 + }, + { + "epoch": 1.8935433330584646, + "grad_norm": 2.2594966888427734, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7389247417449951, + "num_tokens": 580361366.0, + "step": 22963 + }, + { + "epoch": 1.8936257936835161, + "grad_norm": 2.1684393882751465, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7575389742851257, + "num_tokens": 580386334.0, + "step": 22964 + }, + { + "epoch": 1.8937082543085677, + "grad_norm": 1.8780481815338135, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7325643301010132, + "num_tokens": 580421210.0, + "step": 22965 + }, + { + "epoch": 1.893790714933619, + "grad_norm": 1.9627729654312134, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7737630605697632, + "num_tokens": 580452646.0, + "step": 22966 + }, + { + "epoch": 1.8938731755586709, + "grad_norm": 1.9585093259811401, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7631033658981323, + "num_tokens": 580482693.0, + "step": 22967 + }, + { + "epoch": 1.8939556361837222, + "grad_norm": 2.1941068172454834, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7521989345550537, + "num_tokens": 580508325.0, + "step": 22968 + }, + { + "epoch": 1.8940380968087738, + "grad_norm": 1.8939632177352905, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7614365220069885, + "num_tokens": 580537539.0, + "step": 22969 + }, + { + "epoch": 1.8941205574338253, + "grad_norm": 1.8899569511413574, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.7973635196685791, + "num_tokens": 580569919.0, + "step": 22970 + }, + { + "epoch": 1.894203018058877, + "grad_norm": 1.9273583889007568, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.779489278793335, + "num_tokens": 580603057.0, + "step": 22971 + }, + { + "epoch": 1.8942854786839285, + "grad_norm": 1.958361268043518, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7684007287025452, + "num_tokens": 580634268.0, + "step": 22972 + }, + { + "epoch": 1.8943679393089798, + "grad_norm": 1.849715232849121, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7660443782806396, + "num_tokens": 580667906.0, + "step": 22973 + }, + { + "epoch": 1.8944503999340316, + "grad_norm": 1.7384626865386963, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.774537205696106, + "num_tokens": 580705292.0, + "step": 22974 + }, + { + "epoch": 1.894532860559083, + "grad_norm": 2.119479179382324, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.734768271446228, + "num_tokens": 580734749.0, + "step": 22975 + }, + { + "epoch": 1.8946153211841346, + "grad_norm": 1.7562464475631714, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7808029651641846, + "num_tokens": 580768234.0, + "step": 22976 + }, + { + "epoch": 1.8946977818091861, + "grad_norm": 2.0835535526275635, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7176016569137573, + "num_tokens": 580799818.0, + "step": 22977 + }, + { + "epoch": 1.8947802424342377, + "grad_norm": 1.8123300075531006, + "learning_rate": 1e-06, + "loss": 0.533, + "mean_token_accuracy": 0.8214735984802246, + "num_tokens": 580832451.0, + "step": 22978 + }, + { + "epoch": 1.8948627030592893, + "grad_norm": 1.9336485862731934, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7446378469467163, + "num_tokens": 580867646.0, + "step": 22979 + }, + { + "epoch": 1.8949451636843406, + "grad_norm": 2.254016160964966, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7380038499832153, + "num_tokens": 580892081.0, + "step": 22980 + }, + { + "epoch": 1.8950276243093924, + "grad_norm": 1.9996155500411987, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7497369647026062, + "num_tokens": 580922821.0, + "step": 22981 + }, + { + "epoch": 1.8951100849344438, + "grad_norm": 1.8658596277236938, + "learning_rate": 1e-06, + "loss": 0.594, + "mean_token_accuracy": 0.8032460808753967, + "num_tokens": 580954902.0, + "step": 22982 + }, + { + "epoch": 1.8951925455594953, + "grad_norm": 1.844734787940979, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7753845453262329, + "num_tokens": 580986368.0, + "step": 22983 + }, + { + "epoch": 1.895275006184547, + "grad_norm": 1.9853187799453735, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7631100416183472, + "num_tokens": 581014366.0, + "step": 22984 + }, + { + "epoch": 1.8953574668095983, + "grad_norm": 2.1610281467437744, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7298446297645569, + "num_tokens": 581044871.0, + "step": 22985 + }, + { + "epoch": 1.89543992743465, + "grad_norm": 2.014089822769165, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7265985608100891, + "num_tokens": 581075024.0, + "step": 22986 + }, + { + "epoch": 1.8955223880597014, + "grad_norm": 1.772835373878479, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7876743078231812, + "num_tokens": 581107869.0, + "step": 22987 + }, + { + "epoch": 1.895604848684753, + "grad_norm": 1.967509388923645, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7620725631713867, + "num_tokens": 581136972.0, + "step": 22988 + }, + { + "epoch": 1.8956873093098046, + "grad_norm": 1.9426451921463013, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7589956521987915, + "num_tokens": 581169798.0, + "step": 22989 + }, + { + "epoch": 1.8957697699348561, + "grad_norm": 2.3481011390686035, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7662298679351807, + "num_tokens": 581193460.0, + "step": 22990 + }, + { + "epoch": 1.8958522305599077, + "grad_norm": 2.0407402515411377, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7269518375396729, + "num_tokens": 581222748.0, + "step": 22991 + }, + { + "epoch": 1.895934691184959, + "grad_norm": 1.9633673429489136, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.7844123840332031, + "num_tokens": 581252482.0, + "step": 22992 + }, + { + "epoch": 1.8960171518100108, + "grad_norm": 1.7832478284835815, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7832022905349731, + "num_tokens": 581287460.0, + "step": 22993 + }, + { + "epoch": 1.8960996124350622, + "grad_norm": 2.1277694702148438, + "learning_rate": 1e-06, + "loss": 0.6173, + "mean_token_accuracy": 0.7856249213218689, + "num_tokens": 581313244.0, + "step": 22994 + }, + { + "epoch": 1.8961820730601138, + "grad_norm": 1.977081298828125, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7796429395675659, + "num_tokens": 581346168.0, + "step": 22995 + }, + { + "epoch": 1.8962645336851653, + "grad_norm": 2.100600242614746, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7349308133125305, + "num_tokens": 581375665.0, + "step": 22996 + }, + { + "epoch": 1.896346994310217, + "grad_norm": 1.788923978805542, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7678833603858948, + "num_tokens": 581410800.0, + "step": 22997 + }, + { + "epoch": 1.8964294549352685, + "grad_norm": 2.2180137634277344, + "learning_rate": 1e-06, + "loss": 0.5909, + "mean_token_accuracy": 0.7986881732940674, + "num_tokens": 581433674.0, + "step": 22998 + }, + { + "epoch": 1.8965119155603198, + "grad_norm": 2.0249569416046143, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7819380164146423, + "num_tokens": 581464128.0, + "step": 22999 + }, + { + "epoch": 1.8965943761853716, + "grad_norm": 2.6548893451690674, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7754330635070801, + "num_tokens": 581491337.0, + "step": 23000 + }, + { + "epoch": 1.896676836810423, + "grad_norm": 1.91492760181427, + "learning_rate": 1e-06, + "loss": 0.5745, + "mean_token_accuracy": 0.8045691251754761, + "num_tokens": 581519937.0, + "step": 23001 + }, + { + "epoch": 1.8967592974354746, + "grad_norm": 1.89944589138031, + "learning_rate": 1e-06, + "loss": 0.7868, + "mean_token_accuracy": 0.7536144256591797, + "num_tokens": 581555582.0, + "step": 23002 + }, + { + "epoch": 1.8968417580605261, + "grad_norm": 2.085488796234131, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7245823740959167, + "num_tokens": 581584308.0, + "step": 23003 + }, + { + "epoch": 1.8969242186855775, + "grad_norm": 1.7708746194839478, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7652478218078613, + "num_tokens": 581622347.0, + "step": 23004 + }, + { + "epoch": 1.8970066793106293, + "grad_norm": 2.208378314971924, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7573870420455933, + "num_tokens": 581646993.0, + "step": 23005 + }, + { + "epoch": 1.8970891399356806, + "grad_norm": 2.10467267036438, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.7903856039047241, + "num_tokens": 581672843.0, + "step": 23006 + }, + { + "epoch": 1.8971716005607322, + "grad_norm": 1.9467188119888306, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.7969783544540405, + "num_tokens": 581703128.0, + "step": 23007 + }, + { + "epoch": 1.8972540611857838, + "grad_norm": 2.207249879837036, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7624855041503906, + "num_tokens": 581729453.0, + "step": 23008 + }, + { + "epoch": 1.8973365218108353, + "grad_norm": 1.924878716468811, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7536348700523376, + "num_tokens": 581764159.0, + "step": 23009 + }, + { + "epoch": 1.897418982435887, + "grad_norm": 1.9540483951568604, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7747520208358765, + "num_tokens": 581797288.0, + "step": 23010 + }, + { + "epoch": 1.8975014430609383, + "grad_norm": 1.9600143432617188, + "learning_rate": 1e-06, + "loss": 0.7621, + "mean_token_accuracy": 0.7601770162582397, + "num_tokens": 581827822.0, + "step": 23011 + }, + { + "epoch": 1.89758390368599, + "grad_norm": 1.7384006977081299, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.7918216586112976, + "num_tokens": 581863311.0, + "step": 23012 + }, + { + "epoch": 1.8976663643110414, + "grad_norm": 1.830618143081665, + "learning_rate": 1e-06, + "loss": 0.5603, + "mean_token_accuracy": 0.8167383670806885, + "num_tokens": 581894525.0, + "step": 23013 + }, + { + "epoch": 1.897748824936093, + "grad_norm": 1.7652325630187988, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7829761505126953, + "num_tokens": 581928816.0, + "step": 23014 + }, + { + "epoch": 1.8978312855611446, + "grad_norm": 2.0036497116088867, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7504249811172485, + "num_tokens": 581959542.0, + "step": 23015 + }, + { + "epoch": 1.8979137461861961, + "grad_norm": 2.012655258178711, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7461081147193909, + "num_tokens": 581987146.0, + "step": 23016 + }, + { + "epoch": 1.8979962068112477, + "grad_norm": 2.241227626800537, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7638741731643677, + "num_tokens": 582009867.0, + "step": 23017 + }, + { + "epoch": 1.898078667436299, + "grad_norm": 2.130213737487793, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7531306147575378, + "num_tokens": 582035705.0, + "step": 23018 + }, + { + "epoch": 1.8981611280613508, + "grad_norm": 1.8677648305892944, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7701606750488281, + "num_tokens": 582067652.0, + "step": 23019 + }, + { + "epoch": 1.8982435886864022, + "grad_norm": 2.1714253425598145, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7622120380401611, + "num_tokens": 582092892.0, + "step": 23020 + }, + { + "epoch": 1.8983260493114538, + "grad_norm": 2.007340908050537, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7462955713272095, + "num_tokens": 582123223.0, + "step": 23021 + }, + { + "epoch": 1.8984085099365053, + "grad_norm": 1.9339516162872314, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7795403599739075, + "num_tokens": 582155362.0, + "step": 23022 + }, + { + "epoch": 1.898490970561557, + "grad_norm": 1.8289886713027954, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7576950192451477, + "num_tokens": 582191477.0, + "step": 23023 + }, + { + "epoch": 1.8985734311866085, + "grad_norm": 1.8697378635406494, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.8024200201034546, + "num_tokens": 582223996.0, + "step": 23024 + }, + { + "epoch": 1.8986558918116598, + "grad_norm": 1.8370543718338013, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7472164630889893, + "num_tokens": 582259255.0, + "step": 23025 + }, + { + "epoch": 1.8987383524367116, + "grad_norm": 2.185560464859009, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7548976540565491, + "num_tokens": 582283385.0, + "step": 23026 + }, + { + "epoch": 1.898820813061763, + "grad_norm": 1.8800662755966187, + "learning_rate": 1e-06, + "loss": 0.5983, + "mean_token_accuracy": 0.8145624995231628, + "num_tokens": 582312868.0, + "step": 23027 + }, + { + "epoch": 1.8989032736868146, + "grad_norm": 2.0319366455078125, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7770242691040039, + "num_tokens": 582341641.0, + "step": 23028 + }, + { + "epoch": 1.8989857343118661, + "grad_norm": 1.98577082157135, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7202174663543701, + "num_tokens": 582373829.0, + "step": 23029 + }, + { + "epoch": 1.8990681949369175, + "grad_norm": 2.1745781898498535, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7448129057884216, + "num_tokens": 582402841.0, + "step": 23030 + }, + { + "epoch": 1.8991506555619693, + "grad_norm": 1.9016999006271362, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7385163307189941, + "num_tokens": 582435775.0, + "step": 23031 + }, + { + "epoch": 1.8992331161870206, + "grad_norm": 1.98582923412323, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7389048933982849, + "num_tokens": 582465640.0, + "step": 23032 + }, + { + "epoch": 1.8993155768120722, + "grad_norm": 1.8675998449325562, + "learning_rate": 1e-06, + "loss": 0.5347, + "mean_token_accuracy": 0.8250129818916321, + "num_tokens": 582500465.0, + "step": 23033 + }, + { + "epoch": 1.8993980374371238, + "grad_norm": 1.9456266164779663, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7576435804367065, + "num_tokens": 582529325.0, + "step": 23034 + }, + { + "epoch": 1.8994804980621753, + "grad_norm": 2.1113264560699463, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7580286264419556, + "num_tokens": 582558208.0, + "step": 23035 + }, + { + "epoch": 1.899562958687227, + "grad_norm": 1.9916483163833618, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7615771293640137, + "num_tokens": 582586235.0, + "step": 23036 + }, + { + "epoch": 1.8996454193122783, + "grad_norm": 1.8441683053970337, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7507925629615784, + "num_tokens": 582619410.0, + "step": 23037 + }, + { + "epoch": 1.89972787993733, + "grad_norm": 1.8864364624023438, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7654558420181274, + "num_tokens": 582651497.0, + "step": 23038 + }, + { + "epoch": 1.8998103405623814, + "grad_norm": 2.3257503509521484, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7595374584197998, + "num_tokens": 582675339.0, + "step": 23039 + }, + { + "epoch": 1.899892801187433, + "grad_norm": 2.094087839126587, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7596919536590576, + "num_tokens": 582701651.0, + "step": 23040 + }, + { + "epoch": 1.8999752618124846, + "grad_norm": 2.0697717666625977, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7082804441452026, + "num_tokens": 582730065.0, + "step": 23041 + }, + { + "epoch": 1.9000577224375361, + "grad_norm": 2.1026601791381836, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7708618640899658, + "num_tokens": 582759888.0, + "step": 23042 + }, + { + "epoch": 1.9001401830625877, + "grad_norm": 1.6881060600280762, + "learning_rate": 1e-06, + "loss": 0.6021, + "mean_token_accuracy": 0.8038771152496338, + "num_tokens": 582800199.0, + "step": 23043 + }, + { + "epoch": 1.900222643687639, + "grad_norm": 2.134385347366333, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7596723437309265, + "num_tokens": 582827614.0, + "step": 23044 + }, + { + "epoch": 1.9003051043126908, + "grad_norm": 2.1539969444274902, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7302371263504028, + "num_tokens": 582855145.0, + "step": 23045 + }, + { + "epoch": 1.9003875649377422, + "grad_norm": 2.071002960205078, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7092411518096924, + "num_tokens": 582884571.0, + "step": 23046 + }, + { + "epoch": 1.9004700255627938, + "grad_norm": 1.901653528213501, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7399329543113708, + "num_tokens": 582919422.0, + "step": 23047 + }, + { + "epoch": 1.9005524861878453, + "grad_norm": 2.0500643253326416, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7469632625579834, + "num_tokens": 582946245.0, + "step": 23048 + }, + { + "epoch": 1.9006349468128967, + "grad_norm": 2.1201717853546143, + "learning_rate": 1e-06, + "loss": 0.6083, + "mean_token_accuracy": 0.7916889190673828, + "num_tokens": 582969942.0, + "step": 23049 + }, + { + "epoch": 1.9007174074379485, + "grad_norm": 2.136795997619629, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7632470726966858, + "num_tokens": 582999544.0, + "step": 23050 + }, + { + "epoch": 1.9007998680629998, + "grad_norm": 1.8078358173370361, + "learning_rate": 1e-06, + "loss": 0.6436, + "mean_token_accuracy": 0.797461986541748, + "num_tokens": 583030098.0, + "step": 23051 + }, + { + "epoch": 1.9008823286880514, + "grad_norm": 2.204922676086426, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7470754384994507, + "num_tokens": 583055241.0, + "step": 23052 + }, + { + "epoch": 1.900964789313103, + "grad_norm": 2.099280834197998, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7274415493011475, + "num_tokens": 583084745.0, + "step": 23053 + }, + { + "epoch": 1.9010472499381545, + "grad_norm": 1.9389293193817139, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.781699538230896, + "num_tokens": 583112176.0, + "step": 23054 + }, + { + "epoch": 1.9011297105632061, + "grad_norm": 2.148190975189209, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7694028615951538, + "num_tokens": 583138577.0, + "step": 23055 + }, + { + "epoch": 1.9012121711882575, + "grad_norm": 2.016810655593872, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7491227388381958, + "num_tokens": 583167556.0, + "step": 23056 + }, + { + "epoch": 1.9012946318133093, + "grad_norm": 1.8585513830184937, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7255740165710449, + "num_tokens": 583200400.0, + "step": 23057 + }, + { + "epoch": 1.9013770924383606, + "grad_norm": 1.9816668033599854, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7646336555480957, + "num_tokens": 583232065.0, + "step": 23058 + }, + { + "epoch": 1.9014595530634122, + "grad_norm": 1.7491133213043213, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7662773132324219, + "num_tokens": 583265785.0, + "step": 23059 + }, + { + "epoch": 1.9015420136884638, + "grad_norm": 1.8810532093048096, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7776668071746826, + "num_tokens": 583296551.0, + "step": 23060 + }, + { + "epoch": 1.9016244743135153, + "grad_norm": 1.7731622457504272, + "learning_rate": 1e-06, + "loss": 0.5898, + "mean_token_accuracy": 0.8070192337036133, + "num_tokens": 583329761.0, + "step": 23061 + }, + { + "epoch": 1.901706934938567, + "grad_norm": 1.777343988418579, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.7984696626663208, + "num_tokens": 583361918.0, + "step": 23062 + }, + { + "epoch": 1.9017893955636183, + "grad_norm": 1.9976234436035156, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.7842862010002136, + "num_tokens": 583392242.0, + "step": 23063 + }, + { + "epoch": 1.90187185618867, + "grad_norm": 2.0715487003326416, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7598031759262085, + "num_tokens": 583417418.0, + "step": 23064 + }, + { + "epoch": 1.9019543168137214, + "grad_norm": 1.9040273427963257, + "learning_rate": 1e-06, + "loss": 0.5473, + "mean_token_accuracy": 0.8183860778808594, + "num_tokens": 583447188.0, + "step": 23065 + }, + { + "epoch": 1.902036777438773, + "grad_norm": 2.044809579849243, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.731597363948822, + "num_tokens": 583477520.0, + "step": 23066 + }, + { + "epoch": 1.9021192380638245, + "grad_norm": 1.9025416374206543, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.7863941192626953, + "num_tokens": 583508450.0, + "step": 23067 + }, + { + "epoch": 1.902201698688876, + "grad_norm": 2.0191409587860107, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.784106969833374, + "num_tokens": 583537195.0, + "step": 23068 + }, + { + "epoch": 1.9022841593139277, + "grad_norm": 2.1520862579345703, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7261622548103333, + "num_tokens": 583564138.0, + "step": 23069 + }, + { + "epoch": 1.902366619938979, + "grad_norm": 1.7873808145523071, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7572380900382996, + "num_tokens": 583600297.0, + "step": 23070 + }, + { + "epoch": 1.9024490805640308, + "grad_norm": 2.1533148288726807, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7505075931549072, + "num_tokens": 583625370.0, + "step": 23071 + }, + { + "epoch": 1.9025315411890822, + "grad_norm": 1.991296410560608, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.7544934749603271, + "num_tokens": 583657050.0, + "step": 23072 + }, + { + "epoch": 1.9026140018141338, + "grad_norm": 2.2512385845184326, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7654135823249817, + "num_tokens": 583680140.0, + "step": 23073 + }, + { + "epoch": 1.9026964624391853, + "grad_norm": 1.876816749572754, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.777062714099884, + "num_tokens": 583713128.0, + "step": 23074 + }, + { + "epoch": 1.9027789230642367, + "grad_norm": 1.8209717273712158, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7618057131767273, + "num_tokens": 583751528.0, + "step": 23075 + }, + { + "epoch": 1.9028613836892885, + "grad_norm": 1.8243012428283691, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7314397096633911, + "num_tokens": 583788536.0, + "step": 23076 + }, + { + "epoch": 1.9029438443143398, + "grad_norm": 1.9985007047653198, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7644040584564209, + "num_tokens": 583817947.0, + "step": 23077 + }, + { + "epoch": 1.9030263049393914, + "grad_norm": 1.7777431011199951, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7572754621505737, + "num_tokens": 583856267.0, + "step": 23078 + }, + { + "epoch": 1.903108765564443, + "grad_norm": 1.7205830812454224, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7343512177467346, + "num_tokens": 583894118.0, + "step": 23079 + }, + { + "epoch": 1.9031912261894945, + "grad_norm": 2.149475574493408, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.749513566493988, + "num_tokens": 583918495.0, + "step": 23080 + }, + { + "epoch": 1.9032736868145461, + "grad_norm": 2.052919626235962, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7565178871154785, + "num_tokens": 583946374.0, + "step": 23081 + }, + { + "epoch": 1.9033561474395975, + "grad_norm": 2.0381088256835938, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7667462825775146, + "num_tokens": 583975126.0, + "step": 23082 + }, + { + "epoch": 1.9034386080646493, + "grad_norm": 1.7289161682128906, + "learning_rate": 1e-06, + "loss": 0.5833, + "mean_token_accuracy": 0.8060829639434814, + "num_tokens": 584009590.0, + "step": 23083 + }, + { + "epoch": 1.9035210686897006, + "grad_norm": 1.890741229057312, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7646856307983398, + "num_tokens": 584042683.0, + "step": 23084 + }, + { + "epoch": 1.9036035293147522, + "grad_norm": 1.9368113279342651, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.779944658279419, + "num_tokens": 584073335.0, + "step": 23085 + }, + { + "epoch": 1.9036859899398038, + "grad_norm": 2.19620680809021, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7181771993637085, + "num_tokens": 584100338.0, + "step": 23086 + }, + { + "epoch": 1.9037684505648553, + "grad_norm": 1.8640943765640259, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7834877371788025, + "num_tokens": 584131135.0, + "step": 23087 + }, + { + "epoch": 1.903850911189907, + "grad_norm": 2.2310543060302734, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7657381296157837, + "num_tokens": 584156450.0, + "step": 23088 + }, + { + "epoch": 1.9039333718149583, + "grad_norm": 2.337127208709717, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7624874114990234, + "num_tokens": 584179168.0, + "step": 23089 + }, + { + "epoch": 1.90401583244001, + "grad_norm": 1.6552753448486328, + "learning_rate": 1e-06, + "loss": 0.5882, + "mean_token_accuracy": 0.8048892021179199, + "num_tokens": 584218956.0, + "step": 23090 + }, + { + "epoch": 1.9040982930650614, + "grad_norm": 1.6911383867263794, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7804440259933472, + "num_tokens": 584256754.0, + "step": 23091 + }, + { + "epoch": 1.904180753690113, + "grad_norm": 1.842604160308838, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7421216368675232, + "num_tokens": 584290415.0, + "step": 23092 + }, + { + "epoch": 1.9042632143151645, + "grad_norm": 1.9699983596801758, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.7912083268165588, + "num_tokens": 584322802.0, + "step": 23093 + }, + { + "epoch": 1.904345674940216, + "grad_norm": 1.6854116916656494, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.8077756762504578, + "num_tokens": 584357198.0, + "step": 23094 + }, + { + "epoch": 1.9044281355652677, + "grad_norm": 1.814388632774353, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7592290639877319, + "num_tokens": 584391854.0, + "step": 23095 + }, + { + "epoch": 1.904510596190319, + "grad_norm": 1.7844631671905518, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.7999963760375977, + "num_tokens": 584424015.0, + "step": 23096 + }, + { + "epoch": 1.9045930568153706, + "grad_norm": 1.8206391334533691, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7798300981521606, + "num_tokens": 584458197.0, + "step": 23097 + }, + { + "epoch": 1.9046755174404222, + "grad_norm": 2.1086843013763428, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7754483222961426, + "num_tokens": 584484970.0, + "step": 23098 + }, + { + "epoch": 1.9047579780654738, + "grad_norm": 1.6938687562942505, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7837706804275513, + "num_tokens": 584523779.0, + "step": 23099 + }, + { + "epoch": 1.9048404386905253, + "grad_norm": 2.1872916221618652, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7608886957168579, + "num_tokens": 584551444.0, + "step": 23100 + }, + { + "epoch": 1.9049228993155767, + "grad_norm": 1.7883546352386475, + "learning_rate": 1e-06, + "loss": 0.564, + "mean_token_accuracy": 0.8030881881713867, + "num_tokens": 584584342.0, + "step": 23101 + }, + { + "epoch": 1.9050053599406285, + "grad_norm": 2.0822293758392334, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.8005517721176147, + "num_tokens": 584609488.0, + "step": 23102 + }, + { + "epoch": 1.9050878205656798, + "grad_norm": 1.9395735263824463, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7675087451934814, + "num_tokens": 584639269.0, + "step": 23103 + }, + { + "epoch": 1.9051702811907314, + "grad_norm": 1.9416122436523438, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.7949374318122864, + "num_tokens": 584668741.0, + "step": 23104 + }, + { + "epoch": 1.905252741815783, + "grad_norm": 2.1957778930664062, + "learning_rate": 1e-06, + "loss": 0.7799, + "mean_token_accuracy": 0.7342097163200378, + "num_tokens": 584695372.0, + "step": 23105 + }, + { + "epoch": 1.9053352024408345, + "grad_norm": 2.0991694927215576, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7595539689064026, + "num_tokens": 584720634.0, + "step": 23106 + }, + { + "epoch": 1.9054176630658861, + "grad_norm": 1.8699482679367065, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7399531006813049, + "num_tokens": 584753618.0, + "step": 23107 + }, + { + "epoch": 1.9055001236909375, + "grad_norm": 1.8100066184997559, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7471206188201904, + "num_tokens": 584787390.0, + "step": 23108 + }, + { + "epoch": 1.9055825843159893, + "grad_norm": 2.1159019470214844, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7360507845878601, + "num_tokens": 584813044.0, + "step": 23109 + }, + { + "epoch": 1.9056650449410406, + "grad_norm": 1.9241094589233398, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7577815055847168, + "num_tokens": 584843542.0, + "step": 23110 + }, + { + "epoch": 1.9057475055660922, + "grad_norm": 2.169269323348999, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7856173515319824, + "num_tokens": 584869623.0, + "step": 23111 + }, + { + "epoch": 1.9058299661911438, + "grad_norm": 2.0042126178741455, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.7968388199806213, + "num_tokens": 584896277.0, + "step": 23112 + }, + { + "epoch": 1.905912426816195, + "grad_norm": 1.8945026397705078, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7309973239898682, + "num_tokens": 584927545.0, + "step": 23113 + }, + { + "epoch": 1.905994887441247, + "grad_norm": 1.7269431352615356, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.7943280935287476, + "num_tokens": 584961157.0, + "step": 23114 + }, + { + "epoch": 1.9060773480662982, + "grad_norm": 2.278322458267212, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7721948623657227, + "num_tokens": 584985352.0, + "step": 23115 + }, + { + "epoch": 1.90615980869135, + "grad_norm": 1.9955265522003174, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7612842917442322, + "num_tokens": 585015221.0, + "step": 23116 + }, + { + "epoch": 1.9062422693164014, + "grad_norm": 1.976535677909851, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7523126602172852, + "num_tokens": 585044154.0, + "step": 23117 + }, + { + "epoch": 1.906324729941453, + "grad_norm": 1.9625437259674072, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.77376788854599, + "num_tokens": 585073321.0, + "step": 23118 + }, + { + "epoch": 1.9064071905665045, + "grad_norm": 1.8236289024353027, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7746535539627075, + "num_tokens": 585107751.0, + "step": 23119 + }, + { + "epoch": 1.906489651191556, + "grad_norm": 1.949500560760498, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.7856234312057495, + "num_tokens": 585136545.0, + "step": 23120 + }, + { + "epoch": 1.9065721118166077, + "grad_norm": 1.967010259628296, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.77545166015625, + "num_tokens": 585167580.0, + "step": 23121 + }, + { + "epoch": 1.906654572441659, + "grad_norm": 2.132018566131592, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7361111640930176, + "num_tokens": 585195086.0, + "step": 23122 + }, + { + "epoch": 1.9067370330667106, + "grad_norm": 1.966855525970459, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.761436939239502, + "num_tokens": 585223979.0, + "step": 23123 + }, + { + "epoch": 1.9068194936917622, + "grad_norm": 2.0179200172424316, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7677607536315918, + "num_tokens": 585253968.0, + "step": 23124 + }, + { + "epoch": 1.9069019543168138, + "grad_norm": 1.8781198263168335, + "learning_rate": 1e-06, + "loss": 0.5721, + "mean_token_accuracy": 0.8121563196182251, + "num_tokens": 585283401.0, + "step": 23125 + }, + { + "epoch": 1.9069844149418653, + "grad_norm": 2.1577301025390625, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7328651547431946, + "num_tokens": 585311293.0, + "step": 23126 + }, + { + "epoch": 1.9070668755669167, + "grad_norm": 2.0912866592407227, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7539840936660767, + "num_tokens": 585338701.0, + "step": 23127 + }, + { + "epoch": 1.9071493361919685, + "grad_norm": 1.9179116487503052, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7674516439437866, + "num_tokens": 585370250.0, + "step": 23128 + }, + { + "epoch": 1.9072317968170198, + "grad_norm": 1.850447654724121, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7525688409805298, + "num_tokens": 585404257.0, + "step": 23129 + }, + { + "epoch": 1.9073142574420714, + "grad_norm": 1.9524661302566528, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7506420612335205, + "num_tokens": 585433857.0, + "step": 23130 + }, + { + "epoch": 1.907396718067123, + "grad_norm": 1.9764118194580078, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7515803575515747, + "num_tokens": 585463578.0, + "step": 23131 + }, + { + "epoch": 1.9074791786921745, + "grad_norm": 2.1830813884735107, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7528047561645508, + "num_tokens": 585487823.0, + "step": 23132 + }, + { + "epoch": 1.907561639317226, + "grad_norm": 1.9592149257659912, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7715253829956055, + "num_tokens": 585519061.0, + "step": 23133 + }, + { + "epoch": 1.9076440999422775, + "grad_norm": 1.786555528640747, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.781822681427002, + "num_tokens": 585553218.0, + "step": 23134 + }, + { + "epoch": 1.9077265605673293, + "grad_norm": 2.47393536567688, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.7557882070541382, + "num_tokens": 585574881.0, + "step": 23135 + }, + { + "epoch": 1.9078090211923806, + "grad_norm": 2.049516439437866, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7614181041717529, + "num_tokens": 585603127.0, + "step": 23136 + }, + { + "epoch": 1.9078914818174322, + "grad_norm": 1.8748373985290527, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7480283975601196, + "num_tokens": 585639278.0, + "step": 23137 + }, + { + "epoch": 1.9079739424424838, + "grad_norm": 2.0230631828308105, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7608789801597595, + "num_tokens": 585670354.0, + "step": 23138 + }, + { + "epoch": 1.908056403067535, + "grad_norm": 2.2023890018463135, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7569713592529297, + "num_tokens": 585695049.0, + "step": 23139 + }, + { + "epoch": 1.908138863692587, + "grad_norm": 1.9569567441940308, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.73508220911026, + "num_tokens": 585727289.0, + "step": 23140 + }, + { + "epoch": 1.9082213243176382, + "grad_norm": 2.0743587017059326, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.7900032997131348, + "num_tokens": 585754092.0, + "step": 23141 + }, + { + "epoch": 1.9083037849426898, + "grad_norm": 2.0199382305145264, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7759593725204468, + "num_tokens": 585783138.0, + "step": 23142 + }, + { + "epoch": 1.9083862455677414, + "grad_norm": 2.049149513244629, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7281643748283386, + "num_tokens": 585813240.0, + "step": 23143 + }, + { + "epoch": 1.908468706192793, + "grad_norm": 1.9272966384887695, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7722514867782593, + "num_tokens": 585844631.0, + "step": 23144 + }, + { + "epoch": 1.9085511668178445, + "grad_norm": 1.760425329208374, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7741820812225342, + "num_tokens": 585879462.0, + "step": 23145 + }, + { + "epoch": 1.9086336274428959, + "grad_norm": 2.0245280265808105, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7335045337677002, + "num_tokens": 585909110.0, + "step": 23146 + }, + { + "epoch": 1.9087160880679477, + "grad_norm": 2.1239469051361084, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7369375228881836, + "num_tokens": 585935823.0, + "step": 23147 + }, + { + "epoch": 1.908798548692999, + "grad_norm": 1.83500075340271, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.7946748733520508, + "num_tokens": 585966874.0, + "step": 23148 + }, + { + "epoch": 1.9088810093180506, + "grad_norm": 1.7476285696029663, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7724514007568359, + "num_tokens": 586008056.0, + "step": 23149 + }, + { + "epoch": 1.9089634699431022, + "grad_norm": 1.7719886302947998, + "learning_rate": 1e-06, + "loss": 0.5575, + "mean_token_accuracy": 0.8058496713638306, + "num_tokens": 586039448.0, + "step": 23150 + }, + { + "epoch": 1.9090459305681537, + "grad_norm": 1.706255316734314, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.769871711730957, + "num_tokens": 586078441.0, + "step": 23151 + }, + { + "epoch": 1.9091283911932053, + "grad_norm": 2.2387475967407227, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7796636819839478, + "num_tokens": 586100964.0, + "step": 23152 + }, + { + "epoch": 1.9092108518182567, + "grad_norm": 2.1241655349731445, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7174672484397888, + "num_tokens": 586130509.0, + "step": 23153 + }, + { + "epoch": 1.9092933124433085, + "grad_norm": 2.0437090396881104, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7643764019012451, + "num_tokens": 586157513.0, + "step": 23154 + }, + { + "epoch": 1.9093757730683598, + "grad_norm": 2.120410442352295, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.7896716594696045, + "num_tokens": 586187437.0, + "step": 23155 + }, + { + "epoch": 1.9094582336934114, + "grad_norm": 2.0152387619018555, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7438084483146667, + "num_tokens": 586215980.0, + "step": 23156 + }, + { + "epoch": 1.909540694318463, + "grad_norm": 1.8418220281600952, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7869992852210999, + "num_tokens": 586249238.0, + "step": 23157 + }, + { + "epoch": 1.9096231549435143, + "grad_norm": 2.3446056842803955, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7468093037605286, + "num_tokens": 586272445.0, + "step": 23158 + }, + { + "epoch": 1.909705615568566, + "grad_norm": 1.9817636013031006, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7555450797080994, + "num_tokens": 586303216.0, + "step": 23159 + }, + { + "epoch": 1.9097880761936175, + "grad_norm": 2.5433530807495117, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7606961727142334, + "num_tokens": 586323331.0, + "step": 23160 + }, + { + "epoch": 1.9098705368186693, + "grad_norm": 2.1185824871063232, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7399955987930298, + "num_tokens": 586350659.0, + "step": 23161 + }, + { + "epoch": 1.9099529974437206, + "grad_norm": 2.038944721221924, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7326362729072571, + "num_tokens": 586382185.0, + "step": 23162 + }, + { + "epoch": 1.9100354580687722, + "grad_norm": 2.0347208976745605, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.740820050239563, + "num_tokens": 586412148.0, + "step": 23163 + }, + { + "epoch": 1.9101179186938237, + "grad_norm": 2.2145144939422607, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7513120770454407, + "num_tokens": 586437433.0, + "step": 23164 + }, + { + "epoch": 1.910200379318875, + "grad_norm": 1.9290921688079834, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.773193359375, + "num_tokens": 586471717.0, + "step": 23165 + }, + { + "epoch": 1.910282839943927, + "grad_norm": 2.2032723426818848, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7498053312301636, + "num_tokens": 586496124.0, + "step": 23166 + }, + { + "epoch": 1.9103653005689782, + "grad_norm": 2.084507942199707, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7480510473251343, + "num_tokens": 586524147.0, + "step": 23167 + }, + { + "epoch": 1.9104477611940298, + "grad_norm": 1.9785451889038086, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7451640367507935, + "num_tokens": 586554351.0, + "step": 23168 + }, + { + "epoch": 1.9105302218190814, + "grad_norm": 1.8198293447494507, + "learning_rate": 1e-06, + "loss": 0.6094, + "mean_token_accuracy": 0.7970056533813477, + "num_tokens": 586588519.0, + "step": 23169 + }, + { + "epoch": 1.910612682444133, + "grad_norm": 2.0201053619384766, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7260485887527466, + "num_tokens": 586620784.0, + "step": 23170 + }, + { + "epoch": 1.9106951430691845, + "grad_norm": 1.9378236532211304, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7785366177558899, + "num_tokens": 586651394.0, + "step": 23171 + }, + { + "epoch": 1.9107776036942359, + "grad_norm": 1.9168570041656494, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.7922988533973694, + "num_tokens": 586679674.0, + "step": 23172 + }, + { + "epoch": 1.9108600643192877, + "grad_norm": 1.868934988975525, + "learning_rate": 1e-06, + "loss": 0.5656, + "mean_token_accuracy": 0.8016374111175537, + "num_tokens": 586710930.0, + "step": 23173 + }, + { + "epoch": 1.910942524944339, + "grad_norm": 2.137420415878296, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.757926881313324, + "num_tokens": 586738010.0, + "step": 23174 + }, + { + "epoch": 1.9110249855693906, + "grad_norm": 1.8597455024719238, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7678914666175842, + "num_tokens": 586772129.0, + "step": 23175 + }, + { + "epoch": 1.9111074461944422, + "grad_norm": 1.7335208654403687, + "learning_rate": 1e-06, + "loss": 0.5948, + "mean_token_accuracy": 0.8074502348899841, + "num_tokens": 586804902.0, + "step": 23176 + }, + { + "epoch": 1.9111899068194937, + "grad_norm": 1.6867979764938354, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7714200019836426, + "num_tokens": 586844324.0, + "step": 23177 + }, + { + "epoch": 1.9112723674445453, + "grad_norm": 1.7896605730056763, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7741061449050903, + "num_tokens": 586878882.0, + "step": 23178 + }, + { + "epoch": 1.9113548280695967, + "grad_norm": 2.0427680015563965, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.789579451084137, + "num_tokens": 586908202.0, + "step": 23179 + }, + { + "epoch": 1.9114372886946485, + "grad_norm": 1.9866431951522827, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7496841549873352, + "num_tokens": 586936663.0, + "step": 23180 + }, + { + "epoch": 1.9115197493196998, + "grad_norm": 2.1103127002716064, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7488542199134827, + "num_tokens": 586962020.0, + "step": 23181 + }, + { + "epoch": 1.9116022099447514, + "grad_norm": 2.2680881023406982, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7339361906051636, + "num_tokens": 586988481.0, + "step": 23182 + }, + { + "epoch": 1.911684670569803, + "grad_norm": 2.117774486541748, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7780163884162903, + "num_tokens": 587015955.0, + "step": 23183 + }, + { + "epoch": 1.9117671311948543, + "grad_norm": 2.0100088119506836, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7690879106521606, + "num_tokens": 587043694.0, + "step": 23184 + }, + { + "epoch": 1.911849591819906, + "grad_norm": 1.8516216278076172, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7437841892242432, + "num_tokens": 587077857.0, + "step": 23185 + }, + { + "epoch": 1.9119320524449575, + "grad_norm": 1.9960576295852661, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.7643733024597168, + "num_tokens": 587106064.0, + "step": 23186 + }, + { + "epoch": 1.912014513070009, + "grad_norm": 2.0893609523773193, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7486995458602905, + "num_tokens": 587132135.0, + "step": 23187 + }, + { + "epoch": 1.9120969736950606, + "grad_norm": 2.024949312210083, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.726648211479187, + "num_tokens": 587161323.0, + "step": 23188 + }, + { + "epoch": 1.9121794343201122, + "grad_norm": 1.8287707567214966, + "learning_rate": 1e-06, + "loss": 0.7563, + "mean_token_accuracy": 0.7606710195541382, + "num_tokens": 587196637.0, + "step": 23189 + }, + { + "epoch": 1.9122618949451637, + "grad_norm": 1.9094563722610474, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7232640385627747, + "num_tokens": 587230522.0, + "step": 23190 + }, + { + "epoch": 1.912344355570215, + "grad_norm": 2.1779448986053467, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7387875318527222, + "num_tokens": 587256428.0, + "step": 23191 + }, + { + "epoch": 1.9124268161952669, + "grad_norm": 2.4552018642425537, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7818472981452942, + "num_tokens": 587278104.0, + "step": 23192 + }, + { + "epoch": 1.9125092768203182, + "grad_norm": 1.992849588394165, + "learning_rate": 1e-06, + "loss": 0.7929, + "mean_token_accuracy": 0.7427424192428589, + "num_tokens": 587307721.0, + "step": 23193 + }, + { + "epoch": 1.9125917374453698, + "grad_norm": 1.8291839361190796, + "learning_rate": 1e-06, + "loss": 0.5845, + "mean_token_accuracy": 0.8038021326065063, + "num_tokens": 587339746.0, + "step": 23194 + }, + { + "epoch": 1.9126741980704214, + "grad_norm": 2.3484201431274414, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7412339448928833, + "num_tokens": 587364476.0, + "step": 23195 + }, + { + "epoch": 1.912756658695473, + "grad_norm": 2.122269630432129, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7440582513809204, + "num_tokens": 587394303.0, + "step": 23196 + }, + { + "epoch": 1.9128391193205245, + "grad_norm": 2.1292221546173096, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7691687941551208, + "num_tokens": 587422357.0, + "step": 23197 + }, + { + "epoch": 1.9129215799455759, + "grad_norm": 2.0325706005096436, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7287973761558533, + "num_tokens": 587456182.0, + "step": 23198 + }, + { + "epoch": 1.9130040405706277, + "grad_norm": 1.7475162744522095, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7784920334815979, + "num_tokens": 587489878.0, + "step": 23199 + }, + { + "epoch": 1.913086501195679, + "grad_norm": 1.915258765220642, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.7946410179138184, + "num_tokens": 587517648.0, + "step": 23200 + }, + { + "epoch": 1.9131689618207306, + "grad_norm": 2.146338701248169, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7755866050720215, + "num_tokens": 587545713.0, + "step": 23201 + }, + { + "epoch": 1.9132514224457822, + "grad_norm": 1.898224115371704, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.7929271459579468, + "num_tokens": 587578837.0, + "step": 23202 + }, + { + "epoch": 1.9133338830708335, + "grad_norm": 2.0549159049987793, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7212383151054382, + "num_tokens": 587608285.0, + "step": 23203 + }, + { + "epoch": 1.9134163436958853, + "grad_norm": 1.923822283744812, + "learning_rate": 1e-06, + "loss": 0.5688, + "mean_token_accuracy": 0.800129771232605, + "num_tokens": 587637569.0, + "step": 23204 + }, + { + "epoch": 1.9134988043209367, + "grad_norm": 2.006882667541504, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.7824931144714355, + "num_tokens": 587667995.0, + "step": 23205 + }, + { + "epoch": 1.9135812649459882, + "grad_norm": 1.8283708095550537, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7694398760795593, + "num_tokens": 587700465.0, + "step": 23206 + }, + { + "epoch": 1.9136637255710398, + "grad_norm": 1.839163064956665, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7564772367477417, + "num_tokens": 587732370.0, + "step": 23207 + }, + { + "epoch": 1.9137461861960914, + "grad_norm": 2.315201759338379, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7391488552093506, + "num_tokens": 587755696.0, + "step": 23208 + }, + { + "epoch": 1.913828646821143, + "grad_norm": 1.9316965341567993, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.758013904094696, + "num_tokens": 587786330.0, + "step": 23209 + }, + { + "epoch": 1.9139111074461943, + "grad_norm": 1.9225393533706665, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7670153379440308, + "num_tokens": 587818364.0, + "step": 23210 + }, + { + "epoch": 1.913993568071246, + "grad_norm": 1.8534491062164307, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7581335306167603, + "num_tokens": 587851970.0, + "step": 23211 + }, + { + "epoch": 1.9140760286962974, + "grad_norm": 2.031331777572632, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7240747213363647, + "num_tokens": 587881957.0, + "step": 23212 + }, + { + "epoch": 1.914158489321349, + "grad_norm": 1.7930911779403687, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.7731569409370422, + "num_tokens": 587914550.0, + "step": 23213 + }, + { + "epoch": 1.9142409499464006, + "grad_norm": 2.1131432056427, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7493065595626831, + "num_tokens": 587940959.0, + "step": 23214 + }, + { + "epoch": 1.9143234105714522, + "grad_norm": 1.8146823644638062, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7423789501190186, + "num_tokens": 587976340.0, + "step": 23215 + }, + { + "epoch": 1.9144058711965037, + "grad_norm": 1.7894530296325684, + "learning_rate": 1e-06, + "loss": 0.5604, + "mean_token_accuracy": 0.8055920004844666, + "num_tokens": 588010949.0, + "step": 23216 + }, + { + "epoch": 1.914488331821555, + "grad_norm": 2.232166290283203, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7616522908210754, + "num_tokens": 588034511.0, + "step": 23217 + }, + { + "epoch": 1.9145707924466069, + "grad_norm": 1.942967414855957, + "learning_rate": 1e-06, + "loss": 0.5703, + "mean_token_accuracy": 0.8035192489624023, + "num_tokens": 588064237.0, + "step": 23218 + }, + { + "epoch": 1.9146532530716582, + "grad_norm": 1.9857609272003174, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7854956388473511, + "num_tokens": 588093603.0, + "step": 23219 + }, + { + "epoch": 1.9147357136967098, + "grad_norm": 2.0989317893981934, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7576110363006592, + "num_tokens": 588118204.0, + "step": 23220 + }, + { + "epoch": 1.9148181743217614, + "grad_norm": 1.776065707206726, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.785659909248352, + "num_tokens": 588155806.0, + "step": 23221 + }, + { + "epoch": 1.914900634946813, + "grad_norm": 1.8363348245620728, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7035375833511353, + "num_tokens": 588193299.0, + "step": 23222 + }, + { + "epoch": 1.9149830955718645, + "grad_norm": 2.293302297592163, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.753793478012085, + "num_tokens": 588219732.0, + "step": 23223 + }, + { + "epoch": 1.9150655561969159, + "grad_norm": 1.915272831916809, + "learning_rate": 1e-06, + "loss": 0.5944, + "mean_token_accuracy": 0.793803870677948, + "num_tokens": 588247928.0, + "step": 23224 + }, + { + "epoch": 1.9151480168219677, + "grad_norm": 1.9602519273757935, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.7881296873092651, + "num_tokens": 588275729.0, + "step": 23225 + }, + { + "epoch": 1.915230477447019, + "grad_norm": 2.348814010620117, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7496023774147034, + "num_tokens": 588300888.0, + "step": 23226 + }, + { + "epoch": 1.9153129380720706, + "grad_norm": 1.9983811378479004, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7641598582267761, + "num_tokens": 588330173.0, + "step": 23227 + }, + { + "epoch": 1.9153953986971222, + "grad_norm": 2.07694673538208, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7723511457443237, + "num_tokens": 588358650.0, + "step": 23228 + }, + { + "epoch": 1.9154778593221735, + "grad_norm": 1.8848904371261597, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7483752965927124, + "num_tokens": 588392394.0, + "step": 23229 + }, + { + "epoch": 1.9155603199472253, + "grad_norm": 2.1701018810272217, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7374019622802734, + "num_tokens": 588420357.0, + "step": 23230 + }, + { + "epoch": 1.9156427805722767, + "grad_norm": 1.8418571949005127, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7537651062011719, + "num_tokens": 588455674.0, + "step": 23231 + }, + { + "epoch": 1.9157252411973282, + "grad_norm": 1.9394500255584717, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.775220513343811, + "num_tokens": 588488091.0, + "step": 23232 + }, + { + "epoch": 1.9158077018223798, + "grad_norm": 2.051088809967041, + "learning_rate": 1e-06, + "loss": 0.7852, + "mean_token_accuracy": 0.7475635409355164, + "num_tokens": 588514719.0, + "step": 23233 + }, + { + "epoch": 1.9158901624474314, + "grad_norm": 2.1836845874786377, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7360756993293762, + "num_tokens": 588540174.0, + "step": 23234 + }, + { + "epoch": 1.915972623072483, + "grad_norm": 1.9044052362442017, + "learning_rate": 1e-06, + "loss": 0.5712, + "mean_token_accuracy": 0.8105698823928833, + "num_tokens": 588571632.0, + "step": 23235 + }, + { + "epoch": 1.9160550836975343, + "grad_norm": 1.802600383758545, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7458820939064026, + "num_tokens": 588608059.0, + "step": 23236 + }, + { + "epoch": 1.916137544322586, + "grad_norm": 1.8450254201889038, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7488446235656738, + "num_tokens": 588640479.0, + "step": 23237 + }, + { + "epoch": 1.9162200049476374, + "grad_norm": 2.1605186462402344, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7497339844703674, + "num_tokens": 588665197.0, + "step": 23238 + }, + { + "epoch": 1.916302465572689, + "grad_norm": 1.958143949508667, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7515099048614502, + "num_tokens": 588697937.0, + "step": 23239 + }, + { + "epoch": 1.9163849261977406, + "grad_norm": 2.08699631690979, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.7879414558410645, + "num_tokens": 588724263.0, + "step": 23240 + }, + { + "epoch": 1.9164673868227922, + "grad_norm": 1.914243221282959, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.788679301738739, + "num_tokens": 588756902.0, + "step": 23241 + }, + { + "epoch": 1.9165498474478437, + "grad_norm": 1.9870723485946655, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.753575325012207, + "num_tokens": 588787806.0, + "step": 23242 + }, + { + "epoch": 1.916632308072895, + "grad_norm": 2.1470086574554443, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7730429172515869, + "num_tokens": 588815305.0, + "step": 23243 + }, + { + "epoch": 1.9167147686979469, + "grad_norm": 1.8349993228912354, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.7920603156089783, + "num_tokens": 588844953.0, + "step": 23244 + }, + { + "epoch": 1.9167972293229982, + "grad_norm": 1.9352750778198242, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7843323945999146, + "num_tokens": 588873708.0, + "step": 23245 + }, + { + "epoch": 1.9168796899480498, + "grad_norm": 1.9395126104354858, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7617080807685852, + "num_tokens": 588904386.0, + "step": 23246 + }, + { + "epoch": 1.9169621505731014, + "grad_norm": 2.1025748252868652, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7564672231674194, + "num_tokens": 588929266.0, + "step": 23247 + }, + { + "epoch": 1.9170446111981527, + "grad_norm": 2.2415614128112793, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7414839267730713, + "num_tokens": 588955247.0, + "step": 23248 + }, + { + "epoch": 1.9171270718232045, + "grad_norm": 2.270174741744995, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7461200952529907, + "num_tokens": 588980816.0, + "step": 23249 + }, + { + "epoch": 1.9172095324482559, + "grad_norm": 2.187697172164917, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7483183741569519, + "num_tokens": 589005983.0, + "step": 23250 + }, + { + "epoch": 1.9172919930733074, + "grad_norm": 1.8048591613769531, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7529227137565613, + "num_tokens": 589044718.0, + "step": 23251 + }, + { + "epoch": 1.917374453698359, + "grad_norm": 2.1071114540100098, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7633396983146667, + "num_tokens": 589070981.0, + "step": 23252 + }, + { + "epoch": 1.9174569143234106, + "grad_norm": 2.366039991378784, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7384459972381592, + "num_tokens": 589094379.0, + "step": 23253 + }, + { + "epoch": 1.9175393749484622, + "grad_norm": 2.083873987197876, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.766861617565155, + "num_tokens": 589122919.0, + "step": 23254 + }, + { + "epoch": 1.9176218355735135, + "grad_norm": 1.9863911867141724, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7490332126617432, + "num_tokens": 589151061.0, + "step": 23255 + }, + { + "epoch": 1.9177042961985653, + "grad_norm": 1.9859546422958374, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7450305223464966, + "num_tokens": 589181705.0, + "step": 23256 + }, + { + "epoch": 1.9177867568236167, + "grad_norm": 2.0692737102508545, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7746074199676514, + "num_tokens": 589209536.0, + "step": 23257 + }, + { + "epoch": 1.9178692174486682, + "grad_norm": 2.179166078567505, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7454218864440918, + "num_tokens": 589236220.0, + "step": 23258 + }, + { + "epoch": 1.9179516780737198, + "grad_norm": 1.7917702198028564, + "learning_rate": 1e-06, + "loss": 0.7767, + "mean_token_accuracy": 0.7563682794570923, + "num_tokens": 589270921.0, + "step": 23259 + }, + { + "epoch": 1.9180341386987714, + "grad_norm": 1.9236966371536255, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7738720178604126, + "num_tokens": 589301109.0, + "step": 23260 + }, + { + "epoch": 1.918116599323823, + "grad_norm": 2.2636051177978516, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7732720375061035, + "num_tokens": 589324048.0, + "step": 23261 + }, + { + "epoch": 1.9181990599488743, + "grad_norm": 2.2285308837890625, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.7715016603469849, + "num_tokens": 589349045.0, + "step": 23262 + }, + { + "epoch": 1.918281520573926, + "grad_norm": 2.3076529502868652, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7698222994804382, + "num_tokens": 589371372.0, + "step": 23263 + }, + { + "epoch": 1.9183639811989774, + "grad_norm": 1.8255996704101562, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7695878744125366, + "num_tokens": 589406054.0, + "step": 23264 + }, + { + "epoch": 1.918446441824029, + "grad_norm": 1.7999807596206665, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7202509641647339, + "num_tokens": 589445197.0, + "step": 23265 + }, + { + "epoch": 1.9185289024490806, + "grad_norm": 1.8959578275680542, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7480760812759399, + "num_tokens": 589478309.0, + "step": 23266 + }, + { + "epoch": 1.918611363074132, + "grad_norm": 1.8286701440811157, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7841106653213501, + "num_tokens": 589508690.0, + "step": 23267 + }, + { + "epoch": 1.9186938236991837, + "grad_norm": 2.171459913253784, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.7894961833953857, + "num_tokens": 589533137.0, + "step": 23268 + }, + { + "epoch": 1.918776284324235, + "grad_norm": 2.0366690158843994, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7865380644798279, + "num_tokens": 589561725.0, + "step": 23269 + }, + { + "epoch": 1.9188587449492869, + "grad_norm": 2.0874249935150146, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7777774930000305, + "num_tokens": 589591535.0, + "step": 23270 + }, + { + "epoch": 1.9189412055743382, + "grad_norm": 1.82961106300354, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7702158093452454, + "num_tokens": 589622420.0, + "step": 23271 + }, + { + "epoch": 1.9190236661993898, + "grad_norm": 2.1075737476348877, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7347109317779541, + "num_tokens": 589650052.0, + "step": 23272 + }, + { + "epoch": 1.9191061268244414, + "grad_norm": 1.975427269935608, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7681103944778442, + "num_tokens": 589678139.0, + "step": 23273 + }, + { + "epoch": 1.9191885874494927, + "grad_norm": 1.982804298400879, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7456997036933899, + "num_tokens": 589707902.0, + "step": 23274 + }, + { + "epoch": 1.9192710480745445, + "grad_norm": 1.8165123462677002, + "learning_rate": 1e-06, + "loss": 0.7883, + "mean_token_accuracy": 0.7432796955108643, + "num_tokens": 589743000.0, + "step": 23275 + }, + { + "epoch": 1.9193535086995959, + "grad_norm": 2.012070417404175, + "learning_rate": 1e-06, + "loss": 0.6315, + "mean_token_accuracy": 0.7915095090866089, + "num_tokens": 589773059.0, + "step": 23276 + }, + { + "epoch": 1.9194359693246474, + "grad_norm": 2.2698004245758057, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.7514191269874573, + "num_tokens": 589799045.0, + "step": 23277 + }, + { + "epoch": 1.919518429949699, + "grad_norm": 1.9041454792022705, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7667436599731445, + "num_tokens": 589831237.0, + "step": 23278 + }, + { + "epoch": 1.9196008905747506, + "grad_norm": 2.1058835983276367, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7791058421134949, + "num_tokens": 589859248.0, + "step": 23279 + }, + { + "epoch": 1.9196833511998022, + "grad_norm": 2.1836493015289307, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7436121702194214, + "num_tokens": 589888264.0, + "step": 23280 + }, + { + "epoch": 1.9197658118248535, + "grad_norm": 2.063143730163574, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7436022758483887, + "num_tokens": 589916136.0, + "step": 23281 + }, + { + "epoch": 1.9198482724499053, + "grad_norm": 1.6617581844329834, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.7981505990028381, + "num_tokens": 589953781.0, + "step": 23282 + }, + { + "epoch": 1.9199307330749567, + "grad_norm": 1.944069266319275, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7715791463851929, + "num_tokens": 589982693.0, + "step": 23283 + }, + { + "epoch": 1.9200131937000082, + "grad_norm": 1.7088520526885986, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.7957649827003479, + "num_tokens": 590019296.0, + "step": 23284 + }, + { + "epoch": 1.9200956543250598, + "grad_norm": 2.20516037940979, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7593786716461182, + "num_tokens": 590042871.0, + "step": 23285 + }, + { + "epoch": 1.9201781149501114, + "grad_norm": 2.1541314125061035, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7315390110015869, + "num_tokens": 590068014.0, + "step": 23286 + }, + { + "epoch": 1.920260575575163, + "grad_norm": 2.081717014312744, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7436514496803284, + "num_tokens": 590096435.0, + "step": 23287 + }, + { + "epoch": 1.9203430362002143, + "grad_norm": 1.7490158081054688, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.7987064123153687, + "num_tokens": 590130921.0, + "step": 23288 + }, + { + "epoch": 1.920425496825266, + "grad_norm": 2.0069479942321777, + "learning_rate": 1e-06, + "loss": 0.5189, + "mean_token_accuracy": 0.8121997714042664, + "num_tokens": 590156409.0, + "step": 23289 + }, + { + "epoch": 1.9205079574503174, + "grad_norm": 2.3000025749206543, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7244663238525391, + "num_tokens": 590180187.0, + "step": 23290 + }, + { + "epoch": 1.920590418075369, + "grad_norm": 2.0567760467529297, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7623817920684814, + "num_tokens": 590206321.0, + "step": 23291 + }, + { + "epoch": 1.9206728787004206, + "grad_norm": 1.8334414958953857, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7544934153556824, + "num_tokens": 590237934.0, + "step": 23292 + }, + { + "epoch": 1.920755339325472, + "grad_norm": 1.859968900680542, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.7942240238189697, + "num_tokens": 590270058.0, + "step": 23293 + }, + { + "epoch": 1.9208377999505237, + "grad_norm": 1.8456851243972778, + "learning_rate": 1e-06, + "loss": 0.5437, + "mean_token_accuracy": 0.8103963136672974, + "num_tokens": 590303156.0, + "step": 23294 + }, + { + "epoch": 1.920920260575575, + "grad_norm": 1.9253578186035156, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7701849937438965, + "num_tokens": 590336358.0, + "step": 23295 + }, + { + "epoch": 1.9210027212006267, + "grad_norm": 2.0223588943481445, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7666240930557251, + "num_tokens": 590366263.0, + "step": 23296 + }, + { + "epoch": 1.9210851818256782, + "grad_norm": 2.154202461242676, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7383391857147217, + "num_tokens": 590392390.0, + "step": 23297 + }, + { + "epoch": 1.9211676424507298, + "grad_norm": 1.9928499460220337, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7768656015396118, + "num_tokens": 590421946.0, + "step": 23298 + }, + { + "epoch": 1.9212501030757814, + "grad_norm": 2.00724458694458, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7708612084388733, + "num_tokens": 590452506.0, + "step": 23299 + }, + { + "epoch": 1.9213325637008327, + "grad_norm": 1.8911395072937012, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7206722497940063, + "num_tokens": 590488668.0, + "step": 23300 + }, + { + "epoch": 1.9214150243258845, + "grad_norm": 1.9031972885131836, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.744617223739624, + "num_tokens": 590521451.0, + "step": 23301 + }, + { + "epoch": 1.9214974849509359, + "grad_norm": 1.8424831628799438, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7624744176864624, + "num_tokens": 590556154.0, + "step": 23302 + }, + { + "epoch": 1.9215799455759874, + "grad_norm": 2.2757270336151123, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.785439133644104, + "num_tokens": 590580375.0, + "step": 23303 + }, + { + "epoch": 1.921662406201039, + "grad_norm": 1.7129372358322144, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.7995019555091858, + "num_tokens": 590617312.0, + "step": 23304 + }, + { + "epoch": 1.9217448668260906, + "grad_norm": 1.9986872673034668, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7650677561759949, + "num_tokens": 590650368.0, + "step": 23305 + }, + { + "epoch": 1.9218273274511422, + "grad_norm": 2.0041301250457764, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7454557418823242, + "num_tokens": 590682241.0, + "step": 23306 + }, + { + "epoch": 1.9219097880761935, + "grad_norm": 1.8606261014938354, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7323733568191528, + "num_tokens": 590715355.0, + "step": 23307 + }, + { + "epoch": 1.9219922487012453, + "grad_norm": 2.0215821266174316, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7312212586402893, + "num_tokens": 590743946.0, + "step": 23308 + }, + { + "epoch": 1.9220747093262966, + "grad_norm": 2.14465594291687, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7538066506385803, + "num_tokens": 590769660.0, + "step": 23309 + }, + { + "epoch": 1.9221571699513482, + "grad_norm": 2.0908100605010986, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7744101285934448, + "num_tokens": 590799821.0, + "step": 23310 + }, + { + "epoch": 1.9222396305763998, + "grad_norm": 1.9814823865890503, + "learning_rate": 1e-06, + "loss": 0.5863, + "mean_token_accuracy": 0.7888288497924805, + "num_tokens": 590829024.0, + "step": 23311 + }, + { + "epoch": 1.9223220912014511, + "grad_norm": 1.88784658908844, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.7938379049301147, + "num_tokens": 590857394.0, + "step": 23312 + }, + { + "epoch": 1.922404551826503, + "grad_norm": 2.092139959335327, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7528470754623413, + "num_tokens": 590885178.0, + "step": 23313 + }, + { + "epoch": 1.9224870124515543, + "grad_norm": 2.0473954677581787, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7368263006210327, + "num_tokens": 590914604.0, + "step": 23314 + }, + { + "epoch": 1.922569473076606, + "grad_norm": 2.082618474960327, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7522420883178711, + "num_tokens": 590942951.0, + "step": 23315 + }, + { + "epoch": 1.9226519337016574, + "grad_norm": 1.9260996580123901, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7305399179458618, + "num_tokens": 590975764.0, + "step": 23316 + }, + { + "epoch": 1.922734394326709, + "grad_norm": 2.1846373081207275, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7534615993499756, + "num_tokens": 591003203.0, + "step": 23317 + }, + { + "epoch": 1.9228168549517606, + "grad_norm": 1.7986029386520386, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7729641199111938, + "num_tokens": 591038194.0, + "step": 23318 + }, + { + "epoch": 1.922899315576812, + "grad_norm": 1.9672720432281494, + "learning_rate": 1e-06, + "loss": 0.6068, + "mean_token_accuracy": 0.8039870262145996, + "num_tokens": 591066762.0, + "step": 23319 + }, + { + "epoch": 1.9229817762018637, + "grad_norm": 1.751094102859497, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7759937047958374, + "num_tokens": 591103692.0, + "step": 23320 + }, + { + "epoch": 1.923064236826915, + "grad_norm": 1.898022174835205, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.7558135986328125, + "num_tokens": 591135170.0, + "step": 23321 + }, + { + "epoch": 1.9231466974519666, + "grad_norm": 1.9013475179672241, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7659555077552795, + "num_tokens": 591167302.0, + "step": 23322 + }, + { + "epoch": 1.9232291580770182, + "grad_norm": 2.074495792388916, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.734827995300293, + "num_tokens": 591197750.0, + "step": 23323 + }, + { + "epoch": 1.9233116187020698, + "grad_norm": 2.1534531116485596, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7447001934051514, + "num_tokens": 591223468.0, + "step": 23324 + }, + { + "epoch": 1.9233940793271214, + "grad_norm": 2.034499406814575, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7526839971542358, + "num_tokens": 591252515.0, + "step": 23325 + }, + { + "epoch": 1.9234765399521727, + "grad_norm": 2.2354328632354736, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7171012163162231, + "num_tokens": 591276833.0, + "step": 23326 + }, + { + "epoch": 1.9235590005772245, + "grad_norm": 2.0339975357055664, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.7829816341400146, + "num_tokens": 591302078.0, + "step": 23327 + }, + { + "epoch": 1.9236414612022759, + "grad_norm": 2.0156683921813965, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7703487873077393, + "num_tokens": 591330454.0, + "step": 23328 + }, + { + "epoch": 1.9237239218273274, + "grad_norm": 2.0820424556732178, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7821037173271179, + "num_tokens": 591357823.0, + "step": 23329 + }, + { + "epoch": 1.923806382452379, + "grad_norm": 2.0628416538238525, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7503080368041992, + "num_tokens": 591388652.0, + "step": 23330 + }, + { + "epoch": 1.9238888430774306, + "grad_norm": 1.6869125366210938, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7730530500411987, + "num_tokens": 591426418.0, + "step": 23331 + }, + { + "epoch": 1.9239713037024821, + "grad_norm": 2.2965221405029297, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7785264849662781, + "num_tokens": 591447893.0, + "step": 23332 + }, + { + "epoch": 1.9240537643275335, + "grad_norm": 2.2433080673217773, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7563000917434692, + "num_tokens": 591473904.0, + "step": 23333 + }, + { + "epoch": 1.9241362249525853, + "grad_norm": 1.7116001844406128, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7768316268920898, + "num_tokens": 591508264.0, + "step": 23334 + }, + { + "epoch": 1.9242186855776366, + "grad_norm": 2.4636430740356445, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7417330741882324, + "num_tokens": 591533856.0, + "step": 23335 + }, + { + "epoch": 1.9243011462026882, + "grad_norm": 1.9740769863128662, + "learning_rate": 1e-06, + "loss": 0.5741, + "mean_token_accuracy": 0.8015072345733643, + "num_tokens": 591560498.0, + "step": 23336 + }, + { + "epoch": 1.9243836068277398, + "grad_norm": 1.9515951871871948, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7379459738731384, + "num_tokens": 591593026.0, + "step": 23337 + }, + { + "epoch": 1.9244660674527911, + "grad_norm": 2.1265904903411865, + "learning_rate": 1e-06, + "loss": 0.7926, + "mean_token_accuracy": 0.748201847076416, + "num_tokens": 591622394.0, + "step": 23338 + }, + { + "epoch": 1.924548528077843, + "grad_norm": 1.9810607433319092, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7685209512710571, + "num_tokens": 591656359.0, + "step": 23339 + }, + { + "epoch": 1.9246309887028943, + "grad_norm": 2.09529972076416, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.7406457662582397, + "num_tokens": 591681640.0, + "step": 23340 + }, + { + "epoch": 1.9247134493279459, + "grad_norm": 2.0434553623199463, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7651739120483398, + "num_tokens": 591711025.0, + "step": 23341 + }, + { + "epoch": 1.9247959099529974, + "grad_norm": 1.8697035312652588, + "learning_rate": 1e-06, + "loss": 0.6984, + "mean_token_accuracy": 0.7749683856964111, + "num_tokens": 591743242.0, + "step": 23342 + }, + { + "epoch": 1.924878370578049, + "grad_norm": 1.908668875694275, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7712645530700684, + "num_tokens": 591774600.0, + "step": 23343 + }, + { + "epoch": 1.9249608312031006, + "grad_norm": 1.9191378355026245, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7463014125823975, + "num_tokens": 591806118.0, + "step": 23344 + }, + { + "epoch": 1.925043291828152, + "grad_norm": 1.850461721420288, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.749648928642273, + "num_tokens": 591843591.0, + "step": 23345 + }, + { + "epoch": 1.9251257524532037, + "grad_norm": 1.8216525316238403, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7701870799064636, + "num_tokens": 591875544.0, + "step": 23346 + }, + { + "epoch": 1.925208213078255, + "grad_norm": 2.24851393699646, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.7867841720581055, + "num_tokens": 591899983.0, + "step": 23347 + }, + { + "epoch": 1.9252906737033066, + "grad_norm": 1.8161094188690186, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.7874820232391357, + "num_tokens": 591932257.0, + "step": 23348 + }, + { + "epoch": 1.9253731343283582, + "grad_norm": 2.0677592754364014, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7518690228462219, + "num_tokens": 591960258.0, + "step": 23349 + }, + { + "epoch": 1.9254555949534098, + "grad_norm": 2.2790887355804443, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7274249196052551, + "num_tokens": 591986356.0, + "step": 23350 + }, + { + "epoch": 1.9255380555784614, + "grad_norm": 2.320941925048828, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7397298812866211, + "num_tokens": 592011143.0, + "step": 23351 + }, + { + "epoch": 1.9256205162035127, + "grad_norm": 1.9360198974609375, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7604066133499146, + "num_tokens": 592042725.0, + "step": 23352 + }, + { + "epoch": 1.9257029768285645, + "grad_norm": 1.9986519813537598, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7746248841285706, + "num_tokens": 592071192.0, + "step": 23353 + }, + { + "epoch": 1.9257854374536159, + "grad_norm": 1.9302794933319092, + "learning_rate": 1e-06, + "loss": 0.6066, + "mean_token_accuracy": 0.7904426455497742, + "num_tokens": 592098683.0, + "step": 23354 + }, + { + "epoch": 1.9258678980786674, + "grad_norm": 2.253880739212036, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7530336380004883, + "num_tokens": 592122818.0, + "step": 23355 + }, + { + "epoch": 1.925950358703719, + "grad_norm": 1.8416837453842163, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7555488348007202, + "num_tokens": 592158822.0, + "step": 23356 + }, + { + "epoch": 1.9260328193287704, + "grad_norm": 2.008903741836548, + "learning_rate": 1e-06, + "loss": 0.6074, + "mean_token_accuracy": 0.7941616773605347, + "num_tokens": 592184501.0, + "step": 23357 + }, + { + "epoch": 1.9261152799538221, + "grad_norm": 2.189624547958374, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6887818574905396, + "num_tokens": 592213085.0, + "step": 23358 + }, + { + "epoch": 1.9261977405788735, + "grad_norm": 2.0257673263549805, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7516840696334839, + "num_tokens": 592242314.0, + "step": 23359 + }, + { + "epoch": 1.9262802012039253, + "grad_norm": 2.1452138423919678, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7212511301040649, + "num_tokens": 592270961.0, + "step": 23360 + }, + { + "epoch": 1.9263626618289766, + "grad_norm": 2.1278066635131836, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7507423162460327, + "num_tokens": 592298445.0, + "step": 23361 + }, + { + "epoch": 1.9264451224540282, + "grad_norm": 1.9007039070129395, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7799503803253174, + "num_tokens": 592330497.0, + "step": 23362 + }, + { + "epoch": 1.9265275830790798, + "grad_norm": 2.0027782917022705, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7717186212539673, + "num_tokens": 592360229.0, + "step": 23363 + }, + { + "epoch": 1.9266100437041311, + "grad_norm": 2.0367636680603027, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7595541477203369, + "num_tokens": 592389812.0, + "step": 23364 + }, + { + "epoch": 1.926692504329183, + "grad_norm": 2.166412830352783, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.771736741065979, + "num_tokens": 592414617.0, + "step": 23365 + }, + { + "epoch": 1.9267749649542343, + "grad_norm": 1.7862486839294434, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7772851586341858, + "num_tokens": 592452468.0, + "step": 23366 + }, + { + "epoch": 1.9268574255792859, + "grad_norm": 2.172962188720703, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7605936527252197, + "num_tokens": 592476486.0, + "step": 23367 + }, + { + "epoch": 1.9269398862043374, + "grad_norm": 2.314918279647827, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7352803349494934, + "num_tokens": 592500754.0, + "step": 23368 + }, + { + "epoch": 1.927022346829389, + "grad_norm": 2.0752878189086914, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7472982406616211, + "num_tokens": 592530052.0, + "step": 23369 + }, + { + "epoch": 1.9271048074544406, + "grad_norm": 1.9316034317016602, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.7503482103347778, + "num_tokens": 592563809.0, + "step": 23370 + }, + { + "epoch": 1.927187268079492, + "grad_norm": 1.998181700706482, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7152317762374878, + "num_tokens": 592597121.0, + "step": 23371 + }, + { + "epoch": 1.9272697287045437, + "grad_norm": 2.0127110481262207, + "learning_rate": 1e-06, + "loss": 0.5643, + "mean_token_accuracy": 0.8165521621704102, + "num_tokens": 592629173.0, + "step": 23372 + }, + { + "epoch": 1.927352189329595, + "grad_norm": 1.9477368593215942, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7465823292732239, + "num_tokens": 592660109.0, + "step": 23373 + }, + { + "epoch": 1.9274346499546466, + "grad_norm": 1.896447777748108, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.752173662185669, + "num_tokens": 592692146.0, + "step": 23374 + }, + { + "epoch": 1.9275171105796982, + "grad_norm": 1.8081345558166504, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.7808403968811035, + "num_tokens": 592727840.0, + "step": 23375 + }, + { + "epoch": 1.9275995712047498, + "grad_norm": 2.1759684085845947, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7434836626052856, + "num_tokens": 592753826.0, + "step": 23376 + }, + { + "epoch": 1.9276820318298014, + "grad_norm": 2.1256988048553467, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.745015025138855, + "num_tokens": 592783575.0, + "step": 23377 + }, + { + "epoch": 1.9277644924548527, + "grad_norm": 2.1153626441955566, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7494113445281982, + "num_tokens": 592810775.0, + "step": 23378 + }, + { + "epoch": 1.9278469530799045, + "grad_norm": 1.7764071226119995, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.76871657371521, + "num_tokens": 592845273.0, + "step": 23379 + }, + { + "epoch": 1.9279294137049559, + "grad_norm": 1.852051019668579, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7659855484962463, + "num_tokens": 592876802.0, + "step": 23380 + }, + { + "epoch": 1.9280118743300074, + "grad_norm": 2.308720827102661, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7736181020736694, + "num_tokens": 592898698.0, + "step": 23381 + }, + { + "epoch": 1.928094334955059, + "grad_norm": 1.8130336999893188, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7443028092384338, + "num_tokens": 592935811.0, + "step": 23382 + }, + { + "epoch": 1.9281767955801103, + "grad_norm": 1.9007314443588257, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7545026540756226, + "num_tokens": 592968446.0, + "step": 23383 + }, + { + "epoch": 1.9282592562051621, + "grad_norm": 1.7748301029205322, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7657573223114014, + "num_tokens": 593004144.0, + "step": 23384 + }, + { + "epoch": 1.9283417168302135, + "grad_norm": 1.8734408617019653, + "learning_rate": 1e-06, + "loss": 0.5567, + "mean_token_accuracy": 0.8137680292129517, + "num_tokens": 593037954.0, + "step": 23385 + }, + { + "epoch": 1.928424177455265, + "grad_norm": 2.050548791885376, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7757517695426941, + "num_tokens": 593070552.0, + "step": 23386 + }, + { + "epoch": 1.9285066380803166, + "grad_norm": 1.9345217943191528, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7809979915618896, + "num_tokens": 593101579.0, + "step": 23387 + }, + { + "epoch": 1.9285890987053682, + "grad_norm": 1.791883111000061, + "learning_rate": 1e-06, + "loss": 0.5268, + "mean_token_accuracy": 0.8304456472396851, + "num_tokens": 593133408.0, + "step": 23388 + }, + { + "epoch": 1.9286715593304198, + "grad_norm": 1.9856630563735962, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.7960386276245117, + "num_tokens": 593157937.0, + "step": 23389 + }, + { + "epoch": 1.9287540199554711, + "grad_norm": 2.034303665161133, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7694370746612549, + "num_tokens": 593184841.0, + "step": 23390 + }, + { + "epoch": 1.928836480580523, + "grad_norm": 1.945349931716919, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.8046436905860901, + "num_tokens": 593214121.0, + "step": 23391 + }, + { + "epoch": 1.9289189412055743, + "grad_norm": 1.905360460281372, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7273222804069519, + "num_tokens": 593247027.0, + "step": 23392 + }, + { + "epoch": 1.9290014018306259, + "grad_norm": 1.8642903566360474, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7275851368904114, + "num_tokens": 593280150.0, + "step": 23393 + }, + { + "epoch": 1.9290838624556774, + "grad_norm": 2.1342618465423584, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7518474459648132, + "num_tokens": 593306178.0, + "step": 23394 + }, + { + "epoch": 1.929166323080729, + "grad_norm": 1.956353783607483, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7602168917655945, + "num_tokens": 593339382.0, + "step": 23395 + }, + { + "epoch": 1.9292487837057806, + "grad_norm": 1.767361044883728, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7881036996841431, + "num_tokens": 593374881.0, + "step": 23396 + }, + { + "epoch": 1.929331244330832, + "grad_norm": 2.0018110275268555, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7497248649597168, + "num_tokens": 593405769.0, + "step": 23397 + }, + { + "epoch": 1.9294137049558837, + "grad_norm": 1.9740594625473022, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7507643699645996, + "num_tokens": 593435765.0, + "step": 23398 + }, + { + "epoch": 1.929496165580935, + "grad_norm": 2.035897970199585, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7585999965667725, + "num_tokens": 593463676.0, + "step": 23399 + }, + { + "epoch": 1.9295786262059866, + "grad_norm": 2.1392741203308105, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7533437013626099, + "num_tokens": 593491081.0, + "step": 23400 + }, + { + "epoch": 1.9296610868310382, + "grad_norm": 2.0037622451782227, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7292289137840271, + "num_tokens": 593521432.0, + "step": 23401 + }, + { + "epoch": 1.9297435474560896, + "grad_norm": 2.201219320297241, + "learning_rate": 1e-06, + "loss": 0.5912, + "mean_token_accuracy": 0.8020819425582886, + "num_tokens": 593546341.0, + "step": 23402 + }, + { + "epoch": 1.9298260080811414, + "grad_norm": 1.9211463928222656, + "learning_rate": 1e-06, + "loss": 0.5194, + "mean_token_accuracy": 0.8256657123565674, + "num_tokens": 593575814.0, + "step": 23403 + }, + { + "epoch": 1.9299084687061927, + "grad_norm": 2.2660624980926514, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7453736066818237, + "num_tokens": 593598640.0, + "step": 23404 + }, + { + "epoch": 1.9299909293312443, + "grad_norm": 1.9202154874801636, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7741559147834778, + "num_tokens": 593629004.0, + "step": 23405 + }, + { + "epoch": 1.9300733899562958, + "grad_norm": 1.9304205179214478, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7809127569198608, + "num_tokens": 593658804.0, + "step": 23406 + }, + { + "epoch": 1.9301558505813474, + "grad_norm": 1.8975270986557007, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7713485956192017, + "num_tokens": 593689622.0, + "step": 23407 + }, + { + "epoch": 1.930238311206399, + "grad_norm": 1.8829538822174072, + "learning_rate": 1e-06, + "loss": 0.7702, + "mean_token_accuracy": 0.7515847086906433, + "num_tokens": 593723472.0, + "step": 23408 + }, + { + "epoch": 1.9303207718314503, + "grad_norm": 1.957722544670105, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.7867350578308105, + "num_tokens": 593754188.0, + "step": 23409 + }, + { + "epoch": 1.9304032324565021, + "grad_norm": 1.9679491519927979, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7498222589492798, + "num_tokens": 593787355.0, + "step": 23410 + }, + { + "epoch": 1.9304856930815535, + "grad_norm": 2.005774974822998, + "learning_rate": 1e-06, + "loss": 0.6025, + "mean_token_accuracy": 0.798461377620697, + "num_tokens": 593813956.0, + "step": 23411 + }, + { + "epoch": 1.930568153706605, + "grad_norm": 1.6830397844314575, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7755140066146851, + "num_tokens": 593851821.0, + "step": 23412 + }, + { + "epoch": 1.9306506143316566, + "grad_norm": 1.9961817264556885, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7506391406059265, + "num_tokens": 593881008.0, + "step": 23413 + }, + { + "epoch": 1.9307330749567082, + "grad_norm": 2.1232199668884277, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7555069923400879, + "num_tokens": 593907902.0, + "step": 23414 + }, + { + "epoch": 1.9308155355817598, + "grad_norm": 2.115157127380371, + "learning_rate": 1e-06, + "loss": 0.5994, + "mean_token_accuracy": 0.7925147414207458, + "num_tokens": 593934153.0, + "step": 23415 + }, + { + "epoch": 1.9308979962068111, + "grad_norm": 2.0371594429016113, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7414940595626831, + "num_tokens": 593962607.0, + "step": 23416 + }, + { + "epoch": 1.930980456831863, + "grad_norm": 1.9303325414657593, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7462939620018005, + "num_tokens": 593991185.0, + "step": 23417 + }, + { + "epoch": 1.9310629174569143, + "grad_norm": 1.9322655200958252, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.755810022354126, + "num_tokens": 594020198.0, + "step": 23418 + }, + { + "epoch": 1.9311453780819658, + "grad_norm": 2.044229030609131, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.7783111333847046, + "num_tokens": 594046294.0, + "step": 23419 + }, + { + "epoch": 1.9312278387070174, + "grad_norm": 1.8845384120941162, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7359398603439331, + "num_tokens": 594079938.0, + "step": 23420 + }, + { + "epoch": 1.931310299332069, + "grad_norm": 2.235830783843994, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7618588805198669, + "num_tokens": 594103875.0, + "step": 23421 + }, + { + "epoch": 1.9313927599571206, + "grad_norm": 2.44753098487854, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7223862409591675, + "num_tokens": 594125567.0, + "step": 23422 + }, + { + "epoch": 1.931475220582172, + "grad_norm": 1.8939323425292969, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7734844088554382, + "num_tokens": 594156820.0, + "step": 23423 + }, + { + "epoch": 1.9315576812072237, + "grad_norm": 1.9968031644821167, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7773885726928711, + "num_tokens": 594186268.0, + "step": 23424 + }, + { + "epoch": 1.931640141832275, + "grad_norm": 1.742483139038086, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.7936499118804932, + "num_tokens": 594221004.0, + "step": 23425 + }, + { + "epoch": 1.9317226024573266, + "grad_norm": 1.9128841161727905, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.7914086580276489, + "num_tokens": 594250995.0, + "step": 23426 + }, + { + "epoch": 1.9318050630823782, + "grad_norm": 2.1962153911590576, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7735739946365356, + "num_tokens": 594276398.0, + "step": 23427 + }, + { + "epoch": 1.9318875237074296, + "grad_norm": 1.988397240638733, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7881957292556763, + "num_tokens": 594302199.0, + "step": 23428 + }, + { + "epoch": 1.9319699843324813, + "grad_norm": 2.0320212841033936, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.755242109298706, + "num_tokens": 594329582.0, + "step": 23429 + }, + { + "epoch": 1.9320524449575327, + "grad_norm": 1.839508295059204, + "learning_rate": 1e-06, + "loss": 0.6119, + "mean_token_accuracy": 0.7988143563270569, + "num_tokens": 594358855.0, + "step": 23430 + }, + { + "epoch": 1.9321349055825843, + "grad_norm": 2.0302109718322754, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7198808789253235, + "num_tokens": 594389359.0, + "step": 23431 + }, + { + "epoch": 1.9322173662076358, + "grad_norm": 1.8956725597381592, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7509723901748657, + "num_tokens": 594423137.0, + "step": 23432 + }, + { + "epoch": 1.9322998268326874, + "grad_norm": 1.8743995428085327, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7651124596595764, + "num_tokens": 594457539.0, + "step": 23433 + }, + { + "epoch": 1.932382287457739, + "grad_norm": 2.0030124187469482, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.775810956954956, + "num_tokens": 594488019.0, + "step": 23434 + }, + { + "epoch": 1.9324647480827903, + "grad_norm": 2.0500712394714355, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.745600163936615, + "num_tokens": 594518585.0, + "step": 23435 + }, + { + "epoch": 1.9325472087078421, + "grad_norm": 2.0253002643585205, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7702856063842773, + "num_tokens": 594548745.0, + "step": 23436 + }, + { + "epoch": 1.9326296693328935, + "grad_norm": 2.058551788330078, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7642356157302856, + "num_tokens": 594576604.0, + "step": 23437 + }, + { + "epoch": 1.932712129957945, + "grad_norm": 2.358881950378418, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.6978446841239929, + "num_tokens": 594603405.0, + "step": 23438 + }, + { + "epoch": 1.9327945905829966, + "grad_norm": 1.8773750066757202, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7446073293685913, + "num_tokens": 594638406.0, + "step": 23439 + }, + { + "epoch": 1.9328770512080482, + "grad_norm": 2.160645008087158, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7379465699195862, + "num_tokens": 594667320.0, + "step": 23440 + }, + { + "epoch": 1.9329595118330998, + "grad_norm": 2.1940391063690186, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7527992725372314, + "num_tokens": 594696594.0, + "step": 23441 + }, + { + "epoch": 1.9330419724581511, + "grad_norm": 2.1229965686798096, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7681554555892944, + "num_tokens": 594735091.0, + "step": 23442 + }, + { + "epoch": 1.933124433083203, + "grad_norm": 2.084993600845337, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7816387414932251, + "num_tokens": 594761540.0, + "step": 23443 + }, + { + "epoch": 1.9332068937082543, + "grad_norm": 1.9056520462036133, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7756761312484741, + "num_tokens": 594792379.0, + "step": 23444 + }, + { + "epoch": 1.9332893543333058, + "grad_norm": 2.10793399810791, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7587361931800842, + "num_tokens": 594819659.0, + "step": 23445 + }, + { + "epoch": 1.9333718149583574, + "grad_norm": 1.7398552894592285, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7593545913696289, + "num_tokens": 594858440.0, + "step": 23446 + }, + { + "epoch": 1.9334542755834088, + "grad_norm": 2.247069835662842, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7572959065437317, + "num_tokens": 594885524.0, + "step": 23447 + }, + { + "epoch": 1.9335367362084606, + "grad_norm": 2.0862722396850586, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7736644744873047, + "num_tokens": 594912720.0, + "step": 23448 + }, + { + "epoch": 1.933619196833512, + "grad_norm": 1.749345064163208, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7651737332344055, + "num_tokens": 594949486.0, + "step": 23449 + }, + { + "epoch": 1.9337016574585635, + "grad_norm": 1.907044529914856, + "learning_rate": 1e-06, + "loss": 0.6069, + "mean_token_accuracy": 0.8016596436500549, + "num_tokens": 594980477.0, + "step": 23450 + }, + { + "epoch": 1.933784118083615, + "grad_norm": 2.169902801513672, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7504894137382507, + "num_tokens": 595006532.0, + "step": 23451 + }, + { + "epoch": 1.9338665787086666, + "grad_norm": 2.1016180515289307, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7680277824401855, + "num_tokens": 595034887.0, + "step": 23452 + }, + { + "epoch": 1.9339490393337182, + "grad_norm": 2.0265681743621826, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7583698034286499, + "num_tokens": 595062362.0, + "step": 23453 + }, + { + "epoch": 1.9340314999587696, + "grad_norm": 1.8257445096969604, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.7839638590812683, + "num_tokens": 595095672.0, + "step": 23454 + }, + { + "epoch": 1.9341139605838213, + "grad_norm": 1.8304057121276855, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.7854849100112915, + "num_tokens": 595128659.0, + "step": 23455 + }, + { + "epoch": 1.9341964212088727, + "grad_norm": 2.015484094619751, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7600615620613098, + "num_tokens": 595159215.0, + "step": 23456 + }, + { + "epoch": 1.9342788818339243, + "grad_norm": 1.9212474822998047, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7724312543869019, + "num_tokens": 595189035.0, + "step": 23457 + }, + { + "epoch": 1.9343613424589758, + "grad_norm": 2.0109405517578125, + "learning_rate": 1e-06, + "loss": 0.7706, + "mean_token_accuracy": 0.7559620141983032, + "num_tokens": 595218592.0, + "step": 23458 + }, + { + "epoch": 1.9344438030840274, + "grad_norm": 1.909440279006958, + "learning_rate": 1e-06, + "loss": 0.6021, + "mean_token_accuracy": 0.7969363331794739, + "num_tokens": 595247931.0, + "step": 23459 + }, + { + "epoch": 1.934526263709079, + "grad_norm": 1.8671998977661133, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7710720300674438, + "num_tokens": 595279653.0, + "step": 23460 + }, + { + "epoch": 1.9346087243341303, + "grad_norm": 2.153634786605835, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.7849482297897339, + "num_tokens": 595305454.0, + "step": 23461 + }, + { + "epoch": 1.9346911849591821, + "grad_norm": 2.1755423545837402, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7166633605957031, + "num_tokens": 595333706.0, + "step": 23462 + }, + { + "epoch": 1.9347736455842335, + "grad_norm": 2.189816474914551, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7676342725753784, + "num_tokens": 595358205.0, + "step": 23463 + }, + { + "epoch": 1.934856106209285, + "grad_norm": 2.0330312252044678, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7708996534347534, + "num_tokens": 595386885.0, + "step": 23464 + }, + { + "epoch": 1.9349385668343366, + "grad_norm": 2.0742106437683105, + "learning_rate": 1e-06, + "loss": 0.6017, + "mean_token_accuracy": 0.7896604537963867, + "num_tokens": 595414024.0, + "step": 23465 + }, + { + "epoch": 1.935021027459388, + "grad_norm": 2.026413679122925, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.7933076620101929, + "num_tokens": 595444423.0, + "step": 23466 + }, + { + "epoch": 1.9351034880844398, + "grad_norm": 1.8402053117752075, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.7868393659591675, + "num_tokens": 595477370.0, + "step": 23467 + }, + { + "epoch": 1.9351859487094911, + "grad_norm": 1.8113429546356201, + "learning_rate": 1e-06, + "loss": 0.4407, + "mean_token_accuracy": 0.8366850018501282, + "num_tokens": 595506008.0, + "step": 23468 + }, + { + "epoch": 1.935268409334543, + "grad_norm": 2.1643385887145996, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7744396924972534, + "num_tokens": 595531637.0, + "step": 23469 + }, + { + "epoch": 1.9353508699595943, + "grad_norm": 2.113236665725708, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7796294689178467, + "num_tokens": 595559700.0, + "step": 23470 + }, + { + "epoch": 1.9354333305846458, + "grad_norm": 1.9979100227355957, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.774198591709137, + "num_tokens": 595586794.0, + "step": 23471 + }, + { + "epoch": 1.9355157912096974, + "grad_norm": 2.5663464069366455, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7268049120903015, + "num_tokens": 595618205.0, + "step": 23472 + }, + { + "epoch": 1.9355982518347488, + "grad_norm": 1.917547583580017, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.768675684928894, + "num_tokens": 595649426.0, + "step": 23473 + }, + { + "epoch": 1.9356807124598006, + "grad_norm": 2.1247823238372803, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7639765739440918, + "num_tokens": 595674043.0, + "step": 23474 + }, + { + "epoch": 1.935763173084852, + "grad_norm": 2.0525779724121094, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7337870597839355, + "num_tokens": 595704340.0, + "step": 23475 + }, + { + "epoch": 1.9358456337099035, + "grad_norm": 2.049414873123169, + "learning_rate": 1e-06, + "loss": 0.5886, + "mean_token_accuracy": 0.7893526554107666, + "num_tokens": 595731282.0, + "step": 23476 + }, + { + "epoch": 1.935928094334955, + "grad_norm": 1.9929426908493042, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7572640180587769, + "num_tokens": 595763815.0, + "step": 23477 + }, + { + "epoch": 1.9360105549600066, + "grad_norm": 1.713528037071228, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.7933836579322815, + "num_tokens": 595797590.0, + "step": 23478 + }, + { + "epoch": 1.9360930155850582, + "grad_norm": 2.1040966510772705, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7741612195968628, + "num_tokens": 595823453.0, + "step": 23479 + }, + { + "epoch": 1.9361754762101095, + "grad_norm": 2.2560598850250244, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7374546527862549, + "num_tokens": 595847889.0, + "step": 23480 + }, + { + "epoch": 1.9362579368351613, + "grad_norm": 1.7223131656646729, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7718842029571533, + "num_tokens": 595883193.0, + "step": 23481 + }, + { + "epoch": 1.9363403974602127, + "grad_norm": 1.769003987312317, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7814527750015259, + "num_tokens": 595918713.0, + "step": 23482 + }, + { + "epoch": 1.9364228580852643, + "grad_norm": 2.3279383182525635, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7500964403152466, + "num_tokens": 595942623.0, + "step": 23483 + }, + { + "epoch": 1.9365053187103158, + "grad_norm": 2.0934557914733887, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7548596858978271, + "num_tokens": 595965460.0, + "step": 23484 + }, + { + "epoch": 1.9365877793353674, + "grad_norm": 1.8940898180007935, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7625759840011597, + "num_tokens": 595997690.0, + "step": 23485 + }, + { + "epoch": 1.936670239960419, + "grad_norm": 2.1817238330841064, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7819622755050659, + "num_tokens": 596021682.0, + "step": 23486 + }, + { + "epoch": 1.9367527005854703, + "grad_norm": 2.394108295440674, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.734089732170105, + "num_tokens": 596045751.0, + "step": 23487 + }, + { + "epoch": 1.9368351612105221, + "grad_norm": 1.9472417831420898, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.7944785952568054, + "num_tokens": 596076299.0, + "step": 23488 + }, + { + "epoch": 1.9369176218355735, + "grad_norm": 2.1264359951019287, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7181342840194702, + "num_tokens": 596104843.0, + "step": 23489 + }, + { + "epoch": 1.937000082460625, + "grad_norm": 1.971484661102295, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7582840919494629, + "num_tokens": 596134900.0, + "step": 23490 + }, + { + "epoch": 1.9370825430856766, + "grad_norm": 2.006622314453125, + "learning_rate": 1e-06, + "loss": 0.7988, + "mean_token_accuracy": 0.7399752140045166, + "num_tokens": 596165205.0, + "step": 23491 + }, + { + "epoch": 1.937165003710728, + "grad_norm": 1.7823976278305054, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7788401246070862, + "num_tokens": 596199520.0, + "step": 23492 + }, + { + "epoch": 1.9372474643357798, + "grad_norm": 2.0163474082946777, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7592050433158875, + "num_tokens": 596229878.0, + "step": 23493 + }, + { + "epoch": 1.9373299249608311, + "grad_norm": 1.8245253562927246, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7694167494773865, + "num_tokens": 596265338.0, + "step": 23494 + }, + { + "epoch": 1.9374123855858827, + "grad_norm": 2.1076321601867676, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.789207398891449, + "num_tokens": 596289835.0, + "step": 23495 + }, + { + "epoch": 1.9374948462109343, + "grad_norm": 2.440458297729492, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.761603593826294, + "num_tokens": 596311125.0, + "step": 23496 + }, + { + "epoch": 1.9375773068359858, + "grad_norm": 1.8927282094955444, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.776910662651062, + "num_tokens": 596340011.0, + "step": 23497 + }, + { + "epoch": 1.9376597674610374, + "grad_norm": 2.1376564502716064, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7641064524650574, + "num_tokens": 596368475.0, + "step": 23498 + }, + { + "epoch": 1.9377422280860888, + "grad_norm": 2.1514737606048584, + "learning_rate": 1e-06, + "loss": 0.7396, + "mean_token_accuracy": 0.7614070177078247, + "num_tokens": 596396258.0, + "step": 23499 + }, + { + "epoch": 1.9378246887111406, + "grad_norm": 2.0347397327423096, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7706410884857178, + "num_tokens": 596422263.0, + "step": 23500 + }, + { + "epoch": 1.937907149336192, + "grad_norm": 2.360347270965576, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7578341960906982, + "num_tokens": 596445433.0, + "step": 23501 + }, + { + "epoch": 1.9379896099612435, + "grad_norm": 1.7484554052352905, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7564281225204468, + "num_tokens": 596487872.0, + "step": 23502 + }, + { + "epoch": 1.938072070586295, + "grad_norm": 2.0270655155181885, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.7942681312561035, + "num_tokens": 596515429.0, + "step": 23503 + }, + { + "epoch": 1.9381545312113466, + "grad_norm": 2.0288639068603516, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7486971616744995, + "num_tokens": 596545062.0, + "step": 23504 + }, + { + "epoch": 1.9382369918363982, + "grad_norm": 2.178138494491577, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7300355434417725, + "num_tokens": 596574677.0, + "step": 23505 + }, + { + "epoch": 1.9383194524614495, + "grad_norm": 1.788193702697754, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.773762583732605, + "num_tokens": 596609229.0, + "step": 23506 + }, + { + "epoch": 1.9384019130865013, + "grad_norm": 1.5939223766326904, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.7867510318756104, + "num_tokens": 596649944.0, + "step": 23507 + }, + { + "epoch": 1.9384843737115527, + "grad_norm": 1.9147515296936035, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.7901362180709839, + "num_tokens": 596679956.0, + "step": 23508 + }, + { + "epoch": 1.9385668343366043, + "grad_norm": 1.9720045328140259, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7485170364379883, + "num_tokens": 596712956.0, + "step": 23509 + }, + { + "epoch": 1.9386492949616558, + "grad_norm": 2.0886833667755127, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.781352162361145, + "num_tokens": 596737838.0, + "step": 23510 + }, + { + "epoch": 1.9387317555867072, + "grad_norm": 2.0522000789642334, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.7737802267074585, + "num_tokens": 596767634.0, + "step": 23511 + }, + { + "epoch": 1.938814216211759, + "grad_norm": 2.0354397296905518, + "learning_rate": 1e-06, + "loss": 0.7926, + "mean_token_accuracy": 0.7513779401779175, + "num_tokens": 596796843.0, + "step": 23512 + }, + { + "epoch": 1.9388966768368103, + "grad_norm": 1.7994738817214966, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.745669960975647, + "num_tokens": 596829761.0, + "step": 23513 + }, + { + "epoch": 1.9389791374618621, + "grad_norm": 1.8697717189788818, + "learning_rate": 1e-06, + "loss": 0.5513, + "mean_token_accuracy": 0.8090000748634338, + "num_tokens": 596862021.0, + "step": 23514 + }, + { + "epoch": 1.9390615980869135, + "grad_norm": 1.9481866359710693, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7506238222122192, + "num_tokens": 596893403.0, + "step": 23515 + }, + { + "epoch": 1.939144058711965, + "grad_norm": 1.8372918367385864, + "learning_rate": 1e-06, + "loss": 0.5507, + "mean_token_accuracy": 0.8141340017318726, + "num_tokens": 596923399.0, + "step": 23516 + }, + { + "epoch": 1.9392265193370166, + "grad_norm": 2.2768399715423584, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7135966420173645, + "num_tokens": 596949028.0, + "step": 23517 + }, + { + "epoch": 1.939308979962068, + "grad_norm": 1.8572624921798706, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.7491475939750671, + "num_tokens": 596981890.0, + "step": 23518 + }, + { + "epoch": 1.9393914405871198, + "grad_norm": 1.6742023229599, + "learning_rate": 1e-06, + "loss": 0.5611, + "mean_token_accuracy": 0.812035322189331, + "num_tokens": 597013376.0, + "step": 23519 + }, + { + "epoch": 1.9394739012121711, + "grad_norm": 2.012821912765503, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7287142276763916, + "num_tokens": 597044987.0, + "step": 23520 + }, + { + "epoch": 1.9395563618372227, + "grad_norm": 1.9449331760406494, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.747110903263092, + "num_tokens": 597077274.0, + "step": 23521 + }, + { + "epoch": 1.9396388224622743, + "grad_norm": 1.9018491506576538, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7809216380119324, + "num_tokens": 597105636.0, + "step": 23522 + }, + { + "epoch": 1.9397212830873258, + "grad_norm": 2.005143880844116, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.7847041487693787, + "num_tokens": 597135743.0, + "step": 23523 + }, + { + "epoch": 1.9398037437123774, + "grad_norm": 1.8107777833938599, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7723822593688965, + "num_tokens": 597168519.0, + "step": 23524 + }, + { + "epoch": 1.9398862043374288, + "grad_norm": 2.031437635421753, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7482129335403442, + "num_tokens": 597197615.0, + "step": 23525 + }, + { + "epoch": 1.9399686649624805, + "grad_norm": 2.2075376510620117, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7390670776367188, + "num_tokens": 597223845.0, + "step": 23526 + }, + { + "epoch": 1.940051125587532, + "grad_norm": 1.9072763919830322, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7588067650794983, + "num_tokens": 597256692.0, + "step": 23527 + }, + { + "epoch": 1.9401335862125835, + "grad_norm": 2.2693681716918945, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7704226970672607, + "num_tokens": 597283374.0, + "step": 23528 + }, + { + "epoch": 1.940216046837635, + "grad_norm": 2.0866620540618896, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7690984010696411, + "num_tokens": 597310669.0, + "step": 23529 + }, + { + "epoch": 1.9402985074626866, + "grad_norm": 1.8952231407165527, + "learning_rate": 1e-06, + "loss": 0.5216, + "mean_token_accuracy": 0.8241550922393799, + "num_tokens": 597340514.0, + "step": 23530 + }, + { + "epoch": 1.9403809680877382, + "grad_norm": 1.9884637594223022, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7482202053070068, + "num_tokens": 597371988.0, + "step": 23531 + }, + { + "epoch": 1.9404634287127895, + "grad_norm": 1.9417903423309326, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7705069184303284, + "num_tokens": 597403004.0, + "step": 23532 + }, + { + "epoch": 1.9405458893378413, + "grad_norm": 2.0333411693573, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7349141836166382, + "num_tokens": 597431779.0, + "step": 23533 + }, + { + "epoch": 1.9406283499628927, + "grad_norm": 1.9569168090820312, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7492800951004028, + "num_tokens": 597461536.0, + "step": 23534 + }, + { + "epoch": 1.9407108105879443, + "grad_norm": 2.209381341934204, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7530800104141235, + "num_tokens": 597487474.0, + "step": 23535 + }, + { + "epoch": 1.9407932712129958, + "grad_norm": 1.7950146198272705, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7677009105682373, + "num_tokens": 597523172.0, + "step": 23536 + }, + { + "epoch": 1.9408757318380472, + "grad_norm": 1.8279656171798706, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7733291387557983, + "num_tokens": 597556039.0, + "step": 23537 + }, + { + "epoch": 1.940958192463099, + "grad_norm": 1.8060905933380127, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7898703217506409, + "num_tokens": 597587431.0, + "step": 23538 + }, + { + "epoch": 1.9410406530881503, + "grad_norm": 1.705076813697815, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7301486730575562, + "num_tokens": 597625704.0, + "step": 23539 + }, + { + "epoch": 1.941123113713202, + "grad_norm": 2.0332348346710205, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7380638122558594, + "num_tokens": 597654547.0, + "step": 23540 + }, + { + "epoch": 1.9412055743382535, + "grad_norm": 1.7922879457473755, + "learning_rate": 1e-06, + "loss": 0.5925, + "mean_token_accuracy": 0.8030684590339661, + "num_tokens": 597685315.0, + "step": 23541 + }, + { + "epoch": 1.941288034963305, + "grad_norm": 2.0656423568725586, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7632229328155518, + "num_tokens": 597713761.0, + "step": 23542 + }, + { + "epoch": 1.9413704955883566, + "grad_norm": 2.0659339427948, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7391312122344971, + "num_tokens": 597744125.0, + "step": 23543 + }, + { + "epoch": 1.941452956213408, + "grad_norm": 1.8081365823745728, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.7749921083450317, + "num_tokens": 597778565.0, + "step": 23544 + }, + { + "epoch": 1.9415354168384598, + "grad_norm": 2.368762493133545, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7537485361099243, + "num_tokens": 597805167.0, + "step": 23545 + }, + { + "epoch": 1.941617877463511, + "grad_norm": 1.8500444889068604, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7843093276023865, + "num_tokens": 597838448.0, + "step": 23546 + }, + { + "epoch": 1.9417003380885627, + "grad_norm": 1.9854248762130737, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7782026529312134, + "num_tokens": 597870041.0, + "step": 23547 + }, + { + "epoch": 1.9417827987136143, + "grad_norm": 2.0517165660858154, + "learning_rate": 1e-06, + "loss": 0.5382, + "mean_token_accuracy": 0.8047158122062683, + "num_tokens": 597896365.0, + "step": 23548 + }, + { + "epoch": 1.9418652593386658, + "grad_norm": 2.334874153137207, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7674397230148315, + "num_tokens": 597917634.0, + "step": 23549 + }, + { + "epoch": 1.9419477199637174, + "grad_norm": 2.229344129562378, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7259719371795654, + "num_tokens": 597944729.0, + "step": 23550 + }, + { + "epoch": 1.9420301805887688, + "grad_norm": 2.1428263187408447, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.7838572263717651, + "num_tokens": 597971616.0, + "step": 23551 + }, + { + "epoch": 1.9421126412138205, + "grad_norm": 2.0813944339752197, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7079886198043823, + "num_tokens": 598000320.0, + "step": 23552 + }, + { + "epoch": 1.942195101838872, + "grad_norm": 1.944098711013794, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7511913776397705, + "num_tokens": 598031126.0, + "step": 23553 + }, + { + "epoch": 1.9422775624639235, + "grad_norm": 2.0127458572387695, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7473344802856445, + "num_tokens": 598062804.0, + "step": 23554 + }, + { + "epoch": 1.942360023088975, + "grad_norm": 2.203796148300171, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.732785701751709, + "num_tokens": 598090053.0, + "step": 23555 + }, + { + "epoch": 1.9424424837140264, + "grad_norm": 2.0846359729766846, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7857812643051147, + "num_tokens": 598116699.0, + "step": 23556 + }, + { + "epoch": 1.9425249443390782, + "grad_norm": 1.8434339761734009, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7438574433326721, + "num_tokens": 598150175.0, + "step": 23557 + }, + { + "epoch": 1.9426074049641295, + "grad_norm": 1.8592519760131836, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.7871667742729187, + "num_tokens": 598185149.0, + "step": 23558 + }, + { + "epoch": 1.9426898655891813, + "grad_norm": 2.2732365131378174, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.694351851940155, + "num_tokens": 598211236.0, + "step": 23559 + }, + { + "epoch": 1.9427723262142327, + "grad_norm": 2.025933265686035, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.7942056655883789, + "num_tokens": 598238493.0, + "step": 23560 + }, + { + "epoch": 1.9428547868392843, + "grad_norm": 2.1310815811157227, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7433995604515076, + "num_tokens": 598265760.0, + "step": 23561 + }, + { + "epoch": 1.9429372474643358, + "grad_norm": 1.837497353553772, + "learning_rate": 1e-06, + "loss": 0.6251, + "mean_token_accuracy": 0.7866749167442322, + "num_tokens": 598300511.0, + "step": 23562 + }, + { + "epoch": 1.9430197080893872, + "grad_norm": 1.8079009056091309, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7635855674743652, + "num_tokens": 598336588.0, + "step": 23563 + }, + { + "epoch": 1.943102168714439, + "grad_norm": 2.076751708984375, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.738511323928833, + "num_tokens": 598367302.0, + "step": 23564 + }, + { + "epoch": 1.9431846293394903, + "grad_norm": 2.2801592350006104, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7738326787948608, + "num_tokens": 598389859.0, + "step": 23565 + }, + { + "epoch": 1.943267089964542, + "grad_norm": 2.1987597942352295, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7660382986068726, + "num_tokens": 598411931.0, + "step": 23566 + }, + { + "epoch": 1.9433495505895935, + "grad_norm": 2.0421578884124756, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.794049859046936, + "num_tokens": 598439511.0, + "step": 23567 + }, + { + "epoch": 1.943432011214645, + "grad_norm": 2.001397132873535, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7737632393836975, + "num_tokens": 598466501.0, + "step": 23568 + }, + { + "epoch": 1.9435144718396966, + "grad_norm": 1.9340898990631104, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.7971972227096558, + "num_tokens": 598493729.0, + "step": 23569 + }, + { + "epoch": 1.943596932464748, + "grad_norm": 2.133014440536499, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7785265445709229, + "num_tokens": 598521907.0, + "step": 23570 + }, + { + "epoch": 1.9436793930897998, + "grad_norm": 2.1219184398651123, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7398114204406738, + "num_tokens": 598550022.0, + "step": 23571 + }, + { + "epoch": 1.943761853714851, + "grad_norm": 2.3765931129455566, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7401648759841919, + "num_tokens": 598571425.0, + "step": 23572 + }, + { + "epoch": 1.9438443143399027, + "grad_norm": 1.7941653728485107, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.7993546724319458, + "num_tokens": 598607779.0, + "step": 23573 + }, + { + "epoch": 1.9439267749649543, + "grad_norm": 2.2301197052001953, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7735248804092407, + "num_tokens": 598631995.0, + "step": 23574 + }, + { + "epoch": 1.9440092355900058, + "grad_norm": 2.102667808532715, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7607645392417908, + "num_tokens": 598659453.0, + "step": 23575 + }, + { + "epoch": 1.9440916962150574, + "grad_norm": 1.9888932704925537, + "learning_rate": 1e-06, + "loss": 0.5477, + "mean_token_accuracy": 0.8167977929115295, + "num_tokens": 598686817.0, + "step": 23576 + }, + { + "epoch": 1.9441741568401087, + "grad_norm": 1.7994924783706665, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7682012319564819, + "num_tokens": 598719700.0, + "step": 23577 + }, + { + "epoch": 1.9442566174651605, + "grad_norm": 1.9550954103469849, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7150272130966187, + "num_tokens": 598753783.0, + "step": 23578 + }, + { + "epoch": 1.944339078090212, + "grad_norm": 2.215280294418335, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.749119222164154, + "num_tokens": 598782082.0, + "step": 23579 + }, + { + "epoch": 1.9444215387152635, + "grad_norm": 1.9426002502441406, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.770796537399292, + "num_tokens": 598811676.0, + "step": 23580 + }, + { + "epoch": 1.944503999340315, + "grad_norm": 2.031249761581421, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7321133017539978, + "num_tokens": 598840783.0, + "step": 23581 + }, + { + "epoch": 1.9445864599653664, + "grad_norm": 1.8525444269180298, + "learning_rate": 1e-06, + "loss": 0.5903, + "mean_token_accuracy": 0.8118963241577148, + "num_tokens": 598872163.0, + "step": 23582 + }, + { + "epoch": 1.9446689205904182, + "grad_norm": 2.047569513320923, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7809367775917053, + "num_tokens": 598900207.0, + "step": 23583 + }, + { + "epoch": 1.9447513812154695, + "grad_norm": 1.8735028505325317, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7458304166793823, + "num_tokens": 598932624.0, + "step": 23584 + }, + { + "epoch": 1.944833841840521, + "grad_norm": 1.7929108142852783, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7777883410453796, + "num_tokens": 598969394.0, + "step": 23585 + }, + { + "epoch": 1.9449163024655727, + "grad_norm": 1.9191246032714844, + "learning_rate": 1e-06, + "loss": 0.7598, + "mean_token_accuracy": 0.7519674897193909, + "num_tokens": 599000604.0, + "step": 23586 + }, + { + "epoch": 1.9449987630906242, + "grad_norm": 1.8233641386032104, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7737158536911011, + "num_tokens": 599033377.0, + "step": 23587 + }, + { + "epoch": 1.9450812237156758, + "grad_norm": 1.9169615507125854, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7569130659103394, + "num_tokens": 599065195.0, + "step": 23588 + }, + { + "epoch": 1.9451636843407272, + "grad_norm": 1.9949556589126587, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7611680030822754, + "num_tokens": 599095920.0, + "step": 23589 + }, + { + "epoch": 1.945246144965779, + "grad_norm": 2.160233974456787, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7603148818016052, + "num_tokens": 599120977.0, + "step": 23590 + }, + { + "epoch": 1.9453286055908303, + "grad_norm": 1.8808560371398926, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7391020059585571, + "num_tokens": 599154797.0, + "step": 23591 + }, + { + "epoch": 1.945411066215882, + "grad_norm": 2.1221225261688232, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7635664939880371, + "num_tokens": 599184304.0, + "step": 23592 + }, + { + "epoch": 1.9454935268409335, + "grad_norm": 2.1659815311431885, + "learning_rate": 1e-06, + "loss": 0.6299, + "mean_token_accuracy": 0.7852106094360352, + "num_tokens": 599207791.0, + "step": 23593 + }, + { + "epoch": 1.945575987465985, + "grad_norm": 2.0320513248443604, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7503887414932251, + "num_tokens": 599239111.0, + "step": 23594 + }, + { + "epoch": 1.9456584480910366, + "grad_norm": 1.815134048461914, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.7817350625991821, + "num_tokens": 599272398.0, + "step": 23595 + }, + { + "epoch": 1.945740908716088, + "grad_norm": 2.1377761363983154, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7316260933876038, + "num_tokens": 599299130.0, + "step": 23596 + }, + { + "epoch": 1.9458233693411398, + "grad_norm": 2.3625802993774414, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.6976722478866577, + "num_tokens": 599320630.0, + "step": 23597 + }, + { + "epoch": 1.945905829966191, + "grad_norm": 1.913019061088562, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6878259181976318, + "num_tokens": 599354654.0, + "step": 23598 + }, + { + "epoch": 1.9459882905912427, + "grad_norm": 1.9971758127212524, + "learning_rate": 1e-06, + "loss": 0.5956, + "mean_token_accuracy": 0.799407958984375, + "num_tokens": 599382424.0, + "step": 23599 + }, + { + "epoch": 1.9460707512162942, + "grad_norm": 1.956882357597351, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7377099394798279, + "num_tokens": 599412621.0, + "step": 23600 + }, + { + "epoch": 1.9461532118413456, + "grad_norm": 2.073070764541626, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7561556696891785, + "num_tokens": 599438659.0, + "step": 23601 + }, + { + "epoch": 1.9462356724663974, + "grad_norm": 1.8052623271942139, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7436848878860474, + "num_tokens": 599472650.0, + "step": 23602 + }, + { + "epoch": 1.9463181330914487, + "grad_norm": 1.8823802471160889, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.7905198931694031, + "num_tokens": 599504002.0, + "step": 23603 + }, + { + "epoch": 1.9464005937165003, + "grad_norm": 1.9347634315490723, + "learning_rate": 1e-06, + "loss": 0.6048, + "mean_token_accuracy": 0.8002577424049377, + "num_tokens": 599533158.0, + "step": 23604 + }, + { + "epoch": 1.9464830543415519, + "grad_norm": 2.1443538665771484, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.742051899433136, + "num_tokens": 599561220.0, + "step": 23605 + }, + { + "epoch": 1.9465655149666035, + "grad_norm": 1.802878975868225, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7518036961555481, + "num_tokens": 599597145.0, + "step": 23606 + }, + { + "epoch": 1.946647975591655, + "grad_norm": 2.3680994510650635, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7163107395172119, + "num_tokens": 599619603.0, + "step": 23607 + }, + { + "epoch": 1.9467304362167064, + "grad_norm": 2.148388624191284, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7698724269866943, + "num_tokens": 599645526.0, + "step": 23608 + }, + { + "epoch": 1.9468128968417582, + "grad_norm": 1.8500343561172485, + "learning_rate": 1e-06, + "loss": 0.5594, + "mean_token_accuracy": 0.8114677667617798, + "num_tokens": 599674230.0, + "step": 23609 + }, + { + "epoch": 1.9468953574668095, + "grad_norm": 1.7681357860565186, + "learning_rate": 1e-06, + "loss": 0.5842, + "mean_token_accuracy": 0.7940772175788879, + "num_tokens": 599707926.0, + "step": 23610 + }, + { + "epoch": 1.946977818091861, + "grad_norm": 2.035400390625, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7435541152954102, + "num_tokens": 599735838.0, + "step": 23611 + }, + { + "epoch": 1.9470602787169127, + "grad_norm": 1.9770543575286865, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7442082166671753, + "num_tokens": 599769424.0, + "step": 23612 + }, + { + "epoch": 1.9471427393419642, + "grad_norm": 2.098344326019287, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7637923359870911, + "num_tokens": 599795473.0, + "step": 23613 + }, + { + "epoch": 1.9472251999670158, + "grad_norm": 2.0525622367858887, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.7744760513305664, + "num_tokens": 599820159.0, + "step": 23614 + }, + { + "epoch": 1.9473076605920672, + "grad_norm": 2.0168492794036865, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7408654689788818, + "num_tokens": 599851550.0, + "step": 23615 + }, + { + "epoch": 1.947390121217119, + "grad_norm": 2.2368838787078857, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.752273440361023, + "num_tokens": 599879820.0, + "step": 23616 + }, + { + "epoch": 1.9474725818421703, + "grad_norm": 2.323665142059326, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7435226440429688, + "num_tokens": 599905042.0, + "step": 23617 + }, + { + "epoch": 1.9475550424672219, + "grad_norm": 2.433645486831665, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7313334941864014, + "num_tokens": 599928865.0, + "step": 23618 + }, + { + "epoch": 1.9476375030922735, + "grad_norm": 1.8143469095230103, + "learning_rate": 1e-06, + "loss": 0.7531, + "mean_token_accuracy": 0.7625585794448853, + "num_tokens": 599964258.0, + "step": 23619 + }, + { + "epoch": 1.947719963717325, + "grad_norm": 2.103909969329834, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.7977815866470337, + "num_tokens": 599990724.0, + "step": 23620 + }, + { + "epoch": 1.9478024243423766, + "grad_norm": 2.0239672660827637, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7379793524742126, + "num_tokens": 600019579.0, + "step": 23621 + }, + { + "epoch": 1.947884884967428, + "grad_norm": 2.1827385425567627, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.7541842460632324, + "num_tokens": 600046646.0, + "step": 23622 + }, + { + "epoch": 1.9479673455924797, + "grad_norm": 2.317120313644409, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.772819459438324, + "num_tokens": 600071761.0, + "step": 23623 + }, + { + "epoch": 1.948049806217531, + "grad_norm": 1.902103304862976, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7696431875228882, + "num_tokens": 600104334.0, + "step": 23624 + }, + { + "epoch": 1.9481322668425827, + "grad_norm": 2.2930715084075928, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7626324892044067, + "num_tokens": 600128350.0, + "step": 23625 + }, + { + "epoch": 1.9482147274676342, + "grad_norm": 2.16817045211792, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7711184024810791, + "num_tokens": 600155582.0, + "step": 23626 + }, + { + "epoch": 1.9482971880926856, + "grad_norm": 2.2261619567871094, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7231215238571167, + "num_tokens": 600181280.0, + "step": 23627 + }, + { + "epoch": 1.9483796487177374, + "grad_norm": 2.195664882659912, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.717528223991394, + "num_tokens": 600208176.0, + "step": 23628 + }, + { + "epoch": 1.9484621093427887, + "grad_norm": 1.8257554769515991, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7625973224639893, + "num_tokens": 600245155.0, + "step": 23629 + }, + { + "epoch": 1.9485445699678403, + "grad_norm": 1.902365803718567, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7719951272010803, + "num_tokens": 600275212.0, + "step": 23630 + }, + { + "epoch": 1.9486270305928919, + "grad_norm": 2.0732274055480957, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7633029222488403, + "num_tokens": 600306368.0, + "step": 23631 + }, + { + "epoch": 1.9487094912179435, + "grad_norm": 1.8064030408859253, + "learning_rate": 1e-06, + "loss": 0.7704, + "mean_token_accuracy": 0.7645388841629028, + "num_tokens": 600341017.0, + "step": 23632 + }, + { + "epoch": 1.948791951842995, + "grad_norm": 2.1691863536834717, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.7912101745605469, + "num_tokens": 600363983.0, + "step": 23633 + }, + { + "epoch": 1.9488744124680464, + "grad_norm": 2.0905065536499023, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7200954556465149, + "num_tokens": 600392827.0, + "step": 23634 + }, + { + "epoch": 1.9489568730930982, + "grad_norm": 1.9603540897369385, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7518457770347595, + "num_tokens": 600426536.0, + "step": 23635 + }, + { + "epoch": 1.9490393337181495, + "grad_norm": 1.8286489248275757, + "learning_rate": 1e-06, + "loss": 0.6137, + "mean_token_accuracy": 0.7956777811050415, + "num_tokens": 600459677.0, + "step": 23636 + }, + { + "epoch": 1.949121794343201, + "grad_norm": 1.9120746850967407, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.751236081123352, + "num_tokens": 600490500.0, + "step": 23637 + }, + { + "epoch": 1.9492042549682527, + "grad_norm": 2.143059730529785, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7710643410682678, + "num_tokens": 600517528.0, + "step": 23638 + }, + { + "epoch": 1.9492867155933042, + "grad_norm": 2.189793348312378, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.718638002872467, + "num_tokens": 600543249.0, + "step": 23639 + }, + { + "epoch": 1.9493691762183558, + "grad_norm": 1.7734822034835815, + "learning_rate": 1e-06, + "loss": 0.5624, + "mean_token_accuracy": 0.8183026313781738, + "num_tokens": 600575545.0, + "step": 23640 + }, + { + "epoch": 1.9494516368434072, + "grad_norm": 1.820851445198059, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7844254970550537, + "num_tokens": 600611045.0, + "step": 23641 + }, + { + "epoch": 1.949534097468459, + "grad_norm": 1.9994256496429443, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.8073854446411133, + "num_tokens": 600636677.0, + "step": 23642 + }, + { + "epoch": 1.9496165580935103, + "grad_norm": 1.7991541624069214, + "learning_rate": 1e-06, + "loss": 0.604, + "mean_token_accuracy": 0.7984145283699036, + "num_tokens": 600672261.0, + "step": 23643 + }, + { + "epoch": 1.9496990187185619, + "grad_norm": 1.7862411737442017, + "learning_rate": 1e-06, + "loss": 0.6005, + "mean_token_accuracy": 0.8028706908226013, + "num_tokens": 600704551.0, + "step": 23644 + }, + { + "epoch": 1.9497814793436135, + "grad_norm": 1.9862133264541626, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7618690133094788, + "num_tokens": 600733882.0, + "step": 23645 + }, + { + "epoch": 1.9498639399686648, + "grad_norm": 1.9289445877075195, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7651898860931396, + "num_tokens": 600763021.0, + "step": 23646 + }, + { + "epoch": 1.9499464005937166, + "grad_norm": 1.8556709289550781, + "learning_rate": 1e-06, + "loss": 0.7999, + "mean_token_accuracy": 0.7430106401443481, + "num_tokens": 600798010.0, + "step": 23647 + }, + { + "epoch": 1.950028861218768, + "grad_norm": 1.8571313619613647, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.7980660200119019, + "num_tokens": 600831419.0, + "step": 23648 + }, + { + "epoch": 1.9501113218438195, + "grad_norm": 1.9341626167297363, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7736241817474365, + "num_tokens": 600860478.0, + "step": 23649 + }, + { + "epoch": 1.950193782468871, + "grad_norm": 1.9343924522399902, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.786949872970581, + "num_tokens": 600891192.0, + "step": 23650 + }, + { + "epoch": 1.9502762430939227, + "grad_norm": 1.875931739807129, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7659652829170227, + "num_tokens": 600920594.0, + "step": 23651 + }, + { + "epoch": 1.9503587037189742, + "grad_norm": 2.115809917449951, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7412612438201904, + "num_tokens": 600948565.0, + "step": 23652 + }, + { + "epoch": 1.9504411643440256, + "grad_norm": 1.7553479671478271, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7749098539352417, + "num_tokens": 600984547.0, + "step": 23653 + }, + { + "epoch": 1.9505236249690774, + "grad_norm": 2.0751166343688965, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.7874274253845215, + "num_tokens": 601010883.0, + "step": 23654 + }, + { + "epoch": 1.9506060855941287, + "grad_norm": 1.790209174156189, + "learning_rate": 1e-06, + "loss": 0.565, + "mean_token_accuracy": 0.8221950531005859, + "num_tokens": 601045166.0, + "step": 23655 + }, + { + "epoch": 1.9506885462191803, + "grad_norm": 2.2472143173217773, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7852871417999268, + "num_tokens": 601069853.0, + "step": 23656 + }, + { + "epoch": 1.9507710068442319, + "grad_norm": 1.9830080270767212, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7489281892776489, + "num_tokens": 601103418.0, + "step": 23657 + }, + { + "epoch": 1.9508534674692835, + "grad_norm": 2.159085512161255, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7624141573905945, + "num_tokens": 601129372.0, + "step": 23658 + }, + { + "epoch": 1.950935928094335, + "grad_norm": 2.168767213821411, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7665546536445618, + "num_tokens": 601155109.0, + "step": 23659 + }, + { + "epoch": 1.9510183887193864, + "grad_norm": 2.2008206844329834, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7793624401092529, + "num_tokens": 601177838.0, + "step": 23660 + }, + { + "epoch": 1.9511008493444382, + "grad_norm": 2.265803575515747, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.781532347202301, + "num_tokens": 601200191.0, + "step": 23661 + }, + { + "epoch": 1.9511833099694895, + "grad_norm": 1.9785854816436768, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7626586556434631, + "num_tokens": 601228667.0, + "step": 23662 + }, + { + "epoch": 1.951265770594541, + "grad_norm": 1.9685965776443481, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.7774299383163452, + "num_tokens": 601256391.0, + "step": 23663 + }, + { + "epoch": 1.9513482312195927, + "grad_norm": 2.1099867820739746, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7260938882827759, + "num_tokens": 601284357.0, + "step": 23664 + }, + { + "epoch": 1.951430691844644, + "grad_norm": 2.1912927627563477, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7699629068374634, + "num_tokens": 601311619.0, + "step": 23665 + }, + { + "epoch": 1.9515131524696958, + "grad_norm": 1.981785535812378, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.7839276790618896, + "num_tokens": 601340384.0, + "step": 23666 + }, + { + "epoch": 1.9515956130947472, + "grad_norm": 2.3445825576782227, + "learning_rate": 1e-06, + "loss": 0.7144, + "mean_token_accuracy": 0.7703447937965393, + "num_tokens": 601365791.0, + "step": 23667 + }, + { + "epoch": 1.951678073719799, + "grad_norm": 1.729641079902649, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7562273740768433, + "num_tokens": 601404834.0, + "step": 23668 + }, + { + "epoch": 1.9517605343448503, + "grad_norm": 2.1899867057800293, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.7443537712097168, + "num_tokens": 601434181.0, + "step": 23669 + }, + { + "epoch": 1.9518429949699019, + "grad_norm": 2.0678176879882812, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7809504866600037, + "num_tokens": 601462511.0, + "step": 23670 + }, + { + "epoch": 1.9519254555949535, + "grad_norm": 1.8043410778045654, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.779064416885376, + "num_tokens": 601497795.0, + "step": 23671 + }, + { + "epoch": 1.9520079162200048, + "grad_norm": 1.7114894390106201, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7771683931350708, + "num_tokens": 601534882.0, + "step": 23672 + }, + { + "epoch": 1.9520903768450566, + "grad_norm": 1.9281548261642456, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7805672287940979, + "num_tokens": 601565528.0, + "step": 23673 + }, + { + "epoch": 1.952172837470108, + "grad_norm": 1.8141402006149292, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.779413640499115, + "num_tokens": 601595997.0, + "step": 23674 + }, + { + "epoch": 1.9522552980951595, + "grad_norm": 2.083422899246216, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7473225593566895, + "num_tokens": 601624223.0, + "step": 23675 + }, + { + "epoch": 1.952337758720211, + "grad_norm": 1.877187728881836, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7419201135635376, + "num_tokens": 601657250.0, + "step": 23676 + }, + { + "epoch": 1.9524202193452627, + "grad_norm": 2.170703887939453, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7653729915618896, + "num_tokens": 601682212.0, + "step": 23677 + }, + { + "epoch": 1.9525026799703142, + "grad_norm": 1.9959299564361572, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7301174402236938, + "num_tokens": 601713686.0, + "step": 23678 + }, + { + "epoch": 1.9525851405953656, + "grad_norm": 1.9576337337493896, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.768943190574646, + "num_tokens": 601746482.0, + "step": 23679 + }, + { + "epoch": 1.9526676012204174, + "grad_norm": 1.9077072143554688, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7300499081611633, + "num_tokens": 601778908.0, + "step": 23680 + }, + { + "epoch": 1.9527500618454687, + "grad_norm": 1.846581220626831, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.764469563961029, + "num_tokens": 601812449.0, + "step": 23681 + }, + { + "epoch": 1.9528325224705203, + "grad_norm": 2.0580015182495117, + "learning_rate": 1e-06, + "loss": 0.5626, + "mean_token_accuracy": 0.8149404525756836, + "num_tokens": 601839615.0, + "step": 23682 + }, + { + "epoch": 1.9529149830955719, + "grad_norm": 2.1121742725372314, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.7789996862411499, + "num_tokens": 601866827.0, + "step": 23683 + }, + { + "epoch": 1.9529974437206234, + "grad_norm": 2.027242422103882, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7374891042709351, + "num_tokens": 601896518.0, + "step": 23684 + }, + { + "epoch": 1.953079904345675, + "grad_norm": 2.0804381370544434, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.762172520160675, + "num_tokens": 601923053.0, + "step": 23685 + }, + { + "epoch": 1.9531623649707264, + "grad_norm": 2.070385456085205, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7526392340660095, + "num_tokens": 601949108.0, + "step": 23686 + }, + { + "epoch": 1.9532448255957782, + "grad_norm": 2.066723108291626, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7160353064537048, + "num_tokens": 601977943.0, + "step": 23687 + }, + { + "epoch": 1.9533272862208295, + "grad_norm": 1.8677552938461304, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7731088995933533, + "num_tokens": 602010387.0, + "step": 23688 + }, + { + "epoch": 1.953409746845881, + "grad_norm": 2.018644332885742, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7430218458175659, + "num_tokens": 602041614.0, + "step": 23689 + }, + { + "epoch": 1.9534922074709327, + "grad_norm": 2.1908442974090576, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7752830386161804, + "num_tokens": 602068199.0, + "step": 23690 + }, + { + "epoch": 1.953574668095984, + "grad_norm": 1.926647424697876, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7407952547073364, + "num_tokens": 602100365.0, + "step": 23691 + }, + { + "epoch": 1.9536571287210358, + "grad_norm": 1.978111982345581, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7503122091293335, + "num_tokens": 602130829.0, + "step": 23692 + }, + { + "epoch": 1.9537395893460872, + "grad_norm": 2.1362226009368896, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.7547808885574341, + "num_tokens": 602157658.0, + "step": 23693 + }, + { + "epoch": 1.9538220499711387, + "grad_norm": 1.9008945226669312, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7725324034690857, + "num_tokens": 602189226.0, + "step": 23694 + }, + { + "epoch": 1.9539045105961903, + "grad_norm": 2.2012476921081543, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7444549202919006, + "num_tokens": 602215506.0, + "step": 23695 + }, + { + "epoch": 1.9539869712212419, + "grad_norm": 1.8284034729003906, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7722359895706177, + "num_tokens": 602252055.0, + "step": 23696 + }, + { + "epoch": 1.9540694318462934, + "grad_norm": 2.0652403831481934, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7477869391441345, + "num_tokens": 602280240.0, + "step": 23697 + }, + { + "epoch": 1.9541518924713448, + "grad_norm": 2.014601469039917, + "learning_rate": 1e-06, + "loss": 0.7954, + "mean_token_accuracy": 0.7424858808517456, + "num_tokens": 602309079.0, + "step": 23698 + }, + { + "epoch": 1.9542343530963966, + "grad_norm": 2.1205949783325195, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7599581480026245, + "num_tokens": 602338242.0, + "step": 23699 + }, + { + "epoch": 1.954316813721448, + "grad_norm": 1.943510890007019, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.74069744348526, + "num_tokens": 602371025.0, + "step": 23700 + }, + { + "epoch": 1.9543992743464995, + "grad_norm": 1.8330758810043335, + "learning_rate": 1e-06, + "loss": 0.5672, + "mean_token_accuracy": 0.8162449598312378, + "num_tokens": 602403303.0, + "step": 23701 + }, + { + "epoch": 1.954481734971551, + "grad_norm": 1.965640664100647, + "learning_rate": 1e-06, + "loss": 0.7825, + "mean_token_accuracy": 0.7508847713470459, + "num_tokens": 602434717.0, + "step": 23702 + }, + { + "epoch": 1.9545641955966027, + "grad_norm": 1.8229548931121826, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7504866123199463, + "num_tokens": 602468311.0, + "step": 23703 + }, + { + "epoch": 1.9546466562216542, + "grad_norm": 2.0192129611968994, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7803747653961182, + "num_tokens": 602498238.0, + "step": 23704 + }, + { + "epoch": 1.9547291168467056, + "grad_norm": 1.67414391040802, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7770106196403503, + "num_tokens": 602534490.0, + "step": 23705 + }, + { + "epoch": 1.9548115774717574, + "grad_norm": 2.1456961631774902, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7823225855827332, + "num_tokens": 602559366.0, + "step": 23706 + }, + { + "epoch": 1.9548940380968087, + "grad_norm": 1.907153606414795, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7697620391845703, + "num_tokens": 602587775.0, + "step": 23707 + }, + { + "epoch": 1.9549764987218603, + "grad_norm": 2.167630195617676, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7577952146530151, + "num_tokens": 602612925.0, + "step": 23708 + }, + { + "epoch": 1.9550589593469119, + "grad_norm": 2.042080879211426, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7294524908065796, + "num_tokens": 602642263.0, + "step": 23709 + }, + { + "epoch": 1.9551414199719632, + "grad_norm": 2.0696301460266113, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7519865036010742, + "num_tokens": 602670566.0, + "step": 23710 + }, + { + "epoch": 1.955223880597015, + "grad_norm": 2.095590114593506, + "learning_rate": 1e-06, + "loss": 0.5487, + "mean_token_accuracy": 0.795743465423584, + "num_tokens": 602695011.0, + "step": 23711 + }, + { + "epoch": 1.9553063412220664, + "grad_norm": 1.8527501821517944, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.7899307012557983, + "num_tokens": 602726044.0, + "step": 23712 + }, + { + "epoch": 1.9553888018471182, + "grad_norm": 1.920102596282959, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7840564250946045, + "num_tokens": 602756502.0, + "step": 23713 + }, + { + "epoch": 1.9554712624721695, + "grad_norm": 1.9259310960769653, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.759436845779419, + "num_tokens": 602790039.0, + "step": 23714 + }, + { + "epoch": 1.955553723097221, + "grad_norm": 2.5694007873535156, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7683228254318237, + "num_tokens": 602810008.0, + "step": 23715 + }, + { + "epoch": 1.9556361837222727, + "grad_norm": 1.777413249015808, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7771273255348206, + "num_tokens": 602845421.0, + "step": 23716 + }, + { + "epoch": 1.955718644347324, + "grad_norm": 1.9562653303146362, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7367581129074097, + "num_tokens": 602875173.0, + "step": 23717 + }, + { + "epoch": 1.9558011049723758, + "grad_norm": 1.9887864589691162, + "learning_rate": 1e-06, + "loss": 0.5863, + "mean_token_accuracy": 0.8044695854187012, + "num_tokens": 602903481.0, + "step": 23718 + }, + { + "epoch": 1.9558835655974272, + "grad_norm": 2.240809440612793, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7755999565124512, + "num_tokens": 602928473.0, + "step": 23719 + }, + { + "epoch": 1.9559660262224787, + "grad_norm": 2.27433705329895, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7368658781051636, + "num_tokens": 602952635.0, + "step": 23720 + }, + { + "epoch": 1.9560484868475303, + "grad_norm": 2.098766565322876, + "learning_rate": 1e-06, + "loss": 0.7746, + "mean_token_accuracy": 0.7559880614280701, + "num_tokens": 602980153.0, + "step": 23721 + }, + { + "epoch": 1.9561309474725819, + "grad_norm": 1.870692253112793, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7746728658676147, + "num_tokens": 603012496.0, + "step": 23722 + }, + { + "epoch": 1.9562134080976334, + "grad_norm": 2.2181458473205566, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.7832974195480347, + "num_tokens": 603036265.0, + "step": 23723 + }, + { + "epoch": 1.9562958687226848, + "grad_norm": 1.5779728889465332, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.8007323741912842, + "num_tokens": 603080136.0, + "step": 23724 + }, + { + "epoch": 1.9563783293477366, + "grad_norm": 1.793387770652771, + "learning_rate": 1e-06, + "loss": 0.5833, + "mean_token_accuracy": 0.806840181350708, + "num_tokens": 603112779.0, + "step": 23725 + }, + { + "epoch": 1.956460789972788, + "grad_norm": 1.920568823814392, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7597947120666504, + "num_tokens": 603145254.0, + "step": 23726 + }, + { + "epoch": 1.9565432505978395, + "grad_norm": 1.8721940517425537, + "learning_rate": 1e-06, + "loss": 0.6059, + "mean_token_accuracy": 0.8030140399932861, + "num_tokens": 603173487.0, + "step": 23727 + }, + { + "epoch": 1.956625711222891, + "grad_norm": 1.62260103225708, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7621867656707764, + "num_tokens": 603215704.0, + "step": 23728 + }, + { + "epoch": 1.9567081718479427, + "grad_norm": 1.9588416814804077, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7848812341690063, + "num_tokens": 603249788.0, + "step": 23729 + }, + { + "epoch": 1.9567906324729942, + "grad_norm": 1.9958522319793701, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7459542751312256, + "num_tokens": 603280025.0, + "step": 23730 + }, + { + "epoch": 1.9568730930980456, + "grad_norm": 1.8757641315460205, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7548662424087524, + "num_tokens": 603312197.0, + "step": 23731 + }, + { + "epoch": 1.9569555537230974, + "grad_norm": 2.0219311714172363, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7553880214691162, + "num_tokens": 603340093.0, + "step": 23732 + }, + { + "epoch": 1.9570380143481487, + "grad_norm": 1.9852303266525269, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7718284726142883, + "num_tokens": 603370040.0, + "step": 23733 + }, + { + "epoch": 1.9571204749732003, + "grad_norm": 1.8643693923950195, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7734588384628296, + "num_tokens": 603403159.0, + "step": 23734 + }, + { + "epoch": 1.9572029355982519, + "grad_norm": 1.9962654113769531, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7475656867027283, + "num_tokens": 603433191.0, + "step": 23735 + }, + { + "epoch": 1.9572853962233032, + "grad_norm": 1.9872164726257324, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.7867975234985352, + "num_tokens": 603461814.0, + "step": 23736 + }, + { + "epoch": 1.957367856848355, + "grad_norm": 2.4026641845703125, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7619609832763672, + "num_tokens": 603483040.0, + "step": 23737 + }, + { + "epoch": 1.9574503174734064, + "grad_norm": 2.1823067665100098, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6991380453109741, + "num_tokens": 603513175.0, + "step": 23738 + }, + { + "epoch": 1.957532778098458, + "grad_norm": 2.3379478454589844, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7511675357818604, + "num_tokens": 603537093.0, + "step": 23739 + }, + { + "epoch": 1.9576152387235095, + "grad_norm": 2.014285087585449, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.748911440372467, + "num_tokens": 603567635.0, + "step": 23740 + }, + { + "epoch": 1.957697699348561, + "grad_norm": 2.2388956546783447, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7486530542373657, + "num_tokens": 603593619.0, + "step": 23741 + }, + { + "epoch": 1.9577801599736127, + "grad_norm": 1.8881549835205078, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7870604991912842, + "num_tokens": 603628435.0, + "step": 23742 + }, + { + "epoch": 1.957862620598664, + "grad_norm": 1.989587426185608, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.76369708776474, + "num_tokens": 603658657.0, + "step": 23743 + }, + { + "epoch": 1.9579450812237158, + "grad_norm": 1.927086353302002, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.7697802186012268, + "num_tokens": 603691817.0, + "step": 23744 + }, + { + "epoch": 1.9580275418487672, + "grad_norm": 1.98089599609375, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7758281826972961, + "num_tokens": 603724298.0, + "step": 23745 + }, + { + "epoch": 1.9581100024738187, + "grad_norm": 1.8739293813705444, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.7859508991241455, + "num_tokens": 603754714.0, + "step": 23746 + }, + { + "epoch": 1.9581924630988703, + "grad_norm": 2.2362048625946045, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7501893043518066, + "num_tokens": 603780836.0, + "step": 23747 + }, + { + "epoch": 1.9582749237239219, + "grad_norm": 2.0415711402893066, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.7810943126678467, + "num_tokens": 603808363.0, + "step": 23748 + }, + { + "epoch": 1.9583573843489734, + "grad_norm": 1.7309467792510986, + "learning_rate": 1e-06, + "loss": 0.5973, + "mean_token_accuracy": 0.7982089519500732, + "num_tokens": 603844182.0, + "step": 23749 + }, + { + "epoch": 1.9584398449740248, + "grad_norm": 1.9610743522644043, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.764433741569519, + "num_tokens": 603873794.0, + "step": 23750 + }, + { + "epoch": 1.9585223055990766, + "grad_norm": 1.893996238708496, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7377908229827881, + "num_tokens": 603906639.0, + "step": 23751 + }, + { + "epoch": 1.958604766224128, + "grad_norm": 2.0508804321289062, + "learning_rate": 1e-06, + "loss": 0.5991, + "mean_token_accuracy": 0.7991551160812378, + "num_tokens": 603933332.0, + "step": 23752 + }, + { + "epoch": 1.9586872268491795, + "grad_norm": 1.735729455947876, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7760734558105469, + "num_tokens": 603968853.0, + "step": 23753 + }, + { + "epoch": 1.958769687474231, + "grad_norm": 2.030682325363159, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7610126733779907, + "num_tokens": 603999500.0, + "step": 23754 + }, + { + "epoch": 1.9588521480992824, + "grad_norm": 2.0343096256256104, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7316011786460876, + "num_tokens": 604034163.0, + "step": 23755 + }, + { + "epoch": 1.9589346087243342, + "grad_norm": 1.9644914865493774, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7714580297470093, + "num_tokens": 604066796.0, + "step": 23756 + }, + { + "epoch": 1.9590170693493856, + "grad_norm": 1.8798069953918457, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7349984645843506, + "num_tokens": 604103855.0, + "step": 23757 + }, + { + "epoch": 1.9590995299744374, + "grad_norm": 2.0404536724090576, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7747154235839844, + "num_tokens": 604131968.0, + "step": 23758 + }, + { + "epoch": 1.9591819905994887, + "grad_norm": 1.895004391670227, + "learning_rate": 1e-06, + "loss": 0.5944, + "mean_token_accuracy": 0.8080379962921143, + "num_tokens": 604163291.0, + "step": 23759 + }, + { + "epoch": 1.9592644512245403, + "grad_norm": 1.898964285850525, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.755425214767456, + "num_tokens": 604196678.0, + "step": 23760 + }, + { + "epoch": 1.9593469118495919, + "grad_norm": 2.0164527893066406, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7602518200874329, + "num_tokens": 604226446.0, + "step": 23761 + }, + { + "epoch": 1.9594293724746432, + "grad_norm": 1.928957223892212, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7700742483139038, + "num_tokens": 604256812.0, + "step": 23762 + }, + { + "epoch": 1.959511833099695, + "grad_norm": 1.8630110025405884, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.792079508304596, + "num_tokens": 604286637.0, + "step": 23763 + }, + { + "epoch": 1.9595942937247464, + "grad_norm": 1.7323870658874512, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.7773697972297668, + "num_tokens": 604324678.0, + "step": 23764 + }, + { + "epoch": 1.959676754349798, + "grad_norm": 1.9146904945373535, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7654992938041687, + "num_tokens": 604356656.0, + "step": 23765 + }, + { + "epoch": 1.9597592149748495, + "grad_norm": 1.9144039154052734, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7496695518493652, + "num_tokens": 604391096.0, + "step": 23766 + }, + { + "epoch": 1.959841675599901, + "grad_norm": 1.8729755878448486, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7651602029800415, + "num_tokens": 604424800.0, + "step": 23767 + }, + { + "epoch": 1.9599241362249527, + "grad_norm": 1.8030483722686768, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7697745561599731, + "num_tokens": 604463600.0, + "step": 23768 + }, + { + "epoch": 1.960006596850004, + "grad_norm": 2.124847888946533, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7685645818710327, + "num_tokens": 604489262.0, + "step": 23769 + }, + { + "epoch": 1.9600890574750558, + "grad_norm": 1.816494345664978, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7641047239303589, + "num_tokens": 604523859.0, + "step": 23770 + }, + { + "epoch": 1.9601715181001071, + "grad_norm": 1.9656705856323242, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7571694254875183, + "num_tokens": 604555001.0, + "step": 23771 + }, + { + "epoch": 1.9602539787251587, + "grad_norm": 2.0550222396850586, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7692039012908936, + "num_tokens": 604580759.0, + "step": 23772 + }, + { + "epoch": 1.9603364393502103, + "grad_norm": 1.8101210594177246, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7724223136901855, + "num_tokens": 604615205.0, + "step": 23773 + }, + { + "epoch": 1.9604188999752619, + "grad_norm": 2.1038947105407715, + "learning_rate": 1e-06, + "loss": 0.7698, + "mean_token_accuracy": 0.7553694844245911, + "num_tokens": 604641883.0, + "step": 23774 + }, + { + "epoch": 1.9605013606003134, + "grad_norm": 2.1233301162719727, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.788143515586853, + "num_tokens": 604667970.0, + "step": 23775 + }, + { + "epoch": 1.9605838212253648, + "grad_norm": 2.105257272720337, + "learning_rate": 1e-06, + "loss": 0.7669, + "mean_token_accuracy": 0.7551219463348389, + "num_tokens": 604696924.0, + "step": 23776 + }, + { + "epoch": 1.9606662818504166, + "grad_norm": 2.2995612621307373, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7744691371917725, + "num_tokens": 604717790.0, + "step": 23777 + }, + { + "epoch": 1.960748742475468, + "grad_norm": 2.1643497943878174, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7494726777076721, + "num_tokens": 604745932.0, + "step": 23778 + }, + { + "epoch": 1.9608312031005195, + "grad_norm": 2.2878403663635254, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7082024812698364, + "num_tokens": 604771211.0, + "step": 23779 + }, + { + "epoch": 1.960913663725571, + "grad_norm": 2.146497964859009, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.7746762633323669, + "num_tokens": 604796331.0, + "step": 23780 + }, + { + "epoch": 1.9609961243506224, + "grad_norm": 2.221346378326416, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7751730680465698, + "num_tokens": 604818534.0, + "step": 23781 + }, + { + "epoch": 1.9610785849756742, + "grad_norm": 1.9590826034545898, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7835452556610107, + "num_tokens": 604848583.0, + "step": 23782 + }, + { + "epoch": 1.9611610456007256, + "grad_norm": 1.902993083000183, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7129356265068054, + "num_tokens": 604882183.0, + "step": 23783 + }, + { + "epoch": 1.9612435062257771, + "grad_norm": 2.1597156524658203, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.7902214527130127, + "num_tokens": 604907223.0, + "step": 23784 + }, + { + "epoch": 1.9613259668508287, + "grad_norm": 1.7574046850204468, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7836469411849976, + "num_tokens": 604946313.0, + "step": 23785 + }, + { + "epoch": 1.9614084274758803, + "grad_norm": 1.8561890125274658, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7425954341888428, + "num_tokens": 604978567.0, + "step": 23786 + }, + { + "epoch": 1.9614908881009319, + "grad_norm": 2.0474112033843994, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.7979876399040222, + "num_tokens": 605005455.0, + "step": 23787 + }, + { + "epoch": 1.9615733487259832, + "grad_norm": 2.1879382133483887, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7426654696464539, + "num_tokens": 605031403.0, + "step": 23788 + }, + { + "epoch": 1.961655809351035, + "grad_norm": 2.0101749897003174, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7823556661605835, + "num_tokens": 605058842.0, + "step": 23789 + }, + { + "epoch": 1.9617382699760864, + "grad_norm": 1.9065592288970947, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7652689218521118, + "num_tokens": 605091327.0, + "step": 23790 + }, + { + "epoch": 1.961820730601138, + "grad_norm": 2.051381826400757, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.764519453048706, + "num_tokens": 605121742.0, + "step": 23791 + }, + { + "epoch": 1.9619031912261895, + "grad_norm": 2.0618844032287598, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7426479458808899, + "num_tokens": 605152259.0, + "step": 23792 + }, + { + "epoch": 1.961985651851241, + "grad_norm": 1.8223106861114502, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7775567173957825, + "num_tokens": 605190667.0, + "step": 23793 + }, + { + "epoch": 1.9620681124762926, + "grad_norm": 2.1899502277374268, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7538323402404785, + "num_tokens": 605217535.0, + "step": 23794 + }, + { + "epoch": 1.962150573101344, + "grad_norm": 1.9554070234298706, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7567774057388306, + "num_tokens": 605248261.0, + "step": 23795 + }, + { + "epoch": 1.9622330337263958, + "grad_norm": 1.8158429861068726, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7429220676422119, + "num_tokens": 605284205.0, + "step": 23796 + }, + { + "epoch": 1.9623154943514471, + "grad_norm": 1.9988629817962646, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.7936771512031555, + "num_tokens": 605311140.0, + "step": 23797 + }, + { + "epoch": 1.9623979549764987, + "grad_norm": 2.0976476669311523, + "learning_rate": 1e-06, + "loss": 0.597, + "mean_token_accuracy": 0.7784321904182434, + "num_tokens": 605339728.0, + "step": 23798 + }, + { + "epoch": 1.9624804156015503, + "grad_norm": 1.707747220993042, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.8053262233734131, + "num_tokens": 605377698.0, + "step": 23799 + }, + { + "epoch": 1.9625628762266016, + "grad_norm": 2.1015281677246094, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.73866868019104, + "num_tokens": 605413709.0, + "step": 23800 + }, + { + "epoch": 1.9626453368516534, + "grad_norm": 2.553561210632324, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.769013524055481, + "num_tokens": 605445656.0, + "step": 23801 + }, + { + "epoch": 1.9627277974767048, + "grad_norm": 2.1181559562683105, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7659770846366882, + "num_tokens": 605469652.0, + "step": 23802 + }, + { + "epoch": 1.9628102581017566, + "grad_norm": 1.9483610391616821, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7793751955032349, + "num_tokens": 605502738.0, + "step": 23803 + }, + { + "epoch": 1.962892718726808, + "grad_norm": 2.0000767707824707, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7743980288505554, + "num_tokens": 605530778.0, + "step": 23804 + }, + { + "epoch": 1.9629751793518595, + "grad_norm": 2.0016090869903564, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7828476428985596, + "num_tokens": 605560357.0, + "step": 23805 + }, + { + "epoch": 1.963057639976911, + "grad_norm": 1.8033626079559326, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7424017190933228, + "num_tokens": 605594765.0, + "step": 23806 + }, + { + "epoch": 1.9631401006019624, + "grad_norm": 1.873678207397461, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7796365022659302, + "num_tokens": 605625914.0, + "step": 23807 + }, + { + "epoch": 1.9632225612270142, + "grad_norm": 2.0203704833984375, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7486696243286133, + "num_tokens": 605655731.0, + "step": 23808 + }, + { + "epoch": 1.9633050218520656, + "grad_norm": 1.7508891820907593, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7751678228378296, + "num_tokens": 605693895.0, + "step": 23809 + }, + { + "epoch": 1.9633874824771171, + "grad_norm": 1.9480159282684326, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7575851082801819, + "num_tokens": 605726265.0, + "step": 23810 + }, + { + "epoch": 1.9634699431021687, + "grad_norm": 2.0587289333343506, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.7401968240737915, + "num_tokens": 605755222.0, + "step": 23811 + }, + { + "epoch": 1.9635524037272203, + "grad_norm": 2.199589967727661, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7412391901016235, + "num_tokens": 605782864.0, + "step": 23812 + }, + { + "epoch": 1.9636348643522719, + "grad_norm": 1.9457248449325562, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7322306632995605, + "num_tokens": 605812746.0, + "step": 23813 + }, + { + "epoch": 1.9637173249773232, + "grad_norm": 2.1570963859558105, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7707810997962952, + "num_tokens": 605842134.0, + "step": 23814 + }, + { + "epoch": 1.963799785602375, + "grad_norm": 1.954559326171875, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.7822838425636292, + "num_tokens": 605872499.0, + "step": 23815 + }, + { + "epoch": 1.9638822462274264, + "grad_norm": 1.9325741529464722, + "learning_rate": 1e-06, + "loss": 0.543, + "mean_token_accuracy": 0.8169263005256653, + "num_tokens": 605901800.0, + "step": 23816 + }, + { + "epoch": 1.963964706852478, + "grad_norm": 2.0164592266082764, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.766305685043335, + "num_tokens": 605932910.0, + "step": 23817 + }, + { + "epoch": 1.9640471674775295, + "grad_norm": 1.8963528871536255, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7624698877334595, + "num_tokens": 605964645.0, + "step": 23818 + }, + { + "epoch": 1.964129628102581, + "grad_norm": 1.8823556900024414, + "learning_rate": 1e-06, + "loss": 0.571, + "mean_token_accuracy": 0.8178178071975708, + "num_tokens": 605993467.0, + "step": 23819 + }, + { + "epoch": 1.9642120887276326, + "grad_norm": 2.383168935775757, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7615969181060791, + "num_tokens": 606014813.0, + "step": 23820 + }, + { + "epoch": 1.964294549352684, + "grad_norm": 1.954979419708252, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7719205021858215, + "num_tokens": 606045467.0, + "step": 23821 + }, + { + "epoch": 1.9643770099777358, + "grad_norm": 2.2983384132385254, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7216308116912842, + "num_tokens": 606071423.0, + "step": 23822 + }, + { + "epoch": 1.9644594706027871, + "grad_norm": 2.4168214797973633, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7050285339355469, + "num_tokens": 606093304.0, + "step": 23823 + }, + { + "epoch": 1.9645419312278387, + "grad_norm": 1.9981882572174072, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7707493901252747, + "num_tokens": 606120511.0, + "step": 23824 + }, + { + "epoch": 1.9646243918528903, + "grad_norm": 1.90908944606781, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.738845705986023, + "num_tokens": 606155324.0, + "step": 23825 + }, + { + "epoch": 1.9647068524779416, + "grad_norm": 1.9441502094268799, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7683926224708557, + "num_tokens": 606188177.0, + "step": 23826 + }, + { + "epoch": 1.9647893131029934, + "grad_norm": 1.8123632669448853, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7588528990745544, + "num_tokens": 606223122.0, + "step": 23827 + }, + { + "epoch": 1.9648717737280448, + "grad_norm": 1.9509055614471436, + "learning_rate": 1e-06, + "loss": 0.5884, + "mean_token_accuracy": 0.7928425073623657, + "num_tokens": 606253111.0, + "step": 23828 + }, + { + "epoch": 1.9649542343530964, + "grad_norm": 1.8571152687072754, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.799918532371521, + "num_tokens": 606286516.0, + "step": 23829 + }, + { + "epoch": 1.965036694978148, + "grad_norm": 2.104529857635498, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7822360396385193, + "num_tokens": 606312628.0, + "step": 23830 + }, + { + "epoch": 1.9651191556031995, + "grad_norm": 2.160189151763916, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7563871741294861, + "num_tokens": 606338612.0, + "step": 23831 + }, + { + "epoch": 1.965201616228251, + "grad_norm": 2.019310474395752, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7692376375198364, + "num_tokens": 606368354.0, + "step": 23832 + }, + { + "epoch": 1.9652840768533024, + "grad_norm": 1.9063770771026611, + "learning_rate": 1e-06, + "loss": 0.6223, + "mean_token_accuracy": 0.7966849207878113, + "num_tokens": 606398867.0, + "step": 23833 + }, + { + "epoch": 1.9653665374783542, + "grad_norm": 2.521347761154175, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7613792419433594, + "num_tokens": 606420134.0, + "step": 23834 + }, + { + "epoch": 1.9654489981034056, + "grad_norm": 1.9089863300323486, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7656625509262085, + "num_tokens": 606449844.0, + "step": 23835 + }, + { + "epoch": 1.9655314587284571, + "grad_norm": 1.8467711210250854, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7828636169433594, + "num_tokens": 606484687.0, + "step": 23836 + }, + { + "epoch": 1.9656139193535087, + "grad_norm": 1.9375630617141724, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7432184219360352, + "num_tokens": 606515813.0, + "step": 23837 + }, + { + "epoch": 1.9656963799785603, + "grad_norm": 1.666039228439331, + "learning_rate": 1e-06, + "loss": 0.6075, + "mean_token_accuracy": 0.8010213375091553, + "num_tokens": 606550927.0, + "step": 23838 + }, + { + "epoch": 1.9657788406036119, + "grad_norm": 1.904889464378357, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.7793385982513428, + "num_tokens": 606581805.0, + "step": 23839 + }, + { + "epoch": 1.9658613012286632, + "grad_norm": 1.886310338973999, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7644683122634888, + "num_tokens": 606616307.0, + "step": 23840 + }, + { + "epoch": 1.965943761853715, + "grad_norm": 1.732993721961975, + "learning_rate": 1e-06, + "loss": 0.6097, + "mean_token_accuracy": 0.7924113869667053, + "num_tokens": 606653300.0, + "step": 23841 + }, + { + "epoch": 1.9660262224787663, + "grad_norm": 1.7292546033859253, + "learning_rate": 1e-06, + "loss": 0.5972, + "mean_token_accuracy": 0.7957656979560852, + "num_tokens": 606689208.0, + "step": 23842 + }, + { + "epoch": 1.966108683103818, + "grad_norm": 1.7189642190933228, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7779325246810913, + "num_tokens": 606725585.0, + "step": 23843 + }, + { + "epoch": 1.9661911437288695, + "grad_norm": 1.8289263248443604, + "learning_rate": 1e-06, + "loss": 0.7712, + "mean_token_accuracy": 0.7564675807952881, + "num_tokens": 606758833.0, + "step": 23844 + }, + { + "epoch": 1.9662736043539208, + "grad_norm": 1.8876023292541504, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7519009709358215, + "num_tokens": 606791010.0, + "step": 23845 + }, + { + "epoch": 1.9663560649789726, + "grad_norm": 2.056335926055908, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7198355793952942, + "num_tokens": 606820391.0, + "step": 23846 + }, + { + "epoch": 1.966438525604024, + "grad_norm": 2.110848903656006, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.7819598317146301, + "num_tokens": 606846673.0, + "step": 23847 + }, + { + "epoch": 1.9665209862290756, + "grad_norm": 1.803158164024353, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7749749422073364, + "num_tokens": 606879075.0, + "step": 23848 + }, + { + "epoch": 1.9666034468541271, + "grad_norm": 1.9109522104263306, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7601191997528076, + "num_tokens": 606910241.0, + "step": 23849 + }, + { + "epoch": 1.9666859074791787, + "grad_norm": 2.2956011295318604, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7411150336265564, + "num_tokens": 606935875.0, + "step": 23850 + }, + { + "epoch": 1.9667683681042303, + "grad_norm": 1.7987443208694458, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7647300958633423, + "num_tokens": 606968657.0, + "step": 23851 + }, + { + "epoch": 1.9668508287292816, + "grad_norm": 2.1125380992889404, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7708430290222168, + "num_tokens": 606995493.0, + "step": 23852 + }, + { + "epoch": 1.9669332893543334, + "grad_norm": 1.9779748916625977, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7822410464286804, + "num_tokens": 607024688.0, + "step": 23853 + }, + { + "epoch": 1.9670157499793848, + "grad_norm": 2.3449866771698, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7593772411346436, + "num_tokens": 607046697.0, + "step": 23854 + }, + { + "epoch": 1.9670982106044363, + "grad_norm": 2.111999750137329, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.7568328976631165, + "num_tokens": 607073940.0, + "step": 23855 + }, + { + "epoch": 1.967180671229488, + "grad_norm": 2.1945931911468506, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.7796470522880554, + "num_tokens": 607099110.0, + "step": 23856 + }, + { + "epoch": 1.9672631318545395, + "grad_norm": 2.093444347381592, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7460978031158447, + "num_tokens": 607130631.0, + "step": 23857 + }, + { + "epoch": 1.967345592479591, + "grad_norm": 1.9965652227401733, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7471007108688354, + "num_tokens": 607159352.0, + "step": 23858 + }, + { + "epoch": 1.9674280531046424, + "grad_norm": 1.912000298500061, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7706402540206909, + "num_tokens": 607191168.0, + "step": 23859 + }, + { + "epoch": 1.9675105137296942, + "grad_norm": 2.001251459121704, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7237142324447632, + "num_tokens": 607221696.0, + "step": 23860 + }, + { + "epoch": 1.9675929743547456, + "grad_norm": 2.0644280910491943, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7581776976585388, + "num_tokens": 607249685.0, + "step": 23861 + }, + { + "epoch": 1.9676754349797971, + "grad_norm": 2.276873826980591, + "learning_rate": 1e-06, + "loss": 0.7877, + "mean_token_accuracy": 0.7532712817192078, + "num_tokens": 607272348.0, + "step": 23862 + }, + { + "epoch": 1.9677578956048487, + "grad_norm": 1.9218590259552002, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.761622428894043, + "num_tokens": 607303687.0, + "step": 23863 + }, + { + "epoch": 1.9678403562299, + "grad_norm": 1.6998947858810425, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.805098295211792, + "num_tokens": 607341723.0, + "step": 23864 + }, + { + "epoch": 1.9679228168549519, + "grad_norm": 2.057243585586548, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.75627601146698, + "num_tokens": 607373652.0, + "step": 23865 + }, + { + "epoch": 1.9680052774800032, + "grad_norm": 2.1296796798706055, + "learning_rate": 1e-06, + "loss": 0.5607, + "mean_token_accuracy": 0.812800407409668, + "num_tokens": 607398351.0, + "step": 23866 + }, + { + "epoch": 1.968087738105055, + "grad_norm": 1.9945927858352661, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7680907249450684, + "num_tokens": 607428550.0, + "step": 23867 + }, + { + "epoch": 1.9681701987301063, + "grad_norm": 1.782888650894165, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7853343486785889, + "num_tokens": 607466287.0, + "step": 23868 + }, + { + "epoch": 1.968252659355158, + "grad_norm": 1.8995565176010132, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.760833203792572, + "num_tokens": 607498794.0, + "step": 23869 + }, + { + "epoch": 1.9683351199802095, + "grad_norm": 2.1596031188964844, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7406591176986694, + "num_tokens": 607524842.0, + "step": 23870 + }, + { + "epoch": 1.9684175806052608, + "grad_norm": 1.7833664417266846, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7837569713592529, + "num_tokens": 607559970.0, + "step": 23871 + }, + { + "epoch": 1.9685000412303126, + "grad_norm": 1.9395312070846558, + "learning_rate": 1e-06, + "loss": 0.6056, + "mean_token_accuracy": 0.7966964244842529, + "num_tokens": 607588273.0, + "step": 23872 + }, + { + "epoch": 1.968582501855364, + "grad_norm": 1.7591997385025024, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7816513180732727, + "num_tokens": 607621826.0, + "step": 23873 + }, + { + "epoch": 1.9686649624804156, + "grad_norm": 2.262031078338623, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.7922288179397583, + "num_tokens": 607646971.0, + "step": 23874 + }, + { + "epoch": 1.9687474231054671, + "grad_norm": 1.9124101400375366, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7310566902160645, + "num_tokens": 607679008.0, + "step": 23875 + }, + { + "epoch": 1.9688298837305187, + "grad_norm": 2.059718370437622, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.764494776725769, + "num_tokens": 607707514.0, + "step": 23876 + }, + { + "epoch": 1.9689123443555703, + "grad_norm": 1.8031272888183594, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7610853314399719, + "num_tokens": 607740669.0, + "step": 23877 + }, + { + "epoch": 1.9689948049806216, + "grad_norm": 1.8736916780471802, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.7828296422958374, + "num_tokens": 607770895.0, + "step": 23878 + }, + { + "epoch": 1.9690772656056734, + "grad_norm": 1.9173355102539062, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7599836587905884, + "num_tokens": 607800797.0, + "step": 23879 + }, + { + "epoch": 1.9691597262307248, + "grad_norm": 2.144052505493164, + "learning_rate": 1e-06, + "loss": 0.7767, + "mean_token_accuracy": 0.7527758479118347, + "num_tokens": 607827295.0, + "step": 23880 + }, + { + "epoch": 1.9692421868557763, + "grad_norm": 1.81473708152771, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7604854106903076, + "num_tokens": 607862720.0, + "step": 23881 + }, + { + "epoch": 1.969324647480828, + "grad_norm": 2.2746472358703613, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.7718256711959839, + "num_tokens": 607886897.0, + "step": 23882 + }, + { + "epoch": 1.9694071081058795, + "grad_norm": 2.193688154220581, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.7845445871353149, + "num_tokens": 607913334.0, + "step": 23883 + }, + { + "epoch": 1.969489568730931, + "grad_norm": 1.703301191329956, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7811850309371948, + "num_tokens": 607949278.0, + "step": 23884 + }, + { + "epoch": 1.9695720293559824, + "grad_norm": 2.1272456645965576, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7472865581512451, + "num_tokens": 607978201.0, + "step": 23885 + }, + { + "epoch": 1.9696544899810342, + "grad_norm": 2.240624189376831, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7480055093765259, + "num_tokens": 608005117.0, + "step": 23886 + }, + { + "epoch": 1.9697369506060856, + "grad_norm": 2.1273601055145264, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7355884313583374, + "num_tokens": 608034599.0, + "step": 23887 + }, + { + "epoch": 1.9698194112311371, + "grad_norm": 2.0505130290985107, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7530169486999512, + "num_tokens": 608065060.0, + "step": 23888 + }, + { + "epoch": 1.9699018718561887, + "grad_norm": 1.8603556156158447, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7331160306930542, + "num_tokens": 608097750.0, + "step": 23889 + }, + { + "epoch": 1.96998433248124, + "grad_norm": 1.8396995067596436, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.772142231464386, + "num_tokens": 608131232.0, + "step": 23890 + }, + { + "epoch": 1.9700667931062918, + "grad_norm": 2.155033826828003, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7588024735450745, + "num_tokens": 608158463.0, + "step": 23891 + }, + { + "epoch": 1.9701492537313432, + "grad_norm": 2.1812384128570557, + "learning_rate": 1e-06, + "loss": 0.5924, + "mean_token_accuracy": 0.7942005395889282, + "num_tokens": 608182037.0, + "step": 23892 + }, + { + "epoch": 1.9702317143563948, + "grad_norm": 1.9042161703109741, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7386749982833862, + "num_tokens": 608215725.0, + "step": 23893 + }, + { + "epoch": 1.9703141749814463, + "grad_norm": 1.9535584449768066, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.7853419780731201, + "num_tokens": 608246249.0, + "step": 23894 + }, + { + "epoch": 1.970396635606498, + "grad_norm": 1.6894406080245972, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.785387396812439, + "num_tokens": 608282399.0, + "step": 23895 + }, + { + "epoch": 1.9704790962315495, + "grad_norm": 1.8955183029174805, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7643895149230957, + "num_tokens": 608313391.0, + "step": 23896 + }, + { + "epoch": 1.9705615568566008, + "grad_norm": 1.753635048866272, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7844996452331543, + "num_tokens": 608351062.0, + "step": 23897 + }, + { + "epoch": 1.9706440174816526, + "grad_norm": 2.2957253456115723, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.78471839427948, + "num_tokens": 608376609.0, + "step": 23898 + }, + { + "epoch": 1.970726478106704, + "grad_norm": 1.990006446838379, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7678044438362122, + "num_tokens": 608407056.0, + "step": 23899 + }, + { + "epoch": 1.9708089387317556, + "grad_norm": 2.0018460750579834, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7664349675178528, + "num_tokens": 608437322.0, + "step": 23900 + }, + { + "epoch": 1.9708913993568071, + "grad_norm": 1.8591572046279907, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.7900052070617676, + "num_tokens": 608468922.0, + "step": 23901 + }, + { + "epoch": 1.9709738599818587, + "grad_norm": 1.925222396850586, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7234674692153931, + "num_tokens": 608502443.0, + "step": 23902 + }, + { + "epoch": 1.9710563206069103, + "grad_norm": 2.235960006713867, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7669671177864075, + "num_tokens": 608525375.0, + "step": 23903 + }, + { + "epoch": 1.9711387812319616, + "grad_norm": 1.6329314708709717, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7588922381401062, + "num_tokens": 608564822.0, + "step": 23904 + }, + { + "epoch": 1.9712212418570134, + "grad_norm": 1.8825029134750366, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7861595153808594, + "num_tokens": 608594427.0, + "step": 23905 + }, + { + "epoch": 1.9713037024820648, + "grad_norm": 1.9890165328979492, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.7857005596160889, + "num_tokens": 608622901.0, + "step": 23906 + }, + { + "epoch": 1.9713861631071163, + "grad_norm": 2.3100807666778564, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7648622989654541, + "num_tokens": 608645316.0, + "step": 23907 + }, + { + "epoch": 1.971468623732168, + "grad_norm": 1.6725105047225952, + "learning_rate": 1e-06, + "loss": 0.5603, + "mean_token_accuracy": 0.8027379512786865, + "num_tokens": 608679879.0, + "step": 23908 + }, + { + "epoch": 1.9715510843572193, + "grad_norm": 2.3681411743164062, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.6913257241249084, + "num_tokens": 608704504.0, + "step": 23909 + }, + { + "epoch": 1.971633544982271, + "grad_norm": 1.9325554370880127, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7735140323638916, + "num_tokens": 608732767.0, + "step": 23910 + }, + { + "epoch": 1.9717160056073224, + "grad_norm": 1.9907516241073608, + "learning_rate": 1e-06, + "loss": 0.5937, + "mean_token_accuracy": 0.8015247583389282, + "num_tokens": 608759587.0, + "step": 23911 + }, + { + "epoch": 1.9717984662323742, + "grad_norm": 2.04917311668396, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7674193382263184, + "num_tokens": 608787931.0, + "step": 23912 + }, + { + "epoch": 1.9718809268574256, + "grad_norm": 1.8339933156967163, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7452026605606079, + "num_tokens": 608825312.0, + "step": 23913 + }, + { + "epoch": 1.9719633874824771, + "grad_norm": 1.7440974712371826, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.7933474779129028, + "num_tokens": 608858230.0, + "step": 23914 + }, + { + "epoch": 1.9720458481075287, + "grad_norm": 2.2091188430786133, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7527681589126587, + "num_tokens": 608884830.0, + "step": 23915 + }, + { + "epoch": 1.97212830873258, + "grad_norm": 2.115619421005249, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.74140465259552, + "num_tokens": 608910155.0, + "step": 23916 + }, + { + "epoch": 1.9722107693576318, + "grad_norm": 1.9214229583740234, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7476100921630859, + "num_tokens": 608941689.0, + "step": 23917 + }, + { + "epoch": 1.9722932299826832, + "grad_norm": 1.9280166625976562, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7564108967781067, + "num_tokens": 608974988.0, + "step": 23918 + }, + { + "epoch": 1.9723756906077348, + "grad_norm": 2.022632598876953, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7662751078605652, + "num_tokens": 609006375.0, + "step": 23919 + }, + { + "epoch": 1.9724581512327863, + "grad_norm": 1.7923060655593872, + "learning_rate": 1e-06, + "loss": 0.5736, + "mean_token_accuracy": 0.7940313816070557, + "num_tokens": 609038784.0, + "step": 23920 + }, + { + "epoch": 1.972540611857838, + "grad_norm": 2.0230274200439453, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7375199794769287, + "num_tokens": 609066727.0, + "step": 23921 + }, + { + "epoch": 1.9726230724828895, + "grad_norm": 1.6501611471176147, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7599645853042603, + "num_tokens": 609107479.0, + "step": 23922 + }, + { + "epoch": 1.9727055331079408, + "grad_norm": 1.9204264879226685, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7328485250473022, + "num_tokens": 609139454.0, + "step": 23923 + }, + { + "epoch": 1.9727879937329926, + "grad_norm": 1.9822473526000977, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7532569169998169, + "num_tokens": 609172795.0, + "step": 23924 + }, + { + "epoch": 1.972870454358044, + "grad_norm": 1.8334749937057495, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.7772616147994995, + "num_tokens": 609205404.0, + "step": 23925 + }, + { + "epoch": 1.9729529149830956, + "grad_norm": 1.9185322523117065, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7472009658813477, + "num_tokens": 609236784.0, + "step": 23926 + }, + { + "epoch": 1.9730353756081471, + "grad_norm": 1.7270880937576294, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.7834252119064331, + "num_tokens": 609272773.0, + "step": 23927 + }, + { + "epoch": 1.9731178362331987, + "grad_norm": 2.1324210166931152, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7814989686012268, + "num_tokens": 609296881.0, + "step": 23928 + }, + { + "epoch": 1.9732002968582503, + "grad_norm": 1.9554600715637207, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7702769041061401, + "num_tokens": 609327258.0, + "step": 23929 + }, + { + "epoch": 1.9732827574833016, + "grad_norm": 2.468318462371826, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7446775436401367, + "num_tokens": 609348051.0, + "step": 23930 + }, + { + "epoch": 1.9733652181083534, + "grad_norm": 2.1582841873168945, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7371526956558228, + "num_tokens": 609374969.0, + "step": 23931 + }, + { + "epoch": 1.9734476787334048, + "grad_norm": 2.033114433288574, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.7473199963569641, + "num_tokens": 609403787.0, + "step": 23932 + }, + { + "epoch": 1.9735301393584563, + "grad_norm": 1.9426392316818237, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7239634394645691, + "num_tokens": 609433598.0, + "step": 23933 + }, + { + "epoch": 1.973612599983508, + "grad_norm": 1.8059614896774292, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7371710538864136, + "num_tokens": 609472623.0, + "step": 23934 + }, + { + "epoch": 1.9736950606085593, + "grad_norm": 2.0847713947296143, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.7880549430847168, + "num_tokens": 609497333.0, + "step": 23935 + }, + { + "epoch": 1.973777521233611, + "grad_norm": 1.9378983974456787, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.770165205001831, + "num_tokens": 609524869.0, + "step": 23936 + }, + { + "epoch": 1.9738599818586624, + "grad_norm": 1.782449722290039, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7502423524856567, + "num_tokens": 609560480.0, + "step": 23937 + }, + { + "epoch": 1.973942442483714, + "grad_norm": 2.0242881774902344, + "learning_rate": 1e-06, + "loss": 0.6192, + "mean_token_accuracy": 0.7876251935958862, + "num_tokens": 609587746.0, + "step": 23938 + }, + { + "epoch": 1.9740249031087655, + "grad_norm": 1.8944300413131714, + "learning_rate": 1e-06, + "loss": 0.5675, + "mean_token_accuracy": 0.8097907304763794, + "num_tokens": 609616505.0, + "step": 23939 + }, + { + "epoch": 1.9741073637338171, + "grad_norm": 1.9750680923461914, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.742844820022583, + "num_tokens": 609646235.0, + "step": 23940 + }, + { + "epoch": 1.9741898243588687, + "grad_norm": 2.0000293254852295, + "learning_rate": 1e-06, + "loss": 0.8122, + "mean_token_accuracy": 0.7454419136047363, + "num_tokens": 609675933.0, + "step": 23941 + }, + { + "epoch": 1.97427228498392, + "grad_norm": 1.7675771713256836, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.777388870716095, + "num_tokens": 609713173.0, + "step": 23942 + }, + { + "epoch": 1.9743547456089718, + "grad_norm": 2.0185952186584473, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7774229049682617, + "num_tokens": 609742385.0, + "step": 23943 + }, + { + "epoch": 1.9744372062340232, + "grad_norm": 2.311650276184082, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7621579170227051, + "num_tokens": 609767994.0, + "step": 23944 + }, + { + "epoch": 1.9745196668590748, + "grad_norm": 2.097156286239624, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.7937169671058655, + "num_tokens": 609792552.0, + "step": 23945 + }, + { + "epoch": 1.9746021274841263, + "grad_norm": 2.3672239780426025, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.7571202516555786, + "num_tokens": 609814189.0, + "step": 23946 + }, + { + "epoch": 1.974684588109178, + "grad_norm": 2.034794330596924, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7816370725631714, + "num_tokens": 609838739.0, + "step": 23947 + }, + { + "epoch": 1.9747670487342295, + "grad_norm": 1.9766501188278198, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7625492811203003, + "num_tokens": 609870135.0, + "step": 23948 + }, + { + "epoch": 1.9748495093592808, + "grad_norm": 2.0593459606170654, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7904457449913025, + "num_tokens": 609898697.0, + "step": 23949 + }, + { + "epoch": 1.9749319699843326, + "grad_norm": 2.112884283065796, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7810201644897461, + "num_tokens": 609924508.0, + "step": 23950 + }, + { + "epoch": 1.975014430609384, + "grad_norm": 2.0780351161956787, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7459675073623657, + "num_tokens": 609955846.0, + "step": 23951 + }, + { + "epoch": 1.9750968912344355, + "grad_norm": 1.8808914422988892, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7827528715133667, + "num_tokens": 609987154.0, + "step": 23952 + }, + { + "epoch": 1.9751793518594871, + "grad_norm": 2.065382719039917, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.691234290599823, + "num_tokens": 610020095.0, + "step": 23953 + }, + { + "epoch": 1.9752618124845385, + "grad_norm": 2.3679463863372803, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7320403456687927, + "num_tokens": 610043212.0, + "step": 23954 + }, + { + "epoch": 1.9753442731095903, + "grad_norm": 1.916858196258545, + "learning_rate": 1e-06, + "loss": 0.6047, + "mean_token_accuracy": 0.7967686653137207, + "num_tokens": 610073184.0, + "step": 23955 + }, + { + "epoch": 1.9754267337346416, + "grad_norm": 2.019777774810791, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.7843351364135742, + "num_tokens": 610104915.0, + "step": 23956 + }, + { + "epoch": 1.9755091943596934, + "grad_norm": 1.9042247533798218, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7553659677505493, + "num_tokens": 610136526.0, + "step": 23957 + }, + { + "epoch": 1.9755916549847448, + "grad_norm": 2.0435032844543457, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7525817155838013, + "num_tokens": 610166929.0, + "step": 23958 + }, + { + "epoch": 1.9756741156097963, + "grad_norm": 1.8321515321731567, + "learning_rate": 1e-06, + "loss": 0.6127, + "mean_token_accuracy": 0.7860094904899597, + "num_tokens": 610198152.0, + "step": 23959 + }, + { + "epoch": 1.975756576234848, + "grad_norm": 1.7832854986190796, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.7912600040435791, + "num_tokens": 610231211.0, + "step": 23960 + }, + { + "epoch": 1.9758390368598993, + "grad_norm": 1.853759527206421, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7436805963516235, + "num_tokens": 610264214.0, + "step": 23961 + }, + { + "epoch": 1.975921497484951, + "grad_norm": 1.9159091711044312, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7667298913002014, + "num_tokens": 610297207.0, + "step": 23962 + }, + { + "epoch": 1.9760039581100024, + "grad_norm": 1.899151086807251, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7344213724136353, + "num_tokens": 610330023.0, + "step": 23963 + }, + { + "epoch": 1.976086418735054, + "grad_norm": 2.0902748107910156, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7580091953277588, + "num_tokens": 610360357.0, + "step": 23964 + }, + { + "epoch": 1.9761688793601055, + "grad_norm": 2.006969928741455, + "learning_rate": 1e-06, + "loss": 0.5536, + "mean_token_accuracy": 0.8141377568244934, + "num_tokens": 610387221.0, + "step": 23965 + }, + { + "epoch": 1.9762513399851571, + "grad_norm": 2.0161490440368652, + "learning_rate": 1e-06, + "loss": 0.5604, + "mean_token_accuracy": 0.818163275718689, + "num_tokens": 610411919.0, + "step": 23966 + }, + { + "epoch": 1.9763338006102087, + "grad_norm": 1.8897228240966797, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7643649578094482, + "num_tokens": 610444459.0, + "step": 23967 + }, + { + "epoch": 1.97641626123526, + "grad_norm": 1.945940375328064, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7648165225982666, + "num_tokens": 610476722.0, + "step": 23968 + }, + { + "epoch": 1.9764987218603118, + "grad_norm": 1.9781978130340576, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7365989685058594, + "num_tokens": 610507655.0, + "step": 23969 + }, + { + "epoch": 1.9765811824853632, + "grad_norm": 1.9189363718032837, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.707095205783844, + "num_tokens": 610540091.0, + "step": 23970 + }, + { + "epoch": 1.9766636431104148, + "grad_norm": 2.056211233139038, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.7545260190963745, + "num_tokens": 610569056.0, + "step": 23971 + }, + { + "epoch": 1.9767461037354663, + "grad_norm": 1.892704963684082, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7582035064697266, + "num_tokens": 610603188.0, + "step": 23972 + }, + { + "epoch": 1.976828564360518, + "grad_norm": 2.2197742462158203, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.7797658443450928, + "num_tokens": 610626751.0, + "step": 23973 + }, + { + "epoch": 1.9769110249855695, + "grad_norm": 1.802384614944458, + "learning_rate": 1e-06, + "loss": 0.5841, + "mean_token_accuracy": 0.8100675344467163, + "num_tokens": 610658162.0, + "step": 23974 + }, + { + "epoch": 1.9769934856106208, + "grad_norm": 2.106403112411499, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7720003724098206, + "num_tokens": 610683890.0, + "step": 23975 + }, + { + "epoch": 1.9770759462356726, + "grad_norm": 1.9652254581451416, + "learning_rate": 1e-06, + "loss": 0.5983, + "mean_token_accuracy": 0.791027307510376, + "num_tokens": 610709845.0, + "step": 23976 + }, + { + "epoch": 1.977158406860724, + "grad_norm": 2.199112892150879, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.71098792552948, + "num_tokens": 610735974.0, + "step": 23977 + }, + { + "epoch": 1.9772408674857755, + "grad_norm": 2.0642874240875244, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7616456747055054, + "num_tokens": 610763136.0, + "step": 23978 + }, + { + "epoch": 1.9773233281108271, + "grad_norm": 2.1679725646972656, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.75096595287323, + "num_tokens": 610788130.0, + "step": 23979 + }, + { + "epoch": 1.9774057887358785, + "grad_norm": 1.7492188215255737, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7747567892074585, + "num_tokens": 610823340.0, + "step": 23980 + }, + { + "epoch": 1.9774882493609303, + "grad_norm": 1.9301769733428955, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.7892891764640808, + "num_tokens": 610850433.0, + "step": 23981 + }, + { + "epoch": 1.9775707099859816, + "grad_norm": 1.9146008491516113, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7455745935440063, + "num_tokens": 610880095.0, + "step": 23982 + }, + { + "epoch": 1.9776531706110332, + "grad_norm": 2.311490774154663, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7666933536529541, + "num_tokens": 610902867.0, + "step": 23983 + }, + { + "epoch": 1.9777356312360848, + "grad_norm": 1.9830578565597534, + "learning_rate": 1e-06, + "loss": 0.5483, + "mean_token_accuracy": 0.8141331672668457, + "num_tokens": 610928868.0, + "step": 23984 + }, + { + "epoch": 1.9778180918611363, + "grad_norm": 2.1387217044830322, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7522292733192444, + "num_tokens": 610956007.0, + "step": 23985 + }, + { + "epoch": 1.977900552486188, + "grad_norm": 2.4244444370269775, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7480863928794861, + "num_tokens": 610977637.0, + "step": 23986 + }, + { + "epoch": 1.9779830131112393, + "grad_norm": 2.1402008533477783, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7566993832588196, + "num_tokens": 611002794.0, + "step": 23987 + }, + { + "epoch": 1.978065473736291, + "grad_norm": 2.024038076400757, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.7936311960220337, + "num_tokens": 611031191.0, + "step": 23988 + }, + { + "epoch": 1.9781479343613424, + "grad_norm": 2.0599899291992188, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.746727705001831, + "num_tokens": 611062960.0, + "step": 23989 + }, + { + "epoch": 1.978230394986394, + "grad_norm": 2.116887092590332, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7454575300216675, + "num_tokens": 611091772.0, + "step": 23990 + }, + { + "epoch": 1.9783128556114455, + "grad_norm": 2.223318099975586, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7112980484962463, + "num_tokens": 611118164.0, + "step": 23991 + }, + { + "epoch": 1.9783953162364971, + "grad_norm": 1.83575439453125, + "learning_rate": 1e-06, + "loss": 0.59, + "mean_token_accuracy": 0.809852123260498, + "num_tokens": 611149292.0, + "step": 23992 + }, + { + "epoch": 1.9784777768615487, + "grad_norm": 2.0544867515563965, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7672169208526611, + "num_tokens": 611176652.0, + "step": 23993 + }, + { + "epoch": 1.9785602374866, + "grad_norm": 2.1171252727508545, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7692708969116211, + "num_tokens": 611202836.0, + "step": 23994 + }, + { + "epoch": 1.9786426981116518, + "grad_norm": 1.8696869611740112, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7809415459632874, + "num_tokens": 611233216.0, + "step": 23995 + }, + { + "epoch": 1.9787251587367032, + "grad_norm": 2.3823983669281006, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7601613402366638, + "num_tokens": 611253488.0, + "step": 23996 + }, + { + "epoch": 1.9788076193617548, + "grad_norm": 2.158224582672119, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7344629168510437, + "num_tokens": 611281369.0, + "step": 23997 + }, + { + "epoch": 1.9788900799868063, + "grad_norm": 1.836699366569519, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7578809261322021, + "num_tokens": 611317166.0, + "step": 23998 + }, + { + "epoch": 1.9789725406118577, + "grad_norm": 1.7420716285705566, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7772054672241211, + "num_tokens": 611352687.0, + "step": 23999 + }, + { + "epoch": 1.9790550012369095, + "grad_norm": 1.843240737915039, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.7943322658538818, + "num_tokens": 611384267.0, + "step": 24000 + }, + { + "epoch": 1.9791374618619608, + "grad_norm": 1.8366953134536743, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.7856756448745728, + "num_tokens": 611420622.0, + "step": 24001 + }, + { + "epoch": 1.9792199224870126, + "grad_norm": 1.7113513946533203, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7428661584854126, + "num_tokens": 611461061.0, + "step": 24002 + }, + { + "epoch": 1.979302383112064, + "grad_norm": 1.9327974319458008, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.780700147151947, + "num_tokens": 611490711.0, + "step": 24003 + }, + { + "epoch": 1.9793848437371155, + "grad_norm": 1.9820256233215332, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.7823560237884521, + "num_tokens": 611519589.0, + "step": 24004 + }, + { + "epoch": 1.9794673043621671, + "grad_norm": 1.839676022529602, + "learning_rate": 1e-06, + "loss": 0.5838, + "mean_token_accuracy": 0.8062530755996704, + "num_tokens": 611549535.0, + "step": 24005 + }, + { + "epoch": 1.9795497649872185, + "grad_norm": 1.946314811706543, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7780604362487793, + "num_tokens": 611580794.0, + "step": 24006 + }, + { + "epoch": 1.9796322256122703, + "grad_norm": 1.920820713043213, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7916840314865112, + "num_tokens": 611612075.0, + "step": 24007 + }, + { + "epoch": 1.9797146862373216, + "grad_norm": 1.9632095098495483, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7685989141464233, + "num_tokens": 611643125.0, + "step": 24008 + }, + { + "epoch": 1.9797971468623732, + "grad_norm": 2.0935730934143066, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7762054204940796, + "num_tokens": 611669702.0, + "step": 24009 + }, + { + "epoch": 1.9798796074874248, + "grad_norm": 1.9644490480422974, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7304351329803467, + "num_tokens": 611700196.0, + "step": 24010 + }, + { + "epoch": 1.9799620681124763, + "grad_norm": 1.7695201635360718, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7695846557617188, + "num_tokens": 611733817.0, + "step": 24011 + }, + { + "epoch": 1.980044528737528, + "grad_norm": 1.8480974435806274, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.7953605651855469, + "num_tokens": 611763974.0, + "step": 24012 + }, + { + "epoch": 1.9801269893625792, + "grad_norm": 1.8873878717422485, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7441132664680481, + "num_tokens": 611799357.0, + "step": 24013 + }, + { + "epoch": 1.980209449987631, + "grad_norm": 2.4816277027130127, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7716262340545654, + "num_tokens": 611820108.0, + "step": 24014 + }, + { + "epoch": 1.9802919106126824, + "grad_norm": 1.9117236137390137, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7767048478126526, + "num_tokens": 611853495.0, + "step": 24015 + }, + { + "epoch": 1.980374371237734, + "grad_norm": 1.9688973426818848, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7594505548477173, + "num_tokens": 611881801.0, + "step": 24016 + }, + { + "epoch": 1.9804568318627855, + "grad_norm": 1.8289109468460083, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.7835224866867065, + "num_tokens": 611917798.0, + "step": 24017 + }, + { + "epoch": 1.980539292487837, + "grad_norm": 1.9994467496871948, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7643582820892334, + "num_tokens": 611948947.0, + "step": 24018 + }, + { + "epoch": 1.9806217531128887, + "grad_norm": 2.1977107524871826, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7316120862960815, + "num_tokens": 611976793.0, + "step": 24019 + }, + { + "epoch": 1.98070421373794, + "grad_norm": 2.254413366317749, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7500624656677246, + "num_tokens": 612004112.0, + "step": 24020 + }, + { + "epoch": 1.9807866743629918, + "grad_norm": 2.073936939239502, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7579374313354492, + "num_tokens": 612031650.0, + "step": 24021 + }, + { + "epoch": 1.9808691349880432, + "grad_norm": 2.041835308074951, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.7443767786026001, + "num_tokens": 612061078.0, + "step": 24022 + }, + { + "epoch": 1.9809515956130948, + "grad_norm": 1.841460108757019, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7726027965545654, + "num_tokens": 612095999.0, + "step": 24023 + }, + { + "epoch": 1.9810340562381463, + "grad_norm": 2.039044141769409, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7358250617980957, + "num_tokens": 612123879.0, + "step": 24024 + }, + { + "epoch": 1.9811165168631977, + "grad_norm": 1.9519925117492676, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7629232406616211, + "num_tokens": 612154243.0, + "step": 24025 + }, + { + "epoch": 1.9811989774882495, + "grad_norm": 2.0882532596588135, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.738241970539093, + "num_tokens": 612183636.0, + "step": 24026 + }, + { + "epoch": 1.9812814381133008, + "grad_norm": 1.9774079322814941, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7768328189849854, + "num_tokens": 612214575.0, + "step": 24027 + }, + { + "epoch": 1.9813638987383524, + "grad_norm": 1.8706326484680176, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7332452535629272, + "num_tokens": 612251427.0, + "step": 24028 + }, + { + "epoch": 1.981446359363404, + "grad_norm": 1.740627646446228, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7837545871734619, + "num_tokens": 612287190.0, + "step": 24029 + }, + { + "epoch": 1.9815288199884555, + "grad_norm": 1.7136681079864502, + "learning_rate": 1e-06, + "loss": 0.5345, + "mean_token_accuracy": 0.8211909532546997, + "num_tokens": 612326637.0, + "step": 24030 + }, + { + "epoch": 1.981611280613507, + "grad_norm": 2.086550235748291, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7550379037857056, + "num_tokens": 612354030.0, + "step": 24031 + }, + { + "epoch": 1.9816937412385585, + "grad_norm": 2.1423027515411377, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7135475873947144, + "num_tokens": 612381401.0, + "step": 24032 + }, + { + "epoch": 1.9817762018636103, + "grad_norm": 2.2553274631500244, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.769278347492218, + "num_tokens": 612405795.0, + "step": 24033 + }, + { + "epoch": 1.9818586624886616, + "grad_norm": 1.8456861972808838, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.7940776348114014, + "num_tokens": 612437898.0, + "step": 24034 + }, + { + "epoch": 1.9819411231137132, + "grad_norm": 2.2040810585021973, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.6973212361335754, + "num_tokens": 612464507.0, + "step": 24035 + }, + { + "epoch": 1.9820235837387647, + "grad_norm": 2.1513285636901855, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7081905007362366, + "num_tokens": 612493637.0, + "step": 24036 + }, + { + "epoch": 1.9821060443638163, + "grad_norm": 1.8358824253082275, + "learning_rate": 1e-06, + "loss": 0.5826, + "mean_token_accuracy": 0.8084696531295776, + "num_tokens": 612526940.0, + "step": 24037 + }, + { + "epoch": 1.982188504988868, + "grad_norm": 1.9962412118911743, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7700808048248291, + "num_tokens": 612556051.0, + "step": 24038 + }, + { + "epoch": 1.9822709656139192, + "grad_norm": 1.997214913368225, + "learning_rate": 1e-06, + "loss": 0.8238, + "mean_token_accuracy": 0.7319805026054382, + "num_tokens": 612588513.0, + "step": 24039 + }, + { + "epoch": 1.982353426238971, + "grad_norm": 1.9887479543685913, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7611260414123535, + "num_tokens": 612617271.0, + "step": 24040 + }, + { + "epoch": 1.9824358868640224, + "grad_norm": 2.2006428241729736, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.745118260383606, + "num_tokens": 612644069.0, + "step": 24041 + }, + { + "epoch": 1.982518347489074, + "grad_norm": 2.1393516063690186, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7659051418304443, + "num_tokens": 612671705.0, + "step": 24042 + }, + { + "epoch": 1.9826008081141255, + "grad_norm": 2.037470817565918, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.7839892506599426, + "num_tokens": 612698659.0, + "step": 24043 + }, + { + "epoch": 1.9826832687391769, + "grad_norm": 2.1341254711151123, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7296355366706848, + "num_tokens": 612726248.0, + "step": 24044 + }, + { + "epoch": 1.9827657293642287, + "grad_norm": 2.218440055847168, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7598123550415039, + "num_tokens": 612751156.0, + "step": 24045 + }, + { + "epoch": 1.98284818998928, + "grad_norm": 1.7013050317764282, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7732212543487549, + "num_tokens": 612786435.0, + "step": 24046 + }, + { + "epoch": 1.9829306506143316, + "grad_norm": 1.945030927658081, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7785302400588989, + "num_tokens": 612815836.0, + "step": 24047 + }, + { + "epoch": 1.9830131112393832, + "grad_norm": 1.8571547269821167, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7689825296401978, + "num_tokens": 612847874.0, + "step": 24048 + }, + { + "epoch": 1.9830955718644347, + "grad_norm": 2.136604070663452, + "learning_rate": 1e-06, + "loss": 0.7868, + "mean_token_accuracy": 0.7469489574432373, + "num_tokens": 612877973.0, + "step": 24049 + }, + { + "epoch": 1.9831780324894863, + "grad_norm": 2.105278730392456, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7766023874282837, + "num_tokens": 612906072.0, + "step": 24050 + }, + { + "epoch": 1.9832604931145377, + "grad_norm": 1.6904948949813843, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.7627224922180176, + "num_tokens": 612943256.0, + "step": 24051 + }, + { + "epoch": 1.9833429537395895, + "grad_norm": 2.013840675354004, + "learning_rate": 1e-06, + "loss": 0.5661, + "mean_token_accuracy": 0.8101771473884583, + "num_tokens": 612977519.0, + "step": 24052 + }, + { + "epoch": 1.9834254143646408, + "grad_norm": 1.8443424701690674, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7248388528823853, + "num_tokens": 613015159.0, + "step": 24053 + }, + { + "epoch": 1.9835078749896924, + "grad_norm": 2.169607162475586, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7720288038253784, + "num_tokens": 613042041.0, + "step": 24054 + }, + { + "epoch": 1.983590335614744, + "grad_norm": 2.194624900817871, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7555772662162781, + "num_tokens": 613067498.0, + "step": 24055 + }, + { + "epoch": 1.9836727962397955, + "grad_norm": 2.0714168548583984, + "learning_rate": 1e-06, + "loss": 0.5845, + "mean_token_accuracy": 0.8004432916641235, + "num_tokens": 613093028.0, + "step": 24056 + }, + { + "epoch": 1.983755256864847, + "grad_norm": 1.7112222909927368, + "learning_rate": 1e-06, + "loss": 0.5545, + "mean_token_accuracy": 0.8212525248527527, + "num_tokens": 613126501.0, + "step": 24057 + }, + { + "epoch": 1.9838377174898985, + "grad_norm": 1.9329578876495361, + "learning_rate": 1e-06, + "loss": 0.5883, + "mean_token_accuracy": 0.8054434061050415, + "num_tokens": 613153032.0, + "step": 24058 + }, + { + "epoch": 1.9839201781149503, + "grad_norm": 1.8148845434188843, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7776154279708862, + "num_tokens": 613186934.0, + "step": 24059 + }, + { + "epoch": 1.9840026387400016, + "grad_norm": 2.1632919311523438, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6940592527389526, + "num_tokens": 613215012.0, + "step": 24060 + }, + { + "epoch": 1.9840850993650532, + "grad_norm": 1.733543038368225, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7394100427627563, + "num_tokens": 613251949.0, + "step": 24061 + }, + { + "epoch": 1.9841675599901047, + "grad_norm": 1.9984039068222046, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7651767134666443, + "num_tokens": 613283354.0, + "step": 24062 + }, + { + "epoch": 1.9842500206151563, + "grad_norm": 1.8291795253753662, + "learning_rate": 1e-06, + "loss": 0.5792, + "mean_token_accuracy": 0.7970393896102905, + "num_tokens": 613314927.0, + "step": 24063 + }, + { + "epoch": 1.984332481240208, + "grad_norm": 2.010326623916626, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7769350409507751, + "num_tokens": 613343391.0, + "step": 24064 + }, + { + "epoch": 1.9844149418652592, + "grad_norm": 1.8317897319793701, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7738001346588135, + "num_tokens": 613374850.0, + "step": 24065 + }, + { + "epoch": 1.984497402490311, + "grad_norm": 1.9848661422729492, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7381369471549988, + "num_tokens": 613405739.0, + "step": 24066 + }, + { + "epoch": 1.9845798631153624, + "grad_norm": 1.8890352249145508, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.7802493572235107, + "num_tokens": 613436898.0, + "step": 24067 + }, + { + "epoch": 1.984662323740414, + "grad_norm": 1.8166841268539429, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7580548524856567, + "num_tokens": 613469587.0, + "step": 24068 + }, + { + "epoch": 1.9847447843654655, + "grad_norm": 1.9314414262771606, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.754264235496521, + "num_tokens": 613501221.0, + "step": 24069 + }, + { + "epoch": 1.9848272449905169, + "grad_norm": 1.5889794826507568, + "learning_rate": 1e-06, + "loss": 0.5374, + "mean_token_accuracy": 0.804776668548584, + "num_tokens": 613538447.0, + "step": 24070 + }, + { + "epoch": 1.9849097056155687, + "grad_norm": 2.0330088138580322, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.785301923751831, + "num_tokens": 613565681.0, + "step": 24071 + }, + { + "epoch": 1.98499216624062, + "grad_norm": 1.9734870195388794, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7749935388565063, + "num_tokens": 613594325.0, + "step": 24072 + }, + { + "epoch": 1.9850746268656716, + "grad_norm": 1.9787395000457764, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7737178802490234, + "num_tokens": 613623334.0, + "step": 24073 + }, + { + "epoch": 1.9851570874907232, + "grad_norm": 2.0074009895324707, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7615131139755249, + "num_tokens": 613652576.0, + "step": 24074 + }, + { + "epoch": 1.9852395481157747, + "grad_norm": 1.8576613664627075, + "learning_rate": 1e-06, + "loss": 0.5878, + "mean_token_accuracy": 0.8071224093437195, + "num_tokens": 613682153.0, + "step": 24075 + }, + { + "epoch": 1.9853220087408263, + "grad_norm": 1.951885461807251, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7561378479003906, + "num_tokens": 613712188.0, + "step": 24076 + }, + { + "epoch": 1.9854044693658777, + "grad_norm": 1.9514305591583252, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.78251051902771, + "num_tokens": 613740108.0, + "step": 24077 + }, + { + "epoch": 1.9854869299909295, + "grad_norm": 1.9868606328964233, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7719314694404602, + "num_tokens": 613771528.0, + "step": 24078 + }, + { + "epoch": 1.9855693906159808, + "grad_norm": 2.065406084060669, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.7922670245170593, + "num_tokens": 613799110.0, + "step": 24079 + }, + { + "epoch": 1.9856518512410324, + "grad_norm": 2.160130500793457, + "learning_rate": 1e-06, + "loss": 0.5538, + "mean_token_accuracy": 0.8007106781005859, + "num_tokens": 613824175.0, + "step": 24080 + }, + { + "epoch": 1.985734311866084, + "grad_norm": 1.9833053350448608, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.7927645444869995, + "num_tokens": 613853041.0, + "step": 24081 + }, + { + "epoch": 1.9858167724911355, + "grad_norm": 1.8213459253311157, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7671376466751099, + "num_tokens": 613888781.0, + "step": 24082 + }, + { + "epoch": 1.985899233116187, + "grad_norm": 2.129053831100464, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7297024726867676, + "num_tokens": 613917908.0, + "step": 24083 + }, + { + "epoch": 1.9859816937412385, + "grad_norm": 2.001128911972046, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7605254054069519, + "num_tokens": 613949781.0, + "step": 24084 + }, + { + "epoch": 1.9860641543662902, + "grad_norm": 1.9192227125167847, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7434442639350891, + "num_tokens": 613980748.0, + "step": 24085 + }, + { + "epoch": 1.9861466149913416, + "grad_norm": 1.9869388341903687, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.7807823419570923, + "num_tokens": 614009129.0, + "step": 24086 + }, + { + "epoch": 1.9862290756163932, + "grad_norm": 1.8760343790054321, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.736267626285553, + "num_tokens": 614046684.0, + "step": 24087 + }, + { + "epoch": 1.9863115362414447, + "grad_norm": 2.275027275085449, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7382392883300781, + "num_tokens": 614071064.0, + "step": 24088 + }, + { + "epoch": 1.986393996866496, + "grad_norm": 1.949734091758728, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7558525800704956, + "num_tokens": 614101049.0, + "step": 24089 + }, + { + "epoch": 1.9864764574915479, + "grad_norm": 2.0043439865112305, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7658986449241638, + "num_tokens": 614128937.0, + "step": 24090 + }, + { + "epoch": 1.9865589181165992, + "grad_norm": 2.0678389072418213, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7679729461669922, + "num_tokens": 614155705.0, + "step": 24091 + }, + { + "epoch": 1.9866413787416508, + "grad_norm": 2.188832998275757, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7604333758354187, + "num_tokens": 614183424.0, + "step": 24092 + }, + { + "epoch": 1.9867238393667024, + "grad_norm": 2.02321720123291, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7629494071006775, + "num_tokens": 614209539.0, + "step": 24093 + }, + { + "epoch": 1.986806299991754, + "grad_norm": 2.0428431034088135, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7301892042160034, + "num_tokens": 614240288.0, + "step": 24094 + }, + { + "epoch": 1.9868887606168055, + "grad_norm": 1.8758715391159058, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7676129341125488, + "num_tokens": 614275261.0, + "step": 24095 + }, + { + "epoch": 1.9869712212418569, + "grad_norm": 2.1150314807891846, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7706136703491211, + "num_tokens": 614305834.0, + "step": 24096 + }, + { + "epoch": 1.9870536818669087, + "grad_norm": 1.7232153415679932, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.7852445840835571, + "num_tokens": 614339428.0, + "step": 24097 + }, + { + "epoch": 1.98713614249196, + "grad_norm": 1.9334040880203247, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.745262861251831, + "num_tokens": 614372424.0, + "step": 24098 + }, + { + "epoch": 1.9872186031170116, + "grad_norm": 2.3635482788085938, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7562522888183594, + "num_tokens": 614394369.0, + "step": 24099 + }, + { + "epoch": 1.9873010637420632, + "grad_norm": 1.823109745979309, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7671489119529724, + "num_tokens": 614427847.0, + "step": 24100 + }, + { + "epoch": 1.9873835243671147, + "grad_norm": 2.284923791885376, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7567266225814819, + "num_tokens": 614452509.0, + "step": 24101 + }, + { + "epoch": 1.9874659849921663, + "grad_norm": 2.135753870010376, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7692644596099854, + "num_tokens": 614479996.0, + "step": 24102 + }, + { + "epoch": 1.9875484456172177, + "grad_norm": 1.7252240180969238, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.7807734608650208, + "num_tokens": 614517983.0, + "step": 24103 + }, + { + "epoch": 1.9876309062422695, + "grad_norm": 2.0624420642852783, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.717790961265564, + "num_tokens": 614548367.0, + "step": 24104 + }, + { + "epoch": 1.9877133668673208, + "grad_norm": 1.9666283130645752, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7474303245544434, + "num_tokens": 614580331.0, + "step": 24105 + }, + { + "epoch": 1.9877958274923724, + "grad_norm": 2.0653507709503174, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7765207290649414, + "num_tokens": 614610885.0, + "step": 24106 + }, + { + "epoch": 1.987878288117424, + "grad_norm": 1.9316054582595825, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7466743588447571, + "num_tokens": 614643530.0, + "step": 24107 + }, + { + "epoch": 1.9879607487424753, + "grad_norm": 2.0618627071380615, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.7421846389770508, + "num_tokens": 614671873.0, + "step": 24108 + }, + { + "epoch": 1.988043209367527, + "grad_norm": 1.738983154296875, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.7877355813980103, + "num_tokens": 614706421.0, + "step": 24109 + }, + { + "epoch": 1.9881256699925784, + "grad_norm": 2.1839675903320312, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7237128019332886, + "num_tokens": 614734248.0, + "step": 24110 + }, + { + "epoch": 1.9882081306176302, + "grad_norm": 2.0405843257904053, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7640597820281982, + "num_tokens": 614760285.0, + "step": 24111 + }, + { + "epoch": 1.9882905912426816, + "grad_norm": 2.1877753734588623, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7370151877403259, + "num_tokens": 614787035.0, + "step": 24112 + }, + { + "epoch": 1.9883730518677332, + "grad_norm": 1.5128344297409058, + "learning_rate": 1e-06, + "loss": 0.5911, + "mean_token_accuracy": 0.8075228333473206, + "num_tokens": 614832871.0, + "step": 24113 + }, + { + "epoch": 1.9884555124927847, + "grad_norm": 1.7843745946884155, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7475905418395996, + "num_tokens": 614869568.0, + "step": 24114 + }, + { + "epoch": 1.988537973117836, + "grad_norm": 1.9861950874328613, + "learning_rate": 1e-06, + "loss": 0.5895, + "mean_token_accuracy": 0.798127293586731, + "num_tokens": 614898315.0, + "step": 24115 + }, + { + "epoch": 1.9886204337428879, + "grad_norm": 2.0522477626800537, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.773908793926239, + "num_tokens": 614928208.0, + "step": 24116 + }, + { + "epoch": 1.9887028943679392, + "grad_norm": 1.9935238361358643, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7084444761276245, + "num_tokens": 614958058.0, + "step": 24117 + }, + { + "epoch": 1.9887853549929908, + "grad_norm": 1.775465726852417, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7828723192214966, + "num_tokens": 614994191.0, + "step": 24118 + }, + { + "epoch": 1.9888678156180424, + "grad_norm": 1.8636925220489502, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7903952598571777, + "num_tokens": 615026488.0, + "step": 24119 + }, + { + "epoch": 1.988950276243094, + "grad_norm": 1.8586220741271973, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7469578981399536, + "num_tokens": 615058741.0, + "step": 24120 + }, + { + "epoch": 1.9890327368681455, + "grad_norm": 2.6288647651672363, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7277358174324036, + "num_tokens": 615077639.0, + "step": 24121 + }, + { + "epoch": 1.9891151974931969, + "grad_norm": 1.887954831123352, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7695668339729309, + "num_tokens": 615107715.0, + "step": 24122 + }, + { + "epoch": 1.9891976581182487, + "grad_norm": 1.9890247583389282, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7415591478347778, + "num_tokens": 615139386.0, + "step": 24123 + }, + { + "epoch": 1.9892801187433, + "grad_norm": 2.215937376022339, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.791567325592041, + "num_tokens": 615164030.0, + "step": 24124 + }, + { + "epoch": 1.9893625793683516, + "grad_norm": 2.140843629837036, + "learning_rate": 1e-06, + "loss": 0.5419, + "mean_token_accuracy": 0.8140224814414978, + "num_tokens": 615186777.0, + "step": 24125 + }, + { + "epoch": 1.9894450399934032, + "grad_norm": 1.9499406814575195, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7406387329101562, + "num_tokens": 615219574.0, + "step": 24126 + }, + { + "epoch": 1.9895275006184547, + "grad_norm": 2.140212059020996, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.7861713171005249, + "num_tokens": 615243805.0, + "step": 24127 + }, + { + "epoch": 1.9896099612435063, + "grad_norm": 1.8886381387710571, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7466682195663452, + "num_tokens": 615277512.0, + "step": 24128 + }, + { + "epoch": 1.9896924218685577, + "grad_norm": 2.2451882362365723, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7651715278625488, + "num_tokens": 615302076.0, + "step": 24129 + }, + { + "epoch": 1.9897748824936095, + "grad_norm": 1.9287015199661255, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7522212266921997, + "num_tokens": 615332814.0, + "step": 24130 + }, + { + "epoch": 1.9898573431186608, + "grad_norm": 1.8893730640411377, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7644729018211365, + "num_tokens": 615362222.0, + "step": 24131 + }, + { + "epoch": 1.9899398037437124, + "grad_norm": 2.0786972045898438, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7587589025497437, + "num_tokens": 615391492.0, + "step": 24132 + }, + { + "epoch": 1.990022264368764, + "grad_norm": 2.0153932571411133, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.7852380275726318, + "num_tokens": 615419322.0, + "step": 24133 + }, + { + "epoch": 1.9901047249938153, + "grad_norm": 1.7698683738708496, + "learning_rate": 1e-06, + "loss": 0.5747, + "mean_token_accuracy": 0.8084986209869385, + "num_tokens": 615453081.0, + "step": 24134 + }, + { + "epoch": 1.990187185618867, + "grad_norm": 1.7013647556304932, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7765145897865295, + "num_tokens": 615491770.0, + "step": 24135 + }, + { + "epoch": 1.9902696462439184, + "grad_norm": 2.188457489013672, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7660139203071594, + "num_tokens": 615519844.0, + "step": 24136 + }, + { + "epoch": 1.99035210686897, + "grad_norm": 1.8854997158050537, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7658028602600098, + "num_tokens": 615553087.0, + "step": 24137 + }, + { + "epoch": 1.9904345674940216, + "grad_norm": 1.9958140850067139, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7520692348480225, + "num_tokens": 615578855.0, + "step": 24138 + }, + { + "epoch": 1.9905170281190732, + "grad_norm": 2.0498502254486084, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7061057090759277, + "num_tokens": 615608234.0, + "step": 24139 + }, + { + "epoch": 1.9905994887441247, + "grad_norm": 2.0160977840423584, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7610983848571777, + "num_tokens": 615637634.0, + "step": 24140 + }, + { + "epoch": 1.990681949369176, + "grad_norm": 2.3593556880950928, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7446057796478271, + "num_tokens": 615661425.0, + "step": 24141 + }, + { + "epoch": 1.9907644099942279, + "grad_norm": 2.157841205596924, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7466334104537964, + "num_tokens": 615689753.0, + "step": 24142 + }, + { + "epoch": 1.9908468706192792, + "grad_norm": 1.7190722227096558, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.778206467628479, + "num_tokens": 615725491.0, + "step": 24143 + }, + { + "epoch": 1.9909293312443308, + "grad_norm": 1.78586745262146, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.78312087059021, + "num_tokens": 615759122.0, + "step": 24144 + }, + { + "epoch": 1.9910117918693824, + "grad_norm": 2.071274518966675, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7012426853179932, + "num_tokens": 615787171.0, + "step": 24145 + }, + { + "epoch": 1.991094252494434, + "grad_norm": 1.9116098880767822, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7747149467468262, + "num_tokens": 615818016.0, + "step": 24146 + }, + { + "epoch": 1.9911767131194855, + "grad_norm": 2.1844537258148193, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7412011027336121, + "num_tokens": 615845116.0, + "step": 24147 + }, + { + "epoch": 1.9912591737445369, + "grad_norm": 2.0367255210876465, + "learning_rate": 1e-06, + "loss": 0.7821, + "mean_token_accuracy": 0.7545042037963867, + "num_tokens": 615871831.0, + "step": 24148 + }, + { + "epoch": 1.9913416343695887, + "grad_norm": 1.8089587688446045, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7638465762138367, + "num_tokens": 615907103.0, + "step": 24149 + }, + { + "epoch": 1.99142409499464, + "grad_norm": 1.844765543937683, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7444193363189697, + "num_tokens": 615941394.0, + "step": 24150 + }, + { + "epoch": 1.9915065556196916, + "grad_norm": 2.0403170585632324, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7642459869384766, + "num_tokens": 615969018.0, + "step": 24151 + }, + { + "epoch": 1.9915890162447432, + "grad_norm": 1.9275128841400146, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7593933343887329, + "num_tokens": 616000296.0, + "step": 24152 + }, + { + "epoch": 1.9916714768697945, + "grad_norm": 1.9266291856765747, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7720221281051636, + "num_tokens": 616033337.0, + "step": 24153 + }, + { + "epoch": 1.9917539374948463, + "grad_norm": 1.8952301740646362, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7756595015525818, + "num_tokens": 616067696.0, + "step": 24154 + }, + { + "epoch": 1.9918363981198977, + "grad_norm": 1.7957022190093994, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.7876015305519104, + "num_tokens": 616102191.0, + "step": 24155 + }, + { + "epoch": 1.9919188587449494, + "grad_norm": 1.818772315979004, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7686283588409424, + "num_tokens": 616136106.0, + "step": 24156 + }, + { + "epoch": 1.9920013193700008, + "grad_norm": 1.9476922750473022, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7576482892036438, + "num_tokens": 616166463.0, + "step": 24157 + }, + { + "epoch": 1.9920837799950524, + "grad_norm": 1.8865967988967896, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7613216638565063, + "num_tokens": 616200204.0, + "step": 24158 + }, + { + "epoch": 1.992166240620104, + "grad_norm": 2.3225128650665283, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.724724292755127, + "num_tokens": 616224664.0, + "step": 24159 + }, + { + "epoch": 1.9922487012451553, + "grad_norm": 1.7598106861114502, + "learning_rate": 1e-06, + "loss": 0.5959, + "mean_token_accuracy": 0.7996408939361572, + "num_tokens": 616256961.0, + "step": 24160 + }, + { + "epoch": 1.992331161870207, + "grad_norm": 1.8886799812316895, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7519015073776245, + "num_tokens": 616288200.0, + "step": 24161 + }, + { + "epoch": 1.9924136224952584, + "grad_norm": 1.9235068559646606, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7425495386123657, + "num_tokens": 616320457.0, + "step": 24162 + }, + { + "epoch": 1.99249608312031, + "grad_norm": 1.888697862625122, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.755807638168335, + "num_tokens": 616352857.0, + "step": 24163 + }, + { + "epoch": 1.9925785437453616, + "grad_norm": 2.4002420902252197, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7733726501464844, + "num_tokens": 616374901.0, + "step": 24164 + }, + { + "epoch": 1.9926610043704132, + "grad_norm": 2.0209083557128906, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7601708173751831, + "num_tokens": 616401834.0, + "step": 24165 + }, + { + "epoch": 1.9927434649954647, + "grad_norm": 1.858376145362854, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7648327350616455, + "num_tokens": 616439246.0, + "step": 24166 + }, + { + "epoch": 1.992825925620516, + "grad_norm": 2.026796817779541, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7694512605667114, + "num_tokens": 616466371.0, + "step": 24167 + }, + { + "epoch": 1.9929083862455679, + "grad_norm": 1.7889057397842407, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7763959765434265, + "num_tokens": 616504555.0, + "step": 24168 + }, + { + "epoch": 1.9929908468706192, + "grad_norm": 2.7391560077667236, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7576228380203247, + "num_tokens": 616532895.0, + "step": 24169 + }, + { + "epoch": 1.9930733074956708, + "grad_norm": 1.9403153657913208, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7791967391967773, + "num_tokens": 616562203.0, + "step": 24170 + }, + { + "epoch": 1.9931557681207224, + "grad_norm": 1.6556142568588257, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7870115637779236, + "num_tokens": 616600418.0, + "step": 24171 + }, + { + "epoch": 1.993238228745774, + "grad_norm": 1.8681576251983643, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7894101142883301, + "num_tokens": 616634360.0, + "step": 24172 + }, + { + "epoch": 1.9933206893708255, + "grad_norm": 2.0136280059814453, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.7856137156486511, + "num_tokens": 616663852.0, + "step": 24173 + }, + { + "epoch": 1.9934031499958769, + "grad_norm": 2.0233190059661865, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.7882534265518188, + "num_tokens": 616691538.0, + "step": 24174 + }, + { + "epoch": 1.9934856106209287, + "grad_norm": 1.9453848600387573, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7518900632858276, + "num_tokens": 616723793.0, + "step": 24175 + }, + { + "epoch": 1.99356807124598, + "grad_norm": 2.173436403274536, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7440900802612305, + "num_tokens": 616748713.0, + "step": 24176 + }, + { + "epoch": 1.9936505318710316, + "grad_norm": 1.9082763195037842, + "learning_rate": 1e-06, + "loss": 0.5432, + "mean_token_accuracy": 0.8137645125389099, + "num_tokens": 616779111.0, + "step": 24177 + }, + { + "epoch": 1.9937329924960832, + "grad_norm": 2.232520580291748, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7478581666946411, + "num_tokens": 616806844.0, + "step": 24178 + }, + { + "epoch": 1.9938154531211345, + "grad_norm": 2.0361506938934326, + "learning_rate": 1e-06, + "loss": 0.7683, + "mean_token_accuracy": 0.7520740032196045, + "num_tokens": 616837742.0, + "step": 24179 + }, + { + "epoch": 1.9938979137461863, + "grad_norm": 1.8740731477737427, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.7942860722541809, + "num_tokens": 616868520.0, + "step": 24180 + }, + { + "epoch": 1.9939803743712377, + "grad_norm": 2.2490322589874268, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7529603242874146, + "num_tokens": 616894493.0, + "step": 24181 + }, + { + "epoch": 1.9940628349962892, + "grad_norm": 2.269118547439575, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.745361328125, + "num_tokens": 616917786.0, + "step": 24182 + }, + { + "epoch": 1.9941452956213408, + "grad_norm": 1.9395766258239746, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7836692929267883, + "num_tokens": 616948672.0, + "step": 24183 + }, + { + "epoch": 1.9942277562463924, + "grad_norm": 2.117624044418335, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7306848764419556, + "num_tokens": 616977034.0, + "step": 24184 + }, + { + "epoch": 1.994310216871444, + "grad_norm": 1.7643535137176514, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7849122881889343, + "num_tokens": 617011050.0, + "step": 24185 + }, + { + "epoch": 1.9943926774964953, + "grad_norm": 2.1847035884857178, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7592333555221558, + "num_tokens": 617034159.0, + "step": 24186 + }, + { + "epoch": 1.994475138121547, + "grad_norm": 1.7734180688858032, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7613658905029297, + "num_tokens": 617068813.0, + "step": 24187 + }, + { + "epoch": 1.9945575987465984, + "grad_norm": 1.6555813550949097, + "learning_rate": 1e-06, + "loss": 0.5111, + "mean_token_accuracy": 0.8208438158035278, + "num_tokens": 617105602.0, + "step": 24188 + }, + { + "epoch": 1.99464005937165, + "grad_norm": 2.016157865524292, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.7461615800857544, + "num_tokens": 617136469.0, + "step": 24189 + }, + { + "epoch": 1.9947225199967016, + "grad_norm": 1.7804415225982666, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7577288746833801, + "num_tokens": 617175261.0, + "step": 24190 + }, + { + "epoch": 1.9948049806217532, + "grad_norm": 1.894768238067627, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7699273228645325, + "num_tokens": 617209047.0, + "step": 24191 + }, + { + "epoch": 1.9948874412468047, + "grad_norm": 1.9099394083023071, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7540873885154724, + "num_tokens": 617243815.0, + "step": 24192 + }, + { + "epoch": 1.994969901871856, + "grad_norm": 1.8935874700546265, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7498724460601807, + "num_tokens": 617275668.0, + "step": 24193 + }, + { + "epoch": 1.9950523624969079, + "grad_norm": 2.1919076442718506, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7272108793258667, + "num_tokens": 617302336.0, + "step": 24194 + }, + { + "epoch": 1.9951348231219592, + "grad_norm": 1.8876844644546509, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.7851659059524536, + "num_tokens": 617333448.0, + "step": 24195 + }, + { + "epoch": 1.9952172837470108, + "grad_norm": 1.794844150543213, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7491818070411682, + "num_tokens": 617369208.0, + "step": 24196 + }, + { + "epoch": 1.9952997443720624, + "grad_norm": 1.8943272829055786, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.7869088649749756, + "num_tokens": 617399211.0, + "step": 24197 + }, + { + "epoch": 1.9953822049971137, + "grad_norm": 1.9718068838119507, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7673579454421997, + "num_tokens": 617428636.0, + "step": 24198 + }, + { + "epoch": 1.9954646656221655, + "grad_norm": 2.2746338844299316, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7781518697738647, + "num_tokens": 617452480.0, + "step": 24199 + }, + { + "epoch": 1.9955471262472169, + "grad_norm": 1.858773946762085, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7308580875396729, + "num_tokens": 617485011.0, + "step": 24200 + }, + { + "epoch": 1.9956295868722687, + "grad_norm": 2.0184412002563477, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7464748620986938, + "num_tokens": 617515021.0, + "step": 24201 + }, + { + "epoch": 1.99571204749732, + "grad_norm": 2.06280779838562, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7668799161911011, + "num_tokens": 617541824.0, + "step": 24202 + }, + { + "epoch": 1.9957945081223716, + "grad_norm": 1.948218584060669, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7433949112892151, + "num_tokens": 617571807.0, + "step": 24203 + }, + { + "epoch": 1.9958769687474232, + "grad_norm": 1.7192307710647583, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7657783031463623, + "num_tokens": 617607498.0, + "step": 24204 + }, + { + "epoch": 1.9959594293724745, + "grad_norm": 1.9632295370101929, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.7982600927352905, + "num_tokens": 617635085.0, + "step": 24205 + }, + { + "epoch": 1.9960418899975263, + "grad_norm": 1.654862880706787, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.7929971218109131, + "num_tokens": 617670927.0, + "step": 24206 + }, + { + "epoch": 1.9961243506225776, + "grad_norm": 1.7209014892578125, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.7973524332046509, + "num_tokens": 617707175.0, + "step": 24207 + }, + { + "epoch": 1.9962068112476292, + "grad_norm": 1.9517515897750854, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.778082549571991, + "num_tokens": 617737829.0, + "step": 24208 + }, + { + "epoch": 1.9962892718726808, + "grad_norm": 1.935921311378479, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7524313926696777, + "num_tokens": 617770060.0, + "step": 24209 + }, + { + "epoch": 1.9963717324977324, + "grad_norm": 2.1077115535736084, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7769023776054382, + "num_tokens": 617795416.0, + "step": 24210 + }, + { + "epoch": 1.996454193122784, + "grad_norm": 2.4151341915130615, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7288553714752197, + "num_tokens": 617816594.0, + "step": 24211 + }, + { + "epoch": 1.9965366537478353, + "grad_norm": 1.9913691282272339, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7544596195220947, + "num_tokens": 617847084.0, + "step": 24212 + }, + { + "epoch": 1.996619114372887, + "grad_norm": 2.1020078659057617, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7510605454444885, + "num_tokens": 617874862.0, + "step": 24213 + }, + { + "epoch": 1.9967015749979384, + "grad_norm": 2.0609517097473145, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7383627891540527, + "num_tokens": 617903908.0, + "step": 24214 + }, + { + "epoch": 1.99678403562299, + "grad_norm": 2.090240716934204, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7380993366241455, + "num_tokens": 617931285.0, + "step": 24215 + }, + { + "epoch": 1.9968664962480416, + "grad_norm": 2.0500078201293945, + "learning_rate": 1e-06, + "loss": 0.5926, + "mean_token_accuracy": 0.8017805814743042, + "num_tokens": 617957210.0, + "step": 24216 + }, + { + "epoch": 1.9969489568730932, + "grad_norm": 1.9525468349456787, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7280141115188599, + "num_tokens": 617990900.0, + "step": 24217 + }, + { + "epoch": 1.9970314174981447, + "grad_norm": 2.09885573387146, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.745667576789856, + "num_tokens": 618019039.0, + "step": 24218 + }, + { + "epoch": 1.997113878123196, + "grad_norm": 1.8751389980316162, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7411739230155945, + "num_tokens": 618051948.0, + "step": 24219 + }, + { + "epoch": 1.9971963387482479, + "grad_norm": 1.938758134841919, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7391504049301147, + "num_tokens": 618084546.0, + "step": 24220 + }, + { + "epoch": 1.9972787993732992, + "grad_norm": 2.085822820663452, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7611100673675537, + "num_tokens": 618112301.0, + "step": 24221 + }, + { + "epoch": 1.9973612599983508, + "grad_norm": 1.9728013277053833, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7181982398033142, + "num_tokens": 618145435.0, + "step": 24222 + }, + { + "epoch": 1.9974437206234024, + "grad_norm": 1.7862355709075928, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.7872199416160583, + "num_tokens": 618179881.0, + "step": 24223 + }, + { + "epoch": 1.9975261812484537, + "grad_norm": 1.9881908893585205, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7507343888282776, + "num_tokens": 618210459.0, + "step": 24224 + }, + { + "epoch": 1.9976086418735055, + "grad_norm": 1.8966172933578491, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7641147375106812, + "num_tokens": 618242446.0, + "step": 24225 + }, + { + "epoch": 1.9976911024985569, + "grad_norm": 2.068855047225952, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.7898720502853394, + "num_tokens": 618269915.0, + "step": 24226 + }, + { + "epoch": 1.9977735631236084, + "grad_norm": 1.9525599479675293, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.7743762731552124, + "num_tokens": 618302953.0, + "step": 24227 + }, + { + "epoch": 1.99785602374866, + "grad_norm": 2.1123294830322266, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.7872484922409058, + "num_tokens": 618329682.0, + "step": 24228 + }, + { + "epoch": 1.9979384843737116, + "grad_norm": 2.088231086730957, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7456502914428711, + "num_tokens": 618356878.0, + "step": 24229 + }, + { + "epoch": 1.9980209449987631, + "grad_norm": 2.0674164295196533, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.783074140548706, + "num_tokens": 618385867.0, + "step": 24230 + }, + { + "epoch": 1.9981034056238145, + "grad_norm": 1.9129011631011963, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7176219820976257, + "num_tokens": 618418436.0, + "step": 24231 + }, + { + "epoch": 1.9981858662488663, + "grad_norm": 1.9155505895614624, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7459690570831299, + "num_tokens": 618449890.0, + "step": 24232 + }, + { + "epoch": 1.9982683268739176, + "grad_norm": 2.1509079933166504, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7555424571037292, + "num_tokens": 618474501.0, + "step": 24233 + }, + { + "epoch": 1.9983507874989692, + "grad_norm": 1.8264957666397095, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7662717700004578, + "num_tokens": 618507999.0, + "step": 24234 + }, + { + "epoch": 1.9984332481240208, + "grad_norm": 1.7411175966262817, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7680493593215942, + "num_tokens": 618545614.0, + "step": 24235 + }, + { + "epoch": 1.9985157087490724, + "grad_norm": 1.8172944784164429, + "learning_rate": 1e-06, + "loss": 0.5773, + "mean_token_accuracy": 0.8127975463867188, + "num_tokens": 618578404.0, + "step": 24236 + }, + { + "epoch": 1.998598169374124, + "grad_norm": 1.758944034576416, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.7881264686584473, + "num_tokens": 618612022.0, + "step": 24237 + }, + { + "epoch": 1.9986806299991753, + "grad_norm": 2.0221219062805176, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7589544057846069, + "num_tokens": 618641279.0, + "step": 24238 + }, + { + "epoch": 1.998763090624227, + "grad_norm": 2.3978774547576904, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.752860426902771, + "num_tokens": 618663643.0, + "step": 24239 + }, + { + "epoch": 1.9988455512492784, + "grad_norm": 2.0148704051971436, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7766647338867188, + "num_tokens": 618694410.0, + "step": 24240 + }, + { + "epoch": 1.99892801187433, + "grad_norm": 2.0811514854431152, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7649489641189575, + "num_tokens": 618721819.0, + "step": 24241 + }, + { + "epoch": 1.9990104724993816, + "grad_norm": 1.9150466918945312, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7712960839271545, + "num_tokens": 618753413.0, + "step": 24242 + }, + { + "epoch": 1.999092933124433, + "grad_norm": 1.8081233501434326, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.752922773361206, + "num_tokens": 618788134.0, + "step": 24243 + }, + { + "epoch": 1.9991753937494847, + "grad_norm": 2.067936658859253, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7710116505622864, + "num_tokens": 618814791.0, + "step": 24244 + }, + { + "epoch": 1.999257854374536, + "grad_norm": 1.9276856184005737, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7734023332595825, + "num_tokens": 618843582.0, + "step": 24245 + }, + { + "epoch": 1.9993403149995876, + "grad_norm": 1.8369818925857544, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7863730788230896, + "num_tokens": 618874647.0, + "step": 24246 + }, + { + "epoch": 1.9994227756246392, + "grad_norm": 1.7484911680221558, + "learning_rate": 1e-06, + "loss": 0.588, + "mean_token_accuracy": 0.8105801939964294, + "num_tokens": 618907376.0, + "step": 24247 + }, + { + "epoch": 1.9995052362496908, + "grad_norm": 2.040039539337158, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.7906292676925659, + "num_tokens": 618935035.0, + "step": 24248 + }, + { + "epoch": 1.9995876968747424, + "grad_norm": 1.795030117034912, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.758263349533081, + "num_tokens": 618968162.0, + "step": 24249 + }, + { + "epoch": 1.9996701574997937, + "grad_norm": 1.8033664226531982, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7725444436073303, + "num_tokens": 619003964.0, + "step": 24250 + }, + { + "epoch": 1.9997526181248455, + "grad_norm": 1.947823405265808, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7738039493560791, + "num_tokens": 619032021.0, + "step": 24251 + }, + { + "epoch": 1.9998350787498969, + "grad_norm": 2.2596559524536133, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7427040934562683, + "num_tokens": 619055943.0, + "step": 24252 + }, + { + "epoch": 1.9999175393749484, + "grad_norm": 1.9453370571136475, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7876302599906921, + "num_tokens": 619084779.0, + "step": 24253 + }, + { + "epoch": 2.0, + "grad_norm": 2.2246415615081787, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.737106442451477, + "num_tokens": 619108718.0, + "step": 24254 + }, + { + "epoch": 2.0000824606250514, + "grad_norm": 1.8253880739212036, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.7845464944839478, + "num_tokens": 619139699.0, + "step": 24255 + }, + { + "epoch": 2.000164921250103, + "grad_norm": 2.0206120014190674, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7469451427459717, + "num_tokens": 619167748.0, + "step": 24256 + }, + { + "epoch": 2.0002473818751545, + "grad_norm": 1.9892282485961914, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.770258903503418, + "num_tokens": 619200289.0, + "step": 24257 + }, + { + "epoch": 2.0003298425002063, + "grad_norm": 1.8001340627670288, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7936890125274658, + "num_tokens": 619235001.0, + "step": 24258 + }, + { + "epoch": 2.0004123031252576, + "grad_norm": 2.2089972496032715, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7484570741653442, + "num_tokens": 619259494.0, + "step": 24259 + }, + { + "epoch": 2.0004947637503094, + "grad_norm": 1.8236632347106934, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.7758710980415344, + "num_tokens": 619292227.0, + "step": 24260 + }, + { + "epoch": 2.000577224375361, + "grad_norm": 2.1207334995269775, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7515857219696045, + "num_tokens": 619324519.0, + "step": 24261 + }, + { + "epoch": 2.000659685000412, + "grad_norm": 2.215695381164551, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7502639293670654, + "num_tokens": 619351267.0, + "step": 24262 + }, + { + "epoch": 2.000742145625464, + "grad_norm": 2.0339877605438232, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7912534475326538, + "num_tokens": 619378430.0, + "step": 24263 + }, + { + "epoch": 2.0008246062505153, + "grad_norm": 2.2636499404907227, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7260850667953491, + "num_tokens": 619405338.0, + "step": 24264 + }, + { + "epoch": 2.000907066875567, + "grad_norm": 2.1786649227142334, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7350751161575317, + "num_tokens": 619432221.0, + "step": 24265 + }, + { + "epoch": 2.0009895275006184, + "grad_norm": 1.9700052738189697, + "learning_rate": 1e-06, + "loss": 0.5935, + "mean_token_accuracy": 0.8049495220184326, + "num_tokens": 619461199.0, + "step": 24266 + }, + { + "epoch": 2.0010719881256698, + "grad_norm": 1.9766433238983154, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7737433910369873, + "num_tokens": 619491777.0, + "step": 24267 + }, + { + "epoch": 2.0011544487507216, + "grad_norm": 1.7866963148117065, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7587608098983765, + "num_tokens": 619524795.0, + "step": 24268 + }, + { + "epoch": 2.001236909375773, + "grad_norm": 2.055225372314453, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.7800183296203613, + "num_tokens": 619552768.0, + "step": 24269 + }, + { + "epoch": 2.0013193700008247, + "grad_norm": 2.0969340801239014, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.7374690175056458, + "num_tokens": 619585189.0, + "step": 24270 + }, + { + "epoch": 2.001401830625876, + "grad_norm": 1.8005417585372925, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7799907922744751, + "num_tokens": 619619097.0, + "step": 24271 + }, + { + "epoch": 2.001484291250928, + "grad_norm": 2.100579261779785, + "learning_rate": 1e-06, + "loss": 0.5956, + "mean_token_accuracy": 0.797964334487915, + "num_tokens": 619648130.0, + "step": 24272 + }, + { + "epoch": 2.001566751875979, + "grad_norm": 1.9359164237976074, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.781963586807251, + "num_tokens": 619682811.0, + "step": 24273 + }, + { + "epoch": 2.0016492125010306, + "grad_norm": 2.1643526554107666, + "learning_rate": 1e-06, + "loss": 0.5517, + "mean_token_accuracy": 0.8095521926879883, + "num_tokens": 619707450.0, + "step": 24274 + }, + { + "epoch": 2.0017316731260824, + "grad_norm": 1.8649438619613647, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7675151824951172, + "num_tokens": 619741369.0, + "step": 24275 + }, + { + "epoch": 2.0018141337511337, + "grad_norm": 1.9271399974822998, + "learning_rate": 1e-06, + "loss": 0.5265, + "mean_token_accuracy": 0.8213406205177307, + "num_tokens": 619772885.0, + "step": 24276 + }, + { + "epoch": 2.0018965943761855, + "grad_norm": 1.9684078693389893, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.7855212688446045, + "num_tokens": 619805883.0, + "step": 24277 + }, + { + "epoch": 2.001979055001237, + "grad_norm": 2.065067768096924, + "learning_rate": 1e-06, + "loss": 0.527, + "mean_token_accuracy": 0.8208212852478027, + "num_tokens": 619834900.0, + "step": 24278 + }, + { + "epoch": 2.0020615156262886, + "grad_norm": 2.1599068641662598, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7568415403366089, + "num_tokens": 619862659.0, + "step": 24279 + }, + { + "epoch": 2.00214397625134, + "grad_norm": 2.1250813007354736, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7671732902526855, + "num_tokens": 619891333.0, + "step": 24280 + }, + { + "epoch": 2.0022264368763913, + "grad_norm": 2.284611701965332, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7551621198654175, + "num_tokens": 619917982.0, + "step": 24281 + }, + { + "epoch": 2.002308897501443, + "grad_norm": 2.246932029724121, + "learning_rate": 1e-06, + "loss": 0.5367, + "mean_token_accuracy": 0.8087198734283447, + "num_tokens": 619942014.0, + "step": 24282 + }, + { + "epoch": 2.0023913581264945, + "grad_norm": 2.1558592319488525, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.8001416921615601, + "num_tokens": 619967657.0, + "step": 24283 + }, + { + "epoch": 2.0024738187515463, + "grad_norm": 1.915773630142212, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7769785523414612, + "num_tokens": 620000292.0, + "step": 24284 + }, + { + "epoch": 2.0025562793765976, + "grad_norm": 2.017338752746582, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.7880277633666992, + "num_tokens": 620028319.0, + "step": 24285 + }, + { + "epoch": 2.0026387400016494, + "grad_norm": 2.3577632904052734, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7602657675743103, + "num_tokens": 620050789.0, + "step": 24286 + }, + { + "epoch": 2.002721200626701, + "grad_norm": 2.3464388847351074, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7480029463768005, + "num_tokens": 620077586.0, + "step": 24287 + }, + { + "epoch": 2.002803661251752, + "grad_norm": 2.0386462211608887, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7853306531906128, + "num_tokens": 620109069.0, + "step": 24288 + }, + { + "epoch": 2.002886121876804, + "grad_norm": 1.9292405843734741, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.785772442817688, + "num_tokens": 620140865.0, + "step": 24289 + }, + { + "epoch": 2.0029685825018553, + "grad_norm": 2.000274896621704, + "learning_rate": 1e-06, + "loss": 0.7465, + "mean_token_accuracy": 0.7729228734970093, + "num_tokens": 620172877.0, + "step": 24290 + }, + { + "epoch": 2.003051043126907, + "grad_norm": 2.000199317932129, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.7907459735870361, + "num_tokens": 620203615.0, + "step": 24291 + }, + { + "epoch": 2.0031335037519584, + "grad_norm": 2.114687919616699, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7624170184135437, + "num_tokens": 620232417.0, + "step": 24292 + }, + { + "epoch": 2.0032159643770098, + "grad_norm": 1.6962615251541138, + "learning_rate": 1e-06, + "loss": 0.5703, + "mean_token_accuracy": 0.8133934736251831, + "num_tokens": 620271518.0, + "step": 24293 + }, + { + "epoch": 2.0032984250020616, + "grad_norm": 2.0518720149993896, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7708616256713867, + "num_tokens": 620301448.0, + "step": 24294 + }, + { + "epoch": 2.003380885627113, + "grad_norm": 2.0475854873657227, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7632952928543091, + "num_tokens": 620330614.0, + "step": 24295 + }, + { + "epoch": 2.0034633462521647, + "grad_norm": 2.2862071990966797, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7729188203811646, + "num_tokens": 620354613.0, + "step": 24296 + }, + { + "epoch": 2.003545806877216, + "grad_norm": 2.08622670173645, + "learning_rate": 1e-06, + "loss": 0.5753, + "mean_token_accuracy": 0.8039682507514954, + "num_tokens": 620381136.0, + "step": 24297 + }, + { + "epoch": 2.003628267502268, + "grad_norm": 2.0880000591278076, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7727375030517578, + "num_tokens": 620410233.0, + "step": 24298 + }, + { + "epoch": 2.003710728127319, + "grad_norm": 2.0503790378570557, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7722225189208984, + "num_tokens": 620440916.0, + "step": 24299 + }, + { + "epoch": 2.0037931887523706, + "grad_norm": 2.0839056968688965, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.750267505645752, + "num_tokens": 620469292.0, + "step": 24300 + }, + { + "epoch": 2.0038756493774224, + "grad_norm": 2.1085004806518555, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7162158489227295, + "num_tokens": 620500699.0, + "step": 24301 + }, + { + "epoch": 2.0039581100024737, + "grad_norm": 2.0754306316375732, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.7910169959068298, + "num_tokens": 620529363.0, + "step": 24302 + }, + { + "epoch": 2.0040405706275255, + "grad_norm": 1.8251606225967407, + "learning_rate": 1e-06, + "loss": 0.5923, + "mean_token_accuracy": 0.8005276918411255, + "num_tokens": 620564011.0, + "step": 24303 + }, + { + "epoch": 2.004123031252577, + "grad_norm": 1.7975894212722778, + "learning_rate": 1e-06, + "loss": 0.4968, + "mean_token_accuracy": 0.833621084690094, + "num_tokens": 620601226.0, + "step": 24304 + }, + { + "epoch": 2.0042054918776286, + "grad_norm": 2.1925017833709717, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7438664436340332, + "num_tokens": 620631584.0, + "step": 24305 + }, + { + "epoch": 2.00428795250268, + "grad_norm": 2.1949381828308105, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7748644351959229, + "num_tokens": 620655836.0, + "step": 24306 + }, + { + "epoch": 2.0043704131277313, + "grad_norm": 2.030968189239502, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7458905577659607, + "num_tokens": 620688402.0, + "step": 24307 + }, + { + "epoch": 2.004452873752783, + "grad_norm": 2.1491425037384033, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7766907811164856, + "num_tokens": 620715931.0, + "step": 24308 + }, + { + "epoch": 2.0045353343778345, + "grad_norm": 1.993477702140808, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7698720097541809, + "num_tokens": 620745822.0, + "step": 24309 + }, + { + "epoch": 2.0046177950028863, + "grad_norm": 2.344815731048584, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7361266016960144, + "num_tokens": 620770830.0, + "step": 24310 + }, + { + "epoch": 2.0047002556279376, + "grad_norm": 2.394686222076416, + "learning_rate": 1e-06, + "loss": 0.5792, + "mean_token_accuracy": 0.802916407585144, + "num_tokens": 620792011.0, + "step": 24311 + }, + { + "epoch": 2.004782716252989, + "grad_norm": 2.178436517715454, + "learning_rate": 1e-06, + "loss": 0.5979, + "mean_token_accuracy": 0.7959604263305664, + "num_tokens": 620819709.0, + "step": 24312 + }, + { + "epoch": 2.0048651768780408, + "grad_norm": 1.8891410827636719, + "learning_rate": 1e-06, + "loss": 0.6135, + "mean_token_accuracy": 0.7893310785293579, + "num_tokens": 620848063.0, + "step": 24313 + }, + { + "epoch": 2.004947637503092, + "grad_norm": 1.9606740474700928, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.767469048500061, + "num_tokens": 620879907.0, + "step": 24314 + }, + { + "epoch": 2.005030098128144, + "grad_norm": 1.999390721321106, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7402223348617554, + "num_tokens": 620912864.0, + "step": 24315 + }, + { + "epoch": 2.0051125587531953, + "grad_norm": 1.9254276752471924, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7773022651672363, + "num_tokens": 620945425.0, + "step": 24316 + }, + { + "epoch": 2.005195019378247, + "grad_norm": 1.780708909034729, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.7667039036750793, + "num_tokens": 620986432.0, + "step": 24317 + }, + { + "epoch": 2.0052774800032984, + "grad_norm": 2.314798355102539, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7738826870918274, + "num_tokens": 621012362.0, + "step": 24318 + }, + { + "epoch": 2.0053599406283498, + "grad_norm": 2.0495989322662354, + "learning_rate": 1e-06, + "loss": 0.5676, + "mean_token_accuracy": 0.7937065362930298, + "num_tokens": 621039455.0, + "step": 24319 + }, + { + "epoch": 2.0054424012534016, + "grad_norm": 1.8830362558364868, + "learning_rate": 1e-06, + "loss": 0.481, + "mean_token_accuracy": 0.8382289409637451, + "num_tokens": 621070840.0, + "step": 24320 + }, + { + "epoch": 2.005524861878453, + "grad_norm": 1.9417812824249268, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7723233699798584, + "num_tokens": 621101811.0, + "step": 24321 + }, + { + "epoch": 2.0056073225035047, + "grad_norm": 1.7670092582702637, + "learning_rate": 1e-06, + "loss": 0.5557, + "mean_token_accuracy": 0.8161615133285522, + "num_tokens": 621138582.0, + "step": 24322 + }, + { + "epoch": 2.005689783128556, + "grad_norm": 1.872698187828064, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7911619544029236, + "num_tokens": 621174504.0, + "step": 24323 + }, + { + "epoch": 2.005772243753608, + "grad_norm": 1.9235146045684814, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.746535062789917, + "num_tokens": 621208448.0, + "step": 24324 + }, + { + "epoch": 2.005854704378659, + "grad_norm": 2.4249258041381836, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7605122327804565, + "num_tokens": 621232798.0, + "step": 24325 + }, + { + "epoch": 2.0059371650037106, + "grad_norm": 1.8564940690994263, + "learning_rate": 1e-06, + "loss": 0.5433, + "mean_token_accuracy": 0.8063724040985107, + "num_tokens": 621264172.0, + "step": 24326 + }, + { + "epoch": 2.0060196256287623, + "grad_norm": 2.0024147033691406, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7419571876525879, + "num_tokens": 621295475.0, + "step": 24327 + }, + { + "epoch": 2.0061020862538137, + "grad_norm": 1.8520129919052124, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.7971736788749695, + "num_tokens": 621331778.0, + "step": 24328 + }, + { + "epoch": 2.0061845468788655, + "grad_norm": 2.0046825408935547, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.721770703792572, + "num_tokens": 621362151.0, + "step": 24329 + }, + { + "epoch": 2.006267007503917, + "grad_norm": 2.1988272666931152, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7384060621261597, + "num_tokens": 621390231.0, + "step": 24330 + }, + { + "epoch": 2.0063494681289686, + "grad_norm": 1.952415943145752, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7816438674926758, + "num_tokens": 621421522.0, + "step": 24331 + }, + { + "epoch": 2.00643192875402, + "grad_norm": 2.3840062618255615, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7647976875305176, + "num_tokens": 621446990.0, + "step": 24332 + }, + { + "epoch": 2.0065143893790713, + "grad_norm": 2.280327796936035, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7403093576431274, + "num_tokens": 621473567.0, + "step": 24333 + }, + { + "epoch": 2.006596850004123, + "grad_norm": 1.9968280792236328, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.765312910079956, + "num_tokens": 621505230.0, + "step": 24334 + }, + { + "epoch": 2.0066793106291745, + "grad_norm": 1.9621293544769287, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7446532249450684, + "num_tokens": 621538709.0, + "step": 24335 + }, + { + "epoch": 2.0067617712542263, + "grad_norm": 1.8197021484375, + "learning_rate": 1e-06, + "loss": 0.528, + "mean_token_accuracy": 0.8329787254333496, + "num_tokens": 621576033.0, + "step": 24336 + }, + { + "epoch": 2.0068442318792776, + "grad_norm": 2.2505545616149902, + "learning_rate": 1e-06, + "loss": 0.5681, + "mean_token_accuracy": 0.8112138509750366, + "num_tokens": 621600349.0, + "step": 24337 + }, + { + "epoch": 2.006926692504329, + "grad_norm": 1.9323196411132812, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7240996360778809, + "num_tokens": 621635892.0, + "step": 24338 + }, + { + "epoch": 2.0070091531293808, + "grad_norm": 2.014728307723999, + "learning_rate": 1e-06, + "loss": 0.6132, + "mean_token_accuracy": 0.7891725301742554, + "num_tokens": 621666632.0, + "step": 24339 + }, + { + "epoch": 2.007091613754432, + "grad_norm": 1.7221977710723877, + "learning_rate": 1e-06, + "loss": 0.5594, + "mean_token_accuracy": 0.8093520402908325, + "num_tokens": 621702006.0, + "step": 24340 + }, + { + "epoch": 2.007174074379484, + "grad_norm": 2.203961133956909, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.741402268409729, + "num_tokens": 621728141.0, + "step": 24341 + }, + { + "epoch": 2.0072565350045353, + "grad_norm": 2.6780853271484375, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7761086225509644, + "num_tokens": 621746189.0, + "step": 24342 + }, + { + "epoch": 2.007338995629587, + "grad_norm": 1.9678757190704346, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7747617363929749, + "num_tokens": 621779330.0, + "step": 24343 + }, + { + "epoch": 2.0074214562546384, + "grad_norm": 2.334237813949585, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.7817732691764832, + "num_tokens": 621803809.0, + "step": 24344 + }, + { + "epoch": 2.0075039168796898, + "grad_norm": 1.968644142150879, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7565813064575195, + "num_tokens": 621837127.0, + "step": 24345 + }, + { + "epoch": 2.0075863775047416, + "grad_norm": 1.8713587522506714, + "learning_rate": 1e-06, + "loss": 0.5916, + "mean_token_accuracy": 0.8044135570526123, + "num_tokens": 621868949.0, + "step": 24346 + }, + { + "epoch": 2.007668838129793, + "grad_norm": 1.977903962135315, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7592588663101196, + "num_tokens": 621902628.0, + "step": 24347 + }, + { + "epoch": 2.0077512987548447, + "grad_norm": 2.1280596256256104, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.7900182008743286, + "num_tokens": 621929354.0, + "step": 24348 + }, + { + "epoch": 2.007833759379896, + "grad_norm": 2.014263391494751, + "learning_rate": 1e-06, + "loss": 0.6075, + "mean_token_accuracy": 0.7994188666343689, + "num_tokens": 621959791.0, + "step": 24349 + }, + { + "epoch": 2.007916220004948, + "grad_norm": 2.137158155441284, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.7867136001586914, + "num_tokens": 621990034.0, + "step": 24350 + }, + { + "epoch": 2.007998680629999, + "grad_norm": 1.984634280204773, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7698789834976196, + "num_tokens": 622026121.0, + "step": 24351 + }, + { + "epoch": 2.0080811412550506, + "grad_norm": 2.3350677490234375, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7520474195480347, + "num_tokens": 622051068.0, + "step": 24352 + }, + { + "epoch": 2.0081636018801023, + "grad_norm": 2.099086284637451, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.7574421167373657, + "num_tokens": 622083901.0, + "step": 24353 + }, + { + "epoch": 2.0082460625051537, + "grad_norm": 1.9248563051223755, + "learning_rate": 1e-06, + "loss": 0.5385, + "mean_token_accuracy": 0.8201759457588196, + "num_tokens": 622114000.0, + "step": 24354 + }, + { + "epoch": 2.0083285231302055, + "grad_norm": 1.7691272497177124, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7700825929641724, + "num_tokens": 622148602.0, + "step": 24355 + }, + { + "epoch": 2.008410983755257, + "grad_norm": 2.4202795028686523, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.7767512798309326, + "num_tokens": 622177019.0, + "step": 24356 + }, + { + "epoch": 2.008493444380308, + "grad_norm": 2.0166547298431396, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.7801722288131714, + "num_tokens": 622206451.0, + "step": 24357 + }, + { + "epoch": 2.00857590500536, + "grad_norm": 2.056873083114624, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7716509699821472, + "num_tokens": 622237238.0, + "step": 24358 + }, + { + "epoch": 2.0086583656304113, + "grad_norm": 2.007018566131592, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.781268298625946, + "num_tokens": 622266681.0, + "step": 24359 + }, + { + "epoch": 2.008740826255463, + "grad_norm": 2.2231788635253906, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7674810886383057, + "num_tokens": 622295479.0, + "step": 24360 + }, + { + "epoch": 2.0088232868805145, + "grad_norm": 2.1916658878326416, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7476643919944763, + "num_tokens": 622325695.0, + "step": 24361 + }, + { + "epoch": 2.0089057475055663, + "grad_norm": 2.1370766162872314, + "learning_rate": 1e-06, + "loss": 0.5198, + "mean_token_accuracy": 0.8184493780136108, + "num_tokens": 622352467.0, + "step": 24362 + }, + { + "epoch": 2.0089882081306176, + "grad_norm": 2.3004305362701416, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.7580760717391968, + "num_tokens": 622377192.0, + "step": 24363 + }, + { + "epoch": 2.009070668755669, + "grad_norm": 2.1308774948120117, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7569586038589478, + "num_tokens": 622404787.0, + "step": 24364 + }, + { + "epoch": 2.0091531293807208, + "grad_norm": 1.8652598857879639, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.7809200882911682, + "num_tokens": 622438559.0, + "step": 24365 + }, + { + "epoch": 2.009235590005772, + "grad_norm": 2.2359516620635986, + "learning_rate": 1e-06, + "loss": 0.5985, + "mean_token_accuracy": 0.7937440872192383, + "num_tokens": 622465657.0, + "step": 24366 + }, + { + "epoch": 2.009318050630824, + "grad_norm": 2.3539624214172363, + "learning_rate": 1e-06, + "loss": 0.6053, + "mean_token_accuracy": 0.794471800327301, + "num_tokens": 622488720.0, + "step": 24367 + }, + { + "epoch": 2.0094005112558753, + "grad_norm": 2.2532405853271484, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7590987682342529, + "num_tokens": 622521441.0, + "step": 24368 + }, + { + "epoch": 2.009482971880927, + "grad_norm": 1.9344520568847656, + "learning_rate": 1e-06, + "loss": 0.5783, + "mean_token_accuracy": 0.801232099533081, + "num_tokens": 622553227.0, + "step": 24369 + }, + { + "epoch": 2.0095654325059784, + "grad_norm": 2.3953137397766113, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7804014682769775, + "num_tokens": 622575247.0, + "step": 24370 + }, + { + "epoch": 2.0096478931310298, + "grad_norm": 2.366316318511963, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7740237712860107, + "num_tokens": 622598932.0, + "step": 24371 + }, + { + "epoch": 2.0097303537560816, + "grad_norm": 1.8617746829986572, + "learning_rate": 1e-06, + "loss": 0.5987, + "mean_token_accuracy": 0.8050808310508728, + "num_tokens": 622630065.0, + "step": 24372 + }, + { + "epoch": 2.009812814381133, + "grad_norm": 2.0417213439941406, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7325806617736816, + "num_tokens": 622663130.0, + "step": 24373 + }, + { + "epoch": 2.0098952750061847, + "grad_norm": 2.0274527072906494, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7359654307365417, + "num_tokens": 622694295.0, + "step": 24374 + }, + { + "epoch": 2.009977735631236, + "grad_norm": 2.1844325065612793, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7560123205184937, + "num_tokens": 622721527.0, + "step": 24375 + }, + { + "epoch": 2.010060196256288, + "grad_norm": 2.4684672355651855, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7862330675125122, + "num_tokens": 622747526.0, + "step": 24376 + }, + { + "epoch": 2.010142656881339, + "grad_norm": 2.040499687194824, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7755764722824097, + "num_tokens": 622775183.0, + "step": 24377 + }, + { + "epoch": 2.0102251175063905, + "grad_norm": 2.1132099628448486, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7803918719291687, + "num_tokens": 622801448.0, + "step": 24378 + }, + { + "epoch": 2.0103075781314423, + "grad_norm": 2.183058023452759, + "learning_rate": 1e-06, + "loss": 0.5809, + "mean_token_accuracy": 0.7961598634719849, + "num_tokens": 622826638.0, + "step": 24379 + }, + { + "epoch": 2.0103900387564937, + "grad_norm": 2.131924867630005, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7819461822509766, + "num_tokens": 622852369.0, + "step": 24380 + }, + { + "epoch": 2.0104724993815455, + "grad_norm": 2.4916329383850098, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7508592009544373, + "num_tokens": 622876155.0, + "step": 24381 + }, + { + "epoch": 2.010554960006597, + "grad_norm": 1.9287358522415161, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.7860488891601562, + "num_tokens": 622909183.0, + "step": 24382 + }, + { + "epoch": 2.010637420631648, + "grad_norm": 2.0115811824798584, + "learning_rate": 1e-06, + "loss": 0.5836, + "mean_token_accuracy": 0.8032171726226807, + "num_tokens": 622946159.0, + "step": 24383 + }, + { + "epoch": 2.0107198812567, + "grad_norm": 1.9718117713928223, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7824593782424927, + "num_tokens": 622977075.0, + "step": 24384 + }, + { + "epoch": 2.0108023418817513, + "grad_norm": 2.368743419647217, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7416016459465027, + "num_tokens": 623003654.0, + "step": 24385 + }, + { + "epoch": 2.010884802506803, + "grad_norm": 2.066596746444702, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7821179628372192, + "num_tokens": 623035606.0, + "step": 24386 + }, + { + "epoch": 2.0109672631318545, + "grad_norm": 2.252685070037842, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.750792384147644, + "num_tokens": 623060896.0, + "step": 24387 + }, + { + "epoch": 2.0110497237569063, + "grad_norm": 2.2277979850769043, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7449488043785095, + "num_tokens": 623088549.0, + "step": 24388 + }, + { + "epoch": 2.0111321843819576, + "grad_norm": 2.1277267932891846, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7671425342559814, + "num_tokens": 623117981.0, + "step": 24389 + }, + { + "epoch": 2.011214645007009, + "grad_norm": 2.138385772705078, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7414184808731079, + "num_tokens": 623149525.0, + "step": 24390 + }, + { + "epoch": 2.0112971056320608, + "grad_norm": 2.162323236465454, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.764069676399231, + "num_tokens": 623176850.0, + "step": 24391 + }, + { + "epoch": 2.011379566257112, + "grad_norm": 1.9951132535934448, + "learning_rate": 1e-06, + "loss": 0.5923, + "mean_token_accuracy": 0.8053102493286133, + "num_tokens": 623206800.0, + "step": 24392 + }, + { + "epoch": 2.011462026882164, + "grad_norm": 2.1532630920410156, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7571042776107788, + "num_tokens": 623235942.0, + "step": 24393 + }, + { + "epoch": 2.0115444875072153, + "grad_norm": 2.00692081451416, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7328766584396362, + "num_tokens": 623268950.0, + "step": 24394 + }, + { + "epoch": 2.011626948132267, + "grad_norm": 2.0165531635284424, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7779354453086853, + "num_tokens": 623300094.0, + "step": 24395 + }, + { + "epoch": 2.0117094087573184, + "grad_norm": 1.8859426975250244, + "learning_rate": 1e-06, + "loss": 0.6059, + "mean_token_accuracy": 0.8021785020828247, + "num_tokens": 623333350.0, + "step": 24396 + }, + { + "epoch": 2.0117918693823698, + "grad_norm": 2.0567467212677, + "learning_rate": 1e-06, + "loss": 0.5994, + "mean_token_accuracy": 0.79185551404953, + "num_tokens": 623364635.0, + "step": 24397 + }, + { + "epoch": 2.0118743300074216, + "grad_norm": 1.8977540731430054, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7734462022781372, + "num_tokens": 623402034.0, + "step": 24398 + }, + { + "epoch": 2.011956790632473, + "grad_norm": 1.8984626531600952, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7276627421379089, + "num_tokens": 623436477.0, + "step": 24399 + }, + { + "epoch": 2.0120392512575247, + "grad_norm": 2.145932674407959, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.8004815578460693, + "num_tokens": 623464245.0, + "step": 24400 + }, + { + "epoch": 2.012121711882576, + "grad_norm": 2.0375654697418213, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7866263389587402, + "num_tokens": 623493896.0, + "step": 24401 + }, + { + "epoch": 2.0122041725076274, + "grad_norm": 2.2728965282440186, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.75452721118927, + "num_tokens": 623518396.0, + "step": 24402 + }, + { + "epoch": 2.012286633132679, + "grad_norm": 2.085033655166626, + "learning_rate": 1e-06, + "loss": 0.606, + "mean_token_accuracy": 0.7844808101654053, + "num_tokens": 623548902.0, + "step": 24403 + }, + { + "epoch": 2.0123690937577305, + "grad_norm": 2.4411075115203857, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7793703675270081, + "num_tokens": 623572040.0, + "step": 24404 + }, + { + "epoch": 2.0124515543827823, + "grad_norm": 1.9570363759994507, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.7803767919540405, + "num_tokens": 623605360.0, + "step": 24405 + }, + { + "epoch": 2.0125340150078337, + "grad_norm": 2.1767876148223877, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7609007358551025, + "num_tokens": 623634127.0, + "step": 24406 + }, + { + "epoch": 2.0126164756328855, + "grad_norm": 1.9015402793884277, + "learning_rate": 1e-06, + "loss": 0.5577, + "mean_token_accuracy": 0.8083099126815796, + "num_tokens": 623669237.0, + "step": 24407 + }, + { + "epoch": 2.012698936257937, + "grad_norm": 2.4007604122161865, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7478768825531006, + "num_tokens": 623694735.0, + "step": 24408 + }, + { + "epoch": 2.012781396882988, + "grad_norm": 2.1010282039642334, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.7528324723243713, + "num_tokens": 623723581.0, + "step": 24409 + }, + { + "epoch": 2.01286385750804, + "grad_norm": 2.019592046737671, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.7934746742248535, + "num_tokens": 623756907.0, + "step": 24410 + }, + { + "epoch": 2.0129463181330913, + "grad_norm": 2.260934829711914, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7683358192443848, + "num_tokens": 623784017.0, + "step": 24411 + }, + { + "epoch": 2.013028778758143, + "grad_norm": 2.1439969539642334, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7342057228088379, + "num_tokens": 623813778.0, + "step": 24412 + }, + { + "epoch": 2.0131112393831945, + "grad_norm": 2.033010721206665, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7644528150558472, + "num_tokens": 623844109.0, + "step": 24413 + }, + { + "epoch": 2.0131937000082463, + "grad_norm": 1.8587714433670044, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7501981854438782, + "num_tokens": 623878573.0, + "step": 24414 + }, + { + "epoch": 2.0132761606332976, + "grad_norm": 2.092926025390625, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.7870440483093262, + "num_tokens": 623909033.0, + "step": 24415 + }, + { + "epoch": 2.013358621258349, + "grad_norm": 1.9647687673568726, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7453398108482361, + "num_tokens": 623942141.0, + "step": 24416 + }, + { + "epoch": 2.0134410818834008, + "grad_norm": 2.1912267208099365, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7751976251602173, + "num_tokens": 623969918.0, + "step": 24417 + }, + { + "epoch": 2.013523542508452, + "grad_norm": 2.0471808910369873, + "learning_rate": 1e-06, + "loss": 0.5314, + "mean_token_accuracy": 0.817993700504303, + "num_tokens": 623997173.0, + "step": 24418 + }, + { + "epoch": 2.013606003133504, + "grad_norm": 2.099930763244629, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7628331184387207, + "num_tokens": 624024376.0, + "step": 24419 + }, + { + "epoch": 2.0136884637585553, + "grad_norm": 2.1817705631256104, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7555286884307861, + "num_tokens": 624052008.0, + "step": 24420 + }, + { + "epoch": 2.0137709243836066, + "grad_norm": 1.9003276824951172, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7680774927139282, + "num_tokens": 624085588.0, + "step": 24421 + }, + { + "epoch": 2.0138533850086584, + "grad_norm": 2.0514581203460693, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.740016758441925, + "num_tokens": 624116282.0, + "step": 24422 + }, + { + "epoch": 2.0139358456337098, + "grad_norm": 2.1458792686462402, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7622330784797668, + "num_tokens": 624144185.0, + "step": 24423 + }, + { + "epoch": 2.0140183062587615, + "grad_norm": 1.871940016746521, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7521644234657288, + "num_tokens": 624180230.0, + "step": 24424 + }, + { + "epoch": 2.014100766883813, + "grad_norm": 2.3641602993011475, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7646125555038452, + "num_tokens": 624206830.0, + "step": 24425 + }, + { + "epoch": 2.0141832275088647, + "grad_norm": 2.3648667335510254, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.762316107749939, + "num_tokens": 624231200.0, + "step": 24426 + }, + { + "epoch": 2.014265688133916, + "grad_norm": 2.0656166076660156, + "learning_rate": 1e-06, + "loss": 0.5774, + "mean_token_accuracy": 0.8057761788368225, + "num_tokens": 624264916.0, + "step": 24427 + }, + { + "epoch": 2.0143481487589674, + "grad_norm": 2.1623096466064453, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7659748792648315, + "num_tokens": 624292527.0, + "step": 24428 + }, + { + "epoch": 2.014430609384019, + "grad_norm": 2.033543348312378, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.7857189178466797, + "num_tokens": 624322139.0, + "step": 24429 + }, + { + "epoch": 2.0145130700090705, + "grad_norm": 1.820413589477539, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7577885389328003, + "num_tokens": 624357161.0, + "step": 24430 + }, + { + "epoch": 2.0145955306341223, + "grad_norm": 1.999085545539856, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7619918584823608, + "num_tokens": 624387815.0, + "step": 24431 + }, + { + "epoch": 2.0146779912591737, + "grad_norm": 2.629401206970215, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7653899788856506, + "num_tokens": 624407823.0, + "step": 24432 + }, + { + "epoch": 2.0147604518842255, + "grad_norm": 2.0394363403320312, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7880565524101257, + "num_tokens": 624438028.0, + "step": 24433 + }, + { + "epoch": 2.014842912509277, + "grad_norm": 2.160207986831665, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.787003755569458, + "num_tokens": 624463512.0, + "step": 24434 + }, + { + "epoch": 2.014925373134328, + "grad_norm": 2.057816982269287, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7735908031463623, + "num_tokens": 624495593.0, + "step": 24435 + }, + { + "epoch": 2.01500783375938, + "grad_norm": 2.13619327545166, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7563246488571167, + "num_tokens": 624524785.0, + "step": 24436 + }, + { + "epoch": 2.0150902943844313, + "grad_norm": 2.2953712940216064, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.7768505811691284, + "num_tokens": 624550736.0, + "step": 24437 + }, + { + "epoch": 2.015172755009483, + "grad_norm": 2.0417439937591553, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.7947403192520142, + "num_tokens": 624579282.0, + "step": 24438 + }, + { + "epoch": 2.0152552156345345, + "grad_norm": 1.9272775650024414, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.7875509858131409, + "num_tokens": 624611459.0, + "step": 24439 + }, + { + "epoch": 2.0153376762595863, + "grad_norm": 1.921173334121704, + "learning_rate": 1e-06, + "loss": 0.5629, + "mean_token_accuracy": 0.8129051923751831, + "num_tokens": 624641397.0, + "step": 24440 + }, + { + "epoch": 2.0154201368846376, + "grad_norm": 1.7434183359146118, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7823489904403687, + "num_tokens": 624678620.0, + "step": 24441 + }, + { + "epoch": 2.015502597509689, + "grad_norm": 1.9064749479293823, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7754857540130615, + "num_tokens": 624711280.0, + "step": 24442 + }, + { + "epoch": 2.0155850581347408, + "grad_norm": 2.085279703140259, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7742521166801453, + "num_tokens": 624740814.0, + "step": 24443 + }, + { + "epoch": 2.015667518759792, + "grad_norm": 2.118173122406006, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7755237817764282, + "num_tokens": 624770575.0, + "step": 24444 + }, + { + "epoch": 2.015749979384844, + "grad_norm": 2.1335301399230957, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.7844297885894775, + "num_tokens": 624800866.0, + "step": 24445 + }, + { + "epoch": 2.0158324400098953, + "grad_norm": 2.4330599308013916, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7439365386962891, + "num_tokens": 624825739.0, + "step": 24446 + }, + { + "epoch": 2.0159149006349466, + "grad_norm": 1.8100264072418213, + "learning_rate": 1e-06, + "loss": 0.5752, + "mean_token_accuracy": 0.8102341294288635, + "num_tokens": 624859595.0, + "step": 24447 + }, + { + "epoch": 2.0159973612599984, + "grad_norm": 1.8728535175323486, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7782920598983765, + "num_tokens": 624895369.0, + "step": 24448 + }, + { + "epoch": 2.0160798218850497, + "grad_norm": 2.2211391925811768, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.774135947227478, + "num_tokens": 624922472.0, + "step": 24449 + }, + { + "epoch": 2.0161622825101015, + "grad_norm": 2.0944149494171143, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7664523124694824, + "num_tokens": 624951392.0, + "step": 24450 + }, + { + "epoch": 2.016244743135153, + "grad_norm": 2.2312607765197754, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7549110054969788, + "num_tokens": 624979702.0, + "step": 24451 + }, + { + "epoch": 2.0163272037602047, + "grad_norm": 1.9869500398635864, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7886086106300354, + "num_tokens": 625011584.0, + "step": 24452 + }, + { + "epoch": 2.016409664385256, + "grad_norm": 1.911913275718689, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7651888728141785, + "num_tokens": 625049180.0, + "step": 24453 + }, + { + "epoch": 2.0164921250103074, + "grad_norm": 2.1973319053649902, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7771227359771729, + "num_tokens": 625077568.0, + "step": 24454 + }, + { + "epoch": 2.016574585635359, + "grad_norm": 2.080127000808716, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.7890686392784119, + "num_tokens": 625106229.0, + "step": 24455 + }, + { + "epoch": 2.0166570462604105, + "grad_norm": 2.038194179534912, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.7760459184646606, + "num_tokens": 625134078.0, + "step": 24456 + }, + { + "epoch": 2.0167395068854623, + "grad_norm": 2.3321547508239746, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.7611701488494873, + "num_tokens": 625160018.0, + "step": 24457 + }, + { + "epoch": 2.0168219675105137, + "grad_norm": 2.2277400493621826, + "learning_rate": 1e-06, + "loss": 0.4784, + "mean_token_accuracy": 0.8359967470169067, + "num_tokens": 625185273.0, + "step": 24458 + }, + { + "epoch": 2.0169044281355655, + "grad_norm": 2.2377521991729736, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7677500247955322, + "num_tokens": 625213253.0, + "step": 24459 + }, + { + "epoch": 2.016986888760617, + "grad_norm": 2.249098062515259, + "learning_rate": 1e-06, + "loss": 0.5547, + "mean_token_accuracy": 0.79441237449646, + "num_tokens": 625238446.0, + "step": 24460 + }, + { + "epoch": 2.017069349385668, + "grad_norm": 1.9365918636322021, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.7931010723114014, + "num_tokens": 625269326.0, + "step": 24461 + }, + { + "epoch": 2.01715181001072, + "grad_norm": 1.877923607826233, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7851998805999756, + "num_tokens": 625303106.0, + "step": 24462 + }, + { + "epoch": 2.0172342706357713, + "grad_norm": 2.2668344974517822, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7132823467254639, + "num_tokens": 625331688.0, + "step": 24463 + }, + { + "epoch": 2.017316731260823, + "grad_norm": 1.8124322891235352, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7422686815261841, + "num_tokens": 625370962.0, + "step": 24464 + }, + { + "epoch": 2.0173991918858745, + "grad_norm": 1.9978739023208618, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.7961032390594482, + "num_tokens": 625402398.0, + "step": 24465 + }, + { + "epoch": 2.017481652510926, + "grad_norm": 2.146669626235962, + "learning_rate": 1e-06, + "loss": 0.6124, + "mean_token_accuracy": 0.7799996137619019, + "num_tokens": 625428948.0, + "step": 24466 + }, + { + "epoch": 2.0175641131359776, + "grad_norm": 2.318375825881958, + "learning_rate": 1e-06, + "loss": 0.5958, + "mean_token_accuracy": 0.7990254163742065, + "num_tokens": 625451340.0, + "step": 24467 + }, + { + "epoch": 2.017646573761029, + "grad_norm": 2.0396361351013184, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7539702653884888, + "num_tokens": 625485105.0, + "step": 24468 + }, + { + "epoch": 2.0177290343860808, + "grad_norm": 2.0327606201171875, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7522107362747192, + "num_tokens": 625518430.0, + "step": 24469 + }, + { + "epoch": 2.017811495011132, + "grad_norm": 2.4435975551605225, + "learning_rate": 1e-06, + "loss": 0.6116, + "mean_token_accuracy": 0.7857736945152283, + "num_tokens": 625540802.0, + "step": 24470 + }, + { + "epoch": 2.017893955636184, + "grad_norm": 2.061863422393799, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.7860888242721558, + "num_tokens": 625572841.0, + "step": 24471 + }, + { + "epoch": 2.0179764162612353, + "grad_norm": 1.9462108612060547, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.7712863087654114, + "num_tokens": 625604560.0, + "step": 24472 + }, + { + "epoch": 2.0180588768862866, + "grad_norm": 2.046760320663452, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.7762048244476318, + "num_tokens": 625633915.0, + "step": 24473 + }, + { + "epoch": 2.0181413375113384, + "grad_norm": 2.270965099334717, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.7573768496513367, + "num_tokens": 625657666.0, + "step": 24474 + }, + { + "epoch": 2.0182237981363897, + "grad_norm": 2.0188283920288086, + "learning_rate": 1e-06, + "loss": 0.5753, + "mean_token_accuracy": 0.8082688450813293, + "num_tokens": 625686099.0, + "step": 24475 + }, + { + "epoch": 2.0183062587614415, + "grad_norm": 2.269505500793457, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7620977163314819, + "num_tokens": 625711267.0, + "step": 24476 + }, + { + "epoch": 2.018388719386493, + "grad_norm": 2.0844480991363525, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.7989000082015991, + "num_tokens": 625738006.0, + "step": 24477 + }, + { + "epoch": 2.0184711800115447, + "grad_norm": 2.3346683979034424, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7683121562004089, + "num_tokens": 625760009.0, + "step": 24478 + }, + { + "epoch": 2.018553640636596, + "grad_norm": 1.850667953491211, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7717164754867554, + "num_tokens": 625794849.0, + "step": 24479 + }, + { + "epoch": 2.0186361012616474, + "grad_norm": 1.9283827543258667, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.7877415418624878, + "num_tokens": 625829861.0, + "step": 24480 + }, + { + "epoch": 2.018718561886699, + "grad_norm": 2.2095565795898438, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7323797941207886, + "num_tokens": 625856903.0, + "step": 24481 + }, + { + "epoch": 2.0188010225117505, + "grad_norm": 1.8599238395690918, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.7851831316947937, + "num_tokens": 625891796.0, + "step": 24482 + }, + { + "epoch": 2.0188834831368023, + "grad_norm": 1.9524849653244019, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7542135715484619, + "num_tokens": 625925609.0, + "step": 24483 + }, + { + "epoch": 2.0189659437618537, + "grad_norm": 2.1066272258758545, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7760937809944153, + "num_tokens": 625955555.0, + "step": 24484 + }, + { + "epoch": 2.0190484043869055, + "grad_norm": 1.9647884368896484, + "learning_rate": 1e-06, + "loss": 0.5826, + "mean_token_accuracy": 0.8091332912445068, + "num_tokens": 625986672.0, + "step": 24485 + }, + { + "epoch": 2.019130865011957, + "grad_norm": 2.131810188293457, + "learning_rate": 1e-06, + "loss": 0.5492, + "mean_token_accuracy": 0.8183560371398926, + "num_tokens": 626012371.0, + "step": 24486 + }, + { + "epoch": 2.019213325637008, + "grad_norm": 1.9224188327789307, + "learning_rate": 1e-06, + "loss": 0.5933, + "mean_token_accuracy": 0.8014801740646362, + "num_tokens": 626045917.0, + "step": 24487 + }, + { + "epoch": 2.01929578626206, + "grad_norm": 2.08337140083313, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7311111688613892, + "num_tokens": 626076767.0, + "step": 24488 + }, + { + "epoch": 2.0193782468871113, + "grad_norm": 2.1391854286193848, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.781882107257843, + "num_tokens": 626104706.0, + "step": 24489 + }, + { + "epoch": 2.019460707512163, + "grad_norm": 2.223539113998413, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7800998687744141, + "num_tokens": 626127857.0, + "step": 24490 + }, + { + "epoch": 2.0195431681372145, + "grad_norm": 2.1360480785369873, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7923235893249512, + "num_tokens": 626155925.0, + "step": 24491 + }, + { + "epoch": 2.019625628762266, + "grad_norm": 1.9819893836975098, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.7822890281677246, + "num_tokens": 626187045.0, + "step": 24492 + }, + { + "epoch": 2.0197080893873176, + "grad_norm": 2.0000834465026855, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7541995048522949, + "num_tokens": 626218359.0, + "step": 24493 + }, + { + "epoch": 2.019790550012369, + "grad_norm": 2.0964341163635254, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7432032823562622, + "num_tokens": 626249462.0, + "step": 24494 + }, + { + "epoch": 2.0198730106374208, + "grad_norm": 2.2989726066589355, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7463998794555664, + "num_tokens": 626276809.0, + "step": 24495 + }, + { + "epoch": 2.019955471262472, + "grad_norm": 1.9025726318359375, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7503224611282349, + "num_tokens": 626314782.0, + "step": 24496 + }, + { + "epoch": 2.020037931887524, + "grad_norm": 1.8154220581054688, + "learning_rate": 1e-06, + "loss": 0.5676, + "mean_token_accuracy": 0.8007398843765259, + "num_tokens": 626349372.0, + "step": 24497 + }, + { + "epoch": 2.0201203925125752, + "grad_norm": 2.079901933670044, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7427456378936768, + "num_tokens": 626381771.0, + "step": 24498 + }, + { + "epoch": 2.0202028531376266, + "grad_norm": 2.1268937587738037, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7636611461639404, + "num_tokens": 626412234.0, + "step": 24499 + }, + { + "epoch": 2.0202853137626784, + "grad_norm": 2.0356602668762207, + "learning_rate": 1e-06, + "loss": 0.5771, + "mean_token_accuracy": 0.8084226250648499, + "num_tokens": 626445362.0, + "step": 24500 + }, + { + "epoch": 2.0203677743877297, + "grad_norm": 2.017728090286255, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.763933002948761, + "num_tokens": 626477502.0, + "step": 24501 + }, + { + "epoch": 2.0204502350127815, + "grad_norm": 2.1651692390441895, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7653011679649353, + "num_tokens": 626506113.0, + "step": 24502 + }, + { + "epoch": 2.020532695637833, + "grad_norm": 1.9792813062667847, + "learning_rate": 1e-06, + "loss": 0.504, + "mean_token_accuracy": 0.8225945234298706, + "num_tokens": 626536178.0, + "step": 24503 + }, + { + "epoch": 2.0206151562628847, + "grad_norm": 1.899263620376587, + "learning_rate": 1e-06, + "loss": 0.7482, + "mean_token_accuracy": 0.7512075901031494, + "num_tokens": 626573038.0, + "step": 24504 + }, + { + "epoch": 2.020697616887936, + "grad_norm": 2.0979549884796143, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7765766382217407, + "num_tokens": 626602737.0, + "step": 24505 + }, + { + "epoch": 2.0207800775129874, + "grad_norm": 2.0350773334503174, + "learning_rate": 1e-06, + "loss": 0.5805, + "mean_token_accuracy": 0.805942177772522, + "num_tokens": 626633845.0, + "step": 24506 + }, + { + "epoch": 2.020862538138039, + "grad_norm": 1.8334510326385498, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.7935923337936401, + "num_tokens": 626666275.0, + "step": 24507 + }, + { + "epoch": 2.0209449987630905, + "grad_norm": 2.0266268253326416, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7849215269088745, + "num_tokens": 626695982.0, + "step": 24508 + }, + { + "epoch": 2.0210274593881423, + "grad_norm": 2.181349754333496, + "learning_rate": 1e-06, + "loss": 0.5741, + "mean_token_accuracy": 0.8079280257225037, + "num_tokens": 626723100.0, + "step": 24509 + }, + { + "epoch": 2.0211099200131937, + "grad_norm": 2.3259434700012207, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.75701504945755, + "num_tokens": 626748383.0, + "step": 24510 + }, + { + "epoch": 2.021192380638245, + "grad_norm": 2.188117742538452, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7556045651435852, + "num_tokens": 626775635.0, + "step": 24511 + }, + { + "epoch": 2.021274841263297, + "grad_norm": 2.238205909729004, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7816436290740967, + "num_tokens": 626801747.0, + "step": 24512 + }, + { + "epoch": 2.021357301888348, + "grad_norm": 1.974321722984314, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.7922960519790649, + "num_tokens": 626834149.0, + "step": 24513 + }, + { + "epoch": 2.0214397625134, + "grad_norm": 1.97294020652771, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.7883574962615967, + "num_tokens": 626865778.0, + "step": 24514 + }, + { + "epoch": 2.0215222231384513, + "grad_norm": 2.0617287158966064, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.7968014478683472, + "num_tokens": 626896118.0, + "step": 24515 + }, + { + "epoch": 2.021604683763503, + "grad_norm": 1.810541033744812, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7721061706542969, + "num_tokens": 626931442.0, + "step": 24516 + }, + { + "epoch": 2.0216871443885545, + "grad_norm": 2.0199198722839355, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7739871740341187, + "num_tokens": 626962487.0, + "step": 24517 + }, + { + "epoch": 2.021769605013606, + "grad_norm": 2.3645806312561035, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.774366021156311, + "num_tokens": 626984695.0, + "step": 24518 + }, + { + "epoch": 2.0218520656386576, + "grad_norm": 2.2382166385650635, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7344783544540405, + "num_tokens": 627014038.0, + "step": 24519 + }, + { + "epoch": 2.021934526263709, + "grad_norm": 1.8766427040100098, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7733569145202637, + "num_tokens": 627047168.0, + "step": 24520 + }, + { + "epoch": 2.0220169868887607, + "grad_norm": 1.9186981916427612, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7737237215042114, + "num_tokens": 627081520.0, + "step": 24521 + }, + { + "epoch": 2.022099447513812, + "grad_norm": 2.1003105640411377, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.7981318831443787, + "num_tokens": 627111213.0, + "step": 24522 + }, + { + "epoch": 2.022181908138864, + "grad_norm": 2.1639490127563477, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7660274505615234, + "num_tokens": 627142243.0, + "step": 24523 + }, + { + "epoch": 2.0222643687639152, + "grad_norm": 1.8844647407531738, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.7946704030036926, + "num_tokens": 627173009.0, + "step": 24524 + }, + { + "epoch": 2.0223468293889666, + "grad_norm": 2.176517963409424, + "learning_rate": 1e-06, + "loss": 0.5999, + "mean_token_accuracy": 0.7898409962654114, + "num_tokens": 627203243.0, + "step": 24525 + }, + { + "epoch": 2.0224292900140184, + "grad_norm": 1.8304548263549805, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7895635366439819, + "num_tokens": 627241729.0, + "step": 24526 + }, + { + "epoch": 2.0225117506390697, + "grad_norm": 1.906482219696045, + "learning_rate": 1e-06, + "loss": 0.6079, + "mean_token_accuracy": 0.7978753447532654, + "num_tokens": 627274133.0, + "step": 24527 + }, + { + "epoch": 2.0225942112641215, + "grad_norm": 1.9522322416305542, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.7554948925971985, + "num_tokens": 627310968.0, + "step": 24528 + }, + { + "epoch": 2.022676671889173, + "grad_norm": 2.495046377182007, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7424256801605225, + "num_tokens": 627334590.0, + "step": 24529 + }, + { + "epoch": 2.0227591325142247, + "grad_norm": 2.0529723167419434, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7740806937217712, + "num_tokens": 627364014.0, + "step": 24530 + }, + { + "epoch": 2.022841593139276, + "grad_norm": 2.0327341556549072, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.7848498821258545, + "num_tokens": 627398111.0, + "step": 24531 + }, + { + "epoch": 2.0229240537643274, + "grad_norm": 2.094067096710205, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7757269144058228, + "num_tokens": 627429818.0, + "step": 24532 + }, + { + "epoch": 2.023006514389379, + "grad_norm": 2.211963653564453, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7687758207321167, + "num_tokens": 627458722.0, + "step": 24533 + }, + { + "epoch": 2.0230889750144305, + "grad_norm": 1.81895911693573, + "learning_rate": 1e-06, + "loss": 0.4752, + "mean_token_accuracy": 0.8305371999740601, + "num_tokens": 627490442.0, + "step": 24534 + }, + { + "epoch": 2.0231714356394823, + "grad_norm": 2.4881036281585693, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7503454089164734, + "num_tokens": 627514396.0, + "step": 24535 + }, + { + "epoch": 2.0232538962645337, + "grad_norm": 1.9851843118667603, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7723889350891113, + "num_tokens": 627544840.0, + "step": 24536 + }, + { + "epoch": 2.023336356889585, + "grad_norm": 1.8779020309448242, + "learning_rate": 1e-06, + "loss": 0.5637, + "mean_token_accuracy": 0.8105969429016113, + "num_tokens": 627581252.0, + "step": 24537 + }, + { + "epoch": 2.023418817514637, + "grad_norm": 2.240412473678589, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7472736835479736, + "num_tokens": 627609141.0, + "step": 24538 + }, + { + "epoch": 2.023501278139688, + "grad_norm": 1.8261613845825195, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.7880974411964417, + "num_tokens": 627647994.0, + "step": 24539 + }, + { + "epoch": 2.02358373876474, + "grad_norm": 2.2442867755889893, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8021239042282104, + "num_tokens": 627673250.0, + "step": 24540 + }, + { + "epoch": 2.0236661993897913, + "grad_norm": 1.8790531158447266, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7711475491523743, + "num_tokens": 627713522.0, + "step": 24541 + }, + { + "epoch": 2.023748660014843, + "grad_norm": 1.8510284423828125, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.789336621761322, + "num_tokens": 627747926.0, + "step": 24542 + }, + { + "epoch": 2.0238311206398945, + "grad_norm": 2.088005781173706, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7762724161148071, + "num_tokens": 627778707.0, + "step": 24543 + }, + { + "epoch": 2.023913581264946, + "grad_norm": 2.0668210983276367, + "learning_rate": 1e-06, + "loss": 0.5725, + "mean_token_accuracy": 0.8021539449691772, + "num_tokens": 627806951.0, + "step": 24544 + }, + { + "epoch": 2.0239960418899976, + "grad_norm": 2.1493332386016846, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.789728581905365, + "num_tokens": 627835210.0, + "step": 24545 + }, + { + "epoch": 2.024078502515049, + "grad_norm": 1.838477611541748, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7622777223587036, + "num_tokens": 627871378.0, + "step": 24546 + }, + { + "epoch": 2.0241609631401007, + "grad_norm": 2.2763490676879883, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7627828121185303, + "num_tokens": 627898471.0, + "step": 24547 + }, + { + "epoch": 2.024243423765152, + "grad_norm": 2.0851118564605713, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7859130501747131, + "num_tokens": 627929545.0, + "step": 24548 + }, + { + "epoch": 2.024325884390204, + "grad_norm": 2.312180519104004, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7423515319824219, + "num_tokens": 627954870.0, + "step": 24549 + }, + { + "epoch": 2.0244083450152552, + "grad_norm": 1.8578733205795288, + "learning_rate": 1e-06, + "loss": 0.5207, + "mean_token_accuracy": 0.8297869563102722, + "num_tokens": 627986786.0, + "step": 24550 + }, + { + "epoch": 2.0244908056403066, + "grad_norm": 2.0068068504333496, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7618098258972168, + "num_tokens": 628020003.0, + "step": 24551 + }, + { + "epoch": 2.0245732662653584, + "grad_norm": 1.7442693710327148, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.7904732823371887, + "num_tokens": 628061582.0, + "step": 24552 + }, + { + "epoch": 2.0246557268904097, + "grad_norm": 2.0209121704101562, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.7908211946487427, + "num_tokens": 628094081.0, + "step": 24553 + }, + { + "epoch": 2.0247381875154615, + "grad_norm": 2.2936341762542725, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7475630044937134, + "num_tokens": 628121543.0, + "step": 24554 + }, + { + "epoch": 2.024820648140513, + "grad_norm": 2.1060078144073486, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7590703964233398, + "num_tokens": 628152544.0, + "step": 24555 + }, + { + "epoch": 2.0249031087655642, + "grad_norm": 1.9876790046691895, + "learning_rate": 1e-06, + "loss": 0.5677, + "mean_token_accuracy": 0.8140689134597778, + "num_tokens": 628183237.0, + "step": 24556 + }, + { + "epoch": 2.024985569390616, + "grad_norm": 1.9299710988998413, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7577623724937439, + "num_tokens": 628218413.0, + "step": 24557 + }, + { + "epoch": 2.0250680300156674, + "grad_norm": 2.3424181938171387, + "learning_rate": 1e-06, + "loss": 0.5895, + "mean_token_accuracy": 0.8052244782447815, + "num_tokens": 628242344.0, + "step": 24558 + }, + { + "epoch": 2.025150490640719, + "grad_norm": 2.0775458812713623, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7642104625701904, + "num_tokens": 628273698.0, + "step": 24559 + }, + { + "epoch": 2.0252329512657705, + "grad_norm": 2.214508056640625, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.7869188785552979, + "num_tokens": 628299460.0, + "step": 24560 + }, + { + "epoch": 2.0253154118908223, + "grad_norm": 1.974924087524414, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.7916746139526367, + "num_tokens": 628333003.0, + "step": 24561 + }, + { + "epoch": 2.0253978725158737, + "grad_norm": 1.967450499534607, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.7785636782646179, + "num_tokens": 628366474.0, + "step": 24562 + }, + { + "epoch": 2.025480333140925, + "grad_norm": 2.1409733295440674, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7674028873443604, + "num_tokens": 628393739.0, + "step": 24563 + }, + { + "epoch": 2.025562793765977, + "grad_norm": 2.000913619995117, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.7829701900482178, + "num_tokens": 628425034.0, + "step": 24564 + }, + { + "epoch": 2.025645254391028, + "grad_norm": 2.175105094909668, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.7924436330795288, + "num_tokens": 628451790.0, + "step": 24565 + }, + { + "epoch": 2.02572771501608, + "grad_norm": 1.9487119913101196, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7542755603790283, + "num_tokens": 628483672.0, + "step": 24566 + }, + { + "epoch": 2.0258101756411313, + "grad_norm": 1.9570698738098145, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7368581295013428, + "num_tokens": 628517787.0, + "step": 24567 + }, + { + "epoch": 2.025892636266183, + "grad_norm": 2.0410377979278564, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.780139148235321, + "num_tokens": 628547724.0, + "step": 24568 + }, + { + "epoch": 2.0259750968912345, + "grad_norm": 2.0114216804504395, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7679739594459534, + "num_tokens": 628582196.0, + "step": 24569 + }, + { + "epoch": 2.026057557516286, + "grad_norm": 1.9300458431243896, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7617998123168945, + "num_tokens": 628619542.0, + "step": 24570 + }, + { + "epoch": 2.0261400181413376, + "grad_norm": 2.0218443870544434, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.7798740863800049, + "num_tokens": 628651205.0, + "step": 24571 + }, + { + "epoch": 2.026222478766389, + "grad_norm": 2.2099151611328125, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7392871379852295, + "num_tokens": 628683596.0, + "step": 24572 + }, + { + "epoch": 2.0263049393914407, + "grad_norm": 2.060171604156494, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7532392740249634, + "num_tokens": 628714966.0, + "step": 24573 + }, + { + "epoch": 2.026387400016492, + "grad_norm": 2.206331491470337, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7305256724357605, + "num_tokens": 628743362.0, + "step": 24574 + }, + { + "epoch": 2.0264698606415434, + "grad_norm": 2.1220037937164307, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7682862877845764, + "num_tokens": 628770677.0, + "step": 24575 + }, + { + "epoch": 2.0265523212665952, + "grad_norm": 2.026240348815918, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7821304798126221, + "num_tokens": 628802051.0, + "step": 24576 + }, + { + "epoch": 2.0266347818916466, + "grad_norm": 2.126218795776367, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7672924399375916, + "num_tokens": 628830368.0, + "step": 24577 + }, + { + "epoch": 2.0267172425166984, + "grad_norm": 1.9466378688812256, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.7806186079978943, + "num_tokens": 628866648.0, + "step": 24578 + }, + { + "epoch": 2.0267997031417497, + "grad_norm": 2.1740660667419434, + "learning_rate": 1e-06, + "loss": 0.618, + "mean_token_accuracy": 0.7951930165290833, + "num_tokens": 628895591.0, + "step": 24579 + }, + { + "epoch": 2.0268821637668015, + "grad_norm": 2.096383571624756, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7776525616645813, + "num_tokens": 628924733.0, + "step": 24580 + }, + { + "epoch": 2.026964624391853, + "grad_norm": 2.01116681098938, + "learning_rate": 1e-06, + "loss": 0.5938, + "mean_token_accuracy": 0.8001707196235657, + "num_tokens": 628954029.0, + "step": 24581 + }, + { + "epoch": 2.0270470850169042, + "grad_norm": 1.828851342201233, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.7829654216766357, + "num_tokens": 628988816.0, + "step": 24582 + }, + { + "epoch": 2.027129545641956, + "grad_norm": 2.000215768814087, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7670447826385498, + "num_tokens": 629023519.0, + "step": 24583 + }, + { + "epoch": 2.0272120062670074, + "grad_norm": 2.2848806381225586, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7746180891990662, + "num_tokens": 629049355.0, + "step": 24584 + }, + { + "epoch": 2.027294466892059, + "grad_norm": 2.178694248199463, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7701576948165894, + "num_tokens": 629080495.0, + "step": 24585 + }, + { + "epoch": 2.0273769275171105, + "grad_norm": 2.2621564865112305, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7833713293075562, + "num_tokens": 629108635.0, + "step": 24586 + }, + { + "epoch": 2.0274593881421623, + "grad_norm": 1.8472557067871094, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.8071819543838501, + "num_tokens": 629143670.0, + "step": 24587 + }, + { + "epoch": 2.0275418487672137, + "grad_norm": 1.8590710163116455, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.7855805158615112, + "num_tokens": 629176455.0, + "step": 24588 + }, + { + "epoch": 2.027624309392265, + "grad_norm": 2.071040391921997, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7821972370147705, + "num_tokens": 629204534.0, + "step": 24589 + }, + { + "epoch": 2.027706770017317, + "grad_norm": 2.1764869689941406, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7551954984664917, + "num_tokens": 629231383.0, + "step": 24590 + }, + { + "epoch": 2.027789230642368, + "grad_norm": 2.1915767192840576, + "learning_rate": 1e-06, + "loss": 0.5594, + "mean_token_accuracy": 0.8110512495040894, + "num_tokens": 629258697.0, + "step": 24591 + }, + { + "epoch": 2.02787169126742, + "grad_norm": 2.5512750148773193, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.7832188606262207, + "num_tokens": 629280477.0, + "step": 24592 + }, + { + "epoch": 2.0279541518924713, + "grad_norm": 2.2034385204315186, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7544409036636353, + "num_tokens": 629309341.0, + "step": 24593 + }, + { + "epoch": 2.028036612517523, + "grad_norm": 1.985795021057129, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.7942278385162354, + "num_tokens": 629344598.0, + "step": 24594 + }, + { + "epoch": 2.0281190731425744, + "grad_norm": 2.0355513095855713, + "learning_rate": 1e-06, + "loss": 0.5798, + "mean_token_accuracy": 0.8030864596366882, + "num_tokens": 629376251.0, + "step": 24595 + }, + { + "epoch": 2.028201533767626, + "grad_norm": 2.2216014862060547, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7163200974464417, + "num_tokens": 629406036.0, + "step": 24596 + }, + { + "epoch": 2.0282839943926776, + "grad_norm": 2.2996010780334473, + "learning_rate": 1e-06, + "loss": 0.5545, + "mean_token_accuracy": 0.7963366508483887, + "num_tokens": 629430308.0, + "step": 24597 + }, + { + "epoch": 2.028366455017729, + "grad_norm": 2.0473477840423584, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7767009139060974, + "num_tokens": 629462897.0, + "step": 24598 + }, + { + "epoch": 2.0284489156427807, + "grad_norm": 1.8805572986602783, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7724447846412659, + "num_tokens": 629497719.0, + "step": 24599 + }, + { + "epoch": 2.028531376267832, + "grad_norm": 1.7883641719818115, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7581202387809753, + "num_tokens": 629538682.0, + "step": 24600 + }, + { + "epoch": 2.0286138368928834, + "grad_norm": 1.7708964347839355, + "learning_rate": 1e-06, + "loss": 0.4813, + "mean_token_accuracy": 0.8286683559417725, + "num_tokens": 629571576.0, + "step": 24601 + }, + { + "epoch": 2.0286962975179352, + "grad_norm": 1.9282861948013306, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7676680684089661, + "num_tokens": 629606641.0, + "step": 24602 + }, + { + "epoch": 2.0287787581429866, + "grad_norm": 1.9618855714797974, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.7789242267608643, + "num_tokens": 629636876.0, + "step": 24603 + }, + { + "epoch": 2.0288612187680384, + "grad_norm": 2.3434481620788574, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7555338144302368, + "num_tokens": 629660639.0, + "step": 24604 + }, + { + "epoch": 2.0289436793930897, + "grad_norm": 2.143411874771118, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7619468569755554, + "num_tokens": 629689779.0, + "step": 24605 + }, + { + "epoch": 2.0290261400181415, + "grad_norm": 2.8590247631073, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7520900964736938, + "num_tokens": 629711952.0, + "step": 24606 + }, + { + "epoch": 2.029108600643193, + "grad_norm": 2.2274274826049805, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.79030841588974, + "num_tokens": 629736940.0, + "step": 24607 + }, + { + "epoch": 2.0291910612682442, + "grad_norm": 2.1367902755737305, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7159343361854553, + "num_tokens": 629765996.0, + "step": 24608 + }, + { + "epoch": 2.029273521893296, + "grad_norm": 2.010767698287964, + "learning_rate": 1e-06, + "loss": 0.602, + "mean_token_accuracy": 0.8030562400817871, + "num_tokens": 629797052.0, + "step": 24609 + }, + { + "epoch": 2.0293559825183474, + "grad_norm": 2.3121578693389893, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7293230295181274, + "num_tokens": 629822526.0, + "step": 24610 + }, + { + "epoch": 2.029438443143399, + "grad_norm": 2.20000958442688, + "learning_rate": 1e-06, + "loss": 0.5822, + "mean_token_accuracy": 0.8042017221450806, + "num_tokens": 629848820.0, + "step": 24611 + }, + { + "epoch": 2.0295209037684505, + "grad_norm": 2.0075643062591553, + "learning_rate": 1e-06, + "loss": 0.5799, + "mean_token_accuracy": 0.8041558861732483, + "num_tokens": 629884168.0, + "step": 24612 + }, + { + "epoch": 2.0296033643935023, + "grad_norm": 2.103973627090454, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7735419273376465, + "num_tokens": 629916059.0, + "step": 24613 + }, + { + "epoch": 2.0296858250185537, + "grad_norm": 2.4364466667175293, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.7727069854736328, + "num_tokens": 629938597.0, + "step": 24614 + }, + { + "epoch": 2.029768285643605, + "grad_norm": 2.222752809524536, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7826282382011414, + "num_tokens": 629962358.0, + "step": 24615 + }, + { + "epoch": 2.029850746268657, + "grad_norm": 2.301595687866211, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.737188458442688, + "num_tokens": 629989893.0, + "step": 24616 + }, + { + "epoch": 2.029933206893708, + "grad_norm": 2.0717291831970215, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.7881203293800354, + "num_tokens": 630019004.0, + "step": 24617 + }, + { + "epoch": 2.03001566751876, + "grad_norm": 2.057006597518921, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7722880244255066, + "num_tokens": 630048749.0, + "step": 24618 + }, + { + "epoch": 2.0300981281438113, + "grad_norm": 2.186058282852173, + "learning_rate": 1e-06, + "loss": 0.5956, + "mean_token_accuracy": 0.7954437136650085, + "num_tokens": 630076518.0, + "step": 24619 + }, + { + "epoch": 2.0301805887688626, + "grad_norm": 2.072599411010742, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.7950074672698975, + "num_tokens": 630105778.0, + "step": 24620 + }, + { + "epoch": 2.0302630493939144, + "grad_norm": 1.950000286102295, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7880255579948425, + "num_tokens": 630138548.0, + "step": 24621 + }, + { + "epoch": 2.030345510018966, + "grad_norm": 2.353351593017578, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7765763998031616, + "num_tokens": 630163796.0, + "step": 24622 + }, + { + "epoch": 2.0304279706440176, + "grad_norm": 1.885412573814392, + "learning_rate": 1e-06, + "loss": 0.6029, + "mean_token_accuracy": 0.8000498414039612, + "num_tokens": 630196266.0, + "step": 24623 + }, + { + "epoch": 2.030510431269069, + "grad_norm": 2.271259069442749, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.7784558534622192, + "num_tokens": 630222068.0, + "step": 24624 + }, + { + "epoch": 2.0305928918941207, + "grad_norm": 2.0693929195404053, + "learning_rate": 1e-06, + "loss": 0.5857, + "mean_token_accuracy": 0.7959866523742676, + "num_tokens": 630251989.0, + "step": 24625 + }, + { + "epoch": 2.030675352519172, + "grad_norm": 1.8075642585754395, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7714347839355469, + "num_tokens": 630290972.0, + "step": 24626 + }, + { + "epoch": 2.0307578131442234, + "grad_norm": 2.1931979656219482, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7369682788848877, + "num_tokens": 630319917.0, + "step": 24627 + }, + { + "epoch": 2.0308402737692752, + "grad_norm": 2.024538516998291, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.7940529584884644, + "num_tokens": 630347173.0, + "step": 24628 + }, + { + "epoch": 2.0309227343943266, + "grad_norm": 1.9681519269943237, + "learning_rate": 1e-06, + "loss": 0.5873, + "mean_token_accuracy": 0.7971198558807373, + "num_tokens": 630375649.0, + "step": 24629 + }, + { + "epoch": 2.0310051950193784, + "grad_norm": 1.9892117977142334, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7506504058837891, + "num_tokens": 630405381.0, + "step": 24630 + }, + { + "epoch": 2.0310876556444297, + "grad_norm": 1.8751243352890015, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7458816766738892, + "num_tokens": 630441260.0, + "step": 24631 + }, + { + "epoch": 2.0311701162694815, + "grad_norm": 2.0334279537200928, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7816970944404602, + "num_tokens": 630470691.0, + "step": 24632 + }, + { + "epoch": 2.031252576894533, + "grad_norm": 2.130545139312744, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.7785081267356873, + "num_tokens": 630503407.0, + "step": 24633 + }, + { + "epoch": 2.031335037519584, + "grad_norm": 1.8762025833129883, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.791959285736084, + "num_tokens": 630537697.0, + "step": 24634 + }, + { + "epoch": 2.031417498144636, + "grad_norm": 2.264309883117676, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.782230019569397, + "num_tokens": 630564163.0, + "step": 24635 + }, + { + "epoch": 2.0314999587696874, + "grad_norm": 2.210426092147827, + "learning_rate": 1e-06, + "loss": 0.5448, + "mean_token_accuracy": 0.8110830783843994, + "num_tokens": 630589466.0, + "step": 24636 + }, + { + "epoch": 2.031582419394739, + "grad_norm": 1.8558887243270874, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7790781259536743, + "num_tokens": 630627801.0, + "step": 24637 + }, + { + "epoch": 2.0316648800197905, + "grad_norm": 2.1084811687469482, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7604379057884216, + "num_tokens": 630656898.0, + "step": 24638 + }, + { + "epoch": 2.0317473406448423, + "grad_norm": 2.1042747497558594, + "learning_rate": 1e-06, + "loss": 0.5936, + "mean_token_accuracy": 0.8028692007064819, + "num_tokens": 630684507.0, + "step": 24639 + }, + { + "epoch": 2.0318298012698937, + "grad_norm": 1.9528570175170898, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7690496444702148, + "num_tokens": 630716402.0, + "step": 24640 + }, + { + "epoch": 2.031912261894945, + "grad_norm": 2.2650866508483887, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.7994388341903687, + "num_tokens": 630740069.0, + "step": 24641 + }, + { + "epoch": 2.031994722519997, + "grad_norm": 2.153247833251953, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7416611313819885, + "num_tokens": 630771806.0, + "step": 24642 + }, + { + "epoch": 2.032077183145048, + "grad_norm": 2.386321544647217, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7508644461631775, + "num_tokens": 630798979.0, + "step": 24643 + }, + { + "epoch": 2.0321596437701, + "grad_norm": 1.9458140134811401, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7538059949874878, + "num_tokens": 630833045.0, + "step": 24644 + }, + { + "epoch": 2.0322421043951513, + "grad_norm": 2.373194932937622, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7654309868812561, + "num_tokens": 630857370.0, + "step": 24645 + }, + { + "epoch": 2.0323245650202026, + "grad_norm": 2.087035894393921, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7635160684585571, + "num_tokens": 630886527.0, + "step": 24646 + }, + { + "epoch": 2.0324070256452544, + "grad_norm": 2.020653009414673, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7647523880004883, + "num_tokens": 630915236.0, + "step": 24647 + }, + { + "epoch": 2.032489486270306, + "grad_norm": 2.1143674850463867, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.7945330142974854, + "num_tokens": 630945455.0, + "step": 24648 + }, + { + "epoch": 2.0325719468953576, + "grad_norm": 2.1214847564697266, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7713630199432373, + "num_tokens": 630972353.0, + "step": 24649 + }, + { + "epoch": 2.032654407520409, + "grad_norm": 2.08174467086792, + "learning_rate": 1e-06, + "loss": 0.5712, + "mean_token_accuracy": 0.8056029081344604, + "num_tokens": 630998471.0, + "step": 24650 + }, + { + "epoch": 2.0327368681454607, + "grad_norm": 2.297128915786743, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7207667827606201, + "num_tokens": 631024937.0, + "step": 24651 + }, + { + "epoch": 2.032819328770512, + "grad_norm": 2.0994973182678223, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7740276455879211, + "num_tokens": 631054026.0, + "step": 24652 + }, + { + "epoch": 2.0329017893955634, + "grad_norm": 1.8493826389312744, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7813811302185059, + "num_tokens": 631086340.0, + "step": 24653 + }, + { + "epoch": 2.0329842500206152, + "grad_norm": 2.1499626636505127, + "learning_rate": 1e-06, + "loss": 0.7336, + "mean_token_accuracy": 0.7581478953361511, + "num_tokens": 631117883.0, + "step": 24654 + }, + { + "epoch": 2.0330667106456666, + "grad_norm": 1.948978066444397, + "learning_rate": 1e-06, + "loss": 0.5948, + "mean_token_accuracy": 0.8015878200531006, + "num_tokens": 631149199.0, + "step": 24655 + }, + { + "epoch": 2.0331491712707184, + "grad_norm": 1.9257538318634033, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7700930833816528, + "num_tokens": 631183890.0, + "step": 24656 + }, + { + "epoch": 2.0332316318957697, + "grad_norm": 2.043942928314209, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7805005311965942, + "num_tokens": 631211933.0, + "step": 24657 + }, + { + "epoch": 2.0333140925208215, + "grad_norm": 2.236877679824829, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.7913845777511597, + "num_tokens": 631236253.0, + "step": 24658 + }, + { + "epoch": 2.033396553145873, + "grad_norm": 1.889169454574585, + "learning_rate": 1e-06, + "loss": 0.5954, + "mean_token_accuracy": 0.8073188066482544, + "num_tokens": 631267767.0, + "step": 24659 + }, + { + "epoch": 2.033479013770924, + "grad_norm": 1.8462255001068115, + "learning_rate": 1e-06, + "loss": 0.586, + "mean_token_accuracy": 0.8043262958526611, + "num_tokens": 631303734.0, + "step": 24660 + }, + { + "epoch": 2.033561474395976, + "grad_norm": 2.117769718170166, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7695282697677612, + "num_tokens": 631331922.0, + "step": 24661 + }, + { + "epoch": 2.0336439350210274, + "grad_norm": 1.8173840045928955, + "learning_rate": 1e-06, + "loss": 0.5766, + "mean_token_accuracy": 0.8096253871917725, + "num_tokens": 631361360.0, + "step": 24662 + }, + { + "epoch": 2.033726395646079, + "grad_norm": 2.1924338340759277, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7785026431083679, + "num_tokens": 631387797.0, + "step": 24663 + }, + { + "epoch": 2.0338088562711305, + "grad_norm": 1.8871146440505981, + "learning_rate": 1e-06, + "loss": 0.5908, + "mean_token_accuracy": 0.8004494905471802, + "num_tokens": 631419046.0, + "step": 24664 + }, + { + "epoch": 2.033891316896182, + "grad_norm": 1.6661014556884766, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.8003051280975342, + "num_tokens": 631457162.0, + "step": 24665 + }, + { + "epoch": 2.0339737775212337, + "grad_norm": 1.9703119993209839, + "learning_rate": 1e-06, + "loss": 0.5864, + "mean_token_accuracy": 0.8113687634468079, + "num_tokens": 631489102.0, + "step": 24666 + }, + { + "epoch": 2.034056238146285, + "grad_norm": 2.0051512718200684, + "learning_rate": 1e-06, + "loss": 0.6021, + "mean_token_accuracy": 0.7961515784263611, + "num_tokens": 631517117.0, + "step": 24667 + }, + { + "epoch": 2.034138698771337, + "grad_norm": 2.041576385498047, + "learning_rate": 1e-06, + "loss": 0.5893, + "mean_token_accuracy": 0.809312641620636, + "num_tokens": 631546420.0, + "step": 24668 + }, + { + "epoch": 2.034221159396388, + "grad_norm": 1.9197286367416382, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7632408738136292, + "num_tokens": 631579742.0, + "step": 24669 + }, + { + "epoch": 2.03430362002144, + "grad_norm": 2.2113571166992188, + "learning_rate": 1e-06, + "loss": 0.548, + "mean_token_accuracy": 0.8239233493804932, + "num_tokens": 631604076.0, + "step": 24670 + }, + { + "epoch": 2.0343860806464913, + "grad_norm": 1.9073079824447632, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7423788905143738, + "num_tokens": 631638441.0, + "step": 24671 + }, + { + "epoch": 2.0344685412715426, + "grad_norm": 2.3536152839660645, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7441843748092651, + "num_tokens": 631664741.0, + "step": 24672 + }, + { + "epoch": 2.0345510018965944, + "grad_norm": 1.8842144012451172, + "learning_rate": 1e-06, + "loss": 0.6029, + "mean_token_accuracy": 0.7894860506057739, + "num_tokens": 631699736.0, + "step": 24673 + }, + { + "epoch": 2.034633462521646, + "grad_norm": 2.060384511947632, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7825303077697754, + "num_tokens": 631731008.0, + "step": 24674 + }, + { + "epoch": 2.0347159231466976, + "grad_norm": 1.9850900173187256, + "learning_rate": 1e-06, + "loss": 0.5867, + "mean_token_accuracy": 0.8033462166786194, + "num_tokens": 631761436.0, + "step": 24675 + }, + { + "epoch": 2.034798383771749, + "grad_norm": 2.2416563034057617, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.7931239008903503, + "num_tokens": 631787213.0, + "step": 24676 + }, + { + "epoch": 2.0348808443968007, + "grad_norm": 1.9640356302261353, + "learning_rate": 1e-06, + "loss": 0.5944, + "mean_token_accuracy": 0.8096940517425537, + "num_tokens": 631817089.0, + "step": 24677 + }, + { + "epoch": 2.034963305021852, + "grad_norm": 2.027043581008911, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.7923119068145752, + "num_tokens": 631845383.0, + "step": 24678 + }, + { + "epoch": 2.0350457656469034, + "grad_norm": 2.2115228176116943, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.7904682159423828, + "num_tokens": 631872732.0, + "step": 24679 + }, + { + "epoch": 2.035128226271955, + "grad_norm": 2.1908838748931885, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7803287506103516, + "num_tokens": 631899747.0, + "step": 24680 + }, + { + "epoch": 2.0352106868970066, + "grad_norm": 2.120727777481079, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7736411690711975, + "num_tokens": 631928058.0, + "step": 24681 + }, + { + "epoch": 2.0352931475220584, + "grad_norm": 2.064790964126587, + "learning_rate": 1e-06, + "loss": 0.5841, + "mean_token_accuracy": 0.7912417650222778, + "num_tokens": 631957597.0, + "step": 24682 + }, + { + "epoch": 2.0353756081471097, + "grad_norm": 2.229163408279419, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7337114214897156, + "num_tokens": 631986843.0, + "step": 24683 + }, + { + "epoch": 2.0354580687721615, + "grad_norm": 2.0468297004699707, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.7839975357055664, + "num_tokens": 632018651.0, + "step": 24684 + }, + { + "epoch": 2.035540529397213, + "grad_norm": 2.107762336730957, + "learning_rate": 1e-06, + "loss": 0.4962, + "mean_token_accuracy": 0.8209998607635498, + "num_tokens": 632044189.0, + "step": 24685 + }, + { + "epoch": 2.035622990022264, + "grad_norm": 3.499361991882324, + "learning_rate": 1e-06, + "loss": 0.5787, + "mean_token_accuracy": 0.7971013784408569, + "num_tokens": 632072053.0, + "step": 24686 + }, + { + "epoch": 2.035705450647316, + "grad_norm": 1.9468754529953003, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.7867953777313232, + "num_tokens": 632105904.0, + "step": 24687 + }, + { + "epoch": 2.0357879112723674, + "grad_norm": 1.966809630393982, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.790189266204834, + "num_tokens": 632138223.0, + "step": 24688 + }, + { + "epoch": 2.035870371897419, + "grad_norm": 2.240941286087036, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7801029682159424, + "num_tokens": 632163531.0, + "step": 24689 + }, + { + "epoch": 2.0359528325224705, + "grad_norm": 2.151764392852783, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.7830314636230469, + "num_tokens": 632192862.0, + "step": 24690 + }, + { + "epoch": 2.036035293147522, + "grad_norm": 1.91904878616333, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7475540637969971, + "num_tokens": 632226579.0, + "step": 24691 + }, + { + "epoch": 2.0361177537725736, + "grad_norm": 1.952938199043274, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.795466959476471, + "num_tokens": 632260252.0, + "step": 24692 + }, + { + "epoch": 2.036200214397625, + "grad_norm": 2.1154894828796387, + "learning_rate": 1e-06, + "loss": 0.5874, + "mean_token_accuracy": 0.8128525018692017, + "num_tokens": 632286323.0, + "step": 24693 + }, + { + "epoch": 2.036282675022677, + "grad_norm": 1.8296653032302856, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7716900706291199, + "num_tokens": 632323106.0, + "step": 24694 + }, + { + "epoch": 2.036365135647728, + "grad_norm": 2.077531099319458, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7533597946166992, + "num_tokens": 632355400.0, + "step": 24695 + }, + { + "epoch": 2.03644759627278, + "grad_norm": 2.175936698913574, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7527455687522888, + "num_tokens": 632386040.0, + "step": 24696 + }, + { + "epoch": 2.0365300568978313, + "grad_norm": 2.0606863498687744, + "learning_rate": 1e-06, + "loss": 0.6151, + "mean_token_accuracy": 0.7911720275878906, + "num_tokens": 632415036.0, + "step": 24697 + }, + { + "epoch": 2.0366125175228826, + "grad_norm": 2.2620444297790527, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7675208449363708, + "num_tokens": 632442709.0, + "step": 24698 + }, + { + "epoch": 2.0366949781479344, + "grad_norm": 2.408114194869995, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7767620086669922, + "num_tokens": 632467789.0, + "step": 24699 + }, + { + "epoch": 2.036777438772986, + "grad_norm": 2.238337755203247, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7389077544212341, + "num_tokens": 632495446.0, + "step": 24700 + }, + { + "epoch": 2.0368598993980376, + "grad_norm": 2.180605888366699, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.7702380418777466, + "num_tokens": 632525337.0, + "step": 24701 + }, + { + "epoch": 2.036942360023089, + "grad_norm": 2.493252992630005, + "learning_rate": 1e-06, + "loss": 0.57, + "mean_token_accuracy": 0.8030650615692139, + "num_tokens": 632546759.0, + "step": 24702 + }, + { + "epoch": 2.0370248206481407, + "grad_norm": 1.9150691032409668, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7834041118621826, + "num_tokens": 632580998.0, + "step": 24703 + }, + { + "epoch": 2.037107281273192, + "grad_norm": 2.0531954765319824, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7677940130233765, + "num_tokens": 632609069.0, + "step": 24704 + }, + { + "epoch": 2.0371897418982434, + "grad_norm": 2.0847225189208984, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7789686322212219, + "num_tokens": 632639318.0, + "step": 24705 + }, + { + "epoch": 2.037272202523295, + "grad_norm": 1.9839224815368652, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7808765172958374, + "num_tokens": 632672701.0, + "step": 24706 + }, + { + "epoch": 2.0373546631483466, + "grad_norm": 2.008518695831299, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7694156765937805, + "num_tokens": 632701926.0, + "step": 24707 + }, + { + "epoch": 2.0374371237733984, + "grad_norm": 2.0395898818969727, + "learning_rate": 1e-06, + "loss": 0.4809, + "mean_token_accuracy": 0.8322570323944092, + "num_tokens": 632731486.0, + "step": 24708 + }, + { + "epoch": 2.0375195843984497, + "grad_norm": 2.056594133377075, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7736032009124756, + "num_tokens": 632760258.0, + "step": 24709 + }, + { + "epoch": 2.037602045023501, + "grad_norm": 2.150034189224243, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7329943776130676, + "num_tokens": 632790998.0, + "step": 24710 + }, + { + "epoch": 2.037684505648553, + "grad_norm": 2.3408191204071045, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7646516561508179, + "num_tokens": 632815521.0, + "step": 24711 + }, + { + "epoch": 2.037766966273604, + "grad_norm": 2.071955442428589, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7810711860656738, + "num_tokens": 632845954.0, + "step": 24712 + }, + { + "epoch": 2.037849426898656, + "grad_norm": 1.9755061864852905, + "learning_rate": 1e-06, + "loss": 0.5842, + "mean_token_accuracy": 0.8081325888633728, + "num_tokens": 632876253.0, + "step": 24713 + }, + { + "epoch": 2.0379318875237074, + "grad_norm": 1.9644761085510254, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7580273151397705, + "num_tokens": 632910484.0, + "step": 24714 + }, + { + "epoch": 2.038014348148759, + "grad_norm": 2.3874826431274414, + "learning_rate": 1e-06, + "loss": 0.6008, + "mean_token_accuracy": 0.7938668131828308, + "num_tokens": 632937359.0, + "step": 24715 + }, + { + "epoch": 2.0380968087738105, + "grad_norm": 1.9884330034255981, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7866756916046143, + "num_tokens": 632970528.0, + "step": 24716 + }, + { + "epoch": 2.038179269398862, + "grad_norm": 2.092986583709717, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7586344480514526, + "num_tokens": 633001724.0, + "step": 24717 + }, + { + "epoch": 2.0382617300239136, + "grad_norm": 2.4807496070861816, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7457894086837769, + "num_tokens": 633023937.0, + "step": 24718 + }, + { + "epoch": 2.038344190648965, + "grad_norm": 2.0252556800842285, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.7783334255218506, + "num_tokens": 633055179.0, + "step": 24719 + }, + { + "epoch": 2.038426651274017, + "grad_norm": 2.0646910667419434, + "learning_rate": 1e-06, + "loss": 0.5555, + "mean_token_accuracy": 0.8084978461265564, + "num_tokens": 633085842.0, + "step": 24720 + }, + { + "epoch": 2.038509111899068, + "grad_norm": 1.983185887336731, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.777302622795105, + "num_tokens": 633117596.0, + "step": 24721 + }, + { + "epoch": 2.03859157252412, + "grad_norm": 2.252692937850952, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7634248733520508, + "num_tokens": 633144291.0, + "step": 24722 + }, + { + "epoch": 2.0386740331491713, + "grad_norm": 1.9541910886764526, + "learning_rate": 1e-06, + "loss": 0.5012, + "mean_token_accuracy": 0.8341109156608582, + "num_tokens": 633173487.0, + "step": 24723 + }, + { + "epoch": 2.0387564937742226, + "grad_norm": 1.9988248348236084, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.780752420425415, + "num_tokens": 633207558.0, + "step": 24724 + }, + { + "epoch": 2.0388389543992744, + "grad_norm": 2.2072081565856934, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7352474927902222, + "num_tokens": 633236900.0, + "step": 24725 + }, + { + "epoch": 2.038921415024326, + "grad_norm": 2.12394642829895, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7629163265228271, + "num_tokens": 633266682.0, + "step": 24726 + }, + { + "epoch": 2.0390038756493776, + "grad_norm": 1.9459705352783203, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7415775656700134, + "num_tokens": 633303852.0, + "step": 24727 + }, + { + "epoch": 2.039086336274429, + "grad_norm": 2.0386061668395996, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7767258882522583, + "num_tokens": 633334707.0, + "step": 24728 + }, + { + "epoch": 2.0391687968994807, + "grad_norm": 1.9564448595046997, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7752007246017456, + "num_tokens": 633369388.0, + "step": 24729 + }, + { + "epoch": 2.039251257524532, + "grad_norm": 2.2415945529937744, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.74753338098526, + "num_tokens": 633395999.0, + "step": 24730 + }, + { + "epoch": 2.0393337181495834, + "grad_norm": 2.0974678993225098, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7737009525299072, + "num_tokens": 633426789.0, + "step": 24731 + }, + { + "epoch": 2.039416178774635, + "grad_norm": 2.2102041244506836, + "learning_rate": 1e-06, + "loss": 0.6099, + "mean_token_accuracy": 0.7947252988815308, + "num_tokens": 633454334.0, + "step": 24732 + }, + { + "epoch": 2.0394986393996866, + "grad_norm": 2.057830333709717, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7717430591583252, + "num_tokens": 633486409.0, + "step": 24733 + }, + { + "epoch": 2.0395811000247384, + "grad_norm": 2.3054635524749756, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.7814474105834961, + "num_tokens": 633508750.0, + "step": 24734 + }, + { + "epoch": 2.0396635606497897, + "grad_norm": 1.9048244953155518, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7779330611228943, + "num_tokens": 633541871.0, + "step": 24735 + }, + { + "epoch": 2.039746021274841, + "grad_norm": 1.9973530769348145, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.8034428954124451, + "num_tokens": 633570565.0, + "step": 24736 + }, + { + "epoch": 2.039828481899893, + "grad_norm": 1.9919098615646362, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7759249806404114, + "num_tokens": 633603343.0, + "step": 24737 + }, + { + "epoch": 2.039910942524944, + "grad_norm": 2.507176637649536, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7718145251274109, + "num_tokens": 633623518.0, + "step": 24738 + }, + { + "epoch": 2.039993403149996, + "grad_norm": 1.9152319431304932, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7766039371490479, + "num_tokens": 633660349.0, + "step": 24739 + }, + { + "epoch": 2.0400758637750473, + "grad_norm": 2.0596604347229004, + "learning_rate": 1e-06, + "loss": 0.6002, + "mean_token_accuracy": 0.7955422401428223, + "num_tokens": 633690929.0, + "step": 24740 + }, + { + "epoch": 2.040158324400099, + "grad_norm": 1.8144034147262573, + "learning_rate": 1e-06, + "loss": 0.6361, + "mean_token_accuracy": 0.7920656204223633, + "num_tokens": 633727449.0, + "step": 24741 + }, + { + "epoch": 2.0402407850251505, + "grad_norm": 1.891405463218689, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7872042655944824, + "num_tokens": 633764207.0, + "step": 24742 + }, + { + "epoch": 2.040323245650202, + "grad_norm": 2.176197052001953, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7338330149650574, + "num_tokens": 633794191.0, + "step": 24743 + }, + { + "epoch": 2.0404057062752536, + "grad_norm": 2.086033821105957, + "learning_rate": 1e-06, + "loss": 0.5818, + "mean_token_accuracy": 0.8028518557548523, + "num_tokens": 633822859.0, + "step": 24744 + }, + { + "epoch": 2.040488166900305, + "grad_norm": 1.786817193031311, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.798803448677063, + "num_tokens": 633861640.0, + "step": 24745 + }, + { + "epoch": 2.040570627525357, + "grad_norm": 2.2805464267730713, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7741636037826538, + "num_tokens": 633887877.0, + "step": 24746 + }, + { + "epoch": 2.040653088150408, + "grad_norm": 2.109743595123291, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.745138943195343, + "num_tokens": 633917700.0, + "step": 24747 + }, + { + "epoch": 2.04073554877546, + "grad_norm": 2.2658591270446777, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7667999267578125, + "num_tokens": 633943186.0, + "step": 24748 + }, + { + "epoch": 2.0408180094005113, + "grad_norm": 2.202409029006958, + "learning_rate": 1e-06, + "loss": 0.5998, + "mean_token_accuracy": 0.7937307357788086, + "num_tokens": 633968121.0, + "step": 24749 + }, + { + "epoch": 2.0409004700255626, + "grad_norm": 2.060786247253418, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7802408933639526, + "num_tokens": 633998556.0, + "step": 24750 + }, + { + "epoch": 2.0409829306506144, + "grad_norm": 2.061662435531616, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7768535614013672, + "num_tokens": 634029457.0, + "step": 24751 + }, + { + "epoch": 2.0410653912756658, + "grad_norm": 1.8143266439437866, + "learning_rate": 1e-06, + "loss": 0.5124, + "mean_token_accuracy": 0.821469783782959, + "num_tokens": 634063197.0, + "step": 24752 + }, + { + "epoch": 2.0411478519007176, + "grad_norm": 2.2507834434509277, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7681617736816406, + "num_tokens": 634089100.0, + "step": 24753 + }, + { + "epoch": 2.041230312525769, + "grad_norm": 2.095672845840454, + "learning_rate": 1e-06, + "loss": 0.6271, + "mean_token_accuracy": 0.7915163040161133, + "num_tokens": 634119147.0, + "step": 24754 + }, + { + "epoch": 2.0413127731508203, + "grad_norm": 1.7962452173233032, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.7960727214813232, + "num_tokens": 634157897.0, + "step": 24755 + }, + { + "epoch": 2.041395233775872, + "grad_norm": 1.9113909006118774, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.7906365990638733, + "num_tokens": 634192457.0, + "step": 24756 + }, + { + "epoch": 2.0414776944009234, + "grad_norm": 1.9842394590377808, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.7981100082397461, + "num_tokens": 634225449.0, + "step": 24757 + }, + { + "epoch": 2.041560155025975, + "grad_norm": 1.973899483680725, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7545607089996338, + "num_tokens": 634258025.0, + "step": 24758 + }, + { + "epoch": 2.0416426156510266, + "grad_norm": 2.0097923278808594, + "learning_rate": 1e-06, + "loss": 0.6122, + "mean_token_accuracy": 0.7926511168479919, + "num_tokens": 634290112.0, + "step": 24759 + }, + { + "epoch": 2.0417250762760784, + "grad_norm": 1.9619616270065308, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.7559503316879272, + "num_tokens": 634325322.0, + "step": 24760 + }, + { + "epoch": 2.0418075369011297, + "grad_norm": 2.1321239471435547, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.730819046497345, + "num_tokens": 634353517.0, + "step": 24761 + }, + { + "epoch": 2.041889997526181, + "grad_norm": 2.2672572135925293, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.770936906337738, + "num_tokens": 634379034.0, + "step": 24762 + }, + { + "epoch": 2.041972458151233, + "grad_norm": 2.2574424743652344, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7609351277351379, + "num_tokens": 634405916.0, + "step": 24763 + }, + { + "epoch": 2.042054918776284, + "grad_norm": 2.195859432220459, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7603892087936401, + "num_tokens": 634435010.0, + "step": 24764 + }, + { + "epoch": 2.042137379401336, + "grad_norm": 2.2043466567993164, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7617172002792358, + "num_tokens": 634464295.0, + "step": 24765 + }, + { + "epoch": 2.0422198400263873, + "grad_norm": 2.2774393558502197, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7437951564788818, + "num_tokens": 634489978.0, + "step": 24766 + }, + { + "epoch": 2.042302300651439, + "grad_norm": 2.1605818271636963, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7411719560623169, + "num_tokens": 634519191.0, + "step": 24767 + }, + { + "epoch": 2.0423847612764905, + "grad_norm": 2.10396146774292, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.782916784286499, + "num_tokens": 634549988.0, + "step": 24768 + }, + { + "epoch": 2.042467221901542, + "grad_norm": 2.0949721336364746, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7778792977333069, + "num_tokens": 634579478.0, + "step": 24769 + }, + { + "epoch": 2.0425496825265936, + "grad_norm": 2.3195254802703857, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.760422945022583, + "num_tokens": 634605047.0, + "step": 24770 + }, + { + "epoch": 2.042632143151645, + "grad_norm": 2.284024477005005, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7721962332725525, + "num_tokens": 634630737.0, + "step": 24771 + }, + { + "epoch": 2.042714603776697, + "grad_norm": 1.8519089221954346, + "learning_rate": 1e-06, + "loss": 0.569, + "mean_token_accuracy": 0.7994017601013184, + "num_tokens": 634664268.0, + "step": 24772 + }, + { + "epoch": 2.042797064401748, + "grad_norm": 2.31510591506958, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7580331563949585, + "num_tokens": 634689848.0, + "step": 24773 + }, + { + "epoch": 2.0428795250268, + "grad_norm": 1.9086551666259766, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.789129912853241, + "num_tokens": 634724753.0, + "step": 24774 + }, + { + "epoch": 2.0429619856518513, + "grad_norm": 1.8882921934127808, + "learning_rate": 1e-06, + "loss": 0.5084, + "mean_token_accuracy": 0.825263500213623, + "num_tokens": 634758270.0, + "step": 24775 + }, + { + "epoch": 2.0430444462769026, + "grad_norm": 2.09482741355896, + "learning_rate": 1e-06, + "loss": 0.5926, + "mean_token_accuracy": 0.7924085855484009, + "num_tokens": 634785777.0, + "step": 24776 + }, + { + "epoch": 2.0431269069019544, + "grad_norm": 2.0581021308898926, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.7873658537864685, + "num_tokens": 634816350.0, + "step": 24777 + }, + { + "epoch": 2.0432093675270058, + "grad_norm": 2.2369682788848877, + "learning_rate": 1e-06, + "loss": 0.5589, + "mean_token_accuracy": 0.7938442230224609, + "num_tokens": 634841028.0, + "step": 24778 + }, + { + "epoch": 2.0432918281520576, + "grad_norm": 2.1365668773651123, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7791275978088379, + "num_tokens": 634868647.0, + "step": 24779 + }, + { + "epoch": 2.043374288777109, + "grad_norm": 2.070816993713379, + "learning_rate": 1e-06, + "loss": 0.6361, + "mean_token_accuracy": 0.7922864556312561, + "num_tokens": 634898844.0, + "step": 24780 + }, + { + "epoch": 2.0434567494021603, + "grad_norm": 1.9467183351516724, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7575370073318481, + "num_tokens": 634935201.0, + "step": 24781 + }, + { + "epoch": 2.043539210027212, + "grad_norm": 2.080166816711426, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.7914798855781555, + "num_tokens": 634967671.0, + "step": 24782 + }, + { + "epoch": 2.0436216706522634, + "grad_norm": 1.9776568412780762, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.7937352061271667, + "num_tokens": 634999472.0, + "step": 24783 + }, + { + "epoch": 2.043704131277315, + "grad_norm": 2.244126558303833, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7493584752082825, + "num_tokens": 635027499.0, + "step": 24784 + }, + { + "epoch": 2.0437865919023666, + "grad_norm": 2.208925485610962, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.7904630303382874, + "num_tokens": 635055942.0, + "step": 24785 + }, + { + "epoch": 2.0438690525274184, + "grad_norm": 2.0229930877685547, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7281566858291626, + "num_tokens": 635087172.0, + "step": 24786 + }, + { + "epoch": 2.0439515131524697, + "grad_norm": 1.9723936319351196, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.7916107177734375, + "num_tokens": 635118735.0, + "step": 24787 + }, + { + "epoch": 2.044033973777521, + "grad_norm": 2.2261204719543457, + "learning_rate": 1e-06, + "loss": 0.5901, + "mean_token_accuracy": 0.7973119020462036, + "num_tokens": 635142954.0, + "step": 24788 + }, + { + "epoch": 2.044116434402573, + "grad_norm": 2.0743024349212646, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.7908939123153687, + "num_tokens": 635171939.0, + "step": 24789 + }, + { + "epoch": 2.044198895027624, + "grad_norm": 2.2390475273132324, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.742702066898346, + "num_tokens": 635199093.0, + "step": 24790 + }, + { + "epoch": 2.044281355652676, + "grad_norm": 1.91083824634552, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7616622447967529, + "num_tokens": 635232681.0, + "step": 24791 + }, + { + "epoch": 2.0443638162777273, + "grad_norm": 1.936211109161377, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7785062193870544, + "num_tokens": 635265656.0, + "step": 24792 + }, + { + "epoch": 2.044446276902779, + "grad_norm": 2.0442118644714355, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7823166847229004, + "num_tokens": 635296445.0, + "step": 24793 + }, + { + "epoch": 2.0445287375278305, + "grad_norm": 2.0967519283294678, + "learning_rate": 1e-06, + "loss": 0.5788, + "mean_token_accuracy": 0.808830976486206, + "num_tokens": 635323413.0, + "step": 24794 + }, + { + "epoch": 2.044611198152882, + "grad_norm": 2.3747336864471436, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7700912952423096, + "num_tokens": 635347418.0, + "step": 24795 + }, + { + "epoch": 2.0446936587779336, + "grad_norm": 1.9936619997024536, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.7625527381896973, + "num_tokens": 635378544.0, + "step": 24796 + }, + { + "epoch": 2.044776119402985, + "grad_norm": 1.930840253829956, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7735395431518555, + "num_tokens": 635412378.0, + "step": 24797 + }, + { + "epoch": 2.0448585800280368, + "grad_norm": 1.9913986921310425, + "learning_rate": 1e-06, + "loss": 0.6137, + "mean_token_accuracy": 0.7939288020133972, + "num_tokens": 635445519.0, + "step": 24798 + }, + { + "epoch": 2.044941040653088, + "grad_norm": 1.9503049850463867, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7753576040267944, + "num_tokens": 635476377.0, + "step": 24799 + }, + { + "epoch": 2.0450235012781395, + "grad_norm": 2.2454283237457275, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.7914528846740723, + "num_tokens": 635503983.0, + "step": 24800 + }, + { + "epoch": 2.0451059619031913, + "grad_norm": 2.100911855697632, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.8017141819000244, + "num_tokens": 635532391.0, + "step": 24801 + }, + { + "epoch": 2.0451884225282426, + "grad_norm": 1.787374496459961, + "learning_rate": 1e-06, + "loss": 0.5719, + "mean_token_accuracy": 0.8085713386535645, + "num_tokens": 635573088.0, + "step": 24802 + }, + { + "epoch": 2.0452708831532944, + "grad_norm": 2.1874194145202637, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.754460334777832, + "num_tokens": 635599998.0, + "step": 24803 + }, + { + "epoch": 2.0453533437783458, + "grad_norm": 2.197777509689331, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.7562341690063477, + "num_tokens": 635627950.0, + "step": 24804 + }, + { + "epoch": 2.0454358044033976, + "grad_norm": 1.831712007522583, + "learning_rate": 1e-06, + "loss": 0.6057, + "mean_token_accuracy": 0.8051271438598633, + "num_tokens": 635664704.0, + "step": 24805 + }, + { + "epoch": 2.045518265028449, + "grad_norm": 2.273425340652466, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.7915006279945374, + "num_tokens": 635691287.0, + "step": 24806 + }, + { + "epoch": 2.0456007256535003, + "grad_norm": 2.4854953289031982, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7648080587387085, + "num_tokens": 635713992.0, + "step": 24807 + }, + { + "epoch": 2.045683186278552, + "grad_norm": 2.212583541870117, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.751709520816803, + "num_tokens": 635743828.0, + "step": 24808 + }, + { + "epoch": 2.0457656469036034, + "grad_norm": 2.1048505306243896, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7646300196647644, + "num_tokens": 635773516.0, + "step": 24809 + }, + { + "epoch": 2.045848107528655, + "grad_norm": 2.165961980819702, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7469011545181274, + "num_tokens": 635804553.0, + "step": 24810 + }, + { + "epoch": 2.0459305681537066, + "grad_norm": 2.001424551010132, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7692400813102722, + "num_tokens": 635834270.0, + "step": 24811 + }, + { + "epoch": 2.0460130287787583, + "grad_norm": 2.0579006671905518, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7685789465904236, + "num_tokens": 635863342.0, + "step": 24812 + }, + { + "epoch": 2.0460954894038097, + "grad_norm": 2.1793806552886963, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7829488515853882, + "num_tokens": 635897050.0, + "step": 24813 + }, + { + "epoch": 2.046177950028861, + "grad_norm": 2.349763870239258, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7621608972549438, + "num_tokens": 635921225.0, + "step": 24814 + }, + { + "epoch": 2.046260410653913, + "grad_norm": 2.2285072803497314, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7741909623146057, + "num_tokens": 635947507.0, + "step": 24815 + }, + { + "epoch": 2.046342871278964, + "grad_norm": 2.3850367069244385, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7161108255386353, + "num_tokens": 635974174.0, + "step": 24816 + }, + { + "epoch": 2.046425331904016, + "grad_norm": 1.9294068813323975, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7798436284065247, + "num_tokens": 636008957.0, + "step": 24817 + }, + { + "epoch": 2.0465077925290673, + "grad_norm": 1.9400043487548828, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7808717489242554, + "num_tokens": 636044454.0, + "step": 24818 + }, + { + "epoch": 2.046590253154119, + "grad_norm": 2.327422618865967, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7580193877220154, + "num_tokens": 636069533.0, + "step": 24819 + }, + { + "epoch": 2.0466727137791705, + "grad_norm": 2.0957958698272705, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7454320788383484, + "num_tokens": 636100347.0, + "step": 24820 + }, + { + "epoch": 2.046755174404222, + "grad_norm": 1.9047212600708008, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7671510577201843, + "num_tokens": 636135205.0, + "step": 24821 + }, + { + "epoch": 2.0468376350292736, + "grad_norm": 2.072561025619507, + "learning_rate": 1e-06, + "loss": 0.7303, + "mean_token_accuracy": 0.7582282423973083, + "num_tokens": 636167353.0, + "step": 24822 + }, + { + "epoch": 2.046920095654325, + "grad_norm": 1.920156478881836, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.7845426201820374, + "num_tokens": 636199803.0, + "step": 24823 + }, + { + "epoch": 2.0470025562793768, + "grad_norm": 2.305532693862915, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7578948736190796, + "num_tokens": 636228313.0, + "step": 24824 + }, + { + "epoch": 2.047085016904428, + "grad_norm": 2.2853176593780518, + "learning_rate": 1e-06, + "loss": 0.6147, + "mean_token_accuracy": 0.7913034558296204, + "num_tokens": 636255308.0, + "step": 24825 + }, + { + "epoch": 2.0471674775294795, + "grad_norm": 1.9649296998977661, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.776767373085022, + "num_tokens": 636290986.0, + "step": 24826 + }, + { + "epoch": 2.0472499381545313, + "grad_norm": 2.2735276222229004, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7702106237411499, + "num_tokens": 636318598.0, + "step": 24827 + }, + { + "epoch": 2.0473323987795826, + "grad_norm": 1.8683987855911255, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.7840096354484558, + "num_tokens": 636354540.0, + "step": 24828 + }, + { + "epoch": 2.0474148594046344, + "grad_norm": 2.0457067489624023, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.7907394766807556, + "num_tokens": 636385111.0, + "step": 24829 + }, + { + "epoch": 2.0474973200296858, + "grad_norm": 2.245579957962036, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.778362512588501, + "num_tokens": 636412982.0, + "step": 24830 + }, + { + "epoch": 2.0475797806547376, + "grad_norm": 2.3396713733673096, + "learning_rate": 1e-06, + "loss": 0.5279, + "mean_token_accuracy": 0.8134099245071411, + "num_tokens": 636436752.0, + "step": 24831 + }, + { + "epoch": 2.047662241279789, + "grad_norm": 1.882439374923706, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7790815830230713, + "num_tokens": 636470269.0, + "step": 24832 + }, + { + "epoch": 2.0477447019048403, + "grad_norm": 2.0548524856567383, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7892935872077942, + "num_tokens": 636498727.0, + "step": 24833 + }, + { + "epoch": 2.047827162529892, + "grad_norm": 1.716933012008667, + "learning_rate": 1e-06, + "loss": 0.5928, + "mean_token_accuracy": 0.8062300682067871, + "num_tokens": 636537619.0, + "step": 24834 + }, + { + "epoch": 2.0479096231549434, + "grad_norm": 1.950867772102356, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7471668124198914, + "num_tokens": 636571198.0, + "step": 24835 + }, + { + "epoch": 2.047992083779995, + "grad_norm": 2.2013494968414307, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.7972036600112915, + "num_tokens": 636597856.0, + "step": 24836 + }, + { + "epoch": 2.0480745444050465, + "grad_norm": 1.9887900352478027, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7669946551322937, + "num_tokens": 636630539.0, + "step": 24837 + }, + { + "epoch": 2.0481570050300983, + "grad_norm": 2.1227922439575195, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7755577564239502, + "num_tokens": 636659633.0, + "step": 24838 + }, + { + "epoch": 2.0482394656551497, + "grad_norm": 2.1030399799346924, + "learning_rate": 1e-06, + "loss": 0.587, + "mean_token_accuracy": 0.809855580329895, + "num_tokens": 636688454.0, + "step": 24839 + }, + { + "epoch": 2.048321926280201, + "grad_norm": 1.8370611667633057, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.7817878723144531, + "num_tokens": 636726285.0, + "step": 24840 + }, + { + "epoch": 2.048404386905253, + "grad_norm": 1.8962907791137695, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7815848588943481, + "num_tokens": 636756665.0, + "step": 24841 + }, + { + "epoch": 2.048486847530304, + "grad_norm": 2.0185129642486572, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.7825014591217041, + "num_tokens": 636787840.0, + "step": 24842 + }, + { + "epoch": 2.048569308155356, + "grad_norm": 2.182002067565918, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.7608115673065186, + "num_tokens": 636814862.0, + "step": 24843 + }, + { + "epoch": 2.0486517687804073, + "grad_norm": 1.665211796760559, + "learning_rate": 1e-06, + "loss": 0.5169, + "mean_token_accuracy": 0.8318557143211365, + "num_tokens": 636853679.0, + "step": 24844 + }, + { + "epoch": 2.0487342294054587, + "grad_norm": 2.39699649810791, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7718384265899658, + "num_tokens": 636875045.0, + "step": 24845 + }, + { + "epoch": 2.0488166900305105, + "grad_norm": 2.052185535430908, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7783863544464111, + "num_tokens": 636902931.0, + "step": 24846 + }, + { + "epoch": 2.048899150655562, + "grad_norm": 2.0446109771728516, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.7938535213470459, + "num_tokens": 636933971.0, + "step": 24847 + }, + { + "epoch": 2.0489816112806136, + "grad_norm": 1.9996988773345947, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.7921942472457886, + "num_tokens": 636962700.0, + "step": 24848 + }, + { + "epoch": 2.049064071905665, + "grad_norm": 2.2532222270965576, + "learning_rate": 1e-06, + "loss": 0.604, + "mean_token_accuracy": 0.7966133952140808, + "num_tokens": 636989255.0, + "step": 24849 + }, + { + "epoch": 2.0491465325307168, + "grad_norm": 1.8280022144317627, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7857944965362549, + "num_tokens": 637025995.0, + "step": 24850 + }, + { + "epoch": 2.049228993155768, + "grad_norm": 2.0689382553100586, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7412700653076172, + "num_tokens": 637056349.0, + "step": 24851 + }, + { + "epoch": 2.0493114537808195, + "grad_norm": 1.9453667402267456, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7741605043411255, + "num_tokens": 637090392.0, + "step": 24852 + }, + { + "epoch": 2.0493939144058713, + "grad_norm": 2.2030954360961914, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7718230485916138, + "num_tokens": 637120661.0, + "step": 24853 + }, + { + "epoch": 2.0494763750309226, + "grad_norm": 2.066509485244751, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7774986028671265, + "num_tokens": 637151166.0, + "step": 24854 + }, + { + "epoch": 2.0495588356559744, + "grad_norm": 2.0309648513793945, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7652277946472168, + "num_tokens": 637183208.0, + "step": 24855 + }, + { + "epoch": 2.0496412962810258, + "grad_norm": 2.1170191764831543, + "learning_rate": 1e-06, + "loss": 0.5424, + "mean_token_accuracy": 0.8107980489730835, + "num_tokens": 637212210.0, + "step": 24856 + }, + { + "epoch": 2.0497237569060776, + "grad_norm": 2.1911556720733643, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7797247171401978, + "num_tokens": 637239013.0, + "step": 24857 + }, + { + "epoch": 2.049806217531129, + "grad_norm": 2.116727352142334, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7620623707771301, + "num_tokens": 637272714.0, + "step": 24858 + }, + { + "epoch": 2.0498886781561803, + "grad_norm": 2.445030450820923, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7356247901916504, + "num_tokens": 637296499.0, + "step": 24859 + }, + { + "epoch": 2.049971138781232, + "grad_norm": 1.9967172145843506, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7802115678787231, + "num_tokens": 637328388.0, + "step": 24860 + }, + { + "epoch": 2.0500535994062834, + "grad_norm": 1.8655935525894165, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7782241106033325, + "num_tokens": 637362983.0, + "step": 24861 + }, + { + "epoch": 2.050136060031335, + "grad_norm": 2.118030071258545, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.7815618515014648, + "num_tokens": 637393217.0, + "step": 24862 + }, + { + "epoch": 2.0502185206563865, + "grad_norm": 1.861776351928711, + "learning_rate": 1e-06, + "loss": 0.5964, + "mean_token_accuracy": 0.8037467002868652, + "num_tokens": 637430546.0, + "step": 24863 + }, + { + "epoch": 2.050300981281438, + "grad_norm": 1.903978705406189, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.7963084578514099, + "num_tokens": 637462991.0, + "step": 24864 + }, + { + "epoch": 2.0503834419064897, + "grad_norm": 1.9586867094039917, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.7840713262557983, + "num_tokens": 637493280.0, + "step": 24865 + }, + { + "epoch": 2.050465902531541, + "grad_norm": 2.0081276893615723, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.782328188419342, + "num_tokens": 637524278.0, + "step": 24866 + }, + { + "epoch": 2.050548363156593, + "grad_norm": 2.022653579711914, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.7842267751693726, + "num_tokens": 637556871.0, + "step": 24867 + }, + { + "epoch": 2.050630823781644, + "grad_norm": 2.183443784713745, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7268162965774536, + "num_tokens": 637584501.0, + "step": 24868 + }, + { + "epoch": 2.050713284406696, + "grad_norm": 2.3902387619018555, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7308495044708252, + "num_tokens": 637611225.0, + "step": 24869 + }, + { + "epoch": 2.0507957450317473, + "grad_norm": 1.8918124437332153, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.7935422658920288, + "num_tokens": 637643264.0, + "step": 24870 + }, + { + "epoch": 2.0508782056567987, + "grad_norm": 1.908469557762146, + "learning_rate": 1e-06, + "loss": 0.6016, + "mean_token_accuracy": 0.8073667287826538, + "num_tokens": 637679719.0, + "step": 24871 + }, + { + "epoch": 2.0509606662818505, + "grad_norm": 1.8476580381393433, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7506999969482422, + "num_tokens": 637712969.0, + "step": 24872 + }, + { + "epoch": 2.051043126906902, + "grad_norm": 2.089827299118042, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.7837432622909546, + "num_tokens": 637742629.0, + "step": 24873 + }, + { + "epoch": 2.0511255875319536, + "grad_norm": 2.1813511848449707, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7445769309997559, + "num_tokens": 637772259.0, + "step": 24874 + }, + { + "epoch": 2.051208048157005, + "grad_norm": 2.1642391681671143, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7819831967353821, + "num_tokens": 637800853.0, + "step": 24875 + }, + { + "epoch": 2.0512905087820568, + "grad_norm": 2.4619433879852295, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7166580557823181, + "num_tokens": 637826965.0, + "step": 24876 + }, + { + "epoch": 2.051372969407108, + "grad_norm": 2.352064609527588, + "learning_rate": 1e-06, + "loss": 0.5566, + "mean_token_accuracy": 0.81136554479599, + "num_tokens": 637849932.0, + "step": 24877 + }, + { + "epoch": 2.0514554300321595, + "grad_norm": 2.098883867263794, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.776458740234375, + "num_tokens": 637880520.0, + "step": 24878 + }, + { + "epoch": 2.0515378906572113, + "grad_norm": 2.2870354652404785, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7705685496330261, + "num_tokens": 637906011.0, + "step": 24879 + }, + { + "epoch": 2.0516203512822626, + "grad_norm": 1.8531378507614136, + "learning_rate": 1e-06, + "loss": 0.5763, + "mean_token_accuracy": 0.8057153820991516, + "num_tokens": 637941281.0, + "step": 24880 + }, + { + "epoch": 2.0517028119073144, + "grad_norm": 2.1992855072021484, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7449625134468079, + "num_tokens": 637969052.0, + "step": 24881 + }, + { + "epoch": 2.0517852725323658, + "grad_norm": 2.154775857925415, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.7960460186004639, + "num_tokens": 637996146.0, + "step": 24882 + }, + { + "epoch": 2.0518677331574176, + "grad_norm": 2.037545680999756, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.7914460897445679, + "num_tokens": 638026455.0, + "step": 24883 + }, + { + "epoch": 2.051950193782469, + "grad_norm": 2.0698797702789307, + "learning_rate": 1e-06, + "loss": 0.5921, + "mean_token_accuracy": 0.7829368114471436, + "num_tokens": 638058314.0, + "step": 24884 + }, + { + "epoch": 2.0520326544075203, + "grad_norm": 2.108668327331543, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.789400577545166, + "num_tokens": 638088312.0, + "step": 24885 + }, + { + "epoch": 2.052115115032572, + "grad_norm": 2.066122055053711, + "learning_rate": 1e-06, + "loss": 0.5296, + "mean_token_accuracy": 0.8171759247779846, + "num_tokens": 638118561.0, + "step": 24886 + }, + { + "epoch": 2.0521975756576234, + "grad_norm": 2.185692310333252, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.773312509059906, + "num_tokens": 638145328.0, + "step": 24887 + }, + { + "epoch": 2.052280036282675, + "grad_norm": 2.095916509628296, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7639034986495972, + "num_tokens": 638176868.0, + "step": 24888 + }, + { + "epoch": 2.0523624969077265, + "grad_norm": 1.8837240934371948, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7804341912269592, + "num_tokens": 638211234.0, + "step": 24889 + }, + { + "epoch": 2.052444957532778, + "grad_norm": 2.0598785877227783, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7562046051025391, + "num_tokens": 638244592.0, + "step": 24890 + }, + { + "epoch": 2.0525274181578297, + "grad_norm": 2.5314786434173584, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7725235223770142, + "num_tokens": 638265918.0, + "step": 24891 + }, + { + "epoch": 2.052609878782881, + "grad_norm": 1.822795033454895, + "learning_rate": 1e-06, + "loss": 0.6002, + "mean_token_accuracy": 0.8114185333251953, + "num_tokens": 638297387.0, + "step": 24892 + }, + { + "epoch": 2.052692339407933, + "grad_norm": 1.9530045986175537, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.764909565448761, + "num_tokens": 638329691.0, + "step": 24893 + }, + { + "epoch": 2.052774800032984, + "grad_norm": 2.0567100048065186, + "learning_rate": 1e-06, + "loss": 0.5006, + "mean_token_accuracy": 0.8301956057548523, + "num_tokens": 638359985.0, + "step": 24894 + }, + { + "epoch": 2.052857260658036, + "grad_norm": 2.151750087738037, + "learning_rate": 1e-06, + "loss": 0.5978, + "mean_token_accuracy": 0.8002879619598389, + "num_tokens": 638389394.0, + "step": 24895 + }, + { + "epoch": 2.0529397212830873, + "grad_norm": 1.951228380203247, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7828872799873352, + "num_tokens": 638422159.0, + "step": 24896 + }, + { + "epoch": 2.0530221819081387, + "grad_norm": 2.0802316665649414, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7570052742958069, + "num_tokens": 638455167.0, + "step": 24897 + }, + { + "epoch": 2.0531046425331905, + "grad_norm": 1.9613209962844849, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7822262048721313, + "num_tokens": 638488835.0, + "step": 24898 + }, + { + "epoch": 2.053187103158242, + "grad_norm": 1.924072265625, + "learning_rate": 1e-06, + "loss": 0.6261, + "mean_token_accuracy": 0.7975346446037292, + "num_tokens": 638525416.0, + "step": 24899 + }, + { + "epoch": 2.0532695637832936, + "grad_norm": 2.490668535232544, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7567853927612305, + "num_tokens": 638548279.0, + "step": 24900 + }, + { + "epoch": 2.053352024408345, + "grad_norm": 2.0086259841918945, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7808222770690918, + "num_tokens": 638581573.0, + "step": 24901 + }, + { + "epoch": 2.0534344850333968, + "grad_norm": 2.0574092864990234, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7564278244972229, + "num_tokens": 638612355.0, + "step": 24902 + }, + { + "epoch": 2.053516945658448, + "grad_norm": 2.2243456840515137, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7455117702484131, + "num_tokens": 638640165.0, + "step": 24903 + }, + { + "epoch": 2.0535994062834995, + "grad_norm": 2.0517897605895996, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7630050778388977, + "num_tokens": 638671354.0, + "step": 24904 + }, + { + "epoch": 2.0536818669085513, + "grad_norm": 2.0354015827178955, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.759709358215332, + "num_tokens": 638700470.0, + "step": 24905 + }, + { + "epoch": 2.0537643275336026, + "grad_norm": 2.05454158782959, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7342007756233215, + "num_tokens": 638731347.0, + "step": 24906 + }, + { + "epoch": 2.0538467881586544, + "grad_norm": 2.0417706966400146, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.756865382194519, + "num_tokens": 638761954.0, + "step": 24907 + }, + { + "epoch": 2.0539292487837058, + "grad_norm": 1.8502672910690308, + "learning_rate": 1e-06, + "loss": 0.5295, + "mean_token_accuracy": 0.8205133676528931, + "num_tokens": 638797926.0, + "step": 24908 + }, + { + "epoch": 2.054011709408757, + "grad_norm": 2.041952133178711, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7679906487464905, + "num_tokens": 638828636.0, + "step": 24909 + }, + { + "epoch": 2.054094170033809, + "grad_norm": 2.158127546310425, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.759676456451416, + "num_tokens": 638857143.0, + "step": 24910 + }, + { + "epoch": 2.0541766306588602, + "grad_norm": 2.000432014465332, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7607958316802979, + "num_tokens": 638890282.0, + "step": 24911 + }, + { + "epoch": 2.054259091283912, + "grad_norm": 1.9209327697753906, + "learning_rate": 1e-06, + "loss": 0.5944, + "mean_token_accuracy": 0.8053518533706665, + "num_tokens": 638921218.0, + "step": 24912 + }, + { + "epoch": 2.0543415519089634, + "grad_norm": 2.411741018295288, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7555866837501526, + "num_tokens": 638946282.0, + "step": 24913 + }, + { + "epoch": 2.054424012534015, + "grad_norm": 2.247667074203491, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7463021278381348, + "num_tokens": 638972199.0, + "step": 24914 + }, + { + "epoch": 2.0545064731590665, + "grad_norm": 2.190368890762329, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7653025388717651, + "num_tokens": 639000061.0, + "step": 24915 + }, + { + "epoch": 2.054588933784118, + "grad_norm": 2.0425214767456055, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7761722803115845, + "num_tokens": 639030632.0, + "step": 24916 + }, + { + "epoch": 2.0546713944091697, + "grad_norm": 1.918839693069458, + "learning_rate": 1e-06, + "loss": 0.5627, + "mean_token_accuracy": 0.8081620931625366, + "num_tokens": 639063627.0, + "step": 24917 + }, + { + "epoch": 2.054753855034221, + "grad_norm": 2.000465154647827, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7632158994674683, + "num_tokens": 639096627.0, + "step": 24918 + }, + { + "epoch": 2.054836315659273, + "grad_norm": 2.1397883892059326, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7293201088905334, + "num_tokens": 639126017.0, + "step": 24919 + }, + { + "epoch": 2.054918776284324, + "grad_norm": 2.2076165676116943, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7810590267181396, + "num_tokens": 639152198.0, + "step": 24920 + }, + { + "epoch": 2.055001236909376, + "grad_norm": 2.1894006729125977, + "learning_rate": 1e-06, + "loss": 0.6037, + "mean_token_accuracy": 0.8052336573600769, + "num_tokens": 639178064.0, + "step": 24921 + }, + { + "epoch": 2.0550836975344273, + "grad_norm": 2.080632209777832, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7811452150344849, + "num_tokens": 639210051.0, + "step": 24922 + }, + { + "epoch": 2.0551661581594787, + "grad_norm": 2.3317530155181885, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.783227264881134, + "num_tokens": 639235030.0, + "step": 24923 + }, + { + "epoch": 2.0552486187845305, + "grad_norm": 2.0210142135620117, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.7987462282180786, + "num_tokens": 639265958.0, + "step": 24924 + }, + { + "epoch": 2.055331079409582, + "grad_norm": 2.1662678718566895, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.7846253514289856, + "num_tokens": 639294207.0, + "step": 24925 + }, + { + "epoch": 2.0554135400346336, + "grad_norm": 2.0234081745147705, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7416407465934753, + "num_tokens": 639325466.0, + "step": 24926 + }, + { + "epoch": 2.055496000659685, + "grad_norm": 1.9920648336410522, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7561960220336914, + "num_tokens": 639358575.0, + "step": 24927 + }, + { + "epoch": 2.0555784612847363, + "grad_norm": 2.1010444164276123, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.7822479605674744, + "num_tokens": 639388021.0, + "step": 24928 + }, + { + "epoch": 2.055660921909788, + "grad_norm": 2.3901567459106445, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7627350091934204, + "num_tokens": 639414826.0, + "step": 24929 + }, + { + "epoch": 2.0557433825348395, + "grad_norm": 2.0597753524780273, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7614481449127197, + "num_tokens": 639443474.0, + "step": 24930 + }, + { + "epoch": 2.0558258431598913, + "grad_norm": 2.018176555633545, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7821602821350098, + "num_tokens": 639473938.0, + "step": 24931 + }, + { + "epoch": 2.0559083037849426, + "grad_norm": 1.9716707468032837, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7707135677337646, + "num_tokens": 639506186.0, + "step": 24932 + }, + { + "epoch": 2.0559907644099944, + "grad_norm": 2.284316301345825, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7647882699966431, + "num_tokens": 639531296.0, + "step": 24933 + }, + { + "epoch": 2.0560732250350457, + "grad_norm": 1.9737052917480469, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.749761700630188, + "num_tokens": 639564734.0, + "step": 24934 + }, + { + "epoch": 2.056155685660097, + "grad_norm": 2.2321388721466064, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7702158093452454, + "num_tokens": 639591863.0, + "step": 24935 + }, + { + "epoch": 2.056238146285149, + "grad_norm": 2.1565847396850586, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.7976378798484802, + "num_tokens": 639618864.0, + "step": 24936 + }, + { + "epoch": 2.0563206069102002, + "grad_norm": 2.103752613067627, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7314262390136719, + "num_tokens": 639650580.0, + "step": 24937 + }, + { + "epoch": 2.056403067535252, + "grad_norm": 1.6408792734146118, + "learning_rate": 1e-06, + "loss": 0.5539, + "mean_token_accuracy": 0.8214510679244995, + "num_tokens": 639691071.0, + "step": 24938 + }, + { + "epoch": 2.0564855281603034, + "grad_norm": 2.5370242595672607, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7561577558517456, + "num_tokens": 639712622.0, + "step": 24939 + }, + { + "epoch": 2.056567988785355, + "grad_norm": 1.9068711996078491, + "learning_rate": 1e-06, + "loss": 0.5456, + "mean_token_accuracy": 0.8066428899765015, + "num_tokens": 639742852.0, + "step": 24940 + }, + { + "epoch": 2.0566504494104065, + "grad_norm": 2.6144728660583496, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7477173805236816, + "num_tokens": 639763916.0, + "step": 24941 + }, + { + "epoch": 2.056732910035458, + "grad_norm": 2.1367409229278564, + "learning_rate": 1e-06, + "loss": 0.5114, + "mean_token_accuracy": 0.8199743628501892, + "num_tokens": 639788265.0, + "step": 24942 + }, + { + "epoch": 2.0568153706605097, + "grad_norm": 2.2125115394592285, + "learning_rate": 1e-06, + "loss": 0.5957, + "mean_token_accuracy": 0.7985305786132812, + "num_tokens": 639816419.0, + "step": 24943 + }, + { + "epoch": 2.056897831285561, + "grad_norm": 2.193234920501709, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7508749961853027, + "num_tokens": 639843789.0, + "step": 24944 + }, + { + "epoch": 2.056980291910613, + "grad_norm": 1.8899468183517456, + "learning_rate": 1e-06, + "loss": 0.5642, + "mean_token_accuracy": 0.8135370016098022, + "num_tokens": 639874306.0, + "step": 24945 + }, + { + "epoch": 2.057062752535664, + "grad_norm": 2.210174083709717, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7658861875534058, + "num_tokens": 639901339.0, + "step": 24946 + }, + { + "epoch": 2.057145213160716, + "grad_norm": 2.024834394454956, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7757183909416199, + "num_tokens": 639934402.0, + "step": 24947 + }, + { + "epoch": 2.0572276737857673, + "grad_norm": 1.8521357774734497, + "learning_rate": 1e-06, + "loss": 0.509, + "mean_token_accuracy": 0.8241719007492065, + "num_tokens": 639970159.0, + "step": 24948 + }, + { + "epoch": 2.0573101344108187, + "grad_norm": 1.8133845329284668, + "learning_rate": 1e-06, + "loss": 0.5382, + "mean_token_accuracy": 0.8112022280693054, + "num_tokens": 640004290.0, + "step": 24949 + }, + { + "epoch": 2.0573925950358705, + "grad_norm": 2.018016815185547, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.7802827954292297, + "num_tokens": 640033566.0, + "step": 24950 + }, + { + "epoch": 2.057475055660922, + "grad_norm": 2.1357791423797607, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.790764570236206, + "num_tokens": 640062132.0, + "step": 24951 + }, + { + "epoch": 2.0575575162859736, + "grad_norm": 1.981852412223816, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7624050378799438, + "num_tokens": 640097380.0, + "step": 24952 + }, + { + "epoch": 2.057639976911025, + "grad_norm": 1.8511784076690674, + "learning_rate": 1e-06, + "loss": 0.5995, + "mean_token_accuracy": 0.7964839935302734, + "num_tokens": 640129632.0, + "step": 24953 + }, + { + "epoch": 2.0577224375360763, + "grad_norm": 1.9838918447494507, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.7931971549987793, + "num_tokens": 640160259.0, + "step": 24954 + }, + { + "epoch": 2.057804898161128, + "grad_norm": 1.9540181159973145, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7876485586166382, + "num_tokens": 640191759.0, + "step": 24955 + }, + { + "epoch": 2.0578873587861795, + "grad_norm": 2.037003993988037, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.781730055809021, + "num_tokens": 640222219.0, + "step": 24956 + }, + { + "epoch": 2.0579698194112312, + "grad_norm": 1.9958305358886719, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7650890350341797, + "num_tokens": 640256882.0, + "step": 24957 + }, + { + "epoch": 2.0580522800362826, + "grad_norm": 1.8786143064498901, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7651998996734619, + "num_tokens": 640290843.0, + "step": 24958 + }, + { + "epoch": 2.0581347406613344, + "grad_norm": 2.019818067550659, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7782440781593323, + "num_tokens": 640321688.0, + "step": 24959 + }, + { + "epoch": 2.0582172012863857, + "grad_norm": 1.8103622198104858, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.789487361907959, + "num_tokens": 640354904.0, + "step": 24960 + }, + { + "epoch": 2.058299661911437, + "grad_norm": 2.036086082458496, + "learning_rate": 1e-06, + "loss": 0.5926, + "mean_token_accuracy": 0.8048760890960693, + "num_tokens": 640386871.0, + "step": 24961 + }, + { + "epoch": 2.058382122536489, + "grad_norm": 2.24869966506958, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7607204914093018, + "num_tokens": 640421564.0, + "step": 24962 + }, + { + "epoch": 2.0584645831615402, + "grad_norm": 2.082329034805298, + "learning_rate": 1e-06, + "loss": 0.4959, + "mean_token_accuracy": 0.8241479396820068, + "num_tokens": 640450946.0, + "step": 24963 + }, + { + "epoch": 2.058547043786592, + "grad_norm": 2.3857781887054443, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7496552467346191, + "num_tokens": 640476945.0, + "step": 24964 + }, + { + "epoch": 2.0586295044116434, + "grad_norm": 2.3102452754974365, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7754073739051819, + "num_tokens": 640501711.0, + "step": 24965 + }, + { + "epoch": 2.058711965036695, + "grad_norm": 1.9255766868591309, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.7865010499954224, + "num_tokens": 640533552.0, + "step": 24966 + }, + { + "epoch": 2.0587944256617465, + "grad_norm": 1.6888612508773804, + "learning_rate": 1e-06, + "loss": 0.5467, + "mean_token_accuracy": 0.8148176670074463, + "num_tokens": 640573784.0, + "step": 24967 + }, + { + "epoch": 2.058876886286798, + "grad_norm": 1.9829431772232056, + "learning_rate": 1e-06, + "loss": 0.5577, + "mean_token_accuracy": 0.8109662532806396, + "num_tokens": 640604156.0, + "step": 24968 + }, + { + "epoch": 2.0589593469118497, + "grad_norm": 2.0953266620635986, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7558774948120117, + "num_tokens": 640634673.0, + "step": 24969 + }, + { + "epoch": 2.059041807536901, + "grad_norm": 2.116847276687622, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7778228521347046, + "num_tokens": 640664456.0, + "step": 24970 + }, + { + "epoch": 2.059124268161953, + "grad_norm": 2.042754888534546, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7899059057235718, + "num_tokens": 640698603.0, + "step": 24971 + }, + { + "epoch": 2.059206728787004, + "grad_norm": 1.9654041528701782, + "learning_rate": 1e-06, + "loss": 0.5524, + "mean_token_accuracy": 0.8102007508277893, + "num_tokens": 640727815.0, + "step": 24972 + }, + { + "epoch": 2.0592891894120555, + "grad_norm": 1.9774904251098633, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7725574970245361, + "num_tokens": 640761898.0, + "step": 24973 + }, + { + "epoch": 2.0593716500371073, + "grad_norm": 1.79801607131958, + "learning_rate": 1e-06, + "loss": 0.5422, + "mean_token_accuracy": 0.8154788017272949, + "num_tokens": 640798362.0, + "step": 24974 + }, + { + "epoch": 2.0594541106621587, + "grad_norm": 2.304901361465454, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.747215747833252, + "num_tokens": 640825196.0, + "step": 24975 + }, + { + "epoch": 2.0595365712872105, + "grad_norm": 2.0488297939300537, + "learning_rate": 1e-06, + "loss": 0.4667, + "mean_token_accuracy": 0.8367688655853271, + "num_tokens": 640853090.0, + "step": 24976 + }, + { + "epoch": 2.059619031912262, + "grad_norm": 2.02890682220459, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7420758008956909, + "num_tokens": 640885089.0, + "step": 24977 + }, + { + "epoch": 2.0597014925373136, + "grad_norm": 2.094666004180908, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.734755277633667, + "num_tokens": 640916257.0, + "step": 24978 + }, + { + "epoch": 2.059783953162365, + "grad_norm": 2.0017642974853516, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.7832691073417664, + "num_tokens": 640946346.0, + "step": 24979 + }, + { + "epoch": 2.0598664137874163, + "grad_norm": 1.9117623567581177, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.77745121717453, + "num_tokens": 640981809.0, + "step": 24980 + }, + { + "epoch": 2.059948874412468, + "grad_norm": 2.005186080932617, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7865485548973083, + "num_tokens": 641013331.0, + "step": 24981 + }, + { + "epoch": 2.0600313350375195, + "grad_norm": 1.8190782070159912, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7756353616714478, + "num_tokens": 641049470.0, + "step": 24982 + }, + { + "epoch": 2.0601137956625712, + "grad_norm": 2.1468164920806885, + "learning_rate": 1e-06, + "loss": 0.5993, + "mean_token_accuracy": 0.789341390132904, + "num_tokens": 641078249.0, + "step": 24983 + }, + { + "epoch": 2.0601962562876226, + "grad_norm": 2.2906384468078613, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7549417018890381, + "num_tokens": 641103325.0, + "step": 24984 + }, + { + "epoch": 2.0602787169126744, + "grad_norm": 2.2647619247436523, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7628811597824097, + "num_tokens": 641129777.0, + "step": 24985 + }, + { + "epoch": 2.0603611775377257, + "grad_norm": 2.0777928829193115, + "learning_rate": 1e-06, + "loss": 0.5269, + "mean_token_accuracy": 0.8161718845367432, + "num_tokens": 641158268.0, + "step": 24986 + }, + { + "epoch": 2.060443638162777, + "grad_norm": 1.8859846591949463, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.7981299757957458, + "num_tokens": 641192059.0, + "step": 24987 + }, + { + "epoch": 2.060526098787829, + "grad_norm": 2.2037744522094727, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7824968695640564, + "num_tokens": 641217961.0, + "step": 24988 + }, + { + "epoch": 2.0606085594128802, + "grad_norm": 2.044851779937744, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7863361835479736, + "num_tokens": 641246602.0, + "step": 24989 + }, + { + "epoch": 2.060691020037932, + "grad_norm": 2.128136396408081, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7823351621627808, + "num_tokens": 641275541.0, + "step": 24990 + }, + { + "epoch": 2.0607734806629834, + "grad_norm": 1.8015861511230469, + "learning_rate": 1e-06, + "loss": 0.5743, + "mean_token_accuracy": 0.8033435344696045, + "num_tokens": 641311321.0, + "step": 24991 + }, + { + "epoch": 2.060855941288035, + "grad_norm": 1.9738988876342773, + "learning_rate": 1e-06, + "loss": 0.5596, + "mean_token_accuracy": 0.8079485297203064, + "num_tokens": 641343326.0, + "step": 24992 + }, + { + "epoch": 2.0609384019130865, + "grad_norm": 2.282827377319336, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.763092041015625, + "num_tokens": 641372176.0, + "step": 24993 + }, + { + "epoch": 2.061020862538138, + "grad_norm": 2.2207388877868652, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7520409822463989, + "num_tokens": 641398451.0, + "step": 24994 + }, + { + "epoch": 2.0611033231631897, + "grad_norm": 2.3375720977783203, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7682147026062012, + "num_tokens": 641423257.0, + "step": 24995 + }, + { + "epoch": 2.061185783788241, + "grad_norm": 2.1841742992401123, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7615796327590942, + "num_tokens": 641449880.0, + "step": 24996 + }, + { + "epoch": 2.061268244413293, + "grad_norm": 2.263017177581787, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7779820561408997, + "num_tokens": 641474884.0, + "step": 24997 + }, + { + "epoch": 2.061350705038344, + "grad_norm": 2.198765993118286, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.7449915409088135, + "num_tokens": 641502180.0, + "step": 24998 + }, + { + "epoch": 2.0614331656633955, + "grad_norm": 1.8408424854278564, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.7915492057800293, + "num_tokens": 641539676.0, + "step": 24999 + }, + { + "epoch": 2.0615156262884473, + "grad_norm": 2.128845691680908, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.7965767979621887, + "num_tokens": 641569676.0, + "step": 25000 + }, + { + "epoch": 2.0615980869134987, + "grad_norm": 1.9767353534698486, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.764045238494873, + "num_tokens": 641600990.0, + "step": 25001 + }, + { + "epoch": 2.0616805475385505, + "grad_norm": 2.0219743251800537, + "learning_rate": 1e-06, + "loss": 0.5372, + "mean_token_accuracy": 0.8131924867630005, + "num_tokens": 641631018.0, + "step": 25002 + }, + { + "epoch": 2.061763008163602, + "grad_norm": 2.3071396350860596, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.773719072341919, + "num_tokens": 641656422.0, + "step": 25003 + }, + { + "epoch": 2.0618454687886536, + "grad_norm": 2.185236930847168, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7492609024047852, + "num_tokens": 641684112.0, + "step": 25004 + }, + { + "epoch": 2.061927929413705, + "grad_norm": 2.0315158367156982, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.7983219623565674, + "num_tokens": 641714408.0, + "step": 25005 + }, + { + "epoch": 2.0620103900387563, + "grad_norm": 2.2582449913024902, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7405592203140259, + "num_tokens": 641741541.0, + "step": 25006 + }, + { + "epoch": 2.062092850663808, + "grad_norm": 2.042609691619873, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7877728939056396, + "num_tokens": 641773149.0, + "step": 25007 + }, + { + "epoch": 2.0621753112888594, + "grad_norm": 2.0873570442199707, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7723599076271057, + "num_tokens": 641800980.0, + "step": 25008 + }, + { + "epoch": 2.0622577719139112, + "grad_norm": 1.906842589378357, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.759415864944458, + "num_tokens": 641837730.0, + "step": 25009 + }, + { + "epoch": 2.0623402325389626, + "grad_norm": 2.1500518321990967, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7790654897689819, + "num_tokens": 641866743.0, + "step": 25010 + }, + { + "epoch": 2.0624226931640144, + "grad_norm": 2.340331792831421, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7430036067962646, + "num_tokens": 641894881.0, + "step": 25011 + }, + { + "epoch": 2.0625051537890657, + "grad_norm": 1.9658501148223877, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7583554983139038, + "num_tokens": 641926575.0, + "step": 25012 + }, + { + "epoch": 2.062587614414117, + "grad_norm": 2.206441640853882, + "learning_rate": 1e-06, + "loss": 0.5894, + "mean_token_accuracy": 0.8050376176834106, + "num_tokens": 641952520.0, + "step": 25013 + }, + { + "epoch": 2.062670075039169, + "grad_norm": 2.3796913623809814, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7229114770889282, + "num_tokens": 641977761.0, + "step": 25014 + }, + { + "epoch": 2.0627525356642202, + "grad_norm": 2.1927311420440674, + "learning_rate": 1e-06, + "loss": 0.5781, + "mean_token_accuracy": 0.8140973448753357, + "num_tokens": 642003825.0, + "step": 25015 + }, + { + "epoch": 2.062834996289272, + "grad_norm": 1.9230064153671265, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7665777206420898, + "num_tokens": 642036729.0, + "step": 25016 + }, + { + "epoch": 2.0629174569143234, + "grad_norm": 2.410093307495117, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.772728443145752, + "num_tokens": 642057975.0, + "step": 25017 + }, + { + "epoch": 2.0629999175393747, + "grad_norm": 2.0950002670288086, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7775960564613342, + "num_tokens": 642090728.0, + "step": 25018 + }, + { + "epoch": 2.0630823781644265, + "grad_norm": 2.143892526626587, + "learning_rate": 1e-06, + "loss": 0.5708, + "mean_token_accuracy": 0.8087414503097534, + "num_tokens": 642117880.0, + "step": 25019 + }, + { + "epoch": 2.063164838789478, + "grad_norm": 2.1433887481689453, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7506356239318848, + "num_tokens": 642148184.0, + "step": 25020 + }, + { + "epoch": 2.0632472994145297, + "grad_norm": 1.9909638166427612, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7805259227752686, + "num_tokens": 642178551.0, + "step": 25021 + }, + { + "epoch": 2.063329760039581, + "grad_norm": 2.0023555755615234, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.773888111114502, + "num_tokens": 642209296.0, + "step": 25022 + }, + { + "epoch": 2.063412220664633, + "grad_norm": 2.166281223297119, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7591531872749329, + "num_tokens": 642237441.0, + "step": 25023 + }, + { + "epoch": 2.063494681289684, + "grad_norm": 1.9197771549224854, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.7658160924911499, + "num_tokens": 642270715.0, + "step": 25024 + }, + { + "epoch": 2.0635771419147355, + "grad_norm": 2.1849236488342285, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.735530436038971, + "num_tokens": 642300745.0, + "step": 25025 + }, + { + "epoch": 2.0636596025397873, + "grad_norm": 2.223787784576416, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7732478976249695, + "num_tokens": 642326876.0, + "step": 25026 + }, + { + "epoch": 2.0637420631648387, + "grad_norm": 1.8945695161819458, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.7884016036987305, + "num_tokens": 642358483.0, + "step": 25027 + }, + { + "epoch": 2.0638245237898905, + "grad_norm": 1.8153408765792847, + "learning_rate": 1e-06, + "loss": 0.4005, + "mean_token_accuracy": 0.8542969226837158, + "num_tokens": 642390698.0, + "step": 25028 + }, + { + "epoch": 2.063906984414942, + "grad_norm": 2.2266244888305664, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.768784761428833, + "num_tokens": 642417842.0, + "step": 25029 + }, + { + "epoch": 2.0639894450399936, + "grad_norm": 2.104111671447754, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7511348724365234, + "num_tokens": 642447257.0, + "step": 25030 + }, + { + "epoch": 2.064071905665045, + "grad_norm": 2.034428596496582, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7741959095001221, + "num_tokens": 642479576.0, + "step": 25031 + }, + { + "epoch": 2.0641543662900963, + "grad_norm": 1.8645920753479004, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.7931579351425171, + "num_tokens": 642517816.0, + "step": 25032 + }, + { + "epoch": 2.064236826915148, + "grad_norm": 2.1227035522460938, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7663931846618652, + "num_tokens": 642548824.0, + "step": 25033 + }, + { + "epoch": 2.0643192875401994, + "grad_norm": 2.219820022583008, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7790905237197876, + "num_tokens": 642576294.0, + "step": 25034 + }, + { + "epoch": 2.0644017481652512, + "grad_norm": 1.8631699085235596, + "learning_rate": 1e-06, + "loss": 0.5858, + "mean_token_accuracy": 0.8043911457061768, + "num_tokens": 642608931.0, + "step": 25035 + }, + { + "epoch": 2.0644842087903026, + "grad_norm": 2.1605308055877686, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.7888438701629639, + "num_tokens": 642636949.0, + "step": 25036 + }, + { + "epoch": 2.0645666694153544, + "grad_norm": 2.003263235092163, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.7852540612220764, + "num_tokens": 642668880.0, + "step": 25037 + }, + { + "epoch": 2.0646491300404057, + "grad_norm": 2.12795352935791, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.768419623374939, + "num_tokens": 642699893.0, + "step": 25038 + }, + { + "epoch": 2.064731590665457, + "grad_norm": 1.8865541219711304, + "learning_rate": 1e-06, + "loss": 0.5436, + "mean_token_accuracy": 0.8195361495018005, + "num_tokens": 642732480.0, + "step": 25039 + }, + { + "epoch": 2.064814051290509, + "grad_norm": 2.286715507507324, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7696566581726074, + "num_tokens": 642758809.0, + "step": 25040 + }, + { + "epoch": 2.0648965119155602, + "grad_norm": 2.1511292457580566, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7697368860244751, + "num_tokens": 642786844.0, + "step": 25041 + }, + { + "epoch": 2.064978972540612, + "grad_norm": 2.065748929977417, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7706835269927979, + "num_tokens": 642815817.0, + "step": 25042 + }, + { + "epoch": 2.0650614331656634, + "grad_norm": 2.3948774337768555, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.7842074036598206, + "num_tokens": 642838344.0, + "step": 25043 + }, + { + "epoch": 2.0651438937907147, + "grad_norm": 2.021287202835083, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7792969942092896, + "num_tokens": 642868867.0, + "step": 25044 + }, + { + "epoch": 2.0652263544157665, + "grad_norm": 2.0993711948394775, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7739653587341309, + "num_tokens": 642901482.0, + "step": 25045 + }, + { + "epoch": 2.065308815040818, + "grad_norm": 1.840979814529419, + "learning_rate": 1e-06, + "loss": 0.5424, + "mean_token_accuracy": 0.8168660402297974, + "num_tokens": 642937348.0, + "step": 25046 + }, + { + "epoch": 2.0653912756658697, + "grad_norm": 2.0665669441223145, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.763211727142334, + "num_tokens": 642968379.0, + "step": 25047 + }, + { + "epoch": 2.065473736290921, + "grad_norm": 1.8587157726287842, + "learning_rate": 1e-06, + "loss": 0.5595, + "mean_token_accuracy": 0.8201537132263184, + "num_tokens": 642999582.0, + "step": 25048 + }, + { + "epoch": 2.065556196915973, + "grad_norm": 2.5017166137695312, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7386999130249023, + "num_tokens": 643023928.0, + "step": 25049 + }, + { + "epoch": 2.065638657541024, + "grad_norm": 1.965967059135437, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7883068919181824, + "num_tokens": 643056244.0, + "step": 25050 + }, + { + "epoch": 2.0657211181660755, + "grad_norm": 2.109004020690918, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7676185369491577, + "num_tokens": 643083811.0, + "step": 25051 + }, + { + "epoch": 2.0658035787911273, + "grad_norm": 1.8857930898666382, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.7967956066131592, + "num_tokens": 643119827.0, + "step": 25052 + }, + { + "epoch": 2.0658860394161787, + "grad_norm": 1.9564305543899536, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7818922996520996, + "num_tokens": 643151346.0, + "step": 25053 + }, + { + "epoch": 2.0659685000412304, + "grad_norm": 2.1423838138580322, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7658773064613342, + "num_tokens": 643181906.0, + "step": 25054 + }, + { + "epoch": 2.066050960666282, + "grad_norm": 1.98600172996521, + "learning_rate": 1e-06, + "loss": 0.5484, + "mean_token_accuracy": 0.8204230070114136, + "num_tokens": 643212675.0, + "step": 25055 + }, + { + "epoch": 2.0661334212913336, + "grad_norm": 2.213850975036621, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7753780484199524, + "num_tokens": 643241297.0, + "step": 25056 + }, + { + "epoch": 2.066215881916385, + "grad_norm": 1.9306734800338745, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7688093781471252, + "num_tokens": 643276776.0, + "step": 25057 + }, + { + "epoch": 2.0662983425414363, + "grad_norm": 2.0212135314941406, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7653976082801819, + "num_tokens": 643310868.0, + "step": 25058 + }, + { + "epoch": 2.066380803166488, + "grad_norm": 2.2890326976776123, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7216659784317017, + "num_tokens": 643339139.0, + "step": 25059 + }, + { + "epoch": 2.0664632637915394, + "grad_norm": 2.3764727115631104, + "learning_rate": 1e-06, + "loss": 0.5705, + "mean_token_accuracy": 0.8077225685119629, + "num_tokens": 643362755.0, + "step": 25060 + }, + { + "epoch": 2.0665457244165912, + "grad_norm": 2.099536657333374, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7686238288879395, + "num_tokens": 643395527.0, + "step": 25061 + }, + { + "epoch": 2.0666281850416426, + "grad_norm": 2.0867559909820557, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7485802173614502, + "num_tokens": 643424364.0, + "step": 25062 + }, + { + "epoch": 2.066710645666694, + "grad_norm": 2.0052433013916016, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7169772386550903, + "num_tokens": 643459797.0, + "step": 25063 + }, + { + "epoch": 2.0667931062917457, + "grad_norm": 2.098919153213501, + "learning_rate": 1e-06, + "loss": 0.6139, + "mean_token_accuracy": 0.7977163791656494, + "num_tokens": 643487690.0, + "step": 25064 + }, + { + "epoch": 2.066875566916797, + "grad_norm": 2.2517189979553223, + "learning_rate": 1e-06, + "loss": 0.6088, + "mean_token_accuracy": 0.790042519569397, + "num_tokens": 643510434.0, + "step": 25065 + }, + { + "epoch": 2.066958027541849, + "grad_norm": 2.033003568649292, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7672717571258545, + "num_tokens": 643540574.0, + "step": 25066 + }, + { + "epoch": 2.0670404881669002, + "grad_norm": 1.9954575300216675, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.758908748626709, + "num_tokens": 643573912.0, + "step": 25067 + }, + { + "epoch": 2.067122948791952, + "grad_norm": 2.0001652240753174, + "learning_rate": 1e-06, + "loss": 0.5663, + "mean_token_accuracy": 0.8045240640640259, + "num_tokens": 643602942.0, + "step": 25068 + }, + { + "epoch": 2.0672054094170034, + "grad_norm": 2.070984125137329, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.8015127778053284, + "num_tokens": 643636747.0, + "step": 25069 + }, + { + "epoch": 2.0672878700420547, + "grad_norm": 1.8793903589248657, + "learning_rate": 1e-06, + "loss": 0.5885, + "mean_token_accuracy": 0.8105591535568237, + "num_tokens": 643671615.0, + "step": 25070 + }, + { + "epoch": 2.0673703306671065, + "grad_norm": 2.094837188720703, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7416731119155884, + "num_tokens": 643700551.0, + "step": 25071 + }, + { + "epoch": 2.067452791292158, + "grad_norm": 2.072547197341919, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7850805521011353, + "num_tokens": 643729073.0, + "step": 25072 + }, + { + "epoch": 2.0675352519172097, + "grad_norm": 2.071830987930298, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.745476245880127, + "num_tokens": 643760942.0, + "step": 25073 + }, + { + "epoch": 2.067617712542261, + "grad_norm": 2.249008893966675, + "learning_rate": 1e-06, + "loss": 0.5411, + "mean_token_accuracy": 0.8132286667823792, + "num_tokens": 643783932.0, + "step": 25074 + }, + { + "epoch": 2.067700173167313, + "grad_norm": 1.9578192234039307, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.7938427925109863, + "num_tokens": 643816857.0, + "step": 25075 + }, + { + "epoch": 2.067782633792364, + "grad_norm": 2.059704065322876, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.7901747226715088, + "num_tokens": 643847453.0, + "step": 25076 + }, + { + "epoch": 2.0678650944174155, + "grad_norm": 2.4872093200683594, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7470405697822571, + "num_tokens": 643881376.0, + "step": 25077 + }, + { + "epoch": 2.0679475550424673, + "grad_norm": 2.2812814712524414, + "learning_rate": 1e-06, + "loss": 0.7465, + "mean_token_accuracy": 0.76387619972229, + "num_tokens": 643906658.0, + "step": 25078 + }, + { + "epoch": 2.0680300156675187, + "grad_norm": 2.1493191719055176, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7861920595169067, + "num_tokens": 643936439.0, + "step": 25079 + }, + { + "epoch": 2.0681124762925704, + "grad_norm": 2.185154438018799, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7557269930839539, + "num_tokens": 643964172.0, + "step": 25080 + }, + { + "epoch": 2.068194936917622, + "grad_norm": 1.741205096244812, + "learning_rate": 1e-06, + "loss": 0.5543, + "mean_token_accuracy": 0.8162490129470825, + "num_tokens": 644002528.0, + "step": 25081 + }, + { + "epoch": 2.0682773975426736, + "grad_norm": 2.0064237117767334, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7591367959976196, + "num_tokens": 644035080.0, + "step": 25082 + }, + { + "epoch": 2.068359858167725, + "grad_norm": 2.066957712173462, + "learning_rate": 1e-06, + "loss": 0.556, + "mean_token_accuracy": 0.816332995891571, + "num_tokens": 644065346.0, + "step": 25083 + }, + { + "epoch": 2.0684423187927763, + "grad_norm": 1.9438822269439697, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7772861123085022, + "num_tokens": 644097035.0, + "step": 25084 + }, + { + "epoch": 2.068524779417828, + "grad_norm": 2.1932520866394043, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7244061231613159, + "num_tokens": 644125466.0, + "step": 25085 + }, + { + "epoch": 2.0686072400428794, + "grad_norm": 1.9666181802749634, + "learning_rate": 1e-06, + "loss": 0.5705, + "mean_token_accuracy": 0.808975338935852, + "num_tokens": 644156771.0, + "step": 25086 + }, + { + "epoch": 2.0686897006679312, + "grad_norm": 2.070158004760742, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7410742044448853, + "num_tokens": 644189446.0, + "step": 25087 + }, + { + "epoch": 2.0687721612929826, + "grad_norm": 2.2905797958374023, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7282394170761108, + "num_tokens": 644215975.0, + "step": 25088 + }, + { + "epoch": 2.068854621918034, + "grad_norm": 2.3272902965545654, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7538483142852783, + "num_tokens": 644241960.0, + "step": 25089 + }, + { + "epoch": 2.0689370825430857, + "grad_norm": 2.225628137588501, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7645425796508789, + "num_tokens": 644268793.0, + "step": 25090 + }, + { + "epoch": 2.069019543168137, + "grad_norm": 2.137728214263916, + "learning_rate": 1e-06, + "loss": 0.7891, + "mean_token_accuracy": 0.7455462217330933, + "num_tokens": 644298237.0, + "step": 25091 + }, + { + "epoch": 2.069102003793189, + "grad_norm": 2.2320029735565186, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7530220150947571, + "num_tokens": 644326861.0, + "step": 25092 + }, + { + "epoch": 2.06918446441824, + "grad_norm": 2.193352460861206, + "learning_rate": 1e-06, + "loss": 0.5749, + "mean_token_accuracy": 0.8112807869911194, + "num_tokens": 644353891.0, + "step": 25093 + }, + { + "epoch": 2.069266925043292, + "grad_norm": 2.0255727767944336, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.7833645343780518, + "num_tokens": 644381127.0, + "step": 25094 + }, + { + "epoch": 2.0693493856683434, + "grad_norm": 2.073943853378296, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7733564972877502, + "num_tokens": 644412568.0, + "step": 25095 + }, + { + "epoch": 2.0694318462933947, + "grad_norm": 2.0919175148010254, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7564936876296997, + "num_tokens": 644445222.0, + "step": 25096 + }, + { + "epoch": 2.0695143069184465, + "grad_norm": 2.240164279937744, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7314484715461731, + "num_tokens": 644473808.0, + "step": 25097 + }, + { + "epoch": 2.069596767543498, + "grad_norm": 2.0212509632110596, + "learning_rate": 1e-06, + "loss": 0.6112, + "mean_token_accuracy": 0.7911419868469238, + "num_tokens": 644504224.0, + "step": 25098 + }, + { + "epoch": 2.0696792281685497, + "grad_norm": 2.0176522731781006, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.7777496576309204, + "num_tokens": 644534006.0, + "step": 25099 + }, + { + "epoch": 2.069761688793601, + "grad_norm": 1.9580345153808594, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.7822684049606323, + "num_tokens": 644566009.0, + "step": 25100 + }, + { + "epoch": 2.069844149418653, + "grad_norm": 2.152620315551758, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7584617733955383, + "num_tokens": 644593130.0, + "step": 25101 + }, + { + "epoch": 2.069926610043704, + "grad_norm": 2.0579211711883545, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.7999095916748047, + "num_tokens": 644623236.0, + "step": 25102 + }, + { + "epoch": 2.0700090706687555, + "grad_norm": 1.7469203472137451, + "learning_rate": 1e-06, + "loss": 0.5493, + "mean_token_accuracy": 0.8225879669189453, + "num_tokens": 644662064.0, + "step": 25103 + }, + { + "epoch": 2.0700915312938073, + "grad_norm": 1.8969056606292725, + "learning_rate": 1e-06, + "loss": 0.5715, + "mean_token_accuracy": 0.8110705018043518, + "num_tokens": 644695910.0, + "step": 25104 + }, + { + "epoch": 2.0701739919188586, + "grad_norm": 2.4554619789123535, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7649743556976318, + "num_tokens": 644720003.0, + "step": 25105 + }, + { + "epoch": 2.0702564525439104, + "grad_norm": 2.3627097606658936, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.7878890037536621, + "num_tokens": 644745369.0, + "step": 25106 + }, + { + "epoch": 2.070338913168962, + "grad_norm": 2.0660228729248047, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7359700798988342, + "num_tokens": 644776472.0, + "step": 25107 + }, + { + "epoch": 2.070421373794013, + "grad_norm": 1.9541479349136353, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.767663300037384, + "num_tokens": 644810518.0, + "step": 25108 + }, + { + "epoch": 2.070503834419065, + "grad_norm": 1.8660509586334229, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7578507661819458, + "num_tokens": 644844973.0, + "step": 25109 + }, + { + "epoch": 2.0705862950441163, + "grad_norm": 2.207357883453369, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.7604108452796936, + "num_tokens": 644875831.0, + "step": 25110 + }, + { + "epoch": 2.070668755669168, + "grad_norm": 2.1428744792938232, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7780182361602783, + "num_tokens": 644906110.0, + "step": 25111 + }, + { + "epoch": 2.0707512162942194, + "grad_norm": 2.137369394302368, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7751653790473938, + "num_tokens": 644937658.0, + "step": 25112 + }, + { + "epoch": 2.0708336769192712, + "grad_norm": 2.1261329650878906, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.7816154360771179, + "num_tokens": 644968526.0, + "step": 25113 + }, + { + "epoch": 2.0709161375443226, + "grad_norm": 2.1946825981140137, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7731063365936279, + "num_tokens": 644994833.0, + "step": 25114 + }, + { + "epoch": 2.070998598169374, + "grad_norm": 1.9853554964065552, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7751035690307617, + "num_tokens": 645026701.0, + "step": 25115 + }, + { + "epoch": 2.0710810587944257, + "grad_norm": 2.1063766479492188, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7810274362564087, + "num_tokens": 645056681.0, + "step": 25116 + }, + { + "epoch": 2.071163519419477, + "grad_norm": 2.5049171447753906, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.7843600511550903, + "num_tokens": 645080005.0, + "step": 25117 + }, + { + "epoch": 2.071245980044529, + "grad_norm": 2.209343433380127, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.78684401512146, + "num_tokens": 645106441.0, + "step": 25118 + }, + { + "epoch": 2.07132844066958, + "grad_norm": 1.872349500656128, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7787427306175232, + "num_tokens": 645141054.0, + "step": 25119 + }, + { + "epoch": 2.071410901294632, + "grad_norm": 1.876003623008728, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.7925549745559692, + "num_tokens": 645175106.0, + "step": 25120 + }, + { + "epoch": 2.0714933619196834, + "grad_norm": 2.128779888153076, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7653307318687439, + "num_tokens": 645202881.0, + "step": 25121 + }, + { + "epoch": 2.0715758225447347, + "grad_norm": 1.945332407951355, + "learning_rate": 1e-06, + "loss": 0.6147, + "mean_token_accuracy": 0.8009849190711975, + "num_tokens": 645234421.0, + "step": 25122 + }, + { + "epoch": 2.0716582831697865, + "grad_norm": 2.4314396381378174, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7628470659255981, + "num_tokens": 645257527.0, + "step": 25123 + }, + { + "epoch": 2.071740743794838, + "grad_norm": 2.0364718437194824, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.7858442068099976, + "num_tokens": 645290249.0, + "step": 25124 + }, + { + "epoch": 2.0718232044198897, + "grad_norm": 2.18133282661438, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7692365646362305, + "num_tokens": 645318595.0, + "step": 25125 + }, + { + "epoch": 2.071905665044941, + "grad_norm": 2.3109257221221924, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7627947926521301, + "num_tokens": 645345745.0, + "step": 25126 + }, + { + "epoch": 2.071988125669993, + "grad_norm": 1.853563904762268, + "learning_rate": 1e-06, + "loss": 0.5041, + "mean_token_accuracy": 0.8276183605194092, + "num_tokens": 645378083.0, + "step": 25127 + }, + { + "epoch": 2.072070586295044, + "grad_norm": 2.0975635051727295, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.7919557094573975, + "num_tokens": 645406668.0, + "step": 25128 + }, + { + "epoch": 2.0721530469200955, + "grad_norm": 2.012178421020508, + "learning_rate": 1e-06, + "loss": 0.6094, + "mean_token_accuracy": 0.7962549328804016, + "num_tokens": 645437519.0, + "step": 25129 + }, + { + "epoch": 2.0722355075451473, + "grad_norm": 1.8144440650939941, + "learning_rate": 1e-06, + "loss": 0.6145, + "mean_token_accuracy": 0.7993066310882568, + "num_tokens": 645472754.0, + "step": 25130 + }, + { + "epoch": 2.0723179681701986, + "grad_norm": 1.9423760175704956, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.7871102094650269, + "num_tokens": 645510667.0, + "step": 25131 + }, + { + "epoch": 2.0724004287952504, + "grad_norm": 2.1537885665893555, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.753921389579773, + "num_tokens": 645538444.0, + "step": 25132 + }, + { + "epoch": 2.072482889420302, + "grad_norm": 2.09261155128479, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7610126733779907, + "num_tokens": 645567837.0, + "step": 25133 + }, + { + "epoch": 2.072565350045353, + "grad_norm": 1.7531511783599854, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.773749053478241, + "num_tokens": 645605987.0, + "step": 25134 + }, + { + "epoch": 2.072647810670405, + "grad_norm": 2.258054733276367, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.7572886943817139, + "num_tokens": 645634464.0, + "step": 25135 + }, + { + "epoch": 2.0727302712954563, + "grad_norm": 2.093916893005371, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7260093092918396, + "num_tokens": 645667214.0, + "step": 25136 + }, + { + "epoch": 2.072812731920508, + "grad_norm": 2.12424898147583, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7497985363006592, + "num_tokens": 645697287.0, + "step": 25137 + }, + { + "epoch": 2.0728951925455594, + "grad_norm": 2.1241397857666016, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.7919223308563232, + "num_tokens": 645726749.0, + "step": 25138 + }, + { + "epoch": 2.0729776531706112, + "grad_norm": 2.3931221961975098, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7534889578819275, + "num_tokens": 645753398.0, + "step": 25139 + }, + { + "epoch": 2.0730601137956626, + "grad_norm": 1.851058840751648, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.7867406606674194, + "num_tokens": 645789152.0, + "step": 25140 + }, + { + "epoch": 2.073142574420714, + "grad_norm": 2.4595329761505127, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7458894848823547, + "num_tokens": 645811810.0, + "step": 25141 + }, + { + "epoch": 2.0732250350457657, + "grad_norm": 1.949493646621704, + "learning_rate": 1e-06, + "loss": 0.6077, + "mean_token_accuracy": 0.797569990158081, + "num_tokens": 645845085.0, + "step": 25142 + }, + { + "epoch": 2.073307495670817, + "grad_norm": 2.397446393966675, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7736064195632935, + "num_tokens": 645868499.0, + "step": 25143 + }, + { + "epoch": 2.073389956295869, + "grad_norm": 2.2505810260772705, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7659900188446045, + "num_tokens": 645892854.0, + "step": 25144 + }, + { + "epoch": 2.07347241692092, + "grad_norm": 2.2929646968841553, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7699189782142639, + "num_tokens": 645918283.0, + "step": 25145 + }, + { + "epoch": 2.073554877545972, + "grad_norm": 1.9896477460861206, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.753241240978241, + "num_tokens": 645951144.0, + "step": 25146 + }, + { + "epoch": 2.0736373381710234, + "grad_norm": 1.8814970254898071, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7492866516113281, + "num_tokens": 645986749.0, + "step": 25147 + }, + { + "epoch": 2.0737197987960747, + "grad_norm": 2.1287097930908203, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7677661776542664, + "num_tokens": 646016868.0, + "step": 25148 + }, + { + "epoch": 2.0738022594211265, + "grad_norm": 2.1281168460845947, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.7987140417098999, + "num_tokens": 646043730.0, + "step": 25149 + }, + { + "epoch": 2.073884720046178, + "grad_norm": 1.9301897287368774, + "learning_rate": 1e-06, + "loss": 0.5669, + "mean_token_accuracy": 0.8144282102584839, + "num_tokens": 646077533.0, + "step": 25150 + }, + { + "epoch": 2.0739671806712296, + "grad_norm": 2.2544400691986084, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7661062479019165, + "num_tokens": 646108843.0, + "step": 25151 + }, + { + "epoch": 2.074049641296281, + "grad_norm": 2.061530113220215, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.7864605188369751, + "num_tokens": 646137411.0, + "step": 25152 + }, + { + "epoch": 2.0741321019213323, + "grad_norm": 1.8497859239578247, + "learning_rate": 1e-06, + "loss": 0.5877, + "mean_token_accuracy": 0.793394923210144, + "num_tokens": 646173491.0, + "step": 25153 + }, + { + "epoch": 2.074214562546384, + "grad_norm": 2.2433791160583496, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7415894269943237, + "num_tokens": 646202961.0, + "step": 25154 + }, + { + "epoch": 2.0742970231714355, + "grad_norm": 2.051708936691284, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.7957954406738281, + "num_tokens": 646234078.0, + "step": 25155 + }, + { + "epoch": 2.0743794837964873, + "grad_norm": 1.9990284442901611, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.7929120659828186, + "num_tokens": 646267771.0, + "step": 25156 + }, + { + "epoch": 2.0744619444215386, + "grad_norm": 2.2627289295196533, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.764553964138031, + "num_tokens": 646293243.0, + "step": 25157 + }, + { + "epoch": 2.0745444050465904, + "grad_norm": 1.9939748048782349, + "learning_rate": 1e-06, + "loss": 0.542, + "mean_token_accuracy": 0.8095704317092896, + "num_tokens": 646322656.0, + "step": 25158 + }, + { + "epoch": 2.074626865671642, + "grad_norm": 1.9226545095443726, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.7812833786010742, + "num_tokens": 646358575.0, + "step": 25159 + }, + { + "epoch": 2.074709326296693, + "grad_norm": 2.1093969345092773, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7603475451469421, + "num_tokens": 646388188.0, + "step": 25160 + }, + { + "epoch": 2.074791786921745, + "grad_norm": 2.0664875507354736, + "learning_rate": 1e-06, + "loss": 0.6044, + "mean_token_accuracy": 0.797467827796936, + "num_tokens": 646420003.0, + "step": 25161 + }, + { + "epoch": 2.0748742475467963, + "grad_norm": 2.0652949810028076, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7632389664649963, + "num_tokens": 646449650.0, + "step": 25162 + }, + { + "epoch": 2.074956708171848, + "grad_norm": 1.9841282367706299, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.7846310138702393, + "num_tokens": 646483293.0, + "step": 25163 + }, + { + "epoch": 2.0750391687968994, + "grad_norm": 2.337846040725708, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.779204785823822, + "num_tokens": 646506099.0, + "step": 25164 + }, + { + "epoch": 2.075121629421951, + "grad_norm": 1.947468876838684, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7612578868865967, + "num_tokens": 646542194.0, + "step": 25165 + }, + { + "epoch": 2.0752040900470026, + "grad_norm": 2.369823932647705, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7264204621315002, + "num_tokens": 646570414.0, + "step": 25166 + }, + { + "epoch": 2.075286550672054, + "grad_norm": 2.2980549335479736, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.7888132929801941, + "num_tokens": 646596112.0, + "step": 25167 + }, + { + "epoch": 2.0753690112971057, + "grad_norm": 2.0377731323242188, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7434098720550537, + "num_tokens": 646627661.0, + "step": 25168 + }, + { + "epoch": 2.075451471922157, + "grad_norm": 2.0395243167877197, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.7838960886001587, + "num_tokens": 646661372.0, + "step": 25169 + }, + { + "epoch": 2.075533932547209, + "grad_norm": 2.0674238204956055, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.783319354057312, + "num_tokens": 646690588.0, + "step": 25170 + }, + { + "epoch": 2.07561639317226, + "grad_norm": 2.0974693298339844, + "learning_rate": 1e-06, + "loss": 0.7712, + "mean_token_accuracy": 0.7537809014320374, + "num_tokens": 646721271.0, + "step": 25171 + }, + { + "epoch": 2.075698853797312, + "grad_norm": 1.9636272192001343, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7534945011138916, + "num_tokens": 646756210.0, + "step": 25172 + }, + { + "epoch": 2.0757813144223634, + "grad_norm": 2.023346424102783, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7759553790092468, + "num_tokens": 646790545.0, + "step": 25173 + }, + { + "epoch": 2.0758637750474147, + "grad_norm": 2.036085844039917, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.760431170463562, + "num_tokens": 646823388.0, + "step": 25174 + }, + { + "epoch": 2.0759462356724665, + "grad_norm": 2.2106809616088867, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7585563063621521, + "num_tokens": 646851312.0, + "step": 25175 + }, + { + "epoch": 2.076028696297518, + "grad_norm": 2.0152268409729004, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.7838307023048401, + "num_tokens": 646882077.0, + "step": 25176 + }, + { + "epoch": 2.0761111569225696, + "grad_norm": 2.326357126235962, + "learning_rate": 1e-06, + "loss": 0.771, + "mean_token_accuracy": 0.7485277652740479, + "num_tokens": 646908440.0, + "step": 25177 + }, + { + "epoch": 2.076193617547621, + "grad_norm": 2.0502772331237793, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7656217813491821, + "num_tokens": 646940057.0, + "step": 25178 + }, + { + "epoch": 2.0762760781726723, + "grad_norm": 2.255397319793701, + "learning_rate": 1e-06, + "loss": 0.6039, + "mean_token_accuracy": 0.789766788482666, + "num_tokens": 646966713.0, + "step": 25179 + }, + { + "epoch": 2.076358538797724, + "grad_norm": 2.143723487854004, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7869800329208374, + "num_tokens": 646994485.0, + "step": 25180 + }, + { + "epoch": 2.0764409994227755, + "grad_norm": 2.107621192932129, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7820110321044922, + "num_tokens": 647025731.0, + "step": 25181 + }, + { + "epoch": 2.0765234600478273, + "grad_norm": 2.256549835205078, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7483925819396973, + "num_tokens": 647053172.0, + "step": 25182 + }, + { + "epoch": 2.0766059206728786, + "grad_norm": 1.7766656875610352, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.775019645690918, + "num_tokens": 647095623.0, + "step": 25183 + }, + { + "epoch": 2.0766883812979304, + "grad_norm": 2.1400890350341797, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7727007269859314, + "num_tokens": 647124627.0, + "step": 25184 + }, + { + "epoch": 2.076770841922982, + "grad_norm": 2.4613871574401855, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7527389526367188, + "num_tokens": 647147059.0, + "step": 25185 + }, + { + "epoch": 2.076853302548033, + "grad_norm": 2.0907340049743652, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.7941586375236511, + "num_tokens": 647177571.0, + "step": 25186 + }, + { + "epoch": 2.076935763173085, + "grad_norm": 2.3979902267456055, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.7475717067718506, + "num_tokens": 647201792.0, + "step": 25187 + }, + { + "epoch": 2.0770182237981363, + "grad_norm": 2.148482084274292, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7495914101600647, + "num_tokens": 647230696.0, + "step": 25188 + }, + { + "epoch": 2.077100684423188, + "grad_norm": 2.223513603210449, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.7507845163345337, + "num_tokens": 647259045.0, + "step": 25189 + }, + { + "epoch": 2.0771831450482394, + "grad_norm": 2.4063751697540283, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7605050206184387, + "num_tokens": 647282343.0, + "step": 25190 + }, + { + "epoch": 2.077265605673291, + "grad_norm": 2.1276097297668457, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7639414668083191, + "num_tokens": 647310973.0, + "step": 25191 + }, + { + "epoch": 2.0773480662983426, + "grad_norm": 2.0208868980407715, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.773836612701416, + "num_tokens": 647342770.0, + "step": 25192 + }, + { + "epoch": 2.077430526923394, + "grad_norm": 2.0059163570404053, + "learning_rate": 1e-06, + "loss": 0.5753, + "mean_token_accuracy": 0.8065610527992249, + "num_tokens": 647374392.0, + "step": 25193 + }, + { + "epoch": 2.0775129875484457, + "grad_norm": 2.2346696853637695, + "learning_rate": 1e-06, + "loss": 0.5332, + "mean_token_accuracy": 0.8137387037277222, + "num_tokens": 647399944.0, + "step": 25194 + }, + { + "epoch": 2.077595448173497, + "grad_norm": 1.8426371812820435, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.7821468114852905, + "num_tokens": 647432158.0, + "step": 25195 + }, + { + "epoch": 2.077677908798549, + "grad_norm": 2.09601092338562, + "learning_rate": 1e-06, + "loss": 0.5634, + "mean_token_accuracy": 0.8092715740203857, + "num_tokens": 647458975.0, + "step": 25196 + }, + { + "epoch": 2.0777603694236, + "grad_norm": 1.8689543008804321, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7501949071884155, + "num_tokens": 647496519.0, + "step": 25197 + }, + { + "epoch": 2.0778428300486516, + "grad_norm": 2.1403307914733887, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7725182771682739, + "num_tokens": 647526032.0, + "step": 25198 + }, + { + "epoch": 2.0779252906737034, + "grad_norm": 2.0141844749450684, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7729429602622986, + "num_tokens": 647558622.0, + "step": 25199 + }, + { + "epoch": 2.0780077512987547, + "grad_norm": 2.247650146484375, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7800905108451843, + "num_tokens": 647586954.0, + "step": 25200 + }, + { + "epoch": 2.0780902119238065, + "grad_norm": 2.018770456314087, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7864770889282227, + "num_tokens": 647618027.0, + "step": 25201 + }, + { + "epoch": 2.078172672548858, + "grad_norm": 2.1263232231140137, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.7969693541526794, + "num_tokens": 647647425.0, + "step": 25202 + }, + { + "epoch": 2.0782551331739096, + "grad_norm": 2.630300283432007, + "learning_rate": 1e-06, + "loss": 0.5344, + "mean_token_accuracy": 0.821155309677124, + "num_tokens": 647671510.0, + "step": 25203 + }, + { + "epoch": 2.078337593798961, + "grad_norm": 1.9822490215301514, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.766124427318573, + "num_tokens": 647706082.0, + "step": 25204 + }, + { + "epoch": 2.0784200544240123, + "grad_norm": 2.1666290760040283, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.767676591873169, + "num_tokens": 647734559.0, + "step": 25205 + }, + { + "epoch": 2.078502515049064, + "grad_norm": 2.374385356903076, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7589352130889893, + "num_tokens": 647759718.0, + "step": 25206 + }, + { + "epoch": 2.0785849756741155, + "grad_norm": 2.273503303527832, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.7877274751663208, + "num_tokens": 647787212.0, + "step": 25207 + }, + { + "epoch": 2.0786674362991673, + "grad_norm": 1.996687650680542, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.7873811721801758, + "num_tokens": 647817511.0, + "step": 25208 + }, + { + "epoch": 2.0787498969242186, + "grad_norm": 2.390247344970703, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7743145227432251, + "num_tokens": 647843798.0, + "step": 25209 + }, + { + "epoch": 2.0788323575492704, + "grad_norm": 2.4580025672912598, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7476566433906555, + "num_tokens": 647867997.0, + "step": 25210 + }, + { + "epoch": 2.0789148181743218, + "grad_norm": 2.3598995208740234, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7686339616775513, + "num_tokens": 647894558.0, + "step": 25211 + }, + { + "epoch": 2.078997278799373, + "grad_norm": 2.300936460494995, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7804169654846191, + "num_tokens": 647919332.0, + "step": 25212 + }, + { + "epoch": 2.079079739424425, + "grad_norm": 2.2502267360687256, + "learning_rate": 1e-06, + "loss": 0.6139, + "mean_token_accuracy": 0.7868729829788208, + "num_tokens": 647944797.0, + "step": 25213 + }, + { + "epoch": 2.0791622000494763, + "grad_norm": 2.1092605590820312, + "learning_rate": 1e-06, + "loss": 0.5686, + "mean_token_accuracy": 0.8019177913665771, + "num_tokens": 647971709.0, + "step": 25214 + }, + { + "epoch": 2.079244660674528, + "grad_norm": 1.9971188306808472, + "learning_rate": 1e-06, + "loss": 0.5511, + "mean_token_accuracy": 0.8144081830978394, + "num_tokens": 647998804.0, + "step": 25215 + }, + { + "epoch": 2.0793271212995794, + "grad_norm": 2.3361330032348633, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7671220898628235, + "num_tokens": 648020829.0, + "step": 25216 + }, + { + "epoch": 2.079409581924631, + "grad_norm": 2.0470564365386963, + "learning_rate": 1e-06, + "loss": 0.5976, + "mean_token_accuracy": 0.7968891859054565, + "num_tokens": 648050938.0, + "step": 25217 + }, + { + "epoch": 2.0794920425496826, + "grad_norm": 2.5362465381622314, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.7913188934326172, + "num_tokens": 648072381.0, + "step": 25218 + }, + { + "epoch": 2.079574503174734, + "grad_norm": 2.134892702102661, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7770224809646606, + "num_tokens": 648103809.0, + "step": 25219 + }, + { + "epoch": 2.0796569637997857, + "grad_norm": 2.0952913761138916, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7630139589309692, + "num_tokens": 648133869.0, + "step": 25220 + }, + { + "epoch": 2.079739424424837, + "grad_norm": 1.9263668060302734, + "learning_rate": 1e-06, + "loss": 0.5121, + "mean_token_accuracy": 0.8276166915893555, + "num_tokens": 648167022.0, + "step": 25221 + }, + { + "epoch": 2.079821885049889, + "grad_norm": 2.20147442817688, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7441132068634033, + "num_tokens": 648195819.0, + "step": 25222 + }, + { + "epoch": 2.07990434567494, + "grad_norm": 2.5589840412139893, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7723506689071655, + "num_tokens": 648218631.0, + "step": 25223 + }, + { + "epoch": 2.0799868062999916, + "grad_norm": 2.0264923572540283, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.8050998449325562, + "num_tokens": 648249381.0, + "step": 25224 + }, + { + "epoch": 2.0800692669250433, + "grad_norm": 2.127094268798828, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.7924293279647827, + "num_tokens": 648278702.0, + "step": 25225 + }, + { + "epoch": 2.0801517275500947, + "grad_norm": 2.3195431232452393, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7693661451339722, + "num_tokens": 648301350.0, + "step": 25226 + }, + { + "epoch": 2.0802341881751465, + "grad_norm": 2.2580108642578125, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7595868706703186, + "num_tokens": 648327965.0, + "step": 25227 + }, + { + "epoch": 2.080316648800198, + "grad_norm": 2.2809295654296875, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.7776522040367126, + "num_tokens": 648357646.0, + "step": 25228 + }, + { + "epoch": 2.0803991094252496, + "grad_norm": 2.212785243988037, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.8009806871414185, + "num_tokens": 648386825.0, + "step": 25229 + }, + { + "epoch": 2.080481570050301, + "grad_norm": 2.226731777191162, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7562994956970215, + "num_tokens": 648415606.0, + "step": 25230 + }, + { + "epoch": 2.0805640306753523, + "grad_norm": 1.9623275995254517, + "learning_rate": 1e-06, + "loss": 0.593, + "mean_token_accuracy": 0.7928133010864258, + "num_tokens": 648446680.0, + "step": 25231 + }, + { + "epoch": 2.080646491300404, + "grad_norm": 2.146223306655884, + "learning_rate": 1e-06, + "loss": 0.594, + "mean_token_accuracy": 0.8039423227310181, + "num_tokens": 648473295.0, + "step": 25232 + }, + { + "epoch": 2.0807289519254555, + "grad_norm": 2.0078158378601074, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7738145589828491, + "num_tokens": 648502005.0, + "step": 25233 + }, + { + "epoch": 2.0808114125505073, + "grad_norm": 2.108487129211426, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7827380895614624, + "num_tokens": 648531083.0, + "step": 25234 + }, + { + "epoch": 2.0808938731755586, + "grad_norm": 1.9680137634277344, + "learning_rate": 1e-06, + "loss": 0.5826, + "mean_token_accuracy": 0.8061254024505615, + "num_tokens": 648559754.0, + "step": 25235 + }, + { + "epoch": 2.0809763338006104, + "grad_norm": 2.017798662185669, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7770543098449707, + "num_tokens": 648592344.0, + "step": 25236 + }, + { + "epoch": 2.0810587944256618, + "grad_norm": 2.2281341552734375, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7405457496643066, + "num_tokens": 648621531.0, + "step": 25237 + }, + { + "epoch": 2.081141255050713, + "grad_norm": 2.0242273807525635, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.7874049544334412, + "num_tokens": 648653437.0, + "step": 25238 + }, + { + "epoch": 2.081223715675765, + "grad_norm": 2.0580978393554688, + "learning_rate": 1e-06, + "loss": 0.5809, + "mean_token_accuracy": 0.8074632883071899, + "num_tokens": 648681587.0, + "step": 25239 + }, + { + "epoch": 2.0813061763008163, + "grad_norm": 1.8754688501358032, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7854191064834595, + "num_tokens": 648715484.0, + "step": 25240 + }, + { + "epoch": 2.081388636925868, + "grad_norm": 1.943063735961914, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.7865829467773438, + "num_tokens": 648748505.0, + "step": 25241 + }, + { + "epoch": 2.0814710975509194, + "grad_norm": 2.1145198345184326, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7558701634407043, + "num_tokens": 648777849.0, + "step": 25242 + }, + { + "epoch": 2.0815535581759708, + "grad_norm": 2.102614164352417, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7682932615280151, + "num_tokens": 648806664.0, + "step": 25243 + }, + { + "epoch": 2.0816360188010226, + "grad_norm": 1.9570882320404053, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7652571797370911, + "num_tokens": 648841939.0, + "step": 25244 + }, + { + "epoch": 2.081718479426074, + "grad_norm": 1.9401122331619263, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7656838297843933, + "num_tokens": 648873393.0, + "step": 25245 + }, + { + "epoch": 2.0818009400511257, + "grad_norm": 2.0813333988189697, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7577167749404907, + "num_tokens": 648903541.0, + "step": 25246 + }, + { + "epoch": 2.081883400676177, + "grad_norm": 2.227254867553711, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7637711763381958, + "num_tokens": 648928153.0, + "step": 25247 + }, + { + "epoch": 2.081965861301229, + "grad_norm": 2.1932499408721924, + "learning_rate": 1e-06, + "loss": 0.4986, + "mean_token_accuracy": 0.8315526843070984, + "num_tokens": 648952252.0, + "step": 25248 + }, + { + "epoch": 2.08204832192628, + "grad_norm": 1.9757294654846191, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7622044086456299, + "num_tokens": 648983257.0, + "step": 25249 + }, + { + "epoch": 2.0821307825513315, + "grad_norm": 2.0078136920928955, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.7924380898475647, + "num_tokens": 649016629.0, + "step": 25250 + }, + { + "epoch": 2.0822132431763833, + "grad_norm": 1.854634165763855, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7647799849510193, + "num_tokens": 649051244.0, + "step": 25251 + }, + { + "epoch": 2.0822957038014347, + "grad_norm": 2.1586549282073975, + "learning_rate": 1e-06, + "loss": 0.8549, + "mean_token_accuracy": 0.7348894476890564, + "num_tokens": 649079221.0, + "step": 25252 + }, + { + "epoch": 2.0823781644264865, + "grad_norm": 2.4331281185150146, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7726202011108398, + "num_tokens": 649102429.0, + "step": 25253 + }, + { + "epoch": 2.082460625051538, + "grad_norm": 2.126089334487915, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7759509086608887, + "num_tokens": 649130136.0, + "step": 25254 + }, + { + "epoch": 2.0825430856765896, + "grad_norm": 2.3779733180999756, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7799304723739624, + "num_tokens": 649154169.0, + "step": 25255 + }, + { + "epoch": 2.082625546301641, + "grad_norm": 2.2572643756866455, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7699143886566162, + "num_tokens": 649178266.0, + "step": 25256 + }, + { + "epoch": 2.0827080069266923, + "grad_norm": 2.3841371536254883, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.728229284286499, + "num_tokens": 649205083.0, + "step": 25257 + }, + { + "epoch": 2.082790467551744, + "grad_norm": 1.9457755088806152, + "learning_rate": 1e-06, + "loss": 0.5746, + "mean_token_accuracy": 0.8102452158927917, + "num_tokens": 649238639.0, + "step": 25258 + }, + { + "epoch": 2.0828729281767955, + "grad_norm": 1.8945469856262207, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7243348360061646, + "num_tokens": 649274697.0, + "step": 25259 + }, + { + "epoch": 2.0829553888018473, + "grad_norm": 2.069063901901245, + "learning_rate": 1e-06, + "loss": 0.7967, + "mean_token_accuracy": 0.7461214065551758, + "num_tokens": 649305501.0, + "step": 25260 + }, + { + "epoch": 2.0830378494268986, + "grad_norm": 1.9258759021759033, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7767698764801025, + "num_tokens": 649340052.0, + "step": 25261 + }, + { + "epoch": 2.0831203100519504, + "grad_norm": 2.0420258045196533, + "learning_rate": 1e-06, + "loss": 0.5884, + "mean_token_accuracy": 0.7998181581497192, + "num_tokens": 649372294.0, + "step": 25262 + }, + { + "epoch": 2.0832027706770018, + "grad_norm": 2.3708088397979736, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7494220733642578, + "num_tokens": 649399189.0, + "step": 25263 + }, + { + "epoch": 2.083285231302053, + "grad_norm": 2.194248914718628, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7673666477203369, + "num_tokens": 649429688.0, + "step": 25264 + }, + { + "epoch": 2.083367691927105, + "grad_norm": 2.074140787124634, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7626035213470459, + "num_tokens": 649460249.0, + "step": 25265 + }, + { + "epoch": 2.0834501525521563, + "grad_norm": 2.034885883331299, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7314211130142212, + "num_tokens": 649494012.0, + "step": 25266 + }, + { + "epoch": 2.083532613177208, + "grad_norm": 2.014410972595215, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.7868971228599548, + "num_tokens": 649522969.0, + "step": 25267 + }, + { + "epoch": 2.0836150738022594, + "grad_norm": 1.9458469152450562, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.775322437286377, + "num_tokens": 649556987.0, + "step": 25268 + }, + { + "epoch": 2.0836975344273108, + "grad_norm": 2.014688730239868, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.7943087220191956, + "num_tokens": 649586516.0, + "step": 25269 + }, + { + "epoch": 2.0837799950523626, + "grad_norm": 2.24619722366333, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7563307285308838, + "num_tokens": 649613948.0, + "step": 25270 + }, + { + "epoch": 2.083862455677414, + "grad_norm": 1.9956549406051636, + "learning_rate": 1e-06, + "loss": 0.5679, + "mean_token_accuracy": 0.8038536906242371, + "num_tokens": 649642827.0, + "step": 25271 + }, + { + "epoch": 2.0839449163024657, + "grad_norm": 2.0819272994995117, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7512997388839722, + "num_tokens": 649674181.0, + "step": 25272 + }, + { + "epoch": 2.084027376927517, + "grad_norm": 2.3001532554626465, + "learning_rate": 1e-06, + "loss": 0.5795, + "mean_token_accuracy": 0.8034510612487793, + "num_tokens": 649697625.0, + "step": 25273 + }, + { + "epoch": 2.084109837552569, + "grad_norm": 2.0872700214385986, + "learning_rate": 1e-06, + "loss": 0.6075, + "mean_token_accuracy": 0.7958971261978149, + "num_tokens": 649728965.0, + "step": 25274 + }, + { + "epoch": 2.08419229817762, + "grad_norm": 2.145785331726074, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7756418585777283, + "num_tokens": 649758265.0, + "step": 25275 + }, + { + "epoch": 2.0842747588026715, + "grad_norm": 2.084019422531128, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.7846874594688416, + "num_tokens": 649787539.0, + "step": 25276 + }, + { + "epoch": 2.0843572194277233, + "grad_norm": 1.966936707496643, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7613369226455688, + "num_tokens": 649821326.0, + "step": 25277 + }, + { + "epoch": 2.0844396800527747, + "grad_norm": 2.0819356441497803, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7755722999572754, + "num_tokens": 649852496.0, + "step": 25278 + }, + { + "epoch": 2.0845221406778265, + "grad_norm": 2.1345374584198, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7546507716178894, + "num_tokens": 649881711.0, + "step": 25279 + }, + { + "epoch": 2.084604601302878, + "grad_norm": 1.7869616746902466, + "learning_rate": 1e-06, + "loss": 0.5536, + "mean_token_accuracy": 0.8075583577156067, + "num_tokens": 649918316.0, + "step": 25280 + }, + { + "epoch": 2.084687061927929, + "grad_norm": 2.033019542694092, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7628403306007385, + "num_tokens": 649949174.0, + "step": 25281 + }, + { + "epoch": 2.084769522552981, + "grad_norm": 2.13822865486145, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7836401462554932, + "num_tokens": 649976466.0, + "step": 25282 + }, + { + "epoch": 2.0848519831780323, + "grad_norm": 2.0620994567871094, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7700729370117188, + "num_tokens": 650005681.0, + "step": 25283 + }, + { + "epoch": 2.084934443803084, + "grad_norm": 2.2086338996887207, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7549982070922852, + "num_tokens": 650033361.0, + "step": 25284 + }, + { + "epoch": 2.0850169044281355, + "grad_norm": 1.952956199645996, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7670167684555054, + "num_tokens": 650066096.0, + "step": 25285 + }, + { + "epoch": 2.0850993650531873, + "grad_norm": 2.009606122970581, + "learning_rate": 1e-06, + "loss": 0.5808, + "mean_token_accuracy": 0.7985454797744751, + "num_tokens": 650094556.0, + "step": 25286 + }, + { + "epoch": 2.0851818256782386, + "grad_norm": 2.1726388931274414, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.7915985584259033, + "num_tokens": 650124060.0, + "step": 25287 + }, + { + "epoch": 2.08526428630329, + "grad_norm": 2.526674509048462, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.7867565751075745, + "num_tokens": 650144628.0, + "step": 25288 + }, + { + "epoch": 2.0853467469283418, + "grad_norm": 1.918912410736084, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7664570212364197, + "num_tokens": 650178994.0, + "step": 25289 + }, + { + "epoch": 2.085429207553393, + "grad_norm": 2.0904500484466553, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.7887909412384033, + "num_tokens": 650210660.0, + "step": 25290 + }, + { + "epoch": 2.085511668178445, + "grad_norm": 2.05312180519104, + "learning_rate": 1e-06, + "loss": 0.5902, + "mean_token_accuracy": 0.8047380447387695, + "num_tokens": 650242743.0, + "step": 25291 + }, + { + "epoch": 2.0855941288034963, + "grad_norm": 2.178675651550293, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7349292039871216, + "num_tokens": 650270700.0, + "step": 25292 + }, + { + "epoch": 2.085676589428548, + "grad_norm": 2.1222124099731445, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7651246786117554, + "num_tokens": 650299520.0, + "step": 25293 + }, + { + "epoch": 2.0857590500535994, + "grad_norm": 2.0120351314544678, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7543533444404602, + "num_tokens": 650329696.0, + "step": 25294 + }, + { + "epoch": 2.0858415106786508, + "grad_norm": 2.1960458755493164, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.795779824256897, + "num_tokens": 650355625.0, + "step": 25295 + }, + { + "epoch": 2.0859239713037026, + "grad_norm": 1.9066616296768188, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7620327472686768, + "num_tokens": 650390404.0, + "step": 25296 + }, + { + "epoch": 2.086006431928754, + "grad_norm": 2.1679701805114746, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7680150270462036, + "num_tokens": 650419295.0, + "step": 25297 + }, + { + "epoch": 2.0860888925538057, + "grad_norm": 1.8467618227005005, + "learning_rate": 1e-06, + "loss": 0.57, + "mean_token_accuracy": 0.8130268454551697, + "num_tokens": 650452477.0, + "step": 25298 + }, + { + "epoch": 2.086171353178857, + "grad_norm": 2.236380100250244, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7328990697860718, + "num_tokens": 650477121.0, + "step": 25299 + }, + { + "epoch": 2.086253813803909, + "grad_norm": 2.3865203857421875, + "learning_rate": 1e-06, + "loss": 0.823, + "mean_token_accuracy": 0.7339698672294617, + "num_tokens": 650501328.0, + "step": 25300 + }, + { + "epoch": 2.08633627442896, + "grad_norm": 2.1738762855529785, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.761383056640625, + "num_tokens": 650532547.0, + "step": 25301 + }, + { + "epoch": 2.0864187350540115, + "grad_norm": 2.052412509918213, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.767106831073761, + "num_tokens": 650564270.0, + "step": 25302 + }, + { + "epoch": 2.0865011956790633, + "grad_norm": 2.2342782020568848, + "learning_rate": 1e-06, + "loss": 0.5547, + "mean_token_accuracy": 0.8071595430374146, + "num_tokens": 650589721.0, + "step": 25303 + }, + { + "epoch": 2.0865836563041147, + "grad_norm": 2.257463216781616, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7494534850120544, + "num_tokens": 650616053.0, + "step": 25304 + }, + { + "epoch": 2.0866661169291665, + "grad_norm": 1.8398939371109009, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7761731743812561, + "num_tokens": 650652256.0, + "step": 25305 + }, + { + "epoch": 2.086748577554218, + "grad_norm": 1.9725680351257324, + "learning_rate": 1e-06, + "loss": 0.4975, + "mean_token_accuracy": 0.8283164501190186, + "num_tokens": 650682188.0, + "step": 25306 + }, + { + "epoch": 2.0868310381792696, + "grad_norm": 1.8812118768692017, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.8022907972335815, + "num_tokens": 650715645.0, + "step": 25307 + }, + { + "epoch": 2.086913498804321, + "grad_norm": 2.1329970359802246, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.7845299243927002, + "num_tokens": 650745734.0, + "step": 25308 + }, + { + "epoch": 2.0869959594293723, + "grad_norm": 2.0404012203216553, + "learning_rate": 1e-06, + "loss": 0.5885, + "mean_token_accuracy": 0.7926389575004578, + "num_tokens": 650775357.0, + "step": 25309 + }, + { + "epoch": 2.087078420054424, + "grad_norm": 1.9963977336883545, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.7936116456985474, + "num_tokens": 650808062.0, + "step": 25310 + }, + { + "epoch": 2.0871608806794755, + "grad_norm": 2.095872402191162, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7487239837646484, + "num_tokens": 650839628.0, + "step": 25311 + }, + { + "epoch": 2.0872433413045273, + "grad_norm": 1.9440927505493164, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7798023223876953, + "num_tokens": 650871134.0, + "step": 25312 + }, + { + "epoch": 2.0873258019295786, + "grad_norm": 2.1894402503967285, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7664781808853149, + "num_tokens": 650897374.0, + "step": 25313 + }, + { + "epoch": 2.08740826255463, + "grad_norm": 1.9402720928192139, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.7914682626724243, + "num_tokens": 650932521.0, + "step": 25314 + }, + { + "epoch": 2.0874907231796818, + "grad_norm": 2.229451894760132, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.7844899892807007, + "num_tokens": 650959050.0, + "step": 25315 + }, + { + "epoch": 2.087573183804733, + "grad_norm": 1.997457504272461, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7786866426467896, + "num_tokens": 650989531.0, + "step": 25316 + }, + { + "epoch": 2.087655644429785, + "grad_norm": 2.4598186016082764, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7748239040374756, + "num_tokens": 651010941.0, + "step": 25317 + }, + { + "epoch": 2.0877381050548363, + "grad_norm": 1.8565375804901123, + "learning_rate": 1e-06, + "loss": 0.5548, + "mean_token_accuracy": 0.8102654218673706, + "num_tokens": 651047635.0, + "step": 25318 + }, + { + "epoch": 2.087820565679888, + "grad_norm": 2.088484764099121, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7755913138389587, + "num_tokens": 651077597.0, + "step": 25319 + }, + { + "epoch": 2.0879030263049394, + "grad_norm": 1.9794197082519531, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.7918126583099365, + "num_tokens": 651109081.0, + "step": 25320 + }, + { + "epoch": 2.0879854869299908, + "grad_norm": 2.0386884212493896, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7679944038391113, + "num_tokens": 651139128.0, + "step": 25321 + }, + { + "epoch": 2.0880679475550425, + "grad_norm": 2.076817035675049, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7485308647155762, + "num_tokens": 651171233.0, + "step": 25322 + }, + { + "epoch": 2.088150408180094, + "grad_norm": 2.3778176307678223, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7543432712554932, + "num_tokens": 651196602.0, + "step": 25323 + }, + { + "epoch": 2.0882328688051457, + "grad_norm": 1.8023715019226074, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7675037384033203, + "num_tokens": 651236534.0, + "step": 25324 + }, + { + "epoch": 2.088315329430197, + "grad_norm": 2.1688613891601562, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7728891968727112, + "num_tokens": 651265534.0, + "step": 25325 + }, + { + "epoch": 2.0883977900552484, + "grad_norm": 2.2769322395324707, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.7852714657783508, + "num_tokens": 651289032.0, + "step": 25326 + }, + { + "epoch": 2.0884802506803, + "grad_norm": 2.1876697540283203, + "learning_rate": 1e-06, + "loss": 0.6361, + "mean_token_accuracy": 0.7937148213386536, + "num_tokens": 651315207.0, + "step": 25327 + }, + { + "epoch": 2.0885627113053515, + "grad_norm": 2.1038169860839844, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7550903558731079, + "num_tokens": 651347909.0, + "step": 25328 + }, + { + "epoch": 2.0886451719304033, + "grad_norm": 1.9182519912719727, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7999587655067444, + "num_tokens": 651378674.0, + "step": 25329 + }, + { + "epoch": 2.0887276325554547, + "grad_norm": 1.9184825420379639, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7618234753608704, + "num_tokens": 651413051.0, + "step": 25330 + }, + { + "epoch": 2.0888100931805065, + "grad_norm": 2.4303174018859863, + "learning_rate": 1e-06, + "loss": 0.5539, + "mean_token_accuracy": 0.8040704727172852, + "num_tokens": 651434069.0, + "step": 25331 + }, + { + "epoch": 2.088892553805558, + "grad_norm": 1.9078083038330078, + "learning_rate": 1e-06, + "loss": 0.5926, + "mean_token_accuracy": 0.8043482303619385, + "num_tokens": 651467334.0, + "step": 25332 + }, + { + "epoch": 2.088975014430609, + "grad_norm": 1.989794373512268, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.7801555395126343, + "num_tokens": 651495025.0, + "step": 25333 + }, + { + "epoch": 2.089057475055661, + "grad_norm": 2.381222724914551, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7783004641532898, + "num_tokens": 651521329.0, + "step": 25334 + }, + { + "epoch": 2.0891399356807123, + "grad_norm": 2.125699996948242, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.7863966226577759, + "num_tokens": 651549170.0, + "step": 25335 + }, + { + "epoch": 2.089222396305764, + "grad_norm": 1.7189867496490479, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7819420695304871, + "num_tokens": 651589765.0, + "step": 25336 + }, + { + "epoch": 2.0893048569308155, + "grad_norm": 1.9987117052078247, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.7900252342224121, + "num_tokens": 651622196.0, + "step": 25337 + }, + { + "epoch": 2.0893873175558673, + "grad_norm": 2.3084027767181396, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7756661772727966, + "num_tokens": 651648588.0, + "step": 25338 + }, + { + "epoch": 2.0894697781809186, + "grad_norm": 1.8755873441696167, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7817379236221313, + "num_tokens": 651683245.0, + "step": 25339 + }, + { + "epoch": 2.08955223880597, + "grad_norm": 2.0773510932922363, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7536075115203857, + "num_tokens": 651715675.0, + "step": 25340 + }, + { + "epoch": 2.0896346994310218, + "grad_norm": 2.0258147716522217, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.7918264269828796, + "num_tokens": 651744457.0, + "step": 25341 + }, + { + "epoch": 2.089717160056073, + "grad_norm": 1.665984034538269, + "learning_rate": 1e-06, + "loss": 0.4735, + "mean_token_accuracy": 0.8385941982269287, + "num_tokens": 651782282.0, + "step": 25342 + }, + { + "epoch": 2.089799620681125, + "grad_norm": 2.051462173461914, + "learning_rate": 1e-06, + "loss": 0.5569, + "mean_token_accuracy": 0.8144163489341736, + "num_tokens": 651810926.0, + "step": 25343 + }, + { + "epoch": 2.0898820813061763, + "grad_norm": 1.8780651092529297, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7861347198486328, + "num_tokens": 651844913.0, + "step": 25344 + }, + { + "epoch": 2.089964541931228, + "grad_norm": 1.8948818445205688, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7736889719963074, + "num_tokens": 651877170.0, + "step": 25345 + }, + { + "epoch": 2.0900470025562794, + "grad_norm": 2.0833775997161865, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7542039155960083, + "num_tokens": 651910337.0, + "step": 25346 + }, + { + "epoch": 2.0901294631813307, + "grad_norm": 2.0079712867736816, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7481521368026733, + "num_tokens": 651946239.0, + "step": 25347 + }, + { + "epoch": 2.0902119238063825, + "grad_norm": 2.0214345455169678, + "learning_rate": 1e-06, + "loss": 0.5799, + "mean_token_accuracy": 0.80255126953125, + "num_tokens": 651979992.0, + "step": 25348 + }, + { + "epoch": 2.090294384431434, + "grad_norm": 1.9574772119522095, + "learning_rate": 1e-06, + "loss": 0.5969, + "mean_token_accuracy": 0.7953262329101562, + "num_tokens": 652010289.0, + "step": 25349 + }, + { + "epoch": 2.0903768450564857, + "grad_norm": 2.1920907497406006, + "learning_rate": 1e-06, + "loss": 0.784, + "mean_token_accuracy": 0.7460614442825317, + "num_tokens": 652039621.0, + "step": 25350 + }, + { + "epoch": 2.090459305681537, + "grad_norm": 2.686593532562256, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.731295108795166, + "num_tokens": 652059563.0, + "step": 25351 + }, + { + "epoch": 2.0905417663065884, + "grad_norm": 2.2874984741210938, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7779577970504761, + "num_tokens": 652084655.0, + "step": 25352 + }, + { + "epoch": 2.09062422693164, + "grad_norm": 1.9021213054656982, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.7815187573432922, + "num_tokens": 652122905.0, + "step": 25353 + }, + { + "epoch": 2.0907066875566915, + "grad_norm": 2.306446075439453, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.7741108536720276, + "num_tokens": 652146069.0, + "step": 25354 + }, + { + "epoch": 2.0907891481817433, + "grad_norm": 2.338146924972534, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7580447196960449, + "num_tokens": 652174834.0, + "step": 25355 + }, + { + "epoch": 2.0908716088067947, + "grad_norm": 2.156254529953003, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7715691328048706, + "num_tokens": 652206730.0, + "step": 25356 + }, + { + "epoch": 2.0909540694318465, + "grad_norm": 1.9843090772628784, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7604271173477173, + "num_tokens": 652236687.0, + "step": 25357 + }, + { + "epoch": 2.091036530056898, + "grad_norm": 2.2045400142669678, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.763924777507782, + "num_tokens": 652261190.0, + "step": 25358 + }, + { + "epoch": 2.091118990681949, + "grad_norm": 2.149223566055298, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.7935296893119812, + "num_tokens": 652288960.0, + "step": 25359 + }, + { + "epoch": 2.091201451307001, + "grad_norm": 2.2441182136535645, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.7829419374465942, + "num_tokens": 652315685.0, + "step": 25360 + }, + { + "epoch": 2.0912839119320523, + "grad_norm": 2.071743965148926, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7671747207641602, + "num_tokens": 652348221.0, + "step": 25361 + }, + { + "epoch": 2.091366372557104, + "grad_norm": 1.9633679389953613, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7641850709915161, + "num_tokens": 652380322.0, + "step": 25362 + }, + { + "epoch": 2.0914488331821555, + "grad_norm": 2.1571404933929443, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.796955406665802, + "num_tokens": 652407345.0, + "step": 25363 + }, + { + "epoch": 2.0915312938072073, + "grad_norm": 2.106016159057617, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7653138637542725, + "num_tokens": 652441083.0, + "step": 25364 + }, + { + "epoch": 2.0916137544322586, + "grad_norm": 2.07175874710083, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7483333349227905, + "num_tokens": 652471811.0, + "step": 25365 + }, + { + "epoch": 2.09169621505731, + "grad_norm": 2.234114408493042, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7631998658180237, + "num_tokens": 652499434.0, + "step": 25366 + }, + { + "epoch": 2.0917786756823618, + "grad_norm": 2.112440586090088, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7699953317642212, + "num_tokens": 652530510.0, + "step": 25367 + }, + { + "epoch": 2.091861136307413, + "grad_norm": 2.287458896636963, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7677779197692871, + "num_tokens": 652556795.0, + "step": 25368 + }, + { + "epoch": 2.091943596932465, + "grad_norm": 1.934719443321228, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7412193417549133, + "num_tokens": 652592062.0, + "step": 25369 + }, + { + "epoch": 2.0920260575575162, + "grad_norm": 2.2405753135681152, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.767852783203125, + "num_tokens": 652619144.0, + "step": 25370 + }, + { + "epoch": 2.0921085181825676, + "grad_norm": 2.1501822471618652, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7702528238296509, + "num_tokens": 652647282.0, + "step": 25371 + }, + { + "epoch": 2.0921909788076194, + "grad_norm": 2.0186350345611572, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7594641447067261, + "num_tokens": 652680198.0, + "step": 25372 + }, + { + "epoch": 2.0922734394326707, + "grad_norm": 2.4993200302124023, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7404114603996277, + "num_tokens": 652703581.0, + "step": 25373 + }, + { + "epoch": 2.0923559000577225, + "grad_norm": 2.0541179180145264, + "learning_rate": 1e-06, + "loss": 0.7585, + "mean_token_accuracy": 0.7544034719467163, + "num_tokens": 652732429.0, + "step": 25374 + }, + { + "epoch": 2.092438360682774, + "grad_norm": 1.9332681894302368, + "learning_rate": 1e-06, + "loss": 0.5706, + "mean_token_accuracy": 0.8139966130256653, + "num_tokens": 652762875.0, + "step": 25375 + }, + { + "epoch": 2.0925208213078257, + "grad_norm": 2.033724069595337, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7698401808738708, + "num_tokens": 652799352.0, + "step": 25376 + }, + { + "epoch": 2.092603281932877, + "grad_norm": 1.8100203275680542, + "learning_rate": 1e-06, + "loss": 0.5807, + "mean_token_accuracy": 0.7934385538101196, + "num_tokens": 652834587.0, + "step": 25377 + }, + { + "epoch": 2.0926857425579284, + "grad_norm": 2.21958589553833, + "learning_rate": 1e-06, + "loss": 0.5715, + "mean_token_accuracy": 0.8037468194961548, + "num_tokens": 652860611.0, + "step": 25378 + }, + { + "epoch": 2.09276820318298, + "grad_norm": 2.0170116424560547, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7753818035125732, + "num_tokens": 652892746.0, + "step": 25379 + }, + { + "epoch": 2.0928506638080315, + "grad_norm": 2.0500380992889404, + "learning_rate": 1e-06, + "loss": 0.5494, + "mean_token_accuracy": 0.8106182813644409, + "num_tokens": 652921916.0, + "step": 25380 + }, + { + "epoch": 2.0929331244330833, + "grad_norm": 2.003675937652588, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.7857388257980347, + "num_tokens": 652954032.0, + "step": 25381 + }, + { + "epoch": 2.0930155850581347, + "grad_norm": 1.8985320329666138, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.7880277037620544, + "num_tokens": 652989979.0, + "step": 25382 + }, + { + "epoch": 2.0930980456831865, + "grad_norm": 2.0604405403137207, + "learning_rate": 1e-06, + "loss": 0.6147, + "mean_token_accuracy": 0.7961401343345642, + "num_tokens": 653021239.0, + "step": 25383 + }, + { + "epoch": 2.093180506308238, + "grad_norm": 2.2392215728759766, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7638916373252869, + "num_tokens": 653049935.0, + "step": 25384 + }, + { + "epoch": 2.093262966933289, + "grad_norm": 2.3724026679992676, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7438057661056519, + "num_tokens": 653076974.0, + "step": 25385 + }, + { + "epoch": 2.093345427558341, + "grad_norm": 2.3349900245666504, + "learning_rate": 1e-06, + "loss": 0.7445, + "mean_token_accuracy": 0.7611687183380127, + "num_tokens": 653102947.0, + "step": 25386 + }, + { + "epoch": 2.0934278881833923, + "grad_norm": 1.9209691286087036, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7767013311386108, + "num_tokens": 653136083.0, + "step": 25387 + }, + { + "epoch": 2.093510348808444, + "grad_norm": 1.8705822229385376, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7892366647720337, + "num_tokens": 653170827.0, + "step": 25388 + }, + { + "epoch": 2.0935928094334955, + "grad_norm": 2.33780574798584, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7497894763946533, + "num_tokens": 653195378.0, + "step": 25389 + }, + { + "epoch": 2.0936752700585473, + "grad_norm": 2.3660390377044678, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.7882663607597351, + "num_tokens": 653221658.0, + "step": 25390 + }, + { + "epoch": 2.0937577306835986, + "grad_norm": 1.753404974937439, + "learning_rate": 1e-06, + "loss": 0.5402, + "mean_token_accuracy": 0.8227452039718628, + "num_tokens": 653259405.0, + "step": 25391 + }, + { + "epoch": 2.09384019130865, + "grad_norm": 2.058589220046997, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.8035471439361572, + "num_tokens": 653287916.0, + "step": 25392 + }, + { + "epoch": 2.0939226519337018, + "grad_norm": 2.060875654220581, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7718149423599243, + "num_tokens": 653321209.0, + "step": 25393 + }, + { + "epoch": 2.094005112558753, + "grad_norm": 2.1146934032440186, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7729216814041138, + "num_tokens": 653349300.0, + "step": 25394 + }, + { + "epoch": 2.094087573183805, + "grad_norm": 2.0948076248168945, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.769525408744812, + "num_tokens": 653378783.0, + "step": 25395 + }, + { + "epoch": 2.0941700338088562, + "grad_norm": 2.0106546878814697, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7606824636459351, + "num_tokens": 653412053.0, + "step": 25396 + }, + { + "epoch": 2.0942524944339076, + "grad_norm": 1.994960904121399, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7674484252929688, + "num_tokens": 653443716.0, + "step": 25397 + }, + { + "epoch": 2.0943349550589594, + "grad_norm": 2.2822673320770264, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7585955858230591, + "num_tokens": 653470445.0, + "step": 25398 + }, + { + "epoch": 2.0944174156840107, + "grad_norm": 2.051359176635742, + "learning_rate": 1e-06, + "loss": 0.5722, + "mean_token_accuracy": 0.8004147410392761, + "num_tokens": 653501502.0, + "step": 25399 + }, + { + "epoch": 2.0944998763090625, + "grad_norm": 2.044140577316284, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7393940687179565, + "num_tokens": 653531622.0, + "step": 25400 + }, + { + "epoch": 2.094582336934114, + "grad_norm": 2.0917763710021973, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.759191632270813, + "num_tokens": 653562054.0, + "step": 25401 + }, + { + "epoch": 2.0946647975591657, + "grad_norm": 2.029477596282959, + "learning_rate": 1e-06, + "loss": 0.5663, + "mean_token_accuracy": 0.8090574145317078, + "num_tokens": 653591885.0, + "step": 25402 + }, + { + "epoch": 2.094747258184217, + "grad_norm": 2.196641445159912, + "learning_rate": 1e-06, + "loss": 0.5885, + "mean_token_accuracy": 0.8050209879875183, + "num_tokens": 653617843.0, + "step": 25403 + }, + { + "epoch": 2.0948297188092684, + "grad_norm": 2.168978691101074, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7435106635093689, + "num_tokens": 653648929.0, + "step": 25404 + }, + { + "epoch": 2.09491217943432, + "grad_norm": 2.036311626434326, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7764679193496704, + "num_tokens": 653680759.0, + "step": 25405 + }, + { + "epoch": 2.0949946400593715, + "grad_norm": 2.3634777069091797, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7690384387969971, + "num_tokens": 653704168.0, + "step": 25406 + }, + { + "epoch": 2.0950771006844233, + "grad_norm": 1.9634743928909302, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7508354783058167, + "num_tokens": 653735980.0, + "step": 25407 + }, + { + "epoch": 2.0951595613094747, + "grad_norm": 1.9782867431640625, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7635447382926941, + "num_tokens": 653768430.0, + "step": 25408 + }, + { + "epoch": 2.0952420219345265, + "grad_norm": 2.2834761142730713, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7391242384910583, + "num_tokens": 653795713.0, + "step": 25409 + }, + { + "epoch": 2.095324482559578, + "grad_norm": 2.0325658321380615, + "learning_rate": 1e-06, + "loss": 0.5854, + "mean_token_accuracy": 0.8062578439712524, + "num_tokens": 653825078.0, + "step": 25410 + }, + { + "epoch": 2.095406943184629, + "grad_norm": 2.111046552658081, + "learning_rate": 1e-06, + "loss": 0.591, + "mean_token_accuracy": 0.8110228180885315, + "num_tokens": 653851552.0, + "step": 25411 + }, + { + "epoch": 2.095489403809681, + "grad_norm": 1.9909162521362305, + "learning_rate": 1e-06, + "loss": 0.4783, + "mean_token_accuracy": 0.8302662968635559, + "num_tokens": 653880989.0, + "step": 25412 + }, + { + "epoch": 2.0955718644347323, + "grad_norm": 2.174535036087036, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.779246985912323, + "num_tokens": 653909943.0, + "step": 25413 + }, + { + "epoch": 2.095654325059784, + "grad_norm": 2.1477301120758057, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.7888440489768982, + "num_tokens": 653939598.0, + "step": 25414 + }, + { + "epoch": 2.0957367856848355, + "grad_norm": 2.285630226135254, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7760072946548462, + "num_tokens": 653965111.0, + "step": 25415 + }, + { + "epoch": 2.095819246309887, + "grad_norm": 2.3133091926574707, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7405515909194946, + "num_tokens": 653989998.0, + "step": 25416 + }, + { + "epoch": 2.0959017069349386, + "grad_norm": 1.9374748468399048, + "learning_rate": 1e-06, + "loss": 0.5923, + "mean_token_accuracy": 0.7992050647735596, + "num_tokens": 654023667.0, + "step": 25417 + }, + { + "epoch": 2.09598416755999, + "grad_norm": 1.9697320461273193, + "learning_rate": 1e-06, + "loss": 0.5554, + "mean_token_accuracy": 0.8079527616500854, + "num_tokens": 654056390.0, + "step": 25418 + }, + { + "epoch": 2.0960666281850417, + "grad_norm": 2.09810471534729, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7368829250335693, + "num_tokens": 654087521.0, + "step": 25419 + }, + { + "epoch": 2.096149088810093, + "grad_norm": 2.2195982933044434, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7428086996078491, + "num_tokens": 654117550.0, + "step": 25420 + }, + { + "epoch": 2.096231549435145, + "grad_norm": 2.177323579788208, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7682703733444214, + "num_tokens": 654147002.0, + "step": 25421 + }, + { + "epoch": 2.0963140100601962, + "grad_norm": 2.1297454833984375, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7760822772979736, + "num_tokens": 654176738.0, + "step": 25422 + }, + { + "epoch": 2.0963964706852476, + "grad_norm": 2.121410846710205, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7532044649124146, + "num_tokens": 654209053.0, + "step": 25423 + }, + { + "epoch": 2.0964789313102994, + "grad_norm": 1.856427788734436, + "learning_rate": 1e-06, + "loss": 0.5982, + "mean_token_accuracy": 0.7971068620681763, + "num_tokens": 654244137.0, + "step": 25424 + }, + { + "epoch": 2.0965613919353507, + "grad_norm": 2.505622625350952, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.7845020294189453, + "num_tokens": 654265755.0, + "step": 25425 + }, + { + "epoch": 2.0966438525604025, + "grad_norm": 2.0409510135650635, + "learning_rate": 1e-06, + "loss": 0.7675, + "mean_token_accuracy": 0.7534377574920654, + "num_tokens": 654296705.0, + "step": 25426 + }, + { + "epoch": 2.096726313185454, + "grad_norm": 2.0407421588897705, + "learning_rate": 1e-06, + "loss": 0.5526, + "mean_token_accuracy": 0.806923508644104, + "num_tokens": 654326866.0, + "step": 25427 + }, + { + "epoch": 2.0968087738105057, + "grad_norm": 2.314512252807617, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7322219610214233, + "num_tokens": 654353659.0, + "step": 25428 + }, + { + "epoch": 2.096891234435557, + "grad_norm": 1.9819269180297852, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.7998455166816711, + "num_tokens": 654384057.0, + "step": 25429 + }, + { + "epoch": 2.0969736950606084, + "grad_norm": 2.138587713241577, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7369487881660461, + "num_tokens": 654415474.0, + "step": 25430 + }, + { + "epoch": 2.09705615568566, + "grad_norm": 1.9168964624404907, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7858489155769348, + "num_tokens": 654448620.0, + "step": 25431 + }, + { + "epoch": 2.0971386163107115, + "grad_norm": 2.5683794021606445, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7832760810852051, + "num_tokens": 654468716.0, + "step": 25432 + }, + { + "epoch": 2.0972210769357633, + "grad_norm": 1.976388931274414, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7517735958099365, + "num_tokens": 654501685.0, + "step": 25433 + }, + { + "epoch": 2.0973035375608147, + "grad_norm": 2.1410598754882812, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.762033224105835, + "num_tokens": 654532555.0, + "step": 25434 + }, + { + "epoch": 2.0973859981858665, + "grad_norm": 2.101026773452759, + "learning_rate": 1e-06, + "loss": 0.5999, + "mean_token_accuracy": 0.7982656955718994, + "num_tokens": 654560513.0, + "step": 25435 + }, + { + "epoch": 2.097468458810918, + "grad_norm": 1.9005624055862427, + "learning_rate": 1e-06, + "loss": 0.5825, + "mean_token_accuracy": 0.808501124382019, + "num_tokens": 654592652.0, + "step": 25436 + }, + { + "epoch": 2.097550919435969, + "grad_norm": 1.8756742477416992, + "learning_rate": 1e-06, + "loss": 0.5679, + "mean_token_accuracy": 0.8188174962997437, + "num_tokens": 654628598.0, + "step": 25437 + }, + { + "epoch": 2.097633380061021, + "grad_norm": 2.1968820095062256, + "learning_rate": 1e-06, + "loss": 0.6123, + "mean_token_accuracy": 0.7902321815490723, + "num_tokens": 654655434.0, + "step": 25438 + }, + { + "epoch": 2.0977158406860723, + "grad_norm": 1.9032621383666992, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.7926666140556335, + "num_tokens": 654688831.0, + "step": 25439 + }, + { + "epoch": 2.097798301311124, + "grad_norm": 2.1034774780273438, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.7974734902381897, + "num_tokens": 654717082.0, + "step": 25440 + }, + { + "epoch": 2.0978807619361755, + "grad_norm": 2.1030433177948, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.8008304834365845, + "num_tokens": 654747861.0, + "step": 25441 + }, + { + "epoch": 2.097963222561227, + "grad_norm": 2.3987576961517334, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.750538170337677, + "num_tokens": 654771402.0, + "step": 25442 + }, + { + "epoch": 2.0980456831862786, + "grad_norm": 2.1719491481781006, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.751491367816925, + "num_tokens": 654803014.0, + "step": 25443 + }, + { + "epoch": 2.09812814381133, + "grad_norm": 2.035945177078247, + "learning_rate": 1e-06, + "loss": 0.5802, + "mean_token_accuracy": 0.8068554401397705, + "num_tokens": 654833123.0, + "step": 25444 + }, + { + "epoch": 2.0982106044363817, + "grad_norm": 1.8798291683197021, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7587000131607056, + "num_tokens": 654866322.0, + "step": 25445 + }, + { + "epoch": 2.098293065061433, + "grad_norm": 2.15724515914917, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7840965986251831, + "num_tokens": 654895665.0, + "step": 25446 + }, + { + "epoch": 2.098375525686485, + "grad_norm": 2.219680070877075, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7460641860961914, + "num_tokens": 654925118.0, + "step": 25447 + }, + { + "epoch": 2.0984579863115362, + "grad_norm": 2.08254075050354, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7422725558280945, + "num_tokens": 654956707.0, + "step": 25448 + }, + { + "epoch": 2.0985404469365876, + "grad_norm": 1.9560654163360596, + "learning_rate": 1e-06, + "loss": 0.5795, + "mean_token_accuracy": 0.8057535886764526, + "num_tokens": 654991014.0, + "step": 25449 + }, + { + "epoch": 2.0986229075616394, + "grad_norm": 2.2914693355560303, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7686251401901245, + "num_tokens": 655015464.0, + "step": 25450 + }, + { + "epoch": 2.0987053681866907, + "grad_norm": 2.0690066814422607, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.761695146560669, + "num_tokens": 655047799.0, + "step": 25451 + }, + { + "epoch": 2.0987878288117425, + "grad_norm": 2.0683987140655518, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7812826633453369, + "num_tokens": 655078021.0, + "step": 25452 + }, + { + "epoch": 2.098870289436794, + "grad_norm": 2.0989584922790527, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7600271701812744, + "num_tokens": 655108822.0, + "step": 25453 + }, + { + "epoch": 2.0989527500618457, + "grad_norm": 2.107550621032715, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.7829921245574951, + "num_tokens": 655137708.0, + "step": 25454 + }, + { + "epoch": 2.099035210686897, + "grad_norm": 2.036306619644165, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7731369733810425, + "num_tokens": 655167898.0, + "step": 25455 + }, + { + "epoch": 2.0991176713119484, + "grad_norm": 2.0337610244750977, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7897149920463562, + "num_tokens": 655198248.0, + "step": 25456 + }, + { + "epoch": 2.099200131937, + "grad_norm": 2.1363415718078613, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.761275053024292, + "num_tokens": 655230020.0, + "step": 25457 + }, + { + "epoch": 2.0992825925620515, + "grad_norm": 2.163987159729004, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7817521095275879, + "num_tokens": 655256515.0, + "step": 25458 + }, + { + "epoch": 2.0993650531871033, + "grad_norm": 1.9462597370147705, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7690892815589905, + "num_tokens": 655287348.0, + "step": 25459 + }, + { + "epoch": 2.0994475138121547, + "grad_norm": 1.8813964128494263, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7602276802062988, + "num_tokens": 655321336.0, + "step": 25460 + }, + { + "epoch": 2.099529974437206, + "grad_norm": 2.049344062805176, + "learning_rate": 1e-06, + "loss": 0.5495, + "mean_token_accuracy": 0.8113565444946289, + "num_tokens": 655348705.0, + "step": 25461 + }, + { + "epoch": 2.099612435062258, + "grad_norm": 2.0129282474517822, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.7871091961860657, + "num_tokens": 655380436.0, + "step": 25462 + }, + { + "epoch": 2.099694895687309, + "grad_norm": 2.2471683025360107, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.7853636741638184, + "num_tokens": 655408154.0, + "step": 25463 + }, + { + "epoch": 2.099777356312361, + "grad_norm": 1.9904487133026123, + "learning_rate": 1e-06, + "loss": 0.4991, + "mean_token_accuracy": 0.8293434381484985, + "num_tokens": 655435920.0, + "step": 25464 + }, + { + "epoch": 2.0998598169374123, + "grad_norm": 2.1425936222076416, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7919555902481079, + "num_tokens": 655465650.0, + "step": 25465 + }, + { + "epoch": 2.099942277562464, + "grad_norm": 1.9864269495010376, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7575783729553223, + "num_tokens": 655495548.0, + "step": 25466 + }, + { + "epoch": 2.1000247381875154, + "grad_norm": 2.113084316253662, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.7790350317955017, + "num_tokens": 655524943.0, + "step": 25467 + }, + { + "epoch": 2.100107198812567, + "grad_norm": 2.205307722091675, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7587147951126099, + "num_tokens": 655553631.0, + "step": 25468 + }, + { + "epoch": 2.1001896594376186, + "grad_norm": 2.141422986984253, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7898109555244446, + "num_tokens": 655579260.0, + "step": 25469 + }, + { + "epoch": 2.10027212006267, + "grad_norm": 2.312701940536499, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.737736165523529, + "num_tokens": 655605629.0, + "step": 25470 + }, + { + "epoch": 2.1003545806877217, + "grad_norm": 2.1845364570617676, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7674651145935059, + "num_tokens": 655632508.0, + "step": 25471 + }, + { + "epoch": 2.100437041312773, + "grad_norm": 2.1830639839172363, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.786164402961731, + "num_tokens": 655661203.0, + "step": 25472 + }, + { + "epoch": 2.100519501937825, + "grad_norm": 2.0685975551605225, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7661210298538208, + "num_tokens": 655691490.0, + "step": 25473 + }, + { + "epoch": 2.1006019625628762, + "grad_norm": 1.925313949584961, + "learning_rate": 1e-06, + "loss": 0.5455, + "mean_token_accuracy": 0.80344158411026, + "num_tokens": 655726816.0, + "step": 25474 + }, + { + "epoch": 2.1006844231879276, + "grad_norm": 2.2562546730041504, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.7835079431533813, + "num_tokens": 655754393.0, + "step": 25475 + }, + { + "epoch": 2.1007668838129794, + "grad_norm": 2.228879928588867, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.8076685667037964, + "num_tokens": 655778999.0, + "step": 25476 + }, + { + "epoch": 2.1008493444380307, + "grad_norm": 2.299943447113037, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.767884373664856, + "num_tokens": 655803957.0, + "step": 25477 + }, + { + "epoch": 2.1009318050630825, + "grad_norm": 2.1120591163635254, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7592816948890686, + "num_tokens": 655832507.0, + "step": 25478 + }, + { + "epoch": 2.101014265688134, + "grad_norm": 2.365501642227173, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7759445905685425, + "num_tokens": 655857409.0, + "step": 25479 + }, + { + "epoch": 2.1010967263131857, + "grad_norm": 2.088534116744995, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7496031522750854, + "num_tokens": 655887061.0, + "step": 25480 + }, + { + "epoch": 2.101179186938237, + "grad_norm": 2.1759140491485596, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7774744033813477, + "num_tokens": 655915854.0, + "step": 25481 + }, + { + "epoch": 2.1012616475632884, + "grad_norm": 2.375187873840332, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7877883315086365, + "num_tokens": 655940021.0, + "step": 25482 + }, + { + "epoch": 2.10134410818834, + "grad_norm": 1.9104626178741455, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.8003138899803162, + "num_tokens": 655973390.0, + "step": 25483 + }, + { + "epoch": 2.1014265688133915, + "grad_norm": 2.091959238052368, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.788415789604187, + "num_tokens": 656000753.0, + "step": 25484 + }, + { + "epoch": 2.1015090294384433, + "grad_norm": 2.0039501190185547, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7902515530586243, + "num_tokens": 656032034.0, + "step": 25485 + }, + { + "epoch": 2.1015914900634947, + "grad_norm": 2.3119993209838867, + "learning_rate": 1e-06, + "loss": 0.5807, + "mean_token_accuracy": 0.8064417839050293, + "num_tokens": 656057872.0, + "step": 25486 + }, + { + "epoch": 2.101673950688546, + "grad_norm": 2.4342265129089355, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7581607699394226, + "num_tokens": 656083215.0, + "step": 25487 + }, + { + "epoch": 2.101756411313598, + "grad_norm": 2.3356869220733643, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.7840370535850525, + "num_tokens": 656108865.0, + "step": 25488 + }, + { + "epoch": 2.101838871938649, + "grad_norm": 2.040078639984131, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7776655554771423, + "num_tokens": 656142421.0, + "step": 25489 + }, + { + "epoch": 2.101921332563701, + "grad_norm": 2.151183843612671, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.7924946546554565, + "num_tokens": 656171324.0, + "step": 25490 + }, + { + "epoch": 2.1020037931887523, + "grad_norm": 2.4532177448272705, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7470598220825195, + "num_tokens": 656194598.0, + "step": 25491 + }, + { + "epoch": 2.102086253813804, + "grad_norm": 2.215212345123291, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.7919381260871887, + "num_tokens": 656223697.0, + "step": 25492 + }, + { + "epoch": 2.1021687144388554, + "grad_norm": 2.6214804649353027, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7593377232551575, + "num_tokens": 656245036.0, + "step": 25493 + }, + { + "epoch": 2.102251175063907, + "grad_norm": 2.3032901287078857, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7743713855743408, + "num_tokens": 656271470.0, + "step": 25494 + }, + { + "epoch": 2.1023336356889586, + "grad_norm": 1.964040756225586, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7759499549865723, + "num_tokens": 656305968.0, + "step": 25495 + }, + { + "epoch": 2.10241609631401, + "grad_norm": 2.383204460144043, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7639991044998169, + "num_tokens": 656331389.0, + "step": 25496 + }, + { + "epoch": 2.1024985569390617, + "grad_norm": 1.7090237140655518, + "learning_rate": 1e-06, + "loss": 0.5307, + "mean_token_accuracy": 0.8186306357383728, + "num_tokens": 656368614.0, + "step": 25497 + }, + { + "epoch": 2.102581017564113, + "grad_norm": 2.1636922359466553, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7688394784927368, + "num_tokens": 656399342.0, + "step": 25498 + }, + { + "epoch": 2.102663478189165, + "grad_norm": 2.2823855876922607, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7399951219558716, + "num_tokens": 656427418.0, + "step": 25499 + }, + { + "epoch": 2.1027459388142162, + "grad_norm": 2.0638577938079834, + "learning_rate": 1e-06, + "loss": 0.5964, + "mean_token_accuracy": 0.8001210689544678, + "num_tokens": 656456425.0, + "step": 25500 + }, + { + "epoch": 2.1028283994392676, + "grad_norm": 2.0417327880859375, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.745438814163208, + "num_tokens": 656490814.0, + "step": 25501 + }, + { + "epoch": 2.1029108600643194, + "grad_norm": 2.202920436859131, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7746045589447021, + "num_tokens": 656517833.0, + "step": 25502 + }, + { + "epoch": 2.1029933206893707, + "grad_norm": 2.0588300228118896, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7417144179344177, + "num_tokens": 656549694.0, + "step": 25503 + }, + { + "epoch": 2.1030757813144225, + "grad_norm": 2.219668388366699, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7651294469833374, + "num_tokens": 656578985.0, + "step": 25504 + }, + { + "epoch": 2.103158241939474, + "grad_norm": 1.8026570081710815, + "learning_rate": 1e-06, + "loss": 0.6004, + "mean_token_accuracy": 0.8079929351806641, + "num_tokens": 656615355.0, + "step": 25505 + }, + { + "epoch": 2.103240702564525, + "grad_norm": 2.059941291809082, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.743591845035553, + "num_tokens": 656644943.0, + "step": 25506 + }, + { + "epoch": 2.103323163189577, + "grad_norm": 2.2954347133636475, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.7847187519073486, + "num_tokens": 656669825.0, + "step": 25507 + }, + { + "epoch": 2.1034056238146284, + "grad_norm": 2.042379856109619, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7857344150543213, + "num_tokens": 656701120.0, + "step": 25508 + }, + { + "epoch": 2.10348808443968, + "grad_norm": 2.014075756072998, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7631174325942993, + "num_tokens": 656732917.0, + "step": 25509 + }, + { + "epoch": 2.1035705450647315, + "grad_norm": 2.0824806690216064, + "learning_rate": 1e-06, + "loss": 0.504, + "mean_token_accuracy": 0.8230041861534119, + "num_tokens": 656760398.0, + "step": 25510 + }, + { + "epoch": 2.1036530056897833, + "grad_norm": 2.2488975524902344, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7438690662384033, + "num_tokens": 656786191.0, + "step": 25511 + }, + { + "epoch": 2.1037354663148347, + "grad_norm": 2.302555561065674, + "learning_rate": 1e-06, + "loss": 0.6144, + "mean_token_accuracy": 0.7918022871017456, + "num_tokens": 656813224.0, + "step": 25512 + }, + { + "epoch": 2.103817926939886, + "grad_norm": 2.305272102355957, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7181704640388489, + "num_tokens": 656841088.0, + "step": 25513 + }, + { + "epoch": 2.103900387564938, + "grad_norm": 2.0285513401031494, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7456774711608887, + "num_tokens": 656871200.0, + "step": 25514 + }, + { + "epoch": 2.103982848189989, + "grad_norm": 2.4619953632354736, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7460304498672485, + "num_tokens": 656894169.0, + "step": 25515 + }, + { + "epoch": 2.104065308815041, + "grad_norm": 2.282688617706299, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7531269788742065, + "num_tokens": 656920414.0, + "step": 25516 + }, + { + "epoch": 2.1041477694400923, + "grad_norm": 1.9795414209365845, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7437527179718018, + "num_tokens": 656952216.0, + "step": 25517 + }, + { + "epoch": 2.104230230065144, + "grad_norm": 1.9248499870300293, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7883739471435547, + "num_tokens": 656984163.0, + "step": 25518 + }, + { + "epoch": 2.1043126906901954, + "grad_norm": 2.1192800998687744, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7529633641242981, + "num_tokens": 657014395.0, + "step": 25519 + }, + { + "epoch": 2.104395151315247, + "grad_norm": 2.005427837371826, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7642954587936401, + "num_tokens": 657046317.0, + "step": 25520 + }, + { + "epoch": 2.1044776119402986, + "grad_norm": 2.476289987564087, + "learning_rate": 1e-06, + "loss": 0.5939, + "mean_token_accuracy": 0.8021509647369385, + "num_tokens": 657070205.0, + "step": 25521 + }, + { + "epoch": 2.10456007256535, + "grad_norm": 1.766681432723999, + "learning_rate": 1e-06, + "loss": 0.4822, + "mean_token_accuracy": 0.8366742134094238, + "num_tokens": 657104840.0, + "step": 25522 + }, + { + "epoch": 2.1046425331904017, + "grad_norm": 2.4229736328125, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7678236961364746, + "num_tokens": 657127136.0, + "step": 25523 + }, + { + "epoch": 2.104724993815453, + "grad_norm": 1.9505783319473267, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7145088911056519, + "num_tokens": 657166480.0, + "step": 25524 + }, + { + "epoch": 2.104807454440505, + "grad_norm": 2.073394298553467, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.7955450415611267, + "num_tokens": 657199061.0, + "step": 25525 + }, + { + "epoch": 2.1048899150655562, + "grad_norm": 1.7915982007980347, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.7972995042800903, + "num_tokens": 657234836.0, + "step": 25526 + }, + { + "epoch": 2.1049723756906076, + "grad_norm": 1.975616216659546, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7597343921661377, + "num_tokens": 657269224.0, + "step": 25527 + }, + { + "epoch": 2.1050548363156594, + "grad_norm": 2.2568469047546387, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7426477670669556, + "num_tokens": 657298619.0, + "step": 25528 + }, + { + "epoch": 2.1051372969407107, + "grad_norm": 2.0445501804351807, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7569922208786011, + "num_tokens": 657327639.0, + "step": 25529 + }, + { + "epoch": 2.1052197575657625, + "grad_norm": 2.3487308025360107, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7526997327804565, + "num_tokens": 657352292.0, + "step": 25530 + }, + { + "epoch": 2.105302218190814, + "grad_norm": 2.0379385948181152, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7659226655960083, + "num_tokens": 657386316.0, + "step": 25531 + }, + { + "epoch": 2.105384678815865, + "grad_norm": 2.31577205657959, + "learning_rate": 1e-06, + "loss": 0.7303, + "mean_token_accuracy": 0.7666201591491699, + "num_tokens": 657412727.0, + "step": 25532 + }, + { + "epoch": 2.105467139440917, + "grad_norm": 2.0674831867218018, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7657859325408936, + "num_tokens": 657443810.0, + "step": 25533 + }, + { + "epoch": 2.1055496000659684, + "grad_norm": 2.157137632369995, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.7719874382019043, + "num_tokens": 657471036.0, + "step": 25534 + }, + { + "epoch": 2.10563206069102, + "grad_norm": 2.196516752243042, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7584524154663086, + "num_tokens": 657498723.0, + "step": 25535 + }, + { + "epoch": 2.1057145213160715, + "grad_norm": 2.0926403999328613, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7795737981796265, + "num_tokens": 657526973.0, + "step": 25536 + }, + { + "epoch": 2.1057969819411233, + "grad_norm": 2.488297462463379, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7540374994277954, + "num_tokens": 657549753.0, + "step": 25537 + }, + { + "epoch": 2.1058794425661747, + "grad_norm": 2.241469144821167, + "learning_rate": 1e-06, + "loss": 0.5898, + "mean_token_accuracy": 0.7995039224624634, + "num_tokens": 657576204.0, + "step": 25538 + }, + { + "epoch": 2.105961903191226, + "grad_norm": 1.899064302444458, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7455732226371765, + "num_tokens": 657614502.0, + "step": 25539 + }, + { + "epoch": 2.106044363816278, + "grad_norm": 2.0799107551574707, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.7884657382965088, + "num_tokens": 657645784.0, + "step": 25540 + }, + { + "epoch": 2.106126824441329, + "grad_norm": 2.277700662612915, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.7756911516189575, + "num_tokens": 657670673.0, + "step": 25541 + }, + { + "epoch": 2.106209285066381, + "grad_norm": 2.1588985919952393, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.7873270511627197, + "num_tokens": 657698987.0, + "step": 25542 + }, + { + "epoch": 2.1062917456914323, + "grad_norm": 2.388141632080078, + "learning_rate": 1e-06, + "loss": 0.5794, + "mean_token_accuracy": 0.7888774871826172, + "num_tokens": 657722022.0, + "step": 25543 + }, + { + "epoch": 2.106374206316484, + "grad_norm": 2.153325080871582, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7657134532928467, + "num_tokens": 657750773.0, + "step": 25544 + }, + { + "epoch": 2.1064566669415354, + "grad_norm": 2.469259738922119, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7304807901382446, + "num_tokens": 657774001.0, + "step": 25545 + }, + { + "epoch": 2.106539127566587, + "grad_norm": 2.0118210315704346, + "learning_rate": 1e-06, + "loss": 0.5188, + "mean_token_accuracy": 0.8272476196289062, + "num_tokens": 657802820.0, + "step": 25546 + }, + { + "epoch": 2.1066215881916386, + "grad_norm": 2.284898519515991, + "learning_rate": 1e-06, + "loss": 0.6019, + "mean_token_accuracy": 0.7907229065895081, + "num_tokens": 657830229.0, + "step": 25547 + }, + { + "epoch": 2.10670404881669, + "grad_norm": 2.0726840496063232, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.759194016456604, + "num_tokens": 657858750.0, + "step": 25548 + }, + { + "epoch": 2.1067865094417417, + "grad_norm": 2.0776216983795166, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.7979981899261475, + "num_tokens": 657889703.0, + "step": 25549 + }, + { + "epoch": 2.106868970066793, + "grad_norm": 2.2187790870666504, + "learning_rate": 1e-06, + "loss": 0.548, + "mean_token_accuracy": 0.8116346597671509, + "num_tokens": 657919204.0, + "step": 25550 + }, + { + "epoch": 2.1069514306918444, + "grad_norm": 1.8099637031555176, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7749505043029785, + "num_tokens": 657958556.0, + "step": 25551 + }, + { + "epoch": 2.1070338913168962, + "grad_norm": 2.2163164615631104, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7445242404937744, + "num_tokens": 657987634.0, + "step": 25552 + }, + { + "epoch": 2.1071163519419476, + "grad_norm": 2.230478525161743, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7741256952285767, + "num_tokens": 658016131.0, + "step": 25553 + }, + { + "epoch": 2.1071988125669994, + "grad_norm": 2.2720274925231934, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7756996154785156, + "num_tokens": 658044496.0, + "step": 25554 + }, + { + "epoch": 2.1072812731920507, + "grad_norm": 2.0583934783935547, + "learning_rate": 1e-06, + "loss": 0.6, + "mean_token_accuracy": 0.8021825551986694, + "num_tokens": 658077238.0, + "step": 25555 + }, + { + "epoch": 2.1073637338171025, + "grad_norm": 2.1729533672332764, + "learning_rate": 1e-06, + "loss": 0.5043, + "mean_token_accuracy": 0.8261001110076904, + "num_tokens": 658105115.0, + "step": 25556 + }, + { + "epoch": 2.107446194442154, + "grad_norm": 2.1350033283233643, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7530983686447144, + "num_tokens": 658133880.0, + "step": 25557 + }, + { + "epoch": 2.107528655067205, + "grad_norm": 1.9529821872711182, + "learning_rate": 1e-06, + "loss": 0.5973, + "mean_token_accuracy": 0.8005053997039795, + "num_tokens": 658165019.0, + "step": 25558 + }, + { + "epoch": 2.107611115692257, + "grad_norm": 1.9885449409484863, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.782045841217041, + "num_tokens": 658199438.0, + "step": 25559 + }, + { + "epoch": 2.1076935763173084, + "grad_norm": 2.0829885005950928, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.781550943851471, + "num_tokens": 658228243.0, + "step": 25560 + }, + { + "epoch": 2.10777603694236, + "grad_norm": 1.8805450201034546, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7527704834938049, + "num_tokens": 658266954.0, + "step": 25561 + }, + { + "epoch": 2.1078584975674115, + "grad_norm": 2.0178940296173096, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7546792030334473, + "num_tokens": 658301663.0, + "step": 25562 + }, + { + "epoch": 2.1079409581924633, + "grad_norm": 2.0297129154205322, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.7891221046447754, + "num_tokens": 658332331.0, + "step": 25563 + }, + { + "epoch": 2.1080234188175146, + "grad_norm": 2.193657636642456, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7802755832672119, + "num_tokens": 658360987.0, + "step": 25564 + }, + { + "epoch": 2.108105879442566, + "grad_norm": 2.139788866043091, + "learning_rate": 1e-06, + "loss": 0.5866, + "mean_token_accuracy": 0.8033636808395386, + "num_tokens": 658386942.0, + "step": 25565 + }, + { + "epoch": 2.108188340067618, + "grad_norm": 2.3282277584075928, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7963590621948242, + "num_tokens": 658413954.0, + "step": 25566 + }, + { + "epoch": 2.108270800692669, + "grad_norm": 1.7854571342468262, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7787691354751587, + "num_tokens": 658449134.0, + "step": 25567 + }, + { + "epoch": 2.108353261317721, + "grad_norm": 2.2604379653930664, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7709616422653198, + "num_tokens": 658476247.0, + "step": 25568 + }, + { + "epoch": 2.1084357219427723, + "grad_norm": 1.80142343044281, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.7921615242958069, + "num_tokens": 658511887.0, + "step": 25569 + }, + { + "epoch": 2.108518182567824, + "grad_norm": 2.1324517726898193, + "learning_rate": 1e-06, + "loss": 0.6106, + "mean_token_accuracy": 0.8006953001022339, + "num_tokens": 658539758.0, + "step": 25570 + }, + { + "epoch": 2.1086006431928754, + "grad_norm": 1.9449189901351929, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.7973936796188354, + "num_tokens": 658573194.0, + "step": 25571 + }, + { + "epoch": 2.108683103817927, + "grad_norm": 1.900477409362793, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.764983594417572, + "num_tokens": 658608463.0, + "step": 25572 + }, + { + "epoch": 2.1087655644429786, + "grad_norm": 2.3486452102661133, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7487595081329346, + "num_tokens": 658632555.0, + "step": 25573 + }, + { + "epoch": 2.10884802506803, + "grad_norm": 2.3585963249206543, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7518005967140198, + "num_tokens": 658658880.0, + "step": 25574 + }, + { + "epoch": 2.1089304856930817, + "grad_norm": 1.904401421546936, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7638531923294067, + "num_tokens": 658694227.0, + "step": 25575 + }, + { + "epoch": 2.109012946318133, + "grad_norm": 2.42720365524292, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.7952236533164978, + "num_tokens": 658717055.0, + "step": 25576 + }, + { + "epoch": 2.1090954069431844, + "grad_norm": 2.5421416759490967, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7478671073913574, + "num_tokens": 658742667.0, + "step": 25577 + }, + { + "epoch": 2.109177867568236, + "grad_norm": 2.136828899383545, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7548015713691711, + "num_tokens": 658772541.0, + "step": 25578 + }, + { + "epoch": 2.1092603281932876, + "grad_norm": 1.8977073431015015, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7814867496490479, + "num_tokens": 658804964.0, + "step": 25579 + }, + { + "epoch": 2.1093427888183394, + "grad_norm": 1.9078792333602905, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.7852247357368469, + "num_tokens": 658841034.0, + "step": 25580 + }, + { + "epoch": 2.1094252494433907, + "grad_norm": 2.097081422805786, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7592553496360779, + "num_tokens": 658873013.0, + "step": 25581 + }, + { + "epoch": 2.1095077100684425, + "grad_norm": 2.245156764984131, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7337019443511963, + "num_tokens": 658901675.0, + "step": 25582 + }, + { + "epoch": 2.109590170693494, + "grad_norm": 2.0639405250549316, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.7815500497817993, + "num_tokens": 658933572.0, + "step": 25583 + }, + { + "epoch": 2.109672631318545, + "grad_norm": 2.178607225418091, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7638274431228638, + "num_tokens": 658961986.0, + "step": 25584 + }, + { + "epoch": 2.109755091943597, + "grad_norm": 2.2025997638702393, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7647531628608704, + "num_tokens": 658989370.0, + "step": 25585 + }, + { + "epoch": 2.1098375525686484, + "grad_norm": 2.277895927429199, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7456890344619751, + "num_tokens": 659013756.0, + "step": 25586 + }, + { + "epoch": 2.1099200131937, + "grad_norm": 1.8744914531707764, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.7809041738510132, + "num_tokens": 659049882.0, + "step": 25587 + }, + { + "epoch": 2.1100024738187515, + "grad_norm": 2.093839406967163, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.792198896408081, + "num_tokens": 659080541.0, + "step": 25588 + }, + { + "epoch": 2.1100849344438033, + "grad_norm": 2.1729578971862793, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7756795883178711, + "num_tokens": 659109051.0, + "step": 25589 + }, + { + "epoch": 2.1101673950688546, + "grad_norm": 2.2116518020629883, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.792016863822937, + "num_tokens": 659134990.0, + "step": 25590 + }, + { + "epoch": 2.110249855693906, + "grad_norm": 2.415163516998291, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.7883814573287964, + "num_tokens": 659158662.0, + "step": 25591 + }, + { + "epoch": 2.110332316318958, + "grad_norm": 2.244382381439209, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7428724765777588, + "num_tokens": 659186974.0, + "step": 25592 + }, + { + "epoch": 2.110414776944009, + "grad_norm": 2.078909397125244, + "learning_rate": 1e-06, + "loss": 0.5684, + "mean_token_accuracy": 0.8139129281044006, + "num_tokens": 659215159.0, + "step": 25593 + }, + { + "epoch": 2.110497237569061, + "grad_norm": 2.1401002407073975, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7774873971939087, + "num_tokens": 659244486.0, + "step": 25594 + }, + { + "epoch": 2.1105796981941123, + "grad_norm": 2.2384583950042725, + "learning_rate": 1e-06, + "loss": 0.558, + "mean_token_accuracy": 0.8077658414840698, + "num_tokens": 659269939.0, + "step": 25595 + }, + { + "epoch": 2.1106621588191636, + "grad_norm": 1.9089415073394775, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.7890625596046448, + "num_tokens": 659303705.0, + "step": 25596 + }, + { + "epoch": 2.1107446194442154, + "grad_norm": 2.1228718757629395, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7672197818756104, + "num_tokens": 659334033.0, + "step": 25597 + }, + { + "epoch": 2.110827080069267, + "grad_norm": 2.0203633308410645, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.7858508229255676, + "num_tokens": 659365709.0, + "step": 25598 + }, + { + "epoch": 2.1109095406943186, + "grad_norm": 1.8139876127243042, + "learning_rate": 1e-06, + "loss": 0.5549, + "mean_token_accuracy": 0.8190221786499023, + "num_tokens": 659400045.0, + "step": 25599 + }, + { + "epoch": 2.11099200131937, + "grad_norm": 1.961220622062683, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7667268514633179, + "num_tokens": 659438848.0, + "step": 25600 + }, + { + "epoch": 2.1110744619444217, + "grad_norm": 2.0628139972686768, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7366020679473877, + "num_tokens": 659470077.0, + "step": 25601 + }, + { + "epoch": 2.111156922569473, + "grad_norm": 1.878836989402771, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.784859299659729, + "num_tokens": 659506453.0, + "step": 25602 + }, + { + "epoch": 2.1112393831945244, + "grad_norm": 1.9458907842636108, + "learning_rate": 1e-06, + "loss": 0.5575, + "mean_token_accuracy": 0.8139247894287109, + "num_tokens": 659537340.0, + "step": 25603 + }, + { + "epoch": 2.111321843819576, + "grad_norm": 2.196967840194702, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7548125982284546, + "num_tokens": 659564226.0, + "step": 25604 + }, + { + "epoch": 2.1114043044446276, + "grad_norm": 2.3056445121765137, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7665339708328247, + "num_tokens": 659590492.0, + "step": 25605 + }, + { + "epoch": 2.1114867650696794, + "grad_norm": 1.997426152229309, + "learning_rate": 1e-06, + "loss": 0.5472, + "mean_token_accuracy": 0.8221569061279297, + "num_tokens": 659619651.0, + "step": 25606 + }, + { + "epoch": 2.1115692256947307, + "grad_norm": 1.9756572246551514, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.7943502068519592, + "num_tokens": 659652322.0, + "step": 25607 + }, + { + "epoch": 2.1116516863197825, + "grad_norm": 2.3560221195220947, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7389551401138306, + "num_tokens": 659677131.0, + "step": 25608 + }, + { + "epoch": 2.111734146944834, + "grad_norm": 2.3173513412475586, + "learning_rate": 1e-06, + "loss": 0.5873, + "mean_token_accuracy": 0.8011878728866577, + "num_tokens": 659702540.0, + "step": 25609 + }, + { + "epoch": 2.111816607569885, + "grad_norm": 2.3239870071411133, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.772159993648529, + "num_tokens": 659726714.0, + "step": 25610 + }, + { + "epoch": 2.111899068194937, + "grad_norm": 2.348635673522949, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7352796196937561, + "num_tokens": 659753149.0, + "step": 25611 + }, + { + "epoch": 2.1119815288199884, + "grad_norm": 1.8415166139602661, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7819339036941528, + "num_tokens": 659789532.0, + "step": 25612 + }, + { + "epoch": 2.11206398944504, + "grad_norm": 2.3968605995178223, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7514515519142151, + "num_tokens": 659813551.0, + "step": 25613 + }, + { + "epoch": 2.1121464500700915, + "grad_norm": 1.8838157653808594, + "learning_rate": 1e-06, + "loss": 0.5535, + "mean_token_accuracy": 0.8192821741104126, + "num_tokens": 659848011.0, + "step": 25614 + }, + { + "epoch": 2.1122289106951433, + "grad_norm": 2.06712007522583, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7626222372055054, + "num_tokens": 659878725.0, + "step": 25615 + }, + { + "epoch": 2.1123113713201946, + "grad_norm": 2.328136682510376, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.7897156476974487, + "num_tokens": 659902698.0, + "step": 25616 + }, + { + "epoch": 2.112393831945246, + "grad_norm": 2.113874673843384, + "learning_rate": 1e-06, + "loss": 0.5081, + "mean_token_accuracy": 0.8250571489334106, + "num_tokens": 659933540.0, + "step": 25617 + }, + { + "epoch": 2.112476292570298, + "grad_norm": 1.945208191871643, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7702994346618652, + "num_tokens": 659966787.0, + "step": 25618 + }, + { + "epoch": 2.112558753195349, + "grad_norm": 2.085026741027832, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7844488620758057, + "num_tokens": 659999910.0, + "step": 25619 + }, + { + "epoch": 2.112641213820401, + "grad_norm": 2.171858549118042, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7309367656707764, + "num_tokens": 660028064.0, + "step": 25620 + }, + { + "epoch": 2.1127236744454523, + "grad_norm": 2.0741384029388428, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7602535486221313, + "num_tokens": 660059425.0, + "step": 25621 + }, + { + "epoch": 2.1128061350705036, + "grad_norm": 2.4568662643432617, + "learning_rate": 1e-06, + "loss": 0.7359, + "mean_token_accuracy": 0.7578229308128357, + "num_tokens": 660089126.0, + "step": 25622 + }, + { + "epoch": 2.1128885956955554, + "grad_norm": 2.593155860900879, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.768538236618042, + "num_tokens": 660110972.0, + "step": 25623 + }, + { + "epoch": 2.1129710563206068, + "grad_norm": 2.0944020748138428, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.768744707107544, + "num_tokens": 660143168.0, + "step": 25624 + }, + { + "epoch": 2.1130535169456586, + "grad_norm": 1.810099482536316, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.7804887294769287, + "num_tokens": 660179896.0, + "step": 25625 + }, + { + "epoch": 2.11313597757071, + "grad_norm": 1.9224801063537598, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7653882503509521, + "num_tokens": 660218328.0, + "step": 25626 + }, + { + "epoch": 2.1132184381957617, + "grad_norm": 1.9200935363769531, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.767156720161438, + "num_tokens": 660252836.0, + "step": 25627 + }, + { + "epoch": 2.113300898820813, + "grad_norm": 2.2396419048309326, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.77752685546875, + "num_tokens": 660281043.0, + "step": 25628 + }, + { + "epoch": 2.1133833594458644, + "grad_norm": 2.132235050201416, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.737373948097229, + "num_tokens": 660311649.0, + "step": 25629 + }, + { + "epoch": 2.113465820070916, + "grad_norm": 1.9643064737319946, + "learning_rate": 1e-06, + "loss": 0.6043, + "mean_token_accuracy": 0.7945901155471802, + "num_tokens": 660344722.0, + "step": 25630 + }, + { + "epoch": 2.1135482806959676, + "grad_norm": 1.9095747470855713, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7750733494758606, + "num_tokens": 660378154.0, + "step": 25631 + }, + { + "epoch": 2.1136307413210194, + "grad_norm": 2.452268600463867, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7672740817070007, + "num_tokens": 660402262.0, + "step": 25632 + }, + { + "epoch": 2.1137132019460707, + "grad_norm": 2.410604238510132, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7892892360687256, + "num_tokens": 660425440.0, + "step": 25633 + }, + { + "epoch": 2.1137956625711225, + "grad_norm": 2.1509077548980713, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7158714532852173, + "num_tokens": 660455506.0, + "step": 25634 + }, + { + "epoch": 2.113878123196174, + "grad_norm": 2.1541528701782227, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7812092900276184, + "num_tokens": 660482702.0, + "step": 25635 + }, + { + "epoch": 2.113960583821225, + "grad_norm": 2.337022542953491, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.8024531602859497, + "num_tokens": 660504971.0, + "step": 25636 + }, + { + "epoch": 2.114043044446277, + "grad_norm": 2.102811098098755, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7681043744087219, + "num_tokens": 660534985.0, + "step": 25637 + }, + { + "epoch": 2.1141255050713283, + "grad_norm": 1.9984205961227417, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7212322950363159, + "num_tokens": 660570678.0, + "step": 25638 + }, + { + "epoch": 2.11420796569638, + "grad_norm": 2.1117146015167236, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.764714241027832, + "num_tokens": 660601254.0, + "step": 25639 + }, + { + "epoch": 2.1142904263214315, + "grad_norm": 2.108053207397461, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7722334861755371, + "num_tokens": 660630880.0, + "step": 25640 + }, + { + "epoch": 2.114372886946483, + "grad_norm": 1.8590072393417358, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.7844278216362, + "num_tokens": 660665964.0, + "step": 25641 + }, + { + "epoch": 2.1144553475715346, + "grad_norm": 2.207221508026123, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7787624597549438, + "num_tokens": 660694083.0, + "step": 25642 + }, + { + "epoch": 2.114537808196586, + "grad_norm": 2.1728689670562744, + "learning_rate": 1e-06, + "loss": 0.5553, + "mean_token_accuracy": 0.8162329196929932, + "num_tokens": 660719485.0, + "step": 25643 + }, + { + "epoch": 2.114620268821638, + "grad_norm": 2.079369068145752, + "learning_rate": 1e-06, + "loss": 0.6079, + "mean_token_accuracy": 0.7941964864730835, + "num_tokens": 660751850.0, + "step": 25644 + }, + { + "epoch": 2.114702729446689, + "grad_norm": 2.2197670936584473, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7876273989677429, + "num_tokens": 660775530.0, + "step": 25645 + }, + { + "epoch": 2.114785190071741, + "grad_norm": 1.985298752784729, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7741905450820923, + "num_tokens": 660810517.0, + "step": 25646 + }, + { + "epoch": 2.1148676506967923, + "grad_norm": 1.87691330909729, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7759240865707397, + "num_tokens": 660847967.0, + "step": 25647 + }, + { + "epoch": 2.1149501113218436, + "grad_norm": 2.107703924179077, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.781233012676239, + "num_tokens": 660877972.0, + "step": 25648 + }, + { + "epoch": 2.1150325719468954, + "grad_norm": 2.261042833328247, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7790579795837402, + "num_tokens": 660903019.0, + "step": 25649 + }, + { + "epoch": 2.1151150325719468, + "grad_norm": 2.398977756500244, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7411483526229858, + "num_tokens": 660929724.0, + "step": 25650 + }, + { + "epoch": 2.1151974931969986, + "grad_norm": 2.2987020015716553, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7263303399085999, + "num_tokens": 660959829.0, + "step": 25651 + }, + { + "epoch": 2.11527995382205, + "grad_norm": 2.0880305767059326, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7719053626060486, + "num_tokens": 660995139.0, + "step": 25652 + }, + { + "epoch": 2.1153624144471017, + "grad_norm": 2.181771993637085, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7752728462219238, + "num_tokens": 661023284.0, + "step": 25653 + }, + { + "epoch": 2.115444875072153, + "grad_norm": 2.1961276531219482, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.749829888343811, + "num_tokens": 661049818.0, + "step": 25654 + }, + { + "epoch": 2.1155273356972044, + "grad_norm": 2.0897598266601562, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7499591112136841, + "num_tokens": 661078752.0, + "step": 25655 + }, + { + "epoch": 2.115609796322256, + "grad_norm": 1.8993067741394043, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7788597941398621, + "num_tokens": 661111988.0, + "step": 25656 + }, + { + "epoch": 2.1156922569473076, + "grad_norm": 2.1504337787628174, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7602971792221069, + "num_tokens": 661140834.0, + "step": 25657 + }, + { + "epoch": 2.1157747175723594, + "grad_norm": 1.9909334182739258, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7485808730125427, + "num_tokens": 661174408.0, + "step": 25658 + }, + { + "epoch": 2.1158571781974107, + "grad_norm": 2.4513516426086426, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7220818400382996, + "num_tokens": 661199220.0, + "step": 25659 + }, + { + "epoch": 2.1159396388224625, + "grad_norm": 2.2553813457489014, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7629560828208923, + "num_tokens": 661226682.0, + "step": 25660 + }, + { + "epoch": 2.116022099447514, + "grad_norm": 2.009536027908325, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.785561203956604, + "num_tokens": 661260857.0, + "step": 25661 + }, + { + "epoch": 2.116104560072565, + "grad_norm": 1.9921661615371704, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.7936193943023682, + "num_tokens": 661290176.0, + "step": 25662 + }, + { + "epoch": 2.116187020697617, + "grad_norm": 1.858332872390747, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7530795335769653, + "num_tokens": 661326066.0, + "step": 25663 + }, + { + "epoch": 2.1162694813226683, + "grad_norm": 1.9413130283355713, + "learning_rate": 1e-06, + "loss": 0.6056, + "mean_token_accuracy": 0.7983382940292358, + "num_tokens": 661361935.0, + "step": 25664 + }, + { + "epoch": 2.11635194194772, + "grad_norm": 2.25177001953125, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7485459446907043, + "num_tokens": 661391114.0, + "step": 25665 + }, + { + "epoch": 2.1164344025727715, + "grad_norm": 2.065723180770874, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7770518064498901, + "num_tokens": 661422907.0, + "step": 25666 + }, + { + "epoch": 2.116516863197823, + "grad_norm": 1.7914767265319824, + "learning_rate": 1e-06, + "loss": 0.5856, + "mean_token_accuracy": 0.8040499687194824, + "num_tokens": 661459825.0, + "step": 25667 + }, + { + "epoch": 2.1165993238228746, + "grad_norm": 1.8913823366165161, + "learning_rate": 1e-06, + "loss": 0.5892, + "mean_token_accuracy": 0.8129299879074097, + "num_tokens": 661492332.0, + "step": 25668 + }, + { + "epoch": 2.116681784447926, + "grad_norm": 2.1181640625, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7486968040466309, + "num_tokens": 661522220.0, + "step": 25669 + }, + { + "epoch": 2.116764245072978, + "grad_norm": 2.093165874481201, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7712869644165039, + "num_tokens": 661551084.0, + "step": 25670 + }, + { + "epoch": 2.116846705698029, + "grad_norm": 1.9828007221221924, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.7861220836639404, + "num_tokens": 661584341.0, + "step": 25671 + }, + { + "epoch": 2.116929166323081, + "grad_norm": 1.9097785949707031, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7793101072311401, + "num_tokens": 661618380.0, + "step": 25672 + }, + { + "epoch": 2.1170116269481323, + "grad_norm": 1.844655990600586, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.8015972971916199, + "num_tokens": 661650793.0, + "step": 25673 + }, + { + "epoch": 2.1170940875731836, + "grad_norm": 1.8579127788543701, + "learning_rate": 1e-06, + "loss": 0.5853, + "mean_token_accuracy": 0.8113232254981995, + "num_tokens": 661684518.0, + "step": 25674 + }, + { + "epoch": 2.1171765481982354, + "grad_norm": 2.384431838989258, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7573477029800415, + "num_tokens": 661711873.0, + "step": 25675 + }, + { + "epoch": 2.1172590088232868, + "grad_norm": 2.2173898220062256, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.7995082139968872, + "num_tokens": 661735933.0, + "step": 25676 + }, + { + "epoch": 2.1173414694483386, + "grad_norm": 2.2886159420013428, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.7626951932907104, + "num_tokens": 661764858.0, + "step": 25677 + }, + { + "epoch": 2.11742393007339, + "grad_norm": 2.0426390171051025, + "learning_rate": 1e-06, + "loss": 0.5588, + "mean_token_accuracy": 0.8220024108886719, + "num_tokens": 661792137.0, + "step": 25678 + }, + { + "epoch": 2.1175063906984413, + "grad_norm": 1.9516500234603882, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.775202751159668, + "num_tokens": 661826347.0, + "step": 25679 + }, + { + "epoch": 2.117588851323493, + "grad_norm": 1.8695006370544434, + "learning_rate": 1e-06, + "loss": 0.5638, + "mean_token_accuracy": 0.8114529848098755, + "num_tokens": 661861380.0, + "step": 25680 + }, + { + "epoch": 2.1176713119485444, + "grad_norm": 1.9881889820098877, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7787415981292725, + "num_tokens": 661893091.0, + "step": 25681 + }, + { + "epoch": 2.117753772573596, + "grad_norm": 2.138550043106079, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7336630821228027, + "num_tokens": 661925480.0, + "step": 25682 + }, + { + "epoch": 2.1178362331986476, + "grad_norm": 1.9275047779083252, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.7572771906852722, + "num_tokens": 661961134.0, + "step": 25683 + }, + { + "epoch": 2.1179186938236993, + "grad_norm": 2.218857526779175, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7346361875534058, + "num_tokens": 661991949.0, + "step": 25684 + }, + { + "epoch": 2.1180011544487507, + "grad_norm": 1.9996451139450073, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7490978240966797, + "num_tokens": 662025494.0, + "step": 25685 + }, + { + "epoch": 2.118083615073802, + "grad_norm": 1.9147740602493286, + "learning_rate": 1e-06, + "loss": 0.4575, + "mean_token_accuracy": 0.8378523588180542, + "num_tokens": 662056511.0, + "step": 25686 + }, + { + "epoch": 2.118166075698854, + "grad_norm": 1.8383079767227173, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.7991912961006165, + "num_tokens": 662093288.0, + "step": 25687 + }, + { + "epoch": 2.118248536323905, + "grad_norm": 2.1000239849090576, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7722927331924438, + "num_tokens": 662120926.0, + "step": 25688 + }, + { + "epoch": 2.118330996948957, + "grad_norm": 2.110570192337036, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.773628294467926, + "num_tokens": 662150458.0, + "step": 25689 + }, + { + "epoch": 2.1184134575740083, + "grad_norm": 1.9480706453323364, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7686893939971924, + "num_tokens": 662182893.0, + "step": 25690 + }, + { + "epoch": 2.11849591819906, + "grad_norm": 2.0392143726348877, + "learning_rate": 1e-06, + "loss": 0.602, + "mean_token_accuracy": 0.7967073917388916, + "num_tokens": 662213526.0, + "step": 25691 + }, + { + "epoch": 2.1185783788241115, + "grad_norm": 2.0154566764831543, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.8023850917816162, + "num_tokens": 662243672.0, + "step": 25692 + }, + { + "epoch": 2.118660839449163, + "grad_norm": 2.0362894535064697, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7729341983795166, + "num_tokens": 662272613.0, + "step": 25693 + }, + { + "epoch": 2.1187433000742146, + "grad_norm": 1.8152132034301758, + "learning_rate": 1e-06, + "loss": 0.6041, + "mean_token_accuracy": 0.7989152669906616, + "num_tokens": 662309184.0, + "step": 25694 + }, + { + "epoch": 2.118825760699266, + "grad_norm": 2.2922964096069336, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7722547054290771, + "num_tokens": 662336507.0, + "step": 25695 + }, + { + "epoch": 2.1189082213243178, + "grad_norm": 2.144867420196533, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7682782411575317, + "num_tokens": 662365815.0, + "step": 25696 + }, + { + "epoch": 2.118990681949369, + "grad_norm": 2.2514076232910156, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.770968496799469, + "num_tokens": 662394207.0, + "step": 25697 + }, + { + "epoch": 2.119073142574421, + "grad_norm": 2.057734727859497, + "learning_rate": 1e-06, + "loss": 0.4731, + "mean_token_accuracy": 0.8287880420684814, + "num_tokens": 662421184.0, + "step": 25698 + }, + { + "epoch": 2.1191556031994723, + "grad_norm": 2.080458402633667, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.7805238366127014, + "num_tokens": 662449508.0, + "step": 25699 + }, + { + "epoch": 2.1192380638245236, + "grad_norm": 2.4361984729766846, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7795605659484863, + "num_tokens": 662471578.0, + "step": 25700 + }, + { + "epoch": 2.1193205244495754, + "grad_norm": 2.244577646255493, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7555330395698547, + "num_tokens": 662500885.0, + "step": 25701 + }, + { + "epoch": 2.1194029850746268, + "grad_norm": 2.3928167819976807, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7562111616134644, + "num_tokens": 662523778.0, + "step": 25702 + }, + { + "epoch": 2.1194854456996786, + "grad_norm": 2.3482141494750977, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7496669292449951, + "num_tokens": 662550809.0, + "step": 25703 + }, + { + "epoch": 2.11956790632473, + "grad_norm": 1.992592453956604, + "learning_rate": 1e-06, + "loss": 0.5753, + "mean_token_accuracy": 0.813150942325592, + "num_tokens": 662581590.0, + "step": 25704 + }, + { + "epoch": 2.1196503669497817, + "grad_norm": 2.230968713760376, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7591703534126282, + "num_tokens": 662610579.0, + "step": 25705 + }, + { + "epoch": 2.119732827574833, + "grad_norm": 2.389153480529785, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7681358456611633, + "num_tokens": 662633381.0, + "step": 25706 + }, + { + "epoch": 2.1198152881998844, + "grad_norm": 2.438666820526123, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.7877767086029053, + "num_tokens": 662658116.0, + "step": 25707 + }, + { + "epoch": 2.119897748824936, + "grad_norm": 2.084925651550293, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.76575767993927, + "num_tokens": 662688039.0, + "step": 25708 + }, + { + "epoch": 2.1199802094499876, + "grad_norm": 2.300879955291748, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.7847691774368286, + "num_tokens": 662713168.0, + "step": 25709 + }, + { + "epoch": 2.1200626700750393, + "grad_norm": 2.407890558242798, + "learning_rate": 1e-06, + "loss": 0.5844, + "mean_token_accuracy": 0.7884073257446289, + "num_tokens": 662735464.0, + "step": 25710 + }, + { + "epoch": 2.1201451307000907, + "grad_norm": 2.14678955078125, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7753517627716064, + "num_tokens": 662765934.0, + "step": 25711 + }, + { + "epoch": 2.120227591325142, + "grad_norm": 1.8895931243896484, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.774283766746521, + "num_tokens": 662802461.0, + "step": 25712 + }, + { + "epoch": 2.120310051950194, + "grad_norm": 2.2070930004119873, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.754550576210022, + "num_tokens": 662831718.0, + "step": 25713 + }, + { + "epoch": 2.120392512575245, + "grad_norm": 1.84734308719635, + "learning_rate": 1e-06, + "loss": 0.5863, + "mean_token_accuracy": 0.7967002987861633, + "num_tokens": 662862680.0, + "step": 25714 + }, + { + "epoch": 2.120474973200297, + "grad_norm": 1.9942257404327393, + "learning_rate": 1e-06, + "loss": 0.615, + "mean_token_accuracy": 0.794413685798645, + "num_tokens": 662895174.0, + "step": 25715 + }, + { + "epoch": 2.1205574338253483, + "grad_norm": 2.120746374130249, + "learning_rate": 1e-06, + "loss": 0.5828, + "mean_token_accuracy": 0.8083544969558716, + "num_tokens": 662923020.0, + "step": 25716 + }, + { + "epoch": 2.1206398944504, + "grad_norm": 1.9544899463653564, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.760053277015686, + "num_tokens": 662957010.0, + "step": 25717 + }, + { + "epoch": 2.1207223550754515, + "grad_norm": 2.153809070587158, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7318544387817383, + "num_tokens": 662987326.0, + "step": 25718 + }, + { + "epoch": 2.120804815700503, + "grad_norm": 2.0622973442077637, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7779619693756104, + "num_tokens": 663018103.0, + "step": 25719 + }, + { + "epoch": 2.1208872763255546, + "grad_norm": 2.206110715866089, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7420620918273926, + "num_tokens": 663048661.0, + "step": 25720 + }, + { + "epoch": 2.120969736950606, + "grad_norm": 1.9993996620178223, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7706835269927979, + "num_tokens": 663081803.0, + "step": 25721 + }, + { + "epoch": 2.1210521975756578, + "grad_norm": 2.0393106937408447, + "learning_rate": 1e-06, + "loss": 0.5984, + "mean_token_accuracy": 0.7955780029296875, + "num_tokens": 663111198.0, + "step": 25722 + }, + { + "epoch": 2.121134658200709, + "grad_norm": 2.0284154415130615, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7722576856613159, + "num_tokens": 663142797.0, + "step": 25723 + }, + { + "epoch": 2.1212171188257605, + "grad_norm": 2.3945071697235107, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7409617304801941, + "num_tokens": 663167503.0, + "step": 25724 + }, + { + "epoch": 2.1212995794508123, + "grad_norm": 1.9707423448562622, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.776178777217865, + "num_tokens": 663200356.0, + "step": 25725 + }, + { + "epoch": 2.1213820400758636, + "grad_norm": 2.0495834350585938, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.7835277318954468, + "num_tokens": 663229882.0, + "step": 25726 + }, + { + "epoch": 2.1214645007009154, + "grad_norm": 2.259708881378174, + "learning_rate": 1e-06, + "loss": 0.5966, + "mean_token_accuracy": 0.8039522171020508, + "num_tokens": 663253956.0, + "step": 25727 + }, + { + "epoch": 2.1215469613259668, + "grad_norm": 2.3613171577453613, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7515954971313477, + "num_tokens": 663280202.0, + "step": 25728 + }, + { + "epoch": 2.1216294219510186, + "grad_norm": 2.548773765563965, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.7426363229751587, + "num_tokens": 663302578.0, + "step": 25729 + }, + { + "epoch": 2.12171188257607, + "grad_norm": 2.0715036392211914, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.7807420492172241, + "num_tokens": 663332497.0, + "step": 25730 + }, + { + "epoch": 2.1217943432011213, + "grad_norm": 2.217810869216919, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.758004367351532, + "num_tokens": 663361790.0, + "step": 25731 + }, + { + "epoch": 2.121876803826173, + "grad_norm": 1.803619146347046, + "learning_rate": 1e-06, + "loss": 0.6221, + "mean_token_accuracy": 0.8015273809432983, + "num_tokens": 663398418.0, + "step": 25732 + }, + { + "epoch": 2.1219592644512244, + "grad_norm": 1.8618956804275513, + "learning_rate": 1e-06, + "loss": 0.5244, + "mean_token_accuracy": 0.8128958940505981, + "num_tokens": 663431827.0, + "step": 25733 + }, + { + "epoch": 2.122041725076276, + "grad_norm": 2.2655889987945557, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7548394203186035, + "num_tokens": 663460426.0, + "step": 25734 + }, + { + "epoch": 2.1221241857013275, + "grad_norm": 1.9946177005767822, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.7959719896316528, + "num_tokens": 663492501.0, + "step": 25735 + }, + { + "epoch": 2.1222066463263793, + "grad_norm": 2.006131649017334, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7850515842437744, + "num_tokens": 663525590.0, + "step": 25736 + }, + { + "epoch": 2.1222891069514307, + "grad_norm": 2.1758780479431152, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.767360508441925, + "num_tokens": 663554756.0, + "step": 25737 + }, + { + "epoch": 2.122371567576482, + "grad_norm": 2.406388282775879, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.7570846080780029, + "num_tokens": 663580233.0, + "step": 25738 + }, + { + "epoch": 2.122454028201534, + "grad_norm": 2.4841806888580322, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7230473756790161, + "num_tokens": 663606502.0, + "step": 25739 + }, + { + "epoch": 2.122536488826585, + "grad_norm": 2.2773900032043457, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.77073073387146, + "num_tokens": 663632285.0, + "step": 25740 + }, + { + "epoch": 2.122618949451637, + "grad_norm": 2.3204877376556396, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.7830244898796082, + "num_tokens": 663657822.0, + "step": 25741 + }, + { + "epoch": 2.1227014100766883, + "grad_norm": 2.1027672290802, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7580897808074951, + "num_tokens": 663691315.0, + "step": 25742 + }, + { + "epoch": 2.12278387070174, + "grad_norm": 1.936906337738037, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.7848219871520996, + "num_tokens": 663727484.0, + "step": 25743 + }, + { + "epoch": 2.1228663313267915, + "grad_norm": 2.390028953552246, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7700339555740356, + "num_tokens": 663752106.0, + "step": 25744 + }, + { + "epoch": 2.122948791951843, + "grad_norm": 1.9405906200408936, + "learning_rate": 1e-06, + "loss": 0.5841, + "mean_token_accuracy": 0.8062204718589783, + "num_tokens": 663783886.0, + "step": 25745 + }, + { + "epoch": 2.1230312525768946, + "grad_norm": 2.0875532627105713, + "learning_rate": 1e-06, + "loss": 0.7825, + "mean_token_accuracy": 0.7493006587028503, + "num_tokens": 663812433.0, + "step": 25746 + }, + { + "epoch": 2.123113713201946, + "grad_norm": 1.884152889251709, + "learning_rate": 1e-06, + "loss": 0.5093, + "mean_token_accuracy": 0.8223289847373962, + "num_tokens": 663843357.0, + "step": 25747 + }, + { + "epoch": 2.1231961738269978, + "grad_norm": 2.066671133041382, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7808975577354431, + "num_tokens": 663873910.0, + "step": 25748 + }, + { + "epoch": 2.123278634452049, + "grad_norm": 2.333911418914795, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7677009105682373, + "num_tokens": 663899426.0, + "step": 25749 + }, + { + "epoch": 2.123361095077101, + "grad_norm": 2.110748291015625, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.773423433303833, + "num_tokens": 663927837.0, + "step": 25750 + }, + { + "epoch": 2.1234435557021523, + "grad_norm": 2.0867552757263184, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7864956259727478, + "num_tokens": 663958795.0, + "step": 25751 + }, + { + "epoch": 2.1235260163272036, + "grad_norm": 1.836190104484558, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7478845119476318, + "num_tokens": 663996078.0, + "step": 25752 + }, + { + "epoch": 2.1236084769522554, + "grad_norm": 2.1760470867156982, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.762654721736908, + "num_tokens": 664025089.0, + "step": 25753 + }, + { + "epoch": 2.1236909375773068, + "grad_norm": 2.029837131500244, + "learning_rate": 1e-06, + "loss": 0.563, + "mean_token_accuracy": 0.8103411793708801, + "num_tokens": 664055309.0, + "step": 25754 + }, + { + "epoch": 2.1237733982023586, + "grad_norm": 1.9594706296920776, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7378416061401367, + "num_tokens": 664090706.0, + "step": 25755 + }, + { + "epoch": 2.12385585882741, + "grad_norm": 1.8407385349273682, + "learning_rate": 1e-06, + "loss": 0.536, + "mean_token_accuracy": 0.8214959502220154, + "num_tokens": 664126333.0, + "step": 25756 + }, + { + "epoch": 2.1239383194524613, + "grad_norm": 2.1590042114257812, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7731290459632874, + "num_tokens": 664156181.0, + "step": 25757 + }, + { + "epoch": 2.124020780077513, + "grad_norm": 2.130037307739258, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.7777788639068604, + "num_tokens": 664186003.0, + "step": 25758 + }, + { + "epoch": 2.1241032407025644, + "grad_norm": 2.204094648361206, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7779229879379272, + "num_tokens": 664214132.0, + "step": 25759 + }, + { + "epoch": 2.124185701327616, + "grad_norm": 1.923304796218872, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7704181671142578, + "num_tokens": 664247949.0, + "step": 25760 + }, + { + "epoch": 2.1242681619526675, + "grad_norm": 2.124131679534912, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7688074111938477, + "num_tokens": 664277766.0, + "step": 25761 + }, + { + "epoch": 2.1243506225777193, + "grad_norm": 1.8318074941635132, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7651418447494507, + "num_tokens": 664316065.0, + "step": 25762 + }, + { + "epoch": 2.1244330832027707, + "grad_norm": 1.859946608543396, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7760501503944397, + "num_tokens": 664351663.0, + "step": 25763 + }, + { + "epoch": 2.124515543827822, + "grad_norm": 2.055492639541626, + "learning_rate": 1e-06, + "loss": 0.7396, + "mean_token_accuracy": 0.7629835605621338, + "num_tokens": 664382140.0, + "step": 25764 + }, + { + "epoch": 2.124598004452874, + "grad_norm": 2.0136020183563232, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.7904188632965088, + "num_tokens": 664415848.0, + "step": 25765 + }, + { + "epoch": 2.124680465077925, + "grad_norm": 1.976353406906128, + "learning_rate": 1e-06, + "loss": 0.57, + "mean_token_accuracy": 0.8164471387863159, + "num_tokens": 664446095.0, + "step": 25766 + }, + { + "epoch": 2.124762925702977, + "grad_norm": 1.9395065307617188, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7760340571403503, + "num_tokens": 664481709.0, + "step": 25767 + }, + { + "epoch": 2.1248453863280283, + "grad_norm": 2.128556966781616, + "learning_rate": 1e-06, + "loss": 0.5499, + "mean_token_accuracy": 0.8063448667526245, + "num_tokens": 664509205.0, + "step": 25768 + }, + { + "epoch": 2.1249278469530797, + "grad_norm": 2.0823891162872314, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7752881050109863, + "num_tokens": 664537385.0, + "step": 25769 + }, + { + "epoch": 2.1250103075781315, + "grad_norm": 2.0519330501556396, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7593530416488647, + "num_tokens": 664569038.0, + "step": 25770 + }, + { + "epoch": 2.125092768203183, + "grad_norm": 2.206369638442993, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.7797831296920776, + "num_tokens": 664597274.0, + "step": 25771 + }, + { + "epoch": 2.1251752288282346, + "grad_norm": 2.1292006969451904, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.763576328754425, + "num_tokens": 664626458.0, + "step": 25772 + }, + { + "epoch": 2.125257689453286, + "grad_norm": 1.9592360258102417, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7565231323242188, + "num_tokens": 664660224.0, + "step": 25773 + }, + { + "epoch": 2.1253401500783378, + "grad_norm": 2.268857479095459, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.7527598142623901, + "num_tokens": 664685195.0, + "step": 25774 + }, + { + "epoch": 2.125422610703389, + "grad_norm": 2.050225257873535, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7857943177223206, + "num_tokens": 664716276.0, + "step": 25775 + }, + { + "epoch": 2.1255050713284405, + "grad_norm": 1.8631021976470947, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.788680374622345, + "num_tokens": 664751068.0, + "step": 25776 + }, + { + "epoch": 2.1255875319534923, + "grad_norm": 2.1586875915527344, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.746442437171936, + "num_tokens": 664779684.0, + "step": 25777 + }, + { + "epoch": 2.1256699925785436, + "grad_norm": 1.9995330572128296, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7685545086860657, + "num_tokens": 664811157.0, + "step": 25778 + }, + { + "epoch": 2.1257524532035954, + "grad_norm": 1.9585609436035156, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7481854557991028, + "num_tokens": 664844172.0, + "step": 25779 + }, + { + "epoch": 2.1258349138286468, + "grad_norm": 2.170232057571411, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7738174200057983, + "num_tokens": 664874707.0, + "step": 25780 + }, + { + "epoch": 2.1259173744536985, + "grad_norm": 2.286083221435547, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7806200981140137, + "num_tokens": 664905436.0, + "step": 25781 + }, + { + "epoch": 2.12599983507875, + "grad_norm": 1.9025609493255615, + "learning_rate": 1e-06, + "loss": 0.5452, + "mean_token_accuracy": 0.82574462890625, + "num_tokens": 664942926.0, + "step": 25782 + }, + { + "epoch": 2.1260822957038013, + "grad_norm": 1.8912606239318848, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7717030644416809, + "num_tokens": 664977460.0, + "step": 25783 + }, + { + "epoch": 2.126164756328853, + "grad_norm": 1.8213400840759277, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7670361399650574, + "num_tokens": 665015751.0, + "step": 25784 + }, + { + "epoch": 2.1262472169539044, + "grad_norm": 2.107882499694824, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7647713422775269, + "num_tokens": 665044432.0, + "step": 25785 + }, + { + "epoch": 2.126329677578956, + "grad_norm": 1.9971022605895996, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.795604944229126, + "num_tokens": 665074267.0, + "step": 25786 + }, + { + "epoch": 2.1264121382040075, + "grad_norm": 2.1831493377685547, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7593340277671814, + "num_tokens": 665103050.0, + "step": 25787 + }, + { + "epoch": 2.1264945988290593, + "grad_norm": 1.9852304458618164, + "learning_rate": 1e-06, + "loss": 0.5589, + "mean_token_accuracy": 0.8122691512107849, + "num_tokens": 665135582.0, + "step": 25788 + }, + { + "epoch": 2.1265770594541107, + "grad_norm": 1.9641599655151367, + "learning_rate": 1e-06, + "loss": 0.572, + "mean_token_accuracy": 0.804772138595581, + "num_tokens": 665168996.0, + "step": 25789 + }, + { + "epoch": 2.126659520079162, + "grad_norm": 2.0367016792297363, + "learning_rate": 1e-06, + "loss": 0.578, + "mean_token_accuracy": 0.8025756478309631, + "num_tokens": 665199227.0, + "step": 25790 + }, + { + "epoch": 2.126741980704214, + "grad_norm": 2.0535929203033447, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.7539381384849548, + "num_tokens": 665230035.0, + "step": 25791 + }, + { + "epoch": 2.126824441329265, + "grad_norm": 2.199320077896118, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7612396478652954, + "num_tokens": 665259465.0, + "step": 25792 + }, + { + "epoch": 2.126906901954317, + "grad_norm": 2.3376340866088867, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7590981721878052, + "num_tokens": 665282764.0, + "step": 25793 + }, + { + "epoch": 2.1269893625793683, + "grad_norm": 2.1652469635009766, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.756259024143219, + "num_tokens": 665311499.0, + "step": 25794 + }, + { + "epoch": 2.12707182320442, + "grad_norm": 1.8488844633102417, + "learning_rate": 1e-06, + "loss": 0.6022, + "mean_token_accuracy": 0.7975994348526001, + "num_tokens": 665347175.0, + "step": 25795 + }, + { + "epoch": 2.1271542838294715, + "grad_norm": 2.1271560192108154, + "learning_rate": 1e-06, + "loss": 0.6257, + "mean_token_accuracy": 0.7930333018302917, + "num_tokens": 665375070.0, + "step": 25796 + }, + { + "epoch": 2.127236744454523, + "grad_norm": 1.84469473361969, + "learning_rate": 1e-06, + "loss": 0.5487, + "mean_token_accuracy": 0.816708505153656, + "num_tokens": 665409188.0, + "step": 25797 + }, + { + "epoch": 2.1273192050795746, + "grad_norm": 2.173302173614502, + "learning_rate": 1e-06, + "loss": 0.6036, + "mean_token_accuracy": 0.7983460426330566, + "num_tokens": 665435739.0, + "step": 25798 + }, + { + "epoch": 2.127401665704626, + "grad_norm": 1.8115578889846802, + "learning_rate": 1e-06, + "loss": 0.5844, + "mean_token_accuracy": 0.8010859489440918, + "num_tokens": 665475305.0, + "step": 25799 + }, + { + "epoch": 2.1274841263296778, + "grad_norm": 1.997575283050537, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7717342972755432, + "num_tokens": 665509293.0, + "step": 25800 + }, + { + "epoch": 2.127566586954729, + "grad_norm": 2.2433876991271973, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7554893493652344, + "num_tokens": 665536385.0, + "step": 25801 + }, + { + "epoch": 2.1276490475797805, + "grad_norm": 2.10097599029541, + "learning_rate": 1e-06, + "loss": 0.5818, + "mean_token_accuracy": 0.8115034103393555, + "num_tokens": 665566262.0, + "step": 25802 + }, + { + "epoch": 2.1277315082048323, + "grad_norm": 1.8701602220535278, + "learning_rate": 1e-06, + "loss": 0.6053, + "mean_token_accuracy": 0.8024129867553711, + "num_tokens": 665599679.0, + "step": 25803 + }, + { + "epoch": 2.1278139688298836, + "grad_norm": 2.544111490249634, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7597118616104126, + "num_tokens": 665625332.0, + "step": 25804 + }, + { + "epoch": 2.1278964294549354, + "grad_norm": 2.44500470161438, + "learning_rate": 1e-06, + "loss": 0.7808, + "mean_token_accuracy": 0.7441173791885376, + "num_tokens": 665649559.0, + "step": 25805 + }, + { + "epoch": 2.1279788900799868, + "grad_norm": 2.2407217025756836, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.785159707069397, + "num_tokens": 665682481.0, + "step": 25806 + }, + { + "epoch": 2.1280613507050385, + "grad_norm": 2.115201473236084, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7349573373794556, + "num_tokens": 665715726.0, + "step": 25807 + }, + { + "epoch": 2.12814381133009, + "grad_norm": 2.0562257766723633, + "learning_rate": 1e-06, + "loss": 0.8072, + "mean_token_accuracy": 0.7504799962043762, + "num_tokens": 665747248.0, + "step": 25808 + }, + { + "epoch": 2.1282262719551412, + "grad_norm": 1.9690648317337036, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7766779065132141, + "num_tokens": 665778554.0, + "step": 25809 + }, + { + "epoch": 2.128308732580193, + "grad_norm": 1.9417264461517334, + "learning_rate": 1e-06, + "loss": 0.7617, + "mean_token_accuracy": 0.7494363784790039, + "num_tokens": 665814376.0, + "step": 25810 + }, + { + "epoch": 2.1283911932052444, + "grad_norm": 2.1322147846221924, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7675427198410034, + "num_tokens": 665842602.0, + "step": 25811 + }, + { + "epoch": 2.128473653830296, + "grad_norm": 2.2031517028808594, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7382791638374329, + "num_tokens": 665871209.0, + "step": 25812 + }, + { + "epoch": 2.1285561144553475, + "grad_norm": 2.115283966064453, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7440212368965149, + "num_tokens": 665901278.0, + "step": 25813 + }, + { + "epoch": 2.128638575080399, + "grad_norm": 1.9019687175750732, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.7920181751251221, + "num_tokens": 665937481.0, + "step": 25814 + }, + { + "epoch": 2.1287210357054507, + "grad_norm": 1.9919688701629639, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7375345230102539, + "num_tokens": 665971865.0, + "step": 25815 + }, + { + "epoch": 2.128803496330502, + "grad_norm": 1.897733449935913, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.8005982041358948, + "num_tokens": 666006201.0, + "step": 25816 + }, + { + "epoch": 2.128885956955554, + "grad_norm": 2.1049585342407227, + "learning_rate": 1e-06, + "loss": 0.5503, + "mean_token_accuracy": 0.8058180809020996, + "num_tokens": 666032958.0, + "step": 25817 + }, + { + "epoch": 2.128968417580605, + "grad_norm": 1.9849859476089478, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.7846401333808899, + "num_tokens": 666062921.0, + "step": 25818 + }, + { + "epoch": 2.129050878205657, + "grad_norm": 2.100299596786499, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7627860307693481, + "num_tokens": 666092424.0, + "step": 25819 + }, + { + "epoch": 2.1291333388307083, + "grad_norm": 2.064710855484009, + "learning_rate": 1e-06, + "loss": 0.5761, + "mean_token_accuracy": 0.8000704646110535, + "num_tokens": 666118189.0, + "step": 25820 + }, + { + "epoch": 2.1292157994557597, + "grad_norm": 2.2225098609924316, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.770147442817688, + "num_tokens": 666146244.0, + "step": 25821 + }, + { + "epoch": 2.1292982600808115, + "grad_norm": 2.2864344120025635, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7832634449005127, + "num_tokens": 666171976.0, + "step": 25822 + }, + { + "epoch": 2.129380720705863, + "grad_norm": 1.873907208442688, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.765653669834137, + "num_tokens": 666205139.0, + "step": 25823 + }, + { + "epoch": 2.1294631813309146, + "grad_norm": 2.0342116355895996, + "learning_rate": 1e-06, + "loss": 0.6049, + "mean_token_accuracy": 0.7933065891265869, + "num_tokens": 666239969.0, + "step": 25824 + }, + { + "epoch": 2.129545641955966, + "grad_norm": 2.0518338680267334, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7463552951812744, + "num_tokens": 666273089.0, + "step": 25825 + }, + { + "epoch": 2.1296281025810178, + "grad_norm": 2.0450241565704346, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7833939790725708, + "num_tokens": 666301579.0, + "step": 25826 + }, + { + "epoch": 2.129710563206069, + "grad_norm": 1.969083309173584, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.7932331562042236, + "num_tokens": 666332739.0, + "step": 25827 + }, + { + "epoch": 2.1297930238311205, + "grad_norm": 2.014129161834717, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7813831567764282, + "num_tokens": 666366332.0, + "step": 25828 + }, + { + "epoch": 2.1298754844561723, + "grad_norm": 2.2687785625457764, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7204104661941528, + "num_tokens": 666394106.0, + "step": 25829 + }, + { + "epoch": 2.1299579450812236, + "grad_norm": 2.1107261180877686, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7490496039390564, + "num_tokens": 666425184.0, + "step": 25830 + }, + { + "epoch": 2.1300404057062754, + "grad_norm": 2.0143399238586426, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7614380121231079, + "num_tokens": 666458272.0, + "step": 25831 + }, + { + "epoch": 2.1301228663313267, + "grad_norm": 2.089843988418579, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.765287458896637, + "num_tokens": 666491367.0, + "step": 25832 + }, + { + "epoch": 2.1302053269563785, + "grad_norm": 1.8886237144470215, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7673273682594299, + "num_tokens": 666523709.0, + "step": 25833 + }, + { + "epoch": 2.13028778758143, + "grad_norm": 2.245649814605713, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7338895797729492, + "num_tokens": 666550002.0, + "step": 25834 + }, + { + "epoch": 2.1303702482064812, + "grad_norm": 1.9948029518127441, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7629936337471008, + "num_tokens": 666581584.0, + "step": 25835 + }, + { + "epoch": 2.130452708831533, + "grad_norm": 2.0304646492004395, + "learning_rate": 1e-06, + "loss": 0.572, + "mean_token_accuracy": 0.8096869587898254, + "num_tokens": 666611047.0, + "step": 25836 + }, + { + "epoch": 2.1305351694565844, + "grad_norm": 2.3224446773529053, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.746266782283783, + "num_tokens": 666636221.0, + "step": 25837 + }, + { + "epoch": 2.130617630081636, + "grad_norm": 2.2897040843963623, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7345775961875916, + "num_tokens": 666662360.0, + "step": 25838 + }, + { + "epoch": 2.1307000907066875, + "grad_norm": 2.467900276184082, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.7868528962135315, + "num_tokens": 666685015.0, + "step": 25839 + }, + { + "epoch": 2.1307825513317393, + "grad_norm": 2.8950793743133545, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7452812194824219, + "num_tokens": 666702631.0, + "step": 25840 + }, + { + "epoch": 2.1308650119567907, + "grad_norm": 2.227407932281494, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.7880952954292297, + "num_tokens": 666729978.0, + "step": 25841 + }, + { + "epoch": 2.130947472581842, + "grad_norm": 2.1483492851257324, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7652890682220459, + "num_tokens": 666757638.0, + "step": 25842 + }, + { + "epoch": 2.131029933206894, + "grad_norm": 2.573967456817627, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7447797060012817, + "num_tokens": 666779826.0, + "step": 25843 + }, + { + "epoch": 2.131112393831945, + "grad_norm": 2.213193893432617, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7712516784667969, + "num_tokens": 666807816.0, + "step": 25844 + }, + { + "epoch": 2.131194854456997, + "grad_norm": 2.0680549144744873, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7362211346626282, + "num_tokens": 666838543.0, + "step": 25845 + }, + { + "epoch": 2.1312773150820483, + "grad_norm": 2.1664018630981445, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7663054466247559, + "num_tokens": 666870777.0, + "step": 25846 + }, + { + "epoch": 2.1313597757070997, + "grad_norm": 2.218456745147705, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.7677142024040222, + "num_tokens": 666898570.0, + "step": 25847 + }, + { + "epoch": 2.1314422363321515, + "grad_norm": 2.1059789657592773, + "learning_rate": 1e-06, + "loss": 0.5965, + "mean_token_accuracy": 0.8002741932868958, + "num_tokens": 666928081.0, + "step": 25848 + }, + { + "epoch": 2.131524696957203, + "grad_norm": 1.9495253562927246, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7718421220779419, + "num_tokens": 666962198.0, + "step": 25849 + }, + { + "epoch": 2.1316071575822546, + "grad_norm": 2.0834567546844482, + "learning_rate": 1e-06, + "loss": 0.5907, + "mean_token_accuracy": 0.7960371971130371, + "num_tokens": 666992839.0, + "step": 25850 + }, + { + "epoch": 2.131689618207306, + "grad_norm": 2.2911481857299805, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7662147283554077, + "num_tokens": 667019559.0, + "step": 25851 + }, + { + "epoch": 2.1317720788323578, + "grad_norm": 2.0136332511901855, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.779945969581604, + "num_tokens": 667053826.0, + "step": 25852 + }, + { + "epoch": 2.131854539457409, + "grad_norm": 2.059391975402832, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7358685731887817, + "num_tokens": 667084431.0, + "step": 25853 + }, + { + "epoch": 2.1319370000824605, + "grad_norm": 2.786607027053833, + "learning_rate": 1e-06, + "loss": 0.586, + "mean_token_accuracy": 0.7981420755386353, + "num_tokens": 667111002.0, + "step": 25854 + }, + { + "epoch": 2.1320194607075122, + "grad_norm": 2.0500752925872803, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.794174313545227, + "num_tokens": 667141864.0, + "step": 25855 + }, + { + "epoch": 2.1321019213325636, + "grad_norm": 1.8167420625686646, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.7796238660812378, + "num_tokens": 667174712.0, + "step": 25856 + }, + { + "epoch": 2.1321843819576154, + "grad_norm": 1.9783347845077515, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.762873649597168, + "num_tokens": 667206684.0, + "step": 25857 + }, + { + "epoch": 2.1322668425826667, + "grad_norm": 2.2486958503723145, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7375293374061584, + "num_tokens": 667232959.0, + "step": 25858 + }, + { + "epoch": 2.132349303207718, + "grad_norm": 2.2767646312713623, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7669476270675659, + "num_tokens": 667260442.0, + "step": 25859 + }, + { + "epoch": 2.13243176383277, + "grad_norm": 2.067415952682495, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.761314868927002, + "num_tokens": 667292263.0, + "step": 25860 + }, + { + "epoch": 2.1325142244578212, + "grad_norm": 2.0969274044036865, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7746974229812622, + "num_tokens": 667320662.0, + "step": 25861 + }, + { + "epoch": 2.132596685082873, + "grad_norm": 1.9618394374847412, + "learning_rate": 1e-06, + "loss": 0.6062, + "mean_token_accuracy": 0.8023141622543335, + "num_tokens": 667357163.0, + "step": 25862 + }, + { + "epoch": 2.1326791457079244, + "grad_norm": 2.222391128540039, + "learning_rate": 1e-06, + "loss": 0.598, + "mean_token_accuracy": 0.7983050346374512, + "num_tokens": 667386122.0, + "step": 25863 + }, + { + "epoch": 2.132761606332976, + "grad_norm": 2.137958288192749, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7698115110397339, + "num_tokens": 667415942.0, + "step": 25864 + }, + { + "epoch": 2.1328440669580275, + "grad_norm": 2.2662513256073, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.771902859210968, + "num_tokens": 667443454.0, + "step": 25865 + }, + { + "epoch": 2.132926527583079, + "grad_norm": 1.9385262727737427, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.7983756065368652, + "num_tokens": 667475144.0, + "step": 25866 + }, + { + "epoch": 2.1330089882081307, + "grad_norm": 2.08699369430542, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.7898881435394287, + "num_tokens": 667506188.0, + "step": 25867 + }, + { + "epoch": 2.133091448833182, + "grad_norm": 2.2828495502471924, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7464170455932617, + "num_tokens": 667533719.0, + "step": 25868 + }, + { + "epoch": 2.133173909458234, + "grad_norm": 2.177920341491699, + "learning_rate": 1e-06, + "loss": 0.6062, + "mean_token_accuracy": 0.8003119230270386, + "num_tokens": 667558767.0, + "step": 25869 + }, + { + "epoch": 2.133256370083285, + "grad_norm": 2.3911068439483643, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.7945051193237305, + "num_tokens": 667582732.0, + "step": 25870 + }, + { + "epoch": 2.133338830708337, + "grad_norm": 2.106450319290161, + "learning_rate": 1e-06, + "loss": 0.5907, + "mean_token_accuracy": 0.7984832525253296, + "num_tokens": 667609409.0, + "step": 25871 + }, + { + "epoch": 2.1334212913333883, + "grad_norm": 2.387662649154663, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7425366640090942, + "num_tokens": 667633431.0, + "step": 25872 + }, + { + "epoch": 2.1335037519584397, + "grad_norm": 2.021111249923706, + "learning_rate": 1e-06, + "loss": 0.5995, + "mean_token_accuracy": 0.807568371295929, + "num_tokens": 667661980.0, + "step": 25873 + }, + { + "epoch": 2.1335862125834915, + "grad_norm": 1.934754490852356, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7772252559661865, + "num_tokens": 667692653.0, + "step": 25874 + }, + { + "epoch": 2.133668673208543, + "grad_norm": 2.487515926361084, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.752404510974884, + "num_tokens": 667717882.0, + "step": 25875 + }, + { + "epoch": 2.1337511338335946, + "grad_norm": 2.184873342514038, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7598137855529785, + "num_tokens": 667747999.0, + "step": 25876 + }, + { + "epoch": 2.133833594458646, + "grad_norm": 2.3820741176605225, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7633992433547974, + "num_tokens": 667772044.0, + "step": 25877 + }, + { + "epoch": 2.1339160550836977, + "grad_norm": 1.9342106580734253, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7436290979385376, + "num_tokens": 667805334.0, + "step": 25878 + }, + { + "epoch": 2.133998515708749, + "grad_norm": 1.92038094997406, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7749395370483398, + "num_tokens": 667840536.0, + "step": 25879 + }, + { + "epoch": 2.1340809763338005, + "grad_norm": 2.0992488861083984, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7717498540878296, + "num_tokens": 667871687.0, + "step": 25880 + }, + { + "epoch": 2.1341634369588522, + "grad_norm": 2.09490704536438, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7638793587684631, + "num_tokens": 667903003.0, + "step": 25881 + }, + { + "epoch": 2.1342458975839036, + "grad_norm": 1.8704522848129272, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7765729427337646, + "num_tokens": 667939455.0, + "step": 25882 + }, + { + "epoch": 2.1343283582089554, + "grad_norm": 2.3066561222076416, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7651225328445435, + "num_tokens": 667967651.0, + "step": 25883 + }, + { + "epoch": 2.1344108188340067, + "grad_norm": 2.0410895347595215, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.772749662399292, + "num_tokens": 668001667.0, + "step": 25884 + }, + { + "epoch": 2.1344932794590585, + "grad_norm": 1.9487327337265015, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7601274251937866, + "num_tokens": 668034759.0, + "step": 25885 + }, + { + "epoch": 2.13457574008411, + "grad_norm": 2.0862698554992676, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7797968983650208, + "num_tokens": 668065072.0, + "step": 25886 + }, + { + "epoch": 2.1346582007091612, + "grad_norm": 2.1036078929901123, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.734214186668396, + "num_tokens": 668097681.0, + "step": 25887 + }, + { + "epoch": 2.134740661334213, + "grad_norm": 1.8978374004364014, + "learning_rate": 1e-06, + "loss": 0.5983, + "mean_token_accuracy": 0.7976573705673218, + "num_tokens": 668133037.0, + "step": 25888 + }, + { + "epoch": 2.1348231219592644, + "grad_norm": 2.311119556427002, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.7829241156578064, + "num_tokens": 668160641.0, + "step": 25889 + }, + { + "epoch": 2.134905582584316, + "grad_norm": 2.1670782566070557, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7552074790000916, + "num_tokens": 668191115.0, + "step": 25890 + }, + { + "epoch": 2.1349880432093675, + "grad_norm": 1.9639629125595093, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.7914972305297852, + "num_tokens": 668228958.0, + "step": 25891 + }, + { + "epoch": 2.135070503834419, + "grad_norm": 2.19600248336792, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7715615034103394, + "num_tokens": 668257183.0, + "step": 25892 + }, + { + "epoch": 2.1351529644594707, + "grad_norm": 2.0038132667541504, + "learning_rate": 1e-06, + "loss": 0.5542, + "mean_token_accuracy": 0.8176189661026001, + "num_tokens": 668285746.0, + "step": 25893 + }, + { + "epoch": 2.135235425084522, + "grad_norm": 2.3267478942871094, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.788774311542511, + "num_tokens": 668310714.0, + "step": 25894 + }, + { + "epoch": 2.135317885709574, + "grad_norm": 2.1801726818084717, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7695580124855042, + "num_tokens": 668337737.0, + "step": 25895 + }, + { + "epoch": 2.135400346334625, + "grad_norm": 2.0219600200653076, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7484740018844604, + "num_tokens": 668368533.0, + "step": 25896 + }, + { + "epoch": 2.135482806959677, + "grad_norm": 2.2102386951446533, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7690371870994568, + "num_tokens": 668397035.0, + "step": 25897 + }, + { + "epoch": 2.1355652675847283, + "grad_norm": 2.2672064304351807, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7379493713378906, + "num_tokens": 668425546.0, + "step": 25898 + }, + { + "epoch": 2.1356477282097797, + "grad_norm": 2.0922024250030518, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7700504660606384, + "num_tokens": 668454595.0, + "step": 25899 + }, + { + "epoch": 2.1357301888348315, + "grad_norm": 2.0896170139312744, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7702957391738892, + "num_tokens": 668483748.0, + "step": 25900 + }, + { + "epoch": 2.135812649459883, + "grad_norm": 2.00227689743042, + "learning_rate": 1e-06, + "loss": 0.5755, + "mean_token_accuracy": 0.8086545467376709, + "num_tokens": 668511600.0, + "step": 25901 + }, + { + "epoch": 2.1358951100849346, + "grad_norm": 2.055164098739624, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7692581415176392, + "num_tokens": 668541298.0, + "step": 25902 + }, + { + "epoch": 2.135977570709986, + "grad_norm": 1.9343466758728027, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.7987831830978394, + "num_tokens": 668574057.0, + "step": 25903 + }, + { + "epoch": 2.1360600313350373, + "grad_norm": 2.1662020683288574, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7640317678451538, + "num_tokens": 668601782.0, + "step": 25904 + }, + { + "epoch": 2.136142491960089, + "grad_norm": 2.5598084926605225, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.792567253112793, + "num_tokens": 668621670.0, + "step": 25905 + }, + { + "epoch": 2.1362249525851404, + "grad_norm": 2.4775826930999756, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.7840771079063416, + "num_tokens": 668646292.0, + "step": 25906 + }, + { + "epoch": 2.1363074132101922, + "grad_norm": 1.94767165184021, + "learning_rate": 1e-06, + "loss": 0.5829, + "mean_token_accuracy": 0.8014037013053894, + "num_tokens": 668676461.0, + "step": 25907 + }, + { + "epoch": 2.1363898738352436, + "grad_norm": 2.213688850402832, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7741045951843262, + "num_tokens": 668703088.0, + "step": 25908 + }, + { + "epoch": 2.1364723344602954, + "grad_norm": 2.0492770671844482, + "learning_rate": 1e-06, + "loss": 0.7954, + "mean_token_accuracy": 0.7532404661178589, + "num_tokens": 668734731.0, + "step": 25909 + }, + { + "epoch": 2.1365547950853467, + "grad_norm": 2.0573465824127197, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7638872265815735, + "num_tokens": 668765184.0, + "step": 25910 + }, + { + "epoch": 2.136637255710398, + "grad_norm": 2.203462839126587, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7627381086349487, + "num_tokens": 668794793.0, + "step": 25911 + }, + { + "epoch": 2.13671971633545, + "grad_norm": 1.9224804639816284, + "learning_rate": 1e-06, + "loss": 0.5942, + "mean_token_accuracy": 0.8020758628845215, + "num_tokens": 668826059.0, + "step": 25912 + }, + { + "epoch": 2.1368021769605012, + "grad_norm": 1.9093148708343506, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7861345410346985, + "num_tokens": 668859546.0, + "step": 25913 + }, + { + "epoch": 2.136884637585553, + "grad_norm": 2.128568649291992, + "learning_rate": 1e-06, + "loss": 0.5114, + "mean_token_accuracy": 0.818474531173706, + "num_tokens": 668886362.0, + "step": 25914 + }, + { + "epoch": 2.1369670982106044, + "grad_norm": 2.0074079036712646, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7587677240371704, + "num_tokens": 668919171.0, + "step": 25915 + }, + { + "epoch": 2.137049558835656, + "grad_norm": 2.1284372806549072, + "learning_rate": 1e-06, + "loss": 0.5661, + "mean_token_accuracy": 0.8072841167449951, + "num_tokens": 668947224.0, + "step": 25916 + }, + { + "epoch": 2.1371320194607075, + "grad_norm": 2.3649232387542725, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.771374523639679, + "num_tokens": 668974003.0, + "step": 25917 + }, + { + "epoch": 2.137214480085759, + "grad_norm": 2.351963996887207, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7615064382553101, + "num_tokens": 668999579.0, + "step": 25918 + }, + { + "epoch": 2.1372969407108107, + "grad_norm": 2.2928149700164795, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7801741361618042, + "num_tokens": 669024107.0, + "step": 25919 + }, + { + "epoch": 2.137379401335862, + "grad_norm": 2.2887253761291504, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7567130327224731, + "num_tokens": 669048805.0, + "step": 25920 + }, + { + "epoch": 2.137461861960914, + "grad_norm": 2.084221124649048, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7561876773834229, + "num_tokens": 669078321.0, + "step": 25921 + }, + { + "epoch": 2.137544322585965, + "grad_norm": 2.1666836738586426, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7772668600082397, + "num_tokens": 669104881.0, + "step": 25922 + }, + { + "epoch": 2.137626783211017, + "grad_norm": 1.937005877494812, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7509573698043823, + "num_tokens": 669141737.0, + "step": 25923 + }, + { + "epoch": 2.1377092438360683, + "grad_norm": 2.1393980979919434, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.729235053062439, + "num_tokens": 669171973.0, + "step": 25924 + }, + { + "epoch": 2.1377917044611197, + "grad_norm": 2.1030843257904053, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7276023030281067, + "num_tokens": 669204149.0, + "step": 25925 + }, + { + "epoch": 2.1378741650861715, + "grad_norm": 1.9384706020355225, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7695578336715698, + "num_tokens": 669237830.0, + "step": 25926 + }, + { + "epoch": 2.137956625711223, + "grad_norm": 1.8563034534454346, + "learning_rate": 1e-06, + "loss": 0.5705, + "mean_token_accuracy": 0.8015509843826294, + "num_tokens": 669269869.0, + "step": 25927 + }, + { + "epoch": 2.1380390863362746, + "grad_norm": 2.0126914978027344, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7636884450912476, + "num_tokens": 669303124.0, + "step": 25928 + }, + { + "epoch": 2.138121546961326, + "grad_norm": 2.055975914001465, + "learning_rate": 1e-06, + "loss": 0.479, + "mean_token_accuracy": 0.8196344971656799, + "num_tokens": 669330379.0, + "step": 25929 + }, + { + "epoch": 2.1382040075863773, + "grad_norm": 2.372537136077881, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7332532405853271, + "num_tokens": 669356675.0, + "step": 25930 + }, + { + "epoch": 2.138286468211429, + "grad_norm": 2.288804531097412, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7587616443634033, + "num_tokens": 669384212.0, + "step": 25931 + }, + { + "epoch": 2.1383689288364804, + "grad_norm": 2.059166431427002, + "learning_rate": 1e-06, + "loss": 0.5418, + "mean_token_accuracy": 0.8170784711837769, + "num_tokens": 669413188.0, + "step": 25932 + }, + { + "epoch": 2.1384513894615322, + "grad_norm": 2.2530510425567627, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.776690661907196, + "num_tokens": 669438075.0, + "step": 25933 + }, + { + "epoch": 2.1385338500865836, + "grad_norm": 2.093102216720581, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.7888213992118835, + "num_tokens": 669469623.0, + "step": 25934 + }, + { + "epoch": 2.1386163107116354, + "grad_norm": 2.3174829483032227, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7776702046394348, + "num_tokens": 669494716.0, + "step": 25935 + }, + { + "epoch": 2.1386987713366867, + "grad_norm": 2.3744747638702393, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7718565464019775, + "num_tokens": 669521751.0, + "step": 25936 + }, + { + "epoch": 2.138781231961738, + "grad_norm": 2.290863275527954, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7621040344238281, + "num_tokens": 669547236.0, + "step": 25937 + }, + { + "epoch": 2.13886369258679, + "grad_norm": 1.9141558408737183, + "learning_rate": 1e-06, + "loss": 0.5009, + "mean_token_accuracy": 0.8251296877861023, + "num_tokens": 669581369.0, + "step": 25938 + }, + { + "epoch": 2.1389461532118412, + "grad_norm": 2.0553178787231445, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.7929682731628418, + "num_tokens": 669610280.0, + "step": 25939 + }, + { + "epoch": 2.139028613836893, + "grad_norm": 2.461127758026123, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7517380714416504, + "num_tokens": 669633781.0, + "step": 25940 + }, + { + "epoch": 2.1391110744619444, + "grad_norm": 2.098723888397217, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7722188234329224, + "num_tokens": 669668567.0, + "step": 25941 + }, + { + "epoch": 2.139193535086996, + "grad_norm": 2.165147066116333, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.7912025451660156, + "num_tokens": 669694695.0, + "step": 25942 + }, + { + "epoch": 2.1392759957120475, + "grad_norm": 1.8869980573654175, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.7606536746025085, + "num_tokens": 669731856.0, + "step": 25943 + }, + { + "epoch": 2.139358456337099, + "grad_norm": 2.170449733734131, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7798104882240295, + "num_tokens": 669760793.0, + "step": 25944 + }, + { + "epoch": 2.1394409169621507, + "grad_norm": 2.3046247959136963, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7932226657867432, + "num_tokens": 669785442.0, + "step": 25945 + }, + { + "epoch": 2.139523377587202, + "grad_norm": 1.8475708961486816, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7680380344390869, + "num_tokens": 669819396.0, + "step": 25946 + }, + { + "epoch": 2.139605838212254, + "grad_norm": 2.198228359222412, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7749245166778564, + "num_tokens": 669846678.0, + "step": 25947 + }, + { + "epoch": 2.139688298837305, + "grad_norm": 2.1515727043151855, + "learning_rate": 1e-06, + "loss": 0.5731, + "mean_token_accuracy": 0.8025913238525391, + "num_tokens": 669871201.0, + "step": 25948 + }, + { + "epoch": 2.1397707594623565, + "grad_norm": 1.9354497194290161, + "learning_rate": 1e-06, + "loss": 0.5922, + "mean_token_accuracy": 0.785851001739502, + "num_tokens": 669903392.0, + "step": 25949 + }, + { + "epoch": 2.1398532200874083, + "grad_norm": 2.199894666671753, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7587158679962158, + "num_tokens": 669935032.0, + "step": 25950 + }, + { + "epoch": 2.1399356807124597, + "grad_norm": 2.096907138824463, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.764382004737854, + "num_tokens": 669966927.0, + "step": 25951 + }, + { + "epoch": 2.1400181413375114, + "grad_norm": 2.2052481174468994, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7477033138275146, + "num_tokens": 669994959.0, + "step": 25952 + }, + { + "epoch": 2.140100601962563, + "grad_norm": 2.1527867317199707, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.777451753616333, + "num_tokens": 670024503.0, + "step": 25953 + }, + { + "epoch": 2.1401830625876146, + "grad_norm": 1.9611711502075195, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7576013207435608, + "num_tokens": 670058292.0, + "step": 25954 + }, + { + "epoch": 2.140265523212666, + "grad_norm": 2.240518808364868, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7465397119522095, + "num_tokens": 670086177.0, + "step": 25955 + }, + { + "epoch": 2.1403479838377173, + "grad_norm": 2.1164045333862305, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7757948040962219, + "num_tokens": 670115055.0, + "step": 25956 + }, + { + "epoch": 2.140430444462769, + "grad_norm": 2.047604560852051, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.7865550518035889, + "num_tokens": 670144254.0, + "step": 25957 + }, + { + "epoch": 2.1405129050878204, + "grad_norm": 1.979289174079895, + "learning_rate": 1e-06, + "loss": 0.6013, + "mean_token_accuracy": 0.796644926071167, + "num_tokens": 670180027.0, + "step": 25958 + }, + { + "epoch": 2.1405953657128722, + "grad_norm": 2.410238265991211, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7359036803245544, + "num_tokens": 670205324.0, + "step": 25959 + }, + { + "epoch": 2.1406778263379236, + "grad_norm": 2.056107759475708, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7798562049865723, + "num_tokens": 670236820.0, + "step": 25960 + }, + { + "epoch": 2.1407602869629754, + "grad_norm": 2.1496543884277344, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.802052915096283, + "num_tokens": 670266982.0, + "step": 25961 + }, + { + "epoch": 2.1408427475880267, + "grad_norm": 1.660345435142517, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7845139503479004, + "num_tokens": 670308872.0, + "step": 25962 + }, + { + "epoch": 2.140925208213078, + "grad_norm": 2.3779537677764893, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7592982053756714, + "num_tokens": 670332691.0, + "step": 25963 + }, + { + "epoch": 2.14100766883813, + "grad_norm": 2.1281633377075195, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7094674110412598, + "num_tokens": 670363832.0, + "step": 25964 + }, + { + "epoch": 2.1410901294631812, + "grad_norm": 2.011502265930176, + "learning_rate": 1e-06, + "loss": 0.6112, + "mean_token_accuracy": 0.7984060049057007, + "num_tokens": 670391831.0, + "step": 25965 + }, + { + "epoch": 2.141172590088233, + "grad_norm": 2.2979636192321777, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7507205009460449, + "num_tokens": 670416618.0, + "step": 25966 + }, + { + "epoch": 2.1412550507132844, + "grad_norm": 2.082153558731079, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7775372862815857, + "num_tokens": 670448697.0, + "step": 25967 + }, + { + "epoch": 2.141337511338336, + "grad_norm": 1.9975571632385254, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7768774032592773, + "num_tokens": 670482259.0, + "step": 25968 + }, + { + "epoch": 2.1414199719633875, + "grad_norm": 2.056468963623047, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.78974449634552, + "num_tokens": 670513148.0, + "step": 25969 + }, + { + "epoch": 2.141502432588439, + "grad_norm": 2.1906325817108154, + "learning_rate": 1e-06, + "loss": 0.5543, + "mean_token_accuracy": 0.8038530349731445, + "num_tokens": 670542904.0, + "step": 25970 + }, + { + "epoch": 2.1415848932134907, + "grad_norm": 1.996424674987793, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7712447047233582, + "num_tokens": 670575800.0, + "step": 25971 + }, + { + "epoch": 2.141667353838542, + "grad_norm": 2.306490421295166, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7552334070205688, + "num_tokens": 670600512.0, + "step": 25972 + }, + { + "epoch": 2.141749814463594, + "grad_norm": 1.9311134815216064, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7597034573554993, + "num_tokens": 670633237.0, + "step": 25973 + }, + { + "epoch": 2.141832275088645, + "grad_norm": 2.0383782386779785, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7720884680747986, + "num_tokens": 670666984.0, + "step": 25974 + }, + { + "epoch": 2.1419147357136965, + "grad_norm": 2.1296627521514893, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7380910515785217, + "num_tokens": 670694746.0, + "step": 25975 + }, + { + "epoch": 2.1419971963387483, + "grad_norm": 2.051422595977783, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.7990210056304932, + "num_tokens": 670724446.0, + "step": 25976 + }, + { + "epoch": 2.1420796569637996, + "grad_norm": 2.1914262771606445, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7574180364608765, + "num_tokens": 670751545.0, + "step": 25977 + }, + { + "epoch": 2.1421621175888514, + "grad_norm": 2.0530166625976562, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7479976415634155, + "num_tokens": 670784814.0, + "step": 25978 + }, + { + "epoch": 2.142244578213903, + "grad_norm": 2.154216766357422, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7587618231773376, + "num_tokens": 670812610.0, + "step": 25979 + }, + { + "epoch": 2.1423270388389546, + "grad_norm": 1.8858609199523926, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.7870736718177795, + "num_tokens": 670846166.0, + "step": 25980 + }, + { + "epoch": 2.142409499464006, + "grad_norm": 2.213993787765503, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.7895141243934631, + "num_tokens": 670872713.0, + "step": 25981 + }, + { + "epoch": 2.1424919600890573, + "grad_norm": 2.4341201782226562, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7392821311950684, + "num_tokens": 670900195.0, + "step": 25982 + }, + { + "epoch": 2.142574420714109, + "grad_norm": 2.375912666320801, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.725920557975769, + "num_tokens": 670925879.0, + "step": 25983 + }, + { + "epoch": 2.1426568813391604, + "grad_norm": 2.0466127395629883, + "learning_rate": 1e-06, + "loss": 0.599, + "mean_token_accuracy": 0.8010551333427429, + "num_tokens": 670955939.0, + "step": 25984 + }, + { + "epoch": 2.1427393419642122, + "grad_norm": 1.9537001848220825, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7718737721443176, + "num_tokens": 670989329.0, + "step": 25985 + }, + { + "epoch": 2.1428218025892636, + "grad_norm": 2.0020880699157715, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7495675086975098, + "num_tokens": 671022693.0, + "step": 25986 + }, + { + "epoch": 2.142904263214315, + "grad_norm": 2.0229644775390625, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7794507741928101, + "num_tokens": 671054087.0, + "step": 25987 + }, + { + "epoch": 2.1429867238393667, + "grad_norm": 2.3114094734191895, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7519859671592712, + "num_tokens": 671081142.0, + "step": 25988 + }, + { + "epoch": 2.143069184464418, + "grad_norm": 1.8953343629837036, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.7911381721496582, + "num_tokens": 671117220.0, + "step": 25989 + }, + { + "epoch": 2.14315164508947, + "grad_norm": 2.3483710289001465, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7426666617393494, + "num_tokens": 671145617.0, + "step": 25990 + }, + { + "epoch": 2.143234105714521, + "grad_norm": 2.0553767681121826, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7753322124481201, + "num_tokens": 671176231.0, + "step": 25991 + }, + { + "epoch": 2.143316566339573, + "grad_norm": 2.2111191749572754, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.7893929481506348, + "num_tokens": 671204566.0, + "step": 25992 + }, + { + "epoch": 2.1433990269646244, + "grad_norm": 2.130751371383667, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7723137140274048, + "num_tokens": 671231408.0, + "step": 25993 + }, + { + "epoch": 2.1434814875896757, + "grad_norm": 2.2551660537719727, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7489279508590698, + "num_tokens": 671258210.0, + "step": 25994 + }, + { + "epoch": 2.1435639482147275, + "grad_norm": 2.0049898624420166, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.783509373664856, + "num_tokens": 671289945.0, + "step": 25995 + }, + { + "epoch": 2.143646408839779, + "grad_norm": 2.15112042427063, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7590286731719971, + "num_tokens": 671321313.0, + "step": 25996 + }, + { + "epoch": 2.1437288694648307, + "grad_norm": 2.0844342708587646, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7298189997673035, + "num_tokens": 671354122.0, + "step": 25997 + }, + { + "epoch": 2.143811330089882, + "grad_norm": 2.38750958442688, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7686195373535156, + "num_tokens": 671381818.0, + "step": 25998 + }, + { + "epoch": 2.143893790714934, + "grad_norm": 2.2363224029541016, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7618534564971924, + "num_tokens": 671409247.0, + "step": 25999 + }, + { + "epoch": 2.143976251339985, + "grad_norm": 2.110196590423584, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7540121078491211, + "num_tokens": 671438875.0, + "step": 26000 + }, + { + "epoch": 2.1440587119650365, + "grad_norm": 2.133341073989868, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7639876008033752, + "num_tokens": 671467587.0, + "step": 26001 + }, + { + "epoch": 2.1441411725900883, + "grad_norm": 2.1666219234466553, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.7977346777915955, + "num_tokens": 671495584.0, + "step": 26002 + }, + { + "epoch": 2.1442236332151396, + "grad_norm": 2.1917264461517334, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7802811861038208, + "num_tokens": 671522490.0, + "step": 26003 + }, + { + "epoch": 2.1443060938401914, + "grad_norm": 2.1968026161193848, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7419910430908203, + "num_tokens": 671552792.0, + "step": 26004 + }, + { + "epoch": 2.144388554465243, + "grad_norm": 2.159553289413452, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.7827239036560059, + "num_tokens": 671582518.0, + "step": 26005 + }, + { + "epoch": 2.1444710150902946, + "grad_norm": 1.8627197742462158, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.7870221138000488, + "num_tokens": 671618499.0, + "step": 26006 + }, + { + "epoch": 2.144553475715346, + "grad_norm": 2.0945630073547363, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7606622576713562, + "num_tokens": 671648436.0, + "step": 26007 + }, + { + "epoch": 2.1446359363403973, + "grad_norm": 2.266038656234741, + "learning_rate": 1e-06, + "loss": 0.5975, + "mean_token_accuracy": 0.7932767271995544, + "num_tokens": 671675244.0, + "step": 26008 + }, + { + "epoch": 2.144718396965449, + "grad_norm": 2.464125633239746, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7752600908279419, + "num_tokens": 671696827.0, + "step": 26009 + }, + { + "epoch": 2.1448008575905004, + "grad_norm": 2.253706693649292, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7511030435562134, + "num_tokens": 671722717.0, + "step": 26010 + }, + { + "epoch": 2.1448833182155522, + "grad_norm": 1.9974173307418823, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7907994985580444, + "num_tokens": 671754871.0, + "step": 26011 + }, + { + "epoch": 2.1449657788406036, + "grad_norm": 2.218532085418701, + "learning_rate": 1e-06, + "loss": 0.6239, + "mean_token_accuracy": 0.7755340337753296, + "num_tokens": 671779605.0, + "step": 26012 + }, + { + "epoch": 2.1450482394656554, + "grad_norm": 2.107583522796631, + "learning_rate": 1e-06, + "loss": 0.5865, + "mean_token_accuracy": 0.8139715194702148, + "num_tokens": 671807218.0, + "step": 26013 + }, + { + "epoch": 2.1451307000907067, + "grad_norm": 2.0196573734283447, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.7581143379211426, + "num_tokens": 671838446.0, + "step": 26014 + }, + { + "epoch": 2.145213160715758, + "grad_norm": 2.0353105068206787, + "learning_rate": 1e-06, + "loss": 0.5758, + "mean_token_accuracy": 0.8017617464065552, + "num_tokens": 671869121.0, + "step": 26015 + }, + { + "epoch": 2.14529562134081, + "grad_norm": 1.9756988286972046, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7703452110290527, + "num_tokens": 671903280.0, + "step": 26016 + }, + { + "epoch": 2.145378081965861, + "grad_norm": 1.8349803686141968, + "learning_rate": 1e-06, + "loss": 0.5509, + "mean_token_accuracy": 0.8149976134300232, + "num_tokens": 671942802.0, + "step": 26017 + }, + { + "epoch": 2.145460542590913, + "grad_norm": 2.176013946533203, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7405712604522705, + "num_tokens": 671970230.0, + "step": 26018 + }, + { + "epoch": 2.1455430032159644, + "grad_norm": 2.139538049697876, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7666609287261963, + "num_tokens": 671997372.0, + "step": 26019 + }, + { + "epoch": 2.1456254638410157, + "grad_norm": 2.110656976699829, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.745855450630188, + "num_tokens": 672025470.0, + "step": 26020 + }, + { + "epoch": 2.1457079244660675, + "grad_norm": 2.191396474838257, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7798276543617249, + "num_tokens": 672054161.0, + "step": 26021 + }, + { + "epoch": 2.145790385091119, + "grad_norm": 2.300206184387207, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7535943388938904, + "num_tokens": 672078715.0, + "step": 26022 + }, + { + "epoch": 2.1458728457161707, + "grad_norm": 1.8626234531402588, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7704017162322998, + "num_tokens": 672115384.0, + "step": 26023 + }, + { + "epoch": 2.145955306341222, + "grad_norm": 2.108959913253784, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7774503827095032, + "num_tokens": 672146334.0, + "step": 26024 + }, + { + "epoch": 2.146037766966274, + "grad_norm": 2.3307952880859375, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7679179906845093, + "num_tokens": 672171269.0, + "step": 26025 + }, + { + "epoch": 2.146120227591325, + "grad_norm": 2.688356876373291, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7700811624526978, + "num_tokens": 672193708.0, + "step": 26026 + }, + { + "epoch": 2.1462026882163765, + "grad_norm": 1.9233185052871704, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7553053498268127, + "num_tokens": 672226651.0, + "step": 26027 + }, + { + "epoch": 2.1462851488414283, + "grad_norm": 2.032285213470459, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7381064891815186, + "num_tokens": 672258526.0, + "step": 26028 + }, + { + "epoch": 2.1463676094664796, + "grad_norm": 2.295959949493408, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7374622821807861, + "num_tokens": 672287951.0, + "step": 26029 + }, + { + "epoch": 2.1464500700915314, + "grad_norm": 1.981674075126648, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.737783670425415, + "num_tokens": 672322089.0, + "step": 26030 + }, + { + "epoch": 2.146532530716583, + "grad_norm": 2.2535266876220703, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7734783887863159, + "num_tokens": 672347834.0, + "step": 26031 + }, + { + "epoch": 2.146614991341634, + "grad_norm": 1.9286425113677979, + "learning_rate": 1e-06, + "loss": 0.5526, + "mean_token_accuracy": 0.8162137866020203, + "num_tokens": 672380227.0, + "step": 26032 + }, + { + "epoch": 2.146697451966686, + "grad_norm": 2.3119096755981445, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.7884413599967957, + "num_tokens": 672406640.0, + "step": 26033 + }, + { + "epoch": 2.1467799125917373, + "grad_norm": 2.085810661315918, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7567098140716553, + "num_tokens": 672435261.0, + "step": 26034 + }, + { + "epoch": 2.146862373216789, + "grad_norm": 2.114809989929199, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.7723988890647888, + "num_tokens": 672464775.0, + "step": 26035 + }, + { + "epoch": 2.1469448338418404, + "grad_norm": 2.232024908065796, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.7871682643890381, + "num_tokens": 672492033.0, + "step": 26036 + }, + { + "epoch": 2.1470272944668922, + "grad_norm": 2.0411155223846436, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7641335725784302, + "num_tokens": 672524019.0, + "step": 26037 + }, + { + "epoch": 2.1471097550919436, + "grad_norm": 2.229956865310669, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.7897484302520752, + "num_tokens": 672555186.0, + "step": 26038 + }, + { + "epoch": 2.147192215716995, + "grad_norm": 2.0062203407287598, + "learning_rate": 1e-06, + "loss": 0.7621, + "mean_token_accuracy": 0.7594706416130066, + "num_tokens": 672588865.0, + "step": 26039 + }, + { + "epoch": 2.1472746763420467, + "grad_norm": 2.071026086807251, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7865293025970459, + "num_tokens": 672618019.0, + "step": 26040 + }, + { + "epoch": 2.147357136967098, + "grad_norm": 2.027460813522339, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7684959173202515, + "num_tokens": 672649661.0, + "step": 26041 + }, + { + "epoch": 2.14743959759215, + "grad_norm": 2.0401461124420166, + "learning_rate": 1e-06, + "loss": 0.8064, + "mean_token_accuracy": 0.7379903793334961, + "num_tokens": 672684987.0, + "step": 26042 + }, + { + "epoch": 2.147522058217201, + "grad_norm": 2.128206729888916, + "learning_rate": 1e-06, + "loss": 0.6098, + "mean_token_accuracy": 0.8025293350219727, + "num_tokens": 672710922.0, + "step": 26043 + }, + { + "epoch": 2.147604518842253, + "grad_norm": 2.132685899734497, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.7432383298873901, + "num_tokens": 672740052.0, + "step": 26044 + }, + { + "epoch": 2.1476869794673044, + "grad_norm": 2.1161091327667236, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7356644868850708, + "num_tokens": 672772632.0, + "step": 26045 + }, + { + "epoch": 2.1477694400923557, + "grad_norm": 2.172501564025879, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.775145411491394, + "num_tokens": 672803381.0, + "step": 26046 + }, + { + "epoch": 2.1478519007174075, + "grad_norm": 2.297076463699341, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.794248640537262, + "num_tokens": 672832560.0, + "step": 26047 + }, + { + "epoch": 2.147934361342459, + "grad_norm": 2.254545211791992, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7650438547134399, + "num_tokens": 672859881.0, + "step": 26048 + }, + { + "epoch": 2.1480168219675106, + "grad_norm": 2.08382248878479, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7297205924987793, + "num_tokens": 672893082.0, + "step": 26049 + }, + { + "epoch": 2.148099282592562, + "grad_norm": 2.150564432144165, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7767423391342163, + "num_tokens": 672924490.0, + "step": 26050 + }, + { + "epoch": 2.148181743217614, + "grad_norm": 1.928849458694458, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7589815258979797, + "num_tokens": 672961191.0, + "step": 26051 + }, + { + "epoch": 2.148264203842665, + "grad_norm": 1.8339042663574219, + "learning_rate": 1e-06, + "loss": 0.6123, + "mean_token_accuracy": 0.8046663403511047, + "num_tokens": 672995932.0, + "step": 26052 + }, + { + "epoch": 2.1483466644677165, + "grad_norm": 1.942656397819519, + "learning_rate": 1e-06, + "loss": 0.5996, + "mean_token_accuracy": 0.7954130172729492, + "num_tokens": 673030119.0, + "step": 26053 + }, + { + "epoch": 2.1484291250927683, + "grad_norm": 1.9385889768600464, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7645312547683716, + "num_tokens": 673062988.0, + "step": 26054 + }, + { + "epoch": 2.1485115857178196, + "grad_norm": 2.060877799987793, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7562747001647949, + "num_tokens": 673093544.0, + "step": 26055 + }, + { + "epoch": 2.1485940463428714, + "grad_norm": 2.0499768257141113, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7625625729560852, + "num_tokens": 673124977.0, + "step": 26056 + }, + { + "epoch": 2.148676506967923, + "grad_norm": 2.110727548599243, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7473429441452026, + "num_tokens": 673157258.0, + "step": 26057 + }, + { + "epoch": 2.1487589675929746, + "grad_norm": 1.9984564781188965, + "learning_rate": 1e-06, + "loss": 0.5721, + "mean_token_accuracy": 0.8063794374465942, + "num_tokens": 673189549.0, + "step": 26058 + }, + { + "epoch": 2.148841428218026, + "grad_norm": 1.740138292312622, + "learning_rate": 1e-06, + "loss": 0.5482, + "mean_token_accuracy": 0.8158457279205322, + "num_tokens": 673229111.0, + "step": 26059 + }, + { + "epoch": 2.1489238888430773, + "grad_norm": 2.1852073669433594, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7717279195785522, + "num_tokens": 673255918.0, + "step": 26060 + }, + { + "epoch": 2.149006349468129, + "grad_norm": 1.8533668518066406, + "learning_rate": 1e-06, + "loss": 0.5692, + "mean_token_accuracy": 0.8045610785484314, + "num_tokens": 673289974.0, + "step": 26061 + }, + { + "epoch": 2.1490888100931804, + "grad_norm": 2.3442275524139404, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7495124936103821, + "num_tokens": 673313805.0, + "step": 26062 + }, + { + "epoch": 2.149171270718232, + "grad_norm": 2.4145658016204834, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7753360271453857, + "num_tokens": 673339034.0, + "step": 26063 + }, + { + "epoch": 2.1492537313432836, + "grad_norm": 1.9724369049072266, + "learning_rate": 1e-06, + "loss": 0.5523, + "mean_token_accuracy": 0.8116633296012878, + "num_tokens": 673372210.0, + "step": 26064 + }, + { + "epoch": 2.149336191968335, + "grad_norm": 1.8121891021728516, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.786318302154541, + "num_tokens": 673408522.0, + "step": 26065 + }, + { + "epoch": 2.1494186525933867, + "grad_norm": 2.2001466751098633, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.7925519943237305, + "num_tokens": 673435407.0, + "step": 26066 + }, + { + "epoch": 2.149501113218438, + "grad_norm": 1.9605016708374023, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.7543683648109436, + "num_tokens": 673468506.0, + "step": 26067 + }, + { + "epoch": 2.14958357384349, + "grad_norm": 1.9557723999023438, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7767565250396729, + "num_tokens": 673505597.0, + "step": 26068 + }, + { + "epoch": 2.149666034468541, + "grad_norm": 2.0213611125946045, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.7920618653297424, + "num_tokens": 673535934.0, + "step": 26069 + }, + { + "epoch": 2.149748495093593, + "grad_norm": 2.2068262100219727, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.7768322825431824, + "num_tokens": 673563915.0, + "step": 26070 + }, + { + "epoch": 2.1498309557186444, + "grad_norm": 2.1161091327667236, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7428412437438965, + "num_tokens": 673596944.0, + "step": 26071 + }, + { + "epoch": 2.1499134163436957, + "grad_norm": 1.9859603643417358, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7865368127822876, + "num_tokens": 673629258.0, + "step": 26072 + }, + { + "epoch": 2.1499958769687475, + "grad_norm": 2.076077461242676, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.7970457077026367, + "num_tokens": 673663881.0, + "step": 26073 + }, + { + "epoch": 2.150078337593799, + "grad_norm": 2.147125244140625, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7776652574539185, + "num_tokens": 673694164.0, + "step": 26074 + }, + { + "epoch": 2.1501607982188506, + "grad_norm": 2.1304450035095215, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7448458671569824, + "num_tokens": 673724940.0, + "step": 26075 + }, + { + "epoch": 2.150243258843902, + "grad_norm": 2.0839767456054688, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7555382251739502, + "num_tokens": 673754123.0, + "step": 26076 + }, + { + "epoch": 2.1503257194689533, + "grad_norm": 2.622504711151123, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7615467309951782, + "num_tokens": 673774365.0, + "step": 26077 + }, + { + "epoch": 2.150408180094005, + "grad_norm": 2.3512864112854004, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7451152801513672, + "num_tokens": 673801089.0, + "step": 26078 + }, + { + "epoch": 2.1504906407190565, + "grad_norm": 2.239429473876953, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7595409154891968, + "num_tokens": 673826837.0, + "step": 26079 + }, + { + "epoch": 2.1505731013441083, + "grad_norm": 1.8749388456344604, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7754034399986267, + "num_tokens": 673863825.0, + "step": 26080 + }, + { + "epoch": 2.1506555619691596, + "grad_norm": 2.0242860317230225, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7402185201644897, + "num_tokens": 673894701.0, + "step": 26081 + }, + { + "epoch": 2.1507380225942114, + "grad_norm": 1.6788766384124756, + "learning_rate": 1e-06, + "loss": 0.5704, + "mean_token_accuracy": 0.8073067665100098, + "num_tokens": 673933602.0, + "step": 26082 + }, + { + "epoch": 2.150820483219263, + "grad_norm": 2.0664479732513428, + "learning_rate": 1e-06, + "loss": 0.5922, + "mean_token_accuracy": 0.7980858087539673, + "num_tokens": 673962972.0, + "step": 26083 + }, + { + "epoch": 2.150902943844314, + "grad_norm": 1.9824258089065552, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7801563739776611, + "num_tokens": 673996797.0, + "step": 26084 + }, + { + "epoch": 2.150985404469366, + "grad_norm": 1.8192862272262573, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7705023884773254, + "num_tokens": 674039664.0, + "step": 26085 + }, + { + "epoch": 2.1510678650944173, + "grad_norm": 2.0516178607940674, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7548165321350098, + "num_tokens": 674072443.0, + "step": 26086 + }, + { + "epoch": 2.151150325719469, + "grad_norm": 2.119506597518921, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7559682130813599, + "num_tokens": 674107138.0, + "step": 26087 + }, + { + "epoch": 2.1512327863445204, + "grad_norm": 2.0236423015594482, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7894673347473145, + "num_tokens": 674139263.0, + "step": 26088 + }, + { + "epoch": 2.151315246969572, + "grad_norm": 1.7767969369888306, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.7883290648460388, + "num_tokens": 674178254.0, + "step": 26089 + }, + { + "epoch": 2.1513977075946236, + "grad_norm": 2.290031671524048, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7330348491668701, + "num_tokens": 674206290.0, + "step": 26090 + }, + { + "epoch": 2.151480168219675, + "grad_norm": 2.213465452194214, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.747493326663971, + "num_tokens": 674238553.0, + "step": 26091 + }, + { + "epoch": 2.1515626288447267, + "grad_norm": 2.038933038711548, + "learning_rate": 1e-06, + "loss": 0.6075, + "mean_token_accuracy": 0.8000028133392334, + "num_tokens": 674265641.0, + "step": 26092 + }, + { + "epoch": 2.151645089469778, + "grad_norm": 2.202763557434082, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.7826176881790161, + "num_tokens": 674292675.0, + "step": 26093 + }, + { + "epoch": 2.15172755009483, + "grad_norm": 2.018024206161499, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7883781790733337, + "num_tokens": 674324685.0, + "step": 26094 + }, + { + "epoch": 2.151810010719881, + "grad_norm": 2.2285120487213135, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.7984347343444824, + "num_tokens": 674349880.0, + "step": 26095 + }, + { + "epoch": 2.151892471344933, + "grad_norm": 1.8536239862442017, + "learning_rate": 1e-06, + "loss": 0.5937, + "mean_token_accuracy": 0.7926496863365173, + "num_tokens": 674385362.0, + "step": 26096 + }, + { + "epoch": 2.1519749319699844, + "grad_norm": 2.25223970413208, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.7617135643959045, + "num_tokens": 674413957.0, + "step": 26097 + }, + { + "epoch": 2.1520573925950357, + "grad_norm": 2.32529354095459, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.768031120300293, + "num_tokens": 674438194.0, + "step": 26098 + }, + { + "epoch": 2.1521398532200875, + "grad_norm": 1.993219017982483, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7536381483078003, + "num_tokens": 674472051.0, + "step": 26099 + }, + { + "epoch": 2.152222313845139, + "grad_norm": 1.9754117727279663, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.786766767501831, + "num_tokens": 674506194.0, + "step": 26100 + }, + { + "epoch": 2.1523047744701906, + "grad_norm": 2.136593818664551, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.7539295554161072, + "num_tokens": 674533909.0, + "step": 26101 + }, + { + "epoch": 2.152387235095242, + "grad_norm": 1.9775395393371582, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7466665506362915, + "num_tokens": 674566873.0, + "step": 26102 + }, + { + "epoch": 2.152469695720294, + "grad_norm": 2.023674964904785, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7608271837234497, + "num_tokens": 674600101.0, + "step": 26103 + }, + { + "epoch": 2.152552156345345, + "grad_norm": 1.9177440404891968, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7710035443305969, + "num_tokens": 674633039.0, + "step": 26104 + }, + { + "epoch": 2.1526346169703965, + "grad_norm": 1.8779945373535156, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7863985300064087, + "num_tokens": 674666185.0, + "step": 26105 + }, + { + "epoch": 2.1527170775954483, + "grad_norm": 2.340838670730591, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7682685852050781, + "num_tokens": 674690666.0, + "step": 26106 + }, + { + "epoch": 2.1527995382204996, + "grad_norm": 2.225346088409424, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7428168654441833, + "num_tokens": 674718257.0, + "step": 26107 + }, + { + "epoch": 2.1528819988455514, + "grad_norm": 1.9725340604782104, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7474117875099182, + "num_tokens": 674751147.0, + "step": 26108 + }, + { + "epoch": 2.1529644594706028, + "grad_norm": 1.8269084692001343, + "learning_rate": 1e-06, + "loss": 0.5737, + "mean_token_accuracy": 0.8058447241783142, + "num_tokens": 674786403.0, + "step": 26109 + }, + { + "epoch": 2.153046920095654, + "grad_norm": 1.8257222175598145, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.7882208824157715, + "num_tokens": 674823526.0, + "step": 26110 + }, + { + "epoch": 2.153129380720706, + "grad_norm": 1.9987614154815674, + "learning_rate": 1e-06, + "loss": 0.6027, + "mean_token_accuracy": 0.7846266627311707, + "num_tokens": 674853623.0, + "step": 26111 + }, + { + "epoch": 2.1532118413457573, + "grad_norm": 2.3614635467529297, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7774521708488464, + "num_tokens": 674877104.0, + "step": 26112 + }, + { + "epoch": 2.153294301970809, + "grad_norm": 2.236130952835083, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7561797499656677, + "num_tokens": 674902772.0, + "step": 26113 + }, + { + "epoch": 2.1533767625958604, + "grad_norm": 1.9838900566101074, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7491411566734314, + "num_tokens": 674936069.0, + "step": 26114 + }, + { + "epoch": 2.153459223220912, + "grad_norm": 2.417080879211426, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7663408517837524, + "num_tokens": 674961840.0, + "step": 26115 + }, + { + "epoch": 2.1535416838459636, + "grad_norm": 2.2658045291900635, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.7894188761711121, + "num_tokens": 674985979.0, + "step": 26116 + }, + { + "epoch": 2.153624144471015, + "grad_norm": 2.109990119934082, + "learning_rate": 1e-06, + "loss": 0.607, + "mean_token_accuracy": 0.798506498336792, + "num_tokens": 675014408.0, + "step": 26117 + }, + { + "epoch": 2.1537066050960667, + "grad_norm": 2.123472213745117, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7713542580604553, + "num_tokens": 675045917.0, + "step": 26118 + }, + { + "epoch": 2.153789065721118, + "grad_norm": 2.0750865936279297, + "learning_rate": 1e-06, + "loss": 0.5386, + "mean_token_accuracy": 0.8225017189979553, + "num_tokens": 675074711.0, + "step": 26119 + }, + { + "epoch": 2.15387152634617, + "grad_norm": 2.0436315536499023, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.7937130928039551, + "num_tokens": 675107064.0, + "step": 26120 + }, + { + "epoch": 2.153953986971221, + "grad_norm": 1.9257315397262573, + "learning_rate": 1e-06, + "loss": 0.597, + "mean_token_accuracy": 0.7985914945602417, + "num_tokens": 675141697.0, + "step": 26121 + }, + { + "epoch": 2.1540364475962726, + "grad_norm": 2.06306529045105, + "learning_rate": 1e-06, + "loss": 0.5956, + "mean_token_accuracy": 0.8026272654533386, + "num_tokens": 675169346.0, + "step": 26122 + }, + { + "epoch": 2.1541189082213243, + "grad_norm": 2.1670491695404053, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7460734248161316, + "num_tokens": 675195415.0, + "step": 26123 + }, + { + "epoch": 2.1542013688463757, + "grad_norm": 1.7956074476242065, + "learning_rate": 1e-06, + "loss": 0.503, + "mean_token_accuracy": 0.837553858757019, + "num_tokens": 675233937.0, + "step": 26124 + }, + { + "epoch": 2.1542838294714275, + "grad_norm": 2.258175849914551, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7800166606903076, + "num_tokens": 675258917.0, + "step": 26125 + }, + { + "epoch": 2.154366290096479, + "grad_norm": 2.045206308364868, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7659851908683777, + "num_tokens": 675289743.0, + "step": 26126 + }, + { + "epoch": 2.1544487507215306, + "grad_norm": 1.9661015272140503, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7778308391571045, + "num_tokens": 675321992.0, + "step": 26127 + }, + { + "epoch": 2.154531211346582, + "grad_norm": 2.159593343734741, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7555978894233704, + "num_tokens": 675350983.0, + "step": 26128 + }, + { + "epoch": 2.1546136719716333, + "grad_norm": 1.9328590631484985, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7712185382843018, + "num_tokens": 675385627.0, + "step": 26129 + }, + { + "epoch": 2.154696132596685, + "grad_norm": 1.994714617729187, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7779601812362671, + "num_tokens": 675422852.0, + "step": 26130 + }, + { + "epoch": 2.1547785932217365, + "grad_norm": 1.9104465246200562, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.774918258190155, + "num_tokens": 675458639.0, + "step": 26131 + }, + { + "epoch": 2.1548610538467883, + "grad_norm": 1.958485722541809, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.765662670135498, + "num_tokens": 675494006.0, + "step": 26132 + }, + { + "epoch": 2.1549435144718396, + "grad_norm": 2.0958564281463623, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.7866612672805786, + "num_tokens": 675525065.0, + "step": 26133 + }, + { + "epoch": 2.1550259750968914, + "grad_norm": 1.9417450428009033, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.7881779670715332, + "num_tokens": 675558414.0, + "step": 26134 + }, + { + "epoch": 2.1551084357219428, + "grad_norm": 2.020329475402832, + "learning_rate": 1e-06, + "loss": 0.5757, + "mean_token_accuracy": 0.8137995004653931, + "num_tokens": 675589521.0, + "step": 26135 + }, + { + "epoch": 2.155190896346994, + "grad_norm": 2.3371734619140625, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7607672214508057, + "num_tokens": 675614673.0, + "step": 26136 + }, + { + "epoch": 2.155273356972046, + "grad_norm": 2.042266607284546, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7434514164924622, + "num_tokens": 675650169.0, + "step": 26137 + }, + { + "epoch": 2.1553558175970973, + "grad_norm": 1.885764241218567, + "learning_rate": 1e-06, + "loss": 0.5456, + "mean_token_accuracy": 0.8131210803985596, + "num_tokens": 675685354.0, + "step": 26138 + }, + { + "epoch": 2.155438278222149, + "grad_norm": 2.0933127403259277, + "learning_rate": 1e-06, + "loss": 0.6021, + "mean_token_accuracy": 0.7899993658065796, + "num_tokens": 675713765.0, + "step": 26139 + }, + { + "epoch": 2.1555207388472004, + "grad_norm": 1.8121192455291748, + "learning_rate": 1e-06, + "loss": 0.5287, + "mean_token_accuracy": 0.8191972374916077, + "num_tokens": 675748641.0, + "step": 26140 + }, + { + "epoch": 2.155603199472252, + "grad_norm": 1.8428484201431274, + "learning_rate": 1e-06, + "loss": 0.5865, + "mean_token_accuracy": 0.8096701502799988, + "num_tokens": 675781475.0, + "step": 26141 + }, + { + "epoch": 2.1556856600973036, + "grad_norm": 2.040769577026367, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.773256778717041, + "num_tokens": 675816050.0, + "step": 26142 + }, + { + "epoch": 2.155768120722355, + "grad_norm": 2.0722272396087646, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7501143217086792, + "num_tokens": 675846781.0, + "step": 26143 + }, + { + "epoch": 2.1558505813474067, + "grad_norm": 2.125004291534424, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7658917903900146, + "num_tokens": 675876404.0, + "step": 26144 + }, + { + "epoch": 2.155933041972458, + "grad_norm": 2.0040013790130615, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.792702853679657, + "num_tokens": 675909738.0, + "step": 26145 + }, + { + "epoch": 2.15601550259751, + "grad_norm": 2.084667921066284, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7616789937019348, + "num_tokens": 675938635.0, + "step": 26146 + }, + { + "epoch": 2.156097963222561, + "grad_norm": 2.104776620864868, + "learning_rate": 1e-06, + "loss": 0.5377, + "mean_token_accuracy": 0.8137997984886169, + "num_tokens": 675964846.0, + "step": 26147 + }, + { + "epoch": 2.156180423847613, + "grad_norm": 1.8965333700180054, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.7828460931777954, + "num_tokens": 675998221.0, + "step": 26148 + }, + { + "epoch": 2.1562628844726643, + "grad_norm": 1.8423900604248047, + "learning_rate": 1e-06, + "loss": 0.5183, + "mean_token_accuracy": 0.8269333243370056, + "num_tokens": 676031352.0, + "step": 26149 + }, + { + "epoch": 2.1563453450977157, + "grad_norm": 1.9990047216415405, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7702493667602539, + "num_tokens": 676062631.0, + "step": 26150 + }, + { + "epoch": 2.1564278057227675, + "grad_norm": 2.1596696376800537, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7484146356582642, + "num_tokens": 676092755.0, + "step": 26151 + }, + { + "epoch": 2.156510266347819, + "grad_norm": 2.3241589069366455, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7233083248138428, + "num_tokens": 676121816.0, + "step": 26152 + }, + { + "epoch": 2.1565927269728706, + "grad_norm": 2.077183246612549, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.76120924949646, + "num_tokens": 676151649.0, + "step": 26153 + }, + { + "epoch": 2.156675187597922, + "grad_norm": 2.1708226203918457, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7681245803833008, + "num_tokens": 676179144.0, + "step": 26154 + }, + { + "epoch": 2.1567576482229733, + "grad_norm": 1.9050910472869873, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.7949482798576355, + "num_tokens": 676215608.0, + "step": 26155 + }, + { + "epoch": 2.156840108848025, + "grad_norm": 2.08354115486145, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.7871608734130859, + "num_tokens": 676245415.0, + "step": 26156 + }, + { + "epoch": 2.1569225694730765, + "grad_norm": 2.2075905799865723, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.7921009063720703, + "num_tokens": 676268770.0, + "step": 26157 + }, + { + "epoch": 2.1570050300981283, + "grad_norm": 2.5444185733795166, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7310587167739868, + "num_tokens": 676292191.0, + "step": 26158 + }, + { + "epoch": 2.1570874907231796, + "grad_norm": 2.3307366371154785, + "learning_rate": 1e-06, + "loss": 0.5968, + "mean_token_accuracy": 0.7910341024398804, + "num_tokens": 676316335.0, + "step": 26159 + }, + { + "epoch": 2.1571699513482314, + "grad_norm": 2.036547899246216, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7693620324134827, + "num_tokens": 676347549.0, + "step": 26160 + }, + { + "epoch": 2.1572524119732828, + "grad_norm": 2.1762337684631348, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7252094149589539, + "num_tokens": 676376510.0, + "step": 26161 + }, + { + "epoch": 2.157334872598334, + "grad_norm": 2.3219497203826904, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.7887237071990967, + "num_tokens": 676401796.0, + "step": 26162 + }, + { + "epoch": 2.157417333223386, + "grad_norm": 1.933090090751648, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7711829543113708, + "num_tokens": 676439694.0, + "step": 26163 + }, + { + "epoch": 2.1574997938484373, + "grad_norm": 2.1558828353881836, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7623710632324219, + "num_tokens": 676468778.0, + "step": 26164 + }, + { + "epoch": 2.157582254473489, + "grad_norm": 2.3143961429595947, + "learning_rate": 1e-06, + "loss": 0.6024, + "mean_token_accuracy": 0.7978074550628662, + "num_tokens": 676491732.0, + "step": 26165 + }, + { + "epoch": 2.1576647150985404, + "grad_norm": 1.9183036088943481, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.7817105054855347, + "num_tokens": 676525988.0, + "step": 26166 + }, + { + "epoch": 2.1577471757235918, + "grad_norm": 1.9780488014221191, + "learning_rate": 1e-06, + "loss": 0.5833, + "mean_token_accuracy": 0.80461585521698, + "num_tokens": 676553521.0, + "step": 26167 + }, + { + "epoch": 2.1578296363486436, + "grad_norm": 2.0994114875793457, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7653079032897949, + "num_tokens": 676582219.0, + "step": 26168 + }, + { + "epoch": 2.157912096973695, + "grad_norm": 2.015918254852295, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7213433980941772, + "num_tokens": 676617002.0, + "step": 26169 + }, + { + "epoch": 2.1579945575987467, + "grad_norm": 2.074488878250122, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7779338359832764, + "num_tokens": 676651832.0, + "step": 26170 + }, + { + "epoch": 2.158077018223798, + "grad_norm": 2.1919355392456055, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.756646454334259, + "num_tokens": 676680897.0, + "step": 26171 + }, + { + "epoch": 2.15815947884885, + "grad_norm": 2.0774953365325928, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7727459669113159, + "num_tokens": 676711507.0, + "step": 26172 + }, + { + "epoch": 2.158241939473901, + "grad_norm": 2.3293278217315674, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7694283723831177, + "num_tokens": 676736506.0, + "step": 26173 + }, + { + "epoch": 2.1583244000989525, + "grad_norm": 2.3477959632873535, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7831697463989258, + "num_tokens": 676762045.0, + "step": 26174 + }, + { + "epoch": 2.1584068607240043, + "grad_norm": 2.199821710586548, + "learning_rate": 1e-06, + "loss": 0.5917, + "mean_token_accuracy": 0.8002111911773682, + "num_tokens": 676790631.0, + "step": 26175 + }, + { + "epoch": 2.1584893213490557, + "grad_norm": 2.025125503540039, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7739185094833374, + "num_tokens": 676823536.0, + "step": 26176 + }, + { + "epoch": 2.1585717819741075, + "grad_norm": 2.1096413135528564, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.738163948059082, + "num_tokens": 676854891.0, + "step": 26177 + }, + { + "epoch": 2.158654242599159, + "grad_norm": 2.4505136013031006, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7467833757400513, + "num_tokens": 676879635.0, + "step": 26178 + }, + { + "epoch": 2.1587367032242106, + "grad_norm": 2.353602409362793, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.774855375289917, + "num_tokens": 676905419.0, + "step": 26179 + }, + { + "epoch": 2.158819163849262, + "grad_norm": 2.1671388149261475, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7799234390258789, + "num_tokens": 676933897.0, + "step": 26180 + }, + { + "epoch": 2.1589016244743133, + "grad_norm": 1.8836253881454468, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7870854139328003, + "num_tokens": 676968982.0, + "step": 26181 + }, + { + "epoch": 2.158984085099365, + "grad_norm": 1.946110725402832, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7675840854644775, + "num_tokens": 677002147.0, + "step": 26182 + }, + { + "epoch": 2.1590665457244165, + "grad_norm": 2.2244620323181152, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7584490776062012, + "num_tokens": 677030751.0, + "step": 26183 + }, + { + "epoch": 2.1591490063494683, + "grad_norm": 2.08864426612854, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.771661639213562, + "num_tokens": 677060654.0, + "step": 26184 + }, + { + "epoch": 2.1592314669745196, + "grad_norm": 2.7156238555908203, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7776370644569397, + "num_tokens": 677081361.0, + "step": 26185 + }, + { + "epoch": 2.1593139275995714, + "grad_norm": 2.4545137882232666, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.7427698373794556, + "num_tokens": 677105817.0, + "step": 26186 + }, + { + "epoch": 2.1593963882246228, + "grad_norm": 2.015793561935425, + "learning_rate": 1e-06, + "loss": 0.4792, + "mean_token_accuracy": 0.8317541480064392, + "num_tokens": 677132191.0, + "step": 26187 + }, + { + "epoch": 2.159478848849674, + "grad_norm": 2.153782367706299, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7802504301071167, + "num_tokens": 677158739.0, + "step": 26188 + }, + { + "epoch": 2.159561309474726, + "grad_norm": 1.9776296615600586, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.7798227667808533, + "num_tokens": 677192708.0, + "step": 26189 + }, + { + "epoch": 2.1596437700997773, + "grad_norm": 2.347576141357422, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7793598771095276, + "num_tokens": 677218141.0, + "step": 26190 + }, + { + "epoch": 2.159726230724829, + "grad_norm": 2.0536484718322754, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7781359553337097, + "num_tokens": 677251171.0, + "step": 26191 + }, + { + "epoch": 2.1598086913498804, + "grad_norm": 2.1233463287353516, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7895507216453552, + "num_tokens": 677278661.0, + "step": 26192 + }, + { + "epoch": 2.159891151974932, + "grad_norm": 2.0219311714172363, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7727203369140625, + "num_tokens": 677310961.0, + "step": 26193 + }, + { + "epoch": 2.1599736125999836, + "grad_norm": 2.292288064956665, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.7604867815971375, + "num_tokens": 677336557.0, + "step": 26194 + }, + { + "epoch": 2.160056073225035, + "grad_norm": 2.089144229888916, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.7939959764480591, + "num_tokens": 677365335.0, + "step": 26195 + }, + { + "epoch": 2.1601385338500867, + "grad_norm": 1.924353837966919, + "learning_rate": 1e-06, + "loss": 0.5505, + "mean_token_accuracy": 0.823544442653656, + "num_tokens": 677399207.0, + "step": 26196 + }, + { + "epoch": 2.160220994475138, + "grad_norm": 2.342721462249756, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7615160942077637, + "num_tokens": 677422880.0, + "step": 26197 + }, + { + "epoch": 2.16030345510019, + "grad_norm": 2.199852466583252, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7716981768608093, + "num_tokens": 677449316.0, + "step": 26198 + }, + { + "epoch": 2.160385915725241, + "grad_norm": 2.0795819759368896, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.7829033136367798, + "num_tokens": 677479474.0, + "step": 26199 + }, + { + "epoch": 2.1604683763502925, + "grad_norm": 2.214303970336914, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7717663645744324, + "num_tokens": 677507828.0, + "step": 26200 + }, + { + "epoch": 2.1605508369753443, + "grad_norm": 2.4400289058685303, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.769243597984314, + "num_tokens": 677531545.0, + "step": 26201 + }, + { + "epoch": 2.1606332976003957, + "grad_norm": 1.9447849988937378, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.7897138595581055, + "num_tokens": 677563009.0, + "step": 26202 + }, + { + "epoch": 2.1607157582254475, + "grad_norm": 1.9086322784423828, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.7914121150970459, + "num_tokens": 677598787.0, + "step": 26203 + }, + { + "epoch": 2.160798218850499, + "grad_norm": 2.1800708770751953, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7335658073425293, + "num_tokens": 677630013.0, + "step": 26204 + }, + { + "epoch": 2.1608806794755506, + "grad_norm": 2.2004849910736084, + "learning_rate": 1e-06, + "loss": 0.5653, + "mean_token_accuracy": 0.8074319362640381, + "num_tokens": 677657820.0, + "step": 26205 + }, + { + "epoch": 2.160963140100602, + "grad_norm": 1.848541259765625, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.7845878005027771, + "num_tokens": 677696989.0, + "step": 26206 + }, + { + "epoch": 2.1610456007256533, + "grad_norm": 2.0806326866149902, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7543092966079712, + "num_tokens": 677730395.0, + "step": 26207 + }, + { + "epoch": 2.161128061350705, + "grad_norm": 2.5936050415039062, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7573003172874451, + "num_tokens": 677751182.0, + "step": 26208 + }, + { + "epoch": 2.1612105219757565, + "grad_norm": 2.0776286125183105, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.7884707450866699, + "num_tokens": 677779082.0, + "step": 26209 + }, + { + "epoch": 2.1612929826008083, + "grad_norm": 2.0504047870635986, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.7865725159645081, + "num_tokens": 677810833.0, + "step": 26210 + }, + { + "epoch": 2.1613754432258596, + "grad_norm": 2.133234977722168, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7640691995620728, + "num_tokens": 677840296.0, + "step": 26211 + }, + { + "epoch": 2.161457903850911, + "grad_norm": 2.0373308658599854, + "learning_rate": 1e-06, + "loss": 0.5566, + "mean_token_accuracy": 0.8048357367515564, + "num_tokens": 677870868.0, + "step": 26212 + }, + { + "epoch": 2.1615403644759628, + "grad_norm": 1.9573566913604736, + "learning_rate": 1e-06, + "loss": 0.4188, + "mean_token_accuracy": 0.8535063862800598, + "num_tokens": 677902229.0, + "step": 26213 + }, + { + "epoch": 2.161622825101014, + "grad_norm": 2.187030553817749, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7322348952293396, + "num_tokens": 677932647.0, + "step": 26214 + }, + { + "epoch": 2.161705285726066, + "grad_norm": 2.0283689498901367, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7445980906486511, + "num_tokens": 677965285.0, + "step": 26215 + }, + { + "epoch": 2.1617877463511173, + "grad_norm": 2.663557529449463, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7380708456039429, + "num_tokens": 677985729.0, + "step": 26216 + }, + { + "epoch": 2.161870206976169, + "grad_norm": 1.8819383382797241, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.7806921005249023, + "num_tokens": 678019896.0, + "step": 26217 + }, + { + "epoch": 2.1619526676012204, + "grad_norm": 1.9527262449264526, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7711247205734253, + "num_tokens": 678051085.0, + "step": 26218 + }, + { + "epoch": 2.1620351282262718, + "grad_norm": 1.9292066097259521, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7624433040618896, + "num_tokens": 678084141.0, + "step": 26219 + }, + { + "epoch": 2.1621175888513235, + "grad_norm": 1.7839739322662354, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7851261496543884, + "num_tokens": 678124689.0, + "step": 26220 + }, + { + "epoch": 2.162200049476375, + "grad_norm": 1.84844970703125, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7626395225524902, + "num_tokens": 678160766.0, + "step": 26221 + }, + { + "epoch": 2.1622825101014267, + "grad_norm": 2.15254807472229, + "learning_rate": 1e-06, + "loss": 0.5283, + "mean_token_accuracy": 0.8197917342185974, + "num_tokens": 678189663.0, + "step": 26222 + }, + { + "epoch": 2.162364970726478, + "grad_norm": 2.285423994064331, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.729519784450531, + "num_tokens": 678215634.0, + "step": 26223 + }, + { + "epoch": 2.16244743135153, + "grad_norm": 2.1479263305664062, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7786448001861572, + "num_tokens": 678243390.0, + "step": 26224 + }, + { + "epoch": 2.162529891976581, + "grad_norm": 1.8877161741256714, + "learning_rate": 1e-06, + "loss": 0.5471, + "mean_token_accuracy": 0.8213348388671875, + "num_tokens": 678275781.0, + "step": 26225 + }, + { + "epoch": 2.1626123526016325, + "grad_norm": 1.7206785678863525, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7407022714614868, + "num_tokens": 678319393.0, + "step": 26226 + }, + { + "epoch": 2.1626948132266843, + "grad_norm": 2.01851749420166, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7613651752471924, + "num_tokens": 678353260.0, + "step": 26227 + }, + { + "epoch": 2.1627772738517357, + "grad_norm": 1.9419028759002686, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7769029140472412, + "num_tokens": 678385650.0, + "step": 26228 + }, + { + "epoch": 2.1628597344767875, + "grad_norm": 2.036043643951416, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.7869666814804077, + "num_tokens": 678417389.0, + "step": 26229 + }, + { + "epoch": 2.162942195101839, + "grad_norm": 2.2064645290374756, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7652472257614136, + "num_tokens": 678444821.0, + "step": 26230 + }, + { + "epoch": 2.1630246557268906, + "grad_norm": 2.130255937576294, + "learning_rate": 1e-06, + "loss": 0.5818, + "mean_token_accuracy": 0.8084577322006226, + "num_tokens": 678472821.0, + "step": 26231 + }, + { + "epoch": 2.163107116351942, + "grad_norm": 2.1303024291992188, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7525688409805298, + "num_tokens": 678500376.0, + "step": 26232 + }, + { + "epoch": 2.1631895769769933, + "grad_norm": 2.242356538772583, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7505589723587036, + "num_tokens": 678529350.0, + "step": 26233 + }, + { + "epoch": 2.163272037602045, + "grad_norm": 2.178008794784546, + "learning_rate": 1e-06, + "loss": 0.5792, + "mean_token_accuracy": 0.8029878735542297, + "num_tokens": 678556327.0, + "step": 26234 + }, + { + "epoch": 2.1633544982270965, + "grad_norm": 2.221208095550537, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.7921732068061829, + "num_tokens": 678584206.0, + "step": 26235 + }, + { + "epoch": 2.1634369588521483, + "grad_norm": 2.125379800796509, + "learning_rate": 1e-06, + "loss": 0.593, + "mean_token_accuracy": 0.8045321702957153, + "num_tokens": 678610538.0, + "step": 26236 + }, + { + "epoch": 2.1635194194771996, + "grad_norm": 2.4062447547912598, + "learning_rate": 1e-06, + "loss": 0.5802, + "mean_token_accuracy": 0.7973601222038269, + "num_tokens": 678633315.0, + "step": 26237 + }, + { + "epoch": 2.1636018801022514, + "grad_norm": 2.2138288021087646, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.78325355052948, + "num_tokens": 678658621.0, + "step": 26238 + }, + { + "epoch": 2.1636843407273028, + "grad_norm": 1.8939194679260254, + "learning_rate": 1e-06, + "loss": 0.4814, + "mean_token_accuracy": 0.8374369144439697, + "num_tokens": 678691136.0, + "step": 26239 + }, + { + "epoch": 2.163766801352354, + "grad_norm": 1.9900449514389038, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7709105610847473, + "num_tokens": 678722742.0, + "step": 26240 + }, + { + "epoch": 2.163849261977406, + "grad_norm": 2.6577236652374268, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7422056198120117, + "num_tokens": 678742744.0, + "step": 26241 + }, + { + "epoch": 2.1639317226024573, + "grad_norm": 2.079695224761963, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.795056164264679, + "num_tokens": 678773352.0, + "step": 26242 + }, + { + "epoch": 2.164014183227509, + "grad_norm": 2.1682872772216797, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7298867106437683, + "num_tokens": 678802378.0, + "step": 26243 + }, + { + "epoch": 2.1640966438525604, + "grad_norm": 2.170898199081421, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.788572371006012, + "num_tokens": 678830451.0, + "step": 26244 + }, + { + "epoch": 2.1641791044776117, + "grad_norm": 2.0594828128814697, + "learning_rate": 1e-06, + "loss": 0.5549, + "mean_token_accuracy": 0.8097561597824097, + "num_tokens": 678860557.0, + "step": 26245 + }, + { + "epoch": 2.1642615651026635, + "grad_norm": 2.1158759593963623, + "learning_rate": 1e-06, + "loss": 0.6001, + "mean_token_accuracy": 0.8082080483436584, + "num_tokens": 678890204.0, + "step": 26246 + }, + { + "epoch": 2.164344025727715, + "grad_norm": 2.106168270111084, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7812366485595703, + "num_tokens": 678917805.0, + "step": 26247 + }, + { + "epoch": 2.1644264863527667, + "grad_norm": 2.2296054363250732, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.793756365776062, + "num_tokens": 678946153.0, + "step": 26248 + }, + { + "epoch": 2.164508946977818, + "grad_norm": 1.9942708015441895, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7630774974822998, + "num_tokens": 678977639.0, + "step": 26249 + }, + { + "epoch": 2.16459140760287, + "grad_norm": 2.218139171600342, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7597199082374573, + "num_tokens": 679004798.0, + "step": 26250 + }, + { + "epoch": 2.164673868227921, + "grad_norm": 2.467496871948242, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7266793847084045, + "num_tokens": 679028943.0, + "step": 26251 + }, + { + "epoch": 2.1647563288529725, + "grad_norm": 2.0308992862701416, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7319488525390625, + "num_tokens": 679062744.0, + "step": 26252 + }, + { + "epoch": 2.1648387894780243, + "grad_norm": 2.0158843994140625, + "learning_rate": 1e-06, + "loss": 0.5765, + "mean_token_accuracy": 0.8009840250015259, + "num_tokens": 679094946.0, + "step": 26253 + }, + { + "epoch": 2.1649212501030757, + "grad_norm": 2.178758144378662, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7554962038993835, + "num_tokens": 679124579.0, + "step": 26254 + }, + { + "epoch": 2.1650037107281275, + "grad_norm": 2.1164255142211914, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.7488021850585938, + "num_tokens": 679156921.0, + "step": 26255 + }, + { + "epoch": 2.165086171353179, + "grad_norm": 2.17073655128479, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.773514986038208, + "num_tokens": 679185954.0, + "step": 26256 + }, + { + "epoch": 2.16516863197823, + "grad_norm": 2.4965498447418213, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7531583309173584, + "num_tokens": 679211083.0, + "step": 26257 + }, + { + "epoch": 2.165251092603282, + "grad_norm": 2.072814464569092, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7745897769927979, + "num_tokens": 679243773.0, + "step": 26258 + }, + { + "epoch": 2.1653335532283333, + "grad_norm": 2.0967466831207275, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7586029767990112, + "num_tokens": 679272305.0, + "step": 26259 + }, + { + "epoch": 2.165416013853385, + "grad_norm": 2.2291619777679443, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7775969505310059, + "num_tokens": 679296948.0, + "step": 26260 + }, + { + "epoch": 2.1654984744784365, + "grad_norm": 2.2385642528533936, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7618242502212524, + "num_tokens": 679327393.0, + "step": 26261 + }, + { + "epoch": 2.1655809351034883, + "grad_norm": 2.3680222034454346, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7723185420036316, + "num_tokens": 679352941.0, + "step": 26262 + }, + { + "epoch": 2.1656633957285396, + "grad_norm": 2.340613603591919, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.7935349345207214, + "num_tokens": 679378236.0, + "step": 26263 + }, + { + "epoch": 2.165745856353591, + "grad_norm": 2.2807531356811523, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7614165544509888, + "num_tokens": 679405557.0, + "step": 26264 + }, + { + "epoch": 2.1658283169786428, + "grad_norm": 1.871455192565918, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7550105452537537, + "num_tokens": 679440196.0, + "step": 26265 + }, + { + "epoch": 2.165910777603694, + "grad_norm": 2.5042331218719482, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7530918121337891, + "num_tokens": 679462336.0, + "step": 26266 + }, + { + "epoch": 2.165993238228746, + "grad_norm": 2.0378305912017822, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7718746662139893, + "num_tokens": 679495756.0, + "step": 26267 + }, + { + "epoch": 2.1660756988537972, + "grad_norm": 2.0875918865203857, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7511272430419922, + "num_tokens": 679527059.0, + "step": 26268 + }, + { + "epoch": 2.166158159478849, + "grad_norm": 2.023616075515747, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7767061591148376, + "num_tokens": 679561472.0, + "step": 26269 + }, + { + "epoch": 2.1662406201039004, + "grad_norm": 2.073397159576416, + "learning_rate": 1e-06, + "loss": 0.7531, + "mean_token_accuracy": 0.7527838349342346, + "num_tokens": 679592947.0, + "step": 26270 + }, + { + "epoch": 2.1663230807289517, + "grad_norm": 1.9220657348632812, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7656304240226746, + "num_tokens": 679631443.0, + "step": 26271 + }, + { + "epoch": 2.1664055413540035, + "grad_norm": 1.9872593879699707, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.8027153015136719, + "num_tokens": 679664223.0, + "step": 26272 + }, + { + "epoch": 2.166488001979055, + "grad_norm": 1.9509469270706177, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7879921197891235, + "num_tokens": 679697026.0, + "step": 26273 + }, + { + "epoch": 2.1665704626041067, + "grad_norm": 2.22658109664917, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7476811408996582, + "num_tokens": 679724600.0, + "step": 26274 + }, + { + "epoch": 2.166652923229158, + "grad_norm": 1.8511130809783936, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7751864194869995, + "num_tokens": 679757831.0, + "step": 26275 + }, + { + "epoch": 2.16673538385421, + "grad_norm": 1.9829516410827637, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.7850909233093262, + "num_tokens": 679787533.0, + "step": 26276 + }, + { + "epoch": 2.166817844479261, + "grad_norm": 2.2679500579833984, + "learning_rate": 1e-06, + "loss": 0.5317, + "mean_token_accuracy": 0.8103301525115967, + "num_tokens": 679813105.0, + "step": 26277 + }, + { + "epoch": 2.1669003051043125, + "grad_norm": 1.936974287033081, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.770611047744751, + "num_tokens": 679844718.0, + "step": 26278 + }, + { + "epoch": 2.1669827657293643, + "grad_norm": 2.4960453510284424, + "learning_rate": 1e-06, + "loss": 0.7452, + "mean_token_accuracy": 0.7581436634063721, + "num_tokens": 679868782.0, + "step": 26279 + }, + { + "epoch": 2.1670652263544157, + "grad_norm": 2.0771095752716064, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7676200866699219, + "num_tokens": 679899775.0, + "step": 26280 + }, + { + "epoch": 2.1671476869794675, + "grad_norm": 1.9974615573883057, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7825295925140381, + "num_tokens": 679930823.0, + "step": 26281 + }, + { + "epoch": 2.167230147604519, + "grad_norm": 2.2727138996124268, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.750556468963623, + "num_tokens": 679958112.0, + "step": 26282 + }, + { + "epoch": 2.1673126082295706, + "grad_norm": 2.3147056102752686, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7825031280517578, + "num_tokens": 679983191.0, + "step": 26283 + }, + { + "epoch": 2.167395068854622, + "grad_norm": 1.8298838138580322, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7735663056373596, + "num_tokens": 680021990.0, + "step": 26284 + }, + { + "epoch": 2.1674775294796733, + "grad_norm": 2.04840087890625, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7881987690925598, + "num_tokens": 680050997.0, + "step": 26285 + }, + { + "epoch": 2.167559990104725, + "grad_norm": 2.143066883087158, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7596362829208374, + "num_tokens": 680080649.0, + "step": 26286 + }, + { + "epoch": 2.1676424507297765, + "grad_norm": 1.994512677192688, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7481780052185059, + "num_tokens": 680116835.0, + "step": 26287 + }, + { + "epoch": 2.1677249113548283, + "grad_norm": 2.230663299560547, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7652567028999329, + "num_tokens": 680143740.0, + "step": 26288 + }, + { + "epoch": 2.1678073719798796, + "grad_norm": 2.1573047637939453, + "learning_rate": 1e-06, + "loss": 0.5898, + "mean_token_accuracy": 0.7916525602340698, + "num_tokens": 680174403.0, + "step": 26289 + }, + { + "epoch": 2.167889832604931, + "grad_norm": 2.3407273292541504, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7604894042015076, + "num_tokens": 680199158.0, + "step": 26290 + }, + { + "epoch": 2.1679722932299827, + "grad_norm": 1.8850435018539429, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7675362825393677, + "num_tokens": 680235003.0, + "step": 26291 + }, + { + "epoch": 2.168054753855034, + "grad_norm": 1.9335790872573853, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7771780490875244, + "num_tokens": 680268982.0, + "step": 26292 + }, + { + "epoch": 2.168137214480086, + "grad_norm": 2.1065292358398438, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.76045823097229, + "num_tokens": 680299802.0, + "step": 26293 + }, + { + "epoch": 2.1682196751051372, + "grad_norm": 2.212897539138794, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7735955715179443, + "num_tokens": 680326853.0, + "step": 26294 + }, + { + "epoch": 2.168302135730189, + "grad_norm": 1.9133777618408203, + "learning_rate": 1e-06, + "loss": 0.6233, + "mean_token_accuracy": 0.7887924909591675, + "num_tokens": 680357496.0, + "step": 26295 + }, + { + "epoch": 2.1683845963552404, + "grad_norm": 1.9011950492858887, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.7829214930534363, + "num_tokens": 680392375.0, + "step": 26296 + }, + { + "epoch": 2.1684670569802917, + "grad_norm": 2.1318485736846924, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7644479274749756, + "num_tokens": 680423216.0, + "step": 26297 + }, + { + "epoch": 2.1685495176053435, + "grad_norm": 2.2417829036712646, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.7876546382904053, + "num_tokens": 680451008.0, + "step": 26298 + }, + { + "epoch": 2.168631978230395, + "grad_norm": 2.176426649093628, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.7509983777999878, + "num_tokens": 680479503.0, + "step": 26299 + }, + { + "epoch": 2.1687144388554467, + "grad_norm": 1.9746215343475342, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.750743567943573, + "num_tokens": 680511512.0, + "step": 26300 + }, + { + "epoch": 2.168796899480498, + "grad_norm": 2.1530861854553223, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7845364212989807, + "num_tokens": 680540562.0, + "step": 26301 + }, + { + "epoch": 2.1688793601055494, + "grad_norm": 2.0448601245880127, + "learning_rate": 1e-06, + "loss": 0.4663, + "mean_token_accuracy": 0.8377302289009094, + "num_tokens": 680569537.0, + "step": 26302 + }, + { + "epoch": 2.168961820730601, + "grad_norm": 2.199404001235962, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.7901175022125244, + "num_tokens": 680597897.0, + "step": 26303 + }, + { + "epoch": 2.1690442813556525, + "grad_norm": 2.0514369010925293, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7556113004684448, + "num_tokens": 680631872.0, + "step": 26304 + }, + { + "epoch": 2.1691267419807043, + "grad_norm": 2.1082987785339355, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.737544059753418, + "num_tokens": 680663256.0, + "step": 26305 + }, + { + "epoch": 2.1692092026057557, + "grad_norm": 1.737657070159912, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.7912473678588867, + "num_tokens": 680700582.0, + "step": 26306 + }, + { + "epoch": 2.1692916632308075, + "grad_norm": 1.9751605987548828, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.7800009846687317, + "num_tokens": 680732602.0, + "step": 26307 + }, + { + "epoch": 2.169374123855859, + "grad_norm": 2.1013283729553223, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.77721107006073, + "num_tokens": 680760350.0, + "step": 26308 + }, + { + "epoch": 2.16945658448091, + "grad_norm": 2.0837111473083496, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.767789363861084, + "num_tokens": 680791611.0, + "step": 26309 + }, + { + "epoch": 2.169539045105962, + "grad_norm": 2.110722541809082, + "learning_rate": 1e-06, + "loss": 0.7712, + "mean_token_accuracy": 0.7536963820457458, + "num_tokens": 680820669.0, + "step": 26310 + }, + { + "epoch": 2.1696215057310133, + "grad_norm": 2.200313091278076, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7755460739135742, + "num_tokens": 680847053.0, + "step": 26311 + }, + { + "epoch": 2.169703966356065, + "grad_norm": 2.247345209121704, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.7914925813674927, + "num_tokens": 680873617.0, + "step": 26312 + }, + { + "epoch": 2.1697864269811165, + "grad_norm": 2.3870913982391357, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7576196193695068, + "num_tokens": 680900824.0, + "step": 26313 + }, + { + "epoch": 2.1698688876061683, + "grad_norm": 2.0412797927856445, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.782268762588501, + "num_tokens": 680934269.0, + "step": 26314 + }, + { + "epoch": 2.1699513482312196, + "grad_norm": 2.1743600368499756, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.7891396284103394, + "num_tokens": 680963342.0, + "step": 26315 + }, + { + "epoch": 2.170033808856271, + "grad_norm": 2.353100299835205, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7575366497039795, + "num_tokens": 680990302.0, + "step": 26316 + }, + { + "epoch": 2.1701162694813227, + "grad_norm": 2.274850606918335, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7643406391143799, + "num_tokens": 681014885.0, + "step": 26317 + }, + { + "epoch": 2.170198730106374, + "grad_norm": 2.120189905166626, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7770081162452698, + "num_tokens": 681044660.0, + "step": 26318 + }, + { + "epoch": 2.170281190731426, + "grad_norm": 1.942782998085022, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7765451669692993, + "num_tokens": 681077837.0, + "step": 26319 + }, + { + "epoch": 2.1703636513564772, + "grad_norm": 2.253173828125, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7632734775543213, + "num_tokens": 681105863.0, + "step": 26320 + }, + { + "epoch": 2.170446111981529, + "grad_norm": 1.9490249156951904, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7821370363235474, + "num_tokens": 681141837.0, + "step": 26321 + }, + { + "epoch": 2.1705285726065804, + "grad_norm": 2.187856435775757, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7550567984580994, + "num_tokens": 681173763.0, + "step": 26322 + }, + { + "epoch": 2.1706110332316317, + "grad_norm": 2.2211239337921143, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.7738958597183228, + "num_tokens": 681199705.0, + "step": 26323 + }, + { + "epoch": 2.1706934938566835, + "grad_norm": 1.9442282915115356, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.7969787120819092, + "num_tokens": 681234217.0, + "step": 26324 + }, + { + "epoch": 2.170775954481735, + "grad_norm": 2.1137313842773438, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7708336710929871, + "num_tokens": 681263532.0, + "step": 26325 + }, + { + "epoch": 2.1708584151067867, + "grad_norm": 2.4864280223846436, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7781648635864258, + "num_tokens": 681286640.0, + "step": 26326 + }, + { + "epoch": 2.170940875731838, + "grad_norm": 2.351300001144409, + "learning_rate": 1e-06, + "loss": 0.5936, + "mean_token_accuracy": 0.7954155206680298, + "num_tokens": 681309430.0, + "step": 26327 + }, + { + "epoch": 2.1710233363568894, + "grad_norm": 2.023801803588867, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7338455319404602, + "num_tokens": 681347913.0, + "step": 26328 + }, + { + "epoch": 2.171105796981941, + "grad_norm": 2.1401526927948, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.7792308926582336, + "num_tokens": 681376103.0, + "step": 26329 + }, + { + "epoch": 2.1711882576069925, + "grad_norm": 2.104410171508789, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7663331031799316, + "num_tokens": 681406170.0, + "step": 26330 + }, + { + "epoch": 2.1712707182320443, + "grad_norm": 2.204582929611206, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.7889194488525391, + "num_tokens": 681436085.0, + "step": 26331 + }, + { + "epoch": 2.1713531788570957, + "grad_norm": 2.1264588832855225, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.7769308686256409, + "num_tokens": 681465579.0, + "step": 26332 + }, + { + "epoch": 2.1714356394821475, + "grad_norm": 2.149036407470703, + "learning_rate": 1e-06, + "loss": 0.5082, + "mean_token_accuracy": 0.8136448860168457, + "num_tokens": 681490027.0, + "step": 26333 + }, + { + "epoch": 2.171518100107199, + "grad_norm": 1.9712587594985962, + "learning_rate": 1e-06, + "loss": 0.6479, + "mean_token_accuracy": 0.787702202796936, + "num_tokens": 681521004.0, + "step": 26334 + }, + { + "epoch": 2.17160056073225, + "grad_norm": 2.1380808353424072, + "learning_rate": 1e-06, + "loss": 0.5474, + "mean_token_accuracy": 0.8126354217529297, + "num_tokens": 681549425.0, + "step": 26335 + }, + { + "epoch": 2.171683021357302, + "grad_norm": 2.197421073913574, + "learning_rate": 1e-06, + "loss": 0.6099, + "mean_token_accuracy": 0.7894889116287231, + "num_tokens": 681576140.0, + "step": 26336 + }, + { + "epoch": 2.1717654819823533, + "grad_norm": 1.9243829250335693, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7679070234298706, + "num_tokens": 681612964.0, + "step": 26337 + }, + { + "epoch": 2.171847942607405, + "grad_norm": 2.2746238708496094, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7837206125259399, + "num_tokens": 681637827.0, + "step": 26338 + }, + { + "epoch": 2.1719304032324565, + "grad_norm": 1.9883185625076294, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.7913066148757935, + "num_tokens": 681671667.0, + "step": 26339 + }, + { + "epoch": 2.1720128638575082, + "grad_norm": 1.8030872344970703, + "learning_rate": 1e-06, + "loss": 0.5874, + "mean_token_accuracy": 0.7882145643234253, + "num_tokens": 681705128.0, + "step": 26340 + }, + { + "epoch": 2.1720953244825596, + "grad_norm": 1.9371534585952759, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.7866804599761963, + "num_tokens": 681735707.0, + "step": 26341 + }, + { + "epoch": 2.172177785107611, + "grad_norm": 2.308117151260376, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7468005418777466, + "num_tokens": 681763834.0, + "step": 26342 + }, + { + "epoch": 2.1722602457326627, + "grad_norm": 1.9572796821594238, + "learning_rate": 1e-06, + "loss": 0.7762, + "mean_token_accuracy": 0.751760721206665, + "num_tokens": 681797988.0, + "step": 26343 + }, + { + "epoch": 2.172342706357714, + "grad_norm": 2.3564975261688232, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7604800462722778, + "num_tokens": 681823523.0, + "step": 26344 + }, + { + "epoch": 2.172425166982766, + "grad_norm": 2.3253695964813232, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7384216785430908, + "num_tokens": 681853715.0, + "step": 26345 + }, + { + "epoch": 2.1725076276078172, + "grad_norm": 2.498424530029297, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7708257436752319, + "num_tokens": 681875164.0, + "step": 26346 + }, + { + "epoch": 2.1725900882328686, + "grad_norm": 1.9864753484725952, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.8017764091491699, + "num_tokens": 681902685.0, + "step": 26347 + }, + { + "epoch": 2.1726725488579204, + "grad_norm": 2.0178184509277344, + "learning_rate": 1e-06, + "loss": 0.6079, + "mean_token_accuracy": 0.7880083322525024, + "num_tokens": 681933073.0, + "step": 26348 + }, + { + "epoch": 2.1727550094829717, + "grad_norm": 1.9050651788711548, + "learning_rate": 1e-06, + "loss": 0.563, + "mean_token_accuracy": 0.8066292405128479, + "num_tokens": 681967325.0, + "step": 26349 + }, + { + "epoch": 2.1728374701080235, + "grad_norm": 2.146967649459839, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7752280235290527, + "num_tokens": 681995996.0, + "step": 26350 + }, + { + "epoch": 2.172919930733075, + "grad_norm": 2.395270824432373, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7645207643508911, + "num_tokens": 682023180.0, + "step": 26351 + }, + { + "epoch": 2.1730023913581267, + "grad_norm": 1.9149047136306763, + "learning_rate": 1e-06, + "loss": 0.5796, + "mean_token_accuracy": 0.8064686059951782, + "num_tokens": 682058237.0, + "step": 26352 + }, + { + "epoch": 2.173084851983178, + "grad_norm": 2.1582963466644287, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.7854663133621216, + "num_tokens": 682089152.0, + "step": 26353 + }, + { + "epoch": 2.1731673126082294, + "grad_norm": 2.1207001209259033, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7433645725250244, + "num_tokens": 682120261.0, + "step": 26354 + }, + { + "epoch": 2.173249773233281, + "grad_norm": 1.9526437520980835, + "learning_rate": 1e-06, + "loss": 0.5853, + "mean_token_accuracy": 0.806126058101654, + "num_tokens": 682152690.0, + "step": 26355 + }, + { + "epoch": 2.1733322338583325, + "grad_norm": 1.9341301918029785, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7909519672393799, + "num_tokens": 682186804.0, + "step": 26356 + }, + { + "epoch": 2.1734146944833843, + "grad_norm": 2.3674724102020264, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7417440414428711, + "num_tokens": 682214459.0, + "step": 26357 + }, + { + "epoch": 2.1734971551084357, + "grad_norm": 1.8806098699569702, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7639602422714233, + "num_tokens": 682249454.0, + "step": 26358 + }, + { + "epoch": 2.1735796157334875, + "grad_norm": 2.4744021892547607, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.773633599281311, + "num_tokens": 682274530.0, + "step": 26359 + }, + { + "epoch": 2.173662076358539, + "grad_norm": 2.287367820739746, + "learning_rate": 1e-06, + "loss": 0.5481, + "mean_token_accuracy": 0.8115983009338379, + "num_tokens": 682297976.0, + "step": 26360 + }, + { + "epoch": 2.17374453698359, + "grad_norm": 2.1202542781829834, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7710752487182617, + "num_tokens": 682326360.0, + "step": 26361 + }, + { + "epoch": 2.173826997608642, + "grad_norm": 2.0971319675445557, + "learning_rate": 1e-06, + "loss": 0.5538, + "mean_token_accuracy": 0.8047117590904236, + "num_tokens": 682354055.0, + "step": 26362 + }, + { + "epoch": 2.1739094582336933, + "grad_norm": 2.0560250282287598, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7630134224891663, + "num_tokens": 682388139.0, + "step": 26363 + }, + { + "epoch": 2.173991918858745, + "grad_norm": 1.80964195728302, + "learning_rate": 1e-06, + "loss": 0.5892, + "mean_token_accuracy": 0.795316219329834, + "num_tokens": 682423498.0, + "step": 26364 + }, + { + "epoch": 2.1740743794837964, + "grad_norm": 2.033708095550537, + "learning_rate": 1e-06, + "loss": 0.6938, + "mean_token_accuracy": 0.7780032157897949, + "num_tokens": 682453097.0, + "step": 26365 + }, + { + "epoch": 2.1741568401088482, + "grad_norm": 2.1921112537384033, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7117273807525635, + "num_tokens": 682483703.0, + "step": 26366 + }, + { + "epoch": 2.1742393007338996, + "grad_norm": 1.9545025825500488, + "learning_rate": 1e-06, + "loss": 0.4678, + "mean_token_accuracy": 0.8407624959945679, + "num_tokens": 682514332.0, + "step": 26367 + }, + { + "epoch": 2.174321761358951, + "grad_norm": 2.2458336353302, + "learning_rate": 1e-06, + "loss": 0.6034, + "mean_token_accuracy": 0.7966151237487793, + "num_tokens": 682539558.0, + "step": 26368 + }, + { + "epoch": 2.1744042219840027, + "grad_norm": 1.9904917478561401, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.7823059558868408, + "num_tokens": 682570562.0, + "step": 26369 + }, + { + "epoch": 2.174486682609054, + "grad_norm": 2.411196231842041, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7489114999771118, + "num_tokens": 682594954.0, + "step": 26370 + }, + { + "epoch": 2.174569143234106, + "grad_norm": 1.9534434080123901, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7315804958343506, + "num_tokens": 682630736.0, + "step": 26371 + }, + { + "epoch": 2.1746516038591572, + "grad_norm": 2.1000211238861084, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7821521162986755, + "num_tokens": 682663921.0, + "step": 26372 + }, + { + "epoch": 2.1747340644842086, + "grad_norm": 2.084317207336426, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7694059610366821, + "num_tokens": 682691578.0, + "step": 26373 + }, + { + "epoch": 2.1748165251092604, + "grad_norm": 2.0204737186431885, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7966970205307007, + "num_tokens": 682721775.0, + "step": 26374 + }, + { + "epoch": 2.1748989857343117, + "grad_norm": 2.0946903228759766, + "learning_rate": 1e-06, + "loss": 0.5515, + "mean_token_accuracy": 0.8098850250244141, + "num_tokens": 682750746.0, + "step": 26375 + }, + { + "epoch": 2.1749814463593635, + "grad_norm": 1.8412609100341797, + "learning_rate": 1e-06, + "loss": 0.4964, + "mean_token_accuracy": 0.8309625387191772, + "num_tokens": 682787043.0, + "step": 26376 + }, + { + "epoch": 2.175063906984415, + "grad_norm": 2.2562685012817383, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.7891848683357239, + "num_tokens": 682813663.0, + "step": 26377 + }, + { + "epoch": 2.1751463676094667, + "grad_norm": 1.8552327156066895, + "learning_rate": 1e-06, + "loss": 0.4802, + "mean_token_accuracy": 0.8387826681137085, + "num_tokens": 682849940.0, + "step": 26378 + }, + { + "epoch": 2.175228828234518, + "grad_norm": 2.394803047180176, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7590866088867188, + "num_tokens": 682872576.0, + "step": 26379 + }, + { + "epoch": 2.1753112888595694, + "grad_norm": 2.199307680130005, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.775800883769989, + "num_tokens": 682900427.0, + "step": 26380 + }, + { + "epoch": 2.175393749484621, + "grad_norm": 2.3458750247955322, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.744813084602356, + "num_tokens": 682926693.0, + "step": 26381 + }, + { + "epoch": 2.1754762101096725, + "grad_norm": 1.8598891496658325, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7910746932029724, + "num_tokens": 682962944.0, + "step": 26382 + }, + { + "epoch": 2.1755586707347243, + "grad_norm": 2.028475522994995, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7731814980506897, + "num_tokens": 682992979.0, + "step": 26383 + }, + { + "epoch": 2.1756411313597757, + "grad_norm": 2.03450345993042, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7759879231452942, + "num_tokens": 683023392.0, + "step": 26384 + }, + { + "epoch": 2.175723591984827, + "grad_norm": 2.117283344268799, + "learning_rate": 1e-06, + "loss": 0.5683, + "mean_token_accuracy": 0.8091244697570801, + "num_tokens": 683049918.0, + "step": 26385 + }, + { + "epoch": 2.175806052609879, + "grad_norm": 2.2029621601104736, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7715346813201904, + "num_tokens": 683079332.0, + "step": 26386 + }, + { + "epoch": 2.17588851323493, + "grad_norm": 1.9898126125335693, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7636159658432007, + "num_tokens": 683111211.0, + "step": 26387 + }, + { + "epoch": 2.175970973859982, + "grad_norm": 2.2207279205322266, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.789050281047821, + "num_tokens": 683138926.0, + "step": 26388 + }, + { + "epoch": 2.1760534344850333, + "grad_norm": 2.078758478164673, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7711302638053894, + "num_tokens": 683172575.0, + "step": 26389 + }, + { + "epoch": 2.176135895110085, + "grad_norm": 2.110957145690918, + "learning_rate": 1e-06, + "loss": 0.5812, + "mean_token_accuracy": 0.8109569549560547, + "num_tokens": 683200935.0, + "step": 26390 + }, + { + "epoch": 2.1762183557351364, + "grad_norm": 2.0041167736053467, + "learning_rate": 1e-06, + "loss": 0.6203, + "mean_token_accuracy": 0.7856155037879944, + "num_tokens": 683232531.0, + "step": 26391 + }, + { + "epoch": 2.176300816360188, + "grad_norm": 1.9085097312927246, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7739395499229431, + "num_tokens": 683265199.0, + "step": 26392 + }, + { + "epoch": 2.1763832769852396, + "grad_norm": 2.3463990688323975, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7871435880661011, + "num_tokens": 683292534.0, + "step": 26393 + }, + { + "epoch": 2.176465737610291, + "grad_norm": 1.9585148096084595, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.7919454574584961, + "num_tokens": 683327861.0, + "step": 26394 + }, + { + "epoch": 2.1765481982353427, + "grad_norm": 1.922385573387146, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7547672986984253, + "num_tokens": 683362960.0, + "step": 26395 + }, + { + "epoch": 2.176630658860394, + "grad_norm": 2.1071126461029053, + "learning_rate": 1e-06, + "loss": 0.5383, + "mean_token_accuracy": 0.8071424961090088, + "num_tokens": 683392157.0, + "step": 26396 + }, + { + "epoch": 2.176713119485446, + "grad_norm": 2.023116111755371, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7722001671791077, + "num_tokens": 683426269.0, + "step": 26397 + }, + { + "epoch": 2.1767955801104972, + "grad_norm": 2.372936487197876, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7738823294639587, + "num_tokens": 683450304.0, + "step": 26398 + }, + { + "epoch": 2.1768780407355486, + "grad_norm": 2.113140106201172, + "learning_rate": 1e-06, + "loss": 0.5738, + "mean_token_accuracy": 0.8021596670150757, + "num_tokens": 683478792.0, + "step": 26399 + }, + { + "epoch": 2.1769605013606004, + "grad_norm": 2.0427353382110596, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7655148506164551, + "num_tokens": 683510233.0, + "step": 26400 + }, + { + "epoch": 2.1770429619856517, + "grad_norm": 1.8657078742980957, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.7789272665977478, + "num_tokens": 683547969.0, + "step": 26401 + }, + { + "epoch": 2.1771254226107035, + "grad_norm": 2.0280802249908447, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7555663585662842, + "num_tokens": 683578879.0, + "step": 26402 + }, + { + "epoch": 2.177207883235755, + "grad_norm": 1.9330432415008545, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7818461656570435, + "num_tokens": 683613115.0, + "step": 26403 + }, + { + "epoch": 2.1772903438608067, + "grad_norm": 2.074913501739502, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.7526452541351318, + "num_tokens": 683644197.0, + "step": 26404 + }, + { + "epoch": 2.177372804485858, + "grad_norm": 2.044708013534546, + "learning_rate": 1e-06, + "loss": 0.5801, + "mean_token_accuracy": 0.801144003868103, + "num_tokens": 683676976.0, + "step": 26405 + }, + { + "epoch": 2.1774552651109094, + "grad_norm": 2.0919017791748047, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7530304193496704, + "num_tokens": 683707272.0, + "step": 26406 + }, + { + "epoch": 2.177537725735961, + "grad_norm": 2.01399302482605, + "learning_rate": 1e-06, + "loss": 0.5612, + "mean_token_accuracy": 0.818793535232544, + "num_tokens": 683736464.0, + "step": 26407 + }, + { + "epoch": 2.1776201863610125, + "grad_norm": 2.099517822265625, + "learning_rate": 1e-06, + "loss": 0.5336, + "mean_token_accuracy": 0.8239414691925049, + "num_tokens": 683769204.0, + "step": 26408 + }, + { + "epoch": 2.1777026469860643, + "grad_norm": 2.4914259910583496, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7330938577651978, + "num_tokens": 683792886.0, + "step": 26409 + }, + { + "epoch": 2.1777851076111157, + "grad_norm": 2.3062071800231934, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7793234586715698, + "num_tokens": 683819392.0, + "step": 26410 + }, + { + "epoch": 2.1778675682361675, + "grad_norm": 2.200355291366577, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.7894923686981201, + "num_tokens": 683847823.0, + "step": 26411 + }, + { + "epoch": 2.177950028861219, + "grad_norm": 2.190434694290161, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.7838982343673706, + "num_tokens": 683874582.0, + "step": 26412 + }, + { + "epoch": 2.17803248948627, + "grad_norm": 2.0818419456481934, + "learning_rate": 1e-06, + "loss": 0.5676, + "mean_token_accuracy": 0.8159912824630737, + "num_tokens": 683903173.0, + "step": 26413 + }, + { + "epoch": 2.178114950111322, + "grad_norm": 2.060939311981201, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.7840610146522522, + "num_tokens": 683934837.0, + "step": 26414 + }, + { + "epoch": 2.1781974107363733, + "grad_norm": 2.1510233879089355, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.749608039855957, + "num_tokens": 683966714.0, + "step": 26415 + }, + { + "epoch": 2.178279871361425, + "grad_norm": 2.41217303276062, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7201874852180481, + "num_tokens": 683991846.0, + "step": 26416 + }, + { + "epoch": 2.1783623319864764, + "grad_norm": 2.268047571182251, + "learning_rate": 1e-06, + "loss": 0.5557, + "mean_token_accuracy": 0.7997564077377319, + "num_tokens": 684016725.0, + "step": 26417 + }, + { + "epoch": 2.178444792611528, + "grad_norm": 2.1980316638946533, + "learning_rate": 1e-06, + "loss": 0.6068, + "mean_token_accuracy": 0.7998514771461487, + "num_tokens": 684043286.0, + "step": 26418 + }, + { + "epoch": 2.1785272532365796, + "grad_norm": 2.433013439178467, + "learning_rate": 1e-06, + "loss": 0.5872, + "mean_token_accuracy": 0.8042342662811279, + "num_tokens": 684065366.0, + "step": 26419 + }, + { + "epoch": 2.178609713861631, + "grad_norm": 2.204150915145874, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7763659954071045, + "num_tokens": 684091745.0, + "step": 26420 + }, + { + "epoch": 2.1786921744866827, + "grad_norm": 2.2201616764068604, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7628136277198792, + "num_tokens": 684120998.0, + "step": 26421 + }, + { + "epoch": 2.178774635111734, + "grad_norm": 2.371093511581421, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.730651319026947, + "num_tokens": 684147259.0, + "step": 26422 + }, + { + "epoch": 2.178857095736786, + "grad_norm": 2.0340747833251953, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7638202905654907, + "num_tokens": 684180958.0, + "step": 26423 + }, + { + "epoch": 2.1789395563618372, + "grad_norm": 2.064891815185547, + "learning_rate": 1e-06, + "loss": 0.5243, + "mean_token_accuracy": 0.8246064186096191, + "num_tokens": 684209965.0, + "step": 26424 + }, + { + "epoch": 2.1790220169868886, + "grad_norm": 2.0184226036071777, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.7890700101852417, + "num_tokens": 684243219.0, + "step": 26425 + }, + { + "epoch": 2.1791044776119404, + "grad_norm": 2.0350587368011475, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7813215255737305, + "num_tokens": 684275546.0, + "step": 26426 + }, + { + "epoch": 2.1791869382369917, + "grad_norm": 2.174955368041992, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7666541337966919, + "num_tokens": 684304064.0, + "step": 26427 + }, + { + "epoch": 2.1792693988620435, + "grad_norm": 1.9313576221466064, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7350714802742004, + "num_tokens": 684341786.0, + "step": 26428 + }, + { + "epoch": 2.179351859487095, + "grad_norm": 2.1974987983703613, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7707980275154114, + "num_tokens": 684369365.0, + "step": 26429 + }, + { + "epoch": 2.179434320112146, + "grad_norm": 2.3731727600097656, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7403031587600708, + "num_tokens": 684395443.0, + "step": 26430 + }, + { + "epoch": 2.179516780737198, + "grad_norm": 2.4410250186920166, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7331365346908569, + "num_tokens": 684421198.0, + "step": 26431 + }, + { + "epoch": 2.1795992413622494, + "grad_norm": 1.764866590499878, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7779985666275024, + "num_tokens": 684460646.0, + "step": 26432 + }, + { + "epoch": 2.179681701987301, + "grad_norm": 2.180497646331787, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7582221031188965, + "num_tokens": 684488260.0, + "step": 26433 + }, + { + "epoch": 2.1797641626123525, + "grad_norm": 1.903386116027832, + "learning_rate": 1e-06, + "loss": 0.6131, + "mean_token_accuracy": 0.7898244857788086, + "num_tokens": 684520478.0, + "step": 26434 + }, + { + "epoch": 2.1798466232374043, + "grad_norm": 2.1932787895202637, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.7569971084594727, + "num_tokens": 684551679.0, + "step": 26435 + }, + { + "epoch": 2.1799290838624557, + "grad_norm": 1.8623666763305664, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7854987382888794, + "num_tokens": 684585791.0, + "step": 26436 + }, + { + "epoch": 2.180011544487507, + "grad_norm": 2.0434625148773193, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7455784678459167, + "num_tokens": 684617681.0, + "step": 26437 + }, + { + "epoch": 2.180094005112559, + "grad_norm": 2.053145170211792, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7838771939277649, + "num_tokens": 684648779.0, + "step": 26438 + }, + { + "epoch": 2.18017646573761, + "grad_norm": 2.208458423614502, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.7830434441566467, + "num_tokens": 684674877.0, + "step": 26439 + }, + { + "epoch": 2.180258926362662, + "grad_norm": 2.323749303817749, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7682802677154541, + "num_tokens": 684702561.0, + "step": 26440 + }, + { + "epoch": 2.1803413869877133, + "grad_norm": 2.44184947013855, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.7894647717475891, + "num_tokens": 684724872.0, + "step": 26441 + }, + { + "epoch": 2.180423847612765, + "grad_norm": 2.168654680252075, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7669359445571899, + "num_tokens": 684753184.0, + "step": 26442 + }, + { + "epoch": 2.1805063082378164, + "grad_norm": 1.8871139287948608, + "learning_rate": 1e-06, + "loss": 0.5716, + "mean_token_accuracy": 0.7978296279907227, + "num_tokens": 684787803.0, + "step": 26443 + }, + { + "epoch": 2.180588768862868, + "grad_norm": 2.247713088989258, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7841921448707581, + "num_tokens": 684815502.0, + "step": 26444 + }, + { + "epoch": 2.1806712294879196, + "grad_norm": 2.3781604766845703, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.7566766142845154, + "num_tokens": 684840827.0, + "step": 26445 + }, + { + "epoch": 2.180753690112971, + "grad_norm": 2.071702480316162, + "learning_rate": 1e-06, + "loss": 0.7842, + "mean_token_accuracy": 0.7438279390335083, + "num_tokens": 684871921.0, + "step": 26446 + }, + { + "epoch": 2.1808361507380227, + "grad_norm": 2.0546317100524902, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7886527180671692, + "num_tokens": 684902947.0, + "step": 26447 + }, + { + "epoch": 2.180918611363074, + "grad_norm": 2.1873865127563477, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7731928825378418, + "num_tokens": 684929839.0, + "step": 26448 + }, + { + "epoch": 2.181001071988126, + "grad_norm": 2.3144590854644775, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7607986927032471, + "num_tokens": 684953693.0, + "step": 26449 + }, + { + "epoch": 2.1810835326131772, + "grad_norm": 2.069922685623169, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7734220623970032, + "num_tokens": 684985085.0, + "step": 26450 + }, + { + "epoch": 2.1811659932382286, + "grad_norm": 2.3120687007904053, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7698241472244263, + "num_tokens": 685009397.0, + "step": 26451 + }, + { + "epoch": 2.1812484538632804, + "grad_norm": 1.8624335527420044, + "learning_rate": 1e-06, + "loss": 0.784, + "mean_token_accuracy": 0.7538772821426392, + "num_tokens": 685047912.0, + "step": 26452 + }, + { + "epoch": 2.1813309144883317, + "grad_norm": 2.2740228176116943, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7650227546691895, + "num_tokens": 685072438.0, + "step": 26453 + }, + { + "epoch": 2.1814133751133835, + "grad_norm": 2.167513847351074, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.7810550928115845, + "num_tokens": 685104962.0, + "step": 26454 + }, + { + "epoch": 2.181495835738435, + "grad_norm": 1.8822026252746582, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.8007662892341614, + "num_tokens": 685142483.0, + "step": 26455 + }, + { + "epoch": 2.1815782963634867, + "grad_norm": 1.9590543508529663, + "learning_rate": 1e-06, + "loss": 0.4774, + "mean_token_accuracy": 0.8284396529197693, + "num_tokens": 685173543.0, + "step": 26456 + }, + { + "epoch": 2.181660756988538, + "grad_norm": 2.14003586769104, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7474627494812012, + "num_tokens": 685205061.0, + "step": 26457 + }, + { + "epoch": 2.1817432176135894, + "grad_norm": 2.2523162364959717, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7614918947219849, + "num_tokens": 685232033.0, + "step": 26458 + }, + { + "epoch": 2.181825678238641, + "grad_norm": 1.9187241792678833, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7791855335235596, + "num_tokens": 685267822.0, + "step": 26459 + }, + { + "epoch": 2.1819081388636925, + "grad_norm": 2.4002816677093506, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7740815877914429, + "num_tokens": 685293459.0, + "step": 26460 + }, + { + "epoch": 2.1819905994887443, + "grad_norm": 1.969265103340149, + "learning_rate": 1e-06, + "loss": 0.6118, + "mean_token_accuracy": 0.799574613571167, + "num_tokens": 685322881.0, + "step": 26461 + }, + { + "epoch": 2.1820730601137956, + "grad_norm": 2.112323760986328, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7489590644836426, + "num_tokens": 685352554.0, + "step": 26462 + }, + { + "epoch": 2.182155520738847, + "grad_norm": 2.260629415512085, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7691371440887451, + "num_tokens": 685379930.0, + "step": 26463 + }, + { + "epoch": 2.182237981363899, + "grad_norm": 2.407597303390503, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7569124102592468, + "num_tokens": 685404626.0, + "step": 26464 + }, + { + "epoch": 2.18232044198895, + "grad_norm": 2.6542415618896484, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7489464282989502, + "num_tokens": 685428543.0, + "step": 26465 + }, + { + "epoch": 2.182402902614002, + "grad_norm": 2.0426228046417236, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7560707330703735, + "num_tokens": 685458932.0, + "step": 26466 + }, + { + "epoch": 2.1824853632390533, + "grad_norm": 2.351978063583374, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.7846751809120178, + "num_tokens": 685484526.0, + "step": 26467 + }, + { + "epoch": 2.182567823864105, + "grad_norm": 1.980063557624817, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7728980779647827, + "num_tokens": 685520608.0, + "step": 26468 + }, + { + "epoch": 2.1826502844891564, + "grad_norm": 2.0355474948883057, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7694041728973389, + "num_tokens": 685554804.0, + "step": 26469 + }, + { + "epoch": 2.182732745114208, + "grad_norm": 2.2188642024993896, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.763958215713501, + "num_tokens": 685581379.0, + "step": 26470 + }, + { + "epoch": 2.1828152057392596, + "grad_norm": 2.4503393173217773, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7358927130699158, + "num_tokens": 685606870.0, + "step": 26471 + }, + { + "epoch": 2.182897666364311, + "grad_norm": 2.2572710514068604, + "learning_rate": 1e-06, + "loss": 0.7724, + "mean_token_accuracy": 0.7475637197494507, + "num_tokens": 685633544.0, + "step": 26472 + }, + { + "epoch": 2.1829801269893627, + "grad_norm": 2.477804660797119, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7719336152076721, + "num_tokens": 685657692.0, + "step": 26473 + }, + { + "epoch": 2.183062587614414, + "grad_norm": 2.384232997894287, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7759312391281128, + "num_tokens": 685682648.0, + "step": 26474 + }, + { + "epoch": 2.1831450482394654, + "grad_norm": 1.9779003858566284, + "learning_rate": 1e-06, + "loss": 0.7774, + "mean_token_accuracy": 0.7462410926818848, + "num_tokens": 685716867.0, + "step": 26475 + }, + { + "epoch": 2.183227508864517, + "grad_norm": 2.1631150245666504, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7882505655288696, + "num_tokens": 685747619.0, + "step": 26476 + }, + { + "epoch": 2.1833099694895686, + "grad_norm": 2.2693920135498047, + "learning_rate": 1e-06, + "loss": 0.7623, + "mean_token_accuracy": 0.7556355595588684, + "num_tokens": 685776499.0, + "step": 26477 + }, + { + "epoch": 2.1833924301146204, + "grad_norm": 2.024487018585205, + "learning_rate": 1e-06, + "loss": 0.5965, + "mean_token_accuracy": 0.802918553352356, + "num_tokens": 685805254.0, + "step": 26478 + }, + { + "epoch": 2.1834748907396717, + "grad_norm": 1.99342679977417, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7698243260383606, + "num_tokens": 685839811.0, + "step": 26479 + }, + { + "epoch": 2.1835573513647235, + "grad_norm": 1.8140020370483398, + "learning_rate": 1e-06, + "loss": 0.5463, + "mean_token_accuracy": 0.8146268129348755, + "num_tokens": 685877372.0, + "step": 26480 + }, + { + "epoch": 2.183639811989775, + "grad_norm": 2.271641969680786, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7572009563446045, + "num_tokens": 685905606.0, + "step": 26481 + }, + { + "epoch": 2.183722272614826, + "grad_norm": 2.4860434532165527, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.758270263671875, + "num_tokens": 685930364.0, + "step": 26482 + }, + { + "epoch": 2.183804733239878, + "grad_norm": 1.9232409000396729, + "learning_rate": 1e-06, + "loss": 0.6106, + "mean_token_accuracy": 0.7978165149688721, + "num_tokens": 685964385.0, + "step": 26483 + }, + { + "epoch": 2.1838871938649294, + "grad_norm": 1.8456804752349854, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7596896886825562, + "num_tokens": 686002772.0, + "step": 26484 + }, + { + "epoch": 2.183969654489981, + "grad_norm": 1.9542371034622192, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7572619318962097, + "num_tokens": 686037603.0, + "step": 26485 + }, + { + "epoch": 2.1840521151150325, + "grad_norm": 2.068504571914673, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7716357707977295, + "num_tokens": 686067753.0, + "step": 26486 + }, + { + "epoch": 2.1841345757400843, + "grad_norm": 1.9376487731933594, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7726848721504211, + "num_tokens": 686100762.0, + "step": 26487 + }, + { + "epoch": 2.1842170363651356, + "grad_norm": 2.1985373497009277, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.7775012254714966, + "num_tokens": 686127067.0, + "step": 26488 + }, + { + "epoch": 2.184299496990187, + "grad_norm": 2.1001968383789062, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.7932531237602234, + "num_tokens": 686155846.0, + "step": 26489 + }, + { + "epoch": 2.184381957615239, + "grad_norm": 2.116563081741333, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.7941684126853943, + "num_tokens": 686182823.0, + "step": 26490 + }, + { + "epoch": 2.18446441824029, + "grad_norm": 2.251033067703247, + "learning_rate": 1e-06, + "loss": 0.5417, + "mean_token_accuracy": 0.8228664398193359, + "num_tokens": 686217391.0, + "step": 26491 + }, + { + "epoch": 2.184546878865342, + "grad_norm": 2.216168165206909, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.78957599401474, + "num_tokens": 686242671.0, + "step": 26492 + }, + { + "epoch": 2.1846293394903933, + "grad_norm": 2.2640461921691895, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7782878279685974, + "num_tokens": 686268448.0, + "step": 26493 + }, + { + "epoch": 2.184711800115445, + "grad_norm": 2.3354361057281494, + "learning_rate": 1e-06, + "loss": 0.5924, + "mean_token_accuracy": 0.8078528046607971, + "num_tokens": 686293285.0, + "step": 26494 + }, + { + "epoch": 2.1847942607404964, + "grad_norm": 2.11657977104187, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7699750661849976, + "num_tokens": 686324546.0, + "step": 26495 + }, + { + "epoch": 2.184876721365548, + "grad_norm": 2.1785645484924316, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7737733125686646, + "num_tokens": 686353168.0, + "step": 26496 + }, + { + "epoch": 2.1849591819905996, + "grad_norm": 2.0101590156555176, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.7867681980133057, + "num_tokens": 686385311.0, + "step": 26497 + }, + { + "epoch": 2.185041642615651, + "grad_norm": 2.1802821159362793, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.7933741807937622, + "num_tokens": 686414180.0, + "step": 26498 + }, + { + "epoch": 2.1851241032407027, + "grad_norm": 2.1319494247436523, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7770707607269287, + "num_tokens": 686446434.0, + "step": 26499 + }, + { + "epoch": 2.185206563865754, + "grad_norm": 2.023329019546509, + "learning_rate": 1e-06, + "loss": 0.608, + "mean_token_accuracy": 0.7996937036514282, + "num_tokens": 686476329.0, + "step": 26500 + }, + { + "epoch": 2.185289024490806, + "grad_norm": 2.469731569290161, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7740131616592407, + "num_tokens": 686501717.0, + "step": 26501 + }, + { + "epoch": 2.185371485115857, + "grad_norm": 1.7520474195480347, + "learning_rate": 1e-06, + "loss": 0.4666, + "mean_token_accuracy": 0.8315896391868591, + "num_tokens": 686537710.0, + "step": 26502 + }, + { + "epoch": 2.1854539457409086, + "grad_norm": 2.1229889392852783, + "learning_rate": 1e-06, + "loss": 0.5708, + "mean_token_accuracy": 0.7995848059654236, + "num_tokens": 686566111.0, + "step": 26503 + }, + { + "epoch": 2.1855364063659604, + "grad_norm": 1.9064075946807861, + "learning_rate": 1e-06, + "loss": 0.569, + "mean_token_accuracy": 0.8098937273025513, + "num_tokens": 686598932.0, + "step": 26504 + }, + { + "epoch": 2.1856188669910117, + "grad_norm": 2.0386791229248047, + "learning_rate": 1e-06, + "loss": 0.5539, + "mean_token_accuracy": 0.805272102355957, + "num_tokens": 686625055.0, + "step": 26505 + }, + { + "epoch": 2.1857013276160635, + "grad_norm": 2.242206573486328, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7761601805686951, + "num_tokens": 686652538.0, + "step": 26506 + }, + { + "epoch": 2.185783788241115, + "grad_norm": 1.8265193700790405, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7635623216629028, + "num_tokens": 686691307.0, + "step": 26507 + }, + { + "epoch": 2.185866248866166, + "grad_norm": 2.162247657775879, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.768412172794342, + "num_tokens": 686721097.0, + "step": 26508 + }, + { + "epoch": 2.185948709491218, + "grad_norm": 2.341862916946411, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7841942310333252, + "num_tokens": 686745809.0, + "step": 26509 + }, + { + "epoch": 2.1860311701162694, + "grad_norm": 2.4726548194885254, + "learning_rate": 1e-06, + "loss": 0.5129, + "mean_token_accuracy": 0.8219995498657227, + "num_tokens": 686766713.0, + "step": 26510 + }, + { + "epoch": 2.186113630741321, + "grad_norm": 2.1549932956695557, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7564088106155396, + "num_tokens": 686797943.0, + "step": 26511 + }, + { + "epoch": 2.1861960913663725, + "grad_norm": 2.226688861846924, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.746238112449646, + "num_tokens": 686825946.0, + "step": 26512 + }, + { + "epoch": 2.1862785519914243, + "grad_norm": 2.6482908725738525, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7626327872276306, + "num_tokens": 686846939.0, + "step": 26513 + }, + { + "epoch": 2.1863610126164756, + "grad_norm": 2.4875564575195312, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7706267237663269, + "num_tokens": 686867895.0, + "step": 26514 + }, + { + "epoch": 2.186443473241527, + "grad_norm": 1.8155150413513184, + "learning_rate": 1e-06, + "loss": 0.6042, + "mean_token_accuracy": 0.7851674556732178, + "num_tokens": 686901344.0, + "step": 26515 + }, + { + "epoch": 2.186525933866579, + "grad_norm": 2.1091067790985107, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.7968206405639648, + "num_tokens": 686929241.0, + "step": 26516 + }, + { + "epoch": 2.18660839449163, + "grad_norm": 1.804396390914917, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7729633450508118, + "num_tokens": 686964169.0, + "step": 26517 + }, + { + "epoch": 2.186690855116682, + "grad_norm": 1.8460931777954102, + "learning_rate": 1e-06, + "loss": 0.5797, + "mean_token_accuracy": 0.8027708530426025, + "num_tokens": 686997952.0, + "step": 26518 + }, + { + "epoch": 2.1867733157417333, + "grad_norm": 1.8765668869018555, + "learning_rate": 1e-06, + "loss": 0.5961, + "mean_token_accuracy": 0.8055027723312378, + "num_tokens": 687032877.0, + "step": 26519 + }, + { + "epoch": 2.1868557763667846, + "grad_norm": 1.9130654335021973, + "learning_rate": 1e-06, + "loss": 0.5647, + "mean_token_accuracy": 0.8153426647186279, + "num_tokens": 687062654.0, + "step": 26520 + }, + { + "epoch": 2.1869382369918364, + "grad_norm": 1.9051765203475952, + "learning_rate": 1e-06, + "loss": 0.5681, + "mean_token_accuracy": 0.799828290939331, + "num_tokens": 687094045.0, + "step": 26521 + }, + { + "epoch": 2.1870206976168878, + "grad_norm": 2.0065767765045166, + "learning_rate": 1e-06, + "loss": 0.6058, + "mean_token_accuracy": 0.8012878894805908, + "num_tokens": 687125114.0, + "step": 26522 + }, + { + "epoch": 2.1871031582419396, + "grad_norm": 2.315140724182129, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7756071090698242, + "num_tokens": 687151505.0, + "step": 26523 + }, + { + "epoch": 2.187185618866991, + "grad_norm": 2.296201705932617, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7790597677230835, + "num_tokens": 687176356.0, + "step": 26524 + }, + { + "epoch": 2.1872680794920427, + "grad_norm": 2.181036949157715, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7500057220458984, + "num_tokens": 687204195.0, + "step": 26525 + }, + { + "epoch": 2.187350540117094, + "grad_norm": 2.2015109062194824, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7779805660247803, + "num_tokens": 687233158.0, + "step": 26526 + }, + { + "epoch": 2.1874330007421454, + "grad_norm": 2.0839672088623047, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7670131921768188, + "num_tokens": 687265306.0, + "step": 26527 + }, + { + "epoch": 2.187515461367197, + "grad_norm": 2.1218109130859375, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.778935432434082, + "num_tokens": 687296823.0, + "step": 26528 + }, + { + "epoch": 2.1875979219922486, + "grad_norm": 2.1463849544525146, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7354859709739685, + "num_tokens": 687328642.0, + "step": 26529 + }, + { + "epoch": 2.1876803826173004, + "grad_norm": 1.8089485168457031, + "learning_rate": 1e-06, + "loss": 0.5818, + "mean_token_accuracy": 0.7969040870666504, + "num_tokens": 687362698.0, + "step": 26530 + }, + { + "epoch": 2.1877628432423517, + "grad_norm": 2.1065640449523926, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7353386282920837, + "num_tokens": 687395534.0, + "step": 26531 + }, + { + "epoch": 2.1878453038674035, + "grad_norm": 1.9982116222381592, + "learning_rate": 1e-06, + "loss": 0.6515, + "mean_token_accuracy": 0.7977251410484314, + "num_tokens": 687428454.0, + "step": 26532 + }, + { + "epoch": 2.187927764492455, + "grad_norm": 2.2213659286499023, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7375128269195557, + "num_tokens": 687455860.0, + "step": 26533 + }, + { + "epoch": 2.188010225117506, + "grad_norm": 2.2584731578826904, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7750803232192993, + "num_tokens": 687483104.0, + "step": 26534 + }, + { + "epoch": 2.188092685742558, + "grad_norm": 2.0555479526519775, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.777880847454071, + "num_tokens": 687512726.0, + "step": 26535 + }, + { + "epoch": 2.1881751463676093, + "grad_norm": 1.9613161087036133, + "learning_rate": 1e-06, + "loss": 0.5314, + "mean_token_accuracy": 0.8172554969787598, + "num_tokens": 687541142.0, + "step": 26536 + }, + { + "epoch": 2.188257606992661, + "grad_norm": 2.297264337539673, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.7881627082824707, + "num_tokens": 687569158.0, + "step": 26537 + }, + { + "epoch": 2.1883400676177125, + "grad_norm": 2.1596057415008545, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7558867931365967, + "num_tokens": 687597990.0, + "step": 26538 + }, + { + "epoch": 2.1884225282427643, + "grad_norm": 1.976163625717163, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.7979938983917236, + "num_tokens": 687630748.0, + "step": 26539 + }, + { + "epoch": 2.1885049888678156, + "grad_norm": 2.231109380722046, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7595800757408142, + "num_tokens": 687657284.0, + "step": 26540 + }, + { + "epoch": 2.188587449492867, + "grad_norm": 2.063875436782837, + "learning_rate": 1e-06, + "loss": 0.5129, + "mean_token_accuracy": 0.8167451620101929, + "num_tokens": 687685052.0, + "step": 26541 + }, + { + "epoch": 2.188669910117919, + "grad_norm": 2.1077473163604736, + "learning_rate": 1e-06, + "loss": 0.5936, + "mean_token_accuracy": 0.7952049970626831, + "num_tokens": 687715235.0, + "step": 26542 + }, + { + "epoch": 2.18875237074297, + "grad_norm": 2.044194221496582, + "learning_rate": 1e-06, + "loss": 0.5961, + "mean_token_accuracy": 0.7967174649238586, + "num_tokens": 687744420.0, + "step": 26543 + }, + { + "epoch": 2.188834831368022, + "grad_norm": 2.0676045417785645, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7606694102287292, + "num_tokens": 687776344.0, + "step": 26544 + }, + { + "epoch": 2.1889172919930733, + "grad_norm": 2.666640043258667, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7484722137451172, + "num_tokens": 687797548.0, + "step": 26545 + }, + { + "epoch": 2.188999752618125, + "grad_norm": 1.9640663862228394, + "learning_rate": 1e-06, + "loss": 0.6069, + "mean_token_accuracy": 0.8012686967849731, + "num_tokens": 687828302.0, + "step": 26546 + }, + { + "epoch": 2.1890822132431764, + "grad_norm": 1.8710086345672607, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.7915835380554199, + "num_tokens": 687860952.0, + "step": 26547 + }, + { + "epoch": 2.1891646738682278, + "grad_norm": 1.932071566581726, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7779934406280518, + "num_tokens": 687897781.0, + "step": 26548 + }, + { + "epoch": 2.1892471344932796, + "grad_norm": 2.2699079513549805, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.7912722826004028, + "num_tokens": 687923967.0, + "step": 26549 + }, + { + "epoch": 2.189329595118331, + "grad_norm": 2.009550094604492, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7766623497009277, + "num_tokens": 687954322.0, + "step": 26550 + }, + { + "epoch": 2.1894120557433827, + "grad_norm": 1.9430570602416992, + "learning_rate": 1e-06, + "loss": 0.7738, + "mean_token_accuracy": 0.7554824352264404, + "num_tokens": 687987908.0, + "step": 26551 + }, + { + "epoch": 2.189494516368434, + "grad_norm": 2.3713557720184326, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7476882338523865, + "num_tokens": 688013040.0, + "step": 26552 + }, + { + "epoch": 2.1895769769934854, + "grad_norm": 2.3094069957733154, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7560628652572632, + "num_tokens": 688039480.0, + "step": 26553 + }, + { + "epoch": 2.189659437618537, + "grad_norm": 2.1110236644744873, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.7795205116271973, + "num_tokens": 688066312.0, + "step": 26554 + }, + { + "epoch": 2.1897418982435886, + "grad_norm": 2.096614360809326, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.7494899034500122, + "num_tokens": 688099778.0, + "step": 26555 + }, + { + "epoch": 2.1898243588686404, + "grad_norm": 2.2770118713378906, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7301356196403503, + "num_tokens": 688126828.0, + "step": 26556 + }, + { + "epoch": 2.1899068194936917, + "grad_norm": 2.311305046081543, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7716060876846313, + "num_tokens": 688154937.0, + "step": 26557 + }, + { + "epoch": 2.1899892801187435, + "grad_norm": 2.1350960731506348, + "learning_rate": 1e-06, + "loss": 0.7702, + "mean_token_accuracy": 0.7465431690216064, + "num_tokens": 688183626.0, + "step": 26558 + }, + { + "epoch": 2.190071740743795, + "grad_norm": 1.9863864183425903, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7529658079147339, + "num_tokens": 688213860.0, + "step": 26559 + }, + { + "epoch": 2.190154201368846, + "grad_norm": 2.3735666275024414, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7477256059646606, + "num_tokens": 688239957.0, + "step": 26560 + }, + { + "epoch": 2.190236661993898, + "grad_norm": 2.401491165161133, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7609416246414185, + "num_tokens": 688264474.0, + "step": 26561 + }, + { + "epoch": 2.1903191226189493, + "grad_norm": 2.2743191719055176, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7575275897979736, + "num_tokens": 688295085.0, + "step": 26562 + }, + { + "epoch": 2.190401583244001, + "grad_norm": 1.8340526819229126, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.7917811870574951, + "num_tokens": 688329882.0, + "step": 26563 + }, + { + "epoch": 2.1904840438690525, + "grad_norm": 1.8176274299621582, + "learning_rate": 1e-06, + "loss": 0.6067, + "mean_token_accuracy": 0.7859461307525635, + "num_tokens": 688367466.0, + "step": 26564 + }, + { + "epoch": 2.190566504494104, + "grad_norm": 2.024195909500122, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7385594248771667, + "num_tokens": 688398567.0, + "step": 26565 + }, + { + "epoch": 2.1906489651191556, + "grad_norm": 2.2877840995788574, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7525032758712769, + "num_tokens": 688426567.0, + "step": 26566 + }, + { + "epoch": 2.190731425744207, + "grad_norm": 1.9086087942123413, + "learning_rate": 1e-06, + "loss": 0.534, + "mean_token_accuracy": 0.8170514106750488, + "num_tokens": 688459158.0, + "step": 26567 + }, + { + "epoch": 2.190813886369259, + "grad_norm": 2.3489089012145996, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7620970010757446, + "num_tokens": 688486124.0, + "step": 26568 + }, + { + "epoch": 2.19089634699431, + "grad_norm": 2.1451847553253174, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7793517112731934, + "num_tokens": 688515843.0, + "step": 26569 + }, + { + "epoch": 2.190978807619362, + "grad_norm": 2.0828685760498047, + "learning_rate": 1e-06, + "loss": 0.583, + "mean_token_accuracy": 0.8011449575424194, + "num_tokens": 688543898.0, + "step": 26570 + }, + { + "epoch": 2.1910612682444133, + "grad_norm": 2.0384223461151123, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7653859853744507, + "num_tokens": 688572679.0, + "step": 26571 + }, + { + "epoch": 2.1911437288694646, + "grad_norm": 2.121293306350708, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.7884082794189453, + "num_tokens": 688601822.0, + "step": 26572 + }, + { + "epoch": 2.1912261894945164, + "grad_norm": 2.203821897506714, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7404806613922119, + "num_tokens": 688633158.0, + "step": 26573 + }, + { + "epoch": 2.1913086501195678, + "grad_norm": 2.1414599418640137, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7418832182884216, + "num_tokens": 688665271.0, + "step": 26574 + }, + { + "epoch": 2.1913911107446196, + "grad_norm": 2.2054178714752197, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7744749188423157, + "num_tokens": 688691822.0, + "step": 26575 + }, + { + "epoch": 2.191473571369671, + "grad_norm": 2.3388073444366455, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.738788902759552, + "num_tokens": 688720692.0, + "step": 26576 + }, + { + "epoch": 2.1915560319947227, + "grad_norm": 2.009223699569702, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7132360339164734, + "num_tokens": 688752491.0, + "step": 26577 + }, + { + "epoch": 2.191638492619774, + "grad_norm": 2.010302782058716, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7788591384887695, + "num_tokens": 688783370.0, + "step": 26578 + }, + { + "epoch": 2.1917209532448254, + "grad_norm": 1.878143548965454, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7714849710464478, + "num_tokens": 688816918.0, + "step": 26579 + }, + { + "epoch": 2.191803413869877, + "grad_norm": 2.1779980659484863, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.7865270376205444, + "num_tokens": 688845955.0, + "step": 26580 + }, + { + "epoch": 2.1918858744949286, + "grad_norm": 1.8874269723892212, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.752961277961731, + "num_tokens": 688879065.0, + "step": 26581 + }, + { + "epoch": 2.1919683351199803, + "grad_norm": 2.019217014312744, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7701718211174011, + "num_tokens": 688910484.0, + "step": 26582 + }, + { + "epoch": 2.1920507957450317, + "grad_norm": 1.8986679315567017, + "learning_rate": 1e-06, + "loss": 0.5658, + "mean_token_accuracy": 0.8053815364837646, + "num_tokens": 688944213.0, + "step": 26583 + }, + { + "epoch": 2.1921332563700835, + "grad_norm": 2.0947964191436768, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7765299081802368, + "num_tokens": 688973116.0, + "step": 26584 + }, + { + "epoch": 2.192215716995135, + "grad_norm": 1.8316789865493774, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.786224365234375, + "num_tokens": 689011053.0, + "step": 26585 + }, + { + "epoch": 2.192298177620186, + "grad_norm": 2.2556869983673096, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7687584757804871, + "num_tokens": 689039916.0, + "step": 26586 + }, + { + "epoch": 2.192380638245238, + "grad_norm": 2.231142044067383, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7621844410896301, + "num_tokens": 689069840.0, + "step": 26587 + }, + { + "epoch": 2.1924630988702893, + "grad_norm": 2.003960609436035, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.7930862903594971, + "num_tokens": 689100191.0, + "step": 26588 + }, + { + "epoch": 2.192545559495341, + "grad_norm": 1.9400219917297363, + "learning_rate": 1e-06, + "loss": 0.5567, + "mean_token_accuracy": 0.8071069717407227, + "num_tokens": 689134967.0, + "step": 26589 + }, + { + "epoch": 2.1926280201203925, + "grad_norm": 2.375166416168213, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7653229236602783, + "num_tokens": 689160224.0, + "step": 26590 + }, + { + "epoch": 2.1927104807454443, + "grad_norm": 2.5475852489471436, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7508516311645508, + "num_tokens": 689185219.0, + "step": 26591 + }, + { + "epoch": 2.1927929413704956, + "grad_norm": 2.138101100921631, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.7915492057800293, + "num_tokens": 689213824.0, + "step": 26592 + }, + { + "epoch": 2.192875401995547, + "grad_norm": 2.036071538925171, + "learning_rate": 1e-06, + "loss": 0.6127, + "mean_token_accuracy": 0.7982437610626221, + "num_tokens": 689243889.0, + "step": 26593 + }, + { + "epoch": 2.1929578626205988, + "grad_norm": 1.8885130882263184, + "learning_rate": 1e-06, + "loss": 0.6013, + "mean_token_accuracy": 0.8037499785423279, + "num_tokens": 689276134.0, + "step": 26594 + }, + { + "epoch": 2.19304032324565, + "grad_norm": 2.051130771636963, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7841867208480835, + "num_tokens": 689306258.0, + "step": 26595 + }, + { + "epoch": 2.193122783870702, + "grad_norm": 1.9771629571914673, + "learning_rate": 1e-06, + "loss": 0.589, + "mean_token_accuracy": 0.7925268411636353, + "num_tokens": 689335673.0, + "step": 26596 + }, + { + "epoch": 2.1932052444957533, + "grad_norm": 2.1396536827087402, + "learning_rate": 1e-06, + "loss": 0.5314, + "mean_token_accuracy": 0.8230184316635132, + "num_tokens": 689362416.0, + "step": 26597 + }, + { + "epoch": 2.1932877051208046, + "grad_norm": 1.8596900701522827, + "learning_rate": 1e-06, + "loss": 0.5672, + "mean_token_accuracy": 0.8108828663825989, + "num_tokens": 689395899.0, + "step": 26598 + }, + { + "epoch": 2.1933701657458564, + "grad_norm": 2.8323464393615723, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.7880591750144958, + "num_tokens": 689415348.0, + "step": 26599 + }, + { + "epoch": 2.1934526263709078, + "grad_norm": 2.0922346115112305, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7540317177772522, + "num_tokens": 689446688.0, + "step": 26600 + }, + { + "epoch": 2.1935350869959596, + "grad_norm": 2.0669877529144287, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.7788981199264526, + "num_tokens": 689474876.0, + "step": 26601 + }, + { + "epoch": 2.193617547621011, + "grad_norm": 1.9870519638061523, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7694531679153442, + "num_tokens": 689507405.0, + "step": 26602 + }, + { + "epoch": 2.1937000082460627, + "grad_norm": 2.1556167602539062, + "learning_rate": 1e-06, + "loss": 0.5089, + "mean_token_accuracy": 0.8269930481910706, + "num_tokens": 689531893.0, + "step": 26603 + }, + { + "epoch": 2.193782468871114, + "grad_norm": 2.3126957416534424, + "learning_rate": 1e-06, + "loss": 0.7702, + "mean_token_accuracy": 0.7453442811965942, + "num_tokens": 689558431.0, + "step": 26604 + }, + { + "epoch": 2.1938649294961654, + "grad_norm": 2.0697314739227295, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7769903540611267, + "num_tokens": 689589850.0, + "step": 26605 + }, + { + "epoch": 2.193947390121217, + "grad_norm": 1.9417634010314941, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7810670137405396, + "num_tokens": 689622755.0, + "step": 26606 + }, + { + "epoch": 2.1940298507462686, + "grad_norm": 2.1269025802612305, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.7851669788360596, + "num_tokens": 689652392.0, + "step": 26607 + }, + { + "epoch": 2.1941123113713203, + "grad_norm": 1.6659058332443237, + "learning_rate": 1e-06, + "loss": 0.5381, + "mean_token_accuracy": 0.817507266998291, + "num_tokens": 689696135.0, + "step": 26608 + }, + { + "epoch": 2.1941947719963717, + "grad_norm": 2.0348265171051025, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.7860987782478333, + "num_tokens": 689725806.0, + "step": 26609 + }, + { + "epoch": 2.194277232621423, + "grad_norm": 1.8533844947814941, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.792681872844696, + "num_tokens": 689764076.0, + "step": 26610 + }, + { + "epoch": 2.194359693246475, + "grad_norm": 1.7548103332519531, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7737806439399719, + "num_tokens": 689803278.0, + "step": 26611 + }, + { + "epoch": 2.194442153871526, + "grad_norm": 2.4906461238861084, + "learning_rate": 1e-06, + "loss": 0.7883, + "mean_token_accuracy": 0.7521264553070068, + "num_tokens": 689826952.0, + "step": 26612 + }, + { + "epoch": 2.194524614496578, + "grad_norm": 1.9212560653686523, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.767627477645874, + "num_tokens": 689859796.0, + "step": 26613 + }, + { + "epoch": 2.1946070751216293, + "grad_norm": 2.0776469707489014, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7626426219940186, + "num_tokens": 689891859.0, + "step": 26614 + }, + { + "epoch": 2.194689535746681, + "grad_norm": 2.603573799133301, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7317412495613098, + "num_tokens": 689913072.0, + "step": 26615 + }, + { + "epoch": 2.1947719963717325, + "grad_norm": 2.228520393371582, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7586094737052917, + "num_tokens": 689939920.0, + "step": 26616 + }, + { + "epoch": 2.194854456996784, + "grad_norm": 2.1642892360687256, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7351348400115967, + "num_tokens": 689968645.0, + "step": 26617 + }, + { + "epoch": 2.1949369176218356, + "grad_norm": 1.9060548543930054, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7716217637062073, + "num_tokens": 690000433.0, + "step": 26618 + }, + { + "epoch": 2.195019378246887, + "grad_norm": 2.1484906673431396, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7686436176300049, + "num_tokens": 690028850.0, + "step": 26619 + }, + { + "epoch": 2.1951018388719388, + "grad_norm": 2.109159231185913, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7251290678977966, + "num_tokens": 690061271.0, + "step": 26620 + }, + { + "epoch": 2.19518429949699, + "grad_norm": 2.142963409423828, + "learning_rate": 1e-06, + "loss": 0.5944, + "mean_token_accuracy": 0.7998477220535278, + "num_tokens": 690085616.0, + "step": 26621 + }, + { + "epoch": 2.195266760122042, + "grad_norm": 1.8303848505020142, + "learning_rate": 1e-06, + "loss": 0.5166, + "mean_token_accuracy": 0.8235958814620972, + "num_tokens": 690122006.0, + "step": 26622 + }, + { + "epoch": 2.1953492207470933, + "grad_norm": 2.263903856277466, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7230851650238037, + "num_tokens": 690150356.0, + "step": 26623 + }, + { + "epoch": 2.1954316813721446, + "grad_norm": 2.16184401512146, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7530592679977417, + "num_tokens": 690182907.0, + "step": 26624 + }, + { + "epoch": 2.1955141419971964, + "grad_norm": 2.115182399749756, + "learning_rate": 1e-06, + "loss": 0.5821, + "mean_token_accuracy": 0.8058089017868042, + "num_tokens": 690209780.0, + "step": 26625 + }, + { + "epoch": 2.1955966026222478, + "grad_norm": 2.146864175796509, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7642730474472046, + "num_tokens": 690242242.0, + "step": 26626 + }, + { + "epoch": 2.1956790632472996, + "grad_norm": 2.1347827911376953, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.7763705849647522, + "num_tokens": 690272854.0, + "step": 26627 + }, + { + "epoch": 2.195761523872351, + "grad_norm": 2.2319676876068115, + "learning_rate": 1e-06, + "loss": 0.5883, + "mean_token_accuracy": 0.8021129369735718, + "num_tokens": 690296560.0, + "step": 26628 + }, + { + "epoch": 2.1958439844974027, + "grad_norm": 2.09169602394104, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7787845134735107, + "num_tokens": 690326826.0, + "step": 26629 + }, + { + "epoch": 2.195926445122454, + "grad_norm": 1.8499236106872559, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.7888103723526001, + "num_tokens": 690362556.0, + "step": 26630 + }, + { + "epoch": 2.1960089057475054, + "grad_norm": 2.411088228225708, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.759943425655365, + "num_tokens": 690387038.0, + "step": 26631 + }, + { + "epoch": 2.196091366372557, + "grad_norm": 2.097214698791504, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7527116537094116, + "num_tokens": 690417501.0, + "step": 26632 + }, + { + "epoch": 2.1961738269976085, + "grad_norm": 1.9687756299972534, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7829672694206238, + "num_tokens": 690448466.0, + "step": 26633 + }, + { + "epoch": 2.1962562876226603, + "grad_norm": 2.0122475624084473, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7827368974685669, + "num_tokens": 690483332.0, + "step": 26634 + }, + { + "epoch": 2.1963387482477117, + "grad_norm": 2.0476253032684326, + "learning_rate": 1e-06, + "loss": 0.6666, + "mean_token_accuracy": 0.7844746112823486, + "num_tokens": 690512024.0, + "step": 26635 + }, + { + "epoch": 2.1964212088727635, + "grad_norm": 2.413062334060669, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7704648971557617, + "num_tokens": 690539140.0, + "step": 26636 + }, + { + "epoch": 2.196503669497815, + "grad_norm": 2.2857348918914795, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7490248680114746, + "num_tokens": 690566831.0, + "step": 26637 + }, + { + "epoch": 2.196586130122866, + "grad_norm": 2.128417491912842, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7378785610198975, + "num_tokens": 690597066.0, + "step": 26638 + }, + { + "epoch": 2.196668590747918, + "grad_norm": 1.8209812641143799, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7860895395278931, + "num_tokens": 690630958.0, + "step": 26639 + }, + { + "epoch": 2.1967510513729693, + "grad_norm": 1.8968156576156616, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7418768405914307, + "num_tokens": 690667003.0, + "step": 26640 + }, + { + "epoch": 2.196833511998021, + "grad_norm": 2.162930727005005, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7751504778862, + "num_tokens": 690694768.0, + "step": 26641 + }, + { + "epoch": 2.1969159726230725, + "grad_norm": 2.4660494327545166, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7434319257736206, + "num_tokens": 690718764.0, + "step": 26642 + }, + { + "epoch": 2.196998433248124, + "grad_norm": 2.088545083999634, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7561240196228027, + "num_tokens": 690751793.0, + "step": 26643 + }, + { + "epoch": 2.1970808938731756, + "grad_norm": 2.016655683517456, + "learning_rate": 1e-06, + "loss": 0.7768, + "mean_token_accuracy": 0.7503049373626709, + "num_tokens": 690784840.0, + "step": 26644 + }, + { + "epoch": 2.197163354498227, + "grad_norm": 1.778207778930664, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.795350968837738, + "num_tokens": 690819869.0, + "step": 26645 + }, + { + "epoch": 2.1972458151232788, + "grad_norm": 2.059534788131714, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7632884979248047, + "num_tokens": 690854372.0, + "step": 26646 + }, + { + "epoch": 2.19732827574833, + "grad_norm": 2.012111186981201, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7803820371627808, + "num_tokens": 690887107.0, + "step": 26647 + }, + { + "epoch": 2.197410736373382, + "grad_norm": 2.402684450149536, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7449380159378052, + "num_tokens": 690909689.0, + "step": 26648 + }, + { + "epoch": 2.1974931969984333, + "grad_norm": 1.8803106546401978, + "learning_rate": 1e-06, + "loss": 0.5432, + "mean_token_accuracy": 0.8245325684547424, + "num_tokens": 690943444.0, + "step": 26649 + }, + { + "epoch": 2.1975756576234846, + "grad_norm": 2.075467109680176, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7585657835006714, + "num_tokens": 690978403.0, + "step": 26650 + }, + { + "epoch": 2.1976581182485364, + "grad_norm": 2.009899616241455, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7442843914031982, + "num_tokens": 691013078.0, + "step": 26651 + }, + { + "epoch": 2.1977405788735878, + "grad_norm": 2.2373435497283936, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7744421362876892, + "num_tokens": 691038231.0, + "step": 26652 + }, + { + "epoch": 2.1978230394986396, + "grad_norm": 2.3730554580688477, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7788070440292358, + "num_tokens": 691061922.0, + "step": 26653 + }, + { + "epoch": 2.197905500123691, + "grad_norm": 2.1133105754852295, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7723146677017212, + "num_tokens": 691090507.0, + "step": 26654 + }, + { + "epoch": 2.1979879607487423, + "grad_norm": 2.119506359100342, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.7920321226119995, + "num_tokens": 691119592.0, + "step": 26655 + }, + { + "epoch": 2.198070421373794, + "grad_norm": 1.927931547164917, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.7871769666671753, + "num_tokens": 691152798.0, + "step": 26656 + }, + { + "epoch": 2.1981528819988454, + "grad_norm": 2.1398353576660156, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.7792972922325134, + "num_tokens": 691183348.0, + "step": 26657 + }, + { + "epoch": 2.198235342623897, + "grad_norm": 2.4326179027557373, + "learning_rate": 1e-06, + "loss": 0.5647, + "mean_token_accuracy": 0.801031231880188, + "num_tokens": 691206694.0, + "step": 26658 + }, + { + "epoch": 2.1983178032489485, + "grad_norm": 2.2167296409606934, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7791392207145691, + "num_tokens": 691235376.0, + "step": 26659 + }, + { + "epoch": 2.1984002638740003, + "grad_norm": 2.2564971446990967, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7244377136230469, + "num_tokens": 691262031.0, + "step": 26660 + }, + { + "epoch": 2.1984827244990517, + "grad_norm": 2.167524576187134, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7791755795478821, + "num_tokens": 691293042.0, + "step": 26661 + }, + { + "epoch": 2.198565185124103, + "grad_norm": 2.284364938735962, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7546384334564209, + "num_tokens": 691318752.0, + "step": 26662 + }, + { + "epoch": 2.198647645749155, + "grad_norm": 1.9943803548812866, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7769854664802551, + "num_tokens": 691351490.0, + "step": 26663 + }, + { + "epoch": 2.198730106374206, + "grad_norm": 1.7953591346740723, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7865344285964966, + "num_tokens": 691387358.0, + "step": 26664 + }, + { + "epoch": 2.198812566999258, + "grad_norm": 2.214218854904175, + "learning_rate": 1e-06, + "loss": 0.6139, + "mean_token_accuracy": 0.7927601933479309, + "num_tokens": 691413297.0, + "step": 26665 + }, + { + "epoch": 2.1988950276243093, + "grad_norm": 2.1429412364959717, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7431453466415405, + "num_tokens": 691444410.0, + "step": 26666 + }, + { + "epoch": 2.198977488249361, + "grad_norm": 2.132413387298584, + "learning_rate": 1e-06, + "loss": 0.5581, + "mean_token_accuracy": 0.8121424913406372, + "num_tokens": 691472722.0, + "step": 26667 + }, + { + "epoch": 2.1990599488744125, + "grad_norm": 2.1070644855499268, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7362647652626038, + "num_tokens": 691503802.0, + "step": 26668 + }, + { + "epoch": 2.199142409499464, + "grad_norm": 2.04006290435791, + "learning_rate": 1e-06, + "loss": 0.5746, + "mean_token_accuracy": 0.8178732395172119, + "num_tokens": 691532379.0, + "step": 26669 + }, + { + "epoch": 2.1992248701245156, + "grad_norm": 2.0347225666046143, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7747501134872437, + "num_tokens": 691561018.0, + "step": 26670 + }, + { + "epoch": 2.199307330749567, + "grad_norm": 2.413647174835205, + "learning_rate": 1e-06, + "loss": 0.7429, + "mean_token_accuracy": 0.763024628162384, + "num_tokens": 691585341.0, + "step": 26671 + }, + { + "epoch": 2.1993897913746188, + "grad_norm": 2.247316837310791, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7715591192245483, + "num_tokens": 691611261.0, + "step": 26672 + }, + { + "epoch": 2.19947225199967, + "grad_norm": 2.1026899814605713, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.7844725847244263, + "num_tokens": 691643158.0, + "step": 26673 + }, + { + "epoch": 2.199554712624722, + "grad_norm": 2.262566566467285, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.7991812825202942, + "num_tokens": 691668329.0, + "step": 26674 + }, + { + "epoch": 2.1996371732497733, + "grad_norm": 2.075652837753296, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7786733508110046, + "num_tokens": 691701106.0, + "step": 26675 + }, + { + "epoch": 2.1997196338748246, + "grad_norm": 1.8510264158248901, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.76533043384552, + "num_tokens": 691737331.0, + "step": 26676 + }, + { + "epoch": 2.1998020944998764, + "grad_norm": 2.501317262649536, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7480548620223999, + "num_tokens": 691762342.0, + "step": 26677 + }, + { + "epoch": 2.1998845551249278, + "grad_norm": 2.071084976196289, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7558856010437012, + "num_tokens": 691793754.0, + "step": 26678 + }, + { + "epoch": 2.1999670157499795, + "grad_norm": 2.578629970550537, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7624223232269287, + "num_tokens": 691817019.0, + "step": 26679 + }, + { + "epoch": 2.200049476375031, + "grad_norm": 1.9983450174331665, + "learning_rate": 1e-06, + "loss": 0.5579, + "mean_token_accuracy": 0.8118447065353394, + "num_tokens": 691847476.0, + "step": 26680 + }, + { + "epoch": 2.2001319370000827, + "grad_norm": 2.1225154399871826, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.7772729396820068, + "num_tokens": 691877538.0, + "step": 26681 + }, + { + "epoch": 2.200214397625134, + "grad_norm": 1.880411148071289, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7741230726242065, + "num_tokens": 691912033.0, + "step": 26682 + }, + { + "epoch": 2.2002968582501854, + "grad_norm": 2.2054429054260254, + "learning_rate": 1e-06, + "loss": 0.5851, + "mean_token_accuracy": 0.8022062182426453, + "num_tokens": 691938455.0, + "step": 26683 + }, + { + "epoch": 2.200379318875237, + "grad_norm": 1.8587144613265991, + "learning_rate": 1e-06, + "loss": 0.5743, + "mean_token_accuracy": 0.8108822703361511, + "num_tokens": 691972296.0, + "step": 26684 + }, + { + "epoch": 2.2004617795002885, + "grad_norm": 2.058868646621704, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7698951959609985, + "num_tokens": 692004310.0, + "step": 26685 + }, + { + "epoch": 2.2005442401253403, + "grad_norm": 2.226018190383911, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7618924379348755, + "num_tokens": 692033655.0, + "step": 26686 + }, + { + "epoch": 2.2006267007503917, + "grad_norm": 2.01101016998291, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7804975509643555, + "num_tokens": 692063783.0, + "step": 26687 + }, + { + "epoch": 2.200709161375443, + "grad_norm": 2.13385272026062, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7918056845664978, + "num_tokens": 692092898.0, + "step": 26688 + }, + { + "epoch": 2.200791622000495, + "grad_norm": 2.101223945617676, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7626304626464844, + "num_tokens": 692123221.0, + "step": 26689 + }, + { + "epoch": 2.200874082625546, + "grad_norm": 2.2347216606140137, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7704020142555237, + "num_tokens": 692151228.0, + "step": 26690 + }, + { + "epoch": 2.200956543250598, + "grad_norm": 2.2094931602478027, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7473161220550537, + "num_tokens": 692182330.0, + "step": 26691 + }, + { + "epoch": 2.2010390038756493, + "grad_norm": 2.142803430557251, + "learning_rate": 1e-06, + "loss": 0.5728, + "mean_token_accuracy": 0.8056126832962036, + "num_tokens": 692208821.0, + "step": 26692 + }, + { + "epoch": 2.201121464500701, + "grad_norm": 2.2330713272094727, + "learning_rate": 1e-06, + "loss": 0.6063, + "mean_token_accuracy": 0.7893053889274597, + "num_tokens": 692237314.0, + "step": 26693 + }, + { + "epoch": 2.2012039251257525, + "grad_norm": 2.133207321166992, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7578722238540649, + "num_tokens": 692268948.0, + "step": 26694 + }, + { + "epoch": 2.201286385750804, + "grad_norm": 2.1920130252838135, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7686402797698975, + "num_tokens": 692297931.0, + "step": 26695 + }, + { + "epoch": 2.2013688463758556, + "grad_norm": 2.0302929878234863, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7720145583152771, + "num_tokens": 692329760.0, + "step": 26696 + }, + { + "epoch": 2.201451307000907, + "grad_norm": 2.0279202461242676, + "learning_rate": 1e-06, + "loss": 0.6001, + "mean_token_accuracy": 0.8088778853416443, + "num_tokens": 692360110.0, + "step": 26697 + }, + { + "epoch": 2.2015337676259588, + "grad_norm": 2.4178645610809326, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7472132444381714, + "num_tokens": 692384826.0, + "step": 26698 + }, + { + "epoch": 2.20161622825101, + "grad_norm": 2.0499355792999268, + "learning_rate": 1e-06, + "loss": 0.5528, + "mean_token_accuracy": 0.8128219842910767, + "num_tokens": 692412660.0, + "step": 26699 + }, + { + "epoch": 2.2016986888760615, + "grad_norm": 1.9079015254974365, + "learning_rate": 1e-06, + "loss": 0.5123, + "mean_token_accuracy": 0.8169591426849365, + "num_tokens": 692444079.0, + "step": 26700 + }, + { + "epoch": 2.2017811495011133, + "grad_norm": 2.2005369663238525, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.7898857593536377, + "num_tokens": 692471167.0, + "step": 26701 + }, + { + "epoch": 2.2018636101261646, + "grad_norm": 2.244874954223633, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7379079461097717, + "num_tokens": 692502977.0, + "step": 26702 + }, + { + "epoch": 2.2019460707512164, + "grad_norm": 2.1506714820861816, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.7837610244750977, + "num_tokens": 692532175.0, + "step": 26703 + }, + { + "epoch": 2.2020285313762678, + "grad_norm": 2.1680238246917725, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7778293490409851, + "num_tokens": 692561090.0, + "step": 26704 + }, + { + "epoch": 2.2021109920013195, + "grad_norm": 1.9654698371887207, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.7887696027755737, + "num_tokens": 692595458.0, + "step": 26705 + }, + { + "epoch": 2.202193452626371, + "grad_norm": 2.3516600131988525, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7288544774055481, + "num_tokens": 692621955.0, + "step": 26706 + }, + { + "epoch": 2.2022759132514222, + "grad_norm": 1.9626388549804688, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7734736204147339, + "num_tokens": 692658507.0, + "step": 26707 + }, + { + "epoch": 2.202358373876474, + "grad_norm": 2.025665760040283, + "learning_rate": 1e-06, + "loss": 0.4974, + "mean_token_accuracy": 0.829487681388855, + "num_tokens": 692687984.0, + "step": 26708 + }, + { + "epoch": 2.2024408345015254, + "grad_norm": 2.1088662147521973, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.785940408706665, + "num_tokens": 692716017.0, + "step": 26709 + }, + { + "epoch": 2.202523295126577, + "grad_norm": 2.067992925643921, + "learning_rate": 1e-06, + "loss": 0.5531, + "mean_token_accuracy": 0.8107220530509949, + "num_tokens": 692748004.0, + "step": 26710 + }, + { + "epoch": 2.2026057557516285, + "grad_norm": 1.9930074214935303, + "learning_rate": 1e-06, + "loss": 0.5676, + "mean_token_accuracy": 0.8101369142532349, + "num_tokens": 692778743.0, + "step": 26711 + }, + { + "epoch": 2.2026882163766803, + "grad_norm": 2.0620312690734863, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7649964690208435, + "num_tokens": 692815197.0, + "step": 26712 + }, + { + "epoch": 2.2027706770017317, + "grad_norm": 1.8211067914962769, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7846037745475769, + "num_tokens": 692850966.0, + "step": 26713 + }, + { + "epoch": 2.202853137626783, + "grad_norm": 2.093465805053711, + "learning_rate": 1e-06, + "loss": 0.5193, + "mean_token_accuracy": 0.7989616394042969, + "num_tokens": 692877034.0, + "step": 26714 + }, + { + "epoch": 2.202935598251835, + "grad_norm": 2.2787845134735107, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.75187087059021, + "num_tokens": 692905128.0, + "step": 26715 + }, + { + "epoch": 2.203018058876886, + "grad_norm": 2.2964022159576416, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7354588508605957, + "num_tokens": 692933570.0, + "step": 26716 + }, + { + "epoch": 2.203100519501938, + "grad_norm": 2.0804569721221924, + "learning_rate": 1e-06, + "loss": 0.6063, + "mean_token_accuracy": 0.7941688895225525, + "num_tokens": 692966171.0, + "step": 26717 + }, + { + "epoch": 2.2031829801269893, + "grad_norm": 1.96975576877594, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7691936492919922, + "num_tokens": 692999075.0, + "step": 26718 + }, + { + "epoch": 2.203265440752041, + "grad_norm": 1.9739887714385986, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7551703453063965, + "num_tokens": 693032923.0, + "step": 26719 + }, + { + "epoch": 2.2033479013770925, + "grad_norm": 2.530010938644409, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7559299468994141, + "num_tokens": 693056487.0, + "step": 26720 + }, + { + "epoch": 2.203430362002144, + "grad_norm": 2.032972574234009, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.7968670129776001, + "num_tokens": 693086616.0, + "step": 26721 + }, + { + "epoch": 2.2035128226271956, + "grad_norm": 1.9679538011550903, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.7815650701522827, + "num_tokens": 693120247.0, + "step": 26722 + }, + { + "epoch": 2.203595283252247, + "grad_norm": 1.9411817789077759, + "learning_rate": 1e-06, + "loss": 0.6004, + "mean_token_accuracy": 0.8039239645004272, + "num_tokens": 693150097.0, + "step": 26723 + }, + { + "epoch": 2.2036777438772988, + "grad_norm": 2.024272918701172, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.7950121164321899, + "num_tokens": 693180356.0, + "step": 26724 + }, + { + "epoch": 2.20376020450235, + "grad_norm": 2.4176886081695557, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7567050457000732, + "num_tokens": 693204236.0, + "step": 26725 + }, + { + "epoch": 2.203842665127402, + "grad_norm": 2.005035638809204, + "learning_rate": 1e-06, + "loss": 0.5566, + "mean_token_accuracy": 0.8118807077407837, + "num_tokens": 693237684.0, + "step": 26726 + }, + { + "epoch": 2.2039251257524533, + "grad_norm": 1.8508262634277344, + "learning_rate": 1e-06, + "loss": 0.5007, + "mean_token_accuracy": 0.8376017808914185, + "num_tokens": 693269344.0, + "step": 26727 + }, + { + "epoch": 2.2040075863775046, + "grad_norm": 1.9801193475723267, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7396835684776306, + "num_tokens": 693301246.0, + "step": 26728 + }, + { + "epoch": 2.2040900470025564, + "grad_norm": 2.0565946102142334, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7442891597747803, + "num_tokens": 693331754.0, + "step": 26729 + }, + { + "epoch": 2.2041725076276077, + "grad_norm": 2.3052539825439453, + "learning_rate": 1e-06, + "loss": 0.607, + "mean_token_accuracy": 0.7959511280059814, + "num_tokens": 693356027.0, + "step": 26730 + }, + { + "epoch": 2.2042549682526595, + "grad_norm": 2.181372880935669, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.7885921001434326, + "num_tokens": 693384539.0, + "step": 26731 + }, + { + "epoch": 2.204337428877711, + "grad_norm": 1.8824461698532104, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.7885649800300598, + "num_tokens": 693420576.0, + "step": 26732 + }, + { + "epoch": 2.2044198895027622, + "grad_norm": 1.8094233274459839, + "learning_rate": 1e-06, + "loss": 0.6021, + "mean_token_accuracy": 0.8026055693626404, + "num_tokens": 693460064.0, + "step": 26733 + }, + { + "epoch": 2.204502350127814, + "grad_norm": 1.9876645803451538, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7738146781921387, + "num_tokens": 693492568.0, + "step": 26734 + }, + { + "epoch": 2.2045848107528654, + "grad_norm": 2.0257210731506348, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7579963207244873, + "num_tokens": 693525553.0, + "step": 26735 + }, + { + "epoch": 2.204667271377917, + "grad_norm": 1.8064579963684082, + "learning_rate": 1e-06, + "loss": 0.5247, + "mean_token_accuracy": 0.8250061273574829, + "num_tokens": 693563221.0, + "step": 26736 + }, + { + "epoch": 2.2047497320029685, + "grad_norm": 2.008223295211792, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.7792625427246094, + "num_tokens": 693594640.0, + "step": 26737 + }, + { + "epoch": 2.2048321926280203, + "grad_norm": 1.8469897508621216, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.7513151168823242, + "num_tokens": 693632105.0, + "step": 26738 + }, + { + "epoch": 2.2049146532530717, + "grad_norm": 2.328840970993042, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7601616382598877, + "num_tokens": 693657990.0, + "step": 26739 + }, + { + "epoch": 2.204997113878123, + "grad_norm": 2.103984832763672, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7614496946334839, + "num_tokens": 693687506.0, + "step": 26740 + }, + { + "epoch": 2.205079574503175, + "grad_norm": 1.9419732093811035, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7663593292236328, + "num_tokens": 693719658.0, + "step": 26741 + }, + { + "epoch": 2.205162035128226, + "grad_norm": 2.489454984664917, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.7751119136810303, + "num_tokens": 693743405.0, + "step": 26742 + }, + { + "epoch": 2.205244495753278, + "grad_norm": 2.3239970207214355, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7784165143966675, + "num_tokens": 693768849.0, + "step": 26743 + }, + { + "epoch": 2.2053269563783293, + "grad_norm": 2.4374537467956543, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7506439685821533, + "num_tokens": 693794495.0, + "step": 26744 + }, + { + "epoch": 2.2054094170033807, + "grad_norm": 2.139896869659424, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7894432544708252, + "num_tokens": 693822866.0, + "step": 26745 + }, + { + "epoch": 2.2054918776284325, + "grad_norm": 2.1653432846069336, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7806773781776428, + "num_tokens": 693852081.0, + "step": 26746 + }, + { + "epoch": 2.205574338253484, + "grad_norm": 2.3436062335968018, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7485864758491516, + "num_tokens": 693878928.0, + "step": 26747 + }, + { + "epoch": 2.2056567988785356, + "grad_norm": 2.3115811347961426, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7745797038078308, + "num_tokens": 693905765.0, + "step": 26748 + }, + { + "epoch": 2.205739259503587, + "grad_norm": 2.3214173316955566, + "learning_rate": 1e-06, + "loss": 0.5721, + "mean_token_accuracy": 0.8007127642631531, + "num_tokens": 693929885.0, + "step": 26749 + }, + { + "epoch": 2.2058217201286388, + "grad_norm": 2.2966015338897705, + "learning_rate": 1e-06, + "loss": 0.4694, + "mean_token_accuracy": 0.8418796062469482, + "num_tokens": 693952339.0, + "step": 26750 + }, + { + "epoch": 2.20590418075369, + "grad_norm": 1.991121768951416, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7813330888748169, + "num_tokens": 693986184.0, + "step": 26751 + }, + { + "epoch": 2.2059866413787415, + "grad_norm": 2.0343215465545654, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.775460958480835, + "num_tokens": 694018228.0, + "step": 26752 + }, + { + "epoch": 2.2060691020037932, + "grad_norm": 2.293673515319824, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7734289169311523, + "num_tokens": 694044822.0, + "step": 26753 + }, + { + "epoch": 2.2061515626288446, + "grad_norm": 2.5734009742736816, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7820817232131958, + "num_tokens": 694066088.0, + "step": 26754 + }, + { + "epoch": 2.2062340232538964, + "grad_norm": 2.1219565868377686, + "learning_rate": 1e-06, + "loss": 0.5222, + "mean_token_accuracy": 0.8199677467346191, + "num_tokens": 694093715.0, + "step": 26755 + }, + { + "epoch": 2.2063164838789477, + "grad_norm": 2.17372989654541, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.7941843867301941, + "num_tokens": 694122947.0, + "step": 26756 + }, + { + "epoch": 2.2063989445039995, + "grad_norm": 2.290463447570801, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.7851759195327759, + "num_tokens": 694148507.0, + "step": 26757 + }, + { + "epoch": 2.206481405129051, + "grad_norm": 1.999624252319336, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7665742635726929, + "num_tokens": 694183169.0, + "step": 26758 + }, + { + "epoch": 2.2065638657541022, + "grad_norm": 2.151191234588623, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.75461745262146, + "num_tokens": 694213837.0, + "step": 26759 + }, + { + "epoch": 2.206646326379154, + "grad_norm": 1.9385298490524292, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7766143083572388, + "num_tokens": 694247477.0, + "step": 26760 + }, + { + "epoch": 2.2067287870042054, + "grad_norm": 2.2295844554901123, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.7823106646537781, + "num_tokens": 694273512.0, + "step": 26761 + }, + { + "epoch": 2.206811247629257, + "grad_norm": 2.1139111518859863, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.7773724794387817, + "num_tokens": 694305059.0, + "step": 26762 + }, + { + "epoch": 2.2068937082543085, + "grad_norm": 2.3060109615325928, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.754339337348938, + "num_tokens": 694330568.0, + "step": 26763 + }, + { + "epoch": 2.2069761688793603, + "grad_norm": 2.1306800842285156, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7350255250930786, + "num_tokens": 694363307.0, + "step": 26764 + }, + { + "epoch": 2.2070586295044117, + "grad_norm": 2.1474738121032715, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7592227458953857, + "num_tokens": 694396385.0, + "step": 26765 + }, + { + "epoch": 2.207141090129463, + "grad_norm": 2.2292556762695312, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7276206016540527, + "num_tokens": 694427172.0, + "step": 26766 + }, + { + "epoch": 2.207223550754515, + "grad_norm": 2.187976360321045, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7423416376113892, + "num_tokens": 694456419.0, + "step": 26767 + }, + { + "epoch": 2.207306011379566, + "grad_norm": 2.520385980606079, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7797566652297974, + "num_tokens": 694481456.0, + "step": 26768 + }, + { + "epoch": 2.207388472004618, + "grad_norm": 2.2226274013519287, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7749248743057251, + "num_tokens": 694507407.0, + "step": 26769 + }, + { + "epoch": 2.2074709326296693, + "grad_norm": 2.2174124717712402, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.7835676074028015, + "num_tokens": 694534614.0, + "step": 26770 + }, + { + "epoch": 2.2075533932547207, + "grad_norm": 1.8170748949050903, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7550138235092163, + "num_tokens": 694573815.0, + "step": 26771 + }, + { + "epoch": 2.2076358538797725, + "grad_norm": 2.2590627670288086, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.7819448709487915, + "num_tokens": 694599473.0, + "step": 26772 + }, + { + "epoch": 2.207718314504824, + "grad_norm": 1.9810550212860107, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7763880491256714, + "num_tokens": 694630070.0, + "step": 26773 + }, + { + "epoch": 2.2078007751298756, + "grad_norm": 2.1963675022125244, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7436583042144775, + "num_tokens": 694658984.0, + "step": 26774 + }, + { + "epoch": 2.207883235754927, + "grad_norm": 2.114917039871216, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.7599679231643677, + "num_tokens": 694688779.0, + "step": 26775 + }, + { + "epoch": 2.2079656963799787, + "grad_norm": 2.1893346309661865, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.7951938509941101, + "num_tokens": 694717724.0, + "step": 26776 + }, + { + "epoch": 2.20804815700503, + "grad_norm": 1.904869794845581, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7561206817626953, + "num_tokens": 694754160.0, + "step": 26777 + }, + { + "epoch": 2.2081306176300814, + "grad_norm": 2.100559949874878, + "learning_rate": 1e-06, + "loss": 0.6007, + "mean_token_accuracy": 0.8010695576667786, + "num_tokens": 694784497.0, + "step": 26778 + }, + { + "epoch": 2.2082130782551332, + "grad_norm": 2.2035560607910156, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7190983891487122, + "num_tokens": 694816714.0, + "step": 26779 + }, + { + "epoch": 2.2082955388801846, + "grad_norm": 2.2906136512756348, + "learning_rate": 1e-06, + "loss": 0.6177, + "mean_token_accuracy": 0.7902405858039856, + "num_tokens": 694840461.0, + "step": 26780 + }, + { + "epoch": 2.2083779995052364, + "grad_norm": 2.0669071674346924, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7575841546058655, + "num_tokens": 694871888.0, + "step": 26781 + }, + { + "epoch": 2.2084604601302877, + "grad_norm": 2.226724147796631, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7371603846549988, + "num_tokens": 694903631.0, + "step": 26782 + }, + { + "epoch": 2.208542920755339, + "grad_norm": 1.9166920185089111, + "learning_rate": 1e-06, + "loss": 0.5633, + "mean_token_accuracy": 0.8082406520843506, + "num_tokens": 694937894.0, + "step": 26783 + }, + { + "epoch": 2.208625381380391, + "grad_norm": 2.014608383178711, + "learning_rate": 1e-06, + "loss": 0.615, + "mean_token_accuracy": 0.7926890254020691, + "num_tokens": 694968405.0, + "step": 26784 + }, + { + "epoch": 2.2087078420054422, + "grad_norm": 2.100555419921875, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.8005033731460571, + "num_tokens": 694995546.0, + "step": 26785 + }, + { + "epoch": 2.208790302630494, + "grad_norm": 2.3235127925872803, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7507988214492798, + "num_tokens": 695024758.0, + "step": 26786 + }, + { + "epoch": 2.2088727632555454, + "grad_norm": 2.5210280418395996, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7572133541107178, + "num_tokens": 695047143.0, + "step": 26787 + }, + { + "epoch": 2.208955223880597, + "grad_norm": 2.461721897125244, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7692452669143677, + "num_tokens": 695075669.0, + "step": 26788 + }, + { + "epoch": 2.2090376845056485, + "grad_norm": 2.0732946395874023, + "learning_rate": 1e-06, + "loss": 0.5562, + "mean_token_accuracy": 0.8083385229110718, + "num_tokens": 695103764.0, + "step": 26789 + }, + { + "epoch": 2.2091201451307, + "grad_norm": 1.8151081800460815, + "learning_rate": 1e-06, + "loss": 0.538, + "mean_token_accuracy": 0.8106852769851685, + "num_tokens": 695139205.0, + "step": 26790 + }, + { + "epoch": 2.2092026057557517, + "grad_norm": 2.3778488636016846, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7364153861999512, + "num_tokens": 695164293.0, + "step": 26791 + }, + { + "epoch": 2.209285066380803, + "grad_norm": 1.8242735862731934, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7803171277046204, + "num_tokens": 695201023.0, + "step": 26792 + }, + { + "epoch": 2.209367527005855, + "grad_norm": 2.4244039058685303, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7786357998847961, + "num_tokens": 695223162.0, + "step": 26793 + }, + { + "epoch": 2.209449987630906, + "grad_norm": 2.277308464050293, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7597655057907104, + "num_tokens": 695249593.0, + "step": 26794 + }, + { + "epoch": 2.209532448255958, + "grad_norm": 1.8085811138153076, + "learning_rate": 1e-06, + "loss": 0.5338, + "mean_token_accuracy": 0.821759819984436, + "num_tokens": 695282659.0, + "step": 26795 + }, + { + "epoch": 2.2096149088810093, + "grad_norm": 2.537139654159546, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7544982433319092, + "num_tokens": 695305835.0, + "step": 26796 + }, + { + "epoch": 2.2096973695060607, + "grad_norm": 2.1076784133911133, + "learning_rate": 1e-06, + "loss": 0.7283, + "mean_token_accuracy": 0.7560833692550659, + "num_tokens": 695336105.0, + "step": 26797 + }, + { + "epoch": 2.2097798301311125, + "grad_norm": 1.9881259202957153, + "learning_rate": 1e-06, + "loss": 0.5273, + "mean_token_accuracy": 0.8197091221809387, + "num_tokens": 695368365.0, + "step": 26798 + }, + { + "epoch": 2.209862290756164, + "grad_norm": 2.349644422531128, + "learning_rate": 1e-06, + "loss": 0.5934, + "mean_token_accuracy": 0.7979987859725952, + "num_tokens": 695392603.0, + "step": 26799 + }, + { + "epoch": 2.2099447513812156, + "grad_norm": 2.1101818084716797, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.8038375973701477, + "num_tokens": 695422174.0, + "step": 26800 + }, + { + "epoch": 2.210027212006267, + "grad_norm": 1.857936978340149, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7694692611694336, + "num_tokens": 695458986.0, + "step": 26801 + }, + { + "epoch": 2.2101096726313187, + "grad_norm": 2.1958205699920654, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.743804931640625, + "num_tokens": 695488464.0, + "step": 26802 + }, + { + "epoch": 2.21019213325637, + "grad_norm": 2.0823142528533936, + "learning_rate": 1e-06, + "loss": 0.5526, + "mean_token_accuracy": 0.8128970861434937, + "num_tokens": 695514083.0, + "step": 26803 + }, + { + "epoch": 2.2102745938814214, + "grad_norm": 2.292422294616699, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.754862904548645, + "num_tokens": 695539619.0, + "step": 26804 + }, + { + "epoch": 2.2103570545064732, + "grad_norm": 2.333564281463623, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.7696319222450256, + "num_tokens": 695566380.0, + "step": 26805 + }, + { + "epoch": 2.2104395151315246, + "grad_norm": 2.290895938873291, + "learning_rate": 1e-06, + "loss": 0.5914, + "mean_token_accuracy": 0.8015032410621643, + "num_tokens": 695589944.0, + "step": 26806 + }, + { + "epoch": 2.2105219757565764, + "grad_norm": 1.9634511470794678, + "learning_rate": 1e-06, + "loss": 0.5844, + "mean_token_accuracy": 0.8033725023269653, + "num_tokens": 695625440.0, + "step": 26807 + }, + { + "epoch": 2.2106044363816277, + "grad_norm": 2.2417922019958496, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7673895955085754, + "num_tokens": 695653136.0, + "step": 26808 + }, + { + "epoch": 2.2106868970066795, + "grad_norm": 2.2666430473327637, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.774274468421936, + "num_tokens": 695678712.0, + "step": 26809 + }, + { + "epoch": 2.210769357631731, + "grad_norm": 2.069044828414917, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7360543012619019, + "num_tokens": 695714549.0, + "step": 26810 + }, + { + "epoch": 2.2108518182567822, + "grad_norm": 2.176693916320801, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.787844181060791, + "num_tokens": 695741202.0, + "step": 26811 + }, + { + "epoch": 2.210934278881834, + "grad_norm": 1.9124221801757812, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7282143831253052, + "num_tokens": 695779346.0, + "step": 26812 + }, + { + "epoch": 2.2110167395068854, + "grad_norm": 1.9396930932998657, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7363021969795227, + "num_tokens": 695818582.0, + "step": 26813 + }, + { + "epoch": 2.211099200131937, + "grad_norm": 2.3363630771636963, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7730764746665955, + "num_tokens": 695843900.0, + "step": 26814 + }, + { + "epoch": 2.2111816607569885, + "grad_norm": 2.1270792484283447, + "learning_rate": 1e-06, + "loss": 0.5618, + "mean_token_accuracy": 0.812706708908081, + "num_tokens": 695871822.0, + "step": 26815 + }, + { + "epoch": 2.21126412138204, + "grad_norm": 2.5041067600250244, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.7773377299308777, + "num_tokens": 695891798.0, + "step": 26816 + }, + { + "epoch": 2.2113465820070917, + "grad_norm": 2.0609726905822754, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.735837459564209, + "num_tokens": 695923292.0, + "step": 26817 + }, + { + "epoch": 2.211429042632143, + "grad_norm": 2.148207902908325, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.7798423171043396, + "num_tokens": 695951558.0, + "step": 26818 + }, + { + "epoch": 2.211511503257195, + "grad_norm": 2.1203691959381104, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.7855143547058105, + "num_tokens": 695986450.0, + "step": 26819 + }, + { + "epoch": 2.211593963882246, + "grad_norm": 2.0080058574676514, + "learning_rate": 1e-06, + "loss": 0.6134, + "mean_token_accuracy": 0.7962620258331299, + "num_tokens": 696017234.0, + "step": 26820 + }, + { + "epoch": 2.211676424507298, + "grad_norm": 1.962010383605957, + "learning_rate": 1e-06, + "loss": 0.584, + "mean_token_accuracy": 0.8030032515525818, + "num_tokens": 696046134.0, + "step": 26821 + }, + { + "epoch": 2.2117588851323493, + "grad_norm": 2.147754669189453, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.777987003326416, + "num_tokens": 696073625.0, + "step": 26822 + }, + { + "epoch": 2.2118413457574007, + "grad_norm": 1.7237906455993652, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.7946979403495789, + "num_tokens": 696113337.0, + "step": 26823 + }, + { + "epoch": 2.2119238063824525, + "grad_norm": 1.9299226999282837, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7462246417999268, + "num_tokens": 696145929.0, + "step": 26824 + }, + { + "epoch": 2.212006267007504, + "grad_norm": 1.8908021450042725, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.7890324592590332, + "num_tokens": 696177470.0, + "step": 26825 + }, + { + "epoch": 2.2120887276325556, + "grad_norm": 2.0149471759796143, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.7979425191879272, + "num_tokens": 696208067.0, + "step": 26826 + }, + { + "epoch": 2.212171188257607, + "grad_norm": 2.6291356086730957, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7498189210891724, + "num_tokens": 696229572.0, + "step": 26827 + }, + { + "epoch": 2.2122536488826583, + "grad_norm": 1.7783080339431763, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7569504976272583, + "num_tokens": 696268348.0, + "step": 26828 + }, + { + "epoch": 2.21233610950771, + "grad_norm": 2.4201273918151855, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7779778242111206, + "num_tokens": 696294535.0, + "step": 26829 + }, + { + "epoch": 2.2124185701327614, + "grad_norm": 2.2658681869506836, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7629787921905518, + "num_tokens": 696321901.0, + "step": 26830 + }, + { + "epoch": 2.2125010307578132, + "grad_norm": 2.131190061569214, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7733323574066162, + "num_tokens": 696351580.0, + "step": 26831 + }, + { + "epoch": 2.2125834913828646, + "grad_norm": 2.0940890312194824, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7737471461296082, + "num_tokens": 696382743.0, + "step": 26832 + }, + { + "epoch": 2.2126659520079164, + "grad_norm": 2.2455358505249023, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7834137678146362, + "num_tokens": 696409754.0, + "step": 26833 + }, + { + "epoch": 2.2127484126329677, + "grad_norm": 2.394505023956299, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7192953824996948, + "num_tokens": 696436552.0, + "step": 26834 + }, + { + "epoch": 2.212830873258019, + "grad_norm": 2.0767269134521484, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.790913462638855, + "num_tokens": 696465915.0, + "step": 26835 + }, + { + "epoch": 2.212913333883071, + "grad_norm": 2.266387701034546, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7450414896011353, + "num_tokens": 696494483.0, + "step": 26836 + }, + { + "epoch": 2.2129957945081222, + "grad_norm": 2.0641233921051025, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7539901733398438, + "num_tokens": 696525928.0, + "step": 26837 + }, + { + "epoch": 2.213078255133174, + "grad_norm": 2.069779872894287, + "learning_rate": 1e-06, + "loss": 0.6, + "mean_token_accuracy": 0.8015735745429993, + "num_tokens": 696556082.0, + "step": 26838 + }, + { + "epoch": 2.2131607157582254, + "grad_norm": 1.7967456579208374, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7623693943023682, + "num_tokens": 696594219.0, + "step": 26839 + }, + { + "epoch": 2.213243176383277, + "grad_norm": 2.027921676635742, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7347850799560547, + "num_tokens": 696628317.0, + "step": 26840 + }, + { + "epoch": 2.2133256370083285, + "grad_norm": 2.02024245262146, + "learning_rate": 1e-06, + "loss": 0.5813, + "mean_token_accuracy": 0.8007147312164307, + "num_tokens": 696658815.0, + "step": 26841 + }, + { + "epoch": 2.21340809763338, + "grad_norm": 2.029510974884033, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7805151343345642, + "num_tokens": 696689950.0, + "step": 26842 + }, + { + "epoch": 2.2134905582584317, + "grad_norm": 2.0149405002593994, + "learning_rate": 1e-06, + "loss": 0.6083, + "mean_token_accuracy": 0.7945727109909058, + "num_tokens": 696720795.0, + "step": 26843 + }, + { + "epoch": 2.213573018883483, + "grad_norm": 2.121229410171509, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7589821815490723, + "num_tokens": 696752594.0, + "step": 26844 + }, + { + "epoch": 2.213655479508535, + "grad_norm": 2.086665630340576, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.8020995259284973, + "num_tokens": 696780321.0, + "step": 26845 + }, + { + "epoch": 2.213737940133586, + "grad_norm": 1.9789172410964966, + "learning_rate": 1e-06, + "loss": 0.6057, + "mean_token_accuracy": 0.7954299449920654, + "num_tokens": 696813217.0, + "step": 26846 + }, + { + "epoch": 2.213820400758638, + "grad_norm": 2.2452645301818848, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7730329036712646, + "num_tokens": 696838848.0, + "step": 26847 + }, + { + "epoch": 2.2139028613836893, + "grad_norm": 1.9315941333770752, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7750716209411621, + "num_tokens": 696874568.0, + "step": 26848 + }, + { + "epoch": 2.2139853220087407, + "grad_norm": 1.9003615379333496, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.77402263879776, + "num_tokens": 696911544.0, + "step": 26849 + }, + { + "epoch": 2.2140677826337924, + "grad_norm": 2.400937795639038, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7705467343330383, + "num_tokens": 696938318.0, + "step": 26850 + }, + { + "epoch": 2.214150243258844, + "grad_norm": 1.9972702264785767, + "learning_rate": 1e-06, + "loss": 0.5526, + "mean_token_accuracy": 0.8158497214317322, + "num_tokens": 696971404.0, + "step": 26851 + }, + { + "epoch": 2.2142327038838956, + "grad_norm": 2.0112955570220947, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7635408639907837, + "num_tokens": 697004089.0, + "step": 26852 + }, + { + "epoch": 2.214315164508947, + "grad_norm": 2.08613920211792, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.776594877243042, + "num_tokens": 697032756.0, + "step": 26853 + }, + { + "epoch": 2.2143976251339987, + "grad_norm": 2.4196510314941406, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7472850680351257, + "num_tokens": 697059538.0, + "step": 26854 + }, + { + "epoch": 2.21448008575905, + "grad_norm": 1.9509954452514648, + "learning_rate": 1e-06, + "loss": 0.5575, + "mean_token_accuracy": 0.8086479902267456, + "num_tokens": 697093899.0, + "step": 26855 + }, + { + "epoch": 2.2145625463841014, + "grad_norm": 2.166714906692505, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.766979992389679, + "num_tokens": 697122700.0, + "step": 26856 + }, + { + "epoch": 2.2146450070091532, + "grad_norm": 1.7828013896942139, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.7891833782196045, + "num_tokens": 697163718.0, + "step": 26857 + }, + { + "epoch": 2.2147274676342046, + "grad_norm": 1.9649219512939453, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.7471702098846436, + "num_tokens": 697202831.0, + "step": 26858 + }, + { + "epoch": 2.2148099282592564, + "grad_norm": 1.999334692955017, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7735466957092285, + "num_tokens": 697235660.0, + "step": 26859 + }, + { + "epoch": 2.2148923888843077, + "grad_norm": 1.8678940534591675, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7321569919586182, + "num_tokens": 697273557.0, + "step": 26860 + }, + { + "epoch": 2.214974849509359, + "grad_norm": 1.8239765167236328, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7800798416137695, + "num_tokens": 697307925.0, + "step": 26861 + }, + { + "epoch": 2.215057310134411, + "grad_norm": 2.1419477462768555, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.7982137203216553, + "num_tokens": 697333330.0, + "step": 26862 + }, + { + "epoch": 2.2151397707594622, + "grad_norm": 1.941620111465454, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7744622230529785, + "num_tokens": 697366301.0, + "step": 26863 + }, + { + "epoch": 2.215222231384514, + "grad_norm": 1.8531280755996704, + "learning_rate": 1e-06, + "loss": 0.624, + "mean_token_accuracy": 0.7880252599716187, + "num_tokens": 697399775.0, + "step": 26864 + }, + { + "epoch": 2.2153046920095654, + "grad_norm": 1.9417935609817505, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7576606273651123, + "num_tokens": 697432625.0, + "step": 26865 + }, + { + "epoch": 2.215387152634617, + "grad_norm": 2.395162343978882, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.762633204460144, + "num_tokens": 697457817.0, + "step": 26866 + }, + { + "epoch": 2.2154696132596685, + "grad_norm": 2.600140333175659, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7785752415657043, + "num_tokens": 697479425.0, + "step": 26867 + }, + { + "epoch": 2.21555207388472, + "grad_norm": 2.212550640106201, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7734609246253967, + "num_tokens": 697505548.0, + "step": 26868 + }, + { + "epoch": 2.2156345345097717, + "grad_norm": 1.928500771522522, + "learning_rate": 1e-06, + "loss": 0.5895, + "mean_token_accuracy": 0.7984481453895569, + "num_tokens": 697538463.0, + "step": 26869 + }, + { + "epoch": 2.215716995134823, + "grad_norm": 2.1400105953216553, + "learning_rate": 1e-06, + "loss": 0.5806, + "mean_token_accuracy": 0.7922032475471497, + "num_tokens": 697565205.0, + "step": 26870 + }, + { + "epoch": 2.215799455759875, + "grad_norm": 2.2238080501556396, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7578384876251221, + "num_tokens": 697592547.0, + "step": 26871 + }, + { + "epoch": 2.215881916384926, + "grad_norm": 2.4274234771728516, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7513469457626343, + "num_tokens": 697618003.0, + "step": 26872 + }, + { + "epoch": 2.2159643770099775, + "grad_norm": 1.8098456859588623, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7921174168586731, + "num_tokens": 697657044.0, + "step": 26873 + }, + { + "epoch": 2.2160468376350293, + "grad_norm": 2.021946907043457, + "learning_rate": 1e-06, + "loss": 0.554, + "mean_token_accuracy": 0.8117713332176208, + "num_tokens": 697690145.0, + "step": 26874 + }, + { + "epoch": 2.2161292982600806, + "grad_norm": 2.093357563018799, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.7939571142196655, + "num_tokens": 697721719.0, + "step": 26875 + }, + { + "epoch": 2.2162117588851324, + "grad_norm": 2.038668155670166, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7387697696685791, + "num_tokens": 697755382.0, + "step": 26876 + }, + { + "epoch": 2.216294219510184, + "grad_norm": 2.0012242794036865, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.7960081100463867, + "num_tokens": 697788163.0, + "step": 26877 + }, + { + "epoch": 2.2163766801352356, + "grad_norm": 2.155090093612671, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7555499076843262, + "num_tokens": 697816613.0, + "step": 26878 + }, + { + "epoch": 2.216459140760287, + "grad_norm": 2.233287811279297, + "learning_rate": 1e-06, + "loss": 0.7017, + "mean_token_accuracy": 0.7751718163490295, + "num_tokens": 697844303.0, + "step": 26879 + }, + { + "epoch": 2.2165416013853383, + "grad_norm": 2.3572545051574707, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7731364965438843, + "num_tokens": 697869215.0, + "step": 26880 + }, + { + "epoch": 2.21662406201039, + "grad_norm": 2.097827911376953, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7659870386123657, + "num_tokens": 697899085.0, + "step": 26881 + }, + { + "epoch": 2.2167065226354414, + "grad_norm": 1.7923521995544434, + "learning_rate": 1e-06, + "loss": 0.5023, + "mean_token_accuracy": 0.8153884410858154, + "num_tokens": 697935552.0, + "step": 26882 + }, + { + "epoch": 2.2167889832604932, + "grad_norm": 2.0849428176879883, + "learning_rate": 1e-06, + "loss": 0.5726, + "mean_token_accuracy": 0.8056464195251465, + "num_tokens": 697964806.0, + "step": 26883 + }, + { + "epoch": 2.2168714438855446, + "grad_norm": 2.266914129257202, + "learning_rate": 1e-06, + "loss": 0.5713, + "mean_token_accuracy": 0.8090874552726746, + "num_tokens": 697990400.0, + "step": 26884 + }, + { + "epoch": 2.2169539045105964, + "grad_norm": 2.425180435180664, + "learning_rate": 1e-06, + "loss": 0.6034, + "mean_token_accuracy": 0.797866940498352, + "num_tokens": 698024458.0, + "step": 26885 + }, + { + "epoch": 2.2170363651356477, + "grad_norm": 1.96022367477417, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7718631029129028, + "num_tokens": 698056449.0, + "step": 26886 + }, + { + "epoch": 2.217118825760699, + "grad_norm": 2.2278947830200195, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.7948354482650757, + "num_tokens": 698081617.0, + "step": 26887 + }, + { + "epoch": 2.217201286385751, + "grad_norm": 1.9083824157714844, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7884395122528076, + "num_tokens": 698113740.0, + "step": 26888 + }, + { + "epoch": 2.217283747010802, + "grad_norm": 2.0993833541870117, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.783268928527832, + "num_tokens": 698142996.0, + "step": 26889 + }, + { + "epoch": 2.217366207635854, + "grad_norm": 2.0108940601348877, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7390943765640259, + "num_tokens": 698175850.0, + "step": 26890 + }, + { + "epoch": 2.2174486682609054, + "grad_norm": 2.276115655899048, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7226110696792603, + "num_tokens": 698201993.0, + "step": 26891 + }, + { + "epoch": 2.217531128885957, + "grad_norm": 2.2340972423553467, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7655496597290039, + "num_tokens": 698230285.0, + "step": 26892 + }, + { + "epoch": 2.2176135895110085, + "grad_norm": 2.3480610847473145, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7831827402114868, + "num_tokens": 698259060.0, + "step": 26893 + }, + { + "epoch": 2.21769605013606, + "grad_norm": 2.0513787269592285, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.7837222814559937, + "num_tokens": 698292471.0, + "step": 26894 + }, + { + "epoch": 2.2177785107611117, + "grad_norm": 2.2155518531799316, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7473093271255493, + "num_tokens": 698322370.0, + "step": 26895 + }, + { + "epoch": 2.217860971386163, + "grad_norm": 2.0565543174743652, + "learning_rate": 1e-06, + "loss": 0.5248, + "mean_token_accuracy": 0.8242899775505066, + "num_tokens": 698353049.0, + "step": 26896 + }, + { + "epoch": 2.217943432011215, + "grad_norm": 1.923307180404663, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7681968808174133, + "num_tokens": 698389525.0, + "step": 26897 + }, + { + "epoch": 2.218025892636266, + "grad_norm": 2.2462568283081055, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7496223449707031, + "num_tokens": 698415995.0, + "step": 26898 + }, + { + "epoch": 2.218108353261318, + "grad_norm": 2.4196560382843018, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7618663311004639, + "num_tokens": 698443645.0, + "step": 26899 + }, + { + "epoch": 2.2181908138863693, + "grad_norm": 2.1920604705810547, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7551218867301941, + "num_tokens": 698474498.0, + "step": 26900 + }, + { + "epoch": 2.2182732745114206, + "grad_norm": 2.278926372528076, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.7891384959220886, + "num_tokens": 698500860.0, + "step": 26901 + }, + { + "epoch": 2.2183557351364724, + "grad_norm": 2.42621111869812, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7689650058746338, + "num_tokens": 698525744.0, + "step": 26902 + }, + { + "epoch": 2.218438195761524, + "grad_norm": 2.168529510498047, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7676081657409668, + "num_tokens": 698557152.0, + "step": 26903 + }, + { + "epoch": 2.2185206563865756, + "grad_norm": 2.2572593688964844, + "learning_rate": 1e-06, + "loss": 0.765, + "mean_token_accuracy": 0.7554243206977844, + "num_tokens": 698585956.0, + "step": 26904 + }, + { + "epoch": 2.218603117011627, + "grad_norm": 1.9432775974273682, + "learning_rate": 1e-06, + "loss": 0.5708, + "mean_token_accuracy": 0.809154212474823, + "num_tokens": 698616966.0, + "step": 26905 + }, + { + "epoch": 2.2186855776366783, + "grad_norm": 2.0803887844085693, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.7994609475135803, + "num_tokens": 698647791.0, + "step": 26906 + }, + { + "epoch": 2.21876803826173, + "grad_norm": 2.1953229904174805, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7491940855979919, + "num_tokens": 698674648.0, + "step": 26907 + }, + { + "epoch": 2.2188504988867814, + "grad_norm": 2.1375725269317627, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7727532982826233, + "num_tokens": 698705740.0, + "step": 26908 + }, + { + "epoch": 2.2189329595118332, + "grad_norm": 2.480715036392212, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7461061477661133, + "num_tokens": 698728808.0, + "step": 26909 + }, + { + "epoch": 2.2190154201368846, + "grad_norm": 1.9425626993179321, + "learning_rate": 1e-06, + "loss": 0.5774, + "mean_token_accuracy": 0.8025311827659607, + "num_tokens": 698763940.0, + "step": 26910 + }, + { + "epoch": 2.2190978807619364, + "grad_norm": 1.8086227178573608, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7644491791725159, + "num_tokens": 698800489.0, + "step": 26911 + }, + { + "epoch": 2.2191803413869877, + "grad_norm": 1.9641973972320557, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7742811441421509, + "num_tokens": 698832935.0, + "step": 26912 + }, + { + "epoch": 2.219262802012039, + "grad_norm": 2.1089727878570557, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7803915739059448, + "num_tokens": 698861235.0, + "step": 26913 + }, + { + "epoch": 2.219345262637091, + "grad_norm": 2.0792505741119385, + "learning_rate": 1e-06, + "loss": 0.5889, + "mean_token_accuracy": 0.8051750063896179, + "num_tokens": 698890615.0, + "step": 26914 + }, + { + "epoch": 2.219427723262142, + "grad_norm": 2.0940439701080322, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7305654883384705, + "num_tokens": 698923801.0, + "step": 26915 + }, + { + "epoch": 2.219510183887194, + "grad_norm": 2.1895861625671387, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.75356525182724, + "num_tokens": 698950925.0, + "step": 26916 + }, + { + "epoch": 2.2195926445122454, + "grad_norm": 2.003847122192383, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7804478406906128, + "num_tokens": 698982623.0, + "step": 26917 + }, + { + "epoch": 2.2196751051372967, + "grad_norm": 2.087474822998047, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.789186418056488, + "num_tokens": 699011439.0, + "step": 26918 + }, + { + "epoch": 2.2197575657623485, + "grad_norm": 1.9895470142364502, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.8055288791656494, + "num_tokens": 699041172.0, + "step": 26919 + }, + { + "epoch": 2.2198400263874, + "grad_norm": 2.004953384399414, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.771481454372406, + "num_tokens": 699071156.0, + "step": 26920 + }, + { + "epoch": 2.2199224870124517, + "grad_norm": 2.1500461101531982, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.7480534911155701, + "num_tokens": 699102788.0, + "step": 26921 + }, + { + "epoch": 2.220004947637503, + "grad_norm": 2.004345178604126, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7655861377716064, + "num_tokens": 699135793.0, + "step": 26922 + }, + { + "epoch": 2.220087408262555, + "grad_norm": 2.3207924365997314, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7727832794189453, + "num_tokens": 699161404.0, + "step": 26923 + }, + { + "epoch": 2.220169868887606, + "grad_norm": 2.13553524017334, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7628368139266968, + "num_tokens": 699188166.0, + "step": 26924 + }, + { + "epoch": 2.2202523295126575, + "grad_norm": 2.438973903656006, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7656295299530029, + "num_tokens": 699212168.0, + "step": 26925 + }, + { + "epoch": 2.2203347901377093, + "grad_norm": 2.15305757522583, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7615975141525269, + "num_tokens": 699241866.0, + "step": 26926 + }, + { + "epoch": 2.2204172507627606, + "grad_norm": 2.2092950344085693, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.7799485921859741, + "num_tokens": 699269566.0, + "step": 26927 + }, + { + "epoch": 2.2204997113878124, + "grad_norm": 2.244424343109131, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.7787760496139526, + "num_tokens": 699295081.0, + "step": 26928 + }, + { + "epoch": 2.220582172012864, + "grad_norm": 1.9978641271591187, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.7937185168266296, + "num_tokens": 699327723.0, + "step": 26929 + }, + { + "epoch": 2.2206646326379156, + "grad_norm": 2.0448708534240723, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7545156478881836, + "num_tokens": 699359899.0, + "step": 26930 + }, + { + "epoch": 2.220747093262967, + "grad_norm": 2.0910866260528564, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.7912648916244507, + "num_tokens": 699384621.0, + "step": 26931 + }, + { + "epoch": 2.2208295538880183, + "grad_norm": 2.129206657409668, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7577605843544006, + "num_tokens": 699414437.0, + "step": 26932 + }, + { + "epoch": 2.22091201451307, + "grad_norm": 2.4419126510620117, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7508835196495056, + "num_tokens": 699439422.0, + "step": 26933 + }, + { + "epoch": 2.2209944751381214, + "grad_norm": 2.3818070888519287, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7511860728263855, + "num_tokens": 699464681.0, + "step": 26934 + }, + { + "epoch": 2.221076935763173, + "grad_norm": 1.9202133417129517, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7707255482673645, + "num_tokens": 699500171.0, + "step": 26935 + }, + { + "epoch": 2.2211593963882246, + "grad_norm": 2.1115164756774902, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7774418592453003, + "num_tokens": 699529606.0, + "step": 26936 + }, + { + "epoch": 2.2212418570132764, + "grad_norm": 2.009564161300659, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7799609899520874, + "num_tokens": 699560501.0, + "step": 26937 + }, + { + "epoch": 2.2213243176383277, + "grad_norm": 1.7336252927780151, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7666240930557251, + "num_tokens": 699598539.0, + "step": 26938 + }, + { + "epoch": 2.221406778263379, + "grad_norm": 2.1905200481414795, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.7789850234985352, + "num_tokens": 699626260.0, + "step": 26939 + }, + { + "epoch": 2.221489238888431, + "grad_norm": 2.046895742416382, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7489734292030334, + "num_tokens": 699658420.0, + "step": 26940 + }, + { + "epoch": 2.221571699513482, + "grad_norm": 1.9249686002731323, + "learning_rate": 1e-06, + "loss": 0.6498, + "mean_token_accuracy": 0.7809901237487793, + "num_tokens": 699692168.0, + "step": 26941 + }, + { + "epoch": 2.221654160138534, + "grad_norm": 1.8481143712997437, + "learning_rate": 1e-06, + "loss": 0.594, + "mean_token_accuracy": 0.8024810552597046, + "num_tokens": 699727826.0, + "step": 26942 + }, + { + "epoch": 2.2217366207635854, + "grad_norm": 2.1324031352996826, + "learning_rate": 1e-06, + "loss": 0.5936, + "mean_token_accuracy": 0.8007425665855408, + "num_tokens": 699756089.0, + "step": 26943 + }, + { + "epoch": 2.221819081388637, + "grad_norm": 2.210975408554077, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7469011545181274, + "num_tokens": 699784495.0, + "step": 26944 + }, + { + "epoch": 2.2219015420136885, + "grad_norm": 2.4909439086914062, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7408904433250427, + "num_tokens": 699806908.0, + "step": 26945 + }, + { + "epoch": 2.22198400263874, + "grad_norm": 2.0280649662017822, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.7923233509063721, + "num_tokens": 699837676.0, + "step": 26946 + }, + { + "epoch": 2.2220664632637916, + "grad_norm": 2.2376322746276855, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7716169357299805, + "num_tokens": 699864339.0, + "step": 26947 + }, + { + "epoch": 2.222148923888843, + "grad_norm": 2.48750901222229, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7565882802009583, + "num_tokens": 699888654.0, + "step": 26948 + }, + { + "epoch": 2.222231384513895, + "grad_norm": 2.5040338039398193, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.7839446663856506, + "num_tokens": 699911282.0, + "step": 26949 + }, + { + "epoch": 2.222313845138946, + "grad_norm": 1.9224703311920166, + "learning_rate": 1e-06, + "loss": 0.4748, + "mean_token_accuracy": 0.8397189378738403, + "num_tokens": 699942017.0, + "step": 26950 + }, + { + "epoch": 2.2223963057639975, + "grad_norm": 1.9116219282150269, + "learning_rate": 1e-06, + "loss": 0.5805, + "mean_token_accuracy": 0.807508647441864, + "num_tokens": 699974869.0, + "step": 26951 + }, + { + "epoch": 2.2224787663890493, + "grad_norm": 2.379326105117798, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.7876795530319214, + "num_tokens": 700001675.0, + "step": 26952 + }, + { + "epoch": 2.2225612270141006, + "grad_norm": 2.0420656204223633, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7474332451820374, + "num_tokens": 700037807.0, + "step": 26953 + }, + { + "epoch": 2.2226436876391524, + "grad_norm": 1.9564433097839355, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7691746950149536, + "num_tokens": 700071469.0, + "step": 26954 + }, + { + "epoch": 2.222726148264204, + "grad_norm": 1.8255082368850708, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.7949850559234619, + "num_tokens": 700108997.0, + "step": 26955 + }, + { + "epoch": 2.2228086088892556, + "grad_norm": 2.512479066848755, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.755308210849762, + "num_tokens": 700134684.0, + "step": 26956 + }, + { + "epoch": 2.222891069514307, + "grad_norm": 2.403555393218994, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7383360862731934, + "num_tokens": 700160290.0, + "step": 26957 + }, + { + "epoch": 2.2229735301393583, + "grad_norm": 2.1213767528533936, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7760411500930786, + "num_tokens": 700189642.0, + "step": 26958 + }, + { + "epoch": 2.22305599076441, + "grad_norm": 1.9954948425292969, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.777894139289856, + "num_tokens": 700223831.0, + "step": 26959 + }, + { + "epoch": 2.2231384513894614, + "grad_norm": 2.0218584537506104, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7653355598449707, + "num_tokens": 700252743.0, + "step": 26960 + }, + { + "epoch": 2.223220912014513, + "grad_norm": 2.3310723304748535, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7645865678787231, + "num_tokens": 700277589.0, + "step": 26961 + }, + { + "epoch": 2.2233033726395646, + "grad_norm": 1.7897870540618896, + "learning_rate": 1e-06, + "loss": 0.546, + "mean_token_accuracy": 0.8168757557868958, + "num_tokens": 700315220.0, + "step": 26962 + }, + { + "epoch": 2.223385833264616, + "grad_norm": 2.1410598754882812, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.7928804159164429, + "num_tokens": 700343268.0, + "step": 26963 + }, + { + "epoch": 2.2234682938896677, + "grad_norm": 2.1894898414611816, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7686102390289307, + "num_tokens": 700368776.0, + "step": 26964 + }, + { + "epoch": 2.223550754514719, + "grad_norm": 1.731107234954834, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7801170349121094, + "num_tokens": 700409901.0, + "step": 26965 + }, + { + "epoch": 2.223633215139771, + "grad_norm": 1.903432846069336, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7656816840171814, + "num_tokens": 700442918.0, + "step": 26966 + }, + { + "epoch": 2.223715675764822, + "grad_norm": 2.1322455406188965, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7749887704849243, + "num_tokens": 700473456.0, + "step": 26967 + }, + { + "epoch": 2.223798136389874, + "grad_norm": 2.0479233264923096, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7747567892074585, + "num_tokens": 700504514.0, + "step": 26968 + }, + { + "epoch": 2.2238805970149254, + "grad_norm": 2.249652862548828, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.7843306660652161, + "num_tokens": 700531318.0, + "step": 26969 + }, + { + "epoch": 2.2239630576399767, + "grad_norm": 2.0027482509613037, + "learning_rate": 1e-06, + "loss": 0.811, + "mean_token_accuracy": 0.7358495593070984, + "num_tokens": 700565774.0, + "step": 26970 + }, + { + "epoch": 2.2240455182650285, + "grad_norm": 2.2258222103118896, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.7822238802909851, + "num_tokens": 700592894.0, + "step": 26971 + }, + { + "epoch": 2.22412797889008, + "grad_norm": 2.0040700435638428, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.768405556678772, + "num_tokens": 700624732.0, + "step": 26972 + }, + { + "epoch": 2.2242104395151316, + "grad_norm": 2.172255039215088, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.8041275143623352, + "num_tokens": 700652955.0, + "step": 26973 + }, + { + "epoch": 2.224292900140183, + "grad_norm": 2.0093307495117188, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7684081792831421, + "num_tokens": 700685878.0, + "step": 26974 + }, + { + "epoch": 2.224375360765235, + "grad_norm": 1.9968420267105103, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.767827033996582, + "num_tokens": 700719670.0, + "step": 26975 + }, + { + "epoch": 2.224457821390286, + "grad_norm": 2.9456942081451416, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7828598618507385, + "num_tokens": 700738599.0, + "step": 26976 + }, + { + "epoch": 2.2245402820153375, + "grad_norm": 2.1950173377990723, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.781768798828125, + "num_tokens": 700764480.0, + "step": 26977 + }, + { + "epoch": 2.2246227426403893, + "grad_norm": 2.147423028945923, + "learning_rate": 1e-06, + "loss": 0.6001, + "mean_token_accuracy": 0.800601065158844, + "num_tokens": 700793408.0, + "step": 26978 + }, + { + "epoch": 2.2247052032654406, + "grad_norm": 2.204955577850342, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.7492156028747559, + "num_tokens": 700823440.0, + "step": 26979 + }, + { + "epoch": 2.2247876638904924, + "grad_norm": 2.149216651916504, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.7575347423553467, + "num_tokens": 700853823.0, + "step": 26980 + }, + { + "epoch": 2.224870124515544, + "grad_norm": 2.4947612285614014, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7487006187438965, + "num_tokens": 700878241.0, + "step": 26981 + }, + { + "epoch": 2.2249525851405956, + "grad_norm": 2.031970262527466, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7448335289955139, + "num_tokens": 700910370.0, + "step": 26982 + }, + { + "epoch": 2.225035045765647, + "grad_norm": 1.7609893083572388, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.7549498081207275, + "num_tokens": 700949953.0, + "step": 26983 + }, + { + "epoch": 2.2251175063906983, + "grad_norm": 2.072523593902588, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.786960244178772, + "num_tokens": 700978153.0, + "step": 26984 + }, + { + "epoch": 2.22519996701575, + "grad_norm": 2.2590363025665283, + "learning_rate": 1e-06, + "loss": 0.7303, + "mean_token_accuracy": 0.7593677639961243, + "num_tokens": 701006993.0, + "step": 26985 + }, + { + "epoch": 2.2252824276408014, + "grad_norm": 2.321958541870117, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.795792818069458, + "num_tokens": 701031748.0, + "step": 26986 + }, + { + "epoch": 2.225364888265853, + "grad_norm": 2.048312187194824, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7446184158325195, + "num_tokens": 701063971.0, + "step": 26987 + }, + { + "epoch": 2.2254473488909046, + "grad_norm": 2.013892412185669, + "learning_rate": 1e-06, + "loss": 0.5732, + "mean_token_accuracy": 0.7997205257415771, + "num_tokens": 701094645.0, + "step": 26988 + }, + { + "epoch": 2.2255298095159564, + "grad_norm": 2.2195286750793457, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7712218761444092, + "num_tokens": 701123822.0, + "step": 26989 + }, + { + "epoch": 2.2256122701410077, + "grad_norm": 2.086940050125122, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7517135143280029, + "num_tokens": 701155187.0, + "step": 26990 + }, + { + "epoch": 2.225694730766059, + "grad_norm": 2.190934181213379, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.7798614501953125, + "num_tokens": 701183791.0, + "step": 26991 + }, + { + "epoch": 2.225777191391111, + "grad_norm": 2.141946792602539, + "learning_rate": 1e-06, + "loss": 0.544, + "mean_token_accuracy": 0.818947434425354, + "num_tokens": 701210603.0, + "step": 26992 + }, + { + "epoch": 2.225859652016162, + "grad_norm": 2.1333673000335693, + "learning_rate": 1e-06, + "loss": 0.4633, + "mean_token_accuracy": 0.8302909135818481, + "num_tokens": 701235862.0, + "step": 26993 + }, + { + "epoch": 2.225942112641214, + "grad_norm": 2.088104248046875, + "learning_rate": 1e-06, + "loss": 0.6047, + "mean_token_accuracy": 0.796705424785614, + "num_tokens": 701263587.0, + "step": 26994 + }, + { + "epoch": 2.2260245732662653, + "grad_norm": 1.7893295288085938, + "learning_rate": 1e-06, + "loss": 0.6193, + "mean_token_accuracy": 0.8008935451507568, + "num_tokens": 701301705.0, + "step": 26995 + }, + { + "epoch": 2.2261070338913167, + "grad_norm": 2.112382411956787, + "learning_rate": 1e-06, + "loss": 0.5779, + "mean_token_accuracy": 0.803009033203125, + "num_tokens": 701330609.0, + "step": 26996 + }, + { + "epoch": 2.2261894945163685, + "grad_norm": 2.286489963531494, + "learning_rate": 1e-06, + "loss": 0.5776, + "mean_token_accuracy": 0.8010327816009521, + "num_tokens": 701353401.0, + "step": 26997 + }, + { + "epoch": 2.22627195514142, + "grad_norm": 1.9734618663787842, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7512820959091187, + "num_tokens": 701387312.0, + "step": 26998 + }, + { + "epoch": 2.2263544157664716, + "grad_norm": 2.0506341457366943, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.7933560609817505, + "num_tokens": 701420413.0, + "step": 26999 + }, + { + "epoch": 2.226436876391523, + "grad_norm": 2.0423836708068848, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.7799286842346191, + "num_tokens": 701452621.0, + "step": 27000 + }, + { + "epoch": 2.226519337016575, + "grad_norm": 1.995002269744873, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7457422018051147, + "num_tokens": 701488933.0, + "step": 27001 + }, + { + "epoch": 2.226601797641626, + "grad_norm": 2.1479740142822266, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.7788543701171875, + "num_tokens": 701518603.0, + "step": 27002 + }, + { + "epoch": 2.2266842582666775, + "grad_norm": 2.234175443649292, + "learning_rate": 1e-06, + "loss": 0.5657, + "mean_token_accuracy": 0.8104139566421509, + "num_tokens": 701545133.0, + "step": 27003 + }, + { + "epoch": 2.2267667188917293, + "grad_norm": 2.2379090785980225, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7765386700630188, + "num_tokens": 701573656.0, + "step": 27004 + }, + { + "epoch": 2.2268491795167806, + "grad_norm": 1.923427939414978, + "learning_rate": 1e-06, + "loss": 0.6089, + "mean_token_accuracy": 0.7944371700286865, + "num_tokens": 701606678.0, + "step": 27005 + }, + { + "epoch": 2.2269316401418324, + "grad_norm": 2.1979100704193115, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7401151657104492, + "num_tokens": 701635646.0, + "step": 27006 + }, + { + "epoch": 2.2270141007668838, + "grad_norm": 2.0501022338867188, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.757723331451416, + "num_tokens": 701668511.0, + "step": 27007 + }, + { + "epoch": 2.227096561391935, + "grad_norm": 2.2798080444335938, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7641240954399109, + "num_tokens": 701697035.0, + "step": 27008 + }, + { + "epoch": 2.227179022016987, + "grad_norm": 2.1407220363616943, + "learning_rate": 1e-06, + "loss": 0.5276, + "mean_token_accuracy": 0.8226356506347656, + "num_tokens": 701722145.0, + "step": 27009 + }, + { + "epoch": 2.2272614826420383, + "grad_norm": 1.8936675786972046, + "learning_rate": 1e-06, + "loss": 0.5816, + "mean_token_accuracy": 0.8039023280143738, + "num_tokens": 701757283.0, + "step": 27010 + }, + { + "epoch": 2.22734394326709, + "grad_norm": 1.9607607126235962, + "learning_rate": 1e-06, + "loss": 0.5768, + "mean_token_accuracy": 0.8105548620223999, + "num_tokens": 701788376.0, + "step": 27011 + }, + { + "epoch": 2.2274264038921414, + "grad_norm": 2.3549177646636963, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7825145721435547, + "num_tokens": 701813575.0, + "step": 27012 + }, + { + "epoch": 2.227508864517193, + "grad_norm": 2.419379234313965, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7626127004623413, + "num_tokens": 701837677.0, + "step": 27013 + }, + { + "epoch": 2.2275913251422446, + "grad_norm": 2.234799861907959, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7661694288253784, + "num_tokens": 701863316.0, + "step": 27014 + }, + { + "epoch": 2.227673785767296, + "grad_norm": 1.9406328201293945, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7642232179641724, + "num_tokens": 701898945.0, + "step": 27015 + }, + { + "epoch": 2.2277562463923477, + "grad_norm": 1.9916530847549438, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.7877345085144043, + "num_tokens": 701932720.0, + "step": 27016 + }, + { + "epoch": 2.227838707017399, + "grad_norm": 2.2608728408813477, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7509243488311768, + "num_tokens": 701960774.0, + "step": 27017 + }, + { + "epoch": 2.227921167642451, + "grad_norm": 2.325603485107422, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.767945408821106, + "num_tokens": 701984752.0, + "step": 27018 + }, + { + "epoch": 2.228003628267502, + "grad_norm": 2.004352331161499, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.782393753528595, + "num_tokens": 702018314.0, + "step": 27019 + }, + { + "epoch": 2.228086088892554, + "grad_norm": 2.5173885822296143, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7751336097717285, + "num_tokens": 702038822.0, + "step": 27020 + }, + { + "epoch": 2.2281685495176053, + "grad_norm": 2.419114351272583, + "learning_rate": 1e-06, + "loss": 0.5989, + "mean_token_accuracy": 0.7940709590911865, + "num_tokens": 702062919.0, + "step": 27021 + }, + { + "epoch": 2.2282510101426567, + "grad_norm": 2.0664725303649902, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7684620022773743, + "num_tokens": 702092466.0, + "step": 27022 + }, + { + "epoch": 2.2283334707677085, + "grad_norm": 2.0349998474121094, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7914666533470154, + "num_tokens": 702121968.0, + "step": 27023 + }, + { + "epoch": 2.22841593139276, + "grad_norm": 2.2112934589385986, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7668688297271729, + "num_tokens": 702148145.0, + "step": 27024 + }, + { + "epoch": 2.2284983920178116, + "grad_norm": 2.122260093688965, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.7907102108001709, + "num_tokens": 702177416.0, + "step": 27025 + }, + { + "epoch": 2.228580852642863, + "grad_norm": 2.087306022644043, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7638362050056458, + "num_tokens": 702206828.0, + "step": 27026 + }, + { + "epoch": 2.228663313267915, + "grad_norm": 2.2097115516662598, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7683139443397522, + "num_tokens": 702234582.0, + "step": 27027 + }, + { + "epoch": 2.228745773892966, + "grad_norm": 2.381702184677124, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7582170963287354, + "num_tokens": 702261650.0, + "step": 27028 + }, + { + "epoch": 2.2288282345180175, + "grad_norm": 2.3105814456939697, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7409865856170654, + "num_tokens": 702286562.0, + "step": 27029 + }, + { + "epoch": 2.2289106951430693, + "grad_norm": 2.0618953704833984, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.7881284952163696, + "num_tokens": 702318070.0, + "step": 27030 + }, + { + "epoch": 2.2289931557681206, + "grad_norm": 1.9306695461273193, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.782861590385437, + "num_tokens": 702353876.0, + "step": 27031 + }, + { + "epoch": 2.2290756163931724, + "grad_norm": 2.5426840782165527, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7499231100082397, + "num_tokens": 702377541.0, + "step": 27032 + }, + { + "epoch": 2.2291580770182238, + "grad_norm": 1.9942781925201416, + "learning_rate": 1e-06, + "loss": 0.5479, + "mean_token_accuracy": 0.8196537494659424, + "num_tokens": 702409085.0, + "step": 27033 + }, + { + "epoch": 2.2292405376432756, + "grad_norm": 2.2690634727478027, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.7942101955413818, + "num_tokens": 702433298.0, + "step": 27034 + }, + { + "epoch": 2.229322998268327, + "grad_norm": 2.032543420791626, + "learning_rate": 1e-06, + "loss": 0.7303, + "mean_token_accuracy": 0.7666369676589966, + "num_tokens": 702466604.0, + "step": 27035 + }, + { + "epoch": 2.2294054588933783, + "grad_norm": 2.200777292251587, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7376682758331299, + "num_tokens": 702497279.0, + "step": 27036 + }, + { + "epoch": 2.22948791951843, + "grad_norm": 2.3079724311828613, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7265893220901489, + "num_tokens": 702526570.0, + "step": 27037 + }, + { + "epoch": 2.2295703801434814, + "grad_norm": 2.0184056758880615, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.7909979820251465, + "num_tokens": 702555796.0, + "step": 27038 + }, + { + "epoch": 2.229652840768533, + "grad_norm": 2.503955364227295, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7582052946090698, + "num_tokens": 702578806.0, + "step": 27039 + }, + { + "epoch": 2.2297353013935846, + "grad_norm": 2.0839085578918457, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.778052031993866, + "num_tokens": 702608153.0, + "step": 27040 + }, + { + "epoch": 2.229817762018636, + "grad_norm": 2.0554232597351074, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7489691972732544, + "num_tokens": 702645125.0, + "step": 27041 + }, + { + "epoch": 2.2299002226436877, + "grad_norm": 2.2248706817626953, + "learning_rate": 1e-06, + "loss": 0.6131, + "mean_token_accuracy": 0.7895351052284241, + "num_tokens": 702670504.0, + "step": 27042 + }, + { + "epoch": 2.229982683268739, + "grad_norm": 2.017300844192505, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.784977912902832, + "num_tokens": 702699880.0, + "step": 27043 + }, + { + "epoch": 2.230065143893791, + "grad_norm": 1.9961824417114258, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7271752953529358, + "num_tokens": 702731826.0, + "step": 27044 + }, + { + "epoch": 2.230147604518842, + "grad_norm": 1.792959451675415, + "learning_rate": 1e-06, + "loss": 0.6183, + "mean_token_accuracy": 0.8012300729751587, + "num_tokens": 702765534.0, + "step": 27045 + }, + { + "epoch": 2.230230065143894, + "grad_norm": 2.00447416305542, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7757192254066467, + "num_tokens": 702800872.0, + "step": 27046 + }, + { + "epoch": 2.2303125257689453, + "grad_norm": 2.037292003631592, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7700774669647217, + "num_tokens": 702833597.0, + "step": 27047 + }, + { + "epoch": 2.2303949863939967, + "grad_norm": 2.0528934001922607, + "learning_rate": 1e-06, + "loss": 0.5892, + "mean_token_accuracy": 0.8063913583755493, + "num_tokens": 702865431.0, + "step": 27048 + }, + { + "epoch": 2.2304774470190485, + "grad_norm": 2.254899024963379, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7399875521659851, + "num_tokens": 702891656.0, + "step": 27049 + }, + { + "epoch": 2.2305599076441, + "grad_norm": 2.1493771076202393, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7780637145042419, + "num_tokens": 702921788.0, + "step": 27050 + }, + { + "epoch": 2.2306423682691516, + "grad_norm": 2.4749155044555664, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.75800621509552, + "num_tokens": 702945717.0, + "step": 27051 + }, + { + "epoch": 2.230724828894203, + "grad_norm": 1.8972002267837524, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7789410352706909, + "num_tokens": 702982547.0, + "step": 27052 + }, + { + "epoch": 2.2308072895192543, + "grad_norm": 1.8434234857559204, + "learning_rate": 1e-06, + "loss": 0.5112, + "mean_token_accuracy": 0.8310114145278931, + "num_tokens": 703018260.0, + "step": 27053 + }, + { + "epoch": 2.230889750144306, + "grad_norm": 1.918747067451477, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7465082406997681, + "num_tokens": 703053948.0, + "step": 27054 + }, + { + "epoch": 2.2309722107693575, + "grad_norm": 1.7973440885543823, + "learning_rate": 1e-06, + "loss": 0.6112, + "mean_token_accuracy": 0.797869086265564, + "num_tokens": 703089912.0, + "step": 27055 + }, + { + "epoch": 2.2310546713944093, + "grad_norm": 2.0252699851989746, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.7892459630966187, + "num_tokens": 703123013.0, + "step": 27056 + }, + { + "epoch": 2.2311371320194606, + "grad_norm": 2.158271551132202, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.7982450723648071, + "num_tokens": 703149742.0, + "step": 27057 + }, + { + "epoch": 2.2312195926445124, + "grad_norm": 2.4407272338867188, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.781578004360199, + "num_tokens": 703171371.0, + "step": 27058 + }, + { + "epoch": 2.2313020532695638, + "grad_norm": 2.174072027206421, + "learning_rate": 1e-06, + "loss": 0.573, + "mean_token_accuracy": 0.8087605237960815, + "num_tokens": 703198440.0, + "step": 27059 + }, + { + "epoch": 2.231384513894615, + "grad_norm": 2.0332372188568115, + "learning_rate": 1e-06, + "loss": 0.5795, + "mean_token_accuracy": 0.8058037757873535, + "num_tokens": 703226259.0, + "step": 27060 + }, + { + "epoch": 2.231466974519667, + "grad_norm": 2.0059807300567627, + "learning_rate": 1e-06, + "loss": 0.612, + "mean_token_accuracy": 0.8006296753883362, + "num_tokens": 703257461.0, + "step": 27061 + }, + { + "epoch": 2.2315494351447183, + "grad_norm": 2.003146171569824, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.7912669777870178, + "num_tokens": 703291143.0, + "step": 27062 + }, + { + "epoch": 2.23163189576977, + "grad_norm": 2.1362125873565674, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.781684398651123, + "num_tokens": 703323056.0, + "step": 27063 + }, + { + "epoch": 2.2317143563948214, + "grad_norm": 2.036719560623169, + "learning_rate": 1e-06, + "loss": 0.5676, + "mean_token_accuracy": 0.791093111038208, + "num_tokens": 703354841.0, + "step": 27064 + }, + { + "epoch": 2.231796817019873, + "grad_norm": 2.313995599746704, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7558494806289673, + "num_tokens": 703379632.0, + "step": 27065 + }, + { + "epoch": 2.2318792776449246, + "grad_norm": 2.4667468070983887, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7358929514884949, + "num_tokens": 703402721.0, + "step": 27066 + }, + { + "epoch": 2.231961738269976, + "grad_norm": 2.2703428268432617, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7857812643051147, + "num_tokens": 703428325.0, + "step": 27067 + }, + { + "epoch": 2.2320441988950277, + "grad_norm": 2.1591861248016357, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7730473875999451, + "num_tokens": 703456788.0, + "step": 27068 + }, + { + "epoch": 2.232126659520079, + "grad_norm": 2.187732219696045, + "learning_rate": 1e-06, + "loss": 0.6072, + "mean_token_accuracy": 0.8074306845664978, + "num_tokens": 703482618.0, + "step": 27069 + }, + { + "epoch": 2.232209120145131, + "grad_norm": 2.1818606853485107, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.770835816860199, + "num_tokens": 703514017.0, + "step": 27070 + }, + { + "epoch": 2.232291580770182, + "grad_norm": 2.3289601802825928, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7330292463302612, + "num_tokens": 703541374.0, + "step": 27071 + }, + { + "epoch": 2.232374041395234, + "grad_norm": 2.1982405185699463, + "learning_rate": 1e-06, + "loss": 0.6101, + "mean_token_accuracy": 0.7932608723640442, + "num_tokens": 703568231.0, + "step": 27072 + }, + { + "epoch": 2.2324565020202853, + "grad_norm": 1.919693112373352, + "learning_rate": 1e-06, + "loss": 0.5213, + "mean_token_accuracy": 0.8241109251976013, + "num_tokens": 703600076.0, + "step": 27073 + }, + { + "epoch": 2.2325389626453367, + "grad_norm": 1.964691162109375, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7630972266197205, + "num_tokens": 703635375.0, + "step": 27074 + }, + { + "epoch": 2.2326214232703885, + "grad_norm": 2.146697521209717, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.753147304058075, + "num_tokens": 703666736.0, + "step": 27075 + }, + { + "epoch": 2.23270388389544, + "grad_norm": 2.0354928970336914, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7733016014099121, + "num_tokens": 703696641.0, + "step": 27076 + }, + { + "epoch": 2.2327863445204916, + "grad_norm": 2.0482113361358643, + "learning_rate": 1e-06, + "loss": 0.6135, + "mean_token_accuracy": 0.7844713926315308, + "num_tokens": 703725943.0, + "step": 27077 + }, + { + "epoch": 2.232868805145543, + "grad_norm": 2.046006202697754, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7793300151824951, + "num_tokens": 703757997.0, + "step": 27078 + }, + { + "epoch": 2.2329512657705948, + "grad_norm": 2.408303737640381, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7530627250671387, + "num_tokens": 703784226.0, + "step": 27079 + }, + { + "epoch": 2.233033726395646, + "grad_norm": 2.2319843769073486, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7764304876327515, + "num_tokens": 703811127.0, + "step": 27080 + }, + { + "epoch": 2.2331161870206975, + "grad_norm": 2.0204918384552, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7782574892044067, + "num_tokens": 703842696.0, + "step": 27081 + }, + { + "epoch": 2.2331986476457493, + "grad_norm": 2.1088223457336426, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7576979398727417, + "num_tokens": 703871621.0, + "step": 27082 + }, + { + "epoch": 2.2332811082708006, + "grad_norm": 2.6754579544067383, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7543027400970459, + "num_tokens": 703894240.0, + "step": 27083 + }, + { + "epoch": 2.2333635688958524, + "grad_norm": 2.074019193649292, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.774664580821991, + "num_tokens": 703925302.0, + "step": 27084 + }, + { + "epoch": 2.2334460295209038, + "grad_norm": 1.8804255723953247, + "learning_rate": 1e-06, + "loss": 0.5933, + "mean_token_accuracy": 0.8072752952575684, + "num_tokens": 703959359.0, + "step": 27085 + }, + { + "epoch": 2.233528490145955, + "grad_norm": 1.9910300970077515, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7777619957923889, + "num_tokens": 703989572.0, + "step": 27086 + }, + { + "epoch": 2.233610950771007, + "grad_norm": 1.9548536539077759, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7594711780548096, + "num_tokens": 704022542.0, + "step": 27087 + }, + { + "epoch": 2.2336934113960583, + "grad_norm": 1.7733606100082397, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.7832095623016357, + "num_tokens": 704061419.0, + "step": 27088 + }, + { + "epoch": 2.23377587202111, + "grad_norm": 1.900915503501892, + "learning_rate": 1e-06, + "loss": 0.445, + "mean_token_accuracy": 0.8335961103439331, + "num_tokens": 704095674.0, + "step": 27089 + }, + { + "epoch": 2.2338583326461614, + "grad_norm": 1.9990177154541016, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7675641775131226, + "num_tokens": 704128340.0, + "step": 27090 + }, + { + "epoch": 2.233940793271213, + "grad_norm": 2.103649377822876, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7682528495788574, + "num_tokens": 704160921.0, + "step": 27091 + }, + { + "epoch": 2.2340232538962645, + "grad_norm": 2.380720615386963, + "learning_rate": 1e-06, + "loss": 0.7601, + "mean_token_accuracy": 0.7509670257568359, + "num_tokens": 704185449.0, + "step": 27092 + }, + { + "epoch": 2.234105714521316, + "grad_norm": 2.264200448989868, + "learning_rate": 1e-06, + "loss": 0.5882, + "mean_token_accuracy": 0.8030633330345154, + "num_tokens": 704211591.0, + "step": 27093 + }, + { + "epoch": 2.2341881751463677, + "grad_norm": 1.906414270401001, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7585119605064392, + "num_tokens": 704242484.0, + "step": 27094 + }, + { + "epoch": 2.234270635771419, + "grad_norm": 2.5061423778533936, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.7852804660797119, + "num_tokens": 704265893.0, + "step": 27095 + }, + { + "epoch": 2.234353096396471, + "grad_norm": 2.0582010746002197, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.768648087978363, + "num_tokens": 704296659.0, + "step": 27096 + }, + { + "epoch": 2.234435557021522, + "grad_norm": 2.002570390701294, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.7478867769241333, + "num_tokens": 704327789.0, + "step": 27097 + }, + { + "epoch": 2.2345180176465735, + "grad_norm": 2.219167470932007, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7765340805053711, + "num_tokens": 704356273.0, + "step": 27098 + }, + { + "epoch": 2.2346004782716253, + "grad_norm": 2.249119281768799, + "learning_rate": 1e-06, + "loss": 0.5371, + "mean_token_accuracy": 0.8152486085891724, + "num_tokens": 704380664.0, + "step": 27099 + }, + { + "epoch": 2.2346829388966767, + "grad_norm": 2.1993770599365234, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.769133448600769, + "num_tokens": 704409374.0, + "step": 27100 + }, + { + "epoch": 2.2347653995217285, + "grad_norm": 2.145948648452759, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7743523120880127, + "num_tokens": 704443142.0, + "step": 27101 + }, + { + "epoch": 2.23484786014678, + "grad_norm": 1.9927200078964233, + "learning_rate": 1e-06, + "loss": 0.5888, + "mean_token_accuracy": 0.8021515607833862, + "num_tokens": 704475135.0, + "step": 27102 + }, + { + "epoch": 2.2349303207718316, + "grad_norm": 1.873327612876892, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7491027116775513, + "num_tokens": 704511300.0, + "step": 27103 + }, + { + "epoch": 2.235012781396883, + "grad_norm": 1.9475243091583252, + "learning_rate": 1e-06, + "loss": 0.5457, + "mean_token_accuracy": 0.8183813095092773, + "num_tokens": 704542034.0, + "step": 27104 + }, + { + "epoch": 2.2350952420219343, + "grad_norm": 1.888764500617981, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7506817579269409, + "num_tokens": 704578400.0, + "step": 27105 + }, + { + "epoch": 2.235177702646986, + "grad_norm": 1.9644461870193481, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7618511915206909, + "num_tokens": 704612550.0, + "step": 27106 + }, + { + "epoch": 2.2352601632720375, + "grad_norm": 2.131605625152588, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.7989092469215393, + "num_tokens": 704643564.0, + "step": 27107 + }, + { + "epoch": 2.2353426238970893, + "grad_norm": 1.9629673957824707, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.7901866436004639, + "num_tokens": 704675479.0, + "step": 27108 + }, + { + "epoch": 2.2354250845221406, + "grad_norm": 2.1816506385803223, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.779983401298523, + "num_tokens": 704700957.0, + "step": 27109 + }, + { + "epoch": 2.2355075451471924, + "grad_norm": 2.2345731258392334, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.7880471348762512, + "num_tokens": 704728372.0, + "step": 27110 + }, + { + "epoch": 2.2355900057722438, + "grad_norm": 2.0000252723693848, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.758811354637146, + "num_tokens": 704762738.0, + "step": 27111 + }, + { + "epoch": 2.235672466397295, + "grad_norm": 2.216139316558838, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7827841639518738, + "num_tokens": 704790290.0, + "step": 27112 + }, + { + "epoch": 2.235754927022347, + "grad_norm": 1.967536449432373, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.7880774736404419, + "num_tokens": 704823472.0, + "step": 27113 + }, + { + "epoch": 2.2358373876473983, + "grad_norm": 2.1243174076080322, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.7805467844009399, + "num_tokens": 704852596.0, + "step": 27114 + }, + { + "epoch": 2.23591984827245, + "grad_norm": 2.446476697921753, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7696114778518677, + "num_tokens": 704876918.0, + "step": 27115 + }, + { + "epoch": 2.2360023088975014, + "grad_norm": 2.2054221630096436, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7814347743988037, + "num_tokens": 704904936.0, + "step": 27116 + }, + { + "epoch": 2.236084769522553, + "grad_norm": 1.9512577056884766, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7825227975845337, + "num_tokens": 704937872.0, + "step": 27117 + }, + { + "epoch": 2.2361672301476045, + "grad_norm": 2.1662638187408447, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7453762292861938, + "num_tokens": 704970261.0, + "step": 27118 + }, + { + "epoch": 2.236249690772656, + "grad_norm": 2.186292886734009, + "learning_rate": 1e-06, + "loss": 0.605, + "mean_token_accuracy": 0.7982273101806641, + "num_tokens": 704998248.0, + "step": 27119 + }, + { + "epoch": 2.2363321513977077, + "grad_norm": 2.0397801399230957, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7471625804901123, + "num_tokens": 705029170.0, + "step": 27120 + }, + { + "epoch": 2.236414612022759, + "grad_norm": 2.0455706119537354, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7582064270973206, + "num_tokens": 705062106.0, + "step": 27121 + }, + { + "epoch": 2.236497072647811, + "grad_norm": 2.3602356910705566, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.7625744342803955, + "num_tokens": 705088160.0, + "step": 27122 + }, + { + "epoch": 2.236579533272862, + "grad_norm": 1.9685271978378296, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.7925445437431335, + "num_tokens": 705119628.0, + "step": 27123 + }, + { + "epoch": 2.236661993897914, + "grad_norm": 2.1745705604553223, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7824256420135498, + "num_tokens": 705144262.0, + "step": 27124 + }, + { + "epoch": 2.2367444545229653, + "grad_norm": 2.04854154586792, + "learning_rate": 1e-06, + "loss": 0.5079, + "mean_token_accuracy": 0.8254358768463135, + "num_tokens": 705173947.0, + "step": 27125 + }, + { + "epoch": 2.2368269151480167, + "grad_norm": 1.9396659135818481, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.7936045527458191, + "num_tokens": 705208743.0, + "step": 27126 + }, + { + "epoch": 2.2369093757730685, + "grad_norm": 2.1834073066711426, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.7875583171844482, + "num_tokens": 705239428.0, + "step": 27127 + }, + { + "epoch": 2.23699183639812, + "grad_norm": 1.9059290885925293, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.7684323787689209, + "num_tokens": 705276428.0, + "step": 27128 + }, + { + "epoch": 2.2370742970231716, + "grad_norm": 2.1665291786193848, + "learning_rate": 1e-06, + "loss": 0.6072, + "mean_token_accuracy": 0.7885440587997437, + "num_tokens": 705305425.0, + "step": 27129 + }, + { + "epoch": 2.237156757648223, + "grad_norm": 2.2822189331054688, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.7811378240585327, + "num_tokens": 705333665.0, + "step": 27130 + }, + { + "epoch": 2.2372392182732743, + "grad_norm": 1.9410157203674316, + "learning_rate": 1e-06, + "loss": 0.5833, + "mean_token_accuracy": 0.7932550311088562, + "num_tokens": 705364715.0, + "step": 27131 + }, + { + "epoch": 2.237321678898326, + "grad_norm": 1.7122809886932373, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.7992709875106812, + "num_tokens": 705402635.0, + "step": 27132 + }, + { + "epoch": 2.2374041395233775, + "grad_norm": 2.8208465576171875, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7317006587982178, + "num_tokens": 705422009.0, + "step": 27133 + }, + { + "epoch": 2.2374866001484293, + "grad_norm": 2.0255465507507324, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7760809659957886, + "num_tokens": 705452012.0, + "step": 27134 + }, + { + "epoch": 2.2375690607734806, + "grad_norm": 2.111445665359497, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7707908153533936, + "num_tokens": 705481714.0, + "step": 27135 + }, + { + "epoch": 2.2376515213985324, + "grad_norm": 2.1795754432678223, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7404471635818481, + "num_tokens": 705513466.0, + "step": 27136 + }, + { + "epoch": 2.2377339820235838, + "grad_norm": 2.0803260803222656, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.773656964302063, + "num_tokens": 705545145.0, + "step": 27137 + }, + { + "epoch": 2.237816442648635, + "grad_norm": 2.189333200454712, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7742885947227478, + "num_tokens": 705572981.0, + "step": 27138 + }, + { + "epoch": 2.237898903273687, + "grad_norm": 2.194232940673828, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.7837789058685303, + "num_tokens": 705600976.0, + "step": 27139 + }, + { + "epoch": 2.2379813638987383, + "grad_norm": 2.3832011222839355, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7372475266456604, + "num_tokens": 705628037.0, + "step": 27140 + }, + { + "epoch": 2.23806382452379, + "grad_norm": 1.99070143699646, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.7864556312561035, + "num_tokens": 705664763.0, + "step": 27141 + }, + { + "epoch": 2.2381462851488414, + "grad_norm": 2.098668336868286, + "learning_rate": 1e-06, + "loss": 0.5388, + "mean_token_accuracy": 0.8198201060295105, + "num_tokens": 705691489.0, + "step": 27142 + }, + { + "epoch": 2.2382287457738927, + "grad_norm": 2.1829886436462402, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.7886872291564941, + "num_tokens": 705719167.0, + "step": 27143 + }, + { + "epoch": 2.2383112063989445, + "grad_norm": 2.323014974594116, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7252746224403381, + "num_tokens": 705746668.0, + "step": 27144 + }, + { + "epoch": 2.238393667023996, + "grad_norm": 1.9652477502822876, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.765119194984436, + "num_tokens": 705778264.0, + "step": 27145 + }, + { + "epoch": 2.2384761276490477, + "grad_norm": 2.440528154373169, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7439563274383545, + "num_tokens": 705800679.0, + "step": 27146 + }, + { + "epoch": 2.238558588274099, + "grad_norm": 1.957772970199585, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.7841407060623169, + "num_tokens": 705836104.0, + "step": 27147 + }, + { + "epoch": 2.238641048899151, + "grad_norm": 1.8856583833694458, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.7818405628204346, + "num_tokens": 705872298.0, + "step": 27148 + }, + { + "epoch": 2.238723509524202, + "grad_norm": 2.08256196975708, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7471478581428528, + "num_tokens": 705903323.0, + "step": 27149 + }, + { + "epoch": 2.2388059701492535, + "grad_norm": 2.06145977973938, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.7891467809677124, + "num_tokens": 705934565.0, + "step": 27150 + }, + { + "epoch": 2.2388884307743053, + "grad_norm": 1.924631118774414, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7701032161712646, + "num_tokens": 705971209.0, + "step": 27151 + }, + { + "epoch": 2.2389708913993567, + "grad_norm": 2.24269700050354, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7396528720855713, + "num_tokens": 705999119.0, + "step": 27152 + }, + { + "epoch": 2.2390533520244085, + "grad_norm": 2.208568572998047, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7269754409790039, + "num_tokens": 706027627.0, + "step": 27153 + }, + { + "epoch": 2.23913581264946, + "grad_norm": 1.9325165748596191, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.7913960218429565, + "num_tokens": 706059443.0, + "step": 27154 + }, + { + "epoch": 2.2392182732745116, + "grad_norm": 2.185835361480713, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7643536329269409, + "num_tokens": 706086547.0, + "step": 27155 + }, + { + "epoch": 2.239300733899563, + "grad_norm": 2.130573272705078, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7407497763633728, + "num_tokens": 706116790.0, + "step": 27156 + }, + { + "epoch": 2.2393831945246143, + "grad_norm": 2.150022268295288, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7570339441299438, + "num_tokens": 706148603.0, + "step": 27157 + }, + { + "epoch": 2.239465655149666, + "grad_norm": 1.990573525428772, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.7942090034484863, + "num_tokens": 706181526.0, + "step": 27158 + }, + { + "epoch": 2.2395481157747175, + "grad_norm": 2.0134823322296143, + "learning_rate": 1e-06, + "loss": 0.5341, + "mean_token_accuracy": 0.8145908117294312, + "num_tokens": 706209159.0, + "step": 27159 + }, + { + "epoch": 2.2396305763997693, + "grad_norm": 2.167304039001465, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7379393577575684, + "num_tokens": 706238217.0, + "step": 27160 + }, + { + "epoch": 2.2397130370248206, + "grad_norm": 2.026444911956787, + "learning_rate": 1e-06, + "loss": 0.7882, + "mean_token_accuracy": 0.7502276301383972, + "num_tokens": 706273570.0, + "step": 27161 + }, + { + "epoch": 2.2397954976498724, + "grad_norm": 2.118769884109497, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7342268228530884, + "num_tokens": 706304019.0, + "step": 27162 + }, + { + "epoch": 2.2398779582749238, + "grad_norm": 2.302675724029541, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7678616046905518, + "num_tokens": 706330398.0, + "step": 27163 + }, + { + "epoch": 2.239960418899975, + "grad_norm": 2.083069086074829, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7629923820495605, + "num_tokens": 706360052.0, + "step": 27164 + }, + { + "epoch": 2.240042879525027, + "grad_norm": 2.0601539611816406, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7411139011383057, + "num_tokens": 706397892.0, + "step": 27165 + }, + { + "epoch": 2.2401253401500782, + "grad_norm": 2.3609771728515625, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.7903876900672913, + "num_tokens": 706420547.0, + "step": 27166 + }, + { + "epoch": 2.24020780077513, + "grad_norm": 2.1989476680755615, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7672860622406006, + "num_tokens": 706449400.0, + "step": 27167 + }, + { + "epoch": 2.2402902614001814, + "grad_norm": 1.8220694065093994, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7774196863174438, + "num_tokens": 706489036.0, + "step": 27168 + }, + { + "epoch": 2.2403727220252327, + "grad_norm": 2.209573268890381, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.7707018852233887, + "num_tokens": 706514629.0, + "step": 27169 + }, + { + "epoch": 2.2404551826502845, + "grad_norm": 1.9112348556518555, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.8010621666908264, + "num_tokens": 706546839.0, + "step": 27170 + }, + { + "epoch": 2.240537643275336, + "grad_norm": 1.9972001314163208, + "learning_rate": 1e-06, + "loss": 0.5636, + "mean_token_accuracy": 0.8125820159912109, + "num_tokens": 706576320.0, + "step": 27171 + }, + { + "epoch": 2.2406201039003877, + "grad_norm": 2.2648637294769287, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7623680233955383, + "num_tokens": 706604173.0, + "step": 27172 + }, + { + "epoch": 2.240702564525439, + "grad_norm": 1.9322996139526367, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7759038209915161, + "num_tokens": 706637542.0, + "step": 27173 + }, + { + "epoch": 2.240785025150491, + "grad_norm": 2.3616786003112793, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7361547350883484, + "num_tokens": 706662787.0, + "step": 27174 + }, + { + "epoch": 2.240867485775542, + "grad_norm": 2.119417428970337, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.7801758050918579, + "num_tokens": 706696432.0, + "step": 27175 + }, + { + "epoch": 2.2409499464005935, + "grad_norm": 1.9653165340423584, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.7943375110626221, + "num_tokens": 706730565.0, + "step": 27176 + }, + { + "epoch": 2.2410324070256453, + "grad_norm": 1.925277590751648, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.7461913228034973, + "num_tokens": 706765941.0, + "step": 27177 + }, + { + "epoch": 2.2411148676506967, + "grad_norm": 2.1880693435668945, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7603407502174377, + "num_tokens": 706797172.0, + "step": 27178 + }, + { + "epoch": 2.2411973282757485, + "grad_norm": 1.969957709312439, + "learning_rate": 1e-06, + "loss": 0.5877, + "mean_token_accuracy": 0.8070670366287231, + "num_tokens": 706826293.0, + "step": 27179 + }, + { + "epoch": 2.2412797889008, + "grad_norm": 2.404433250427246, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7710522413253784, + "num_tokens": 706853185.0, + "step": 27180 + }, + { + "epoch": 2.2413622495258516, + "grad_norm": 1.999037265777588, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7709327936172485, + "num_tokens": 706884352.0, + "step": 27181 + }, + { + "epoch": 2.241444710150903, + "grad_norm": 1.8382017612457275, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.777567446231842, + "num_tokens": 706922228.0, + "step": 27182 + }, + { + "epoch": 2.2415271707759543, + "grad_norm": 2.0296318531036377, + "learning_rate": 1e-06, + "loss": 0.5759, + "mean_token_accuracy": 0.8024668097496033, + "num_tokens": 706953177.0, + "step": 27183 + }, + { + "epoch": 2.241609631401006, + "grad_norm": 2.378211259841919, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7823560237884521, + "num_tokens": 706975972.0, + "step": 27184 + }, + { + "epoch": 2.2416920920260575, + "grad_norm": 2.2447168827056885, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7626943588256836, + "num_tokens": 707003759.0, + "step": 27185 + }, + { + "epoch": 2.2417745526511093, + "grad_norm": 2.124950408935547, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.7875138521194458, + "num_tokens": 707032343.0, + "step": 27186 + }, + { + "epoch": 2.2418570132761606, + "grad_norm": 2.0535051822662354, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7773393392562866, + "num_tokens": 707062567.0, + "step": 27187 + }, + { + "epoch": 2.241939473901212, + "grad_norm": 2.374633550643921, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.7919278740882874, + "num_tokens": 707086195.0, + "step": 27188 + }, + { + "epoch": 2.2420219345262637, + "grad_norm": 1.8646996021270752, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7928133010864258, + "num_tokens": 707118605.0, + "step": 27189 + }, + { + "epoch": 2.242104395151315, + "grad_norm": 2.1818125247955322, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7565063238143921, + "num_tokens": 707145719.0, + "step": 27190 + }, + { + "epoch": 2.242186855776367, + "grad_norm": 2.1839191913604736, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7722735404968262, + "num_tokens": 707173946.0, + "step": 27191 + }, + { + "epoch": 2.2422693164014182, + "grad_norm": 2.2419872283935547, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7476831078529358, + "num_tokens": 707202020.0, + "step": 27192 + }, + { + "epoch": 2.24235177702647, + "grad_norm": 2.0635225772857666, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.735823392868042, + "num_tokens": 707235116.0, + "step": 27193 + }, + { + "epoch": 2.2424342376515214, + "grad_norm": 2.13281512260437, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.784177303314209, + "num_tokens": 707263683.0, + "step": 27194 + }, + { + "epoch": 2.2425166982765727, + "grad_norm": 2.1505002975463867, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7153119444847107, + "num_tokens": 707297115.0, + "step": 27195 + }, + { + "epoch": 2.2425991589016245, + "grad_norm": 2.1285266876220703, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.7867002487182617, + "num_tokens": 707323664.0, + "step": 27196 + }, + { + "epoch": 2.242681619526676, + "grad_norm": 2.194786310195923, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7706848382949829, + "num_tokens": 707349588.0, + "step": 27197 + }, + { + "epoch": 2.2427640801517277, + "grad_norm": 2.5325605869293213, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7392269372940063, + "num_tokens": 707373658.0, + "step": 27198 + }, + { + "epoch": 2.242846540776779, + "grad_norm": 1.9068107604980469, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7740583419799805, + "num_tokens": 707408665.0, + "step": 27199 + }, + { + "epoch": 2.242929001401831, + "grad_norm": 2.286611318588257, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7806110382080078, + "num_tokens": 707434868.0, + "step": 27200 + }, + { + "epoch": 2.243011462026882, + "grad_norm": 2.052008628845215, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.7807924747467041, + "num_tokens": 707463473.0, + "step": 27201 + }, + { + "epoch": 2.2430939226519335, + "grad_norm": 2.3790395259857178, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7793612480163574, + "num_tokens": 707488899.0, + "step": 27202 + }, + { + "epoch": 2.2431763832769853, + "grad_norm": 2.540076732635498, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7580448985099792, + "num_tokens": 707510221.0, + "step": 27203 + }, + { + "epoch": 2.2432588439020367, + "grad_norm": 2.2992842197418213, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7525539398193359, + "num_tokens": 707538502.0, + "step": 27204 + }, + { + "epoch": 2.2433413045270885, + "grad_norm": 2.035205602645874, + "learning_rate": 1e-06, + "loss": 0.6077, + "mean_token_accuracy": 0.7934929132461548, + "num_tokens": 707568910.0, + "step": 27205 + }, + { + "epoch": 2.24342376515214, + "grad_norm": 2.2479660511016846, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.7814144492149353, + "num_tokens": 707595909.0, + "step": 27206 + }, + { + "epoch": 2.2435062257771916, + "grad_norm": 2.0092036724090576, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7800862789154053, + "num_tokens": 707626175.0, + "step": 27207 + }, + { + "epoch": 2.243588686402243, + "grad_norm": 2.080430030822754, + "learning_rate": 1e-06, + "loss": 0.5636, + "mean_token_accuracy": 0.8054091930389404, + "num_tokens": 707659094.0, + "step": 27208 + }, + { + "epoch": 2.2436711470272943, + "grad_norm": 2.385427951812744, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.7919831275939941, + "num_tokens": 707683067.0, + "step": 27209 + }, + { + "epoch": 2.243753607652346, + "grad_norm": 2.0296809673309326, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7624603509902954, + "num_tokens": 707714945.0, + "step": 27210 + }, + { + "epoch": 2.2438360682773975, + "grad_norm": 2.0133845806121826, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7721972465515137, + "num_tokens": 707749076.0, + "step": 27211 + }, + { + "epoch": 2.2439185289024492, + "grad_norm": 2.0118486881256104, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7329320907592773, + "num_tokens": 707785332.0, + "step": 27212 + }, + { + "epoch": 2.2440009895275006, + "grad_norm": 2.0037693977355957, + "learning_rate": 1e-06, + "loss": 0.5432, + "mean_token_accuracy": 0.8171067237854004, + "num_tokens": 707815056.0, + "step": 27213 + }, + { + "epoch": 2.244083450152552, + "grad_norm": 1.972362995147705, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.8043156266212463, + "num_tokens": 707848759.0, + "step": 27214 + }, + { + "epoch": 2.2441659107776037, + "grad_norm": 1.9195283651351929, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7763049602508545, + "num_tokens": 707884054.0, + "step": 27215 + }, + { + "epoch": 2.244248371402655, + "grad_norm": 1.9478881359100342, + "learning_rate": 1e-06, + "loss": 0.54, + "mean_token_accuracy": 0.8152089715003967, + "num_tokens": 707915148.0, + "step": 27216 + }, + { + "epoch": 2.244330832027707, + "grad_norm": 2.2956008911132812, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.787868857383728, + "num_tokens": 707939029.0, + "step": 27217 + }, + { + "epoch": 2.2444132926527582, + "grad_norm": 1.8676446676254272, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.788414478302002, + "num_tokens": 707972627.0, + "step": 27218 + }, + { + "epoch": 2.24449575327781, + "grad_norm": 2.0184452533721924, + "learning_rate": 1e-06, + "loss": 0.5135, + "mean_token_accuracy": 0.8298965692520142, + "num_tokens": 707999415.0, + "step": 27219 + }, + { + "epoch": 2.2445782139028614, + "grad_norm": 2.2681398391723633, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.769635796546936, + "num_tokens": 708026411.0, + "step": 27220 + }, + { + "epoch": 2.2446606745279127, + "grad_norm": 2.214010000228882, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7707099318504333, + "num_tokens": 708053136.0, + "step": 27221 + }, + { + "epoch": 2.2447431351529645, + "grad_norm": 1.9714511632919312, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.7937537431716919, + "num_tokens": 708083298.0, + "step": 27222 + }, + { + "epoch": 2.244825595778016, + "grad_norm": 2.4589762687683105, + "learning_rate": 1e-06, + "loss": 0.5968, + "mean_token_accuracy": 0.7977663278579712, + "num_tokens": 708106011.0, + "step": 27223 + }, + { + "epoch": 2.2449080564030677, + "grad_norm": 2.150667428970337, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7583661079406738, + "num_tokens": 708137352.0, + "step": 27224 + }, + { + "epoch": 2.244990517028119, + "grad_norm": 2.479671001434326, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7484060525894165, + "num_tokens": 708162421.0, + "step": 27225 + }, + { + "epoch": 2.2450729776531704, + "grad_norm": 1.8284084796905518, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.793586015701294, + "num_tokens": 708201490.0, + "step": 27226 + }, + { + "epoch": 2.245155438278222, + "grad_norm": 2.050508975982666, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7607659101486206, + "num_tokens": 708233731.0, + "step": 27227 + }, + { + "epoch": 2.2452378989032735, + "grad_norm": 2.0984585285186768, + "learning_rate": 1e-06, + "loss": 0.5635, + "mean_token_accuracy": 0.8086881041526794, + "num_tokens": 708261683.0, + "step": 27228 + }, + { + "epoch": 2.2453203595283253, + "grad_norm": 2.0237834453582764, + "learning_rate": 1e-06, + "loss": 0.591, + "mean_token_accuracy": 0.798626720905304, + "num_tokens": 708293805.0, + "step": 27229 + }, + { + "epoch": 2.2454028201533767, + "grad_norm": 2.1646270751953125, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7432619333267212, + "num_tokens": 708324971.0, + "step": 27230 + }, + { + "epoch": 2.2454852807784285, + "grad_norm": 2.010355234146118, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7567627429962158, + "num_tokens": 708358429.0, + "step": 27231 + }, + { + "epoch": 2.24556774140348, + "grad_norm": 2.1498470306396484, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7549071311950684, + "num_tokens": 708388218.0, + "step": 27232 + }, + { + "epoch": 2.245650202028531, + "grad_norm": 2.2463183403015137, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7447567582130432, + "num_tokens": 708418346.0, + "step": 27233 + }, + { + "epoch": 2.245732662653583, + "grad_norm": 2.1944563388824463, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.73857581615448, + "num_tokens": 708449583.0, + "step": 27234 + }, + { + "epoch": 2.2458151232786343, + "grad_norm": 2.051332712173462, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7760361433029175, + "num_tokens": 708479916.0, + "step": 27235 + }, + { + "epoch": 2.245897583903686, + "grad_norm": 2.0306272506713867, + "learning_rate": 1e-06, + "loss": 0.5905, + "mean_token_accuracy": 0.8009419441223145, + "num_tokens": 708508057.0, + "step": 27236 + }, + { + "epoch": 2.2459800445287375, + "grad_norm": 2.256455183029175, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.765614926815033, + "num_tokens": 708536200.0, + "step": 27237 + }, + { + "epoch": 2.2460625051537892, + "grad_norm": 2.1340157985687256, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7582150101661682, + "num_tokens": 708564207.0, + "step": 27238 + }, + { + "epoch": 2.2461449657788406, + "grad_norm": 1.888187289237976, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7762627601623535, + "num_tokens": 708598315.0, + "step": 27239 + }, + { + "epoch": 2.246227426403892, + "grad_norm": 2.219696283340454, + "learning_rate": 1e-06, + "loss": 0.5821, + "mean_token_accuracy": 0.8080734014511108, + "num_tokens": 708623728.0, + "step": 27240 + }, + { + "epoch": 2.2463098870289437, + "grad_norm": 2.1886961460113525, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7673672437667847, + "num_tokens": 708650522.0, + "step": 27241 + }, + { + "epoch": 2.246392347653995, + "grad_norm": 2.423159122467041, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.7823525667190552, + "num_tokens": 708673419.0, + "step": 27242 + }, + { + "epoch": 2.246474808279047, + "grad_norm": 2.021684169769287, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.750917911529541, + "num_tokens": 708706006.0, + "step": 27243 + }, + { + "epoch": 2.2465572689040982, + "grad_norm": 1.9415162801742554, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.7884126305580139, + "num_tokens": 708737247.0, + "step": 27244 + }, + { + "epoch": 2.24663972952915, + "grad_norm": 2.2934389114379883, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7711187601089478, + "num_tokens": 708762327.0, + "step": 27245 + }, + { + "epoch": 2.2467221901542014, + "grad_norm": 2.1491761207580566, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7623165249824524, + "num_tokens": 708793719.0, + "step": 27246 + }, + { + "epoch": 2.2468046507792527, + "grad_norm": 2.1473236083984375, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.7831655144691467, + "num_tokens": 708824038.0, + "step": 27247 + }, + { + "epoch": 2.2468871114043045, + "grad_norm": 1.9871891736984253, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.769244909286499, + "num_tokens": 708860530.0, + "step": 27248 + }, + { + "epoch": 2.246969572029356, + "grad_norm": 1.8652455806732178, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.7832406759262085, + "num_tokens": 708896199.0, + "step": 27249 + }, + { + "epoch": 2.2470520326544077, + "grad_norm": 2.126157760620117, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7697769999504089, + "num_tokens": 708927140.0, + "step": 27250 + }, + { + "epoch": 2.247134493279459, + "grad_norm": 2.0726380348205566, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.778560996055603, + "num_tokens": 708959893.0, + "step": 27251 + }, + { + "epoch": 2.247216953904511, + "grad_norm": 2.108492612838745, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7699223160743713, + "num_tokens": 708991663.0, + "step": 27252 + }, + { + "epoch": 2.247299414529562, + "grad_norm": 2.028066396713257, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.762917160987854, + "num_tokens": 709023637.0, + "step": 27253 + }, + { + "epoch": 2.2473818751546135, + "grad_norm": 1.8781497478485107, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7374221086502075, + "num_tokens": 709060868.0, + "step": 27254 + }, + { + "epoch": 2.2474643357796653, + "grad_norm": 2.0894763469696045, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.7981287837028503, + "num_tokens": 709091971.0, + "step": 27255 + }, + { + "epoch": 2.2475467964047167, + "grad_norm": 2.4849178791046143, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7515923976898193, + "num_tokens": 709115342.0, + "step": 27256 + }, + { + "epoch": 2.2476292570297685, + "grad_norm": 2.2313642501831055, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7728948593139648, + "num_tokens": 709142009.0, + "step": 27257 + }, + { + "epoch": 2.24771171765482, + "grad_norm": 1.9670164585113525, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.7788209915161133, + "num_tokens": 709174373.0, + "step": 27258 + }, + { + "epoch": 2.247794178279871, + "grad_norm": 1.956716537475586, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7548835873603821, + "num_tokens": 709207835.0, + "step": 27259 + }, + { + "epoch": 2.247876638904923, + "grad_norm": 2.3665740489959717, + "learning_rate": 1e-06, + "loss": 0.8338, + "mean_token_accuracy": 0.7429164052009583, + "num_tokens": 709233315.0, + "step": 27260 + }, + { + "epoch": 2.2479590995299743, + "grad_norm": 1.9017729759216309, + "learning_rate": 1e-06, + "loss": 0.5273, + "mean_token_accuracy": 0.819701075553894, + "num_tokens": 709264859.0, + "step": 27261 + }, + { + "epoch": 2.248041560155026, + "grad_norm": 2.1581480503082275, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7690619230270386, + "num_tokens": 709295982.0, + "step": 27262 + }, + { + "epoch": 2.2481240207800774, + "grad_norm": 1.9948283433914185, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.7799570560455322, + "num_tokens": 709325563.0, + "step": 27263 + }, + { + "epoch": 2.2482064814051292, + "grad_norm": 1.9194319248199463, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7823795080184937, + "num_tokens": 709362515.0, + "step": 27264 + }, + { + "epoch": 2.2482889420301806, + "grad_norm": 1.9275026321411133, + "learning_rate": 1e-06, + "loss": 0.5765, + "mean_token_accuracy": 0.8048477172851562, + "num_tokens": 709397328.0, + "step": 27265 + }, + { + "epoch": 2.248371402655232, + "grad_norm": 2.176100730895996, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7755573987960815, + "num_tokens": 709422713.0, + "step": 27266 + }, + { + "epoch": 2.2484538632802837, + "grad_norm": 2.0893092155456543, + "learning_rate": 1e-06, + "loss": 0.5534, + "mean_token_accuracy": 0.8095988631248474, + "num_tokens": 709451890.0, + "step": 27267 + }, + { + "epoch": 2.248536323905335, + "grad_norm": 2.0957818031311035, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.767624020576477, + "num_tokens": 709484488.0, + "step": 27268 + }, + { + "epoch": 2.248618784530387, + "grad_norm": 1.9421337842941284, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7328198552131653, + "num_tokens": 709521560.0, + "step": 27269 + }, + { + "epoch": 2.2487012451554382, + "grad_norm": 1.9506644010543823, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.7837706804275513, + "num_tokens": 709556205.0, + "step": 27270 + }, + { + "epoch": 2.2487837057804896, + "grad_norm": 2.215534210205078, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.7939552068710327, + "num_tokens": 709582714.0, + "step": 27271 + }, + { + "epoch": 2.2488661664055414, + "grad_norm": 2.611924171447754, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7509093284606934, + "num_tokens": 709604086.0, + "step": 27272 + }, + { + "epoch": 2.2489486270305927, + "grad_norm": 2.0409255027770996, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.7942295074462891, + "num_tokens": 709633686.0, + "step": 27273 + }, + { + "epoch": 2.2490310876556445, + "grad_norm": 2.2874999046325684, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.729928731918335, + "num_tokens": 709661658.0, + "step": 27274 + }, + { + "epoch": 2.249113548280696, + "grad_norm": 1.7428474426269531, + "learning_rate": 1e-06, + "loss": 0.4717, + "mean_token_accuracy": 0.8383389115333557, + "num_tokens": 709698736.0, + "step": 27275 + }, + { + "epoch": 2.2491960089057477, + "grad_norm": 1.995583176612854, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7774637937545776, + "num_tokens": 709731147.0, + "step": 27276 + }, + { + "epoch": 2.249278469530799, + "grad_norm": 2.2960619926452637, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7758911848068237, + "num_tokens": 709757530.0, + "step": 27277 + }, + { + "epoch": 2.2493609301558504, + "grad_norm": 1.9232878684997559, + "learning_rate": 1e-06, + "loss": 0.5923, + "mean_token_accuracy": 0.8047710657119751, + "num_tokens": 709792396.0, + "step": 27278 + }, + { + "epoch": 2.249443390780902, + "grad_norm": 2.5063464641571045, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7672505378723145, + "num_tokens": 709815900.0, + "step": 27279 + }, + { + "epoch": 2.2495258514059535, + "grad_norm": 2.1140449047088623, + "learning_rate": 1e-06, + "loss": 0.5609, + "mean_token_accuracy": 0.8015269041061401, + "num_tokens": 709845843.0, + "step": 27280 + }, + { + "epoch": 2.2496083120310053, + "grad_norm": 2.211780309677124, + "learning_rate": 1e-06, + "loss": 0.7176, + "mean_token_accuracy": 0.7679978609085083, + "num_tokens": 709877388.0, + "step": 27281 + }, + { + "epoch": 2.2496907726560567, + "grad_norm": 1.8457298278808594, + "learning_rate": 1e-06, + "loss": 0.6023, + "mean_token_accuracy": 0.7997450828552246, + "num_tokens": 709912041.0, + "step": 27282 + }, + { + "epoch": 2.2497732332811085, + "grad_norm": 2.370201587677002, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7905106544494629, + "num_tokens": 709935870.0, + "step": 27283 + }, + { + "epoch": 2.24985569390616, + "grad_norm": 2.2460927963256836, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7394894361495972, + "num_tokens": 709965328.0, + "step": 27284 + }, + { + "epoch": 2.249938154531211, + "grad_norm": 2.2159318923950195, + "learning_rate": 1e-06, + "loss": 0.6192, + "mean_token_accuracy": 0.7877095341682434, + "num_tokens": 709992595.0, + "step": 27285 + }, + { + "epoch": 2.250020615156263, + "grad_norm": 1.8125784397125244, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7639379501342773, + "num_tokens": 710030745.0, + "step": 27286 + }, + { + "epoch": 2.2501030757813143, + "grad_norm": 1.865358591079712, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7544231414794922, + "num_tokens": 710069173.0, + "step": 27287 + }, + { + "epoch": 2.250185536406366, + "grad_norm": 2.1143226623535156, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7781475782394409, + "num_tokens": 710100177.0, + "step": 27288 + }, + { + "epoch": 2.2502679970314174, + "grad_norm": 2.0843026638031006, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7613192200660706, + "num_tokens": 710132989.0, + "step": 27289 + }, + { + "epoch": 2.2503504576564692, + "grad_norm": 1.8430811166763306, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7680217027664185, + "num_tokens": 710173699.0, + "step": 27290 + }, + { + "epoch": 2.2504329182815206, + "grad_norm": 2.275566339492798, + "learning_rate": 1e-06, + "loss": 0.5973, + "mean_token_accuracy": 0.7861932516098022, + "num_tokens": 710198860.0, + "step": 27291 + }, + { + "epoch": 2.250515378906572, + "grad_norm": 1.9519846439361572, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.8041509389877319, + "num_tokens": 710232049.0, + "step": 27292 + }, + { + "epoch": 2.2505978395316237, + "grad_norm": 2.052502155303955, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7774901390075684, + "num_tokens": 710262285.0, + "step": 27293 + }, + { + "epoch": 2.250680300156675, + "grad_norm": 2.3275270462036133, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7733215689659119, + "num_tokens": 710291052.0, + "step": 27294 + }, + { + "epoch": 2.250762760781727, + "grad_norm": 2.4835033416748047, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7480071783065796, + "num_tokens": 710315369.0, + "step": 27295 + }, + { + "epoch": 2.2508452214067782, + "grad_norm": 2.0996315479278564, + "learning_rate": 1e-06, + "loss": 0.603, + "mean_token_accuracy": 0.8005565404891968, + "num_tokens": 710346580.0, + "step": 27296 + }, + { + "epoch": 2.25092768203183, + "grad_norm": 2.039137601852417, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.7227059602737427, + "num_tokens": 710382095.0, + "step": 27297 + }, + { + "epoch": 2.2510101426568814, + "grad_norm": 2.3318939208984375, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7519562244415283, + "num_tokens": 710408759.0, + "step": 27298 + }, + { + "epoch": 2.2510926032819327, + "grad_norm": 2.3506381511688232, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7405146956443787, + "num_tokens": 710435150.0, + "step": 27299 + }, + { + "epoch": 2.2511750639069845, + "grad_norm": 2.5517637729644775, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7835869193077087, + "num_tokens": 710456400.0, + "step": 27300 + }, + { + "epoch": 2.251257524532036, + "grad_norm": 1.855486273765564, + "learning_rate": 1e-06, + "loss": 0.549, + "mean_token_accuracy": 0.8133358955383301, + "num_tokens": 710491851.0, + "step": 27301 + }, + { + "epoch": 2.2513399851570877, + "grad_norm": 1.9533878564834595, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7811181545257568, + "num_tokens": 710524512.0, + "step": 27302 + }, + { + "epoch": 2.251422445782139, + "grad_norm": 2.014481544494629, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.7941209077835083, + "num_tokens": 710552373.0, + "step": 27303 + }, + { + "epoch": 2.251504906407191, + "grad_norm": 1.958449125289917, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.7877297401428223, + "num_tokens": 710583190.0, + "step": 27304 + }, + { + "epoch": 2.251587367032242, + "grad_norm": 2.129823923110962, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7577139735221863, + "num_tokens": 710614086.0, + "step": 27305 + }, + { + "epoch": 2.2516698276572935, + "grad_norm": 2.172710418701172, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7650306224822998, + "num_tokens": 710643846.0, + "step": 27306 + }, + { + "epoch": 2.2517522882823453, + "grad_norm": 2.295644521713257, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.7801220417022705, + "num_tokens": 710669703.0, + "step": 27307 + }, + { + "epoch": 2.2518347489073967, + "grad_norm": 2.179802656173706, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7563931941986084, + "num_tokens": 710700787.0, + "step": 27308 + }, + { + "epoch": 2.2519172095324484, + "grad_norm": 1.890924096107483, + "learning_rate": 1e-06, + "loss": 0.6101, + "mean_token_accuracy": 0.8004999160766602, + "num_tokens": 710732798.0, + "step": 27309 + }, + { + "epoch": 2.2519996701575, + "grad_norm": 2.111677885055542, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.7966586351394653, + "num_tokens": 710762749.0, + "step": 27310 + }, + { + "epoch": 2.252082130782551, + "grad_norm": 2.1999616622924805, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7757078409194946, + "num_tokens": 710789591.0, + "step": 27311 + }, + { + "epoch": 2.252164591407603, + "grad_norm": 1.759610891342163, + "learning_rate": 1e-06, + "loss": 0.5991, + "mean_token_accuracy": 0.8147910833358765, + "num_tokens": 710824037.0, + "step": 27312 + }, + { + "epoch": 2.2522470520326543, + "grad_norm": 2.096709966659546, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7577162981033325, + "num_tokens": 710855371.0, + "step": 27313 + }, + { + "epoch": 2.252329512657706, + "grad_norm": 2.055614709854126, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7645266056060791, + "num_tokens": 710885663.0, + "step": 27314 + }, + { + "epoch": 2.2524119732827574, + "grad_norm": 1.7227725982666016, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7796716690063477, + "num_tokens": 710927362.0, + "step": 27315 + }, + { + "epoch": 2.252494433907809, + "grad_norm": 1.9016008377075195, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7755886912345886, + "num_tokens": 710960480.0, + "step": 27316 + }, + { + "epoch": 2.2525768945328606, + "grad_norm": 1.9881709814071655, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7710931301116943, + "num_tokens": 710991889.0, + "step": 27317 + }, + { + "epoch": 2.252659355157912, + "grad_norm": 2.1482009887695312, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.7714954614639282, + "num_tokens": 711019071.0, + "step": 27318 + }, + { + "epoch": 2.2527418157829637, + "grad_norm": 2.1223063468933105, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7353515028953552, + "num_tokens": 711050922.0, + "step": 27319 + }, + { + "epoch": 2.252824276408015, + "grad_norm": 2.083038091659546, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7724689245223999, + "num_tokens": 711083355.0, + "step": 27320 + }, + { + "epoch": 2.252906737033067, + "grad_norm": 2.5350894927978516, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.7756041288375854, + "num_tokens": 711107137.0, + "step": 27321 + }, + { + "epoch": 2.2529891976581182, + "grad_norm": 2.1653449535369873, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7450504302978516, + "num_tokens": 711136305.0, + "step": 27322 + }, + { + "epoch": 2.2530716582831696, + "grad_norm": 2.0327606201171875, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7725992798805237, + "num_tokens": 711170180.0, + "step": 27323 + }, + { + "epoch": 2.2531541189082214, + "grad_norm": 1.9633618593215942, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7737905979156494, + "num_tokens": 711202580.0, + "step": 27324 + }, + { + "epoch": 2.2532365795332727, + "grad_norm": 2.043398141860962, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7439324259757996, + "num_tokens": 711235398.0, + "step": 27325 + }, + { + "epoch": 2.2533190401583245, + "grad_norm": 2.124976873397827, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7790546417236328, + "num_tokens": 711271095.0, + "step": 27326 + }, + { + "epoch": 2.253401500783376, + "grad_norm": 2.0569472312927246, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7730380892753601, + "num_tokens": 711301810.0, + "step": 27327 + }, + { + "epoch": 2.2534839614084277, + "grad_norm": 2.175882577896118, + "learning_rate": 1e-06, + "loss": 0.6309, + "mean_token_accuracy": 0.7925834655761719, + "num_tokens": 711330610.0, + "step": 27328 + }, + { + "epoch": 2.253566422033479, + "grad_norm": 2.025127410888672, + "learning_rate": 1e-06, + "loss": 0.4568, + "mean_token_accuracy": 0.8413238525390625, + "num_tokens": 711359241.0, + "step": 27329 + }, + { + "epoch": 2.2536488826585304, + "grad_norm": 2.575038433074951, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7654269933700562, + "num_tokens": 711382890.0, + "step": 27330 + }, + { + "epoch": 2.253731343283582, + "grad_norm": 1.9357935190200806, + "learning_rate": 1e-06, + "loss": 0.569, + "mean_token_accuracy": 0.815730094909668, + "num_tokens": 711416124.0, + "step": 27331 + }, + { + "epoch": 2.2538138039086335, + "grad_norm": 1.7811129093170166, + "learning_rate": 1e-06, + "loss": 0.5589, + "mean_token_accuracy": 0.8158767223358154, + "num_tokens": 711450273.0, + "step": 27332 + }, + { + "epoch": 2.2538962645336853, + "grad_norm": 2.1202023029327393, + "learning_rate": 1e-06, + "loss": 0.5974, + "mean_token_accuracy": 0.7998915314674377, + "num_tokens": 711479220.0, + "step": 27333 + }, + { + "epoch": 2.2539787251587367, + "grad_norm": 1.7439649105072021, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7822859287261963, + "num_tokens": 711516385.0, + "step": 27334 + }, + { + "epoch": 2.2540611857837884, + "grad_norm": 1.821010708808899, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.7914324402809143, + "num_tokens": 711551667.0, + "step": 27335 + }, + { + "epoch": 2.25414364640884, + "grad_norm": 2.154792070388794, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7412080764770508, + "num_tokens": 711582046.0, + "step": 27336 + }, + { + "epoch": 2.254226107033891, + "grad_norm": 1.9103165864944458, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7722693681716919, + "num_tokens": 711618496.0, + "step": 27337 + }, + { + "epoch": 2.254308567658943, + "grad_norm": 1.9548306465148926, + "learning_rate": 1e-06, + "loss": 0.5129, + "mean_token_accuracy": 0.8253593444824219, + "num_tokens": 711650931.0, + "step": 27338 + }, + { + "epoch": 2.2543910282839943, + "grad_norm": 2.0520615577697754, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.7574831247329712, + "num_tokens": 711683963.0, + "step": 27339 + }, + { + "epoch": 2.254473488909046, + "grad_norm": 2.4693288803100586, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.8000867366790771, + "num_tokens": 711705794.0, + "step": 27340 + }, + { + "epoch": 2.2545559495340974, + "grad_norm": 1.9313846826553345, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7690843939781189, + "num_tokens": 711739854.0, + "step": 27341 + }, + { + "epoch": 2.2546384101591492, + "grad_norm": 2.585685968399048, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7852827310562134, + "num_tokens": 711762553.0, + "step": 27342 + }, + { + "epoch": 2.2547208707842006, + "grad_norm": 2.2789711952209473, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7783515453338623, + "num_tokens": 711787785.0, + "step": 27343 + }, + { + "epoch": 2.254803331409252, + "grad_norm": 1.8659322261810303, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.7927711606025696, + "num_tokens": 711822429.0, + "step": 27344 + }, + { + "epoch": 2.2548857920343037, + "grad_norm": 1.942922830581665, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7439142465591431, + "num_tokens": 711857183.0, + "step": 27345 + }, + { + "epoch": 2.254968252659355, + "grad_norm": 2.2942264080047607, + "learning_rate": 1e-06, + "loss": 0.7852, + "mean_token_accuracy": 0.753189742565155, + "num_tokens": 711885991.0, + "step": 27346 + }, + { + "epoch": 2.255050713284407, + "grad_norm": 2.17341685295105, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7716889381408691, + "num_tokens": 711912474.0, + "step": 27347 + }, + { + "epoch": 2.255133173909458, + "grad_norm": 2.234614372253418, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.763962984085083, + "num_tokens": 711939617.0, + "step": 27348 + }, + { + "epoch": 2.25521563453451, + "grad_norm": 2.294140100479126, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7403886318206787, + "num_tokens": 711966735.0, + "step": 27349 + }, + { + "epoch": 2.2552980951595614, + "grad_norm": 2.4158363342285156, + "learning_rate": 1e-06, + "loss": 0.7498, + "mean_token_accuracy": 0.7627699375152588, + "num_tokens": 711988464.0, + "step": 27350 + }, + { + "epoch": 2.2553805557846127, + "grad_norm": 2.1107640266418457, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.778769314289093, + "num_tokens": 712019737.0, + "step": 27351 + }, + { + "epoch": 2.2554630164096645, + "grad_norm": 2.2917816638946533, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.7791626453399658, + "num_tokens": 712044912.0, + "step": 27352 + }, + { + "epoch": 2.255545477034716, + "grad_norm": 2.2056963443756104, + "learning_rate": 1e-06, + "loss": 0.5867, + "mean_token_accuracy": 0.8142538070678711, + "num_tokens": 712070753.0, + "step": 27353 + }, + { + "epoch": 2.255627937659767, + "grad_norm": 2.168489694595337, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7671074271202087, + "num_tokens": 712099722.0, + "step": 27354 + }, + { + "epoch": 2.255710398284819, + "grad_norm": 1.8934359550476074, + "learning_rate": 1e-06, + "loss": 0.5995, + "mean_token_accuracy": 0.7828087210655212, + "num_tokens": 712131976.0, + "step": 27355 + }, + { + "epoch": 2.2557928589098704, + "grad_norm": 2.2022573947906494, + "learning_rate": 1e-06, + "loss": 0.5634, + "mean_token_accuracy": 0.8074885010719299, + "num_tokens": 712161818.0, + "step": 27356 + }, + { + "epoch": 2.255875319534922, + "grad_norm": 2.550506353378296, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7631829380989075, + "num_tokens": 712185224.0, + "step": 27357 + }, + { + "epoch": 2.2559577801599735, + "grad_norm": 1.8193756341934204, + "learning_rate": 1e-06, + "loss": 0.5937, + "mean_token_accuracy": 0.8054492473602295, + "num_tokens": 712219660.0, + "step": 27358 + }, + { + "epoch": 2.2560402407850253, + "grad_norm": 1.8262614011764526, + "learning_rate": 1e-06, + "loss": 0.539, + "mean_token_accuracy": 0.8155166506767273, + "num_tokens": 712254916.0, + "step": 27359 + }, + { + "epoch": 2.2561227014100766, + "grad_norm": 2.256566047668457, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.7766594886779785, + "num_tokens": 712279287.0, + "step": 27360 + }, + { + "epoch": 2.256205162035128, + "grad_norm": 1.8891935348510742, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.783223032951355, + "num_tokens": 712316715.0, + "step": 27361 + }, + { + "epoch": 2.25628762266018, + "grad_norm": 2.150423526763916, + "learning_rate": 1e-06, + "loss": 0.7486, + "mean_token_accuracy": 0.7725244164466858, + "num_tokens": 712345333.0, + "step": 27362 + }, + { + "epoch": 2.256370083285231, + "grad_norm": 2.0334134101867676, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.7803931832313538, + "num_tokens": 712377070.0, + "step": 27363 + }, + { + "epoch": 2.256452543910283, + "grad_norm": 2.2803292274475098, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7579838037490845, + "num_tokens": 712401379.0, + "step": 27364 + }, + { + "epoch": 2.2565350045353343, + "grad_norm": 2.420726776123047, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.7893946766853333, + "num_tokens": 712425542.0, + "step": 27365 + }, + { + "epoch": 2.256617465160386, + "grad_norm": 1.9171416759490967, + "learning_rate": 1e-06, + "loss": 0.6029, + "mean_token_accuracy": 0.8001739978790283, + "num_tokens": 712460886.0, + "step": 27366 + }, + { + "epoch": 2.2566999257854374, + "grad_norm": 2.124272346496582, + "learning_rate": 1e-06, + "loss": 0.5426, + "mean_token_accuracy": 0.8183892369270325, + "num_tokens": 712489693.0, + "step": 27367 + }, + { + "epoch": 2.256782386410489, + "grad_norm": 2.4389374256134033, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.7833652496337891, + "num_tokens": 712514671.0, + "step": 27368 + }, + { + "epoch": 2.2568648470355406, + "grad_norm": 2.5856337547302246, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7634868621826172, + "num_tokens": 712537260.0, + "step": 27369 + }, + { + "epoch": 2.256947307660592, + "grad_norm": 1.7212300300598145, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7818652391433716, + "num_tokens": 712578804.0, + "step": 27370 + }, + { + "epoch": 2.2570297682856437, + "grad_norm": 2.2855615615844727, + "learning_rate": 1e-06, + "loss": 0.7724, + "mean_token_accuracy": 0.7531006336212158, + "num_tokens": 712609720.0, + "step": 27371 + }, + { + "epoch": 2.257112228910695, + "grad_norm": 1.8987387418746948, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7737077474594116, + "num_tokens": 712646443.0, + "step": 27372 + }, + { + "epoch": 2.257194689535747, + "grad_norm": 1.8958137035369873, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7630020380020142, + "num_tokens": 712682905.0, + "step": 27373 + }, + { + "epoch": 2.257277150160798, + "grad_norm": 1.8637733459472656, + "learning_rate": 1e-06, + "loss": 0.5806, + "mean_token_accuracy": 0.8116950988769531, + "num_tokens": 712718115.0, + "step": 27374 + }, + { + "epoch": 2.2573596107858496, + "grad_norm": 2.051926374435425, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7649216055870056, + "num_tokens": 712749471.0, + "step": 27375 + }, + { + "epoch": 2.2574420714109014, + "grad_norm": 2.3404784202575684, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7946997284889221, + "num_tokens": 712774395.0, + "step": 27376 + }, + { + "epoch": 2.2575245320359527, + "grad_norm": 1.9488701820373535, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.77098548412323, + "num_tokens": 712806811.0, + "step": 27377 + }, + { + "epoch": 2.2576069926610045, + "grad_norm": 2.167584180831909, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7645504474639893, + "num_tokens": 712837569.0, + "step": 27378 + }, + { + "epoch": 2.257689453286056, + "grad_norm": 1.9957424402236938, + "learning_rate": 1e-06, + "loss": 0.5894, + "mean_token_accuracy": 0.8058880567550659, + "num_tokens": 712869620.0, + "step": 27379 + }, + { + "epoch": 2.2577719139111077, + "grad_norm": 2.1987619400024414, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7798514366149902, + "num_tokens": 712898655.0, + "step": 27380 + }, + { + "epoch": 2.257854374536159, + "grad_norm": 1.8735039234161377, + "learning_rate": 1e-06, + "loss": 0.5797, + "mean_token_accuracy": 0.8002405762672424, + "num_tokens": 712929242.0, + "step": 27381 + }, + { + "epoch": 2.2579368351612104, + "grad_norm": 1.8652517795562744, + "learning_rate": 1e-06, + "loss": 0.5242, + "mean_token_accuracy": 0.8231447339057922, + "num_tokens": 712962324.0, + "step": 27382 + }, + { + "epoch": 2.258019295786262, + "grad_norm": 1.99222731590271, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7718262672424316, + "num_tokens": 712996103.0, + "step": 27383 + }, + { + "epoch": 2.2581017564113135, + "grad_norm": 2.140470504760742, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.7494398355484009, + "num_tokens": 713024740.0, + "step": 27384 + }, + { + "epoch": 2.2581842170363653, + "grad_norm": 2.4124257564544678, + "learning_rate": 1e-06, + "loss": 0.7767, + "mean_token_accuracy": 0.7608863711357117, + "num_tokens": 713048786.0, + "step": 27385 + }, + { + "epoch": 2.2582666776614166, + "grad_norm": 2.1281559467315674, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7781113386154175, + "num_tokens": 713076771.0, + "step": 27386 + }, + { + "epoch": 2.2583491382864684, + "grad_norm": 2.198080062866211, + "learning_rate": 1e-06, + "loss": 0.5695, + "mean_token_accuracy": 0.807669460773468, + "num_tokens": 713101372.0, + "step": 27387 + }, + { + "epoch": 2.25843159891152, + "grad_norm": 2.0831758975982666, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8024324774742126, + "num_tokens": 713132427.0, + "step": 27388 + }, + { + "epoch": 2.258514059536571, + "grad_norm": 2.2484707832336426, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7685202360153198, + "num_tokens": 713158831.0, + "step": 27389 + }, + { + "epoch": 2.258596520161623, + "grad_norm": 2.0385053157806396, + "learning_rate": 1e-06, + "loss": 0.5994, + "mean_token_accuracy": 0.7913585305213928, + "num_tokens": 713189609.0, + "step": 27390 + }, + { + "epoch": 2.2586789807866743, + "grad_norm": 2.059981346130371, + "learning_rate": 1e-06, + "loss": 0.7243, + "mean_token_accuracy": 0.7679075002670288, + "num_tokens": 713219974.0, + "step": 27391 + }, + { + "epoch": 2.258761441411726, + "grad_norm": 2.199967622756958, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7392407655715942, + "num_tokens": 713252834.0, + "step": 27392 + }, + { + "epoch": 2.2588439020367774, + "grad_norm": 2.0278358459472656, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7815117239952087, + "num_tokens": 713283584.0, + "step": 27393 + }, + { + "epoch": 2.2589263626618292, + "grad_norm": 2.2614283561706543, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7498183846473694, + "num_tokens": 713311146.0, + "step": 27394 + }, + { + "epoch": 2.2590088232868806, + "grad_norm": 1.9734408855438232, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.7938819527626038, + "num_tokens": 713343004.0, + "step": 27395 + }, + { + "epoch": 2.259091283911932, + "grad_norm": 2.1467385292053223, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7701429128646851, + "num_tokens": 713371440.0, + "step": 27396 + }, + { + "epoch": 2.2591737445369837, + "grad_norm": 2.381181240081787, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7264597415924072, + "num_tokens": 713397021.0, + "step": 27397 + }, + { + "epoch": 2.259256205162035, + "grad_norm": 2.1217541694641113, + "learning_rate": 1e-06, + "loss": 0.7396, + "mean_token_accuracy": 0.7527551054954529, + "num_tokens": 713426593.0, + "step": 27398 + }, + { + "epoch": 2.2593386657870864, + "grad_norm": 2.4925336837768555, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.772122323513031, + "num_tokens": 713449236.0, + "step": 27399 + }, + { + "epoch": 2.259421126412138, + "grad_norm": 2.1852364540100098, + "learning_rate": 1e-06, + "loss": 0.7762, + "mean_token_accuracy": 0.7478985786437988, + "num_tokens": 713477636.0, + "step": 27400 + }, + { + "epoch": 2.2595035870371896, + "grad_norm": 2.2628731727600098, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7663167715072632, + "num_tokens": 713503502.0, + "step": 27401 + }, + { + "epoch": 2.2595860476622414, + "grad_norm": 2.2375364303588867, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7545888423919678, + "num_tokens": 713528864.0, + "step": 27402 + }, + { + "epoch": 2.2596685082872927, + "grad_norm": 2.190014362335205, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.7811191082000732, + "num_tokens": 713560074.0, + "step": 27403 + }, + { + "epoch": 2.2597509689123445, + "grad_norm": 2.1630353927612305, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7419155240058899, + "num_tokens": 713588545.0, + "step": 27404 + }, + { + "epoch": 2.259833429537396, + "grad_norm": 2.201767921447754, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7344316244125366, + "num_tokens": 713618935.0, + "step": 27405 + }, + { + "epoch": 2.259915890162447, + "grad_norm": 2.3384029865264893, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.771669328212738, + "num_tokens": 713644964.0, + "step": 27406 + }, + { + "epoch": 2.259998350787499, + "grad_norm": 1.9991788864135742, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.7943487167358398, + "num_tokens": 713676295.0, + "step": 27407 + }, + { + "epoch": 2.2600808114125504, + "grad_norm": 1.956251621246338, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.76993727684021, + "num_tokens": 713709236.0, + "step": 27408 + }, + { + "epoch": 2.260163272037602, + "grad_norm": 2.0837700366973877, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7660950422286987, + "num_tokens": 713740348.0, + "step": 27409 + }, + { + "epoch": 2.2602457326626535, + "grad_norm": 2.157825231552124, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7390004396438599, + "num_tokens": 713773209.0, + "step": 27410 + }, + { + "epoch": 2.2603281932877053, + "grad_norm": 2.1030492782592773, + "learning_rate": 1e-06, + "loss": 0.6049, + "mean_token_accuracy": 0.7959792613983154, + "num_tokens": 713800714.0, + "step": 27411 + }, + { + "epoch": 2.2604106539127566, + "grad_norm": 2.004899024963379, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.7950150966644287, + "num_tokens": 713832684.0, + "step": 27412 + }, + { + "epoch": 2.260493114537808, + "grad_norm": 2.208716630935669, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7530137300491333, + "num_tokens": 713860980.0, + "step": 27413 + }, + { + "epoch": 2.26057557516286, + "grad_norm": 2.2268054485321045, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7687770128250122, + "num_tokens": 713887922.0, + "step": 27414 + }, + { + "epoch": 2.260658035787911, + "grad_norm": 2.5033371448516846, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.7842179536819458, + "num_tokens": 713911202.0, + "step": 27415 + }, + { + "epoch": 2.260740496412963, + "grad_norm": 2.1207406520843506, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7576799392700195, + "num_tokens": 713940330.0, + "step": 27416 + }, + { + "epoch": 2.2608229570380143, + "grad_norm": 1.9848686456680298, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7617409229278564, + "num_tokens": 713975314.0, + "step": 27417 + }, + { + "epoch": 2.260905417663066, + "grad_norm": 2.321650743484497, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7660923004150391, + "num_tokens": 714000777.0, + "step": 27418 + }, + { + "epoch": 2.2609878782881174, + "grad_norm": 2.107698440551758, + "learning_rate": 1e-06, + "loss": 0.5552, + "mean_token_accuracy": 0.8053953051567078, + "num_tokens": 714031806.0, + "step": 27419 + }, + { + "epoch": 2.2610703389131688, + "grad_norm": 1.8358933925628662, + "learning_rate": 1e-06, + "loss": 0.5811, + "mean_token_accuracy": 0.8080127835273743, + "num_tokens": 714066016.0, + "step": 27420 + }, + { + "epoch": 2.2611527995382206, + "grad_norm": 1.9338006973266602, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7597430944442749, + "num_tokens": 714098785.0, + "step": 27421 + }, + { + "epoch": 2.261235260163272, + "grad_norm": 2.0647339820861816, + "learning_rate": 1e-06, + "loss": 0.5203, + "mean_token_accuracy": 0.8248834609985352, + "num_tokens": 714127390.0, + "step": 27422 + }, + { + "epoch": 2.2613177207883237, + "grad_norm": 2.2514727115631104, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7858558893203735, + "num_tokens": 714151468.0, + "step": 27423 + }, + { + "epoch": 2.261400181413375, + "grad_norm": 1.9825636148452759, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7793853282928467, + "num_tokens": 714185017.0, + "step": 27424 + }, + { + "epoch": 2.261482642038427, + "grad_norm": 2.173557758331299, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7644107341766357, + "num_tokens": 714214731.0, + "step": 27425 + }, + { + "epoch": 2.261565102663478, + "grad_norm": 2.021730422973633, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7842344045639038, + "num_tokens": 714245387.0, + "step": 27426 + }, + { + "epoch": 2.2616475632885296, + "grad_norm": 2.1128876209259033, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7651739120483398, + "num_tokens": 714276907.0, + "step": 27427 + }, + { + "epoch": 2.2617300239135814, + "grad_norm": 2.0248396396636963, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.769244372844696, + "num_tokens": 714309558.0, + "step": 27428 + }, + { + "epoch": 2.2618124845386327, + "grad_norm": 2.0361149311065674, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.7883340120315552, + "num_tokens": 714340294.0, + "step": 27429 + }, + { + "epoch": 2.2618949451636845, + "grad_norm": 2.271373748779297, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.7955046892166138, + "num_tokens": 714365429.0, + "step": 27430 + }, + { + "epoch": 2.261977405788736, + "grad_norm": 2.156768321990967, + "learning_rate": 1e-06, + "loss": 0.6012, + "mean_token_accuracy": 0.7997155785560608, + "num_tokens": 714394114.0, + "step": 27431 + }, + { + "epoch": 2.2620598664137876, + "grad_norm": 2.1582958698272705, + "learning_rate": 1e-06, + "loss": 0.8127, + "mean_token_accuracy": 0.73683100938797, + "num_tokens": 714420844.0, + "step": 27432 + }, + { + "epoch": 2.262142327038839, + "grad_norm": 1.8753141164779663, + "learning_rate": 1e-06, + "loss": 0.5764, + "mean_token_accuracy": 0.8172051906585693, + "num_tokens": 714457254.0, + "step": 27433 + }, + { + "epoch": 2.2622247876638903, + "grad_norm": 1.8669010400772095, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7770532965660095, + "num_tokens": 714494658.0, + "step": 27434 + }, + { + "epoch": 2.262307248288942, + "grad_norm": 2.0501251220703125, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.798544704914093, + "num_tokens": 714526107.0, + "step": 27435 + }, + { + "epoch": 2.2623897089139935, + "grad_norm": 2.085484027862549, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7206779718399048, + "num_tokens": 714556266.0, + "step": 27436 + }, + { + "epoch": 2.2624721695390453, + "grad_norm": 2.713768243789673, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7629022002220154, + "num_tokens": 714576198.0, + "step": 27437 + }, + { + "epoch": 2.2625546301640966, + "grad_norm": 2.1790428161621094, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.7819822430610657, + "num_tokens": 714604554.0, + "step": 27438 + }, + { + "epoch": 2.2626370907891484, + "grad_norm": 2.093794345855713, + "learning_rate": 1e-06, + "loss": 0.5736, + "mean_token_accuracy": 0.8060002326965332, + "num_tokens": 714634038.0, + "step": 27439 + }, + { + "epoch": 2.2627195514142, + "grad_norm": 1.808206558227539, + "learning_rate": 1e-06, + "loss": 0.464, + "mean_token_accuracy": 0.839409589767456, + "num_tokens": 714667788.0, + "step": 27440 + }, + { + "epoch": 2.262802012039251, + "grad_norm": 2.2503154277801514, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7299866676330566, + "num_tokens": 714696613.0, + "step": 27441 + }, + { + "epoch": 2.262884472664303, + "grad_norm": 2.0873711109161377, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7540580630302429, + "num_tokens": 714727563.0, + "step": 27442 + }, + { + "epoch": 2.2629669332893543, + "grad_norm": 1.8942595720291138, + "learning_rate": 1e-06, + "loss": 0.595, + "mean_token_accuracy": 0.793919563293457, + "num_tokens": 714762003.0, + "step": 27443 + }, + { + "epoch": 2.2630493939144056, + "grad_norm": 1.9637354612350464, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7731912732124329, + "num_tokens": 714794745.0, + "step": 27444 + }, + { + "epoch": 2.2631318545394574, + "grad_norm": 2.409217596054077, + "learning_rate": 1e-06, + "loss": 0.6128, + "mean_token_accuracy": 0.7861943244934082, + "num_tokens": 714816221.0, + "step": 27445 + }, + { + "epoch": 2.2632143151645088, + "grad_norm": 2.1366512775421143, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.761352002620697, + "num_tokens": 714845244.0, + "step": 27446 + }, + { + "epoch": 2.2632967757895606, + "grad_norm": 2.0087058544158936, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7595709562301636, + "num_tokens": 714875620.0, + "step": 27447 + }, + { + "epoch": 2.263379236414612, + "grad_norm": 2.374941349029541, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.771770715713501, + "num_tokens": 714905146.0, + "step": 27448 + }, + { + "epoch": 2.2634616970396637, + "grad_norm": 2.1336746215820312, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7210409045219421, + "num_tokens": 714936115.0, + "step": 27449 + }, + { + "epoch": 2.263544157664715, + "grad_norm": 2.084411382675171, + "learning_rate": 1e-06, + "loss": 0.5876, + "mean_token_accuracy": 0.8099023103713989, + "num_tokens": 714964934.0, + "step": 27450 + }, + { + "epoch": 2.2636266182897664, + "grad_norm": 1.9723011255264282, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7419880628585815, + "num_tokens": 715002324.0, + "step": 27451 + }, + { + "epoch": 2.263709078914818, + "grad_norm": 2.0607056617736816, + "learning_rate": 1e-06, + "loss": 0.7598, + "mean_token_accuracy": 0.7592403888702393, + "num_tokens": 715034792.0, + "step": 27452 + }, + { + "epoch": 2.2637915395398696, + "grad_norm": 2.00274658203125, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.7828851938247681, + "num_tokens": 715069323.0, + "step": 27453 + }, + { + "epoch": 2.2638740001649214, + "grad_norm": 2.135529041290283, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7721606492996216, + "num_tokens": 715098885.0, + "step": 27454 + }, + { + "epoch": 2.2639564607899727, + "grad_norm": 2.1146793365478516, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7808972597122192, + "num_tokens": 715129991.0, + "step": 27455 + }, + { + "epoch": 2.2640389214150245, + "grad_norm": 2.1016347408294678, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7675153017044067, + "num_tokens": 715158379.0, + "step": 27456 + }, + { + "epoch": 2.264121382040076, + "grad_norm": 1.9453327655792236, + "learning_rate": 1e-06, + "loss": 0.5316, + "mean_token_accuracy": 0.8114795088768005, + "num_tokens": 715188125.0, + "step": 27457 + }, + { + "epoch": 2.264203842665127, + "grad_norm": 2.2253670692443848, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7202587127685547, + "num_tokens": 715219619.0, + "step": 27458 + }, + { + "epoch": 2.264286303290179, + "grad_norm": 2.062871217727661, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.7846797108650208, + "num_tokens": 715250331.0, + "step": 27459 + }, + { + "epoch": 2.2643687639152303, + "grad_norm": 2.1476922035217285, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.7927575707435608, + "num_tokens": 715278113.0, + "step": 27460 + }, + { + "epoch": 2.264451224540282, + "grad_norm": 2.3514513969421387, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.7793378233909607, + "num_tokens": 715301852.0, + "step": 27461 + }, + { + "epoch": 2.2645336851653335, + "grad_norm": 2.014535903930664, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7779014110565186, + "num_tokens": 715331700.0, + "step": 27462 + }, + { + "epoch": 2.2646161457903853, + "grad_norm": 2.1355936527252197, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7486820220947266, + "num_tokens": 715359732.0, + "step": 27463 + }, + { + "epoch": 2.2646986064154366, + "grad_norm": 2.162139654159546, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7473152279853821, + "num_tokens": 715388234.0, + "step": 27464 + }, + { + "epoch": 2.264781067040488, + "grad_norm": 1.9069880247116089, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.7789004445075989, + "num_tokens": 715421365.0, + "step": 27465 + }, + { + "epoch": 2.2648635276655398, + "grad_norm": 2.3404054641723633, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7407662272453308, + "num_tokens": 715447978.0, + "step": 27466 + }, + { + "epoch": 2.264945988290591, + "grad_norm": 1.8489078283309937, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.754729151725769, + "num_tokens": 715486126.0, + "step": 27467 + }, + { + "epoch": 2.265028448915643, + "grad_norm": 2.2361576557159424, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7698428630828857, + "num_tokens": 715516593.0, + "step": 27468 + }, + { + "epoch": 2.2651109095406943, + "grad_norm": 2.2354650497436523, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7587823867797852, + "num_tokens": 715541882.0, + "step": 27469 + }, + { + "epoch": 2.265193370165746, + "grad_norm": 2.022148847579956, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7665456533432007, + "num_tokens": 715575011.0, + "step": 27470 + }, + { + "epoch": 2.2652758307907974, + "grad_norm": 2.280081033706665, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.7887415289878845, + "num_tokens": 715597904.0, + "step": 27471 + }, + { + "epoch": 2.2653582914158488, + "grad_norm": 2.101562976837158, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7630101442337036, + "num_tokens": 715626833.0, + "step": 27472 + }, + { + "epoch": 2.2654407520409006, + "grad_norm": 1.9430670738220215, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7754760980606079, + "num_tokens": 715660813.0, + "step": 27473 + }, + { + "epoch": 2.265523212665952, + "grad_norm": 2.2299835681915283, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7613463401794434, + "num_tokens": 715689588.0, + "step": 27474 + }, + { + "epoch": 2.2656056732910037, + "grad_norm": 2.09460186958313, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7420664429664612, + "num_tokens": 715721131.0, + "step": 27475 + }, + { + "epoch": 2.265688133916055, + "grad_norm": 2.391037940979004, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7465813755989075, + "num_tokens": 715746404.0, + "step": 27476 + }, + { + "epoch": 2.265770594541107, + "grad_norm": 2.235790729522705, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7868382930755615, + "num_tokens": 715773231.0, + "step": 27477 + }, + { + "epoch": 2.265853055166158, + "grad_norm": 2.0819571018218994, + "learning_rate": 1e-06, + "loss": 0.7775, + "mean_token_accuracy": 0.7570417523384094, + "num_tokens": 715806837.0, + "step": 27478 + }, + { + "epoch": 2.2659355157912096, + "grad_norm": 2.3714752197265625, + "learning_rate": 1e-06, + "loss": 0.5778, + "mean_token_accuracy": 0.8038418889045715, + "num_tokens": 715831916.0, + "step": 27479 + }, + { + "epoch": 2.2660179764162613, + "grad_norm": 2.275424003601074, + "learning_rate": 1e-06, + "loss": 0.5798, + "mean_token_accuracy": 0.8098140358924866, + "num_tokens": 715858433.0, + "step": 27480 + }, + { + "epoch": 2.2661004370413127, + "grad_norm": 2.140887975692749, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7550699710845947, + "num_tokens": 715888416.0, + "step": 27481 + }, + { + "epoch": 2.2661828976663645, + "grad_norm": 2.1267073154449463, + "learning_rate": 1e-06, + "loss": 0.791, + "mean_token_accuracy": 0.736079216003418, + "num_tokens": 715919134.0, + "step": 27482 + }, + { + "epoch": 2.266265358291416, + "grad_norm": 1.9246299266815186, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7716363072395325, + "num_tokens": 715952834.0, + "step": 27483 + }, + { + "epoch": 2.266347818916467, + "grad_norm": 2.04510760307312, + "learning_rate": 1e-06, + "loss": 0.5971, + "mean_token_accuracy": 0.7921957969665527, + "num_tokens": 715986092.0, + "step": 27484 + }, + { + "epoch": 2.266430279541519, + "grad_norm": 2.270781993865967, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.7852001786231995, + "num_tokens": 716010026.0, + "step": 27485 + }, + { + "epoch": 2.2665127401665703, + "grad_norm": 2.252699136734009, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.7849061489105225, + "num_tokens": 716038961.0, + "step": 27486 + }, + { + "epoch": 2.266595200791622, + "grad_norm": 1.9202312231063843, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7786864638328552, + "num_tokens": 716072237.0, + "step": 27487 + }, + { + "epoch": 2.2666776614166735, + "grad_norm": 1.8965399265289307, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7805807590484619, + "num_tokens": 716110100.0, + "step": 27488 + }, + { + "epoch": 2.266760122041725, + "grad_norm": 2.0494916439056396, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.7857598066329956, + "num_tokens": 716138762.0, + "step": 27489 + }, + { + "epoch": 2.2668425826667766, + "grad_norm": 1.8478631973266602, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7600724697113037, + "num_tokens": 716176033.0, + "step": 27490 + }, + { + "epoch": 2.266925043291828, + "grad_norm": 2.0169334411621094, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.7855464816093445, + "num_tokens": 716208954.0, + "step": 27491 + }, + { + "epoch": 2.2670075039168798, + "grad_norm": 2.306959629058838, + "learning_rate": 1e-06, + "loss": 0.772, + "mean_token_accuracy": 0.7528901100158691, + "num_tokens": 716234901.0, + "step": 27492 + }, + { + "epoch": 2.267089964541931, + "grad_norm": 1.8611668348312378, + "learning_rate": 1e-06, + "loss": 0.5062, + "mean_token_accuracy": 0.8313730955123901, + "num_tokens": 716267763.0, + "step": 27493 + }, + { + "epoch": 2.267172425166983, + "grad_norm": 2.070683002471924, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7705582976341248, + "num_tokens": 716299850.0, + "step": 27494 + }, + { + "epoch": 2.2672548857920343, + "grad_norm": 2.1133878231048584, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7681301236152649, + "num_tokens": 716328223.0, + "step": 27495 + }, + { + "epoch": 2.2673373464170856, + "grad_norm": 2.0810933113098145, + "learning_rate": 1e-06, + "loss": 0.573, + "mean_token_accuracy": 0.8132724761962891, + "num_tokens": 716356326.0, + "step": 27496 + }, + { + "epoch": 2.2674198070421374, + "grad_norm": 1.870496153831482, + "learning_rate": 1e-06, + "loss": 0.5966, + "mean_token_accuracy": 0.7899651527404785, + "num_tokens": 716390720.0, + "step": 27497 + }, + { + "epoch": 2.2675022676671888, + "grad_norm": 2.269646406173706, + "learning_rate": 1e-06, + "loss": 0.5769, + "mean_token_accuracy": 0.7976590991020203, + "num_tokens": 716418119.0, + "step": 27498 + }, + { + "epoch": 2.2675847282922406, + "grad_norm": 2.1001760959625244, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.7885850667953491, + "num_tokens": 716447737.0, + "step": 27499 + }, + { + "epoch": 2.267667188917292, + "grad_norm": 2.50750732421875, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7604295015335083, + "num_tokens": 716470616.0, + "step": 27500 + }, + { + "epoch": 2.2677496495423437, + "grad_norm": 2.3354339599609375, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7677159309387207, + "num_tokens": 716494566.0, + "step": 27501 + }, + { + "epoch": 2.267832110167395, + "grad_norm": 2.0521838665008545, + "learning_rate": 1e-06, + "loss": 0.5984, + "mean_token_accuracy": 0.8021509647369385, + "num_tokens": 716525609.0, + "step": 27502 + }, + { + "epoch": 2.2679145707924464, + "grad_norm": 2.175014019012451, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7303756475448608, + "num_tokens": 716557898.0, + "step": 27503 + }, + { + "epoch": 2.267997031417498, + "grad_norm": 2.313668727874756, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.7866759300231934, + "num_tokens": 716585438.0, + "step": 27504 + }, + { + "epoch": 2.2680794920425495, + "grad_norm": 2.1498305797576904, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7749454975128174, + "num_tokens": 716614633.0, + "step": 27505 + }, + { + "epoch": 2.2681619526676013, + "grad_norm": 2.102238178253174, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7440524101257324, + "num_tokens": 716644883.0, + "step": 27506 + }, + { + "epoch": 2.2682444132926527, + "grad_norm": 1.8716734647750854, + "learning_rate": 1e-06, + "loss": 0.5276, + "mean_token_accuracy": 0.8207179307937622, + "num_tokens": 716678588.0, + "step": 27507 + }, + { + "epoch": 2.2683268739177045, + "grad_norm": 2.0492870807647705, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.768273115158081, + "num_tokens": 716712996.0, + "step": 27508 + }, + { + "epoch": 2.268409334542756, + "grad_norm": 2.2798972129821777, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7838017344474792, + "num_tokens": 716737616.0, + "step": 27509 + }, + { + "epoch": 2.268491795167807, + "grad_norm": 2.2033891677856445, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7283511161804199, + "num_tokens": 716766917.0, + "step": 27510 + }, + { + "epoch": 2.268574255792859, + "grad_norm": 2.00781512260437, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7733043432235718, + "num_tokens": 716800207.0, + "step": 27511 + }, + { + "epoch": 2.2686567164179103, + "grad_norm": 2.096998691558838, + "learning_rate": 1e-06, + "loss": 0.5952, + "mean_token_accuracy": 0.7962692379951477, + "num_tokens": 716830059.0, + "step": 27512 + }, + { + "epoch": 2.268739177042962, + "grad_norm": 1.968390941619873, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7811819314956665, + "num_tokens": 716861739.0, + "step": 27513 + }, + { + "epoch": 2.2688216376680135, + "grad_norm": 1.9624896049499512, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.7521873116493225, + "num_tokens": 716897202.0, + "step": 27514 + }, + { + "epoch": 2.2689040982930653, + "grad_norm": 2.09055757522583, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.783182680606842, + "num_tokens": 716928686.0, + "step": 27515 + }, + { + "epoch": 2.2689865589181166, + "grad_norm": 2.1042845249176025, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7655736207962036, + "num_tokens": 716959186.0, + "step": 27516 + }, + { + "epoch": 2.269069019543168, + "grad_norm": 2.139915943145752, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7510694861412048, + "num_tokens": 716987928.0, + "step": 27517 + }, + { + "epoch": 2.2691514801682198, + "grad_norm": 2.0415055751800537, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7500304579734802, + "num_tokens": 717020490.0, + "step": 27518 + }, + { + "epoch": 2.269233940793271, + "grad_norm": 2.1282262802124023, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7826303243637085, + "num_tokens": 717047916.0, + "step": 27519 + }, + { + "epoch": 2.269316401418323, + "grad_norm": 2.140392541885376, + "learning_rate": 1e-06, + "loss": 0.5899, + "mean_token_accuracy": 0.8033854365348816, + "num_tokens": 717077804.0, + "step": 27520 + }, + { + "epoch": 2.2693988620433743, + "grad_norm": 1.800060510635376, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.7985411882400513, + "num_tokens": 717114275.0, + "step": 27521 + }, + { + "epoch": 2.269481322668426, + "grad_norm": 2.0568621158599854, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.7918179631233215, + "num_tokens": 717144504.0, + "step": 27522 + }, + { + "epoch": 2.2695637832934774, + "grad_norm": 2.1605074405670166, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7562731504440308, + "num_tokens": 717172516.0, + "step": 27523 + }, + { + "epoch": 2.2696462439185288, + "grad_norm": 2.197939395904541, + "learning_rate": 1e-06, + "loss": 0.7888, + "mean_token_accuracy": 0.7434967756271362, + "num_tokens": 717202271.0, + "step": 27524 + }, + { + "epoch": 2.2697287045435806, + "grad_norm": 1.8895034790039062, + "learning_rate": 1e-06, + "loss": 0.6032, + "mean_token_accuracy": 0.8057327270507812, + "num_tokens": 717240787.0, + "step": 27525 + }, + { + "epoch": 2.269811165168632, + "grad_norm": 2.2037816047668457, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.7881407737731934, + "num_tokens": 717269071.0, + "step": 27526 + }, + { + "epoch": 2.2698936257936837, + "grad_norm": 1.9225023984909058, + "learning_rate": 1e-06, + "loss": 0.6132, + "mean_token_accuracy": 0.8031479120254517, + "num_tokens": 717301299.0, + "step": 27527 + }, + { + "epoch": 2.269976086418735, + "grad_norm": 2.1337547302246094, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.7858302593231201, + "num_tokens": 717327943.0, + "step": 27528 + }, + { + "epoch": 2.2700585470437864, + "grad_norm": 2.279115676879883, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7626189589500427, + "num_tokens": 717352896.0, + "step": 27529 + }, + { + "epoch": 2.270141007668838, + "grad_norm": 2.1128740310668945, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.770967960357666, + "num_tokens": 717384312.0, + "step": 27530 + }, + { + "epoch": 2.2702234682938895, + "grad_norm": 1.9404314756393433, + "learning_rate": 1e-06, + "loss": 0.4972, + "mean_token_accuracy": 0.8283739686012268, + "num_tokens": 717413952.0, + "step": 27531 + }, + { + "epoch": 2.2703059289189413, + "grad_norm": 2.1200203895568848, + "learning_rate": 1e-06, + "loss": 0.6011, + "mean_token_accuracy": 0.796772301197052, + "num_tokens": 717442556.0, + "step": 27532 + }, + { + "epoch": 2.2703883895439927, + "grad_norm": 2.0194625854492188, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7659658789634705, + "num_tokens": 717472983.0, + "step": 27533 + }, + { + "epoch": 2.270470850169044, + "grad_norm": 2.0684757232666016, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7799934148788452, + "num_tokens": 717502214.0, + "step": 27534 + }, + { + "epoch": 2.270553310794096, + "grad_norm": 2.3632802963256836, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7312134504318237, + "num_tokens": 717526417.0, + "step": 27535 + }, + { + "epoch": 2.270635771419147, + "grad_norm": 2.279947280883789, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7780401706695557, + "num_tokens": 717554609.0, + "step": 27536 + }, + { + "epoch": 2.270718232044199, + "grad_norm": 2.116063356399536, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.7765398621559143, + "num_tokens": 717584283.0, + "step": 27537 + }, + { + "epoch": 2.2708006926692503, + "grad_norm": 2.524975299835205, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.7916113138198853, + "num_tokens": 717605928.0, + "step": 27538 + }, + { + "epoch": 2.270883153294302, + "grad_norm": 2.262683629989624, + "learning_rate": 1e-06, + "loss": 0.4884, + "mean_token_accuracy": 0.8279473781585693, + "num_tokens": 717631021.0, + "step": 27539 + }, + { + "epoch": 2.2709656139193535, + "grad_norm": 2.1392199993133545, + "learning_rate": 1e-06, + "loss": 0.6165, + "mean_token_accuracy": 0.7939868569374084, + "num_tokens": 717659466.0, + "step": 27540 + }, + { + "epoch": 2.271048074544405, + "grad_norm": 2.142693042755127, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7874128818511963, + "num_tokens": 717687570.0, + "step": 27541 + }, + { + "epoch": 2.2711305351694566, + "grad_norm": 1.9219770431518555, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7549504041671753, + "num_tokens": 717723573.0, + "step": 27542 + }, + { + "epoch": 2.271212995794508, + "grad_norm": 2.446974515914917, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7546406984329224, + "num_tokens": 717749315.0, + "step": 27543 + }, + { + "epoch": 2.2712954564195598, + "grad_norm": 2.25479793548584, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7259631156921387, + "num_tokens": 717781662.0, + "step": 27544 + }, + { + "epoch": 2.271377917044611, + "grad_norm": 2.1848883628845215, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.7961799502372742, + "num_tokens": 717811595.0, + "step": 27545 + }, + { + "epoch": 2.271460377669663, + "grad_norm": 1.920183777809143, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.7859628200531006, + "num_tokens": 717846806.0, + "step": 27546 + }, + { + "epoch": 2.2715428382947143, + "grad_norm": 1.9799836874008179, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.7933622598648071, + "num_tokens": 717880349.0, + "step": 27547 + }, + { + "epoch": 2.2716252989197656, + "grad_norm": 2.108786106109619, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.7956632971763611, + "num_tokens": 717909550.0, + "step": 27548 + }, + { + "epoch": 2.2717077595448174, + "grad_norm": 2.0302603244781494, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7723807096481323, + "num_tokens": 717941585.0, + "step": 27549 + }, + { + "epoch": 2.2717902201698688, + "grad_norm": 2.064565658569336, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7874561548233032, + "num_tokens": 717970740.0, + "step": 27550 + }, + { + "epoch": 2.2718726807949206, + "grad_norm": 1.915695309638977, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7630355954170227, + "num_tokens": 718005663.0, + "step": 27551 + }, + { + "epoch": 2.271955141419972, + "grad_norm": 1.9411355257034302, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7718074321746826, + "num_tokens": 718040483.0, + "step": 27552 + }, + { + "epoch": 2.2720376020450237, + "grad_norm": 2.3495380878448486, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7742375135421753, + "num_tokens": 718065345.0, + "step": 27553 + }, + { + "epoch": 2.272120062670075, + "grad_norm": 2.2759201526641846, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7767736911773682, + "num_tokens": 718090455.0, + "step": 27554 + }, + { + "epoch": 2.2722025232951264, + "grad_norm": 2.0470571517944336, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.731310248374939, + "num_tokens": 718122546.0, + "step": 27555 + }, + { + "epoch": 2.272284983920178, + "grad_norm": 2.138652801513672, + "learning_rate": 1e-06, + "loss": 0.6016, + "mean_token_accuracy": 0.7774980068206787, + "num_tokens": 718150040.0, + "step": 27556 + }, + { + "epoch": 2.2723674445452295, + "grad_norm": 1.8178116083145142, + "learning_rate": 1e-06, + "loss": 0.5944, + "mean_token_accuracy": 0.7958757877349854, + "num_tokens": 718186310.0, + "step": 27557 + }, + { + "epoch": 2.2724499051702813, + "grad_norm": 2.239452600479126, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7484610080718994, + "num_tokens": 718211855.0, + "step": 27558 + }, + { + "epoch": 2.2725323657953327, + "grad_norm": 2.3073949813842773, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7661162614822388, + "num_tokens": 718236339.0, + "step": 27559 + }, + { + "epoch": 2.2726148264203845, + "grad_norm": 1.839120864868164, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7729429602622986, + "num_tokens": 718271343.0, + "step": 27560 + }, + { + "epoch": 2.272697287045436, + "grad_norm": 2.3781957626342773, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7804135680198669, + "num_tokens": 718296399.0, + "step": 27561 + }, + { + "epoch": 2.272779747670487, + "grad_norm": 1.973427414894104, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7895632386207581, + "num_tokens": 718330548.0, + "step": 27562 + }, + { + "epoch": 2.272862208295539, + "grad_norm": 1.8174842596054077, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.7796400785446167, + "num_tokens": 718369719.0, + "step": 27563 + }, + { + "epoch": 2.2729446689205903, + "grad_norm": 2.341796636581421, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7750040292739868, + "num_tokens": 718396021.0, + "step": 27564 + }, + { + "epoch": 2.273027129545642, + "grad_norm": 2.109733819961548, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.7832196950912476, + "num_tokens": 718426487.0, + "step": 27565 + }, + { + "epoch": 2.2731095901706935, + "grad_norm": 1.9696706533432007, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.7840867638587952, + "num_tokens": 718459973.0, + "step": 27566 + }, + { + "epoch": 2.2731920507957453, + "grad_norm": 2.162301778793335, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7754107713699341, + "num_tokens": 718488865.0, + "step": 27567 + }, + { + "epoch": 2.2732745114207966, + "grad_norm": 2.304072618484497, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7756454348564148, + "num_tokens": 718514112.0, + "step": 27568 + }, + { + "epoch": 2.273356972045848, + "grad_norm": 2.097095012664795, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7749303579330444, + "num_tokens": 718546033.0, + "step": 27569 + }, + { + "epoch": 2.2734394326708998, + "grad_norm": 1.9211008548736572, + "learning_rate": 1e-06, + "loss": 0.4997, + "mean_token_accuracy": 0.8182188272476196, + "num_tokens": 718579972.0, + "step": 27570 + }, + { + "epoch": 2.273521893295951, + "grad_norm": 1.9789444208145142, + "learning_rate": 1e-06, + "loss": 0.5938, + "mean_token_accuracy": 0.7935935854911804, + "num_tokens": 718614761.0, + "step": 27571 + }, + { + "epoch": 2.273604353921003, + "grad_norm": 2.096879720687866, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7696502208709717, + "num_tokens": 718646910.0, + "step": 27572 + }, + { + "epoch": 2.2736868145460543, + "grad_norm": 2.6732213497161865, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7505860328674316, + "num_tokens": 718666790.0, + "step": 27573 + }, + { + "epoch": 2.2737692751711056, + "grad_norm": 2.0191116333007812, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7794046998023987, + "num_tokens": 718697465.0, + "step": 27574 + }, + { + "epoch": 2.2738517357961574, + "grad_norm": 2.167280912399292, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.7894166707992554, + "num_tokens": 718726196.0, + "step": 27575 + }, + { + "epoch": 2.2739341964212088, + "grad_norm": 2.1118690967559814, + "learning_rate": 1e-06, + "loss": 0.555, + "mean_token_accuracy": 0.8139329552650452, + "num_tokens": 718753522.0, + "step": 27576 + }, + { + "epoch": 2.2740166570462605, + "grad_norm": 2.130642890930176, + "learning_rate": 1e-06, + "loss": 0.5586, + "mean_token_accuracy": 0.8131107091903687, + "num_tokens": 718779843.0, + "step": 27577 + }, + { + "epoch": 2.274099117671312, + "grad_norm": 2.2430927753448486, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7782146334648132, + "num_tokens": 718808789.0, + "step": 27578 + }, + { + "epoch": 2.2741815782963632, + "grad_norm": 2.0192039012908936, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.766703724861145, + "num_tokens": 718840324.0, + "step": 27579 + }, + { + "epoch": 2.274264038921415, + "grad_norm": 2.3166964054107666, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7336946725845337, + "num_tokens": 718866406.0, + "step": 27580 + }, + { + "epoch": 2.2743464995464664, + "grad_norm": 2.0033600330352783, + "learning_rate": 1e-06, + "loss": 0.5345, + "mean_token_accuracy": 0.8063005208969116, + "num_tokens": 718894925.0, + "step": 27581 + }, + { + "epoch": 2.274428960171518, + "grad_norm": 2.2173614501953125, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7351394891738892, + "num_tokens": 718923806.0, + "step": 27582 + }, + { + "epoch": 2.2745114207965695, + "grad_norm": 2.2084412574768066, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.7816874980926514, + "num_tokens": 718950993.0, + "step": 27583 + }, + { + "epoch": 2.2745938814216213, + "grad_norm": 2.208799123764038, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.7489978075027466, + "num_tokens": 718978647.0, + "step": 27584 + }, + { + "epoch": 2.2746763420466727, + "grad_norm": 2.045846700668335, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7420142889022827, + "num_tokens": 719012253.0, + "step": 27585 + }, + { + "epoch": 2.274758802671724, + "grad_norm": 1.972514033317566, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7403379678726196, + "num_tokens": 719046116.0, + "step": 27586 + }, + { + "epoch": 2.274841263296776, + "grad_norm": 2.2507259845733643, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7626957893371582, + "num_tokens": 719069467.0, + "step": 27587 + }, + { + "epoch": 2.274923723921827, + "grad_norm": 1.9629454612731934, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7778953313827515, + "num_tokens": 719099685.0, + "step": 27588 + }, + { + "epoch": 2.275006184546879, + "grad_norm": 2.1608684062957764, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7573223114013672, + "num_tokens": 719132068.0, + "step": 27589 + }, + { + "epoch": 2.2750886451719303, + "grad_norm": 2.1695120334625244, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.7775005102157593, + "num_tokens": 719159941.0, + "step": 27590 + }, + { + "epoch": 2.275171105796982, + "grad_norm": 2.3299314975738525, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.7847377061843872, + "num_tokens": 719183808.0, + "step": 27591 + }, + { + "epoch": 2.2752535664220335, + "grad_norm": 2.1054422855377197, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7696621417999268, + "num_tokens": 719213377.0, + "step": 27592 + }, + { + "epoch": 2.275336027047085, + "grad_norm": 1.8666480779647827, + "learning_rate": 1e-06, + "loss": 0.5781, + "mean_token_accuracy": 0.7941597700119019, + "num_tokens": 719247470.0, + "step": 27593 + }, + { + "epoch": 2.2754184876721366, + "grad_norm": 1.9111136198043823, + "learning_rate": 1e-06, + "loss": 0.5713, + "mean_token_accuracy": 0.8178349733352661, + "num_tokens": 719277958.0, + "step": 27594 + }, + { + "epoch": 2.275500948297188, + "grad_norm": 2.134432077407837, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7689868211746216, + "num_tokens": 719307952.0, + "step": 27595 + }, + { + "epoch": 2.2755834089222398, + "grad_norm": 2.0027873516082764, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7770949602127075, + "num_tokens": 719341052.0, + "step": 27596 + }, + { + "epoch": 2.275665869547291, + "grad_norm": 1.9900624752044678, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7781080007553101, + "num_tokens": 719375469.0, + "step": 27597 + }, + { + "epoch": 2.275748330172343, + "grad_norm": 2.3071727752685547, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7642102837562561, + "num_tokens": 719401523.0, + "step": 27598 + }, + { + "epoch": 2.2758307907973943, + "grad_norm": 1.9164119958877563, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.7974525690078735, + "num_tokens": 719438035.0, + "step": 27599 + }, + { + "epoch": 2.2759132514224456, + "grad_norm": 1.9939196109771729, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7444711327552795, + "num_tokens": 719471377.0, + "step": 27600 + }, + { + "epoch": 2.2759957120474974, + "grad_norm": 2.3135173320770264, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7617177963256836, + "num_tokens": 719497296.0, + "step": 27601 + }, + { + "epoch": 2.2760781726725487, + "grad_norm": 2.374783754348755, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7530407905578613, + "num_tokens": 719524086.0, + "step": 27602 + }, + { + "epoch": 2.2761606332976005, + "grad_norm": 1.8835101127624512, + "learning_rate": 1e-06, + "loss": 0.575, + "mean_token_accuracy": 0.7998480796813965, + "num_tokens": 719557157.0, + "step": 27603 + }, + { + "epoch": 2.276243093922652, + "grad_norm": 2.2841134071350098, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.761726438999176, + "num_tokens": 719584959.0, + "step": 27604 + }, + { + "epoch": 2.2763255545477037, + "grad_norm": 2.1091766357421875, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7604353427886963, + "num_tokens": 719615592.0, + "step": 27605 + }, + { + "epoch": 2.276408015172755, + "grad_norm": 2.3982293605804443, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.7884845733642578, + "num_tokens": 719641089.0, + "step": 27606 + }, + { + "epoch": 2.2764904757978064, + "grad_norm": 2.0306055545806885, + "learning_rate": 1e-06, + "loss": 0.5725, + "mean_token_accuracy": 0.8050772547721863, + "num_tokens": 719672566.0, + "step": 27607 + }, + { + "epoch": 2.276572936422858, + "grad_norm": 2.2612905502319336, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.738399863243103, + "num_tokens": 719699022.0, + "step": 27608 + }, + { + "epoch": 2.2766553970479095, + "grad_norm": 2.0239710807800293, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7811428308486938, + "num_tokens": 719732699.0, + "step": 27609 + }, + { + "epoch": 2.2767378576729613, + "grad_norm": 1.9530636072158813, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7326911687850952, + "num_tokens": 719765343.0, + "step": 27610 + }, + { + "epoch": 2.2768203182980127, + "grad_norm": 2.395097494125366, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7348734140396118, + "num_tokens": 719792992.0, + "step": 27611 + }, + { + "epoch": 2.2769027789230645, + "grad_norm": 2.0445449352264404, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7707247734069824, + "num_tokens": 719823528.0, + "step": 27612 + }, + { + "epoch": 2.276985239548116, + "grad_norm": 2.1543569564819336, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.7923346161842346, + "num_tokens": 719850417.0, + "step": 27613 + }, + { + "epoch": 2.277067700173167, + "grad_norm": 1.8912451267242432, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7657463550567627, + "num_tokens": 719887499.0, + "step": 27614 + }, + { + "epoch": 2.277150160798219, + "grad_norm": 2.3395845890045166, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7563239336013794, + "num_tokens": 719912942.0, + "step": 27615 + }, + { + "epoch": 2.2772326214232703, + "grad_norm": 2.329545497894287, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7768383026123047, + "num_tokens": 719939588.0, + "step": 27616 + }, + { + "epoch": 2.277315082048322, + "grad_norm": 1.8810468912124634, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7583544850349426, + "num_tokens": 719975830.0, + "step": 27617 + }, + { + "epoch": 2.2773975426733735, + "grad_norm": 2.1583008766174316, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7693177461624146, + "num_tokens": 720007465.0, + "step": 27618 + }, + { + "epoch": 2.277480003298425, + "grad_norm": 2.4088704586029053, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7707360982894897, + "num_tokens": 720034858.0, + "step": 27619 + }, + { + "epoch": 2.2775624639234766, + "grad_norm": 2.100607395172119, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7811070680618286, + "num_tokens": 720067026.0, + "step": 27620 + }, + { + "epoch": 2.277644924548528, + "grad_norm": 2.1304659843444824, + "learning_rate": 1e-06, + "loss": 0.597, + "mean_token_accuracy": 0.8011950850486755, + "num_tokens": 720095517.0, + "step": 27621 + }, + { + "epoch": 2.2777273851735798, + "grad_norm": 2.1261367797851562, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.7948839664459229, + "num_tokens": 720125236.0, + "step": 27622 + }, + { + "epoch": 2.277809845798631, + "grad_norm": 2.195315361022949, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7696593403816223, + "num_tokens": 720151505.0, + "step": 27623 + }, + { + "epoch": 2.2778923064236825, + "grad_norm": 2.137092113494873, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7401527762413025, + "num_tokens": 720182859.0, + "step": 27624 + }, + { + "epoch": 2.2779747670487343, + "grad_norm": 1.9669098854064941, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7610570192337036, + "num_tokens": 720216884.0, + "step": 27625 + }, + { + "epoch": 2.2780572276737856, + "grad_norm": 2.0807864665985107, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7419419884681702, + "num_tokens": 720249305.0, + "step": 27626 + }, + { + "epoch": 2.2781396882988374, + "grad_norm": 1.988494634628296, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7380189895629883, + "num_tokens": 720283937.0, + "step": 27627 + }, + { + "epoch": 2.2782221489238887, + "grad_norm": 1.8628987073898315, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7811501026153564, + "num_tokens": 720318548.0, + "step": 27628 + }, + { + "epoch": 2.2783046095489405, + "grad_norm": 1.964849829673767, + "learning_rate": 1e-06, + "loss": 0.5911, + "mean_token_accuracy": 0.7897382974624634, + "num_tokens": 720350194.0, + "step": 27629 + }, + { + "epoch": 2.278387070173992, + "grad_norm": 2.0431771278381348, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.757284939289093, + "num_tokens": 720382100.0, + "step": 27630 + }, + { + "epoch": 2.2784695307990432, + "grad_norm": 2.1881115436553955, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.779084324836731, + "num_tokens": 720411139.0, + "step": 27631 + }, + { + "epoch": 2.278551991424095, + "grad_norm": 2.223656415939331, + "learning_rate": 1e-06, + "loss": 0.5559, + "mean_token_accuracy": 0.8141682744026184, + "num_tokens": 720437776.0, + "step": 27632 + }, + { + "epoch": 2.2786344520491464, + "grad_norm": 2.017353057861328, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7454587817192078, + "num_tokens": 720471817.0, + "step": 27633 + }, + { + "epoch": 2.278716912674198, + "grad_norm": 2.4672210216522217, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7618021368980408, + "num_tokens": 720493766.0, + "step": 27634 + }, + { + "epoch": 2.2787993732992495, + "grad_norm": 2.033780336380005, + "learning_rate": 1e-06, + "loss": 0.5367, + "mean_token_accuracy": 0.8148427605628967, + "num_tokens": 720520282.0, + "step": 27635 + }, + { + "epoch": 2.2788818339243013, + "grad_norm": 2.5094449520111084, + "learning_rate": 1e-06, + "loss": 0.586, + "mean_token_accuracy": 0.7968717813491821, + "num_tokens": 720543398.0, + "step": 27636 + }, + { + "epoch": 2.2789642945493527, + "grad_norm": 2.3696744441986084, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.751541018486023, + "num_tokens": 720569461.0, + "step": 27637 + }, + { + "epoch": 2.279046755174404, + "grad_norm": 2.179948568344116, + "learning_rate": 1e-06, + "loss": 0.6382, + "mean_token_accuracy": 0.7833648920059204, + "num_tokens": 720597507.0, + "step": 27638 + }, + { + "epoch": 2.279129215799456, + "grad_norm": 1.8309481143951416, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.787615180015564, + "num_tokens": 720633513.0, + "step": 27639 + }, + { + "epoch": 2.279211676424507, + "grad_norm": 2.079310894012451, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7577023506164551, + "num_tokens": 720663962.0, + "step": 27640 + }, + { + "epoch": 2.279294137049559, + "grad_norm": 2.3916471004486084, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7733789682388306, + "num_tokens": 720687471.0, + "step": 27641 + }, + { + "epoch": 2.2793765976746103, + "grad_norm": 2.103823184967041, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7771303653717041, + "num_tokens": 720716816.0, + "step": 27642 + }, + { + "epoch": 2.279459058299662, + "grad_norm": 2.2476115226745605, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7440782785415649, + "num_tokens": 720744797.0, + "step": 27643 + }, + { + "epoch": 2.2795415189247135, + "grad_norm": 2.0326852798461914, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7729697227478027, + "num_tokens": 720777805.0, + "step": 27644 + }, + { + "epoch": 2.279623979549765, + "grad_norm": 2.4800751209259033, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7529552578926086, + "num_tokens": 720799818.0, + "step": 27645 + }, + { + "epoch": 2.2797064401748166, + "grad_norm": 2.084404706954956, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7578907012939453, + "num_tokens": 720831073.0, + "step": 27646 + }, + { + "epoch": 2.279788900799868, + "grad_norm": 1.8956894874572754, + "learning_rate": 1e-06, + "loss": 0.5048, + "mean_token_accuracy": 0.8298390507698059, + "num_tokens": 720860783.0, + "step": 27647 + }, + { + "epoch": 2.2798713614249198, + "grad_norm": 2.3456015586853027, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7720950841903687, + "num_tokens": 720887712.0, + "step": 27648 + }, + { + "epoch": 2.279953822049971, + "grad_norm": 2.069195508956909, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.792898952960968, + "num_tokens": 720916687.0, + "step": 27649 + }, + { + "epoch": 2.280036282675023, + "grad_norm": 2.3657424449920654, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.771201491355896, + "num_tokens": 720942773.0, + "step": 27650 + }, + { + "epoch": 2.2801187433000742, + "grad_norm": 2.3910562992095947, + "learning_rate": 1e-06, + "loss": 0.6079, + "mean_token_accuracy": 0.7883583307266235, + "num_tokens": 720968219.0, + "step": 27651 + }, + { + "epoch": 2.2802012039251256, + "grad_norm": 1.7778913974761963, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.7929346561431885, + "num_tokens": 721008323.0, + "step": 27652 + }, + { + "epoch": 2.2802836645501774, + "grad_norm": 2.1994681358337402, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7685550451278687, + "num_tokens": 721034259.0, + "step": 27653 + }, + { + "epoch": 2.2803661251752287, + "grad_norm": 2.2270989418029785, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.728097677230835, + "num_tokens": 721066029.0, + "step": 27654 + }, + { + "epoch": 2.2804485858002805, + "grad_norm": 2.1473965644836426, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7639435529708862, + "num_tokens": 721094708.0, + "step": 27655 + }, + { + "epoch": 2.280531046425332, + "grad_norm": 1.9757411479949951, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7845834493637085, + "num_tokens": 721123959.0, + "step": 27656 + }, + { + "epoch": 2.2806135070503837, + "grad_norm": 2.2483878135681152, + "learning_rate": 1e-06, + "loss": 0.6126, + "mean_token_accuracy": 0.7932358980178833, + "num_tokens": 721150282.0, + "step": 27657 + }, + { + "epoch": 2.280695967675435, + "grad_norm": 2.3109679222106934, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7723131775856018, + "num_tokens": 721174855.0, + "step": 27658 + }, + { + "epoch": 2.2807784283004864, + "grad_norm": 2.209416389465332, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7710882425308228, + "num_tokens": 721204810.0, + "step": 27659 + }, + { + "epoch": 2.280860888925538, + "grad_norm": 2.115400791168213, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7577722072601318, + "num_tokens": 721237336.0, + "step": 27660 + }, + { + "epoch": 2.2809433495505895, + "grad_norm": 1.9443304538726807, + "learning_rate": 1e-06, + "loss": 0.6307, + "mean_token_accuracy": 0.7981839179992676, + "num_tokens": 721273858.0, + "step": 27661 + }, + { + "epoch": 2.2810258101756413, + "grad_norm": 2.009335517883301, + "learning_rate": 1e-06, + "loss": 0.5648, + "mean_token_accuracy": 0.8017613291740417, + "num_tokens": 721305732.0, + "step": 27662 + }, + { + "epoch": 2.2811082708006927, + "grad_norm": 2.428140640258789, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7468633651733398, + "num_tokens": 721329206.0, + "step": 27663 + }, + { + "epoch": 2.281190731425744, + "grad_norm": 2.3291242122650146, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7423707842826843, + "num_tokens": 721358971.0, + "step": 27664 + }, + { + "epoch": 2.281273192050796, + "grad_norm": 2.153712749481201, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7736226320266724, + "num_tokens": 721386889.0, + "step": 27665 + }, + { + "epoch": 2.281355652675847, + "grad_norm": 1.9184032678604126, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7724246978759766, + "num_tokens": 721421079.0, + "step": 27666 + }, + { + "epoch": 2.281438113300899, + "grad_norm": 2.088102102279663, + "learning_rate": 1e-06, + "loss": 0.5562, + "mean_token_accuracy": 0.7958023548126221, + "num_tokens": 721449242.0, + "step": 27667 + }, + { + "epoch": 2.2815205739259503, + "grad_norm": 1.886450171470642, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7580341100692749, + "num_tokens": 721485246.0, + "step": 27668 + }, + { + "epoch": 2.2816030345510017, + "grad_norm": 2.435940742492676, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7846020460128784, + "num_tokens": 721508243.0, + "step": 27669 + }, + { + "epoch": 2.2816854951760535, + "grad_norm": 2.204246759414673, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7603564262390137, + "num_tokens": 721535781.0, + "step": 27670 + }, + { + "epoch": 2.281767955801105, + "grad_norm": 1.9975197315216064, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.757160484790802, + "num_tokens": 721566584.0, + "step": 27671 + }, + { + "epoch": 2.2818504164261566, + "grad_norm": 2.1351146697998047, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7672165036201477, + "num_tokens": 721592886.0, + "step": 27672 + }, + { + "epoch": 2.281932877051208, + "grad_norm": 2.0829381942749023, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.7514303922653198, + "num_tokens": 721627104.0, + "step": 27673 + }, + { + "epoch": 2.2820153376762597, + "grad_norm": 1.945984959602356, + "learning_rate": 1e-06, + "loss": 0.5794, + "mean_token_accuracy": 0.7982293367385864, + "num_tokens": 721657132.0, + "step": 27674 + }, + { + "epoch": 2.282097798301311, + "grad_norm": 2.0027332305908203, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7717069387435913, + "num_tokens": 721692336.0, + "step": 27675 + }, + { + "epoch": 2.2821802589263624, + "grad_norm": 2.1735622882843018, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7553104162216187, + "num_tokens": 721721627.0, + "step": 27676 + }, + { + "epoch": 2.2822627195514142, + "grad_norm": 1.9657658338546753, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7824456095695496, + "num_tokens": 721751917.0, + "step": 27677 + }, + { + "epoch": 2.2823451801764656, + "grad_norm": 2.314779758453369, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7465369701385498, + "num_tokens": 721776115.0, + "step": 27678 + }, + { + "epoch": 2.2824276408015174, + "grad_norm": 2.0825119018554688, + "learning_rate": 1e-06, + "loss": 0.5628, + "mean_token_accuracy": 0.8070017099380493, + "num_tokens": 721806518.0, + "step": 27679 + }, + { + "epoch": 2.2825101014265687, + "grad_norm": 1.9243665933609009, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.786873996257782, + "num_tokens": 721837841.0, + "step": 27680 + }, + { + "epoch": 2.2825925620516205, + "grad_norm": 1.9455225467681885, + "learning_rate": 1e-06, + "loss": 0.5726, + "mean_token_accuracy": 0.8063418865203857, + "num_tokens": 721869446.0, + "step": 27681 + }, + { + "epoch": 2.282675022676672, + "grad_norm": 1.9295240640640259, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7741485238075256, + "num_tokens": 721902726.0, + "step": 27682 + }, + { + "epoch": 2.2827574833017232, + "grad_norm": 2.2114017009735107, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7675601243972778, + "num_tokens": 721932289.0, + "step": 27683 + }, + { + "epoch": 2.282839943926775, + "grad_norm": 2.1649932861328125, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7575526237487793, + "num_tokens": 721960687.0, + "step": 27684 + }, + { + "epoch": 2.2829224045518264, + "grad_norm": 2.2139930725097656, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7790155410766602, + "num_tokens": 721988071.0, + "step": 27685 + }, + { + "epoch": 2.283004865176878, + "grad_norm": 1.9210050106048584, + "learning_rate": 1e-06, + "loss": 0.5715, + "mean_token_accuracy": 0.8076202869415283, + "num_tokens": 722020649.0, + "step": 27686 + }, + { + "epoch": 2.2830873258019295, + "grad_norm": 2.0349202156066895, + "learning_rate": 1e-06, + "loss": 0.527, + "mean_token_accuracy": 0.8213332295417786, + "num_tokens": 722050474.0, + "step": 27687 + }, + { + "epoch": 2.2831697864269813, + "grad_norm": 1.948448896408081, + "learning_rate": 1e-06, + "loss": 0.5536, + "mean_token_accuracy": 0.8102362155914307, + "num_tokens": 722081322.0, + "step": 27688 + }, + { + "epoch": 2.2832522470520327, + "grad_norm": 1.8650659322738647, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7597710490226746, + "num_tokens": 722118670.0, + "step": 27689 + }, + { + "epoch": 2.283334707677084, + "grad_norm": 2.209667205810547, + "learning_rate": 1e-06, + "loss": 0.7442, + "mean_token_accuracy": 0.7559338212013245, + "num_tokens": 722146572.0, + "step": 27690 + }, + { + "epoch": 2.283417168302136, + "grad_norm": 2.095367431640625, + "learning_rate": 1e-06, + "loss": 0.7175, + "mean_token_accuracy": 0.7715465426445007, + "num_tokens": 722175748.0, + "step": 27691 + }, + { + "epoch": 2.283499628927187, + "grad_norm": 1.978073000907898, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7823165655136108, + "num_tokens": 722207025.0, + "step": 27692 + }, + { + "epoch": 2.283582089552239, + "grad_norm": 2.2082762718200684, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7545432448387146, + "num_tokens": 722233066.0, + "step": 27693 + }, + { + "epoch": 2.2836645501772903, + "grad_norm": 2.0343663692474365, + "learning_rate": 1e-06, + "loss": 0.7656, + "mean_token_accuracy": 0.7512052059173584, + "num_tokens": 722263785.0, + "step": 27694 + }, + { + "epoch": 2.283747010802342, + "grad_norm": 1.9363764524459839, + "learning_rate": 1e-06, + "loss": 0.6252, + "mean_token_accuracy": 0.7945426106452942, + "num_tokens": 722298024.0, + "step": 27695 + }, + { + "epoch": 2.2838294714273935, + "grad_norm": 2.11246919631958, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7732260227203369, + "num_tokens": 722330511.0, + "step": 27696 + }, + { + "epoch": 2.283911932052445, + "grad_norm": 2.3493285179138184, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7419216632843018, + "num_tokens": 722356531.0, + "step": 27697 + }, + { + "epoch": 2.2839943926774966, + "grad_norm": 2.0298826694488525, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7672234773635864, + "num_tokens": 722388097.0, + "step": 27698 + }, + { + "epoch": 2.284076853302548, + "grad_norm": 2.339625120162964, + "learning_rate": 1e-06, + "loss": 0.6014, + "mean_token_accuracy": 0.7885314226150513, + "num_tokens": 722412658.0, + "step": 27699 + }, + { + "epoch": 2.2841593139275997, + "grad_norm": 1.9090063571929932, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7839047908782959, + "num_tokens": 722445487.0, + "step": 27700 + }, + { + "epoch": 2.284241774552651, + "grad_norm": 2.0168724060058594, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7760982513427734, + "num_tokens": 722477793.0, + "step": 27701 + }, + { + "epoch": 2.284324235177703, + "grad_norm": 1.8753373622894287, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7675184011459351, + "num_tokens": 722513269.0, + "step": 27702 + }, + { + "epoch": 2.2844066958027542, + "grad_norm": 2.0936594009399414, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.769442617893219, + "num_tokens": 722544516.0, + "step": 27703 + }, + { + "epoch": 2.2844891564278056, + "grad_norm": 2.519867420196533, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7686060667037964, + "num_tokens": 722569307.0, + "step": 27704 + }, + { + "epoch": 2.2845716170528574, + "grad_norm": 2.3059463500976562, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7678801417350769, + "num_tokens": 722598804.0, + "step": 27705 + }, + { + "epoch": 2.2846540776779087, + "grad_norm": 2.3569252490997314, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7433038949966431, + "num_tokens": 722626463.0, + "step": 27706 + }, + { + "epoch": 2.2847365383029605, + "grad_norm": 1.869448184967041, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.7863451838493347, + "num_tokens": 722662108.0, + "step": 27707 + }, + { + "epoch": 2.284818998928012, + "grad_norm": 2.424121379852295, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7738717198371887, + "num_tokens": 722686962.0, + "step": 27708 + }, + { + "epoch": 2.2849014595530632, + "grad_norm": 2.349768877029419, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.763168454170227, + "num_tokens": 722712539.0, + "step": 27709 + }, + { + "epoch": 2.284983920178115, + "grad_norm": 2.0217339992523193, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.794630229473114, + "num_tokens": 722742487.0, + "step": 27710 + }, + { + "epoch": 2.2850663808031664, + "grad_norm": 2.0866546630859375, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7360667586326599, + "num_tokens": 722772878.0, + "step": 27711 + }, + { + "epoch": 2.285148841428218, + "grad_norm": 1.987170696258545, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.750327467918396, + "num_tokens": 722807540.0, + "step": 27712 + }, + { + "epoch": 2.2852313020532695, + "grad_norm": 2.017979383468628, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7560312747955322, + "num_tokens": 722838877.0, + "step": 27713 + }, + { + "epoch": 2.285313762678321, + "grad_norm": 2.2853810787200928, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7524492144584656, + "num_tokens": 722863396.0, + "step": 27714 + }, + { + "epoch": 2.2853962233033727, + "grad_norm": 2.183842420578003, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7525074481964111, + "num_tokens": 722893242.0, + "step": 27715 + }, + { + "epoch": 2.285478683928424, + "grad_norm": 2.250270128250122, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7660446166992188, + "num_tokens": 722919968.0, + "step": 27716 + }, + { + "epoch": 2.285561144553476, + "grad_norm": 2.1575703620910645, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7362146973609924, + "num_tokens": 722948846.0, + "step": 27717 + }, + { + "epoch": 2.285643605178527, + "grad_norm": 2.0382063388824463, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7486859560012817, + "num_tokens": 722983987.0, + "step": 27718 + }, + { + "epoch": 2.285726065803579, + "grad_norm": 2.157360315322876, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7708774209022522, + "num_tokens": 723014972.0, + "step": 27719 + }, + { + "epoch": 2.2858085264286303, + "grad_norm": 1.92146635055542, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7855449914932251, + "num_tokens": 723047801.0, + "step": 27720 + }, + { + "epoch": 2.2858909870536817, + "grad_norm": 2.092651844024658, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.7838271856307983, + "num_tokens": 723075638.0, + "step": 27721 + }, + { + "epoch": 2.2859734476787335, + "grad_norm": 2.1699228286743164, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7682138085365295, + "num_tokens": 723106212.0, + "step": 27722 + }, + { + "epoch": 2.286055908303785, + "grad_norm": 1.9726568460464478, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.7948823571205139, + "num_tokens": 723137211.0, + "step": 27723 + }, + { + "epoch": 2.2861383689288366, + "grad_norm": 2.341000556945801, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.7821308970451355, + "num_tokens": 723163336.0, + "step": 27724 + }, + { + "epoch": 2.286220829553888, + "grad_norm": 2.092167615890503, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.7817543148994446, + "num_tokens": 723192423.0, + "step": 27725 + }, + { + "epoch": 2.2863032901789397, + "grad_norm": 2.0598409175872803, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7838674783706665, + "num_tokens": 723222747.0, + "step": 27726 + }, + { + "epoch": 2.286385750803991, + "grad_norm": 1.870567798614502, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7803781628608704, + "num_tokens": 723253685.0, + "step": 27727 + }, + { + "epoch": 2.2864682114290424, + "grad_norm": 2.4362692832946777, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7652724385261536, + "num_tokens": 723278321.0, + "step": 27728 + }, + { + "epoch": 2.2865506720540942, + "grad_norm": 2.1149234771728516, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7603334188461304, + "num_tokens": 723309007.0, + "step": 27729 + }, + { + "epoch": 2.2866331326791456, + "grad_norm": 2.395263910293579, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7556637525558472, + "num_tokens": 723334513.0, + "step": 27730 + }, + { + "epoch": 2.2867155933041974, + "grad_norm": 1.8329792022705078, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.7970547676086426, + "num_tokens": 723370588.0, + "step": 27731 + }, + { + "epoch": 2.2867980539292487, + "grad_norm": 2.3250503540039062, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7418997287750244, + "num_tokens": 723398311.0, + "step": 27732 + }, + { + "epoch": 2.2868805145543005, + "grad_norm": 2.3183069229125977, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7616602182388306, + "num_tokens": 723422073.0, + "step": 27733 + }, + { + "epoch": 2.286962975179352, + "grad_norm": 1.9469470977783203, + "learning_rate": 1e-06, + "loss": 0.5871, + "mean_token_accuracy": 0.7965137362480164, + "num_tokens": 723453993.0, + "step": 27734 + }, + { + "epoch": 2.2870454358044032, + "grad_norm": 2.074840784072876, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7526732683181763, + "num_tokens": 723486062.0, + "step": 27735 + }, + { + "epoch": 2.287127896429455, + "grad_norm": 2.2468101978302, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.7920600175857544, + "num_tokens": 723510482.0, + "step": 27736 + }, + { + "epoch": 2.2872103570545064, + "grad_norm": 2.0377066135406494, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7466155290603638, + "num_tokens": 723545004.0, + "step": 27737 + }, + { + "epoch": 2.287292817679558, + "grad_norm": 1.885088324546814, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.7836371660232544, + "num_tokens": 723578289.0, + "step": 27738 + }, + { + "epoch": 2.2873752783046095, + "grad_norm": 2.3118896484375, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.7824933528900146, + "num_tokens": 723603517.0, + "step": 27739 + }, + { + "epoch": 2.2874577389296613, + "grad_norm": 2.2650887966156006, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.788782000541687, + "num_tokens": 723631818.0, + "step": 27740 + }, + { + "epoch": 2.2875401995547127, + "grad_norm": 2.4757416248321533, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7724336385726929, + "num_tokens": 723657274.0, + "step": 27741 + }, + { + "epoch": 2.287622660179764, + "grad_norm": 2.253563642501831, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7715287804603577, + "num_tokens": 723685011.0, + "step": 27742 + }, + { + "epoch": 2.287705120804816, + "grad_norm": 2.0668177604675293, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.7869094610214233, + "num_tokens": 723715543.0, + "step": 27743 + }, + { + "epoch": 2.287787581429867, + "grad_norm": 1.9364265203475952, + "learning_rate": 1e-06, + "loss": 0.6083, + "mean_token_accuracy": 0.8008789420127869, + "num_tokens": 723748981.0, + "step": 27744 + }, + { + "epoch": 2.287870042054919, + "grad_norm": 2.0233263969421387, + "learning_rate": 1e-06, + "loss": 0.4886, + "mean_token_accuracy": 0.8233809471130371, + "num_tokens": 723778343.0, + "step": 27745 + }, + { + "epoch": 2.2879525026799703, + "grad_norm": 1.871646523475647, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7687790393829346, + "num_tokens": 723816511.0, + "step": 27746 + }, + { + "epoch": 2.288034963305022, + "grad_norm": 1.7392274141311646, + "learning_rate": 1e-06, + "loss": 0.6095, + "mean_token_accuracy": 0.7988266944885254, + "num_tokens": 723854162.0, + "step": 27747 + }, + { + "epoch": 2.2881174239300734, + "grad_norm": 2.373756170272827, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7691015601158142, + "num_tokens": 723877895.0, + "step": 27748 + }, + { + "epoch": 2.288199884555125, + "grad_norm": 2.379986524581909, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7567568421363831, + "num_tokens": 723902535.0, + "step": 27749 + }, + { + "epoch": 2.2882823451801766, + "grad_norm": 2.4511406421661377, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7537481784820557, + "num_tokens": 723925130.0, + "step": 27750 + }, + { + "epoch": 2.288364805805228, + "grad_norm": 2.0312583446502686, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7697255611419678, + "num_tokens": 723956824.0, + "step": 27751 + }, + { + "epoch": 2.2884472664302793, + "grad_norm": 2.0866177082061768, + "learning_rate": 1e-06, + "loss": 0.7043, + "mean_token_accuracy": 0.7756069898605347, + "num_tokens": 723989864.0, + "step": 27752 + }, + { + "epoch": 2.288529727055331, + "grad_norm": 1.9505308866500854, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.7897560596466064, + "num_tokens": 724025004.0, + "step": 27753 + }, + { + "epoch": 2.2886121876803824, + "grad_norm": 2.0754597187042236, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.7946686744689941, + "num_tokens": 724054925.0, + "step": 27754 + }, + { + "epoch": 2.2886946483054342, + "grad_norm": 2.191420555114746, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.799090564250946, + "num_tokens": 724083325.0, + "step": 27755 + }, + { + "epoch": 2.2887771089304856, + "grad_norm": 2.1747524738311768, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.778167188167572, + "num_tokens": 724110702.0, + "step": 27756 + }, + { + "epoch": 2.2888595695555374, + "grad_norm": 1.98660147190094, + "learning_rate": 1e-06, + "loss": 0.5636, + "mean_token_accuracy": 0.8060126304626465, + "num_tokens": 724142832.0, + "step": 27757 + }, + { + "epoch": 2.2889420301805887, + "grad_norm": 2.295703649520874, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.7941293716430664, + "num_tokens": 724169437.0, + "step": 27758 + }, + { + "epoch": 2.28902449080564, + "grad_norm": 1.9503841400146484, + "learning_rate": 1e-06, + "loss": 0.5897, + "mean_token_accuracy": 0.8029664754867554, + "num_tokens": 724201716.0, + "step": 27759 + }, + { + "epoch": 2.289106951430692, + "grad_norm": 1.9033355712890625, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.763360857963562, + "num_tokens": 724235635.0, + "step": 27760 + }, + { + "epoch": 2.2891894120557432, + "grad_norm": 2.2928872108459473, + "learning_rate": 1e-06, + "loss": 0.78, + "mean_token_accuracy": 0.7453179955482483, + "num_tokens": 724263958.0, + "step": 27761 + }, + { + "epoch": 2.289271872680795, + "grad_norm": 2.31768536567688, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7538964748382568, + "num_tokens": 724290666.0, + "step": 27762 + }, + { + "epoch": 2.2893543333058464, + "grad_norm": 2.3370041847229004, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.7868826389312744, + "num_tokens": 724317324.0, + "step": 27763 + }, + { + "epoch": 2.289436793930898, + "grad_norm": 1.9752261638641357, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.7836172580718994, + "num_tokens": 724347831.0, + "step": 27764 + }, + { + "epoch": 2.2895192545559495, + "grad_norm": 2.083033561706543, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7637758255004883, + "num_tokens": 724377010.0, + "step": 27765 + }, + { + "epoch": 2.289601715181001, + "grad_norm": 2.2369565963745117, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7532211542129517, + "num_tokens": 724406638.0, + "step": 27766 + }, + { + "epoch": 2.2896841758060527, + "grad_norm": 2.481189727783203, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7643202543258667, + "num_tokens": 724428596.0, + "step": 27767 + }, + { + "epoch": 2.289766636431104, + "grad_norm": 2.3282675743103027, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7759262323379517, + "num_tokens": 724453554.0, + "step": 27768 + }, + { + "epoch": 2.289849097056156, + "grad_norm": 2.0059151649475098, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7652552127838135, + "num_tokens": 724488439.0, + "step": 27769 + }, + { + "epoch": 2.289931557681207, + "grad_norm": 1.8885945081710815, + "learning_rate": 1e-06, + "loss": 0.6137, + "mean_token_accuracy": 0.7949389815330505, + "num_tokens": 724521447.0, + "step": 27770 + }, + { + "epoch": 2.290014018306259, + "grad_norm": 2.209465503692627, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.7859306335449219, + "num_tokens": 724549543.0, + "step": 27771 + }, + { + "epoch": 2.2900964789313103, + "grad_norm": 2.305223226547241, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7718217372894287, + "num_tokens": 724573842.0, + "step": 27772 + }, + { + "epoch": 2.2901789395563616, + "grad_norm": 1.9341381788253784, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7682300806045532, + "num_tokens": 724605079.0, + "step": 27773 + }, + { + "epoch": 2.2902614001814134, + "grad_norm": 1.9040675163269043, + "learning_rate": 1e-06, + "loss": 0.615, + "mean_token_accuracy": 0.795344352722168, + "num_tokens": 724638133.0, + "step": 27774 + }, + { + "epoch": 2.290343860806465, + "grad_norm": 1.9866864681243896, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7814123630523682, + "num_tokens": 724671297.0, + "step": 27775 + }, + { + "epoch": 2.2904263214315166, + "grad_norm": 2.1194565296173096, + "learning_rate": 1e-06, + "loss": 0.5374, + "mean_token_accuracy": 0.8204598426818848, + "num_tokens": 724699577.0, + "step": 27776 + }, + { + "epoch": 2.290508782056568, + "grad_norm": 2.149834394454956, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.7833248376846313, + "num_tokens": 724726024.0, + "step": 27777 + }, + { + "epoch": 2.2905912426816197, + "grad_norm": 1.9074292182922363, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7779738903045654, + "num_tokens": 724761372.0, + "step": 27778 + }, + { + "epoch": 2.290673703306671, + "grad_norm": 2.1888427734375, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.78192138671875, + "num_tokens": 724789272.0, + "step": 27779 + }, + { + "epoch": 2.2907561639317224, + "grad_norm": 2.208401679992676, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7645758390426636, + "num_tokens": 724818495.0, + "step": 27780 + }, + { + "epoch": 2.2908386245567742, + "grad_norm": 2.220965623855591, + "learning_rate": 1e-06, + "loss": 0.5854, + "mean_token_accuracy": 0.7940689325332642, + "num_tokens": 724844818.0, + "step": 27781 + }, + { + "epoch": 2.2909210851818256, + "grad_norm": 2.2891030311584473, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7904502153396606, + "num_tokens": 724870308.0, + "step": 27782 + }, + { + "epoch": 2.2910035458068774, + "grad_norm": 1.8979432582855225, + "learning_rate": 1e-06, + "loss": 0.7506, + "mean_token_accuracy": 0.7605565190315247, + "num_tokens": 724907855.0, + "step": 27783 + }, + { + "epoch": 2.2910860064319287, + "grad_norm": 2.146561622619629, + "learning_rate": 1e-06, + "loss": 0.6013, + "mean_token_accuracy": 0.8053830862045288, + "num_tokens": 724935472.0, + "step": 27784 + }, + { + "epoch": 2.2911684670569805, + "grad_norm": 1.9511244297027588, + "learning_rate": 1e-06, + "loss": 0.5545, + "mean_token_accuracy": 0.8186897039413452, + "num_tokens": 724966986.0, + "step": 27785 + }, + { + "epoch": 2.291250927682032, + "grad_norm": 1.889344334602356, + "learning_rate": 1e-06, + "loss": 0.6172, + "mean_token_accuracy": 0.7951964139938354, + "num_tokens": 725002394.0, + "step": 27786 + }, + { + "epoch": 2.291333388307083, + "grad_norm": 2.1718716621398926, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7610543966293335, + "num_tokens": 725029622.0, + "step": 27787 + }, + { + "epoch": 2.291415848932135, + "grad_norm": 2.3024938106536865, + "learning_rate": 1e-06, + "loss": 0.6033, + "mean_token_accuracy": 0.7918352484703064, + "num_tokens": 725056555.0, + "step": 27788 + }, + { + "epoch": 2.2914983095571864, + "grad_norm": 2.239346504211426, + "learning_rate": 1e-06, + "loss": 0.5375, + "mean_token_accuracy": 0.8138589859008789, + "num_tokens": 725081699.0, + "step": 27789 + }, + { + "epoch": 2.291580770182238, + "grad_norm": 2.296556234359741, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7749801874160767, + "num_tokens": 725105847.0, + "step": 27790 + }, + { + "epoch": 2.2916632308072895, + "grad_norm": 1.9483871459960938, + "learning_rate": 1e-06, + "loss": 0.4904, + "mean_token_accuracy": 0.8283495306968689, + "num_tokens": 725135636.0, + "step": 27791 + }, + { + "epoch": 2.2917456914323413, + "grad_norm": 2.106557846069336, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7565533518791199, + "num_tokens": 725166660.0, + "step": 27792 + }, + { + "epoch": 2.2918281520573927, + "grad_norm": 2.3016319274902344, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7673176527023315, + "num_tokens": 725194048.0, + "step": 27793 + }, + { + "epoch": 2.291910612682444, + "grad_norm": 2.0392470359802246, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7809591293334961, + "num_tokens": 725225252.0, + "step": 27794 + }, + { + "epoch": 2.291993073307496, + "grad_norm": 2.0051050186157227, + "learning_rate": 1e-06, + "loss": 0.567, + "mean_token_accuracy": 0.8080267906188965, + "num_tokens": 725255488.0, + "step": 27795 + }, + { + "epoch": 2.292075533932547, + "grad_norm": 1.9905269145965576, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7796743512153625, + "num_tokens": 725288157.0, + "step": 27796 + }, + { + "epoch": 2.2921579945575985, + "grad_norm": 2.0388002395629883, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.717118501663208, + "num_tokens": 725320960.0, + "step": 27797 + }, + { + "epoch": 2.2922404551826503, + "grad_norm": 1.9674527645111084, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7652204036712646, + "num_tokens": 725349243.0, + "step": 27798 + }, + { + "epoch": 2.2923229158077016, + "grad_norm": 2.381361484527588, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7359093427658081, + "num_tokens": 725376453.0, + "step": 27799 + }, + { + "epoch": 2.2924053764327534, + "grad_norm": 2.034188985824585, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7738308906555176, + "num_tokens": 725407977.0, + "step": 27800 + }, + { + "epoch": 2.292487837057805, + "grad_norm": 2.22471022605896, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7314698696136475, + "num_tokens": 725436807.0, + "step": 27801 + }, + { + "epoch": 2.2925702976828566, + "grad_norm": 2.0435850620269775, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7618312835693359, + "num_tokens": 725466954.0, + "step": 27802 + }, + { + "epoch": 2.292652758307908, + "grad_norm": 2.1395177841186523, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.7708956003189087, + "num_tokens": 725495455.0, + "step": 27803 + }, + { + "epoch": 2.2927352189329593, + "grad_norm": 2.084909677505493, + "learning_rate": 1e-06, + "loss": 0.4938, + "mean_token_accuracy": 0.829874575138092, + "num_tokens": 725521953.0, + "step": 27804 + }, + { + "epoch": 2.292817679558011, + "grad_norm": 2.097048282623291, + "learning_rate": 1e-06, + "loss": 0.6219, + "mean_token_accuracy": 0.7936409711837769, + "num_tokens": 725552611.0, + "step": 27805 + }, + { + "epoch": 2.2929001401830624, + "grad_norm": 2.0932137966156006, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7579116821289062, + "num_tokens": 725581886.0, + "step": 27806 + }, + { + "epoch": 2.2929826008081142, + "grad_norm": 2.1278605461120605, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7751798033714294, + "num_tokens": 725609620.0, + "step": 27807 + }, + { + "epoch": 2.2930650614331656, + "grad_norm": 2.172999143600464, + "learning_rate": 1e-06, + "loss": 0.5916, + "mean_token_accuracy": 0.8029476404190063, + "num_tokens": 725633019.0, + "step": 27808 + }, + { + "epoch": 2.2931475220582174, + "grad_norm": 2.7806875705718994, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7489370703697205, + "num_tokens": 725652819.0, + "step": 27809 + }, + { + "epoch": 2.2932299826832687, + "grad_norm": 2.240588426589966, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.7706334590911865, + "num_tokens": 725678458.0, + "step": 27810 + }, + { + "epoch": 2.29331244330832, + "grad_norm": 2.2092058658599854, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.7912864089012146, + "num_tokens": 725705398.0, + "step": 27811 + }, + { + "epoch": 2.293394903933372, + "grad_norm": 2.059948205947876, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7706604599952698, + "num_tokens": 725737214.0, + "step": 27812 + }, + { + "epoch": 2.293477364558423, + "grad_norm": 2.167107105255127, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.7455736398696899, + "num_tokens": 725766320.0, + "step": 27813 + }, + { + "epoch": 2.293559825183475, + "grad_norm": 2.1555750370025635, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7442061901092529, + "num_tokens": 725801205.0, + "step": 27814 + }, + { + "epoch": 2.2936422858085264, + "grad_norm": 2.2179346084594727, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7624138593673706, + "num_tokens": 725829848.0, + "step": 27815 + }, + { + "epoch": 2.293724746433578, + "grad_norm": 2.2914865016937256, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7757402658462524, + "num_tokens": 725861636.0, + "step": 27816 + }, + { + "epoch": 2.2938072070586295, + "grad_norm": 2.078882932662964, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7827398777008057, + "num_tokens": 725890088.0, + "step": 27817 + }, + { + "epoch": 2.293889667683681, + "grad_norm": 2.205462694168091, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7658743858337402, + "num_tokens": 725918954.0, + "step": 27818 + }, + { + "epoch": 2.2939721283087326, + "grad_norm": 2.0818450450897217, + "learning_rate": 1e-06, + "loss": 0.594, + "mean_token_accuracy": 0.7903778553009033, + "num_tokens": 725947798.0, + "step": 27819 + }, + { + "epoch": 2.294054588933784, + "grad_norm": 2.09138822555542, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7558347582817078, + "num_tokens": 725979454.0, + "step": 27820 + }, + { + "epoch": 2.294137049558836, + "grad_norm": 2.1366355419158936, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7489201426506042, + "num_tokens": 726010579.0, + "step": 27821 + }, + { + "epoch": 2.294219510183887, + "grad_norm": 1.9356560707092285, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7787092924118042, + "num_tokens": 726042720.0, + "step": 27822 + }, + { + "epoch": 2.294301970808939, + "grad_norm": 1.8633060455322266, + "learning_rate": 1e-06, + "loss": 0.7563, + "mean_token_accuracy": 0.7568174004554749, + "num_tokens": 726081954.0, + "step": 27823 + }, + { + "epoch": 2.2943844314339903, + "grad_norm": 2.5442209243774414, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.726067066192627, + "num_tokens": 726104848.0, + "step": 27824 + }, + { + "epoch": 2.2944668920590416, + "grad_norm": 2.191581964492798, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7359783053398132, + "num_tokens": 726134398.0, + "step": 27825 + }, + { + "epoch": 2.2945493526840934, + "grad_norm": 1.9866653680801392, + "learning_rate": 1e-06, + "loss": 0.5575, + "mean_token_accuracy": 0.8150312900543213, + "num_tokens": 726166323.0, + "step": 27826 + }, + { + "epoch": 2.294631813309145, + "grad_norm": 1.8229225873947144, + "learning_rate": 1e-06, + "loss": 0.5774, + "mean_token_accuracy": 0.8041801452636719, + "num_tokens": 726202188.0, + "step": 27827 + }, + { + "epoch": 2.2947142739341966, + "grad_norm": 2.102437734603882, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7762737274169922, + "num_tokens": 726232656.0, + "step": 27828 + }, + { + "epoch": 2.294796734559248, + "grad_norm": 1.7801554203033447, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7724097967147827, + "num_tokens": 726267580.0, + "step": 27829 + }, + { + "epoch": 2.2948791951842997, + "grad_norm": 2.06302809715271, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7659851312637329, + "num_tokens": 726298346.0, + "step": 27830 + }, + { + "epoch": 2.294961655809351, + "grad_norm": 2.245661973953247, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7763121128082275, + "num_tokens": 726327069.0, + "step": 27831 + }, + { + "epoch": 2.2950441164344024, + "grad_norm": 2.1872522830963135, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7839056849479675, + "num_tokens": 726355444.0, + "step": 27832 + }, + { + "epoch": 2.295126577059454, + "grad_norm": 1.9436205625534058, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8027215003967285, + "num_tokens": 726389133.0, + "step": 27833 + }, + { + "epoch": 2.2952090376845056, + "grad_norm": 2.1836905479431152, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7738974094390869, + "num_tokens": 726415377.0, + "step": 27834 + }, + { + "epoch": 2.2952914983095574, + "grad_norm": 1.978232741355896, + "learning_rate": 1e-06, + "loss": 0.5768, + "mean_token_accuracy": 0.8114964962005615, + "num_tokens": 726447953.0, + "step": 27835 + }, + { + "epoch": 2.2953739589346087, + "grad_norm": 1.8187724351882935, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.789876401424408, + "num_tokens": 726485328.0, + "step": 27836 + }, + { + "epoch": 2.2954564195596605, + "grad_norm": 2.0842347145080566, + "learning_rate": 1e-06, + "loss": 0.6278, + "mean_token_accuracy": 0.793096661567688, + "num_tokens": 726513297.0, + "step": 27837 + }, + { + "epoch": 2.295538880184712, + "grad_norm": 2.219489097595215, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7701859474182129, + "num_tokens": 726542377.0, + "step": 27838 + }, + { + "epoch": 2.295621340809763, + "grad_norm": 2.066664695739746, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7798789739608765, + "num_tokens": 726572919.0, + "step": 27839 + }, + { + "epoch": 2.295703801434815, + "grad_norm": 2.1818130016326904, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7342016696929932, + "num_tokens": 726603264.0, + "step": 27840 + }, + { + "epoch": 2.2957862620598664, + "grad_norm": 2.036876678466797, + "learning_rate": 1e-06, + "loss": 0.5563, + "mean_token_accuracy": 0.8120632767677307, + "num_tokens": 726635134.0, + "step": 27841 + }, + { + "epoch": 2.2958687226849177, + "grad_norm": 2.0761542320251465, + "learning_rate": 1e-06, + "loss": 0.6177, + "mean_token_accuracy": 0.7992232441902161, + "num_tokens": 726662042.0, + "step": 27842 + }, + { + "epoch": 2.2959511833099695, + "grad_norm": 2.595507860183716, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.779125452041626, + "num_tokens": 726681960.0, + "step": 27843 + }, + { + "epoch": 2.296033643935021, + "grad_norm": 2.200098752975464, + "learning_rate": 1e-06, + "loss": 0.5377, + "mean_token_accuracy": 0.8164910078048706, + "num_tokens": 726707861.0, + "step": 27844 + }, + { + "epoch": 2.2961161045600726, + "grad_norm": 2.0451252460479736, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7508212327957153, + "num_tokens": 726740008.0, + "step": 27845 + }, + { + "epoch": 2.296198565185124, + "grad_norm": 2.4206953048706055, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7403539419174194, + "num_tokens": 726764070.0, + "step": 27846 + }, + { + "epoch": 2.296281025810176, + "grad_norm": 2.1138789653778076, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7699341773986816, + "num_tokens": 726794341.0, + "step": 27847 + }, + { + "epoch": 2.296363486435227, + "grad_norm": 2.186630964279175, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7749599814414978, + "num_tokens": 726824705.0, + "step": 27848 + }, + { + "epoch": 2.2964459470602785, + "grad_norm": 1.992356300354004, + "learning_rate": 1e-06, + "loss": 0.6069, + "mean_token_accuracy": 0.8079790472984314, + "num_tokens": 726855767.0, + "step": 27849 + }, + { + "epoch": 2.2965284076853303, + "grad_norm": 2.428731918334961, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.7562307119369507, + "num_tokens": 726880224.0, + "step": 27850 + }, + { + "epoch": 2.2966108683103816, + "grad_norm": 2.2661759853363037, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7595114707946777, + "num_tokens": 726905877.0, + "step": 27851 + }, + { + "epoch": 2.2966933289354334, + "grad_norm": 2.1012842655181885, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7784233689308167, + "num_tokens": 726936237.0, + "step": 27852 + }, + { + "epoch": 2.296775789560485, + "grad_norm": 1.9911705255508423, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7729261517524719, + "num_tokens": 726967146.0, + "step": 27853 + }, + { + "epoch": 2.2968582501855366, + "grad_norm": 2.3441290855407715, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.7332793474197388, + "num_tokens": 726992610.0, + "step": 27854 + }, + { + "epoch": 2.296940710810588, + "grad_norm": 1.762919545173645, + "learning_rate": 1e-06, + "loss": 0.4004, + "mean_token_accuracy": 0.8649940490722656, + "num_tokens": 727025191.0, + "step": 27855 + }, + { + "epoch": 2.2970231714356393, + "grad_norm": 2.280590534210205, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7687532901763916, + "num_tokens": 727053929.0, + "step": 27856 + }, + { + "epoch": 2.297105632060691, + "grad_norm": 2.0135626792907715, + "learning_rate": 1e-06, + "loss": 0.6119, + "mean_token_accuracy": 0.7962301969528198, + "num_tokens": 727087047.0, + "step": 27857 + }, + { + "epoch": 2.2971880926857424, + "grad_norm": 2.1058149337768555, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7632589340209961, + "num_tokens": 727117755.0, + "step": 27858 + }, + { + "epoch": 2.297270553310794, + "grad_norm": 1.9442768096923828, + "learning_rate": 1e-06, + "loss": 0.5825, + "mean_token_accuracy": 0.8086751699447632, + "num_tokens": 727149278.0, + "step": 27859 + }, + { + "epoch": 2.2973530139358456, + "grad_norm": 2.225332736968994, + "learning_rate": 1e-06, + "loss": 0.471, + "mean_token_accuracy": 0.836824893951416, + "num_tokens": 727173421.0, + "step": 27860 + }, + { + "epoch": 2.2974354745608974, + "grad_norm": 1.88666832447052, + "learning_rate": 1e-06, + "loss": 0.5979, + "mean_token_accuracy": 0.805317759513855, + "num_tokens": 727206574.0, + "step": 27861 + }, + { + "epoch": 2.2975179351859487, + "grad_norm": 2.28658390045166, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7663435935974121, + "num_tokens": 727230661.0, + "step": 27862 + }, + { + "epoch": 2.297600395811, + "grad_norm": 1.8641674518585205, + "learning_rate": 1e-06, + "loss": 0.6178, + "mean_token_accuracy": 0.7916107773780823, + "num_tokens": 727266710.0, + "step": 27863 + }, + { + "epoch": 2.297682856436052, + "grad_norm": 2.15759015083313, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.7726951241493225, + "num_tokens": 727295478.0, + "step": 27864 + }, + { + "epoch": 2.297765317061103, + "grad_norm": 2.2674410343170166, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.7904673218727112, + "num_tokens": 727322314.0, + "step": 27865 + }, + { + "epoch": 2.297847777686155, + "grad_norm": 2.2464540004730225, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7631998062133789, + "num_tokens": 727349038.0, + "step": 27866 + }, + { + "epoch": 2.2979302383112064, + "grad_norm": 2.110382080078125, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7644591331481934, + "num_tokens": 727376404.0, + "step": 27867 + }, + { + "epoch": 2.298012698936258, + "grad_norm": 1.900290846824646, + "learning_rate": 1e-06, + "loss": 0.5694, + "mean_token_accuracy": 0.8153419494628906, + "num_tokens": 727407168.0, + "step": 27868 + }, + { + "epoch": 2.2980951595613095, + "grad_norm": 2.128039836883545, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.7822901010513306, + "num_tokens": 727435143.0, + "step": 27869 + }, + { + "epoch": 2.298177620186361, + "grad_norm": 2.1533045768737793, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7431390285491943, + "num_tokens": 727462500.0, + "step": 27870 + }, + { + "epoch": 2.2982600808114126, + "grad_norm": 1.9810584783554077, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.7882044315338135, + "num_tokens": 727492311.0, + "step": 27871 + }, + { + "epoch": 2.298342541436464, + "grad_norm": 1.9455844163894653, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7379989624023438, + "num_tokens": 727526596.0, + "step": 27872 + }, + { + "epoch": 2.298425002061516, + "grad_norm": 2.128068447113037, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7388769388198853, + "num_tokens": 727555917.0, + "step": 27873 + }, + { + "epoch": 2.298507462686567, + "grad_norm": 2.2236688137054443, + "learning_rate": 1e-06, + "loss": 0.7302, + "mean_token_accuracy": 0.7628583312034607, + "num_tokens": 727584778.0, + "step": 27874 + }, + { + "epoch": 2.298589923311619, + "grad_norm": 1.9191827774047852, + "learning_rate": 1e-06, + "loss": 0.5636, + "mean_token_accuracy": 0.8106734752655029, + "num_tokens": 727620001.0, + "step": 27875 + }, + { + "epoch": 2.2986723839366703, + "grad_norm": 2.291553020477295, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7514483332633972, + "num_tokens": 727651131.0, + "step": 27876 + }, + { + "epoch": 2.2987548445617216, + "grad_norm": 2.4545392990112305, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7684035301208496, + "num_tokens": 727675178.0, + "step": 27877 + }, + { + "epoch": 2.2988373051867734, + "grad_norm": 2.1233651638031006, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.7879989147186279, + "num_tokens": 727703409.0, + "step": 27878 + }, + { + "epoch": 2.298919765811825, + "grad_norm": 2.0993242263793945, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.74709552526474, + "num_tokens": 727738091.0, + "step": 27879 + }, + { + "epoch": 2.2990022264368766, + "grad_norm": 2.2614762783050537, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7725008130073547, + "num_tokens": 727764811.0, + "step": 27880 + }, + { + "epoch": 2.299084687061928, + "grad_norm": 2.137468099594116, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7465962767601013, + "num_tokens": 727793954.0, + "step": 27881 + }, + { + "epoch": 2.2991671476869793, + "grad_norm": 2.135986328125, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.762345552444458, + "num_tokens": 727824606.0, + "step": 27882 + }, + { + "epoch": 2.299249608312031, + "grad_norm": 2.242872476577759, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7794517874717712, + "num_tokens": 727851936.0, + "step": 27883 + }, + { + "epoch": 2.2993320689370824, + "grad_norm": 2.2809462547302246, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7776246070861816, + "num_tokens": 727878367.0, + "step": 27884 + }, + { + "epoch": 2.299414529562134, + "grad_norm": 2.194878578186035, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7416971325874329, + "num_tokens": 727907793.0, + "step": 27885 + }, + { + "epoch": 2.2994969901871856, + "grad_norm": 2.246826410293579, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.729925274848938, + "num_tokens": 727937698.0, + "step": 27886 + }, + { + "epoch": 2.299579450812237, + "grad_norm": 2.2396790981292725, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7689837217330933, + "num_tokens": 727965837.0, + "step": 27887 + }, + { + "epoch": 2.2996619114372887, + "grad_norm": 1.9936567544937134, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7601909041404724, + "num_tokens": 727998768.0, + "step": 27888 + }, + { + "epoch": 2.29974437206234, + "grad_norm": 2.2261667251586914, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7567307949066162, + "num_tokens": 728027797.0, + "step": 27889 + }, + { + "epoch": 2.299826832687392, + "grad_norm": 1.903356909751892, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.7965089082717896, + "num_tokens": 728064352.0, + "step": 27890 + }, + { + "epoch": 2.299909293312443, + "grad_norm": 1.9841455221176147, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.7787624001502991, + "num_tokens": 728097527.0, + "step": 27891 + }, + { + "epoch": 2.299991753937495, + "grad_norm": 2.1707100868225098, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7735675573348999, + "num_tokens": 728128243.0, + "step": 27892 + }, + { + "epoch": 2.3000742145625463, + "grad_norm": 2.282681941986084, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7624979019165039, + "num_tokens": 728153339.0, + "step": 27893 + }, + { + "epoch": 2.3001566751875977, + "grad_norm": 2.3155505657196045, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7604036331176758, + "num_tokens": 728180414.0, + "step": 27894 + }, + { + "epoch": 2.3002391358126495, + "grad_norm": 1.7951641082763672, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7715541124343872, + "num_tokens": 728218164.0, + "step": 27895 + }, + { + "epoch": 2.300321596437701, + "grad_norm": 2.2077012062072754, + "learning_rate": 1e-06, + "loss": 0.5864, + "mean_token_accuracy": 0.7972105741500854, + "num_tokens": 728244694.0, + "step": 27896 + }, + { + "epoch": 2.3004040570627526, + "grad_norm": 1.9747276306152344, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7909991145133972, + "num_tokens": 728278132.0, + "step": 27897 + }, + { + "epoch": 2.300486517687804, + "grad_norm": 2.0155344009399414, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7878642678260803, + "num_tokens": 728313861.0, + "step": 27898 + }, + { + "epoch": 2.300568978312856, + "grad_norm": 2.1964893341064453, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7607154846191406, + "num_tokens": 728343454.0, + "step": 27899 + }, + { + "epoch": 2.300651438937907, + "grad_norm": 2.022225856781006, + "learning_rate": 1e-06, + "loss": 0.6019, + "mean_token_accuracy": 0.8052828311920166, + "num_tokens": 728376881.0, + "step": 27900 + }, + { + "epoch": 2.3007338995629585, + "grad_norm": 2.060579299926758, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.785391628742218, + "num_tokens": 728409795.0, + "step": 27901 + }, + { + "epoch": 2.3008163601880103, + "grad_norm": 2.098564863204956, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7778978943824768, + "num_tokens": 728442194.0, + "step": 27902 + }, + { + "epoch": 2.3008988208130616, + "grad_norm": 2.0403714179992676, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7533738613128662, + "num_tokens": 728475890.0, + "step": 27903 + }, + { + "epoch": 2.3009812814381134, + "grad_norm": 2.136352777481079, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7702335119247437, + "num_tokens": 728503582.0, + "step": 27904 + }, + { + "epoch": 2.3010637420631648, + "grad_norm": 2.0273184776306152, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7765481472015381, + "num_tokens": 728532664.0, + "step": 27905 + }, + { + "epoch": 2.3011462026882166, + "grad_norm": 2.1143453121185303, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7619847655296326, + "num_tokens": 728563973.0, + "step": 27906 + }, + { + "epoch": 2.301228663313268, + "grad_norm": 1.7792963981628418, + "learning_rate": 1e-06, + "loss": 0.6046, + "mean_token_accuracy": 0.7987170219421387, + "num_tokens": 728601674.0, + "step": 27907 + }, + { + "epoch": 2.3013111239383193, + "grad_norm": 2.0369677543640137, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7760650515556335, + "num_tokens": 728632784.0, + "step": 27908 + }, + { + "epoch": 2.301393584563371, + "grad_norm": 2.1099112033843994, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7133307456970215, + "num_tokens": 728662993.0, + "step": 27909 + }, + { + "epoch": 2.3014760451884224, + "grad_norm": 2.5395255088806152, + "learning_rate": 1e-06, + "loss": 0.5583, + "mean_token_accuracy": 0.8088762760162354, + "num_tokens": 728683906.0, + "step": 27910 + }, + { + "epoch": 2.301558505813474, + "grad_norm": 2.3218634128570557, + "learning_rate": 1e-06, + "loss": 0.5843, + "mean_token_accuracy": 0.8030890226364136, + "num_tokens": 728708583.0, + "step": 27911 + }, + { + "epoch": 2.3016409664385256, + "grad_norm": 2.1671817302703857, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7588591575622559, + "num_tokens": 728734782.0, + "step": 27912 + }, + { + "epoch": 2.3017234270635774, + "grad_norm": 2.0181870460510254, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.7878676652908325, + "num_tokens": 728765546.0, + "step": 27913 + }, + { + "epoch": 2.3018058876886287, + "grad_norm": 2.327889919281006, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7431109547615051, + "num_tokens": 728792211.0, + "step": 27914 + }, + { + "epoch": 2.30188834831368, + "grad_norm": 2.155144453048706, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.7906595468521118, + "num_tokens": 728821195.0, + "step": 27915 + }, + { + "epoch": 2.301970808938732, + "grad_norm": 2.153782367706299, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.7917349338531494, + "num_tokens": 728850861.0, + "step": 27916 + }, + { + "epoch": 2.302053269563783, + "grad_norm": 1.9058657884597778, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7662789225578308, + "num_tokens": 728885786.0, + "step": 27917 + }, + { + "epoch": 2.302135730188835, + "grad_norm": 1.8213802576065063, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7731571197509766, + "num_tokens": 728922836.0, + "step": 27918 + }, + { + "epoch": 2.3022181908138863, + "grad_norm": 1.7804882526397705, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7801214456558228, + "num_tokens": 728961856.0, + "step": 27919 + }, + { + "epoch": 2.302300651438938, + "grad_norm": 2.2362020015716553, + "learning_rate": 1e-06, + "loss": 0.5302, + "mean_token_accuracy": 0.810327410697937, + "num_tokens": 728988264.0, + "step": 27920 + }, + { + "epoch": 2.3023831120639895, + "grad_norm": 2.0083703994750977, + "learning_rate": 1e-06, + "loss": 0.6139, + "mean_token_accuracy": 0.797205924987793, + "num_tokens": 729020298.0, + "step": 27921 + }, + { + "epoch": 2.302465572689041, + "grad_norm": 2.4442481994628906, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7704655528068542, + "num_tokens": 729047199.0, + "step": 27922 + }, + { + "epoch": 2.3025480333140926, + "grad_norm": 1.9717769622802734, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.7879327535629272, + "num_tokens": 729079894.0, + "step": 27923 + }, + { + "epoch": 2.302630493939144, + "grad_norm": 2.127411365509033, + "learning_rate": 1e-06, + "loss": 0.5736, + "mean_token_accuracy": 0.8027641773223877, + "num_tokens": 729108383.0, + "step": 27924 + }, + { + "epoch": 2.302712954564196, + "grad_norm": 2.159924268722534, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.792394757270813, + "num_tokens": 729138242.0, + "step": 27925 + }, + { + "epoch": 2.302795415189247, + "grad_norm": 1.9794392585754395, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.763197660446167, + "num_tokens": 729172042.0, + "step": 27926 + }, + { + "epoch": 2.3028778758142985, + "grad_norm": 1.9893810749053955, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7435568571090698, + "num_tokens": 729205768.0, + "step": 27927 + }, + { + "epoch": 2.3029603364393503, + "grad_norm": 2.3724403381347656, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7679950594902039, + "num_tokens": 729232274.0, + "step": 27928 + }, + { + "epoch": 2.3030427970644016, + "grad_norm": 1.9685677289962769, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7751919627189636, + "num_tokens": 729264688.0, + "step": 27929 + }, + { + "epoch": 2.3031252576894534, + "grad_norm": 2.2178468704223633, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.7811893820762634, + "num_tokens": 729289926.0, + "step": 27930 + }, + { + "epoch": 2.3032077183145048, + "grad_norm": 2.238657236099243, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7279046773910522, + "num_tokens": 729318353.0, + "step": 27931 + }, + { + "epoch": 2.303290178939556, + "grad_norm": 1.898952841758728, + "learning_rate": 1e-06, + "loss": 0.5297, + "mean_token_accuracy": 0.8246967792510986, + "num_tokens": 729351317.0, + "step": 27932 + }, + { + "epoch": 2.303372639564608, + "grad_norm": 2.226998805999756, + "learning_rate": 1e-06, + "loss": 0.5709, + "mean_token_accuracy": 0.8037871718406677, + "num_tokens": 729376988.0, + "step": 27933 + }, + { + "epoch": 2.3034551001896593, + "grad_norm": 1.98789381980896, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7806534171104431, + "num_tokens": 729411238.0, + "step": 27934 + }, + { + "epoch": 2.303537560814711, + "grad_norm": 2.1119911670684814, + "learning_rate": 1e-06, + "loss": 0.5269, + "mean_token_accuracy": 0.8234233856201172, + "num_tokens": 729441226.0, + "step": 27935 + }, + { + "epoch": 2.3036200214397624, + "grad_norm": 1.932045578956604, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.7990685701370239, + "num_tokens": 729474292.0, + "step": 27936 + }, + { + "epoch": 2.303702482064814, + "grad_norm": 2.154787540435791, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7537562251091003, + "num_tokens": 729506861.0, + "step": 27937 + }, + { + "epoch": 2.3037849426898656, + "grad_norm": 2.1035430431365967, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.781578779220581, + "num_tokens": 729534953.0, + "step": 27938 + }, + { + "epoch": 2.303867403314917, + "grad_norm": 2.3294310569763184, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7724448442459106, + "num_tokens": 729560842.0, + "step": 27939 + }, + { + "epoch": 2.3039498639399687, + "grad_norm": 1.910014033317566, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7732754945755005, + "num_tokens": 729594248.0, + "step": 27940 + }, + { + "epoch": 2.30403232456502, + "grad_norm": 2.03452205657959, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7802251577377319, + "num_tokens": 729625304.0, + "step": 27941 + }, + { + "epoch": 2.304114785190072, + "grad_norm": 2.056633949279785, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7589029669761658, + "num_tokens": 729658335.0, + "step": 27942 + }, + { + "epoch": 2.304197245815123, + "grad_norm": 2.045804500579834, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.7873682975769043, + "num_tokens": 729691601.0, + "step": 27943 + }, + { + "epoch": 2.304279706440175, + "grad_norm": 2.16336989402771, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7724149823188782, + "num_tokens": 729721570.0, + "step": 27944 + }, + { + "epoch": 2.3043621670652263, + "grad_norm": 1.9581397771835327, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.7959522604942322, + "num_tokens": 729751547.0, + "step": 27945 + }, + { + "epoch": 2.3044446276902777, + "grad_norm": 1.7843308448791504, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.7845043540000916, + "num_tokens": 729786807.0, + "step": 27946 + }, + { + "epoch": 2.3045270883153295, + "grad_norm": 1.7319493293762207, + "learning_rate": 1e-06, + "loss": 0.5196, + "mean_token_accuracy": 0.8153907656669617, + "num_tokens": 729823773.0, + "step": 27947 + }, + { + "epoch": 2.304609548940381, + "grad_norm": 2.0669353008270264, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.7986753582954407, + "num_tokens": 729855732.0, + "step": 27948 + }, + { + "epoch": 2.3046920095654326, + "grad_norm": 2.2437236309051514, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.7880630493164062, + "num_tokens": 729882209.0, + "step": 27949 + }, + { + "epoch": 2.304774470190484, + "grad_norm": 2.123878240585327, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7566455006599426, + "num_tokens": 729914294.0, + "step": 27950 + }, + { + "epoch": 2.3048569308155358, + "grad_norm": 2.0629191398620605, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.777831494808197, + "num_tokens": 729945887.0, + "step": 27951 + }, + { + "epoch": 2.304939391440587, + "grad_norm": 2.254082202911377, + "learning_rate": 1e-06, + "loss": 0.6126, + "mean_token_accuracy": 0.7870315313339233, + "num_tokens": 729972582.0, + "step": 27952 + }, + { + "epoch": 2.3050218520656385, + "grad_norm": 2.2359437942504883, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7494181394577026, + "num_tokens": 730002263.0, + "step": 27953 + }, + { + "epoch": 2.3051043126906903, + "grad_norm": 2.6008846759796143, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7534434199333191, + "num_tokens": 730024619.0, + "step": 27954 + }, + { + "epoch": 2.3051867733157416, + "grad_norm": 2.660736560821533, + "learning_rate": 1e-06, + "loss": 0.5633, + "mean_token_accuracy": 0.8086245059967041, + "num_tokens": 730057729.0, + "step": 27955 + }, + { + "epoch": 2.3052692339407934, + "grad_norm": 2.332947015762329, + "learning_rate": 1e-06, + "loss": 0.6145, + "mean_token_accuracy": 0.7959820032119751, + "num_tokens": 730081523.0, + "step": 27956 + }, + { + "epoch": 2.3053516945658448, + "grad_norm": 1.948377013206482, + "learning_rate": 1e-06, + "loss": 0.5928, + "mean_token_accuracy": 0.7983863353729248, + "num_tokens": 730116469.0, + "step": 27957 + }, + { + "epoch": 2.3054341551908966, + "grad_norm": 1.9335484504699707, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7785203456878662, + "num_tokens": 730150190.0, + "step": 27958 + }, + { + "epoch": 2.305516615815948, + "grad_norm": 2.19119930267334, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7789558172225952, + "num_tokens": 730180872.0, + "step": 27959 + }, + { + "epoch": 2.3055990764409993, + "grad_norm": 1.8295148611068726, + "learning_rate": 1e-06, + "loss": 0.6095, + "mean_token_accuracy": 0.7983558177947998, + "num_tokens": 730216966.0, + "step": 27960 + }, + { + "epoch": 2.305681537066051, + "grad_norm": 2.2759957313537598, + "learning_rate": 1e-06, + "loss": 0.7236, + "mean_token_accuracy": 0.7712392807006836, + "num_tokens": 730243848.0, + "step": 27961 + }, + { + "epoch": 2.3057639976911024, + "grad_norm": 2.2000834941864014, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7691983580589294, + "num_tokens": 730274364.0, + "step": 27962 + }, + { + "epoch": 2.305846458316154, + "grad_norm": 1.9209316968917847, + "learning_rate": 1e-06, + "loss": 0.5431, + "mean_token_accuracy": 0.8084031343460083, + "num_tokens": 730307076.0, + "step": 27963 + }, + { + "epoch": 2.3059289189412056, + "grad_norm": 2.262077808380127, + "learning_rate": 1e-06, + "loss": 0.7441, + "mean_token_accuracy": 0.7631644010543823, + "num_tokens": 730334818.0, + "step": 27964 + }, + { + "epoch": 2.3060113795662573, + "grad_norm": 1.762423038482666, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.797145426273346, + "num_tokens": 730373907.0, + "step": 27965 + }, + { + "epoch": 2.3060938401913087, + "grad_norm": 2.0332534313201904, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7537953853607178, + "num_tokens": 730403843.0, + "step": 27966 + }, + { + "epoch": 2.30617630081636, + "grad_norm": 2.014644145965576, + "learning_rate": 1e-06, + "loss": 0.5717, + "mean_token_accuracy": 0.7977825403213501, + "num_tokens": 730436229.0, + "step": 27967 + }, + { + "epoch": 2.306258761441412, + "grad_norm": 2.4339489936828613, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7464045286178589, + "num_tokens": 730461586.0, + "step": 27968 + }, + { + "epoch": 2.306341222066463, + "grad_norm": 2.457974672317505, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7538991570472717, + "num_tokens": 730485835.0, + "step": 27969 + }, + { + "epoch": 2.306423682691515, + "grad_norm": 2.262885332107544, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.762497067451477, + "num_tokens": 730514567.0, + "step": 27970 + }, + { + "epoch": 2.3065061433165663, + "grad_norm": 1.9472894668579102, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.7701879143714905, + "num_tokens": 730549523.0, + "step": 27971 + }, + { + "epoch": 2.3065886039416177, + "grad_norm": 1.8802497386932373, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.764586329460144, + "num_tokens": 730585199.0, + "step": 27972 + }, + { + "epoch": 2.3066710645666695, + "grad_norm": 2.2511608600616455, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7443726062774658, + "num_tokens": 730616787.0, + "step": 27973 + }, + { + "epoch": 2.306753525191721, + "grad_norm": 2.2723584175109863, + "learning_rate": 1e-06, + "loss": 0.5988, + "mean_token_accuracy": 0.7970477342605591, + "num_tokens": 730641475.0, + "step": 27974 + }, + { + "epoch": 2.3068359858167726, + "grad_norm": 2.1592118740081787, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7638390064239502, + "num_tokens": 730670328.0, + "step": 27975 + }, + { + "epoch": 2.306918446441824, + "grad_norm": 2.0167887210845947, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.7585383057594299, + "num_tokens": 730702159.0, + "step": 27976 + }, + { + "epoch": 2.3070009070668753, + "grad_norm": 2.075331926345825, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.792976438999176, + "num_tokens": 730730499.0, + "step": 27977 + }, + { + "epoch": 2.307083367691927, + "grad_norm": 1.92037034034729, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.760155975818634, + "num_tokens": 730768438.0, + "step": 27978 + }, + { + "epoch": 2.3071658283169785, + "grad_norm": 2.1574974060058594, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.7857245802879333, + "num_tokens": 730799007.0, + "step": 27979 + }, + { + "epoch": 2.3072482889420303, + "grad_norm": 1.9154983758926392, + "learning_rate": 1e-06, + "loss": 0.5163, + "mean_token_accuracy": 0.824211597442627, + "num_tokens": 730827921.0, + "step": 27980 + }, + { + "epoch": 2.3073307495670816, + "grad_norm": 2.010883092880249, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7699165940284729, + "num_tokens": 730859068.0, + "step": 27981 + }, + { + "epoch": 2.3074132101921334, + "grad_norm": 1.8263307809829712, + "learning_rate": 1e-06, + "loss": 0.5243, + "mean_token_accuracy": 0.8235554099082947, + "num_tokens": 730894343.0, + "step": 27982 + }, + { + "epoch": 2.3074956708171848, + "grad_norm": 2.2027034759521484, + "learning_rate": 1e-06, + "loss": 0.5937, + "mean_token_accuracy": 0.7957656383514404, + "num_tokens": 730922636.0, + "step": 27983 + }, + { + "epoch": 2.307578131442236, + "grad_norm": 2.0138535499572754, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7805036306381226, + "num_tokens": 730954762.0, + "step": 27984 + }, + { + "epoch": 2.307660592067288, + "grad_norm": 1.953079342842102, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.7888044118881226, + "num_tokens": 730985225.0, + "step": 27985 + }, + { + "epoch": 2.3077430526923393, + "grad_norm": 1.9593448638916016, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7730414271354675, + "num_tokens": 731021718.0, + "step": 27986 + }, + { + "epoch": 2.307825513317391, + "grad_norm": 2.005944013595581, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.76451176404953, + "num_tokens": 731056367.0, + "step": 27987 + }, + { + "epoch": 2.3079079739424424, + "grad_norm": 2.0095088481903076, + "learning_rate": 1e-06, + "loss": 0.6042, + "mean_token_accuracy": 0.7926506996154785, + "num_tokens": 731089860.0, + "step": 27988 + }, + { + "epoch": 2.307990434567494, + "grad_norm": 2.0826761722564697, + "learning_rate": 1e-06, + "loss": 0.5181, + "mean_token_accuracy": 0.8316524028778076, + "num_tokens": 731116828.0, + "step": 27989 + }, + { + "epoch": 2.3080728951925455, + "grad_norm": 1.9817030429840088, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.7979139089584351, + "num_tokens": 731151211.0, + "step": 27990 + }, + { + "epoch": 2.308155355817597, + "grad_norm": 2.597978353500366, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7721183896064758, + "num_tokens": 731172772.0, + "step": 27991 + }, + { + "epoch": 2.3082378164426487, + "grad_norm": 1.9922562837600708, + "learning_rate": 1e-06, + "loss": 0.6004, + "mean_token_accuracy": 0.8001502156257629, + "num_tokens": 731205725.0, + "step": 27992 + }, + { + "epoch": 2.3083202770677, + "grad_norm": 2.1337761878967285, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7764568328857422, + "num_tokens": 731233878.0, + "step": 27993 + }, + { + "epoch": 2.308402737692752, + "grad_norm": 2.3872973918914795, + "learning_rate": 1e-06, + "loss": 0.6084, + "mean_token_accuracy": 0.7940196990966797, + "num_tokens": 731255522.0, + "step": 27994 + }, + { + "epoch": 2.308485198317803, + "grad_norm": 1.8450534343719482, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7362080812454224, + "num_tokens": 731294498.0, + "step": 27995 + }, + { + "epoch": 2.308567658942855, + "grad_norm": 2.15649676322937, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.7835841178894043, + "num_tokens": 731320948.0, + "step": 27996 + }, + { + "epoch": 2.3086501195679063, + "grad_norm": 2.1569368839263916, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.7843651175498962, + "num_tokens": 731348630.0, + "step": 27997 + }, + { + "epoch": 2.3087325801929577, + "grad_norm": 2.009215831756592, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7577312588691711, + "num_tokens": 731381563.0, + "step": 27998 + }, + { + "epoch": 2.3088150408180095, + "grad_norm": 2.137136697769165, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.7726877927780151, + "num_tokens": 731414136.0, + "step": 27999 + }, + { + "epoch": 2.308897501443061, + "grad_norm": 1.930345058441162, + "learning_rate": 1e-06, + "loss": 0.6093, + "mean_token_accuracy": 0.7961173057556152, + "num_tokens": 731444973.0, + "step": 28000 + }, + { + "epoch": 2.3089799620681126, + "grad_norm": 2.206693649291992, + "learning_rate": 1e-06, + "loss": 0.5868, + "mean_token_accuracy": 0.8033390045166016, + "num_tokens": 731472349.0, + "step": 28001 + }, + { + "epoch": 2.309062422693164, + "grad_norm": 2.244274616241455, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.7913727164268494, + "num_tokens": 731498512.0, + "step": 28002 + }, + { + "epoch": 2.3091448833182158, + "grad_norm": 2.1766490936279297, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.7572804093360901, + "num_tokens": 731529225.0, + "step": 28003 + }, + { + "epoch": 2.309227343943267, + "grad_norm": 2.279552936553955, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.7813084125518799, + "num_tokens": 731555241.0, + "step": 28004 + }, + { + "epoch": 2.3093098045683185, + "grad_norm": 1.9744666814804077, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.7889903783798218, + "num_tokens": 731586846.0, + "step": 28005 + }, + { + "epoch": 2.3093922651933703, + "grad_norm": 2.081854820251465, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7628912925720215, + "num_tokens": 731616553.0, + "step": 28006 + }, + { + "epoch": 2.3094747258184216, + "grad_norm": 2.390913724899292, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7538607120513916, + "num_tokens": 731641159.0, + "step": 28007 + }, + { + "epoch": 2.3095571864434734, + "grad_norm": 1.9657552242279053, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7635972499847412, + "num_tokens": 731675482.0, + "step": 28008 + }, + { + "epoch": 2.3096396470685248, + "grad_norm": 2.1293275356292725, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7777270078659058, + "num_tokens": 731702257.0, + "step": 28009 + }, + { + "epoch": 2.3097221076935766, + "grad_norm": 2.030287504196167, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7909843325614929, + "num_tokens": 731735701.0, + "step": 28010 + }, + { + "epoch": 2.309804568318628, + "grad_norm": 2.6600074768066406, + "learning_rate": 1e-06, + "loss": 0.5681, + "mean_token_accuracy": 0.7979327440261841, + "num_tokens": 731755259.0, + "step": 28011 + }, + { + "epoch": 2.3098870289436793, + "grad_norm": 1.9897286891937256, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.7947260141372681, + "num_tokens": 731784221.0, + "step": 28012 + }, + { + "epoch": 2.309969489568731, + "grad_norm": 2.1762850284576416, + "learning_rate": 1e-06, + "loss": 0.5049, + "mean_token_accuracy": 0.82878577709198, + "num_tokens": 731810542.0, + "step": 28013 + }, + { + "epoch": 2.3100519501937824, + "grad_norm": 2.1941378116607666, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7774935364723206, + "num_tokens": 731840227.0, + "step": 28014 + }, + { + "epoch": 2.310134410818834, + "grad_norm": 1.900841474533081, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7433849573135376, + "num_tokens": 731877710.0, + "step": 28015 + }, + { + "epoch": 2.3102168714438855, + "grad_norm": 2.108219861984253, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7862122058868408, + "num_tokens": 731905414.0, + "step": 28016 + }, + { + "epoch": 2.310299332068937, + "grad_norm": 1.8963299989700317, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.7918347716331482, + "num_tokens": 731940646.0, + "step": 28017 + }, + { + "epoch": 2.3103817926939887, + "grad_norm": 2.2170302867889404, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7653346657752991, + "num_tokens": 731969713.0, + "step": 28018 + }, + { + "epoch": 2.31046425331904, + "grad_norm": 1.9150974750518799, + "learning_rate": 1e-06, + "loss": 0.5409, + "mean_token_accuracy": 0.8047630786895752, + "num_tokens": 732002388.0, + "step": 28019 + }, + { + "epoch": 2.310546713944092, + "grad_norm": 2.1756629943847656, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7321594953536987, + "num_tokens": 732033204.0, + "step": 28020 + }, + { + "epoch": 2.310629174569143, + "grad_norm": 2.271679639816284, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7538852691650391, + "num_tokens": 732059935.0, + "step": 28021 + }, + { + "epoch": 2.3107116351941945, + "grad_norm": 2.6757004261016846, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7363423705101013, + "num_tokens": 732080336.0, + "step": 28022 + }, + { + "epoch": 2.3107940958192463, + "grad_norm": 2.1007063388824463, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.7938014268875122, + "num_tokens": 732108384.0, + "step": 28023 + }, + { + "epoch": 2.3108765564442977, + "grad_norm": 1.8556931018829346, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.793166995048523, + "num_tokens": 732146147.0, + "step": 28024 + }, + { + "epoch": 2.3109590170693495, + "grad_norm": 2.0430614948272705, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7738004922866821, + "num_tokens": 732181424.0, + "step": 28025 + }, + { + "epoch": 2.311041477694401, + "grad_norm": 2.345109224319458, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7403422594070435, + "num_tokens": 732207210.0, + "step": 28026 + }, + { + "epoch": 2.3111239383194526, + "grad_norm": 2.161137104034424, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7445067167282104, + "num_tokens": 732235692.0, + "step": 28027 + }, + { + "epoch": 2.311206398944504, + "grad_norm": 2.506113290786743, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7459651827812195, + "num_tokens": 732257782.0, + "step": 28028 + }, + { + "epoch": 2.3112888595695553, + "grad_norm": 2.0005650520324707, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7648744583129883, + "num_tokens": 732289535.0, + "step": 28029 + }, + { + "epoch": 2.311371320194607, + "grad_norm": 2.128473997116089, + "learning_rate": 1e-06, + "loss": 0.7698, + "mean_token_accuracy": 0.7511861324310303, + "num_tokens": 732320715.0, + "step": 28030 + }, + { + "epoch": 2.3114537808196585, + "grad_norm": 2.0275185108184814, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.7816641330718994, + "num_tokens": 732348690.0, + "step": 28031 + }, + { + "epoch": 2.3115362414447103, + "grad_norm": 2.1070539951324463, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7593256235122681, + "num_tokens": 732379275.0, + "step": 28032 + }, + { + "epoch": 2.3116187020697616, + "grad_norm": 2.2410614490509033, + "learning_rate": 1e-06, + "loss": 0.6925, + "mean_token_accuracy": 0.7798889875411987, + "num_tokens": 732407772.0, + "step": 28033 + }, + { + "epoch": 2.3117011626948134, + "grad_norm": 2.005309581756592, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.7781838774681091, + "num_tokens": 732438432.0, + "step": 28034 + }, + { + "epoch": 2.3117836233198648, + "grad_norm": 2.420149087905884, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7844076156616211, + "num_tokens": 732463636.0, + "step": 28035 + }, + { + "epoch": 2.311866083944916, + "grad_norm": 2.068464994430542, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.7944044470787048, + "num_tokens": 732492357.0, + "step": 28036 + }, + { + "epoch": 2.311948544569968, + "grad_norm": 1.9773255586624146, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.7621562480926514, + "num_tokens": 732527898.0, + "step": 28037 + }, + { + "epoch": 2.3120310051950193, + "grad_norm": 2.240584135055542, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.7839822173118591, + "num_tokens": 732554932.0, + "step": 28038 + }, + { + "epoch": 2.312113465820071, + "grad_norm": 2.049147844314575, + "learning_rate": 1e-06, + "loss": 0.5104, + "mean_token_accuracy": 0.8260805010795593, + "num_tokens": 732586249.0, + "step": 28039 + }, + { + "epoch": 2.3121959264451224, + "grad_norm": 1.9387246370315552, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7375762462615967, + "num_tokens": 732620640.0, + "step": 28040 + }, + { + "epoch": 2.312278387070174, + "grad_norm": 2.0335893630981445, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7447335720062256, + "num_tokens": 732654560.0, + "step": 28041 + }, + { + "epoch": 2.3123608476952255, + "grad_norm": 2.265960931777954, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7749847173690796, + "num_tokens": 732680159.0, + "step": 28042 + }, + { + "epoch": 2.312443308320277, + "grad_norm": 1.800787329673767, + "learning_rate": 1e-06, + "loss": 0.518, + "mean_token_accuracy": 0.8231099843978882, + "num_tokens": 732715997.0, + "step": 28043 + }, + { + "epoch": 2.3125257689453287, + "grad_norm": 2.3623080253601074, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.7818083763122559, + "num_tokens": 732739725.0, + "step": 28044 + }, + { + "epoch": 2.31260822957038, + "grad_norm": 2.1338050365448, + "learning_rate": 1e-06, + "loss": 0.5811, + "mean_token_accuracy": 0.806432843208313, + "num_tokens": 732768028.0, + "step": 28045 + }, + { + "epoch": 2.312690690195432, + "grad_norm": 2.0614070892333984, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7605533599853516, + "num_tokens": 732798318.0, + "step": 28046 + }, + { + "epoch": 2.312773150820483, + "grad_norm": 2.297579288482666, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7413879036903381, + "num_tokens": 732826353.0, + "step": 28047 + }, + { + "epoch": 2.312855611445535, + "grad_norm": 2.1286981105804443, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7575981616973877, + "num_tokens": 732858310.0, + "step": 28048 + }, + { + "epoch": 2.3129380720705863, + "grad_norm": 2.0186166763305664, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.7857017517089844, + "num_tokens": 732891573.0, + "step": 28049 + }, + { + "epoch": 2.3130205326956377, + "grad_norm": 1.8677297830581665, + "learning_rate": 1e-06, + "loss": 0.5423, + "mean_token_accuracy": 0.8208246231079102, + "num_tokens": 732922979.0, + "step": 28050 + }, + { + "epoch": 2.3131029933206895, + "grad_norm": 2.0919549465179443, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.767230212688446, + "num_tokens": 732956925.0, + "step": 28051 + }, + { + "epoch": 2.313185453945741, + "grad_norm": 2.1997063159942627, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.8032403588294983, + "num_tokens": 732983297.0, + "step": 28052 + }, + { + "epoch": 2.3132679145707926, + "grad_norm": 2.0263454914093018, + "learning_rate": 1e-06, + "loss": 0.6212, + "mean_token_accuracy": 0.7958276271820068, + "num_tokens": 733013431.0, + "step": 28053 + }, + { + "epoch": 2.313350375195844, + "grad_norm": 2.2738728523254395, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7747534513473511, + "num_tokens": 733037496.0, + "step": 28054 + }, + { + "epoch": 2.3134328358208958, + "grad_norm": 1.9339548349380493, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7487195730209351, + "num_tokens": 733071771.0, + "step": 28055 + }, + { + "epoch": 2.313515296445947, + "grad_norm": 2.4826772212982178, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.7794609665870667, + "num_tokens": 733092172.0, + "step": 28056 + }, + { + "epoch": 2.3135977570709985, + "grad_norm": 1.979810357093811, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.749670147895813, + "num_tokens": 733124552.0, + "step": 28057 + }, + { + "epoch": 2.3136802176960503, + "grad_norm": 2.2789485454559326, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.769964337348938, + "num_tokens": 733151122.0, + "step": 28058 + }, + { + "epoch": 2.3137626783211016, + "grad_norm": 2.092085123062134, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7823113799095154, + "num_tokens": 733179914.0, + "step": 28059 + }, + { + "epoch": 2.3138451389461534, + "grad_norm": 2.2271835803985596, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7415382862091064, + "num_tokens": 733208270.0, + "step": 28060 + }, + { + "epoch": 2.3139275995712048, + "grad_norm": 2.438608407974243, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7477298974990845, + "num_tokens": 733232391.0, + "step": 28061 + }, + { + "epoch": 2.314010060196256, + "grad_norm": 2.4536616802215576, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7758194208145142, + "num_tokens": 733256682.0, + "step": 28062 + }, + { + "epoch": 2.314092520821308, + "grad_norm": 2.2285478115081787, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.7878894805908203, + "num_tokens": 733283012.0, + "step": 28063 + }, + { + "epoch": 2.3141749814463592, + "grad_norm": 2.1062848567962646, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7685997486114502, + "num_tokens": 733312149.0, + "step": 28064 + }, + { + "epoch": 2.314257442071411, + "grad_norm": 2.0349483489990234, + "learning_rate": 1e-06, + "loss": 0.5411, + "mean_token_accuracy": 0.8108684420585632, + "num_tokens": 733342815.0, + "step": 28065 + }, + { + "epoch": 2.3143399026964624, + "grad_norm": 2.0088932514190674, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7843470573425293, + "num_tokens": 733374312.0, + "step": 28066 + }, + { + "epoch": 2.3144223633215137, + "grad_norm": 1.9989511966705322, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.773307204246521, + "num_tokens": 733410356.0, + "step": 28067 + }, + { + "epoch": 2.3145048239465655, + "grad_norm": 2.0651752948760986, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.7755805253982544, + "num_tokens": 733440799.0, + "step": 28068 + }, + { + "epoch": 2.314587284571617, + "grad_norm": 1.9445483684539795, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.7958562970161438, + "num_tokens": 733474584.0, + "step": 28069 + }, + { + "epoch": 2.3146697451966687, + "grad_norm": 1.992875576019287, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7525818347930908, + "num_tokens": 733509044.0, + "step": 28070 + }, + { + "epoch": 2.31475220582172, + "grad_norm": 2.146632671356201, + "learning_rate": 1e-06, + "loss": 0.6424, + "mean_token_accuracy": 0.8019905090332031, + "num_tokens": 733535314.0, + "step": 28071 + }, + { + "epoch": 2.314834666446772, + "grad_norm": 2.067276954650879, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7497454285621643, + "num_tokens": 733567173.0, + "step": 28072 + }, + { + "epoch": 2.314917127071823, + "grad_norm": 2.186668872833252, + "learning_rate": 1e-06, + "loss": 0.5847, + "mean_token_accuracy": 0.7937110662460327, + "num_tokens": 733592801.0, + "step": 28073 + }, + { + "epoch": 2.3149995876968745, + "grad_norm": 1.9821975231170654, + "learning_rate": 1e-06, + "loss": 0.494, + "mean_token_accuracy": 0.825154185295105, + "num_tokens": 733620870.0, + "step": 28074 + }, + { + "epoch": 2.3150820483219263, + "grad_norm": 2.0002152919769287, + "learning_rate": 1e-06, + "loss": 0.5299, + "mean_token_accuracy": 0.8197630047798157, + "num_tokens": 733652666.0, + "step": 28075 + }, + { + "epoch": 2.3151645089469777, + "grad_norm": 1.8636680841445923, + "learning_rate": 1e-06, + "loss": 0.5747, + "mean_token_accuracy": 0.8127855658531189, + "num_tokens": 733686456.0, + "step": 28076 + }, + { + "epoch": 2.3152469695720295, + "grad_norm": 2.009761095046997, + "learning_rate": 1e-06, + "loss": 0.5698, + "mean_token_accuracy": 0.8095685243606567, + "num_tokens": 733718606.0, + "step": 28077 + }, + { + "epoch": 2.315329430197081, + "grad_norm": 2.0206830501556396, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7439956665039062, + "num_tokens": 733752675.0, + "step": 28078 + }, + { + "epoch": 2.3154118908221326, + "grad_norm": 2.272078037261963, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7541746497154236, + "num_tokens": 733778947.0, + "step": 28079 + }, + { + "epoch": 2.315494351447184, + "grad_norm": 2.283634901046753, + "learning_rate": 1e-06, + "loss": 0.79, + "mean_token_accuracy": 0.7426930665969849, + "num_tokens": 733806285.0, + "step": 28080 + }, + { + "epoch": 2.3155768120722353, + "grad_norm": 2.2951695919036865, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7659502625465393, + "num_tokens": 733831423.0, + "step": 28081 + }, + { + "epoch": 2.315659272697287, + "grad_norm": 1.9097120761871338, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.7982097268104553, + "num_tokens": 733864601.0, + "step": 28082 + }, + { + "epoch": 2.3157417333223385, + "grad_norm": 1.981230616569519, + "learning_rate": 1e-06, + "loss": 0.6303, + "mean_token_accuracy": 0.7948063015937805, + "num_tokens": 733897787.0, + "step": 28083 + }, + { + "epoch": 2.3158241939473903, + "grad_norm": 2.038158655166626, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7575051784515381, + "num_tokens": 733929207.0, + "step": 28084 + }, + { + "epoch": 2.3159066545724416, + "grad_norm": 1.916527509689331, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.7965542078018188, + "num_tokens": 733963273.0, + "step": 28085 + }, + { + "epoch": 2.3159891151974934, + "grad_norm": 2.000234842300415, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7463801503181458, + "num_tokens": 733995785.0, + "step": 28086 + }, + { + "epoch": 2.3160715758225447, + "grad_norm": 2.005383253097534, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7781177163124084, + "num_tokens": 734032113.0, + "step": 28087 + }, + { + "epoch": 2.316154036447596, + "grad_norm": 2.056199073791504, + "learning_rate": 1e-06, + "loss": 0.5933, + "mean_token_accuracy": 0.7992104291915894, + "num_tokens": 734060421.0, + "step": 28088 + }, + { + "epoch": 2.316236497072648, + "grad_norm": 2.1617796421051025, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7259891629219055, + "num_tokens": 734091718.0, + "step": 28089 + }, + { + "epoch": 2.3163189576976992, + "grad_norm": 2.2041187286376953, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7683682441711426, + "num_tokens": 734120982.0, + "step": 28090 + }, + { + "epoch": 2.316401418322751, + "grad_norm": 2.142759323120117, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7829590439796448, + "num_tokens": 734148930.0, + "step": 28091 + }, + { + "epoch": 2.3164838789478024, + "grad_norm": 1.8559107780456543, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7575072050094604, + "num_tokens": 734189187.0, + "step": 28092 + }, + { + "epoch": 2.316566339572854, + "grad_norm": 2.2333571910858154, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7628977298736572, + "num_tokens": 734219042.0, + "step": 28093 + }, + { + "epoch": 2.3166488001979055, + "grad_norm": 1.985429286956787, + "learning_rate": 1e-06, + "loss": 0.5934, + "mean_token_accuracy": 0.8005363941192627, + "num_tokens": 734250212.0, + "step": 28094 + }, + { + "epoch": 2.316731260822957, + "grad_norm": 2.1426842212677, + "learning_rate": 1e-06, + "loss": 0.5535, + "mean_token_accuracy": 0.8046116232872009, + "num_tokens": 734280165.0, + "step": 28095 + }, + { + "epoch": 2.3168137214480087, + "grad_norm": 2.0323729515075684, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7834008932113647, + "num_tokens": 734313654.0, + "step": 28096 + }, + { + "epoch": 2.31689618207306, + "grad_norm": 2.0144121646881104, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7947113513946533, + "num_tokens": 734344098.0, + "step": 28097 + }, + { + "epoch": 2.316978642698112, + "grad_norm": 2.208834409713745, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.7811112999916077, + "num_tokens": 734373113.0, + "step": 28098 + }, + { + "epoch": 2.317061103323163, + "grad_norm": 2.2632393836975098, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.7951081991195679, + "num_tokens": 734401605.0, + "step": 28099 + }, + { + "epoch": 2.317143563948215, + "grad_norm": 1.9363043308258057, + "learning_rate": 1e-06, + "loss": 0.5254, + "mean_token_accuracy": 0.81315678358078, + "num_tokens": 734433933.0, + "step": 28100 + }, + { + "epoch": 2.3172260245732663, + "grad_norm": 2.031893730163574, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.7615644931793213, + "num_tokens": 734462881.0, + "step": 28101 + }, + { + "epoch": 2.3173084851983177, + "grad_norm": 1.9609845876693726, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7725865244865417, + "num_tokens": 734493482.0, + "step": 28102 + }, + { + "epoch": 2.3173909458233695, + "grad_norm": 2.4493260383605957, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7442461252212524, + "num_tokens": 734517000.0, + "step": 28103 + }, + { + "epoch": 2.317473406448421, + "grad_norm": 2.0742344856262207, + "learning_rate": 1e-06, + "loss": 0.5238, + "mean_token_accuracy": 0.8179061412811279, + "num_tokens": 734547461.0, + "step": 28104 + }, + { + "epoch": 2.3175558670734726, + "grad_norm": 2.0413618087768555, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7734645009040833, + "num_tokens": 734579757.0, + "step": 28105 + }, + { + "epoch": 2.317638327698524, + "grad_norm": 1.8974045515060425, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.7971224784851074, + "num_tokens": 734613640.0, + "step": 28106 + }, + { + "epoch": 2.3177207883235753, + "grad_norm": 1.9358913898468018, + "learning_rate": 1e-06, + "loss": 0.6177, + "mean_token_accuracy": 0.7988433837890625, + "num_tokens": 734647132.0, + "step": 28107 + }, + { + "epoch": 2.317803248948627, + "grad_norm": 2.4347755908966064, + "learning_rate": 1e-06, + "loss": 0.7588, + "mean_token_accuracy": 0.7524153590202332, + "num_tokens": 734672616.0, + "step": 28108 + }, + { + "epoch": 2.3178857095736785, + "grad_norm": 2.060558557510376, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7585422396659851, + "num_tokens": 734702462.0, + "step": 28109 + }, + { + "epoch": 2.3179681701987302, + "grad_norm": 2.332542896270752, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7760650515556335, + "num_tokens": 734727264.0, + "step": 28110 + }, + { + "epoch": 2.3180506308237816, + "grad_norm": 2.1810734272003174, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7393265962600708, + "num_tokens": 734757525.0, + "step": 28111 + }, + { + "epoch": 2.318133091448833, + "grad_norm": 2.4753713607788086, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7611460089683533, + "num_tokens": 734781253.0, + "step": 28112 + }, + { + "epoch": 2.3182155520738847, + "grad_norm": 2.211961507797241, + "learning_rate": 1e-06, + "loss": 0.5894, + "mean_token_accuracy": 0.8059970140457153, + "num_tokens": 734807038.0, + "step": 28113 + }, + { + "epoch": 2.318298012698936, + "grad_norm": 2.103200674057007, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7728418111801147, + "num_tokens": 734838741.0, + "step": 28114 + }, + { + "epoch": 2.318380473323988, + "grad_norm": 1.989261507987976, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7687990069389343, + "num_tokens": 734870288.0, + "step": 28115 + }, + { + "epoch": 2.3184629339490392, + "grad_norm": 1.8022964000701904, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.7837861180305481, + "num_tokens": 734910506.0, + "step": 28116 + }, + { + "epoch": 2.318545394574091, + "grad_norm": 1.994003176689148, + "learning_rate": 1e-06, + "loss": 0.6, + "mean_token_accuracy": 0.8102153539657593, + "num_tokens": 734937391.0, + "step": 28117 + }, + { + "epoch": 2.3186278551991424, + "grad_norm": 1.9602893590927124, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7931225299835205, + "num_tokens": 734970668.0, + "step": 28118 + }, + { + "epoch": 2.3187103158241937, + "grad_norm": 2.1602864265441895, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7790583372116089, + "num_tokens": 734998449.0, + "step": 28119 + }, + { + "epoch": 2.3187927764492455, + "grad_norm": 2.289687395095825, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7675844430923462, + "num_tokens": 735026750.0, + "step": 28120 + }, + { + "epoch": 2.318875237074297, + "grad_norm": 2.291093349456787, + "learning_rate": 1e-06, + "loss": 0.7384, + "mean_token_accuracy": 0.7564754486083984, + "num_tokens": 735051980.0, + "step": 28121 + }, + { + "epoch": 2.3189576976993487, + "grad_norm": 2.2395105361938477, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7617461681365967, + "num_tokens": 735080892.0, + "step": 28122 + }, + { + "epoch": 2.3190401583244, + "grad_norm": 1.946163535118103, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.7931549549102783, + "num_tokens": 735117417.0, + "step": 28123 + }, + { + "epoch": 2.319122618949452, + "grad_norm": 1.8579576015472412, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.7864314317703247, + "num_tokens": 735156393.0, + "step": 28124 + }, + { + "epoch": 2.319205079574503, + "grad_norm": 2.246821403503418, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7483001947402954, + "num_tokens": 735182027.0, + "step": 28125 + }, + { + "epoch": 2.3192875401995545, + "grad_norm": 2.018184185028076, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.7896599769592285, + "num_tokens": 735213417.0, + "step": 28126 + }, + { + "epoch": 2.3193700008246063, + "grad_norm": 2.0261430740356445, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.7943598031997681, + "num_tokens": 735246005.0, + "step": 28127 + }, + { + "epoch": 2.3194524614496577, + "grad_norm": 2.270071268081665, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7654075622558594, + "num_tokens": 735274891.0, + "step": 28128 + }, + { + "epoch": 2.3195349220747095, + "grad_norm": 2.1778383255004883, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.7885276079177856, + "num_tokens": 735303678.0, + "step": 28129 + }, + { + "epoch": 2.319617382699761, + "grad_norm": 1.8449617624282837, + "learning_rate": 1e-06, + "loss": 0.7762, + "mean_token_accuracy": 0.7520588636398315, + "num_tokens": 735339250.0, + "step": 28130 + }, + { + "epoch": 2.3196998433248126, + "grad_norm": 2.2840960025787354, + "learning_rate": 1e-06, + "loss": 0.8188, + "mean_token_accuracy": 0.7438240051269531, + "num_tokens": 735366306.0, + "step": 28131 + }, + { + "epoch": 2.319782303949864, + "grad_norm": 2.3909847736358643, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7746847867965698, + "num_tokens": 735390743.0, + "step": 28132 + }, + { + "epoch": 2.3198647645749153, + "grad_norm": 2.6271233558654785, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7276779413223267, + "num_tokens": 735412554.0, + "step": 28133 + }, + { + "epoch": 2.319947225199967, + "grad_norm": 2.408540964126587, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7675070762634277, + "num_tokens": 735437703.0, + "step": 28134 + }, + { + "epoch": 2.3200296858250185, + "grad_norm": 2.140697717666626, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7731630802154541, + "num_tokens": 735467421.0, + "step": 28135 + }, + { + "epoch": 2.3201121464500702, + "grad_norm": 2.1847991943359375, + "learning_rate": 1e-06, + "loss": 0.6776, + "mean_token_accuracy": 0.7821646928787231, + "num_tokens": 735494122.0, + "step": 28136 + }, + { + "epoch": 2.3201946070751216, + "grad_norm": 2.5726308822631836, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7482717037200928, + "num_tokens": 735515579.0, + "step": 28137 + }, + { + "epoch": 2.3202770677001734, + "grad_norm": 2.5327768325805664, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7533957958221436, + "num_tokens": 735540161.0, + "step": 28138 + }, + { + "epoch": 2.3203595283252247, + "grad_norm": 1.7286711931228638, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.7825857400894165, + "num_tokens": 735580779.0, + "step": 28139 + }, + { + "epoch": 2.320441988950276, + "grad_norm": 2.2056782245635986, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.736722469329834, + "num_tokens": 735609727.0, + "step": 28140 + }, + { + "epoch": 2.320524449575328, + "grad_norm": 1.9661587476730347, + "learning_rate": 1e-06, + "loss": 0.5862, + "mean_token_accuracy": 0.810450553894043, + "num_tokens": 735641233.0, + "step": 28141 + }, + { + "epoch": 2.3206069102003792, + "grad_norm": 2.126765251159668, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7600212097167969, + "num_tokens": 735671983.0, + "step": 28142 + }, + { + "epoch": 2.320689370825431, + "grad_norm": 2.2695157527923584, + "learning_rate": 1e-06, + "loss": 0.5718, + "mean_token_accuracy": 0.7992510199546814, + "num_tokens": 735698198.0, + "step": 28143 + }, + { + "epoch": 2.3207718314504824, + "grad_norm": 2.2098143100738525, + "learning_rate": 1e-06, + "loss": 0.5886, + "mean_token_accuracy": 0.7941503524780273, + "num_tokens": 735725162.0, + "step": 28144 + }, + { + "epoch": 2.320854292075534, + "grad_norm": 2.0063116550445557, + "learning_rate": 1e-06, + "loss": 0.6127, + "mean_token_accuracy": 0.7963972091674805, + "num_tokens": 735754936.0, + "step": 28145 + }, + { + "epoch": 2.3209367527005855, + "grad_norm": 1.6645054817199707, + "learning_rate": 1e-06, + "loss": 0.4866, + "mean_token_accuracy": 0.8327390551567078, + "num_tokens": 735793519.0, + "step": 28146 + }, + { + "epoch": 2.321019213325637, + "grad_norm": 2.0865931510925293, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7553290724754333, + "num_tokens": 735826282.0, + "step": 28147 + }, + { + "epoch": 2.3211016739506887, + "grad_norm": 2.5677638053894043, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.7899324893951416, + "num_tokens": 735848189.0, + "step": 28148 + }, + { + "epoch": 2.32118413457574, + "grad_norm": 2.258992910385132, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.7843843102455139, + "num_tokens": 735874188.0, + "step": 28149 + }, + { + "epoch": 2.321266595200792, + "grad_norm": 1.9526997804641724, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7838166952133179, + "num_tokens": 735906137.0, + "step": 28150 + }, + { + "epoch": 2.321349055825843, + "grad_norm": 2.1820228099823, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7726503014564514, + "num_tokens": 735936174.0, + "step": 28151 + }, + { + "epoch": 2.3214315164508945, + "grad_norm": 2.1034910678863525, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.7994310259819031, + "num_tokens": 735965189.0, + "step": 28152 + }, + { + "epoch": 2.3215139770759463, + "grad_norm": 2.3692593574523926, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.7790588140487671, + "num_tokens": 735991639.0, + "step": 28153 + }, + { + "epoch": 2.3215964377009977, + "grad_norm": 2.334872245788574, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.791077733039856, + "num_tokens": 736017477.0, + "step": 28154 + }, + { + "epoch": 2.3216788983260495, + "grad_norm": 2.1133289337158203, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.7889215350151062, + "num_tokens": 736046279.0, + "step": 28155 + }, + { + "epoch": 2.321761358951101, + "grad_norm": 2.1476364135742188, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7587858438491821, + "num_tokens": 736073609.0, + "step": 28156 + }, + { + "epoch": 2.321843819576152, + "grad_norm": 2.1032443046569824, + "learning_rate": 1e-06, + "loss": 0.599, + "mean_token_accuracy": 0.7914156913757324, + "num_tokens": 736101515.0, + "step": 28157 + }, + { + "epoch": 2.321926280201204, + "grad_norm": 2.321338176727295, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7532252073287964, + "num_tokens": 736126598.0, + "step": 28158 + }, + { + "epoch": 2.3220087408262553, + "grad_norm": 2.067789316177368, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7752554416656494, + "num_tokens": 736159052.0, + "step": 28159 + }, + { + "epoch": 2.322091201451307, + "grad_norm": 2.150367498397827, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7656776905059814, + "num_tokens": 736186916.0, + "step": 28160 + }, + { + "epoch": 2.3221736620763584, + "grad_norm": 1.89373779296875, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.7817097902297974, + "num_tokens": 736224644.0, + "step": 28161 + }, + { + "epoch": 2.3222561227014102, + "grad_norm": 2.0538880825042725, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.7731566429138184, + "num_tokens": 736255933.0, + "step": 28162 + }, + { + "epoch": 2.3223385833264616, + "grad_norm": 2.0339436531066895, + "learning_rate": 1e-06, + "loss": 0.6054, + "mean_token_accuracy": 0.8044673800468445, + "num_tokens": 736285408.0, + "step": 28163 + }, + { + "epoch": 2.322421043951513, + "grad_norm": 2.2132375240325928, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7399996519088745, + "num_tokens": 736314774.0, + "step": 28164 + }, + { + "epoch": 2.3225035045765647, + "grad_norm": 2.0720059871673584, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7750190496444702, + "num_tokens": 736345140.0, + "step": 28165 + }, + { + "epoch": 2.322585965201616, + "grad_norm": 2.3190410137176514, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.77678382396698, + "num_tokens": 736377188.0, + "step": 28166 + }, + { + "epoch": 2.322668425826668, + "grad_norm": 2.0378575325012207, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7377135753631592, + "num_tokens": 736408102.0, + "step": 28167 + }, + { + "epoch": 2.3227508864517192, + "grad_norm": 2.0615041255950928, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7654316425323486, + "num_tokens": 736436451.0, + "step": 28168 + }, + { + "epoch": 2.322833347076771, + "grad_norm": 2.2332944869995117, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.7857123613357544, + "num_tokens": 736463458.0, + "step": 28169 + }, + { + "epoch": 2.3229158077018224, + "grad_norm": 2.3929927349090576, + "learning_rate": 1e-06, + "loss": 0.5935, + "mean_token_accuracy": 0.7994929552078247, + "num_tokens": 736489724.0, + "step": 28170 + }, + { + "epoch": 2.3229982683268737, + "grad_norm": 1.9775151014328003, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7848798632621765, + "num_tokens": 736521871.0, + "step": 28171 + }, + { + "epoch": 2.3230807289519255, + "grad_norm": 2.1800413131713867, + "learning_rate": 1e-06, + "loss": 0.5218, + "mean_token_accuracy": 0.8120422959327698, + "num_tokens": 736548709.0, + "step": 28172 + }, + { + "epoch": 2.323163189576977, + "grad_norm": 1.8424586057662964, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7640204429626465, + "num_tokens": 736587120.0, + "step": 28173 + }, + { + "epoch": 2.3232456502020287, + "grad_norm": 2.055506467819214, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7651073336601257, + "num_tokens": 736621972.0, + "step": 28174 + }, + { + "epoch": 2.32332811082708, + "grad_norm": 2.2427117824554443, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7629622220993042, + "num_tokens": 736649222.0, + "step": 28175 + }, + { + "epoch": 2.323410571452132, + "grad_norm": 1.9883357286453247, + "learning_rate": 1e-06, + "loss": 0.5151, + "mean_token_accuracy": 0.8143081068992615, + "num_tokens": 736677654.0, + "step": 28176 + }, + { + "epoch": 2.323493032077183, + "grad_norm": 2.31781005859375, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7819699048995972, + "num_tokens": 736703783.0, + "step": 28177 + }, + { + "epoch": 2.3235754927022345, + "grad_norm": 2.036210536956787, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.7979272603988647, + "num_tokens": 736737116.0, + "step": 28178 + }, + { + "epoch": 2.3236579533272863, + "grad_norm": 2.252671957015991, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.712572455406189, + "num_tokens": 736766071.0, + "step": 28179 + }, + { + "epoch": 2.3237404139523377, + "grad_norm": 2.0777082443237305, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7842042446136475, + "num_tokens": 736796197.0, + "step": 28180 + }, + { + "epoch": 2.3238228745773895, + "grad_norm": 1.9101146459579468, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.7846305966377258, + "num_tokens": 736831323.0, + "step": 28181 + }, + { + "epoch": 2.323905335202441, + "grad_norm": 2.0165021419525146, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.794273853302002, + "num_tokens": 736867403.0, + "step": 28182 + }, + { + "epoch": 2.3239877958274926, + "grad_norm": 2.1402924060821533, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.749109148979187, + "num_tokens": 736897155.0, + "step": 28183 + }, + { + "epoch": 2.324070256452544, + "grad_norm": 1.7539958953857422, + "learning_rate": 1e-06, + "loss": 0.5872, + "mean_token_accuracy": 0.8109248876571655, + "num_tokens": 736934117.0, + "step": 28184 + }, + { + "epoch": 2.3241527170775953, + "grad_norm": 2.3304741382598877, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7222907543182373, + "num_tokens": 736961637.0, + "step": 28185 + }, + { + "epoch": 2.324235177702647, + "grad_norm": 1.9860374927520752, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7727469205856323, + "num_tokens": 736995021.0, + "step": 28186 + }, + { + "epoch": 2.3243176383276984, + "grad_norm": 1.7880122661590576, + "learning_rate": 1e-06, + "loss": 0.4964, + "mean_token_accuracy": 0.8325071334838867, + "num_tokens": 737033891.0, + "step": 28187 + }, + { + "epoch": 2.3244000989527502, + "grad_norm": 2.3087081909179688, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7774312496185303, + "num_tokens": 737058344.0, + "step": 28188 + }, + { + "epoch": 2.3244825595778016, + "grad_norm": 1.7790457010269165, + "learning_rate": 1e-06, + "loss": 0.5874, + "mean_token_accuracy": 0.8008217811584473, + "num_tokens": 737099176.0, + "step": 28189 + }, + { + "epoch": 2.3245650202028534, + "grad_norm": 2.305238723754883, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7487748861312866, + "num_tokens": 737127614.0, + "step": 28190 + }, + { + "epoch": 2.3246474808279047, + "grad_norm": 2.173758029937744, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7251923680305481, + "num_tokens": 737158421.0, + "step": 28191 + }, + { + "epoch": 2.324729941452956, + "grad_norm": 2.0369720458984375, + "learning_rate": 1e-06, + "loss": 0.7569, + "mean_token_accuracy": 0.7497960329055786, + "num_tokens": 737192942.0, + "step": 28192 + }, + { + "epoch": 2.324812402078008, + "grad_norm": 2.1630048751831055, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.7959340810775757, + "num_tokens": 737219567.0, + "step": 28193 + }, + { + "epoch": 2.3248948627030592, + "grad_norm": 1.841845989227295, + "learning_rate": 1e-06, + "loss": 0.5151, + "mean_token_accuracy": 0.8171064257621765, + "num_tokens": 737253348.0, + "step": 28194 + }, + { + "epoch": 2.3249773233281106, + "grad_norm": 2.187556743621826, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7725088596343994, + "num_tokens": 737282696.0, + "step": 28195 + }, + { + "epoch": 2.3250597839531624, + "grad_norm": 2.150944709777832, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7400444746017456, + "num_tokens": 737315135.0, + "step": 28196 + }, + { + "epoch": 2.3251422445782137, + "grad_norm": 2.0411934852600098, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7775521278381348, + "num_tokens": 737348911.0, + "step": 28197 + }, + { + "epoch": 2.3252247052032655, + "grad_norm": 2.069279432296753, + "learning_rate": 1e-06, + "loss": 0.6082, + "mean_token_accuracy": 0.7989562749862671, + "num_tokens": 737380154.0, + "step": 28198 + }, + { + "epoch": 2.325307165828317, + "grad_norm": 2.2258975505828857, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7599915862083435, + "num_tokens": 737406504.0, + "step": 28199 + }, + { + "epoch": 2.3253896264533687, + "grad_norm": 2.0226118564605713, + "learning_rate": 1e-06, + "loss": 0.784, + "mean_token_accuracy": 0.7536823749542236, + "num_tokens": 737439266.0, + "step": 28200 + }, + { + "epoch": 2.32547208707842, + "grad_norm": 1.9078549146652222, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7705261707305908, + "num_tokens": 737472367.0, + "step": 28201 + }, + { + "epoch": 2.3255545477034714, + "grad_norm": 2.263396739959717, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.7904748916625977, + "num_tokens": 737496542.0, + "step": 28202 + }, + { + "epoch": 2.325637008328523, + "grad_norm": 2.071308135986328, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.716835618019104, + "num_tokens": 737530289.0, + "step": 28203 + }, + { + "epoch": 2.3257194689535745, + "grad_norm": 2.111116886138916, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7735150456428528, + "num_tokens": 737558046.0, + "step": 28204 + }, + { + "epoch": 2.3258019295786263, + "grad_norm": 2.0721592903137207, + "learning_rate": 1e-06, + "loss": 0.705, + "mean_token_accuracy": 0.7733831405639648, + "num_tokens": 737590853.0, + "step": 28205 + }, + { + "epoch": 2.3258843902036777, + "grad_norm": 1.978528618812561, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7816663384437561, + "num_tokens": 737623941.0, + "step": 28206 + }, + { + "epoch": 2.3259668508287294, + "grad_norm": 2.160712718963623, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7650551199913025, + "num_tokens": 737654155.0, + "step": 28207 + }, + { + "epoch": 2.326049311453781, + "grad_norm": 2.4031167030334473, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7636866569519043, + "num_tokens": 737678360.0, + "step": 28208 + }, + { + "epoch": 2.326131772078832, + "grad_norm": 2.2183713912963867, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.7824573516845703, + "num_tokens": 737707151.0, + "step": 28209 + }, + { + "epoch": 2.326214232703884, + "grad_norm": 2.1579058170318604, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.7711824774742126, + "num_tokens": 737736410.0, + "step": 28210 + }, + { + "epoch": 2.3262966933289353, + "grad_norm": 2.1001434326171875, + "learning_rate": 1e-06, + "loss": 0.5461, + "mean_token_accuracy": 0.8192882537841797, + "num_tokens": 737763053.0, + "step": 28211 + }, + { + "epoch": 2.326379153953987, + "grad_norm": 2.1352221965789795, + "learning_rate": 1e-06, + "loss": 0.5835, + "mean_token_accuracy": 0.7999827861785889, + "num_tokens": 737792133.0, + "step": 28212 + }, + { + "epoch": 2.3264616145790384, + "grad_norm": 2.245023012161255, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.75669264793396, + "num_tokens": 737819132.0, + "step": 28213 + }, + { + "epoch": 2.3265440752040902, + "grad_norm": 1.947587251663208, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7737938761711121, + "num_tokens": 737851991.0, + "step": 28214 + }, + { + "epoch": 2.3266265358291416, + "grad_norm": 2.1442878246307373, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7645541429519653, + "num_tokens": 737882509.0, + "step": 28215 + }, + { + "epoch": 2.326708996454193, + "grad_norm": 1.8489447832107544, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7648857831954956, + "num_tokens": 737916486.0, + "step": 28216 + }, + { + "epoch": 2.3267914570792447, + "grad_norm": 1.8226639032363892, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7528473734855652, + "num_tokens": 737954646.0, + "step": 28217 + }, + { + "epoch": 2.326873917704296, + "grad_norm": 2.302687644958496, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7676332592964172, + "num_tokens": 737982196.0, + "step": 28218 + }, + { + "epoch": 2.326956378329348, + "grad_norm": 2.080807685852051, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7598443031311035, + "num_tokens": 738012272.0, + "step": 28219 + }, + { + "epoch": 2.3270388389543992, + "grad_norm": 2.231527805328369, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7673852443695068, + "num_tokens": 738038937.0, + "step": 28220 + }, + { + "epoch": 2.327121299579451, + "grad_norm": 1.8563463687896729, + "learning_rate": 1e-06, + "loss": 0.5847, + "mean_token_accuracy": 0.8075007200241089, + "num_tokens": 738073022.0, + "step": 28221 + }, + { + "epoch": 2.3272037602045024, + "grad_norm": 2.5581767559051514, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7562867403030396, + "num_tokens": 738095326.0, + "step": 28222 + }, + { + "epoch": 2.3272862208295537, + "grad_norm": 1.9824944734573364, + "learning_rate": 1e-06, + "loss": 0.5166, + "mean_token_accuracy": 0.8267142176628113, + "num_tokens": 738122557.0, + "step": 28223 + }, + { + "epoch": 2.3273686814546055, + "grad_norm": 2.158407211303711, + "learning_rate": 1e-06, + "loss": 0.5355, + "mean_token_accuracy": 0.807695746421814, + "num_tokens": 738150340.0, + "step": 28224 + }, + { + "epoch": 2.327451142079657, + "grad_norm": 2.01192045211792, + "learning_rate": 1e-06, + "loss": 0.6083, + "mean_token_accuracy": 0.8056979775428772, + "num_tokens": 738180733.0, + "step": 28225 + }, + { + "epoch": 2.3275336027047087, + "grad_norm": 1.9929965734481812, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7703076004981995, + "num_tokens": 738210837.0, + "step": 28226 + }, + { + "epoch": 2.32761606332976, + "grad_norm": 2.0978918075561523, + "learning_rate": 1e-06, + "loss": 0.6123, + "mean_token_accuracy": 0.8016601800918579, + "num_tokens": 738238578.0, + "step": 28227 + }, + { + "epoch": 2.327698523954812, + "grad_norm": 2.077871084213257, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.773442268371582, + "num_tokens": 738268839.0, + "step": 28228 + }, + { + "epoch": 2.327780984579863, + "grad_norm": 2.1429896354675293, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7806082367897034, + "num_tokens": 738298345.0, + "step": 28229 + }, + { + "epoch": 2.3278634452049145, + "grad_norm": 2.067272663116455, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7749723196029663, + "num_tokens": 738332558.0, + "step": 28230 + }, + { + "epoch": 2.3279459058299663, + "grad_norm": 2.356618881225586, + "learning_rate": 1e-06, + "loss": 0.5521, + "mean_token_accuracy": 0.8060879707336426, + "num_tokens": 738355389.0, + "step": 28231 + }, + { + "epoch": 2.3280283664550177, + "grad_norm": 1.9783509969711304, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.7945394515991211, + "num_tokens": 738389665.0, + "step": 28232 + }, + { + "epoch": 2.3281108270800694, + "grad_norm": 2.0620481967926025, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7606387138366699, + "num_tokens": 738418728.0, + "step": 28233 + }, + { + "epoch": 2.328193287705121, + "grad_norm": 2.0115959644317627, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7641783952713013, + "num_tokens": 738452946.0, + "step": 28234 + }, + { + "epoch": 2.3282757483301726, + "grad_norm": 2.0586843490600586, + "learning_rate": 1e-06, + "loss": 0.5525, + "mean_token_accuracy": 0.8121234178543091, + "num_tokens": 738481040.0, + "step": 28235 + }, + { + "epoch": 2.328358208955224, + "grad_norm": 2.1043143272399902, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7642468810081482, + "num_tokens": 738512603.0, + "step": 28236 + }, + { + "epoch": 2.3284406695802753, + "grad_norm": 2.2252039909362793, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7419873476028442, + "num_tokens": 738541804.0, + "step": 28237 + }, + { + "epoch": 2.328523130205327, + "grad_norm": 2.1212165355682373, + "learning_rate": 1e-06, + "loss": 0.6005, + "mean_token_accuracy": 0.7984297275543213, + "num_tokens": 738567227.0, + "step": 28238 + }, + { + "epoch": 2.3286055908303784, + "grad_norm": 2.201022148132324, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7585598826408386, + "num_tokens": 738596972.0, + "step": 28239 + }, + { + "epoch": 2.32868805145543, + "grad_norm": 2.186056613922119, + "learning_rate": 1e-06, + "loss": 0.6422, + "mean_token_accuracy": 0.7905237078666687, + "num_tokens": 738622440.0, + "step": 28240 + }, + { + "epoch": 2.3287705120804816, + "grad_norm": 2.1179237365722656, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7726687788963318, + "num_tokens": 738652588.0, + "step": 28241 + }, + { + "epoch": 2.328852972705533, + "grad_norm": 1.8697608709335327, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.8007284998893738, + "num_tokens": 738688200.0, + "step": 28242 + }, + { + "epoch": 2.3289354333305847, + "grad_norm": 1.8965612649917603, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.7887228727340698, + "num_tokens": 738727510.0, + "step": 28243 + }, + { + "epoch": 2.329017893955636, + "grad_norm": 2.323765277862549, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7320426106452942, + "num_tokens": 738755057.0, + "step": 28244 + }, + { + "epoch": 2.329100354580688, + "grad_norm": 2.0384011268615723, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7521946430206299, + "num_tokens": 738788861.0, + "step": 28245 + }, + { + "epoch": 2.329182815205739, + "grad_norm": 2.166391611099243, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7415158748626709, + "num_tokens": 738819113.0, + "step": 28246 + }, + { + "epoch": 2.3292652758307906, + "grad_norm": 2.250678777694702, + "learning_rate": 1e-06, + "loss": 0.76, + "mean_token_accuracy": 0.7566510438919067, + "num_tokens": 738846230.0, + "step": 28247 + }, + { + "epoch": 2.3293477364558424, + "grad_norm": 2.2623891830444336, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7761544585227966, + "num_tokens": 738875761.0, + "step": 28248 + }, + { + "epoch": 2.3294301970808937, + "grad_norm": 2.140425443649292, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7297073602676392, + "num_tokens": 738907843.0, + "step": 28249 + }, + { + "epoch": 2.3295126577059455, + "grad_norm": 1.7977535724639893, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.7988300919532776, + "num_tokens": 738945789.0, + "step": 28250 + }, + { + "epoch": 2.329595118330997, + "grad_norm": 1.7642501592636108, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7797845005989075, + "num_tokens": 738987050.0, + "step": 28251 + }, + { + "epoch": 2.3296775789560487, + "grad_norm": 1.8189841508865356, + "learning_rate": 1e-06, + "loss": 0.5489, + "mean_token_accuracy": 0.8124868273735046, + "num_tokens": 739020289.0, + "step": 28252 + }, + { + "epoch": 2.3297600395811, + "grad_norm": 2.3257501125335693, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7623975276947021, + "num_tokens": 739046939.0, + "step": 28253 + }, + { + "epoch": 2.3298425002061514, + "grad_norm": 1.9735138416290283, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7722146511077881, + "num_tokens": 739077984.0, + "step": 28254 + }, + { + "epoch": 2.329924960831203, + "grad_norm": 2.0290377140045166, + "learning_rate": 1e-06, + "loss": 0.5766, + "mean_token_accuracy": 0.8043990135192871, + "num_tokens": 739107062.0, + "step": 28255 + }, + { + "epoch": 2.3300074214562545, + "grad_norm": 2.143326759338379, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.7798330187797546, + "num_tokens": 739137171.0, + "step": 28256 + }, + { + "epoch": 2.3300898820813063, + "grad_norm": 1.862566590309143, + "learning_rate": 1e-06, + "loss": 0.5938, + "mean_token_accuracy": 0.8029404878616333, + "num_tokens": 739175487.0, + "step": 28257 + }, + { + "epoch": 2.3301723427063576, + "grad_norm": 2.0589888095855713, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7574002742767334, + "num_tokens": 739210301.0, + "step": 28258 + }, + { + "epoch": 2.3302548033314094, + "grad_norm": 2.165374755859375, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.780731201171875, + "num_tokens": 739238349.0, + "step": 28259 + }, + { + "epoch": 2.330337263956461, + "grad_norm": 2.1576101779937744, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7701743841171265, + "num_tokens": 739270399.0, + "step": 28260 + }, + { + "epoch": 2.330419724581512, + "grad_norm": 2.105725049972534, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7254534363746643, + "num_tokens": 739302086.0, + "step": 28261 + }, + { + "epoch": 2.330502185206564, + "grad_norm": 2.1021804809570312, + "learning_rate": 1e-06, + "loss": 0.6026, + "mean_token_accuracy": 0.8006094694137573, + "num_tokens": 739330663.0, + "step": 28262 + }, + { + "epoch": 2.3305846458316153, + "grad_norm": 1.9348950386047363, + "learning_rate": 1e-06, + "loss": 0.7063, + "mean_token_accuracy": 0.7737008929252625, + "num_tokens": 739363379.0, + "step": 28263 + }, + { + "epoch": 2.330667106456667, + "grad_norm": 2.266892433166504, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7220095992088318, + "num_tokens": 739393838.0, + "step": 28264 + }, + { + "epoch": 2.3307495670817184, + "grad_norm": 1.886651873588562, + "learning_rate": 1e-06, + "loss": 0.5913, + "mean_token_accuracy": 0.8085732460021973, + "num_tokens": 739425936.0, + "step": 28265 + }, + { + "epoch": 2.3308320277067702, + "grad_norm": 2.617396593093872, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7579176425933838, + "num_tokens": 739448451.0, + "step": 28266 + }, + { + "epoch": 2.3309144883318216, + "grad_norm": 2.2341527938842773, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7482033967971802, + "num_tokens": 739477072.0, + "step": 28267 + }, + { + "epoch": 2.330996948956873, + "grad_norm": 1.993098258972168, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7728550434112549, + "num_tokens": 739508717.0, + "step": 28268 + }, + { + "epoch": 2.3310794095819247, + "grad_norm": 2.1207714080810547, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7507948875427246, + "num_tokens": 739539615.0, + "step": 28269 + }, + { + "epoch": 2.331161870206976, + "grad_norm": 2.2314248085021973, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.751729428768158, + "num_tokens": 739566905.0, + "step": 28270 + }, + { + "epoch": 2.331244330832028, + "grad_norm": 2.141552686691284, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.7869561910629272, + "num_tokens": 739595650.0, + "step": 28271 + }, + { + "epoch": 2.331326791457079, + "grad_norm": 2.2213382720947266, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.7911701202392578, + "num_tokens": 739621661.0, + "step": 28272 + }, + { + "epoch": 2.331409252082131, + "grad_norm": 2.2398557662963867, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7588526010513306, + "num_tokens": 739648614.0, + "step": 28273 + }, + { + "epoch": 2.3314917127071824, + "grad_norm": 2.1013340950012207, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7493687868118286, + "num_tokens": 739678527.0, + "step": 28274 + }, + { + "epoch": 2.3315741733322337, + "grad_norm": 1.7952042818069458, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7388919591903687, + "num_tokens": 739718111.0, + "step": 28275 + }, + { + "epoch": 2.3316566339572855, + "grad_norm": 2.2305338382720947, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.7905077934265137, + "num_tokens": 739745525.0, + "step": 28276 + }, + { + "epoch": 2.331739094582337, + "grad_norm": 2.249873399734497, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7558275461196899, + "num_tokens": 739772000.0, + "step": 28277 + }, + { + "epoch": 2.3318215552073887, + "grad_norm": 2.186061382293701, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7490469217300415, + "num_tokens": 739800935.0, + "step": 28278 + }, + { + "epoch": 2.33190401583244, + "grad_norm": 2.114180326461792, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7593822479248047, + "num_tokens": 739830570.0, + "step": 28279 + }, + { + "epoch": 2.331986476457492, + "grad_norm": 2.5920135974884033, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7466918230056763, + "num_tokens": 739852939.0, + "step": 28280 + }, + { + "epoch": 2.332068937082543, + "grad_norm": 1.827927589416504, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7659974098205566, + "num_tokens": 739890778.0, + "step": 28281 + }, + { + "epoch": 2.3321513977075945, + "grad_norm": 2.196486473083496, + "learning_rate": 1e-06, + "loss": 0.5792, + "mean_token_accuracy": 0.8070489168167114, + "num_tokens": 739915664.0, + "step": 28282 + }, + { + "epoch": 2.3322338583326463, + "grad_norm": 1.8007084131240845, + "learning_rate": 1e-06, + "loss": 0.6705, + "mean_token_accuracy": 0.781896710395813, + "num_tokens": 739951203.0, + "step": 28283 + }, + { + "epoch": 2.3323163189576976, + "grad_norm": 2.171175003051758, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7766905426979065, + "num_tokens": 739978899.0, + "step": 28284 + }, + { + "epoch": 2.332398779582749, + "grad_norm": 1.8107352256774902, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7767685651779175, + "num_tokens": 740016242.0, + "step": 28285 + }, + { + "epoch": 2.332481240207801, + "grad_norm": 2.0064444541931152, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.790352463722229, + "num_tokens": 740047100.0, + "step": 28286 + }, + { + "epoch": 2.332563700832852, + "grad_norm": 1.9403120279312134, + "learning_rate": 1e-06, + "loss": 0.5919, + "mean_token_accuracy": 0.7963404059410095, + "num_tokens": 740078881.0, + "step": 28287 + }, + { + "epoch": 2.332646161457904, + "grad_norm": 2.100642204284668, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7589184641838074, + "num_tokens": 740109463.0, + "step": 28288 + }, + { + "epoch": 2.3327286220829553, + "grad_norm": 1.9841418266296387, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7693657875061035, + "num_tokens": 740145684.0, + "step": 28289 + }, + { + "epoch": 2.332811082708007, + "grad_norm": 2.313554048538208, + "learning_rate": 1e-06, + "loss": 0.617, + "mean_token_accuracy": 0.7882867455482483, + "num_tokens": 740172308.0, + "step": 28290 + }, + { + "epoch": 2.3328935433330584, + "grad_norm": 1.9532387256622314, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7724744081497192, + "num_tokens": 740205717.0, + "step": 28291 + }, + { + "epoch": 2.33297600395811, + "grad_norm": 2.194056510925293, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7744032740592957, + "num_tokens": 740234595.0, + "step": 28292 + }, + { + "epoch": 2.3330584645831616, + "grad_norm": 1.9488991498947144, + "learning_rate": 1e-06, + "loss": 0.5047, + "mean_token_accuracy": 0.8190656900405884, + "num_tokens": 740265223.0, + "step": 28293 + }, + { + "epoch": 2.333140925208213, + "grad_norm": 2.125857353210449, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7795507907867432, + "num_tokens": 740295653.0, + "step": 28294 + }, + { + "epoch": 2.3332233858332647, + "grad_norm": 2.0160093307495117, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7647650241851807, + "num_tokens": 740328867.0, + "step": 28295 + }, + { + "epoch": 2.333305846458316, + "grad_norm": 1.8991564512252808, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.7728356719017029, + "num_tokens": 740367148.0, + "step": 28296 + }, + { + "epoch": 2.333388307083368, + "grad_norm": 1.9795728921890259, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.7900394201278687, + "num_tokens": 740402911.0, + "step": 28297 + }, + { + "epoch": 2.333470767708419, + "grad_norm": 1.9386593103408813, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7608453631401062, + "num_tokens": 740436054.0, + "step": 28298 + }, + { + "epoch": 2.3335532283334706, + "grad_norm": 2.0219123363494873, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7738137245178223, + "num_tokens": 740466614.0, + "step": 28299 + }, + { + "epoch": 2.3336356889585224, + "grad_norm": 2.042905569076538, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7788590788841248, + "num_tokens": 740499920.0, + "step": 28300 + }, + { + "epoch": 2.3337181495835737, + "grad_norm": 2.21425461769104, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.707196831703186, + "num_tokens": 740527919.0, + "step": 28301 + }, + { + "epoch": 2.3338006102086255, + "grad_norm": 1.8969451189041138, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.75771164894104, + "num_tokens": 740565349.0, + "step": 28302 + }, + { + "epoch": 2.333883070833677, + "grad_norm": 2.3094656467437744, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7495591640472412, + "num_tokens": 740592824.0, + "step": 28303 + }, + { + "epoch": 2.3339655314587286, + "grad_norm": 2.150709867477417, + "learning_rate": 1e-06, + "loss": 0.5661, + "mean_token_accuracy": 0.798643946647644, + "num_tokens": 740620935.0, + "step": 28304 + }, + { + "epoch": 2.33404799208378, + "grad_norm": 2.212643623352051, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.781916618347168, + "num_tokens": 740650159.0, + "step": 28305 + }, + { + "epoch": 2.3341304527088313, + "grad_norm": 1.881205677986145, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7704651355743408, + "num_tokens": 740685604.0, + "step": 28306 + }, + { + "epoch": 2.334212913333883, + "grad_norm": 2.0450496673583984, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7601585388183594, + "num_tokens": 740717927.0, + "step": 28307 + }, + { + "epoch": 2.3342953739589345, + "grad_norm": 1.9962483644485474, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7618885040283203, + "num_tokens": 740749976.0, + "step": 28308 + }, + { + "epoch": 2.3343778345839863, + "grad_norm": 2.0499703884124756, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7377779483795166, + "num_tokens": 740781863.0, + "step": 28309 + }, + { + "epoch": 2.3344602952090376, + "grad_norm": 1.9900572299957275, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.786990761756897, + "num_tokens": 740814996.0, + "step": 28310 + }, + { + "epoch": 2.3345427558340894, + "grad_norm": 2.1688344478607178, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.784382700920105, + "num_tokens": 740843811.0, + "step": 28311 + }, + { + "epoch": 2.334625216459141, + "grad_norm": 2.252274513244629, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.7871876358985901, + "num_tokens": 740870274.0, + "step": 28312 + }, + { + "epoch": 2.334707677084192, + "grad_norm": 2.02630352973938, + "learning_rate": 1e-06, + "loss": 0.5936, + "mean_token_accuracy": 0.7966654300689697, + "num_tokens": 740902476.0, + "step": 28313 + }, + { + "epoch": 2.334790137709244, + "grad_norm": 2.209815740585327, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7714853286743164, + "num_tokens": 740930981.0, + "step": 28314 + }, + { + "epoch": 2.3348725983342953, + "grad_norm": 1.995950698852539, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.7872679233551025, + "num_tokens": 740962747.0, + "step": 28315 + }, + { + "epoch": 2.334955058959347, + "grad_norm": 2.1084144115448, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7438278198242188, + "num_tokens": 740994564.0, + "step": 28316 + }, + { + "epoch": 2.3350375195843984, + "grad_norm": 2.3081283569335938, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.748200535774231, + "num_tokens": 741021336.0, + "step": 28317 + }, + { + "epoch": 2.33511998020945, + "grad_norm": 2.197643518447876, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7734692096710205, + "num_tokens": 741049706.0, + "step": 28318 + }, + { + "epoch": 2.3352024408345016, + "grad_norm": 2.2315633296966553, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7679259181022644, + "num_tokens": 741077400.0, + "step": 28319 + }, + { + "epoch": 2.335284901459553, + "grad_norm": 2.189901351928711, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.8026742935180664, + "num_tokens": 741105264.0, + "step": 28320 + }, + { + "epoch": 2.3353673620846047, + "grad_norm": 2.1289684772491455, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7698341608047485, + "num_tokens": 741134227.0, + "step": 28321 + }, + { + "epoch": 2.335449822709656, + "grad_norm": 2.3359668254852295, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7779792547225952, + "num_tokens": 741161194.0, + "step": 28322 + }, + { + "epoch": 2.335532283334708, + "grad_norm": 3.913217544555664, + "learning_rate": 1e-06, + "loss": 0.5832, + "mean_token_accuracy": 0.8060017824172974, + "num_tokens": 741187071.0, + "step": 28323 + }, + { + "epoch": 2.335614743959759, + "grad_norm": 2.011898994445801, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.7827736735343933, + "num_tokens": 741218508.0, + "step": 28324 + }, + { + "epoch": 2.3356972045848106, + "grad_norm": 2.258300304412842, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.7859233617782593, + "num_tokens": 741243743.0, + "step": 28325 + }, + { + "epoch": 2.3357796652098624, + "grad_norm": 2.163689374923706, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7775186896324158, + "num_tokens": 741272347.0, + "step": 28326 + }, + { + "epoch": 2.3358621258349137, + "grad_norm": 2.2705466747283936, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7516615986824036, + "num_tokens": 741300218.0, + "step": 28327 + }, + { + "epoch": 2.3359445864599655, + "grad_norm": 2.1887028217315674, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7682738304138184, + "num_tokens": 741331868.0, + "step": 28328 + }, + { + "epoch": 2.336027047085017, + "grad_norm": 2.209955930709839, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7492406368255615, + "num_tokens": 741359526.0, + "step": 28329 + }, + { + "epoch": 2.336109507710068, + "grad_norm": 1.9017232656478882, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.8033300042152405, + "num_tokens": 741390818.0, + "step": 28330 + }, + { + "epoch": 2.33619196833512, + "grad_norm": 2.2868692874908447, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7612992525100708, + "num_tokens": 741416404.0, + "step": 28331 + }, + { + "epoch": 2.3362744289601713, + "grad_norm": 1.8474642038345337, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7678499817848206, + "num_tokens": 741451291.0, + "step": 28332 + }, + { + "epoch": 2.336356889585223, + "grad_norm": 1.779788613319397, + "learning_rate": 1e-06, + "loss": 0.4612, + "mean_token_accuracy": 0.8397818207740784, + "num_tokens": 741488004.0, + "step": 28333 + }, + { + "epoch": 2.3364393502102745, + "grad_norm": 2.139842987060547, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7413266897201538, + "num_tokens": 741517930.0, + "step": 28334 + }, + { + "epoch": 2.3365218108353263, + "grad_norm": 2.296221971511841, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7395706176757812, + "num_tokens": 741549363.0, + "step": 28335 + }, + { + "epoch": 2.3366042714603776, + "grad_norm": 2.0200581550598145, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7395222187042236, + "num_tokens": 741585449.0, + "step": 28336 + }, + { + "epoch": 2.336686732085429, + "grad_norm": 2.249072313308716, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7529927492141724, + "num_tokens": 741615591.0, + "step": 28337 + }, + { + "epoch": 2.336769192710481, + "grad_norm": 2.135164976119995, + "learning_rate": 1e-06, + "loss": 0.5817, + "mean_token_accuracy": 0.8044013381004333, + "num_tokens": 741647023.0, + "step": 28338 + }, + { + "epoch": 2.336851653335532, + "grad_norm": 2.1268842220306396, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.7949119806289673, + "num_tokens": 741676534.0, + "step": 28339 + }, + { + "epoch": 2.336934113960584, + "grad_norm": 2.1533989906311035, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7454128265380859, + "num_tokens": 741704480.0, + "step": 28340 + }, + { + "epoch": 2.3370165745856353, + "grad_norm": 2.1098403930664062, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7841197848320007, + "num_tokens": 741734520.0, + "step": 28341 + }, + { + "epoch": 2.337099035210687, + "grad_norm": 1.842236876487732, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.7677010297775269, + "num_tokens": 741770331.0, + "step": 28342 + }, + { + "epoch": 2.3371814958357384, + "grad_norm": 2.110651731491089, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7615104913711548, + "num_tokens": 741800251.0, + "step": 28343 + }, + { + "epoch": 2.3372639564607898, + "grad_norm": 1.878164529800415, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7674180269241333, + "num_tokens": 741837695.0, + "step": 28344 + }, + { + "epoch": 2.3373464170858416, + "grad_norm": 2.0795795917510986, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7929571866989136, + "num_tokens": 741868940.0, + "step": 28345 + }, + { + "epoch": 2.337428877710893, + "grad_norm": 2.0768203735351562, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7838000655174255, + "num_tokens": 741897273.0, + "step": 28346 + }, + { + "epoch": 2.3375113383359447, + "grad_norm": 1.998016595840454, + "learning_rate": 1e-06, + "loss": 0.4992, + "mean_token_accuracy": 0.8273294568061829, + "num_tokens": 741926827.0, + "step": 28347 + }, + { + "epoch": 2.337593798960996, + "grad_norm": 2.141439199447632, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7654941082000732, + "num_tokens": 741955382.0, + "step": 28348 + }, + { + "epoch": 2.337676259586048, + "grad_norm": 2.326582431793213, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.748488187789917, + "num_tokens": 741981308.0, + "step": 28349 + }, + { + "epoch": 2.337758720211099, + "grad_norm": 2.2388291358947754, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.7834346890449524, + "num_tokens": 742007254.0, + "step": 28350 + }, + { + "epoch": 2.3378411808361506, + "grad_norm": 2.262617588043213, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7865414619445801, + "num_tokens": 742032303.0, + "step": 28351 + }, + { + "epoch": 2.3379236414612024, + "grad_norm": 1.9627234935760498, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7631089687347412, + "num_tokens": 742065686.0, + "step": 28352 + }, + { + "epoch": 2.3380061020862537, + "grad_norm": 2.1174182891845703, + "learning_rate": 1e-06, + "loss": 0.766, + "mean_token_accuracy": 0.7565728425979614, + "num_tokens": 742095382.0, + "step": 28353 + }, + { + "epoch": 2.3380885627113055, + "grad_norm": 1.934091329574585, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.793063759803772, + "num_tokens": 742131490.0, + "step": 28354 + }, + { + "epoch": 2.338171023336357, + "grad_norm": 2.098210096359253, + "learning_rate": 1e-06, + "loss": 0.6033, + "mean_token_accuracy": 0.8004207611083984, + "num_tokens": 742159368.0, + "step": 28355 + }, + { + "epoch": 2.3382534839614086, + "grad_norm": 2.2519454956054688, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7386429309844971, + "num_tokens": 742186640.0, + "step": 28356 + }, + { + "epoch": 2.33833594458646, + "grad_norm": 2.264296770095825, + "learning_rate": 1e-06, + "loss": 0.6166, + "mean_token_accuracy": 0.7831388711929321, + "num_tokens": 742214580.0, + "step": 28357 + }, + { + "epoch": 2.3384184052115113, + "grad_norm": 2.0276989936828613, + "learning_rate": 1e-06, + "loss": 0.6173, + "mean_token_accuracy": 0.7966378927230835, + "num_tokens": 742245702.0, + "step": 28358 + }, + { + "epoch": 2.338500865836563, + "grad_norm": 2.0370819568634033, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7829120755195618, + "num_tokens": 742274703.0, + "step": 28359 + }, + { + "epoch": 2.3385833264616145, + "grad_norm": 2.0794692039489746, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.761843204498291, + "num_tokens": 742305552.0, + "step": 28360 + }, + { + "epoch": 2.3386657870866663, + "grad_norm": 2.111422300338745, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.7933662533760071, + "num_tokens": 742334954.0, + "step": 28361 + }, + { + "epoch": 2.3387482477117176, + "grad_norm": 2.3973419666290283, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7672450542449951, + "num_tokens": 742359812.0, + "step": 28362 + }, + { + "epoch": 2.3388307083367694, + "grad_norm": 2.5221619606018066, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7815179824829102, + "num_tokens": 742385957.0, + "step": 28363 + }, + { + "epoch": 2.3389131689618208, + "grad_norm": 2.2352116107940674, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7451308965682983, + "num_tokens": 742415207.0, + "step": 28364 + }, + { + "epoch": 2.338995629586872, + "grad_norm": 2.0524797439575195, + "learning_rate": 1e-06, + "loss": 0.5628, + "mean_token_accuracy": 0.8018907904624939, + "num_tokens": 742447082.0, + "step": 28365 + }, + { + "epoch": 2.339078090211924, + "grad_norm": 2.2717483043670654, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7533713579177856, + "num_tokens": 742473195.0, + "step": 28366 + }, + { + "epoch": 2.3391605508369753, + "grad_norm": 2.3243534564971924, + "learning_rate": 1e-06, + "loss": 0.6038, + "mean_token_accuracy": 0.801289975643158, + "num_tokens": 742500223.0, + "step": 28367 + }, + { + "epoch": 2.339243011462027, + "grad_norm": 2.132753610610962, + "learning_rate": 1e-06, + "loss": 0.5919, + "mean_token_accuracy": 0.7876870632171631, + "num_tokens": 742532776.0, + "step": 28368 + }, + { + "epoch": 2.3393254720870784, + "grad_norm": 2.2777645587921143, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7565724849700928, + "num_tokens": 742558807.0, + "step": 28369 + }, + { + "epoch": 2.3394079327121298, + "grad_norm": 2.1899354457855225, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7585641145706177, + "num_tokens": 742587720.0, + "step": 28370 + }, + { + "epoch": 2.3394903933371816, + "grad_norm": 2.4516899585723877, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7673213481903076, + "num_tokens": 742613349.0, + "step": 28371 + }, + { + "epoch": 2.339572853962233, + "grad_norm": 2.076434850692749, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7873814105987549, + "num_tokens": 742646872.0, + "step": 28372 + }, + { + "epoch": 2.3396553145872847, + "grad_norm": 2.1632156372070312, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7772462368011475, + "num_tokens": 742675669.0, + "step": 28373 + }, + { + "epoch": 2.339737775212336, + "grad_norm": 1.8822815418243408, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7769877910614014, + "num_tokens": 742708992.0, + "step": 28374 + }, + { + "epoch": 2.3398202358373874, + "grad_norm": 1.7775685787200928, + "learning_rate": 1e-06, + "loss": 0.5981, + "mean_token_accuracy": 0.7950899004936218, + "num_tokens": 742747591.0, + "step": 28375 + }, + { + "epoch": 2.339902696462439, + "grad_norm": 2.2195653915405273, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.7926507592201233, + "num_tokens": 742778021.0, + "step": 28376 + }, + { + "epoch": 2.3399851570874906, + "grad_norm": 2.34993052482605, + "learning_rate": 1e-06, + "loss": 0.5864, + "mean_token_accuracy": 0.7988438606262207, + "num_tokens": 742800416.0, + "step": 28377 + }, + { + "epoch": 2.3400676177125423, + "grad_norm": 2.1203060150146484, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7755742073059082, + "num_tokens": 742830325.0, + "step": 28378 + }, + { + "epoch": 2.3401500783375937, + "grad_norm": 1.9751416444778442, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.745711088180542, + "num_tokens": 742862164.0, + "step": 28379 + }, + { + "epoch": 2.3402325389626455, + "grad_norm": 1.9360597133636475, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7739289999008179, + "num_tokens": 742896199.0, + "step": 28380 + }, + { + "epoch": 2.340314999587697, + "grad_norm": 2.0363476276397705, + "learning_rate": 1e-06, + "loss": 0.5881, + "mean_token_accuracy": 0.7953900694847107, + "num_tokens": 742928420.0, + "step": 28381 + }, + { + "epoch": 2.340397460212748, + "grad_norm": 1.9423028230667114, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7698937654495239, + "num_tokens": 742964649.0, + "step": 28382 + }, + { + "epoch": 2.3404799208378, + "grad_norm": 2.2528605461120605, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.747159481048584, + "num_tokens": 742992068.0, + "step": 28383 + }, + { + "epoch": 2.3405623814628513, + "grad_norm": 1.9156981706619263, + "learning_rate": 1e-06, + "loss": 0.572, + "mean_token_accuracy": 0.7972689270973206, + "num_tokens": 743027842.0, + "step": 28384 + }, + { + "epoch": 2.340644842087903, + "grad_norm": 2.368144989013672, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7106049656867981, + "num_tokens": 743055291.0, + "step": 28385 + }, + { + "epoch": 2.3407273027129545, + "grad_norm": 2.0303969383239746, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7677180767059326, + "num_tokens": 743088255.0, + "step": 28386 + }, + { + "epoch": 2.3408097633380063, + "grad_norm": 1.859841227531433, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7799335718154907, + "num_tokens": 743126898.0, + "step": 28387 + }, + { + "epoch": 2.3408922239630576, + "grad_norm": 2.061922550201416, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.7844408750534058, + "num_tokens": 743158463.0, + "step": 28388 + }, + { + "epoch": 2.340974684588109, + "grad_norm": 2.2098398208618164, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7794126868247986, + "num_tokens": 743185806.0, + "step": 28389 + }, + { + "epoch": 2.3410571452131608, + "grad_norm": 2.215733289718628, + "learning_rate": 1e-06, + "loss": 0.6219, + "mean_token_accuracy": 0.7776557207107544, + "num_tokens": 743212021.0, + "step": 28390 + }, + { + "epoch": 2.341139605838212, + "grad_norm": 2.096017360687256, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7851299047470093, + "num_tokens": 743239817.0, + "step": 28391 + }, + { + "epoch": 2.341222066463264, + "grad_norm": 2.17573618888855, + "learning_rate": 1e-06, + "loss": 0.582, + "mean_token_accuracy": 0.8083393573760986, + "num_tokens": 743267047.0, + "step": 28392 + }, + { + "epoch": 2.3413045270883153, + "grad_norm": 2.115363359451294, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.788889467716217, + "num_tokens": 743297813.0, + "step": 28393 + }, + { + "epoch": 2.341386987713367, + "grad_norm": 2.2313144207000732, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.7821700572967529, + "num_tokens": 743325094.0, + "step": 28394 + }, + { + "epoch": 2.3414694483384184, + "grad_norm": 2.1525189876556396, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7418843507766724, + "num_tokens": 743356979.0, + "step": 28395 + }, + { + "epoch": 2.3415519089634698, + "grad_norm": 2.1047637462615967, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7774415612220764, + "num_tokens": 743385526.0, + "step": 28396 + }, + { + "epoch": 2.3416343695885216, + "grad_norm": 2.1626086235046387, + "learning_rate": 1e-06, + "loss": 0.6027, + "mean_token_accuracy": 0.8004565238952637, + "num_tokens": 743414885.0, + "step": 28397 + }, + { + "epoch": 2.341716830213573, + "grad_norm": 2.08321213722229, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7737261652946472, + "num_tokens": 743447625.0, + "step": 28398 + }, + { + "epoch": 2.3417992908386247, + "grad_norm": 2.1974329948425293, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.7834821939468384, + "num_tokens": 743474734.0, + "step": 28399 + }, + { + "epoch": 2.341881751463676, + "grad_norm": 1.8009260892868042, + "learning_rate": 1e-06, + "loss": 0.57, + "mean_token_accuracy": 0.8048297166824341, + "num_tokens": 743513103.0, + "step": 28400 + }, + { + "epoch": 2.341964212088728, + "grad_norm": 2.1040122509002686, + "learning_rate": 1e-06, + "loss": 0.6093, + "mean_token_accuracy": 0.7841095924377441, + "num_tokens": 743543621.0, + "step": 28401 + }, + { + "epoch": 2.342046672713779, + "grad_norm": 1.9329993724822998, + "learning_rate": 1e-06, + "loss": 0.5427, + "mean_token_accuracy": 0.8192620277404785, + "num_tokens": 743576870.0, + "step": 28402 + }, + { + "epoch": 2.3421291333388305, + "grad_norm": 2.153001546859741, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.8008196949958801, + "num_tokens": 743603037.0, + "step": 28403 + }, + { + "epoch": 2.3422115939638823, + "grad_norm": 2.047558069229126, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.783911406993866, + "num_tokens": 743636623.0, + "step": 28404 + }, + { + "epoch": 2.3422940545889337, + "grad_norm": 2.321232557296753, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7768692374229431, + "num_tokens": 743662469.0, + "step": 28405 + }, + { + "epoch": 2.3423765152139855, + "grad_norm": 2.0207018852233887, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7535806894302368, + "num_tokens": 743694137.0, + "step": 28406 + }, + { + "epoch": 2.342458975839037, + "grad_norm": 2.044492721557617, + "learning_rate": 1e-06, + "loss": 0.5291, + "mean_token_accuracy": 0.8194080591201782, + "num_tokens": 743725311.0, + "step": 28407 + }, + { + "epoch": 2.3425414364640886, + "grad_norm": 1.8577487468719482, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.7805605530738831, + "num_tokens": 743758143.0, + "step": 28408 + }, + { + "epoch": 2.34262389708914, + "grad_norm": 2.2874844074249268, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7729564905166626, + "num_tokens": 743782691.0, + "step": 28409 + }, + { + "epoch": 2.3427063577141913, + "grad_norm": 2.0449941158294678, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7604213953018188, + "num_tokens": 743814151.0, + "step": 28410 + }, + { + "epoch": 2.342788818339243, + "grad_norm": 1.797922968864441, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7773599624633789, + "num_tokens": 743851776.0, + "step": 28411 + }, + { + "epoch": 2.3428712789642945, + "grad_norm": 2.4252357482910156, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7466607093811035, + "num_tokens": 743876359.0, + "step": 28412 + }, + { + "epoch": 2.3429537395893463, + "grad_norm": 2.6001474857330322, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7639720439910889, + "num_tokens": 743896353.0, + "step": 28413 + }, + { + "epoch": 2.3430362002143976, + "grad_norm": 2.322727918624878, + "learning_rate": 1e-06, + "loss": 0.6012, + "mean_token_accuracy": 0.7987359762191772, + "num_tokens": 743919775.0, + "step": 28414 + }, + { + "epoch": 2.343118660839449, + "grad_norm": 1.8011975288391113, + "learning_rate": 1e-06, + "loss": 0.5343, + "mean_token_accuracy": 0.8083056211471558, + "num_tokens": 743955717.0, + "step": 28415 + }, + { + "epoch": 2.3432011214645008, + "grad_norm": 2.068439483642578, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.7854283452033997, + "num_tokens": 743983655.0, + "step": 28416 + }, + { + "epoch": 2.343283582089552, + "grad_norm": 2.260425090789795, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.7653014659881592, + "num_tokens": 744013593.0, + "step": 28417 + }, + { + "epoch": 2.343366042714604, + "grad_norm": 1.9802539348602295, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.7950964570045471, + "num_tokens": 744047948.0, + "step": 28418 + }, + { + "epoch": 2.3434485033396553, + "grad_norm": 2.1641035079956055, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7426352500915527, + "num_tokens": 744077710.0, + "step": 28419 + }, + { + "epoch": 2.3435309639647066, + "grad_norm": 2.1001884937286377, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7523829936981201, + "num_tokens": 744109939.0, + "step": 28420 + }, + { + "epoch": 2.3436134245897584, + "grad_norm": 2.0243210792541504, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7728935480117798, + "num_tokens": 744139728.0, + "step": 28421 + }, + { + "epoch": 2.3436958852148098, + "grad_norm": 2.1846840381622314, + "learning_rate": 1e-06, + "loss": 0.6166, + "mean_token_accuracy": 0.7908012866973877, + "num_tokens": 744165135.0, + "step": 28422 + }, + { + "epoch": 2.3437783458398616, + "grad_norm": 1.9601999521255493, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.7986195683479309, + "num_tokens": 744195786.0, + "step": 28423 + }, + { + "epoch": 2.343860806464913, + "grad_norm": 2.081061601638794, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.760947585105896, + "num_tokens": 744228224.0, + "step": 28424 + }, + { + "epoch": 2.3439432670899647, + "grad_norm": 2.1613237857818604, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7373157739639282, + "num_tokens": 744257200.0, + "step": 28425 + }, + { + "epoch": 2.344025727715016, + "grad_norm": 2.1307992935180664, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7793426513671875, + "num_tokens": 744285090.0, + "step": 28426 + }, + { + "epoch": 2.3441081883400674, + "grad_norm": 2.126286506652832, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7648451328277588, + "num_tokens": 744312452.0, + "step": 28427 + }, + { + "epoch": 2.344190648965119, + "grad_norm": 1.9829035997390747, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.795073390007019, + "num_tokens": 744344116.0, + "step": 28428 + }, + { + "epoch": 2.3442731095901705, + "grad_norm": 2.0841407775878906, + "learning_rate": 1e-06, + "loss": 0.7862, + "mean_token_accuracy": 0.7497330904006958, + "num_tokens": 744376975.0, + "step": 28429 + }, + { + "epoch": 2.3443555702152223, + "grad_norm": 1.9834458827972412, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7701206207275391, + "num_tokens": 744409184.0, + "step": 28430 + }, + { + "epoch": 2.3444380308402737, + "grad_norm": 1.937718152999878, + "learning_rate": 1e-06, + "loss": 0.5753, + "mean_token_accuracy": 0.803400456905365, + "num_tokens": 744441791.0, + "step": 28431 + }, + { + "epoch": 2.3445204914653255, + "grad_norm": 2.1355032920837402, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7386527061462402, + "num_tokens": 744473747.0, + "step": 28432 + }, + { + "epoch": 2.344602952090377, + "grad_norm": 2.2554616928100586, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7380220890045166, + "num_tokens": 744501281.0, + "step": 28433 + }, + { + "epoch": 2.344685412715428, + "grad_norm": 1.7354599237442017, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.7940307259559631, + "num_tokens": 744537983.0, + "step": 28434 + }, + { + "epoch": 2.34476787334048, + "grad_norm": 2.06847882270813, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.7943441867828369, + "num_tokens": 744567138.0, + "step": 28435 + }, + { + "epoch": 2.3448503339655313, + "grad_norm": 1.9418480396270752, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7845298051834106, + "num_tokens": 744601199.0, + "step": 28436 + }, + { + "epoch": 2.344932794590583, + "grad_norm": 1.7783633470535278, + "learning_rate": 1e-06, + "loss": 0.5704, + "mean_token_accuracy": 0.8052412271499634, + "num_tokens": 744638872.0, + "step": 28437 + }, + { + "epoch": 2.3450152552156345, + "grad_norm": 2.113175868988037, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7250766754150391, + "num_tokens": 744669053.0, + "step": 28438 + }, + { + "epoch": 2.3450977158406863, + "grad_norm": 2.210022211074829, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7678501009941101, + "num_tokens": 744694401.0, + "step": 28439 + }, + { + "epoch": 2.3451801764657376, + "grad_norm": 2.0783913135528564, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7428597807884216, + "num_tokens": 744727922.0, + "step": 28440 + }, + { + "epoch": 2.345262637090789, + "grad_norm": 2.0280511379241943, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7722313404083252, + "num_tokens": 744759108.0, + "step": 28441 + }, + { + "epoch": 2.3453450977158408, + "grad_norm": 2.324878454208374, + "learning_rate": 1e-06, + "loss": 0.5518, + "mean_token_accuracy": 0.8060493469238281, + "num_tokens": 744783040.0, + "step": 28442 + }, + { + "epoch": 2.345427558340892, + "grad_norm": 1.9803266525268555, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7722153067588806, + "num_tokens": 744813671.0, + "step": 28443 + }, + { + "epoch": 2.345510018965944, + "grad_norm": 2.2340340614318848, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7554478645324707, + "num_tokens": 744839688.0, + "step": 28444 + }, + { + "epoch": 2.3455924795909953, + "grad_norm": 2.0619866847991943, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7557511925697327, + "num_tokens": 744869195.0, + "step": 28445 + }, + { + "epoch": 2.345674940216047, + "grad_norm": 2.4815757274627686, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7473942041397095, + "num_tokens": 744890922.0, + "step": 28446 + }, + { + "epoch": 2.3457574008410984, + "grad_norm": 2.288276195526123, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7562481760978699, + "num_tokens": 744916990.0, + "step": 28447 + }, + { + "epoch": 2.3458398614661498, + "grad_norm": 1.7195148468017578, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7834063768386841, + "num_tokens": 744954042.0, + "step": 28448 + }, + { + "epoch": 2.3459223220912016, + "grad_norm": 1.8786816596984863, + "learning_rate": 1e-06, + "loss": 0.5183, + "mean_token_accuracy": 0.8196366429328918, + "num_tokens": 744986311.0, + "step": 28449 + }, + { + "epoch": 2.346004782716253, + "grad_norm": 2.1325507164001465, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7504416108131409, + "num_tokens": 745017418.0, + "step": 28450 + }, + { + "epoch": 2.3460872433413047, + "grad_norm": 1.9312744140625, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7762624621391296, + "num_tokens": 745052910.0, + "step": 28451 + }, + { + "epoch": 2.346169703966356, + "grad_norm": 2.2195076942443848, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7543028593063354, + "num_tokens": 745081568.0, + "step": 28452 + }, + { + "epoch": 2.346252164591408, + "grad_norm": 1.9689867496490479, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7238575220108032, + "num_tokens": 745114350.0, + "step": 28453 + }, + { + "epoch": 2.346334625216459, + "grad_norm": 2.4814443588256836, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7434762716293335, + "num_tokens": 745140010.0, + "step": 28454 + }, + { + "epoch": 2.3464170858415105, + "grad_norm": 2.135408878326416, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7703399062156677, + "num_tokens": 745171033.0, + "step": 28455 + }, + { + "epoch": 2.3464995464665623, + "grad_norm": 2.0418879985809326, + "learning_rate": 1e-06, + "loss": 0.5481, + "mean_token_accuracy": 0.8105315566062927, + "num_tokens": 745199198.0, + "step": 28456 + }, + { + "epoch": 2.3465820070916137, + "grad_norm": 2.102524995803833, + "learning_rate": 1e-06, + "loss": 0.5861, + "mean_token_accuracy": 0.8071527481079102, + "num_tokens": 745226164.0, + "step": 28457 + }, + { + "epoch": 2.3466644677166655, + "grad_norm": 2.0880417823791504, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7464404106140137, + "num_tokens": 745256259.0, + "step": 28458 + }, + { + "epoch": 2.346746928341717, + "grad_norm": 2.326809883117676, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.766308069229126, + "num_tokens": 745283918.0, + "step": 28459 + }, + { + "epoch": 2.346829388966768, + "grad_norm": 1.9062598943710327, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7944611310958862, + "num_tokens": 745319316.0, + "step": 28460 + }, + { + "epoch": 2.34691184959182, + "grad_norm": 1.979297399520874, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.7941864728927612, + "num_tokens": 745350608.0, + "step": 28461 + }, + { + "epoch": 2.3469943102168713, + "grad_norm": 2.4063878059387207, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7512719631195068, + "num_tokens": 745375993.0, + "step": 28462 + }, + { + "epoch": 2.347076770841923, + "grad_norm": 2.041034698486328, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7624261975288391, + "num_tokens": 745410665.0, + "step": 28463 + }, + { + "epoch": 2.3471592314669745, + "grad_norm": 2.1799609661102295, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7385292649269104, + "num_tokens": 745441064.0, + "step": 28464 + }, + { + "epoch": 2.347241692092026, + "grad_norm": 2.362647771835327, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7458051443099976, + "num_tokens": 745465561.0, + "step": 28465 + }, + { + "epoch": 2.3473241527170776, + "grad_norm": 2.272899866104126, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.7924576997756958, + "num_tokens": 745491601.0, + "step": 28466 + }, + { + "epoch": 2.347406613342129, + "grad_norm": 2.437807083129883, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.770209789276123, + "num_tokens": 745516198.0, + "step": 28467 + }, + { + "epoch": 2.3474890739671808, + "grad_norm": 2.5306525230407715, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.7794893980026245, + "num_tokens": 745539627.0, + "step": 28468 + }, + { + "epoch": 2.347571534592232, + "grad_norm": 1.9730901718139648, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7847174406051636, + "num_tokens": 745575121.0, + "step": 28469 + }, + { + "epoch": 2.347653995217284, + "grad_norm": 2.0901076793670654, + "learning_rate": 1e-06, + "loss": 0.734, + "mean_token_accuracy": 0.769116222858429, + "num_tokens": 745604386.0, + "step": 28470 + }, + { + "epoch": 2.3477364558423353, + "grad_norm": 2.345442295074463, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7283381223678589, + "num_tokens": 745628671.0, + "step": 28471 + }, + { + "epoch": 2.3478189164673866, + "grad_norm": 2.2350611686706543, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7897025942802429, + "num_tokens": 745654650.0, + "step": 28472 + }, + { + "epoch": 2.3479013770924384, + "grad_norm": 2.0370571613311768, + "learning_rate": 1e-06, + "loss": 0.6889, + "mean_token_accuracy": 0.7819730639457703, + "num_tokens": 745687420.0, + "step": 28473 + }, + { + "epoch": 2.3479838377174898, + "grad_norm": 2.0834505558013916, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7573230266571045, + "num_tokens": 745717140.0, + "step": 28474 + }, + { + "epoch": 2.3480662983425415, + "grad_norm": 2.005812883377075, + "learning_rate": 1e-06, + "loss": 0.5715, + "mean_token_accuracy": 0.8096271753311157, + "num_tokens": 745748009.0, + "step": 28475 + }, + { + "epoch": 2.348148758967593, + "grad_norm": 2.2161266803741455, + "learning_rate": 1e-06, + "loss": 0.7598, + "mean_token_accuracy": 0.74920654296875, + "num_tokens": 745774614.0, + "step": 28476 + }, + { + "epoch": 2.3482312195926447, + "grad_norm": 2.269261360168457, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7786351442337036, + "num_tokens": 745807466.0, + "step": 28477 + }, + { + "epoch": 2.348313680217696, + "grad_norm": 2.3605587482452393, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7620983719825745, + "num_tokens": 745833854.0, + "step": 28478 + }, + { + "epoch": 2.3483961408427474, + "grad_norm": 2.260984420776367, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.73577880859375, + "num_tokens": 745860537.0, + "step": 28479 + }, + { + "epoch": 2.348478601467799, + "grad_norm": 2.289440155029297, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.742925226688385, + "num_tokens": 745886948.0, + "step": 28480 + }, + { + "epoch": 2.3485610620928505, + "grad_norm": 1.6905505657196045, + "learning_rate": 1e-06, + "loss": 0.5826, + "mean_token_accuracy": 0.7964598536491394, + "num_tokens": 745928582.0, + "step": 28481 + }, + { + "epoch": 2.3486435227179023, + "grad_norm": 2.147705316543579, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.7898569703102112, + "num_tokens": 745958218.0, + "step": 28482 + }, + { + "epoch": 2.3487259833429537, + "grad_norm": 1.8885068893432617, + "learning_rate": 1e-06, + "loss": 0.5755, + "mean_token_accuracy": 0.8147687911987305, + "num_tokens": 745992198.0, + "step": 28483 + }, + { + "epoch": 2.3488084439680055, + "grad_norm": 1.942197322845459, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7540280222892761, + "num_tokens": 746031496.0, + "step": 28484 + }, + { + "epoch": 2.348890904593057, + "grad_norm": 2.1923575401306152, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.772964358329773, + "num_tokens": 746059413.0, + "step": 28485 + }, + { + "epoch": 2.348973365218108, + "grad_norm": 2.058485507965088, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7878705859184265, + "num_tokens": 746090262.0, + "step": 28486 + }, + { + "epoch": 2.34905582584316, + "grad_norm": 2.1403818130493164, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7726514339447021, + "num_tokens": 746118770.0, + "step": 28487 + }, + { + "epoch": 2.3491382864682113, + "grad_norm": 2.014753818511963, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7648472785949707, + "num_tokens": 746150073.0, + "step": 28488 + }, + { + "epoch": 2.349220747093263, + "grad_norm": 2.3651201725006104, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7676789164543152, + "num_tokens": 746174626.0, + "step": 28489 + }, + { + "epoch": 2.3493032077183145, + "grad_norm": 2.048196315765381, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.7869679927825928, + "num_tokens": 746204762.0, + "step": 28490 + }, + { + "epoch": 2.3493856683433663, + "grad_norm": 2.2174630165100098, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.772088348865509, + "num_tokens": 746230813.0, + "step": 28491 + }, + { + "epoch": 2.3494681289684176, + "grad_norm": 2.069725275039673, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7739512324333191, + "num_tokens": 746260726.0, + "step": 28492 + }, + { + "epoch": 2.349550589593469, + "grad_norm": 1.910992980003357, + "learning_rate": 1e-06, + "loss": 0.6053, + "mean_token_accuracy": 0.7978416085243225, + "num_tokens": 746294632.0, + "step": 28493 + }, + { + "epoch": 2.3496330502185208, + "grad_norm": 1.8181002140045166, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.7723609209060669, + "num_tokens": 746331334.0, + "step": 28494 + }, + { + "epoch": 2.349715510843572, + "grad_norm": 1.9602715969085693, + "learning_rate": 1e-06, + "loss": 0.4953, + "mean_token_accuracy": 0.829677939414978, + "num_tokens": 746363732.0, + "step": 28495 + }, + { + "epoch": 2.349797971468624, + "grad_norm": 2.0707221031188965, + "learning_rate": 1e-06, + "loss": 0.582, + "mean_token_accuracy": 0.8065922856330872, + "num_tokens": 746393571.0, + "step": 28496 + }, + { + "epoch": 2.3498804320936753, + "grad_norm": 1.9022150039672852, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7795587778091431, + "num_tokens": 746430796.0, + "step": 28497 + }, + { + "epoch": 2.349962892718727, + "grad_norm": 1.9265378713607788, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.783541738986969, + "num_tokens": 746463688.0, + "step": 28498 + }, + { + "epoch": 2.3500453533437784, + "grad_norm": 2.0624003410339355, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7701565027236938, + "num_tokens": 746494140.0, + "step": 28499 + }, + { + "epoch": 2.3501278139688297, + "grad_norm": 1.9976800680160522, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7678273320198059, + "num_tokens": 746527945.0, + "step": 28500 + }, + { + "epoch": 2.3502102745938815, + "grad_norm": 2.1975743770599365, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7605977058410645, + "num_tokens": 746556644.0, + "step": 28501 + }, + { + "epoch": 2.350292735218933, + "grad_norm": 2.6486499309539795, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7386679649353027, + "num_tokens": 746577947.0, + "step": 28502 + }, + { + "epoch": 2.3503751958439847, + "grad_norm": 2.1435647010803223, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7347713112831116, + "num_tokens": 746611441.0, + "step": 28503 + }, + { + "epoch": 2.350457656469036, + "grad_norm": 2.0520389080047607, + "learning_rate": 1e-06, + "loss": 0.5202, + "mean_token_accuracy": 0.8243526220321655, + "num_tokens": 746640074.0, + "step": 28504 + }, + { + "epoch": 2.3505401170940874, + "grad_norm": 2.1690595149993896, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7582100033760071, + "num_tokens": 746667264.0, + "step": 28505 + }, + { + "epoch": 2.350622577719139, + "grad_norm": 2.028205156326294, + "learning_rate": 1e-06, + "loss": 0.5552, + "mean_token_accuracy": 0.8167970180511475, + "num_tokens": 746698246.0, + "step": 28506 + }, + { + "epoch": 2.3507050383441905, + "grad_norm": 2.104417085647583, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.7880235910415649, + "num_tokens": 746727239.0, + "step": 28507 + }, + { + "epoch": 2.3507874989692423, + "grad_norm": 1.8024024963378906, + "learning_rate": 1e-06, + "loss": 0.6088, + "mean_token_accuracy": 0.8033442497253418, + "num_tokens": 746763329.0, + "step": 28508 + }, + { + "epoch": 2.3508699595942937, + "grad_norm": 2.008953094482422, + "learning_rate": 1e-06, + "loss": 0.6038, + "mean_token_accuracy": 0.8038094639778137, + "num_tokens": 746792802.0, + "step": 28509 + }, + { + "epoch": 2.350952420219345, + "grad_norm": 2.3118438720703125, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.7731568813323975, + "num_tokens": 746819654.0, + "step": 28510 + }, + { + "epoch": 2.351034880844397, + "grad_norm": 2.31904673576355, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7805837988853455, + "num_tokens": 746846946.0, + "step": 28511 + }, + { + "epoch": 2.351117341469448, + "grad_norm": 2.3030343055725098, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.7982826232910156, + "num_tokens": 746872757.0, + "step": 28512 + }, + { + "epoch": 2.3511998020945, + "grad_norm": 2.027034044265747, + "learning_rate": 1e-06, + "loss": 0.6046, + "mean_token_accuracy": 0.8029974699020386, + "num_tokens": 746903639.0, + "step": 28513 + }, + { + "epoch": 2.3512822627195513, + "grad_norm": 1.9545142650604248, + "learning_rate": 1e-06, + "loss": 0.6051, + "mean_token_accuracy": 0.7971795797348022, + "num_tokens": 746935816.0, + "step": 28514 + }, + { + "epoch": 2.351364723344603, + "grad_norm": 2.0607991218566895, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7388757467269897, + "num_tokens": 746969067.0, + "step": 28515 + }, + { + "epoch": 2.3514471839696545, + "grad_norm": 2.1388041973114014, + "learning_rate": 1e-06, + "loss": 0.6048, + "mean_token_accuracy": 0.7933164238929749, + "num_tokens": 746995643.0, + "step": 28516 + }, + { + "epoch": 2.351529644594706, + "grad_norm": 2.247636318206787, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7667843103408813, + "num_tokens": 747022545.0, + "step": 28517 + }, + { + "epoch": 2.3516121052197576, + "grad_norm": 2.1856088638305664, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.7930450439453125, + "num_tokens": 747047911.0, + "step": 28518 + }, + { + "epoch": 2.351694565844809, + "grad_norm": 2.092392683029175, + "learning_rate": 1e-06, + "loss": 0.7113, + "mean_token_accuracy": 0.7645443081855774, + "num_tokens": 747080205.0, + "step": 28519 + }, + { + "epoch": 2.3517770264698608, + "grad_norm": 2.086982488632202, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.7742185592651367, + "num_tokens": 747109027.0, + "step": 28520 + }, + { + "epoch": 2.351859487094912, + "grad_norm": 2.247818946838379, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7497007846832275, + "num_tokens": 747135823.0, + "step": 28521 + }, + { + "epoch": 2.351941947719964, + "grad_norm": 2.2036826610565186, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7538498640060425, + "num_tokens": 747166808.0, + "step": 28522 + }, + { + "epoch": 2.3520244083450152, + "grad_norm": 1.9606910943984985, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7578047513961792, + "num_tokens": 747199059.0, + "step": 28523 + }, + { + "epoch": 2.3521068689700666, + "grad_norm": 2.088142156600952, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.7776256799697876, + "num_tokens": 747230049.0, + "step": 28524 + }, + { + "epoch": 2.3521893295951184, + "grad_norm": 2.130396604537964, + "learning_rate": 1e-06, + "loss": 0.6076, + "mean_token_accuracy": 0.8018514513969421, + "num_tokens": 747258640.0, + "step": 28525 + }, + { + "epoch": 2.3522717902201697, + "grad_norm": 2.5853211879730225, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7498633861541748, + "num_tokens": 747282745.0, + "step": 28526 + }, + { + "epoch": 2.3523542508452215, + "grad_norm": 2.0297350883483887, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7706572413444519, + "num_tokens": 747318944.0, + "step": 28527 + }, + { + "epoch": 2.352436711470273, + "grad_norm": 2.142078161239624, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7338923215866089, + "num_tokens": 747349864.0, + "step": 28528 + }, + { + "epoch": 2.3525191720953247, + "grad_norm": 2.0409364700317383, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7472396492958069, + "num_tokens": 747381334.0, + "step": 28529 + }, + { + "epoch": 2.352601632720376, + "grad_norm": 2.2318880558013916, + "learning_rate": 1e-06, + "loss": 0.8128, + "mean_token_accuracy": 0.7487800121307373, + "num_tokens": 747410129.0, + "step": 28530 + }, + { + "epoch": 2.3526840933454274, + "grad_norm": 2.2929775714874268, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7588198781013489, + "num_tokens": 747436764.0, + "step": 28531 + }, + { + "epoch": 2.352766553970479, + "grad_norm": 2.1364927291870117, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.8042598962783813, + "num_tokens": 747463467.0, + "step": 28532 + }, + { + "epoch": 2.3528490145955305, + "grad_norm": 1.825003743171692, + "learning_rate": 1e-06, + "loss": 0.5947, + "mean_token_accuracy": 0.79267418384552, + "num_tokens": 747498118.0, + "step": 28533 + }, + { + "epoch": 2.3529314752205823, + "grad_norm": 2.0145018100738525, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.7843595743179321, + "num_tokens": 747530327.0, + "step": 28534 + }, + { + "epoch": 2.3530139358456337, + "grad_norm": 2.165858030319214, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7565512657165527, + "num_tokens": 747562343.0, + "step": 28535 + }, + { + "epoch": 2.3530963964706855, + "grad_norm": 2.118508815765381, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7448539733886719, + "num_tokens": 747592605.0, + "step": 28536 + }, + { + "epoch": 2.353178857095737, + "grad_norm": 2.1825318336486816, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7635970115661621, + "num_tokens": 747620308.0, + "step": 28537 + }, + { + "epoch": 2.353261317720788, + "grad_norm": 1.970287561416626, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7617676854133606, + "num_tokens": 747653854.0, + "step": 28538 + }, + { + "epoch": 2.35334377834584, + "grad_norm": 2.512434720993042, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7712802886962891, + "num_tokens": 747678819.0, + "step": 28539 + }, + { + "epoch": 2.3534262389708913, + "grad_norm": 2.263312339782715, + "learning_rate": 1e-06, + "loss": 0.7858, + "mean_token_accuracy": 0.7510538101196289, + "num_tokens": 747706148.0, + "step": 28540 + }, + { + "epoch": 2.353508699595943, + "grad_norm": 2.3760182857513428, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7410903573036194, + "num_tokens": 747731765.0, + "step": 28541 + }, + { + "epoch": 2.3535911602209945, + "grad_norm": 2.203097105026245, + "learning_rate": 1e-06, + "loss": 0.5414, + "mean_token_accuracy": 0.8096150159835815, + "num_tokens": 747756389.0, + "step": 28542 + }, + { + "epoch": 2.3536736208460463, + "grad_norm": 2.286921977996826, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7747732400894165, + "num_tokens": 747779841.0, + "step": 28543 + }, + { + "epoch": 2.3537560814710976, + "grad_norm": 2.1603381633758545, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7552177906036377, + "num_tokens": 747807548.0, + "step": 28544 + }, + { + "epoch": 2.353838542096149, + "grad_norm": 1.9515193700790405, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.7880408763885498, + "num_tokens": 747842997.0, + "step": 28545 + }, + { + "epoch": 2.3539210027212008, + "grad_norm": 2.135777235031128, + "learning_rate": 1e-06, + "loss": 0.588, + "mean_token_accuracy": 0.8008623123168945, + "num_tokens": 747874276.0, + "step": 28546 + }, + { + "epoch": 2.354003463346252, + "grad_norm": 1.81792151927948, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7738236784934998, + "num_tokens": 747912937.0, + "step": 28547 + }, + { + "epoch": 2.354085923971304, + "grad_norm": 2.0034945011138916, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7665316462516785, + "num_tokens": 747946683.0, + "step": 28548 + }, + { + "epoch": 2.3541683845963552, + "grad_norm": 1.8970893621444702, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.7902355194091797, + "num_tokens": 747980395.0, + "step": 28549 + }, + { + "epoch": 2.3542508452214066, + "grad_norm": 1.9196497201919556, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7547251582145691, + "num_tokens": 748015979.0, + "step": 28550 + }, + { + "epoch": 2.3543333058464584, + "grad_norm": 2.194632053375244, + "learning_rate": 1e-06, + "loss": 0.5628, + "mean_token_accuracy": 0.8104411363601685, + "num_tokens": 748040661.0, + "step": 28551 + }, + { + "epoch": 2.3544157664715097, + "grad_norm": 2.6046669483184814, + "learning_rate": 1e-06, + "loss": 0.6046, + "mean_token_accuracy": 0.7972869873046875, + "num_tokens": 748059862.0, + "step": 28552 + }, + { + "epoch": 2.3544982270965615, + "grad_norm": 1.7831406593322754, + "learning_rate": 1e-06, + "loss": 0.5976, + "mean_token_accuracy": 0.8023761510848999, + "num_tokens": 748098228.0, + "step": 28553 + }, + { + "epoch": 2.354580687721613, + "grad_norm": 2.3853321075439453, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7603319883346558, + "num_tokens": 748125813.0, + "step": 28554 + }, + { + "epoch": 2.3546631483466642, + "grad_norm": 2.2330799102783203, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7377846240997314, + "num_tokens": 748153985.0, + "step": 28555 + }, + { + "epoch": 2.354745608971716, + "grad_norm": 2.060608148574829, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7821385860443115, + "num_tokens": 748184077.0, + "step": 28556 + }, + { + "epoch": 2.3548280695967674, + "grad_norm": 2.238858938217163, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.7801876664161682, + "num_tokens": 748211098.0, + "step": 28557 + }, + { + "epoch": 2.354910530221819, + "grad_norm": 2.0918381214141846, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7757121324539185, + "num_tokens": 748242064.0, + "step": 28558 + }, + { + "epoch": 2.3549929908468705, + "grad_norm": 2.059542417526245, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.771642804145813, + "num_tokens": 748272578.0, + "step": 28559 + }, + { + "epoch": 2.3550754514719223, + "grad_norm": 2.028874635696411, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7449820041656494, + "num_tokens": 748305603.0, + "step": 28560 + }, + { + "epoch": 2.3551579120969737, + "grad_norm": 1.9792141914367676, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.7863141298294067, + "num_tokens": 748336919.0, + "step": 28561 + }, + { + "epoch": 2.355240372722025, + "grad_norm": 2.3067352771759033, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.7922344207763672, + "num_tokens": 748362876.0, + "step": 28562 + }, + { + "epoch": 2.355322833347077, + "grad_norm": 1.8762128353118896, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7825162410736084, + "num_tokens": 748397031.0, + "step": 28563 + }, + { + "epoch": 2.355405293972128, + "grad_norm": 2.2589495182037354, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7489405870437622, + "num_tokens": 748423522.0, + "step": 28564 + }, + { + "epoch": 2.35548775459718, + "grad_norm": 2.0461349487304688, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.7828742265701294, + "num_tokens": 748451132.0, + "step": 28565 + }, + { + "epoch": 2.3555702152222313, + "grad_norm": 2.250915765762329, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7445169687271118, + "num_tokens": 748481148.0, + "step": 28566 + }, + { + "epoch": 2.355652675847283, + "grad_norm": 2.274284601211548, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7781497240066528, + "num_tokens": 748507352.0, + "step": 28567 + }, + { + "epoch": 2.3557351364723345, + "grad_norm": 2.05173921585083, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7563824653625488, + "num_tokens": 748539192.0, + "step": 28568 + }, + { + "epoch": 2.355817597097386, + "grad_norm": 2.240931987762451, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.7776731252670288, + "num_tokens": 748566772.0, + "step": 28569 + }, + { + "epoch": 2.3559000577224376, + "grad_norm": 2.3376166820526123, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7775774002075195, + "num_tokens": 748593479.0, + "step": 28570 + }, + { + "epoch": 2.355982518347489, + "grad_norm": 2.3177056312561035, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7657626271247864, + "num_tokens": 748620096.0, + "step": 28571 + }, + { + "epoch": 2.3560649789725407, + "grad_norm": 2.2382209300994873, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.792090654373169, + "num_tokens": 748646203.0, + "step": 28572 + }, + { + "epoch": 2.356147439597592, + "grad_norm": 2.0266313552856445, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7791867256164551, + "num_tokens": 748679537.0, + "step": 28573 + }, + { + "epoch": 2.356229900222644, + "grad_norm": 2.1840572357177734, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7587267160415649, + "num_tokens": 748708211.0, + "step": 28574 + }, + { + "epoch": 2.3563123608476952, + "grad_norm": 2.0992181301116943, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.7834424376487732, + "num_tokens": 748735552.0, + "step": 28575 + }, + { + "epoch": 2.3563948214727466, + "grad_norm": 2.433826446533203, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.750343382358551, + "num_tokens": 748761244.0, + "step": 28576 + }, + { + "epoch": 2.3564772820977984, + "grad_norm": 2.152862310409546, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7742879390716553, + "num_tokens": 748794939.0, + "step": 28577 + }, + { + "epoch": 2.3565597427228497, + "grad_norm": 1.9008649587631226, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7741935849189758, + "num_tokens": 748830226.0, + "step": 28578 + }, + { + "epoch": 2.3566422033479015, + "grad_norm": 2.0368847846984863, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7782674431800842, + "num_tokens": 748864862.0, + "step": 28579 + }, + { + "epoch": 2.356724663972953, + "grad_norm": 1.9614061117172241, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7526264190673828, + "num_tokens": 748900901.0, + "step": 28580 + }, + { + "epoch": 2.3568071245980047, + "grad_norm": 1.989632487297058, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.7907108068466187, + "num_tokens": 748933148.0, + "step": 28581 + }, + { + "epoch": 2.356889585223056, + "grad_norm": 1.9854921102523804, + "learning_rate": 1e-06, + "loss": 0.5355, + "mean_token_accuracy": 0.8222593665122986, + "num_tokens": 748963745.0, + "step": 28582 + }, + { + "epoch": 2.3569720458481074, + "grad_norm": 2.3240745067596436, + "learning_rate": 1e-06, + "loss": 0.719, + "mean_token_accuracy": 0.7755601406097412, + "num_tokens": 748990435.0, + "step": 28583 + }, + { + "epoch": 2.357054506473159, + "grad_norm": 2.1675760746002197, + "learning_rate": 1e-06, + "loss": 0.4975, + "mean_token_accuracy": 0.8250589370727539, + "num_tokens": 749016405.0, + "step": 28584 + }, + { + "epoch": 2.3571369670982105, + "grad_norm": 2.046072483062744, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7560594081878662, + "num_tokens": 749049124.0, + "step": 28585 + }, + { + "epoch": 2.3572194277232623, + "grad_norm": 2.1606805324554443, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7368355989456177, + "num_tokens": 749078099.0, + "step": 28586 + }, + { + "epoch": 2.3573018883483137, + "grad_norm": 2.0878090858459473, + "learning_rate": 1e-06, + "loss": 0.5749, + "mean_token_accuracy": 0.8007892370223999, + "num_tokens": 749108745.0, + "step": 28587 + }, + { + "epoch": 2.3573843489733655, + "grad_norm": 1.8807474374771118, + "learning_rate": 1e-06, + "loss": 0.5066, + "mean_token_accuracy": 0.8170788288116455, + "num_tokens": 749140577.0, + "step": 28588 + }, + { + "epoch": 2.357466809598417, + "grad_norm": 1.9713438749313354, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7456114292144775, + "num_tokens": 749173388.0, + "step": 28589 + }, + { + "epoch": 2.357549270223468, + "grad_norm": 2.4797918796539307, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.763028621673584, + "num_tokens": 749195820.0, + "step": 28590 + }, + { + "epoch": 2.35763173084852, + "grad_norm": 2.178138494491577, + "learning_rate": 1e-06, + "loss": 0.5675, + "mean_token_accuracy": 0.8179320096969604, + "num_tokens": 749223602.0, + "step": 28591 + }, + { + "epoch": 2.3577141914735713, + "grad_norm": 1.8472169637680054, + "learning_rate": 1e-06, + "loss": 0.5973, + "mean_token_accuracy": 0.7989361882209778, + "num_tokens": 749261272.0, + "step": 28592 + }, + { + "epoch": 2.3577966520986227, + "grad_norm": 1.9902784824371338, + "learning_rate": 1e-06, + "loss": 0.597, + "mean_token_accuracy": 0.807640552520752, + "num_tokens": 749292637.0, + "step": 28593 + }, + { + "epoch": 2.3578791127236745, + "grad_norm": 2.120728015899658, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7539573311805725, + "num_tokens": 749323265.0, + "step": 28594 + }, + { + "epoch": 2.357961573348726, + "grad_norm": 2.091296434402466, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.736652135848999, + "num_tokens": 749354645.0, + "step": 28595 + }, + { + "epoch": 2.3580440339737776, + "grad_norm": 2.0936663150787354, + "learning_rate": 1e-06, + "loss": 0.5936, + "mean_token_accuracy": 0.7980813980102539, + "num_tokens": 749384874.0, + "step": 28596 + }, + { + "epoch": 2.358126494598829, + "grad_norm": 2.276646852493286, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.7509785890579224, + "num_tokens": 749412685.0, + "step": 28597 + }, + { + "epoch": 2.3582089552238807, + "grad_norm": 1.9324021339416504, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.7904347777366638, + "num_tokens": 749448458.0, + "step": 28598 + }, + { + "epoch": 2.358291415848932, + "grad_norm": 2.0441231727600098, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7551599740982056, + "num_tokens": 749480511.0, + "step": 28599 + }, + { + "epoch": 2.3583738764739834, + "grad_norm": 2.179046869277954, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7520340085029602, + "num_tokens": 749509587.0, + "step": 28600 + }, + { + "epoch": 2.3584563370990352, + "grad_norm": 2.0663955211639404, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7319214940071106, + "num_tokens": 749542455.0, + "step": 28601 + }, + { + "epoch": 2.3585387977240866, + "grad_norm": 2.307973861694336, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.7904448509216309, + "num_tokens": 749570162.0, + "step": 28602 + }, + { + "epoch": 2.3586212583491384, + "grad_norm": 1.9450973272323608, + "learning_rate": 1e-06, + "loss": 0.6123, + "mean_token_accuracy": 0.796309232711792, + "num_tokens": 749603628.0, + "step": 28603 + }, + { + "epoch": 2.3587037189741897, + "grad_norm": 2.1880288124084473, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7484521865844727, + "num_tokens": 749633180.0, + "step": 28604 + }, + { + "epoch": 2.3587861795992415, + "grad_norm": 2.1693625450134277, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.7963453531265259, + "num_tokens": 749662299.0, + "step": 28605 + }, + { + "epoch": 2.358868640224293, + "grad_norm": 2.170076370239258, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7573475241661072, + "num_tokens": 749693106.0, + "step": 28606 + }, + { + "epoch": 2.3589511008493442, + "grad_norm": 1.7498924732208252, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7743973731994629, + "num_tokens": 749732754.0, + "step": 28607 + }, + { + "epoch": 2.359033561474396, + "grad_norm": 2.233474016189575, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7743215560913086, + "num_tokens": 749760314.0, + "step": 28608 + }, + { + "epoch": 2.3591160220994474, + "grad_norm": 2.268247365951538, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7618510723114014, + "num_tokens": 749786786.0, + "step": 28609 + }, + { + "epoch": 2.359198482724499, + "grad_norm": 1.8578394651412964, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.801740288734436, + "num_tokens": 749822052.0, + "step": 28610 + }, + { + "epoch": 2.3592809433495505, + "grad_norm": 2.0122432708740234, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7722465395927429, + "num_tokens": 749854957.0, + "step": 28611 + }, + { + "epoch": 2.3593634039746023, + "grad_norm": 2.484821319580078, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7604448795318604, + "num_tokens": 749877741.0, + "step": 28612 + }, + { + "epoch": 2.3594458645996537, + "grad_norm": 2.2010269165039062, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7776528596878052, + "num_tokens": 749907075.0, + "step": 28613 + }, + { + "epoch": 2.359528325224705, + "grad_norm": 1.8767448663711548, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7646990418434143, + "num_tokens": 749943933.0, + "step": 28614 + }, + { + "epoch": 2.359610785849757, + "grad_norm": 2.060116767883301, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7328753471374512, + "num_tokens": 749975646.0, + "step": 28615 + }, + { + "epoch": 2.359693246474808, + "grad_norm": 2.2334413528442383, + "learning_rate": 1e-06, + "loss": 0.7564, + "mean_token_accuracy": 0.7581433653831482, + "num_tokens": 750004385.0, + "step": 28616 + }, + { + "epoch": 2.35977570709986, + "grad_norm": 1.8479907512664795, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.8004547357559204, + "num_tokens": 750045702.0, + "step": 28617 + }, + { + "epoch": 2.3598581677249113, + "grad_norm": 1.7478069067001343, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.7921161651611328, + "num_tokens": 750086459.0, + "step": 28618 + }, + { + "epoch": 2.359940628349963, + "grad_norm": 2.0073673725128174, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7751550078392029, + "num_tokens": 750119197.0, + "step": 28619 + }, + { + "epoch": 2.3600230889750144, + "grad_norm": 1.9970468282699585, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7851474285125732, + "num_tokens": 750149623.0, + "step": 28620 + }, + { + "epoch": 2.360105549600066, + "grad_norm": 1.944577693939209, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.770641028881073, + "num_tokens": 750182114.0, + "step": 28621 + }, + { + "epoch": 2.3601880102251176, + "grad_norm": 2.0111379623413086, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.786218523979187, + "num_tokens": 750218065.0, + "step": 28622 + }, + { + "epoch": 2.360270470850169, + "grad_norm": 2.548103094100952, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7251918315887451, + "num_tokens": 750241839.0, + "step": 28623 + }, + { + "epoch": 2.3603529314752207, + "grad_norm": 2.287801742553711, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7743263244628906, + "num_tokens": 750268250.0, + "step": 28624 + }, + { + "epoch": 2.360435392100272, + "grad_norm": 2.127753257751465, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.7847388982772827, + "num_tokens": 750300519.0, + "step": 28625 + }, + { + "epoch": 2.360517852725324, + "grad_norm": 2.129154682159424, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7539342045783997, + "num_tokens": 750329298.0, + "step": 28626 + }, + { + "epoch": 2.3606003133503752, + "grad_norm": 2.168673038482666, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7674980163574219, + "num_tokens": 750359516.0, + "step": 28627 + }, + { + "epoch": 2.3606827739754266, + "grad_norm": 2.3911733627319336, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7468873858451843, + "num_tokens": 750383782.0, + "step": 28628 + }, + { + "epoch": 2.3607652346004784, + "grad_norm": 1.9269708395004272, + "learning_rate": 1e-06, + "loss": 0.6011, + "mean_token_accuracy": 0.790177583694458, + "num_tokens": 750414225.0, + "step": 28629 + }, + { + "epoch": 2.3608476952255297, + "grad_norm": 2.291774272918701, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7449175715446472, + "num_tokens": 750439826.0, + "step": 28630 + }, + { + "epoch": 2.3609301558505815, + "grad_norm": 2.089646100997925, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7439459562301636, + "num_tokens": 750474829.0, + "step": 28631 + }, + { + "epoch": 2.361012616475633, + "grad_norm": 2.015610456466675, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7842239737510681, + "num_tokens": 750506980.0, + "step": 28632 + }, + { + "epoch": 2.3610950771006847, + "grad_norm": 2.019749164581299, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7631511688232422, + "num_tokens": 750541145.0, + "step": 28633 + }, + { + "epoch": 2.361177537725736, + "grad_norm": 1.7593765258789062, + "learning_rate": 1e-06, + "loss": 0.5059, + "mean_token_accuracy": 0.8337152600288391, + "num_tokens": 750577412.0, + "step": 28634 + }, + { + "epoch": 2.3612599983507874, + "grad_norm": 2.0047049522399902, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.7957262396812439, + "num_tokens": 750608588.0, + "step": 28635 + }, + { + "epoch": 2.361342458975839, + "grad_norm": 2.092177152633667, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7749807238578796, + "num_tokens": 750637868.0, + "step": 28636 + }, + { + "epoch": 2.3614249196008905, + "grad_norm": 2.099156379699707, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7428419589996338, + "num_tokens": 750669607.0, + "step": 28637 + }, + { + "epoch": 2.361507380225942, + "grad_norm": 2.1487176418304443, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.7897803783416748, + "num_tokens": 750700548.0, + "step": 28638 + }, + { + "epoch": 2.3615898408509937, + "grad_norm": 2.1730308532714844, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7455285787582397, + "num_tokens": 750729189.0, + "step": 28639 + }, + { + "epoch": 2.361672301476045, + "grad_norm": 2.3590803146362305, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7601473331451416, + "num_tokens": 750754017.0, + "step": 28640 + }, + { + "epoch": 2.361754762101097, + "grad_norm": 2.489655017852783, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.7753493189811707, + "num_tokens": 750776329.0, + "step": 28641 + }, + { + "epoch": 2.361837222726148, + "grad_norm": 1.9870941638946533, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7677958011627197, + "num_tokens": 750809739.0, + "step": 28642 + }, + { + "epoch": 2.3619196833512, + "grad_norm": 2.14313006401062, + "learning_rate": 1e-06, + "loss": 0.5744, + "mean_token_accuracy": 0.8108595609664917, + "num_tokens": 750837189.0, + "step": 28643 + }, + { + "epoch": 2.3620021439762513, + "grad_norm": 1.944215178489685, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7811287641525269, + "num_tokens": 750874877.0, + "step": 28644 + }, + { + "epoch": 2.3620846046013027, + "grad_norm": 2.2555994987487793, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7506980895996094, + "num_tokens": 750901017.0, + "step": 28645 + }, + { + "epoch": 2.3621670652263544, + "grad_norm": 2.157252550125122, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7769085764884949, + "num_tokens": 750927905.0, + "step": 28646 + }, + { + "epoch": 2.362249525851406, + "grad_norm": 2.0164706707000732, + "learning_rate": 1e-06, + "loss": 0.4629, + "mean_token_accuracy": 0.8406757116317749, + "num_tokens": 750959546.0, + "step": 28647 + }, + { + "epoch": 2.3623319864764576, + "grad_norm": 2.158459186553955, + "learning_rate": 1e-06, + "loss": 0.6024, + "mean_token_accuracy": 0.8061376214027405, + "num_tokens": 750986942.0, + "step": 28648 + }, + { + "epoch": 2.362414447101509, + "grad_norm": 1.9544448852539062, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7735950350761414, + "num_tokens": 751019463.0, + "step": 28649 + }, + { + "epoch": 2.3624969077265607, + "grad_norm": 2.1335222721099854, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7840280532836914, + "num_tokens": 751048109.0, + "step": 28650 + }, + { + "epoch": 2.362579368351612, + "grad_norm": 2.1513376235961914, + "learning_rate": 1e-06, + "loss": 0.5122, + "mean_token_accuracy": 0.8251692056655884, + "num_tokens": 751074862.0, + "step": 28651 + }, + { + "epoch": 2.3626618289766634, + "grad_norm": 1.8732960224151611, + "learning_rate": 1e-06, + "loss": 0.6014, + "mean_token_accuracy": 0.791880190372467, + "num_tokens": 751107925.0, + "step": 28652 + }, + { + "epoch": 2.3627442896017152, + "grad_norm": 2.158311605453491, + "learning_rate": 1e-06, + "loss": 0.7483, + "mean_token_accuracy": 0.7720952033996582, + "num_tokens": 751135841.0, + "step": 28653 + }, + { + "epoch": 2.3628267502267666, + "grad_norm": 2.1805310249328613, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7549027800559998, + "num_tokens": 751162777.0, + "step": 28654 + }, + { + "epoch": 2.3629092108518184, + "grad_norm": 2.0552892684936523, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7866240739822388, + "num_tokens": 751195399.0, + "step": 28655 + }, + { + "epoch": 2.3629916714768697, + "grad_norm": 2.057002305984497, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.7823234796524048, + "num_tokens": 751227386.0, + "step": 28656 + }, + { + "epoch": 2.3630741321019215, + "grad_norm": 2.2446508407592773, + "learning_rate": 1e-06, + "loss": 0.6082, + "mean_token_accuracy": 0.7966247797012329, + "num_tokens": 751254438.0, + "step": 28657 + }, + { + "epoch": 2.363156592726973, + "grad_norm": 2.221465826034546, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7480080127716064, + "num_tokens": 751283630.0, + "step": 28658 + }, + { + "epoch": 2.363239053352024, + "grad_norm": 2.444247007369995, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7396955490112305, + "num_tokens": 751308883.0, + "step": 28659 + }, + { + "epoch": 2.363321513977076, + "grad_norm": 2.1915369033813477, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7754385471343994, + "num_tokens": 751338507.0, + "step": 28660 + }, + { + "epoch": 2.3634039746021274, + "grad_norm": 2.2354934215545654, + "learning_rate": 1e-06, + "loss": 0.5438, + "mean_token_accuracy": 0.805347204208374, + "num_tokens": 751365664.0, + "step": 28661 + }, + { + "epoch": 2.363486435227179, + "grad_norm": 1.9564145803451538, + "learning_rate": 1e-06, + "loss": 0.5831, + "mean_token_accuracy": 0.8042503595352173, + "num_tokens": 751396639.0, + "step": 28662 + }, + { + "epoch": 2.3635688958522305, + "grad_norm": 1.9709035158157349, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.7973656058311462, + "num_tokens": 751430329.0, + "step": 28663 + }, + { + "epoch": 2.3636513564772823, + "grad_norm": 2.2046778202056885, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.789836585521698, + "num_tokens": 751460374.0, + "step": 28664 + }, + { + "epoch": 2.3637338171023337, + "grad_norm": 1.8869943618774414, + "learning_rate": 1e-06, + "loss": 0.4793, + "mean_token_accuracy": 0.834270715713501, + "num_tokens": 751492142.0, + "step": 28665 + }, + { + "epoch": 2.363816277727385, + "grad_norm": 1.9007281064987183, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.750988245010376, + "num_tokens": 751528220.0, + "step": 28666 + }, + { + "epoch": 2.363898738352437, + "grad_norm": 1.9610861539840698, + "learning_rate": 1e-06, + "loss": 0.574, + "mean_token_accuracy": 0.8088586330413818, + "num_tokens": 751558626.0, + "step": 28667 + }, + { + "epoch": 2.363981198977488, + "grad_norm": 2.5341641902923584, + "learning_rate": 1e-06, + "loss": 0.7116, + "mean_token_accuracy": 0.7632603645324707, + "num_tokens": 751580919.0, + "step": 28668 + }, + { + "epoch": 2.36406365960254, + "grad_norm": 2.13458514213562, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7621257305145264, + "num_tokens": 751610412.0, + "step": 28669 + }, + { + "epoch": 2.3641461202275913, + "grad_norm": 1.9380578994750977, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.7957241535186768, + "num_tokens": 751643974.0, + "step": 28670 + }, + { + "epoch": 2.364228580852643, + "grad_norm": 1.9046835899353027, + "learning_rate": 1e-06, + "loss": 0.6067, + "mean_token_accuracy": 0.8092831373214722, + "num_tokens": 751678114.0, + "step": 28671 + }, + { + "epoch": 2.3643110414776944, + "grad_norm": 2.034123659133911, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7701748609542847, + "num_tokens": 751710966.0, + "step": 28672 + }, + { + "epoch": 2.364393502102746, + "grad_norm": 2.022759437561035, + "learning_rate": 1e-06, + "loss": 0.5612, + "mean_token_accuracy": 0.8057702779769897, + "num_tokens": 751746316.0, + "step": 28673 + }, + { + "epoch": 2.3644759627277976, + "grad_norm": 2.124260187149048, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.7763018608093262, + "num_tokens": 751773617.0, + "step": 28674 + }, + { + "epoch": 2.364558423352849, + "grad_norm": 2.2382774353027344, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7731562256813049, + "num_tokens": 751799659.0, + "step": 28675 + }, + { + "epoch": 2.3646408839779007, + "grad_norm": 1.9017579555511475, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7688745260238647, + "num_tokens": 751836561.0, + "step": 28676 + }, + { + "epoch": 2.364723344602952, + "grad_norm": 2.3119935989379883, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7679001092910767, + "num_tokens": 751860971.0, + "step": 28677 + }, + { + "epoch": 2.364805805228004, + "grad_norm": 2.1836729049682617, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.764256477355957, + "num_tokens": 751892421.0, + "step": 28678 + }, + { + "epoch": 2.3648882658530552, + "grad_norm": 2.2970595359802246, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7416713237762451, + "num_tokens": 751918053.0, + "step": 28679 + }, + { + "epoch": 2.3649707264781066, + "grad_norm": 1.8257774114608765, + "learning_rate": 1e-06, + "loss": 0.5911, + "mean_token_accuracy": 0.8065487146377563, + "num_tokens": 751952085.0, + "step": 28680 + }, + { + "epoch": 2.3650531871031584, + "grad_norm": 2.0279550552368164, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7757833003997803, + "num_tokens": 751981106.0, + "step": 28681 + }, + { + "epoch": 2.3651356477282097, + "grad_norm": 1.8933478593826294, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.7816543579101562, + "num_tokens": 752017941.0, + "step": 28682 + }, + { + "epoch": 2.365218108353261, + "grad_norm": 2.048536777496338, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.7975322604179382, + "num_tokens": 752049844.0, + "step": 28683 + }, + { + "epoch": 2.365300568978313, + "grad_norm": 2.478172779083252, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7646600008010864, + "num_tokens": 752072144.0, + "step": 28684 + }, + { + "epoch": 2.365383029603364, + "grad_norm": 2.2334837913513184, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7593117952346802, + "num_tokens": 752098990.0, + "step": 28685 + }, + { + "epoch": 2.365465490228416, + "grad_norm": 1.9865416288375854, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7643400430679321, + "num_tokens": 752134748.0, + "step": 28686 + }, + { + "epoch": 2.3655479508534674, + "grad_norm": 2.307687997817993, + "learning_rate": 1e-06, + "loss": 0.7617, + "mean_token_accuracy": 0.7525651454925537, + "num_tokens": 752161993.0, + "step": 28687 + }, + { + "epoch": 2.365630411478519, + "grad_norm": 2.4029393196105957, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7394800186157227, + "num_tokens": 752186891.0, + "step": 28688 + }, + { + "epoch": 2.3657128721035705, + "grad_norm": 2.4012105464935303, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7740885019302368, + "num_tokens": 752211689.0, + "step": 28689 + }, + { + "epoch": 2.365795332728622, + "grad_norm": 2.318336009979248, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7783992886543274, + "num_tokens": 752233916.0, + "step": 28690 + }, + { + "epoch": 2.3658777933536737, + "grad_norm": 2.01696515083313, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7695318460464478, + "num_tokens": 752267988.0, + "step": 28691 + }, + { + "epoch": 2.365960253978725, + "grad_norm": 1.9764842987060547, + "learning_rate": 1e-06, + "loss": 0.7598, + "mean_token_accuracy": 0.7611255645751953, + "num_tokens": 752303868.0, + "step": 28692 + }, + { + "epoch": 2.366042714603777, + "grad_norm": 2.2406351566314697, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7678743004798889, + "num_tokens": 752331165.0, + "step": 28693 + }, + { + "epoch": 2.366125175228828, + "grad_norm": 2.0581581592559814, + "learning_rate": 1e-06, + "loss": 0.6053, + "mean_token_accuracy": 0.7925149202346802, + "num_tokens": 752360526.0, + "step": 28694 + }, + { + "epoch": 2.36620763585388, + "grad_norm": 2.042149543762207, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7779794335365295, + "num_tokens": 752392658.0, + "step": 28695 + }, + { + "epoch": 2.3662900964789313, + "grad_norm": 2.1731889247894287, + "learning_rate": 1e-06, + "loss": 0.519, + "mean_token_accuracy": 0.8192031979560852, + "num_tokens": 752419145.0, + "step": 28696 + }, + { + "epoch": 2.3663725571039826, + "grad_norm": 1.821573257446289, + "learning_rate": 1e-06, + "loss": 0.6083, + "mean_token_accuracy": 0.8023162484169006, + "num_tokens": 752458724.0, + "step": 28697 + }, + { + "epoch": 2.3664550177290344, + "grad_norm": 2.1916205883026123, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.7834154367446899, + "num_tokens": 752486356.0, + "step": 28698 + }, + { + "epoch": 2.366537478354086, + "grad_norm": 1.8736016750335693, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.7846624255180359, + "num_tokens": 752522139.0, + "step": 28699 + }, + { + "epoch": 2.3666199389791376, + "grad_norm": 2.093294143676758, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.7845157980918884, + "num_tokens": 752551575.0, + "step": 28700 + }, + { + "epoch": 2.366702399604189, + "grad_norm": 2.053969383239746, + "learning_rate": 1e-06, + "loss": 0.5806, + "mean_token_accuracy": 0.8098049163818359, + "num_tokens": 752583385.0, + "step": 28701 + }, + { + "epoch": 2.3667848602292407, + "grad_norm": 2.037029266357422, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7443892955780029, + "num_tokens": 752615503.0, + "step": 28702 + }, + { + "epoch": 2.366867320854292, + "grad_norm": 1.812280297279358, + "learning_rate": 1e-06, + "loss": 0.59, + "mean_token_accuracy": 0.808651864528656, + "num_tokens": 752651004.0, + "step": 28703 + }, + { + "epoch": 2.3669497814793434, + "grad_norm": 1.9325907230377197, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.7824214696884155, + "num_tokens": 752683850.0, + "step": 28704 + }, + { + "epoch": 2.3670322421043952, + "grad_norm": 2.0336904525756836, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.7718960046768188, + "num_tokens": 752717002.0, + "step": 28705 + }, + { + "epoch": 2.3671147027294466, + "grad_norm": 1.9806822538375854, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.765631377696991, + "num_tokens": 752749607.0, + "step": 28706 + }, + { + "epoch": 2.3671971633544984, + "grad_norm": 1.9084997177124023, + "learning_rate": 1e-06, + "loss": 0.6785, + "mean_token_accuracy": 0.7830955982208252, + "num_tokens": 752784381.0, + "step": 28707 + }, + { + "epoch": 2.3672796239795497, + "grad_norm": 2.129122734069824, + "learning_rate": 1e-06, + "loss": 0.5928, + "mean_token_accuracy": 0.7983536124229431, + "num_tokens": 752813847.0, + "step": 28708 + }, + { + "epoch": 2.3673620846046015, + "grad_norm": 2.198382616043091, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7805342674255371, + "num_tokens": 752841806.0, + "step": 28709 + }, + { + "epoch": 2.367444545229653, + "grad_norm": 2.348181962966919, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7494689226150513, + "num_tokens": 752866647.0, + "step": 28710 + }, + { + "epoch": 2.367527005854704, + "grad_norm": 2.284602165222168, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.767139196395874, + "num_tokens": 752891928.0, + "step": 28711 + }, + { + "epoch": 2.367609466479756, + "grad_norm": 2.0971739292144775, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7588683366775513, + "num_tokens": 752920732.0, + "step": 28712 + }, + { + "epoch": 2.3676919271048074, + "grad_norm": 1.9300897121429443, + "learning_rate": 1e-06, + "loss": 0.5627, + "mean_token_accuracy": 0.8006797432899475, + "num_tokens": 752950663.0, + "step": 28713 + }, + { + "epoch": 2.367774387729859, + "grad_norm": 2.1236445903778076, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.7418166995048523, + "num_tokens": 752978938.0, + "step": 28714 + }, + { + "epoch": 2.3678568483549105, + "grad_norm": 2.1715810298919678, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7620394825935364, + "num_tokens": 753009099.0, + "step": 28715 + }, + { + "epoch": 2.3679393089799623, + "grad_norm": 2.1737630367279053, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7781597971916199, + "num_tokens": 753035907.0, + "step": 28716 + }, + { + "epoch": 2.3680217696050136, + "grad_norm": 2.2357420921325684, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7810893654823303, + "num_tokens": 753061971.0, + "step": 28717 + }, + { + "epoch": 2.368104230230065, + "grad_norm": 2.5915935039520264, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7380058765411377, + "num_tokens": 753094873.0, + "step": 28718 + }, + { + "epoch": 2.368186690855117, + "grad_norm": 1.8895000219345093, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7537567615509033, + "num_tokens": 753133530.0, + "step": 28719 + }, + { + "epoch": 2.368269151480168, + "grad_norm": 1.9163568019866943, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7664964199066162, + "num_tokens": 753169879.0, + "step": 28720 + }, + { + "epoch": 2.36835161210522, + "grad_norm": 2.329653024673462, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.777378261089325, + "num_tokens": 753196849.0, + "step": 28721 + }, + { + "epoch": 2.3684340727302713, + "grad_norm": 2.1225268840789795, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.779074490070343, + "num_tokens": 753225137.0, + "step": 28722 + }, + { + "epoch": 2.3685165333553226, + "grad_norm": 2.3839046955108643, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.780636727809906, + "num_tokens": 753251577.0, + "step": 28723 + }, + { + "epoch": 2.3685989939803744, + "grad_norm": 2.0278563499450684, + "learning_rate": 1e-06, + "loss": 0.5291, + "mean_token_accuracy": 0.8076673150062561, + "num_tokens": 753280724.0, + "step": 28724 + }, + { + "epoch": 2.368681454605426, + "grad_norm": 2.072147846221924, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7654602527618408, + "num_tokens": 753312070.0, + "step": 28725 + }, + { + "epoch": 2.3687639152304776, + "grad_norm": 2.4043030738830566, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7711684703826904, + "num_tokens": 753338089.0, + "step": 28726 + }, + { + "epoch": 2.368846375855529, + "grad_norm": 2.0161283016204834, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.7871608734130859, + "num_tokens": 753371864.0, + "step": 28727 + }, + { + "epoch": 2.3689288364805803, + "grad_norm": 2.1035122871398926, + "learning_rate": 1e-06, + "loss": 0.5835, + "mean_token_accuracy": 0.807997465133667, + "num_tokens": 753402657.0, + "step": 28728 + }, + { + "epoch": 2.369011297105632, + "grad_norm": 2.2108232975006104, + "learning_rate": 1e-06, + "loss": 0.5957, + "mean_token_accuracy": 0.7996236085891724, + "num_tokens": 753432300.0, + "step": 28729 + }, + { + "epoch": 2.3690937577306834, + "grad_norm": 2.2910358905792236, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7706794142723083, + "num_tokens": 753458806.0, + "step": 28730 + }, + { + "epoch": 2.369176218355735, + "grad_norm": 2.4134135246276855, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7471606135368347, + "num_tokens": 753483872.0, + "step": 28731 + }, + { + "epoch": 2.3692586789807866, + "grad_norm": 1.9718366861343384, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7589730024337769, + "num_tokens": 753518264.0, + "step": 28732 + }, + { + "epoch": 2.3693411396058384, + "grad_norm": 2.040203094482422, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.7915176153182983, + "num_tokens": 753548633.0, + "step": 28733 + }, + { + "epoch": 2.3694236002308897, + "grad_norm": 2.562980890274048, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7731204628944397, + "num_tokens": 753568852.0, + "step": 28734 + }, + { + "epoch": 2.369506060855941, + "grad_norm": 2.2203025817871094, + "learning_rate": 1e-06, + "loss": 0.5962, + "mean_token_accuracy": 0.799889326095581, + "num_tokens": 753598608.0, + "step": 28735 + }, + { + "epoch": 2.369588521480993, + "grad_norm": 2.1593897342681885, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7519392371177673, + "num_tokens": 753627895.0, + "step": 28736 + }, + { + "epoch": 2.369670982106044, + "grad_norm": 2.1151347160339355, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7773392200469971, + "num_tokens": 753659373.0, + "step": 28737 + }, + { + "epoch": 2.369753442731096, + "grad_norm": 2.1406702995300293, + "learning_rate": 1e-06, + "loss": 0.8208, + "mean_token_accuracy": 0.7412569522857666, + "num_tokens": 753692584.0, + "step": 28738 + }, + { + "epoch": 2.3698359033561474, + "grad_norm": 1.9430468082427979, + "learning_rate": 1e-06, + "loss": 0.5558, + "mean_token_accuracy": 0.8108930587768555, + "num_tokens": 753727755.0, + "step": 28739 + }, + { + "epoch": 2.369918363981199, + "grad_norm": 2.027082920074463, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.8014580011367798, + "num_tokens": 753757534.0, + "step": 28740 + }, + { + "epoch": 2.3700008246062505, + "grad_norm": 2.3020176887512207, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.791703462600708, + "num_tokens": 753784101.0, + "step": 28741 + }, + { + "epoch": 2.370083285231302, + "grad_norm": 2.252610921859741, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.7512093186378479, + "num_tokens": 753810218.0, + "step": 28742 + }, + { + "epoch": 2.3701657458563536, + "grad_norm": 2.4727463722229004, + "learning_rate": 1e-06, + "loss": 0.7486, + "mean_token_accuracy": 0.755478024482727, + "num_tokens": 753834185.0, + "step": 28743 + }, + { + "epoch": 2.370248206481405, + "grad_norm": 1.9913712739944458, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.793246865272522, + "num_tokens": 753867460.0, + "step": 28744 + }, + { + "epoch": 2.370330667106457, + "grad_norm": 2.0857431888580322, + "learning_rate": 1e-06, + "loss": 0.5769, + "mean_token_accuracy": 0.8079107999801636, + "num_tokens": 753900739.0, + "step": 28745 + }, + { + "epoch": 2.370413127731508, + "grad_norm": 2.295708656311035, + "learning_rate": 1e-06, + "loss": 0.5221, + "mean_token_accuracy": 0.8125113248825073, + "num_tokens": 753926804.0, + "step": 28746 + }, + { + "epoch": 2.37049558835656, + "grad_norm": 2.251044511795044, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.785295307636261, + "num_tokens": 753954274.0, + "step": 28747 + }, + { + "epoch": 2.3705780489816113, + "grad_norm": 1.9654768705368042, + "learning_rate": 1e-06, + "loss": 0.5914, + "mean_token_accuracy": 0.795361340045929, + "num_tokens": 753986204.0, + "step": 28748 + }, + { + "epoch": 2.3706605096066626, + "grad_norm": 1.7636762857437134, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7564347386360168, + "num_tokens": 754027545.0, + "step": 28749 + }, + { + "epoch": 2.3707429702317144, + "grad_norm": 2.183884859085083, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7618975639343262, + "num_tokens": 754056410.0, + "step": 28750 + }, + { + "epoch": 2.370825430856766, + "grad_norm": 1.986333966255188, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7775870561599731, + "num_tokens": 754089202.0, + "step": 28751 + }, + { + "epoch": 2.3709078914818176, + "grad_norm": 2.3636634349823, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7789584398269653, + "num_tokens": 754114424.0, + "step": 28752 + }, + { + "epoch": 2.370990352106869, + "grad_norm": 2.089698076248169, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7565501928329468, + "num_tokens": 754147837.0, + "step": 28753 + }, + { + "epoch": 2.3710728127319207, + "grad_norm": 1.9230589866638184, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.7930915951728821, + "num_tokens": 754181550.0, + "step": 28754 + }, + { + "epoch": 2.371155273356972, + "grad_norm": 2.021364212036133, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.782130241394043, + "num_tokens": 754215119.0, + "step": 28755 + }, + { + "epoch": 2.3712377339820234, + "grad_norm": 1.8798524141311646, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.792935848236084, + "num_tokens": 754249795.0, + "step": 28756 + }, + { + "epoch": 2.371320194607075, + "grad_norm": 2.0774118900299072, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7639226913452148, + "num_tokens": 754279239.0, + "step": 28757 + }, + { + "epoch": 2.3714026552321266, + "grad_norm": 1.8965495824813843, + "learning_rate": 1e-06, + "loss": 0.5956, + "mean_token_accuracy": 0.8180367350578308, + "num_tokens": 754310306.0, + "step": 28758 + }, + { + "epoch": 2.3714851158571784, + "grad_norm": 2.264253854751587, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7665876150131226, + "num_tokens": 754343810.0, + "step": 28759 + }, + { + "epoch": 2.3715675764822297, + "grad_norm": 1.849208950996399, + "learning_rate": 1e-06, + "loss": 0.5638, + "mean_token_accuracy": 0.812656044960022, + "num_tokens": 754376004.0, + "step": 28760 + }, + { + "epoch": 2.3716500371072815, + "grad_norm": 2.048408269882202, + "learning_rate": 1e-06, + "loss": 0.5947, + "mean_token_accuracy": 0.7961709499359131, + "num_tokens": 754408838.0, + "step": 28761 + }, + { + "epoch": 2.371732497732333, + "grad_norm": 2.346829652786255, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7411555051803589, + "num_tokens": 754437948.0, + "step": 28762 + }, + { + "epoch": 2.371814958357384, + "grad_norm": 2.2145400047302246, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7239072322845459, + "num_tokens": 754467843.0, + "step": 28763 + }, + { + "epoch": 2.371897418982436, + "grad_norm": 1.9494905471801758, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7831957936286926, + "num_tokens": 754501344.0, + "step": 28764 + }, + { + "epoch": 2.3719798796074874, + "grad_norm": 1.9227274656295776, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7708252668380737, + "num_tokens": 754538048.0, + "step": 28765 + }, + { + "epoch": 2.372062340232539, + "grad_norm": 2.2953710556030273, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.7829700708389282, + "num_tokens": 754561344.0, + "step": 28766 + }, + { + "epoch": 2.3721448008575905, + "grad_norm": 1.8764749765396118, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.7864091992378235, + "num_tokens": 754594315.0, + "step": 28767 + }, + { + "epoch": 2.372227261482642, + "grad_norm": 2.081685781478882, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7641709446907043, + "num_tokens": 754629316.0, + "step": 28768 + }, + { + "epoch": 2.3723097221076936, + "grad_norm": 2.1937448978424072, + "learning_rate": 1e-06, + "loss": 0.7134, + "mean_token_accuracy": 0.7687386870384216, + "num_tokens": 754659415.0, + "step": 28769 + }, + { + "epoch": 2.372392182732745, + "grad_norm": 2.254655122756958, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.7805741429328918, + "num_tokens": 754687910.0, + "step": 28770 + }, + { + "epoch": 2.372474643357797, + "grad_norm": 1.909944772720337, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7660820484161377, + "num_tokens": 754723772.0, + "step": 28771 + }, + { + "epoch": 2.372557103982848, + "grad_norm": 1.9437520503997803, + "learning_rate": 1e-06, + "loss": 0.4592, + "mean_token_accuracy": 0.8243346214294434, + "num_tokens": 754754925.0, + "step": 28772 + }, + { + "epoch": 2.3726395646078995, + "grad_norm": 1.9490350484848022, + "learning_rate": 1e-06, + "loss": 0.5699, + "mean_token_accuracy": 0.8106497526168823, + "num_tokens": 754785279.0, + "step": 28773 + }, + { + "epoch": 2.3727220252329513, + "grad_norm": 2.2378809452056885, + "learning_rate": 1e-06, + "loss": 0.5949, + "mean_token_accuracy": 0.7914731502532959, + "num_tokens": 754812613.0, + "step": 28774 + }, + { + "epoch": 2.3728044858580026, + "grad_norm": 1.9776402711868286, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7871711254119873, + "num_tokens": 754842461.0, + "step": 28775 + }, + { + "epoch": 2.3728869464830544, + "grad_norm": 2.1015114784240723, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7401924133300781, + "num_tokens": 754875237.0, + "step": 28776 + }, + { + "epoch": 2.3729694071081058, + "grad_norm": 2.4120559692382812, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7583919763565063, + "num_tokens": 754897822.0, + "step": 28777 + }, + { + "epoch": 2.3730518677331576, + "grad_norm": 1.9822521209716797, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7653974294662476, + "num_tokens": 754929716.0, + "step": 28778 + }, + { + "epoch": 2.373134328358209, + "grad_norm": 1.9412883520126343, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7737575769424438, + "num_tokens": 754965938.0, + "step": 28779 + }, + { + "epoch": 2.3732167889832603, + "grad_norm": 1.911893367767334, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7774913311004639, + "num_tokens": 755002137.0, + "step": 28780 + }, + { + "epoch": 2.373299249608312, + "grad_norm": 2.0582711696624756, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.777396559715271, + "num_tokens": 755036284.0, + "step": 28781 + }, + { + "epoch": 2.3733817102333634, + "grad_norm": 2.0008699893951416, + "learning_rate": 1e-06, + "loss": 0.6094, + "mean_token_accuracy": 0.7942673563957214, + "num_tokens": 755068864.0, + "step": 28782 + }, + { + "epoch": 2.373464170858415, + "grad_norm": 2.1607258319854736, + "learning_rate": 1e-06, + "loss": 0.6066, + "mean_token_accuracy": 0.7983588576316833, + "num_tokens": 755099740.0, + "step": 28783 + }, + { + "epoch": 2.3735466314834666, + "grad_norm": 1.8680403232574463, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7408670783042908, + "num_tokens": 755136104.0, + "step": 28784 + }, + { + "epoch": 2.3736290921085184, + "grad_norm": 2.10758638381958, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7619609832763672, + "num_tokens": 755168397.0, + "step": 28785 + }, + { + "epoch": 2.3737115527335697, + "grad_norm": 2.284608840942383, + "learning_rate": 1e-06, + "loss": 0.5625, + "mean_token_accuracy": 0.8058470487594604, + "num_tokens": 755194950.0, + "step": 28786 + }, + { + "epoch": 2.373794013358621, + "grad_norm": 1.9881932735443115, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7690541744232178, + "num_tokens": 755226550.0, + "step": 28787 + }, + { + "epoch": 2.373876473983673, + "grad_norm": 1.8377013206481934, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7410227656364441, + "num_tokens": 755264835.0, + "step": 28788 + }, + { + "epoch": 2.373958934608724, + "grad_norm": 2.1724677085876465, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.796906590461731, + "num_tokens": 755292218.0, + "step": 28789 + }, + { + "epoch": 2.374041395233776, + "grad_norm": 1.9635757207870483, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7771823406219482, + "num_tokens": 755323826.0, + "step": 28790 + }, + { + "epoch": 2.3741238558588273, + "grad_norm": 2.0655670166015625, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.7925611138343811, + "num_tokens": 755354372.0, + "step": 28791 + }, + { + "epoch": 2.374206316483879, + "grad_norm": 1.8955787420272827, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7748562097549438, + "num_tokens": 755389084.0, + "step": 28792 + }, + { + "epoch": 2.3742887771089305, + "grad_norm": 2.302703857421875, + "learning_rate": 1e-06, + "loss": 0.5854, + "mean_token_accuracy": 0.8100445866584778, + "num_tokens": 755413352.0, + "step": 28793 + }, + { + "epoch": 2.374371237733982, + "grad_norm": 1.9542381763458252, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7296386957168579, + "num_tokens": 755450140.0, + "step": 28794 + }, + { + "epoch": 2.3744536983590336, + "grad_norm": 2.1099560260772705, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7537671327590942, + "num_tokens": 755482340.0, + "step": 28795 + }, + { + "epoch": 2.374536158984085, + "grad_norm": 2.106004476547241, + "learning_rate": 1e-06, + "loss": 0.5276, + "mean_token_accuracy": 0.8110730648040771, + "num_tokens": 755511834.0, + "step": 28796 + }, + { + "epoch": 2.374618619609137, + "grad_norm": 1.8277473449707031, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7812271118164062, + "num_tokens": 755551247.0, + "step": 28797 + }, + { + "epoch": 2.374701080234188, + "grad_norm": 2.141327381134033, + "learning_rate": 1e-06, + "loss": 0.5831, + "mean_token_accuracy": 0.8086342215538025, + "num_tokens": 755577553.0, + "step": 28798 + }, + { + "epoch": 2.37478354085924, + "grad_norm": 2.2902090549468994, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7752101421356201, + "num_tokens": 755605471.0, + "step": 28799 + }, + { + "epoch": 2.3748660014842913, + "grad_norm": 1.8621320724487305, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.7817173004150391, + "num_tokens": 755642299.0, + "step": 28800 + }, + { + "epoch": 2.3749484621093426, + "grad_norm": 2.0315606594085693, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7275258898735046, + "num_tokens": 755673480.0, + "step": 28801 + }, + { + "epoch": 2.3750309227343944, + "grad_norm": 2.036482572555542, + "learning_rate": 1e-06, + "loss": 0.5732, + "mean_token_accuracy": 0.8087137937545776, + "num_tokens": 755705548.0, + "step": 28802 + }, + { + "epoch": 2.3751133833594458, + "grad_norm": 2.321943759918213, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7753932476043701, + "num_tokens": 755731277.0, + "step": 28803 + }, + { + "epoch": 2.3751958439844976, + "grad_norm": 2.2289769649505615, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.7805840969085693, + "num_tokens": 755757491.0, + "step": 28804 + }, + { + "epoch": 2.375278304609549, + "grad_norm": 2.0229287147521973, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7320095300674438, + "num_tokens": 755790002.0, + "step": 28805 + }, + { + "epoch": 2.3753607652346007, + "grad_norm": 2.118990898132324, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7818208932876587, + "num_tokens": 755819088.0, + "step": 28806 + }, + { + "epoch": 2.375443225859652, + "grad_norm": 1.9167392253875732, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.7989830374717712, + "num_tokens": 755852860.0, + "step": 28807 + }, + { + "epoch": 2.3755256864847034, + "grad_norm": 2.3669991493225098, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7535232305526733, + "num_tokens": 755878954.0, + "step": 28808 + }, + { + "epoch": 2.375608147109755, + "grad_norm": 2.0851247310638428, + "learning_rate": 1e-06, + "loss": 0.6751, + "mean_token_accuracy": 0.7845518589019775, + "num_tokens": 755907248.0, + "step": 28809 + }, + { + "epoch": 2.3756906077348066, + "grad_norm": 1.9395380020141602, + "learning_rate": 1e-06, + "loss": 0.7327, + "mean_token_accuracy": 0.7540117502212524, + "num_tokens": 755943537.0, + "step": 28810 + }, + { + "epoch": 2.3757730683598584, + "grad_norm": 1.9481676816940308, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7550703883171082, + "num_tokens": 755976944.0, + "step": 28811 + }, + { + "epoch": 2.3758555289849097, + "grad_norm": 2.3350231647491455, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7666903734207153, + "num_tokens": 756008085.0, + "step": 28812 + }, + { + "epoch": 2.375937989609961, + "grad_norm": 1.94431471824646, + "learning_rate": 1e-06, + "loss": 0.5929, + "mean_token_accuracy": 0.7989421486854553, + "num_tokens": 756037328.0, + "step": 28813 + }, + { + "epoch": 2.376020450235013, + "grad_norm": 2.137683629989624, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7553545236587524, + "num_tokens": 756066914.0, + "step": 28814 + }, + { + "epoch": 2.376102910860064, + "grad_norm": 2.3211538791656494, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7787823677062988, + "num_tokens": 756094314.0, + "step": 28815 + }, + { + "epoch": 2.376185371485116, + "grad_norm": 2.106234312057495, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7814942598342896, + "num_tokens": 756121973.0, + "step": 28816 + }, + { + "epoch": 2.3762678321101673, + "grad_norm": 2.1231346130371094, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7237462997436523, + "num_tokens": 756154453.0, + "step": 28817 + }, + { + "epoch": 2.3763502927352187, + "grad_norm": 1.9625780582427979, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.779211699962616, + "num_tokens": 756187208.0, + "step": 28818 + }, + { + "epoch": 2.3764327533602705, + "grad_norm": 2.08162784576416, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7621657848358154, + "num_tokens": 756218945.0, + "step": 28819 + }, + { + "epoch": 2.376515213985322, + "grad_norm": 1.8678762912750244, + "learning_rate": 1e-06, + "loss": 0.6367, + "mean_token_accuracy": 0.7972474098205566, + "num_tokens": 756253616.0, + "step": 28820 + }, + { + "epoch": 2.3765976746103736, + "grad_norm": 2.0907652378082275, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.7946861982345581, + "num_tokens": 756283022.0, + "step": 28821 + }, + { + "epoch": 2.376680135235425, + "grad_norm": 2.092160224914551, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7585852146148682, + "num_tokens": 756312073.0, + "step": 28822 + }, + { + "epoch": 2.376762595860477, + "grad_norm": 2.0551438331604004, + "learning_rate": 1e-06, + "loss": 0.5988, + "mean_token_accuracy": 0.8007973432540894, + "num_tokens": 756339918.0, + "step": 28823 + }, + { + "epoch": 2.376845056485528, + "grad_norm": 1.9964208602905273, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7660097479820251, + "num_tokens": 756372181.0, + "step": 28824 + }, + { + "epoch": 2.3769275171105795, + "grad_norm": 1.967067003250122, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.762201189994812, + "num_tokens": 756406647.0, + "step": 28825 + }, + { + "epoch": 2.3770099777356313, + "grad_norm": 2.3142457008361816, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7722480297088623, + "num_tokens": 756434364.0, + "step": 28826 + }, + { + "epoch": 2.3770924383606826, + "grad_norm": 2.3261735439300537, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.7554531097412109, + "num_tokens": 756461924.0, + "step": 28827 + }, + { + "epoch": 2.3771748989857344, + "grad_norm": 1.9548598527908325, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.76823890209198, + "num_tokens": 756497819.0, + "step": 28828 + }, + { + "epoch": 2.3772573596107858, + "grad_norm": 2.060213327407837, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7508541345596313, + "num_tokens": 756530611.0, + "step": 28829 + }, + { + "epoch": 2.3773398202358376, + "grad_norm": 2.2602808475494385, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7660353183746338, + "num_tokens": 756558850.0, + "step": 28830 + }, + { + "epoch": 2.377422280860889, + "grad_norm": 2.0827784538269043, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7715526819229126, + "num_tokens": 756589393.0, + "step": 28831 + }, + { + "epoch": 2.3775047414859403, + "grad_norm": 2.1407792568206787, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7715761661529541, + "num_tokens": 756616999.0, + "step": 28832 + }, + { + "epoch": 2.377587202110992, + "grad_norm": 2.107612371444702, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.7866171598434448, + "num_tokens": 756646169.0, + "step": 28833 + }, + { + "epoch": 2.3776696627360434, + "grad_norm": 2.271951913833618, + "learning_rate": 1e-06, + "loss": 0.5222, + "mean_token_accuracy": 0.8093117475509644, + "num_tokens": 756671113.0, + "step": 28834 + }, + { + "epoch": 2.377752123361095, + "grad_norm": 1.9743224382400513, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7784696817398071, + "num_tokens": 756706157.0, + "step": 28835 + }, + { + "epoch": 2.3778345839861466, + "grad_norm": 1.9006041288375854, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.7893053889274597, + "num_tokens": 756740180.0, + "step": 28836 + }, + { + "epoch": 2.3779170446111983, + "grad_norm": 1.9957348108291626, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7493188381195068, + "num_tokens": 756774767.0, + "step": 28837 + }, + { + "epoch": 2.3779995052362497, + "grad_norm": 2.1592934131622314, + "learning_rate": 1e-06, + "loss": 0.6041, + "mean_token_accuracy": 0.803406834602356, + "num_tokens": 756803909.0, + "step": 28838 + }, + { + "epoch": 2.378081965861301, + "grad_norm": 1.8608983755111694, + "learning_rate": 1e-06, + "loss": 0.593, + "mean_token_accuracy": 0.8062466979026794, + "num_tokens": 756838644.0, + "step": 28839 + }, + { + "epoch": 2.378164426486353, + "grad_norm": 2.242022752761841, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7796809673309326, + "num_tokens": 756864460.0, + "step": 28840 + }, + { + "epoch": 2.378246887111404, + "grad_norm": 2.159928798675537, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7485249042510986, + "num_tokens": 756893163.0, + "step": 28841 + }, + { + "epoch": 2.378329347736456, + "grad_norm": 2.168527126312256, + "learning_rate": 1e-06, + "loss": 0.607, + "mean_token_accuracy": 0.786915123462677, + "num_tokens": 756923681.0, + "step": 28842 + }, + { + "epoch": 2.3784118083615073, + "grad_norm": 1.9208914041519165, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.7799251079559326, + "num_tokens": 756959308.0, + "step": 28843 + }, + { + "epoch": 2.378494268986559, + "grad_norm": 2.0782759189605713, + "learning_rate": 1e-06, + "loss": 0.5027, + "mean_token_accuracy": 0.8232420682907104, + "num_tokens": 756989317.0, + "step": 28844 + }, + { + "epoch": 2.3785767296116105, + "grad_norm": 2.2533624172210693, + "learning_rate": 1e-06, + "loss": 0.6683, + "mean_token_accuracy": 0.7954078912734985, + "num_tokens": 757016054.0, + "step": 28845 + }, + { + "epoch": 2.378659190236662, + "grad_norm": 1.7935562133789062, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7748368978500366, + "num_tokens": 757058682.0, + "step": 28846 + }, + { + "epoch": 2.3787416508617136, + "grad_norm": 2.19134521484375, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.7871814370155334, + "num_tokens": 757086866.0, + "step": 28847 + }, + { + "epoch": 2.378824111486765, + "grad_norm": 2.442882537841797, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7782024145126343, + "num_tokens": 757110297.0, + "step": 28848 + }, + { + "epoch": 2.3789065721118168, + "grad_norm": 2.1742405891418457, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7620962858200073, + "num_tokens": 757137963.0, + "step": 28849 + }, + { + "epoch": 2.378989032736868, + "grad_norm": 2.4864790439605713, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7503337264060974, + "num_tokens": 757162715.0, + "step": 28850 + }, + { + "epoch": 2.37907149336192, + "grad_norm": 2.027566909790039, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7463668584823608, + "num_tokens": 757195622.0, + "step": 28851 + }, + { + "epoch": 2.3791539539869713, + "grad_norm": 2.0078883171081543, + "learning_rate": 1e-06, + "loss": 0.5743, + "mean_token_accuracy": 0.809291422367096, + "num_tokens": 757225205.0, + "step": 28852 + }, + { + "epoch": 2.3792364146120226, + "grad_norm": 1.7987364530563354, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7757149934768677, + "num_tokens": 757263273.0, + "step": 28853 + }, + { + "epoch": 2.3793188752370744, + "grad_norm": 2.28720760345459, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.783566951751709, + "num_tokens": 757287759.0, + "step": 28854 + }, + { + "epoch": 2.3794013358621258, + "grad_norm": 1.904349684715271, + "learning_rate": 1e-06, + "loss": 0.5672, + "mean_token_accuracy": 0.809688150882721, + "num_tokens": 757321999.0, + "step": 28855 + }, + { + "epoch": 2.3794837964871776, + "grad_norm": 2.21606183052063, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7555451393127441, + "num_tokens": 757349468.0, + "step": 28856 + }, + { + "epoch": 2.379566257112229, + "grad_norm": 1.9026036262512207, + "learning_rate": 1e-06, + "loss": 0.4724, + "mean_token_accuracy": 0.840499997138977, + "num_tokens": 757384490.0, + "step": 28857 + }, + { + "epoch": 2.3796487177372803, + "grad_norm": 2.165170431137085, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7829710245132446, + "num_tokens": 757413299.0, + "step": 28858 + }, + { + "epoch": 2.379731178362332, + "grad_norm": 2.350689649581909, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7757176160812378, + "num_tokens": 757440144.0, + "step": 28859 + }, + { + "epoch": 2.3798136389873834, + "grad_norm": 2.0135791301727295, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7788923978805542, + "num_tokens": 757475487.0, + "step": 28860 + }, + { + "epoch": 2.379896099612435, + "grad_norm": 1.9851161241531372, + "learning_rate": 1e-06, + "loss": 0.6587, + "mean_token_accuracy": 0.7849408388137817, + "num_tokens": 757508423.0, + "step": 28861 + }, + { + "epoch": 2.3799785602374866, + "grad_norm": 2.1331064701080322, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.7762277722358704, + "num_tokens": 757537830.0, + "step": 28862 + }, + { + "epoch": 2.380061020862538, + "grad_norm": 2.2019054889678955, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.7929688096046448, + "num_tokens": 757567934.0, + "step": 28863 + }, + { + "epoch": 2.3801434814875897, + "grad_norm": 2.181074857711792, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.7850567102432251, + "num_tokens": 757596219.0, + "step": 28864 + }, + { + "epoch": 2.380225942112641, + "grad_norm": 2.068096399307251, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7417863607406616, + "num_tokens": 757629870.0, + "step": 28865 + }, + { + "epoch": 2.380308402737693, + "grad_norm": 2.1815273761749268, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7665702700614929, + "num_tokens": 757658751.0, + "step": 28866 + }, + { + "epoch": 2.380390863362744, + "grad_norm": 2.1417694091796875, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7828981876373291, + "num_tokens": 757690220.0, + "step": 28867 + }, + { + "epoch": 2.380473323987796, + "grad_norm": 2.2115962505340576, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7596956491470337, + "num_tokens": 757720620.0, + "step": 28868 + }, + { + "epoch": 2.3805557846128473, + "grad_norm": 2.269205093383789, + "learning_rate": 1e-06, + "loss": 0.5833, + "mean_token_accuracy": 0.8035289645195007, + "num_tokens": 757744755.0, + "step": 28869 + }, + { + "epoch": 2.3806382452378987, + "grad_norm": 2.1230766773223877, + "learning_rate": 1e-06, + "loss": 0.5436, + "mean_token_accuracy": 0.816267728805542, + "num_tokens": 757770765.0, + "step": 28870 + }, + { + "epoch": 2.3807207058629505, + "grad_norm": 1.880297064781189, + "learning_rate": 1e-06, + "loss": 0.5829, + "mean_token_accuracy": 0.8032200336456299, + "num_tokens": 757802615.0, + "step": 28871 + }, + { + "epoch": 2.380803166488002, + "grad_norm": 2.087987184524536, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7820671200752258, + "num_tokens": 757833698.0, + "step": 28872 + }, + { + "epoch": 2.3808856271130536, + "grad_norm": 2.255007266998291, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.7841378450393677, + "num_tokens": 757859825.0, + "step": 28873 + }, + { + "epoch": 2.380968087738105, + "grad_norm": 2.1684157848358154, + "learning_rate": 1e-06, + "loss": 0.5589, + "mean_token_accuracy": 0.8096722364425659, + "num_tokens": 757885890.0, + "step": 28874 + }, + { + "epoch": 2.3810505483631568, + "grad_norm": 2.2344915866851807, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.7781034111976624, + "num_tokens": 757913651.0, + "step": 28875 + }, + { + "epoch": 2.381133008988208, + "grad_norm": 2.1975302696228027, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7536238431930542, + "num_tokens": 757940504.0, + "step": 28876 + }, + { + "epoch": 2.3812154696132595, + "grad_norm": 2.0226614475250244, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7226861119270325, + "num_tokens": 757973451.0, + "step": 28877 + }, + { + "epoch": 2.3812979302383113, + "grad_norm": 2.5485079288482666, + "learning_rate": 1e-06, + "loss": 0.6072, + "mean_token_accuracy": 0.7959121465682983, + "num_tokens": 757999408.0, + "step": 28878 + }, + { + "epoch": 2.3813803908633626, + "grad_norm": 2.038808584213257, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.7888942956924438, + "num_tokens": 758031417.0, + "step": 28879 + }, + { + "epoch": 2.3814628514884144, + "grad_norm": 2.0354223251342773, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7801235318183899, + "num_tokens": 758065101.0, + "step": 28880 + }, + { + "epoch": 2.3815453121134658, + "grad_norm": 2.0056538581848145, + "learning_rate": 1e-06, + "loss": 0.7706, + "mean_token_accuracy": 0.7594460248947144, + "num_tokens": 758095887.0, + "step": 28881 + }, + { + "epoch": 2.3816277727385176, + "grad_norm": 2.281947612762451, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7577502727508545, + "num_tokens": 758121935.0, + "step": 28882 + }, + { + "epoch": 2.381710233363569, + "grad_norm": 2.4051334857940674, + "learning_rate": 1e-06, + "loss": 0.5935, + "mean_token_accuracy": 0.799089789390564, + "num_tokens": 758144641.0, + "step": 28883 + }, + { + "epoch": 2.3817926939886203, + "grad_norm": 2.1998891830444336, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7731406688690186, + "num_tokens": 758174024.0, + "step": 28884 + }, + { + "epoch": 2.381875154613672, + "grad_norm": 2.2115840911865234, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7586521506309509, + "num_tokens": 758201093.0, + "step": 28885 + }, + { + "epoch": 2.3819576152387234, + "grad_norm": 2.1159555912017822, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7470618486404419, + "num_tokens": 758231265.0, + "step": 28886 + }, + { + "epoch": 2.382040075863775, + "grad_norm": 2.075331449508667, + "learning_rate": 1e-06, + "loss": 0.5149, + "mean_token_accuracy": 0.8092249035835266, + "num_tokens": 758257181.0, + "step": 28887 + }, + { + "epoch": 2.3821225364888265, + "grad_norm": 2.2317304611206055, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7830209732055664, + "num_tokens": 758283369.0, + "step": 28888 + }, + { + "epoch": 2.3822049971138783, + "grad_norm": 2.3482120037078857, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7777957916259766, + "num_tokens": 758307403.0, + "step": 28889 + }, + { + "epoch": 2.3822874577389297, + "grad_norm": 2.260802984237671, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.738378643989563, + "num_tokens": 758336474.0, + "step": 28890 + }, + { + "epoch": 2.382369918363981, + "grad_norm": 2.261321544647217, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7607892751693726, + "num_tokens": 758363378.0, + "step": 28891 + }, + { + "epoch": 2.382452378989033, + "grad_norm": 2.001586675643921, + "learning_rate": 1e-06, + "loss": 0.5657, + "mean_token_accuracy": 0.8151849508285522, + "num_tokens": 758396494.0, + "step": 28892 + }, + { + "epoch": 2.382534839614084, + "grad_norm": 1.9638832807540894, + "learning_rate": 1e-06, + "loss": 0.4327, + "mean_token_accuracy": 0.8443517684936523, + "num_tokens": 758429515.0, + "step": 28893 + }, + { + "epoch": 2.382617300239136, + "grad_norm": 1.8737537860870361, + "learning_rate": 1e-06, + "loss": 0.6194, + "mean_token_accuracy": 0.7903540134429932, + "num_tokens": 758465364.0, + "step": 28894 + }, + { + "epoch": 2.3826997608641873, + "grad_norm": 2.2814457416534424, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7528785467147827, + "num_tokens": 758491575.0, + "step": 28895 + }, + { + "epoch": 2.382782221489239, + "grad_norm": 2.25298810005188, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7659293413162231, + "num_tokens": 758518409.0, + "step": 28896 + }, + { + "epoch": 2.3828646821142905, + "grad_norm": 1.9443800449371338, + "learning_rate": 1e-06, + "loss": 0.5416, + "mean_token_accuracy": 0.819008469581604, + "num_tokens": 758550225.0, + "step": 28897 + }, + { + "epoch": 2.382947142739342, + "grad_norm": 2.02026629447937, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7412517070770264, + "num_tokens": 758585133.0, + "step": 28898 + }, + { + "epoch": 2.3830296033643936, + "grad_norm": 2.3230977058410645, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7402524352073669, + "num_tokens": 758611463.0, + "step": 28899 + }, + { + "epoch": 2.383112063989445, + "grad_norm": 1.890389084815979, + "learning_rate": 1e-06, + "loss": 0.4965, + "mean_token_accuracy": 0.8270359039306641, + "num_tokens": 758643708.0, + "step": 28900 + }, + { + "epoch": 2.3831945246144968, + "grad_norm": 2.1894278526306152, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7564675807952881, + "num_tokens": 758673804.0, + "step": 28901 + }, + { + "epoch": 2.383276985239548, + "grad_norm": 2.6748619079589844, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.777556300163269, + "num_tokens": 758697033.0, + "step": 28902 + }, + { + "epoch": 2.3833594458645995, + "grad_norm": 2.1660099029541016, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7758965492248535, + "num_tokens": 758728786.0, + "step": 28903 + }, + { + "epoch": 2.3834419064896513, + "grad_norm": 2.071608066558838, + "learning_rate": 1e-06, + "loss": 0.7623, + "mean_token_accuracy": 0.7618499994277954, + "num_tokens": 758759620.0, + "step": 28904 + }, + { + "epoch": 2.3835243671147026, + "grad_norm": 2.3836231231689453, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7730423212051392, + "num_tokens": 758787208.0, + "step": 28905 + }, + { + "epoch": 2.3836068277397544, + "grad_norm": 2.2669498920440674, + "learning_rate": 1e-06, + "loss": 0.5625, + "mean_token_accuracy": 0.8003922700881958, + "num_tokens": 758812323.0, + "step": 28906 + }, + { + "epoch": 2.3836892883648058, + "grad_norm": 2.122981071472168, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.7895798683166504, + "num_tokens": 758841973.0, + "step": 28907 + }, + { + "epoch": 2.383771748989857, + "grad_norm": 1.9040582180023193, + "learning_rate": 1e-06, + "loss": 0.6223, + "mean_token_accuracy": 0.7923505902290344, + "num_tokens": 758874841.0, + "step": 28908 + }, + { + "epoch": 2.383854209614909, + "grad_norm": 2.1030447483062744, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7489017248153687, + "num_tokens": 758906889.0, + "step": 28909 + }, + { + "epoch": 2.3839366702399603, + "grad_norm": 1.9602171182632446, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7684361934661865, + "num_tokens": 758939272.0, + "step": 28910 + }, + { + "epoch": 2.384019130865012, + "grad_norm": 1.9279636144638062, + "learning_rate": 1e-06, + "loss": 0.5498, + "mean_token_accuracy": 0.8115899562835693, + "num_tokens": 758970948.0, + "step": 28911 + }, + { + "epoch": 2.3841015914900634, + "grad_norm": 1.7974241971969604, + "learning_rate": 1e-06, + "loss": 0.3989, + "mean_token_accuracy": 0.8589702248573303, + "num_tokens": 759006156.0, + "step": 28912 + }, + { + "epoch": 2.384184052115115, + "grad_norm": 2.432863473892212, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.7865511178970337, + "num_tokens": 759027919.0, + "step": 28913 + }, + { + "epoch": 2.3842665127401665, + "grad_norm": 2.341966390609741, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7523369193077087, + "num_tokens": 759053054.0, + "step": 28914 + }, + { + "epoch": 2.384348973365218, + "grad_norm": 2.3014540672302246, + "learning_rate": 1e-06, + "loss": 0.5517, + "mean_token_accuracy": 0.8016037940979004, + "num_tokens": 759077251.0, + "step": 28915 + }, + { + "epoch": 2.3844314339902697, + "grad_norm": 2.2661988735198975, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.780325710773468, + "num_tokens": 759105638.0, + "step": 28916 + }, + { + "epoch": 2.384513894615321, + "grad_norm": 2.3200881481170654, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7552397847175598, + "num_tokens": 759134222.0, + "step": 28917 + }, + { + "epoch": 2.384596355240373, + "grad_norm": 2.1808879375457764, + "learning_rate": 1e-06, + "loss": 0.4876, + "mean_token_accuracy": 0.8210234642028809, + "num_tokens": 759163887.0, + "step": 28918 + }, + { + "epoch": 2.384678815865424, + "grad_norm": 2.1445438861846924, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7692968845367432, + "num_tokens": 759192518.0, + "step": 28919 + }, + { + "epoch": 2.384761276490476, + "grad_norm": 2.090855598449707, + "learning_rate": 1e-06, + "loss": 0.605, + "mean_token_accuracy": 0.7902280688285828, + "num_tokens": 759219541.0, + "step": 28920 + }, + { + "epoch": 2.3848437371155273, + "grad_norm": 1.8825501203536987, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7821599245071411, + "num_tokens": 759258113.0, + "step": 28921 + }, + { + "epoch": 2.3849261977405787, + "grad_norm": 2.1121857166290283, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7656222581863403, + "num_tokens": 759292248.0, + "step": 28922 + }, + { + "epoch": 2.3850086583656305, + "grad_norm": 2.2122581005096436, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.7910246849060059, + "num_tokens": 759321983.0, + "step": 28923 + }, + { + "epoch": 2.385091118990682, + "grad_norm": 2.473102569580078, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7677330374717712, + "num_tokens": 759347733.0, + "step": 28924 + }, + { + "epoch": 2.3851735796157336, + "grad_norm": 2.1046900749206543, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7792102098464966, + "num_tokens": 759378680.0, + "step": 28925 + }, + { + "epoch": 2.385256040240785, + "grad_norm": 2.2146503925323486, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7654783725738525, + "num_tokens": 759406340.0, + "step": 28926 + }, + { + "epoch": 2.3853385008658368, + "grad_norm": 2.6323819160461426, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7508224248886108, + "num_tokens": 759427556.0, + "step": 28927 + }, + { + "epoch": 2.385420961490888, + "grad_norm": 2.103335380554199, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7475769519805908, + "num_tokens": 759458150.0, + "step": 28928 + }, + { + "epoch": 2.3855034221159395, + "grad_norm": 2.2597463130950928, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7688411474227905, + "num_tokens": 759483701.0, + "step": 28929 + }, + { + "epoch": 2.3855858827409913, + "grad_norm": 1.878593921661377, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7350164651870728, + "num_tokens": 759519781.0, + "step": 28930 + }, + { + "epoch": 2.3856683433660426, + "grad_norm": 1.722161054611206, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.7883043885231018, + "num_tokens": 759560787.0, + "step": 28931 + }, + { + "epoch": 2.3857508039910944, + "grad_norm": 2.058295965194702, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7491539716720581, + "num_tokens": 759590617.0, + "step": 28932 + }, + { + "epoch": 2.3858332646161458, + "grad_norm": 1.9483903646469116, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.7899436354637146, + "num_tokens": 759625460.0, + "step": 28933 + }, + { + "epoch": 2.3859157252411975, + "grad_norm": 2.2124743461608887, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.7859190106391907, + "num_tokens": 759654622.0, + "step": 28934 + }, + { + "epoch": 2.385998185866249, + "grad_norm": 1.9931329488754272, + "learning_rate": 1e-06, + "loss": 0.5781, + "mean_token_accuracy": 0.8153560161590576, + "num_tokens": 759687288.0, + "step": 28935 + }, + { + "epoch": 2.3860806464913003, + "grad_norm": 2.4784109592437744, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.7888915538787842, + "num_tokens": 759709233.0, + "step": 28936 + }, + { + "epoch": 2.386163107116352, + "grad_norm": 2.4116580486297607, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.7790958881378174, + "num_tokens": 759736052.0, + "step": 28937 + }, + { + "epoch": 2.3862455677414034, + "grad_norm": 2.1879026889801025, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.7952920198440552, + "num_tokens": 759763277.0, + "step": 28938 + }, + { + "epoch": 2.386328028366455, + "grad_norm": 1.8260992765426636, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.7894796133041382, + "num_tokens": 759800422.0, + "step": 28939 + }, + { + "epoch": 2.3864104889915065, + "grad_norm": 2.1163105964660645, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7707996368408203, + "num_tokens": 759830119.0, + "step": 28940 + }, + { + "epoch": 2.3864929496165583, + "grad_norm": 2.2110719680786133, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7448477745056152, + "num_tokens": 759857917.0, + "step": 28941 + }, + { + "epoch": 2.3865754102416097, + "grad_norm": 1.9141234159469604, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7730101346969604, + "num_tokens": 759894792.0, + "step": 28942 + }, + { + "epoch": 2.386657870866661, + "grad_norm": 1.8970106840133667, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.7934324741363525, + "num_tokens": 759929536.0, + "step": 28943 + }, + { + "epoch": 2.386740331491713, + "grad_norm": 1.8562815189361572, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.7736151218414307, + "num_tokens": 759968212.0, + "step": 28944 + }, + { + "epoch": 2.386822792116764, + "grad_norm": 1.9628183841705322, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7759882807731628, + "num_tokens": 760002991.0, + "step": 28945 + }, + { + "epoch": 2.386905252741816, + "grad_norm": 1.8779792785644531, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7832815647125244, + "num_tokens": 760040962.0, + "step": 28946 + }, + { + "epoch": 2.3869877133668673, + "grad_norm": 2.001450300216675, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.788876473903656, + "num_tokens": 760073611.0, + "step": 28947 + }, + { + "epoch": 2.3870701739919187, + "grad_norm": 2.1076231002807617, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.7959601879119873, + "num_tokens": 760100976.0, + "step": 28948 + }, + { + "epoch": 2.3871526346169705, + "grad_norm": 1.9757927656173706, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7581856846809387, + "num_tokens": 760134181.0, + "step": 28949 + }, + { + "epoch": 2.387235095242022, + "grad_norm": 2.1416382789611816, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7529917359352112, + "num_tokens": 760163242.0, + "step": 28950 + }, + { + "epoch": 2.3873175558670736, + "grad_norm": 1.9483981132507324, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7450885772705078, + "num_tokens": 760200925.0, + "step": 28951 + }, + { + "epoch": 2.387400016492125, + "grad_norm": 2.465790033340454, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7825438976287842, + "num_tokens": 760222767.0, + "step": 28952 + }, + { + "epoch": 2.3874824771171763, + "grad_norm": 2.230206251144409, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7262207269668579, + "num_tokens": 760248699.0, + "step": 28953 + }, + { + "epoch": 2.387564937742228, + "grad_norm": 2.4454448223114014, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.7776341438293457, + "num_tokens": 760270168.0, + "step": 28954 + }, + { + "epoch": 2.3876473983672795, + "grad_norm": 1.7878941297531128, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7727054357528687, + "num_tokens": 760306757.0, + "step": 28955 + }, + { + "epoch": 2.3877298589923313, + "grad_norm": 2.2502601146698, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7708221673965454, + "num_tokens": 760335237.0, + "step": 28956 + }, + { + "epoch": 2.3878123196173826, + "grad_norm": 2.012334108352661, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7667670249938965, + "num_tokens": 760365637.0, + "step": 28957 + }, + { + "epoch": 2.3878947802424344, + "grad_norm": 2.334141731262207, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.756149172782898, + "num_tokens": 760390785.0, + "step": 28958 + }, + { + "epoch": 2.3879772408674858, + "grad_norm": 2.300668239593506, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7464573383331299, + "num_tokens": 760416726.0, + "step": 28959 + }, + { + "epoch": 2.388059701492537, + "grad_norm": 2.128127098083496, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7619174718856812, + "num_tokens": 760445549.0, + "step": 28960 + }, + { + "epoch": 2.388142162117589, + "grad_norm": 2.1094002723693848, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7751007080078125, + "num_tokens": 760475193.0, + "step": 28961 + }, + { + "epoch": 2.3882246227426402, + "grad_norm": 2.023284912109375, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7736856937408447, + "num_tokens": 760504085.0, + "step": 28962 + }, + { + "epoch": 2.388307083367692, + "grad_norm": 2.2351903915405273, + "learning_rate": 1e-06, + "loss": 0.5626, + "mean_token_accuracy": 0.8155632615089417, + "num_tokens": 760528636.0, + "step": 28963 + }, + { + "epoch": 2.3883895439927434, + "grad_norm": 2.149310350418091, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7833313941955566, + "num_tokens": 760561452.0, + "step": 28964 + }, + { + "epoch": 2.388472004617795, + "grad_norm": 2.138972520828247, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7676073908805847, + "num_tokens": 760588115.0, + "step": 28965 + }, + { + "epoch": 2.3885544652428465, + "grad_norm": 2.0214970111846924, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7695552706718445, + "num_tokens": 760621240.0, + "step": 28966 + }, + { + "epoch": 2.388636925867898, + "grad_norm": 2.204397439956665, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7747045755386353, + "num_tokens": 760649875.0, + "step": 28967 + }, + { + "epoch": 2.3887193864929497, + "grad_norm": 2.1044111251831055, + "learning_rate": 1e-06, + "loss": 0.5949, + "mean_token_accuracy": 0.7884016633033752, + "num_tokens": 760679433.0, + "step": 28968 + }, + { + "epoch": 2.388801847118001, + "grad_norm": 1.9936225414276123, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7476974725723267, + "num_tokens": 760716376.0, + "step": 28969 + }, + { + "epoch": 2.388884307743053, + "grad_norm": 2.30515718460083, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7283843755722046, + "num_tokens": 760744228.0, + "step": 28970 + }, + { + "epoch": 2.388966768368104, + "grad_norm": 2.2869508266448975, + "learning_rate": 1e-06, + "loss": 0.7886, + "mean_token_accuracy": 0.7464605569839478, + "num_tokens": 760770325.0, + "step": 28971 + }, + { + "epoch": 2.389049228993156, + "grad_norm": 2.3361899852752686, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7648265957832336, + "num_tokens": 760797727.0, + "step": 28972 + }, + { + "epoch": 2.3891316896182073, + "grad_norm": 2.0739543437957764, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7320160269737244, + "num_tokens": 760831020.0, + "step": 28973 + }, + { + "epoch": 2.3892141502432587, + "grad_norm": 1.9976459741592407, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7614490985870361, + "num_tokens": 760862922.0, + "step": 28974 + }, + { + "epoch": 2.3892966108683105, + "grad_norm": 1.933860421180725, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7718006372451782, + "num_tokens": 760896160.0, + "step": 28975 + }, + { + "epoch": 2.389379071493362, + "grad_norm": 2.043447256088257, + "learning_rate": 1e-06, + "loss": 0.545, + "mean_token_accuracy": 0.8125684261322021, + "num_tokens": 760927979.0, + "step": 28976 + }, + { + "epoch": 2.3894615321184136, + "grad_norm": 1.8941110372543335, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.7864778637886047, + "num_tokens": 760960759.0, + "step": 28977 + }, + { + "epoch": 2.389543992743465, + "grad_norm": 2.092526912689209, + "learning_rate": 1e-06, + "loss": 0.6219, + "mean_token_accuracy": 0.7978244423866272, + "num_tokens": 760991040.0, + "step": 28978 + }, + { + "epoch": 2.3896264533685168, + "grad_norm": 2.0225775241851807, + "learning_rate": 1e-06, + "loss": 0.5887, + "mean_token_accuracy": 0.8035765886306763, + "num_tokens": 761023490.0, + "step": 28979 + }, + { + "epoch": 2.389708913993568, + "grad_norm": 2.1325294971466064, + "learning_rate": 1e-06, + "loss": 0.5458, + "mean_token_accuracy": 0.8172045350074768, + "num_tokens": 761051568.0, + "step": 28980 + }, + { + "epoch": 2.3897913746186195, + "grad_norm": 2.3647117614746094, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.811928927898407, + "num_tokens": 761077764.0, + "step": 28981 + }, + { + "epoch": 2.3898738352436713, + "grad_norm": 2.1104423999786377, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7672194838523865, + "num_tokens": 761107058.0, + "step": 28982 + }, + { + "epoch": 2.3899562958687226, + "grad_norm": 2.398531913757324, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7208578586578369, + "num_tokens": 761134474.0, + "step": 28983 + }, + { + "epoch": 2.3900387564937744, + "grad_norm": 2.451404571533203, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7382784485816956, + "num_tokens": 761159822.0, + "step": 28984 + }, + { + "epoch": 2.3901212171188257, + "grad_norm": 2.594118356704712, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7657475471496582, + "num_tokens": 761181787.0, + "step": 28985 + }, + { + "epoch": 2.3902036777438775, + "grad_norm": 2.2032885551452637, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.77020263671875, + "num_tokens": 761211805.0, + "step": 28986 + }, + { + "epoch": 2.390286138368929, + "grad_norm": 2.030533790588379, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.771303653717041, + "num_tokens": 761242036.0, + "step": 28987 + }, + { + "epoch": 2.3903685989939802, + "grad_norm": 2.266812562942505, + "learning_rate": 1e-06, + "loss": 0.5822, + "mean_token_accuracy": 0.8024213314056396, + "num_tokens": 761269578.0, + "step": 28988 + }, + { + "epoch": 2.390451059619032, + "grad_norm": 2.379747152328491, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.7872929573059082, + "num_tokens": 761293126.0, + "step": 28989 + }, + { + "epoch": 2.3905335202440834, + "grad_norm": 2.124300241470337, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.77553391456604, + "num_tokens": 761326719.0, + "step": 28990 + }, + { + "epoch": 2.390615980869135, + "grad_norm": 2.006002902984619, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.7919914722442627, + "num_tokens": 761357399.0, + "step": 28991 + }, + { + "epoch": 2.3906984414941865, + "grad_norm": 1.9184986352920532, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.789172887802124, + "num_tokens": 761391351.0, + "step": 28992 + }, + { + "epoch": 2.390780902119238, + "grad_norm": 2.0299360752105713, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7807761430740356, + "num_tokens": 761422424.0, + "step": 28993 + }, + { + "epoch": 2.3908633627442897, + "grad_norm": 2.5526134967803955, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.785833477973938, + "num_tokens": 761443782.0, + "step": 28994 + }, + { + "epoch": 2.390945823369341, + "grad_norm": 2.4173998832702637, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7466086745262146, + "num_tokens": 761468545.0, + "step": 28995 + }, + { + "epoch": 2.391028283994393, + "grad_norm": 2.0481278896331787, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.782135009765625, + "num_tokens": 761501081.0, + "step": 28996 + }, + { + "epoch": 2.391110744619444, + "grad_norm": 1.9583662748336792, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.7912408709526062, + "num_tokens": 761537335.0, + "step": 28997 + }, + { + "epoch": 2.3911932052444955, + "grad_norm": 2.581292152404785, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7558712959289551, + "num_tokens": 761559705.0, + "step": 28998 + }, + { + "epoch": 2.3912756658695473, + "grad_norm": 1.9759190082550049, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.7862898111343384, + "num_tokens": 761590722.0, + "step": 28999 + }, + { + "epoch": 2.3913581264945987, + "grad_norm": 2.1867172718048096, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7464001774787903, + "num_tokens": 761619512.0, + "step": 29000 + }, + { + "epoch": 2.3914405871196505, + "grad_norm": 2.1828434467315674, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7867811918258667, + "num_tokens": 761648800.0, + "step": 29001 + }, + { + "epoch": 2.391523047744702, + "grad_norm": 2.255378007888794, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7864673137664795, + "num_tokens": 761675891.0, + "step": 29002 + }, + { + "epoch": 2.3916055083697536, + "grad_norm": 1.9100744724273682, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7732930779457092, + "num_tokens": 761707642.0, + "step": 29003 + }, + { + "epoch": 2.391687968994805, + "grad_norm": 1.9129072427749634, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7503037452697754, + "num_tokens": 761749434.0, + "step": 29004 + }, + { + "epoch": 2.3917704296198563, + "grad_norm": 1.9814602136611938, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7674638628959656, + "num_tokens": 761781763.0, + "step": 29005 + }, + { + "epoch": 2.391852890244908, + "grad_norm": 2.2498035430908203, + "learning_rate": 1e-06, + "loss": 0.5641, + "mean_token_accuracy": 0.8150291442871094, + "num_tokens": 761805501.0, + "step": 29006 + }, + { + "epoch": 2.3919353508699595, + "grad_norm": 2.098845958709717, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.7864206433296204, + "num_tokens": 761835793.0, + "step": 29007 + }, + { + "epoch": 2.3920178114950112, + "grad_norm": 1.9587956666946411, + "learning_rate": 1e-06, + "loss": 0.5306, + "mean_token_accuracy": 0.816034197807312, + "num_tokens": 761870347.0, + "step": 29008 + }, + { + "epoch": 2.3921002721200626, + "grad_norm": 2.1594748497009277, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7653594017028809, + "num_tokens": 761897606.0, + "step": 29009 + }, + { + "epoch": 2.3921827327451144, + "grad_norm": 2.1413867473602295, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.766582190990448, + "num_tokens": 761928947.0, + "step": 29010 + }, + { + "epoch": 2.3922651933701657, + "grad_norm": 2.1547656059265137, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7837046384811401, + "num_tokens": 761961536.0, + "step": 29011 + }, + { + "epoch": 2.392347653995217, + "grad_norm": 2.1085567474365234, + "learning_rate": 1e-06, + "loss": 0.5416, + "mean_token_accuracy": 0.8122367858886719, + "num_tokens": 761992670.0, + "step": 29012 + }, + { + "epoch": 2.392430114620269, + "grad_norm": 2.3126864433288574, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7571356296539307, + "num_tokens": 762020117.0, + "step": 29013 + }, + { + "epoch": 2.3925125752453202, + "grad_norm": 2.039130926132202, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7265533208847046, + "num_tokens": 762056899.0, + "step": 29014 + }, + { + "epoch": 2.392595035870372, + "grad_norm": 2.5518805980682373, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.766951858997345, + "num_tokens": 762078339.0, + "step": 29015 + }, + { + "epoch": 2.3926774964954234, + "grad_norm": 2.0706679821014404, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7524666786193848, + "num_tokens": 762114842.0, + "step": 29016 + }, + { + "epoch": 2.392759957120475, + "grad_norm": 2.2127771377563477, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7576614618301392, + "num_tokens": 762143606.0, + "step": 29017 + }, + { + "epoch": 2.3928424177455265, + "grad_norm": 2.0324714183807373, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7518552541732788, + "num_tokens": 762175375.0, + "step": 29018 + }, + { + "epoch": 2.392924878370578, + "grad_norm": 2.397487163543701, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7639660835266113, + "num_tokens": 762200633.0, + "step": 29019 + }, + { + "epoch": 2.3930073389956297, + "grad_norm": 2.5468320846557617, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7456773519515991, + "num_tokens": 762223137.0, + "step": 29020 + }, + { + "epoch": 2.393089799620681, + "grad_norm": 2.0494391918182373, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7757833003997803, + "num_tokens": 762253291.0, + "step": 29021 + }, + { + "epoch": 2.393172260245733, + "grad_norm": 2.1723556518554688, + "learning_rate": 1e-06, + "loss": 0.5632, + "mean_token_accuracy": 0.8101681470870972, + "num_tokens": 762280615.0, + "step": 29022 + }, + { + "epoch": 2.393254720870784, + "grad_norm": 1.9322645664215088, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.803243100643158, + "num_tokens": 762315135.0, + "step": 29023 + }, + { + "epoch": 2.393337181495836, + "grad_norm": 1.818229079246521, + "learning_rate": 1e-06, + "loss": 0.6178, + "mean_token_accuracy": 0.7933155298233032, + "num_tokens": 762352453.0, + "step": 29024 + }, + { + "epoch": 2.3934196421208873, + "grad_norm": 2.1057629585266113, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.7820989489555359, + "num_tokens": 762382458.0, + "step": 29025 + }, + { + "epoch": 2.3935021027459387, + "grad_norm": 2.1620752811431885, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.7937338352203369, + "num_tokens": 762410152.0, + "step": 29026 + }, + { + "epoch": 2.3935845633709905, + "grad_norm": 2.225343704223633, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7684275507926941, + "num_tokens": 762437002.0, + "step": 29027 + }, + { + "epoch": 2.393667023996042, + "grad_norm": 1.7014291286468506, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.7929996252059937, + "num_tokens": 762478810.0, + "step": 29028 + }, + { + "epoch": 2.3937494846210936, + "grad_norm": 2.326218843460083, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7495995759963989, + "num_tokens": 762503849.0, + "step": 29029 + }, + { + "epoch": 2.393831945246145, + "grad_norm": 1.9975032806396484, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.7868914008140564, + "num_tokens": 762536691.0, + "step": 29030 + }, + { + "epoch": 2.3939144058711967, + "grad_norm": 2.1773109436035156, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7723326683044434, + "num_tokens": 762565512.0, + "step": 29031 + }, + { + "epoch": 2.393996866496248, + "grad_norm": 2.272592306137085, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7585042119026184, + "num_tokens": 762594772.0, + "step": 29032 + }, + { + "epoch": 2.3940793271212994, + "grad_norm": 1.9707235097885132, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7804611921310425, + "num_tokens": 762629896.0, + "step": 29033 + }, + { + "epoch": 2.3941617877463512, + "grad_norm": 2.3724706172943115, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7679778337478638, + "num_tokens": 762656697.0, + "step": 29034 + }, + { + "epoch": 2.3942442483714026, + "grad_norm": 2.1854665279388428, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7288737297058105, + "num_tokens": 762687583.0, + "step": 29035 + }, + { + "epoch": 2.394326708996454, + "grad_norm": 1.9502383470535278, + "learning_rate": 1e-06, + "loss": 0.6429, + "mean_token_accuracy": 0.7880967855453491, + "num_tokens": 762722016.0, + "step": 29036 + }, + { + "epoch": 2.3944091696215057, + "grad_norm": 2.008197784423828, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7823023200035095, + "num_tokens": 762753060.0, + "step": 29037 + }, + { + "epoch": 2.394491630246557, + "grad_norm": 2.0825705528259277, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7842574119567871, + "num_tokens": 762783227.0, + "step": 29038 + }, + { + "epoch": 2.394574090871609, + "grad_norm": 2.3024260997772217, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7509599924087524, + "num_tokens": 762812232.0, + "step": 29039 + }, + { + "epoch": 2.3946565514966602, + "grad_norm": 2.1468019485473633, + "learning_rate": 1e-06, + "loss": 0.6205, + "mean_token_accuracy": 0.7907391786575317, + "num_tokens": 762841798.0, + "step": 29040 + }, + { + "epoch": 2.394739012121712, + "grad_norm": 2.241809606552124, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7558239698410034, + "num_tokens": 762871731.0, + "step": 29041 + }, + { + "epoch": 2.3948214727467634, + "grad_norm": 2.3650712966918945, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.773684024810791, + "num_tokens": 762896926.0, + "step": 29042 + }, + { + "epoch": 2.3949039333718147, + "grad_norm": 2.156167507171631, + "learning_rate": 1e-06, + "loss": 0.5054, + "mean_token_accuracy": 0.8301876783370972, + "num_tokens": 762921644.0, + "step": 29043 + }, + { + "epoch": 2.3949863939968665, + "grad_norm": 2.0241386890411377, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7522013187408447, + "num_tokens": 762954564.0, + "step": 29044 + }, + { + "epoch": 2.395068854621918, + "grad_norm": 2.218738317489624, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7621331810951233, + "num_tokens": 762978962.0, + "step": 29045 + }, + { + "epoch": 2.3951513152469697, + "grad_norm": 2.0166726112365723, + "learning_rate": 1e-06, + "loss": 0.5828, + "mean_token_accuracy": 0.8064471483230591, + "num_tokens": 763010368.0, + "step": 29046 + }, + { + "epoch": 2.395233775872021, + "grad_norm": 1.8015326261520386, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7857350707054138, + "num_tokens": 763049909.0, + "step": 29047 + }, + { + "epoch": 2.395316236497073, + "grad_norm": 2.3994479179382324, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7761988639831543, + "num_tokens": 763071792.0, + "step": 29048 + }, + { + "epoch": 2.395398697122124, + "grad_norm": 1.971268892288208, + "learning_rate": 1e-06, + "loss": 0.57, + "mean_token_accuracy": 0.8093414902687073, + "num_tokens": 763101494.0, + "step": 29049 + }, + { + "epoch": 2.3954811577471755, + "grad_norm": 2.0485806465148926, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.7847012281417847, + "num_tokens": 763132478.0, + "step": 29050 + }, + { + "epoch": 2.3955636183722273, + "grad_norm": 2.0439469814300537, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7556848526000977, + "num_tokens": 763164993.0, + "step": 29051 + }, + { + "epoch": 2.3956460789972787, + "grad_norm": 2.0223305225372314, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.7991976141929626, + "num_tokens": 763196096.0, + "step": 29052 + }, + { + "epoch": 2.3957285396223305, + "grad_norm": 1.9892562627792358, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7652052044868469, + "num_tokens": 763230506.0, + "step": 29053 + }, + { + "epoch": 2.395811000247382, + "grad_norm": 2.0300838947296143, + "learning_rate": 1e-06, + "loss": 0.6522, + "mean_token_accuracy": 0.7847279906272888, + "num_tokens": 763264524.0, + "step": 29054 + }, + { + "epoch": 2.3958934608724336, + "grad_norm": 2.267777681350708, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7717626690864563, + "num_tokens": 763292624.0, + "step": 29055 + }, + { + "epoch": 2.395975921497485, + "grad_norm": 1.9879933595657349, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7591802477836609, + "num_tokens": 763325768.0, + "step": 29056 + }, + { + "epoch": 2.3960583821225363, + "grad_norm": 1.7721129655838013, + "learning_rate": 1e-06, + "loss": 0.543, + "mean_token_accuracy": 0.8172585964202881, + "num_tokens": 763361234.0, + "step": 29057 + }, + { + "epoch": 2.396140842747588, + "grad_norm": 2.1568949222564697, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.7888957262039185, + "num_tokens": 763389798.0, + "step": 29058 + }, + { + "epoch": 2.3962233033726394, + "grad_norm": 2.1742517948150635, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7754136919975281, + "num_tokens": 763420162.0, + "step": 29059 + }, + { + "epoch": 2.3963057639976912, + "grad_norm": 2.0274300575256348, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7583168745040894, + "num_tokens": 763458231.0, + "step": 29060 + }, + { + "epoch": 2.3963882246227426, + "grad_norm": 2.265460252761841, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7139896750450134, + "num_tokens": 763490209.0, + "step": 29061 + }, + { + "epoch": 2.3964706852477944, + "grad_norm": 2.090322971343994, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7438995242118835, + "num_tokens": 763520320.0, + "step": 29062 + }, + { + "epoch": 2.3965531458728457, + "grad_norm": 1.9396064281463623, + "learning_rate": 1e-06, + "loss": 0.5487, + "mean_token_accuracy": 0.8050587177276611, + "num_tokens": 763554106.0, + "step": 29063 + }, + { + "epoch": 2.396635606497897, + "grad_norm": 2.0515940189361572, + "learning_rate": 1e-06, + "loss": 0.6991, + "mean_token_accuracy": 0.7773826718330383, + "num_tokens": 763584686.0, + "step": 29064 + }, + { + "epoch": 2.396718067122949, + "grad_norm": 1.9895482063293457, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7857679128646851, + "num_tokens": 763615721.0, + "step": 29065 + }, + { + "epoch": 2.3968005277480002, + "grad_norm": 2.028172254562378, + "learning_rate": 1e-06, + "loss": 0.615, + "mean_token_accuracy": 0.7863402962684631, + "num_tokens": 763649028.0, + "step": 29066 + }, + { + "epoch": 2.396882988373052, + "grad_norm": 1.997405767440796, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7816153168678284, + "num_tokens": 763682646.0, + "step": 29067 + }, + { + "epoch": 2.3969654489981034, + "grad_norm": 2.1585514545440674, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.7833309173583984, + "num_tokens": 763709828.0, + "step": 29068 + }, + { + "epoch": 2.397047909623155, + "grad_norm": 1.9871909618377686, + "learning_rate": 1e-06, + "loss": 0.502, + "mean_token_accuracy": 0.8275513648986816, + "num_tokens": 763737823.0, + "step": 29069 + }, + { + "epoch": 2.3971303702482065, + "grad_norm": 2.367401123046875, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7197983264923096, + "num_tokens": 763764032.0, + "step": 29070 + }, + { + "epoch": 2.397212830873258, + "grad_norm": 2.0625829696655273, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7609238624572754, + "num_tokens": 763795040.0, + "step": 29071 + }, + { + "epoch": 2.3972952914983097, + "grad_norm": 2.044847011566162, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7646671533584595, + "num_tokens": 763823926.0, + "step": 29072 + }, + { + "epoch": 2.397377752123361, + "grad_norm": 2.067378282546997, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7778300046920776, + "num_tokens": 763854341.0, + "step": 29073 + }, + { + "epoch": 2.397460212748413, + "grad_norm": 2.0526933670043945, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.766151487827301, + "num_tokens": 763885501.0, + "step": 29074 + }, + { + "epoch": 2.397542673373464, + "grad_norm": 2.019218921661377, + "learning_rate": 1e-06, + "loss": 0.5523, + "mean_token_accuracy": 0.8011745810508728, + "num_tokens": 763914491.0, + "step": 29075 + }, + { + "epoch": 2.397625133998516, + "grad_norm": 2.119736909866333, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7829813957214355, + "num_tokens": 763940778.0, + "step": 29076 + }, + { + "epoch": 2.3977075946235673, + "grad_norm": 2.0095295906066895, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7702299952507019, + "num_tokens": 763972550.0, + "step": 29077 + }, + { + "epoch": 2.3977900552486187, + "grad_norm": 2.1836795806884766, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.7875037789344788, + "num_tokens": 763999351.0, + "step": 29078 + }, + { + "epoch": 2.3978725158736705, + "grad_norm": 2.1778564453125, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7688428163528442, + "num_tokens": 764029183.0, + "step": 29079 + }, + { + "epoch": 2.397954976498722, + "grad_norm": 2.2019364833831787, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.7923007011413574, + "num_tokens": 764055828.0, + "step": 29080 + }, + { + "epoch": 2.398037437123773, + "grad_norm": 2.044818878173828, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7672240734100342, + "num_tokens": 764086494.0, + "step": 29081 + }, + { + "epoch": 2.398119897748825, + "grad_norm": 1.8383963108062744, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.781017541885376, + "num_tokens": 764123949.0, + "step": 29082 + }, + { + "epoch": 2.3982023583738763, + "grad_norm": 1.9843940734863281, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7731369733810425, + "num_tokens": 764157574.0, + "step": 29083 + }, + { + "epoch": 2.398284818998928, + "grad_norm": 2.2454097270965576, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7525822520256042, + "num_tokens": 764189190.0, + "step": 29084 + }, + { + "epoch": 2.3983672796239794, + "grad_norm": 2.2189249992370605, + "learning_rate": 1e-06, + "loss": 0.525, + "mean_token_accuracy": 0.8146772384643555, + "num_tokens": 764214282.0, + "step": 29085 + }, + { + "epoch": 2.3984497402490312, + "grad_norm": 2.33091139793396, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7680485248565674, + "num_tokens": 764238455.0, + "step": 29086 + }, + { + "epoch": 2.3985322008740826, + "grad_norm": 2.1728549003601074, + "learning_rate": 1e-06, + "loss": 0.6214, + "mean_token_accuracy": 0.7849010825157166, + "num_tokens": 764266059.0, + "step": 29087 + }, + { + "epoch": 2.398614661499134, + "grad_norm": 2.395479440689087, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7563438415527344, + "num_tokens": 764289137.0, + "step": 29088 + }, + { + "epoch": 2.3986971221241857, + "grad_norm": 2.0922577381134033, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.7837010622024536, + "num_tokens": 764318194.0, + "step": 29089 + }, + { + "epoch": 2.398779582749237, + "grad_norm": 1.9238767623901367, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.7788532972335815, + "num_tokens": 764351953.0, + "step": 29090 + }, + { + "epoch": 2.398862043374289, + "grad_norm": 2.445573568344116, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.753048300743103, + "num_tokens": 764377567.0, + "step": 29091 + }, + { + "epoch": 2.3989445039993402, + "grad_norm": 2.261030435562134, + "learning_rate": 1e-06, + "loss": 0.6118, + "mean_token_accuracy": 0.791107177734375, + "num_tokens": 764402881.0, + "step": 29092 + }, + { + "epoch": 2.399026964624392, + "grad_norm": 2.0366718769073486, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.7890362739562988, + "num_tokens": 764433838.0, + "step": 29093 + }, + { + "epoch": 2.3991094252494434, + "grad_norm": 2.1013131141662598, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7594531774520874, + "num_tokens": 764466778.0, + "step": 29094 + }, + { + "epoch": 2.3991918858744947, + "grad_norm": 2.163783311843872, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7484787702560425, + "num_tokens": 764496234.0, + "step": 29095 + }, + { + "epoch": 2.3992743464995465, + "grad_norm": 2.234351396560669, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7257364988327026, + "num_tokens": 764523458.0, + "step": 29096 + }, + { + "epoch": 2.399356807124598, + "grad_norm": 1.9950517416000366, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7795847654342651, + "num_tokens": 764555796.0, + "step": 29097 + }, + { + "epoch": 2.3994392677496497, + "grad_norm": 2.4279744625091553, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7690483331680298, + "num_tokens": 764577691.0, + "step": 29098 + }, + { + "epoch": 2.399521728374701, + "grad_norm": 2.111394166946411, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7380478978157043, + "num_tokens": 764606604.0, + "step": 29099 + }, + { + "epoch": 2.399604188999753, + "grad_norm": 2.1336400508880615, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.783979058265686, + "num_tokens": 764633256.0, + "step": 29100 + }, + { + "epoch": 2.399686649624804, + "grad_norm": 2.237875461578369, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7676586508750916, + "num_tokens": 764661685.0, + "step": 29101 + }, + { + "epoch": 2.3997691102498555, + "grad_norm": 2.1262662410736084, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7608731985092163, + "num_tokens": 764690330.0, + "step": 29102 + }, + { + "epoch": 2.3998515708749073, + "grad_norm": 1.9859561920166016, + "learning_rate": 1e-06, + "loss": 0.572, + "mean_token_accuracy": 0.8070521354675293, + "num_tokens": 764726141.0, + "step": 29103 + }, + { + "epoch": 2.3999340314999587, + "grad_norm": 2.250030040740967, + "learning_rate": 1e-06, + "loss": 0.5939, + "mean_token_accuracy": 0.7988003492355347, + "num_tokens": 764752662.0, + "step": 29104 + }, + { + "epoch": 2.4000164921250104, + "grad_norm": 2.0921590328216553, + "learning_rate": 1e-06, + "loss": 0.6027, + "mean_token_accuracy": 0.7998164892196655, + "num_tokens": 764781628.0, + "step": 29105 + }, + { + "epoch": 2.400098952750062, + "grad_norm": 1.7890520095825195, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7474876642227173, + "num_tokens": 764819595.0, + "step": 29106 + }, + { + "epoch": 2.4001814133751136, + "grad_norm": 1.9794976711273193, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7804235816001892, + "num_tokens": 764850770.0, + "step": 29107 + }, + { + "epoch": 2.400263874000165, + "grad_norm": 1.8900701999664307, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.7897685766220093, + "num_tokens": 764884334.0, + "step": 29108 + }, + { + "epoch": 2.4003463346252163, + "grad_norm": 2.264143943786621, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7417995929718018, + "num_tokens": 764910248.0, + "step": 29109 + }, + { + "epoch": 2.400428795250268, + "grad_norm": 2.266085386276245, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7855291366577148, + "num_tokens": 764936108.0, + "step": 29110 + }, + { + "epoch": 2.4005112558753194, + "grad_norm": 2.855525016784668, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7552905678749084, + "num_tokens": 764971032.0, + "step": 29111 + }, + { + "epoch": 2.4005937165003712, + "grad_norm": 2.181711435317993, + "learning_rate": 1e-06, + "loss": 0.6487, + "mean_token_accuracy": 0.7891478538513184, + "num_tokens": 764997573.0, + "step": 29112 + }, + { + "epoch": 2.4006761771254226, + "grad_norm": 1.8167153596878052, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7617002129554749, + "num_tokens": 765035566.0, + "step": 29113 + }, + { + "epoch": 2.4007586377504744, + "grad_norm": 2.101595401763916, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7475416660308838, + "num_tokens": 765065677.0, + "step": 29114 + }, + { + "epoch": 2.4008410983755257, + "grad_norm": 2.2100460529327393, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.763115644454956, + "num_tokens": 765094490.0, + "step": 29115 + }, + { + "epoch": 2.400923559000577, + "grad_norm": 2.1125733852386475, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7418695688247681, + "num_tokens": 765125003.0, + "step": 29116 + }, + { + "epoch": 2.401006019625629, + "grad_norm": 2.190934896469116, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7624771595001221, + "num_tokens": 765153980.0, + "step": 29117 + }, + { + "epoch": 2.4010884802506802, + "grad_norm": 2.3793697357177734, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7519287467002869, + "num_tokens": 765177319.0, + "step": 29118 + }, + { + "epoch": 2.401170940875732, + "grad_norm": 2.292684555053711, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.7924505472183228, + "num_tokens": 765203458.0, + "step": 29119 + }, + { + "epoch": 2.4012534015007834, + "grad_norm": 2.039229154586792, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.7971251010894775, + "num_tokens": 765232476.0, + "step": 29120 + }, + { + "epoch": 2.401335862125835, + "grad_norm": 2.148805856704712, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7595103979110718, + "num_tokens": 765261902.0, + "step": 29121 + }, + { + "epoch": 2.4014183227508865, + "grad_norm": 2.418593168258667, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7793893814086914, + "num_tokens": 765286325.0, + "step": 29122 + }, + { + "epoch": 2.401500783375938, + "grad_norm": 2.1640491485595703, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.7905702590942383, + "num_tokens": 765313565.0, + "step": 29123 + }, + { + "epoch": 2.4015832440009897, + "grad_norm": 2.1894845962524414, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7711002826690674, + "num_tokens": 765343303.0, + "step": 29124 + }, + { + "epoch": 2.401665704626041, + "grad_norm": 1.9410420656204224, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7437252402305603, + "num_tokens": 765377120.0, + "step": 29125 + }, + { + "epoch": 2.4017481652510924, + "grad_norm": 2.0466489791870117, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.7965373396873474, + "num_tokens": 765409377.0, + "step": 29126 + }, + { + "epoch": 2.401830625876144, + "grad_norm": 1.978810429573059, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7524641752243042, + "num_tokens": 765443016.0, + "step": 29127 + }, + { + "epoch": 2.4019130865011955, + "grad_norm": 2.252276659011841, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7617629766464233, + "num_tokens": 765470726.0, + "step": 29128 + }, + { + "epoch": 2.4019955471262473, + "grad_norm": 2.0498125553131104, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7770611047744751, + "num_tokens": 765503485.0, + "step": 29129 + }, + { + "epoch": 2.4020780077512986, + "grad_norm": 2.167910575866699, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.7962158918380737, + "num_tokens": 765532414.0, + "step": 29130 + }, + { + "epoch": 2.4021604683763504, + "grad_norm": 2.2443854808807373, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7673630714416504, + "num_tokens": 765557518.0, + "step": 29131 + }, + { + "epoch": 2.402242929001402, + "grad_norm": 1.9838204383850098, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.7794891595840454, + "num_tokens": 765590226.0, + "step": 29132 + }, + { + "epoch": 2.402325389626453, + "grad_norm": 2.1458969116210938, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7586244940757751, + "num_tokens": 765619742.0, + "step": 29133 + }, + { + "epoch": 2.402407850251505, + "grad_norm": 1.9465091228485107, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.7944400310516357, + "num_tokens": 765653171.0, + "step": 29134 + }, + { + "epoch": 2.4024903108765563, + "grad_norm": 2.230877637863159, + "learning_rate": 1e-06, + "loss": 0.5663, + "mean_token_accuracy": 0.8056151270866394, + "num_tokens": 765679003.0, + "step": 29135 + }, + { + "epoch": 2.402572771501608, + "grad_norm": 2.1416776180267334, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.7948054075241089, + "num_tokens": 765705283.0, + "step": 29136 + }, + { + "epoch": 2.4026552321266594, + "grad_norm": 2.003633737564087, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7768169641494751, + "num_tokens": 765733513.0, + "step": 29137 + }, + { + "epoch": 2.4027376927517112, + "grad_norm": 1.835523247718811, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.7882009744644165, + "num_tokens": 765772299.0, + "step": 29138 + }, + { + "epoch": 2.4028201533767626, + "grad_norm": 1.9624255895614624, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7415305376052856, + "num_tokens": 765806208.0, + "step": 29139 + }, + { + "epoch": 2.402902614001814, + "grad_norm": 2.061624765396118, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.7946421504020691, + "num_tokens": 765835375.0, + "step": 29140 + }, + { + "epoch": 2.4029850746268657, + "grad_norm": 2.113316297531128, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.77753084897995, + "num_tokens": 765864474.0, + "step": 29141 + }, + { + "epoch": 2.403067535251917, + "grad_norm": 2.20031476020813, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7686964273452759, + "num_tokens": 765891715.0, + "step": 29142 + }, + { + "epoch": 2.403149995876969, + "grad_norm": 2.2009377479553223, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7302766442298889, + "num_tokens": 765923978.0, + "step": 29143 + }, + { + "epoch": 2.40323245650202, + "grad_norm": 1.9607044458389282, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7862739562988281, + "num_tokens": 765959746.0, + "step": 29144 + }, + { + "epoch": 2.403314917127072, + "grad_norm": 1.985232949256897, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7426979541778564, + "num_tokens": 765992645.0, + "step": 29145 + }, + { + "epoch": 2.4033973777521234, + "grad_norm": 2.4722440242767334, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7840195894241333, + "num_tokens": 766015481.0, + "step": 29146 + }, + { + "epoch": 2.4034798383771747, + "grad_norm": 2.2031922340393066, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.789078950881958, + "num_tokens": 766042048.0, + "step": 29147 + }, + { + "epoch": 2.4035622990022265, + "grad_norm": 2.0716869831085205, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.7892085313796997, + "num_tokens": 766071544.0, + "step": 29148 + }, + { + "epoch": 2.403644759627278, + "grad_norm": 2.1245367527008057, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.7982301712036133, + "num_tokens": 766100850.0, + "step": 29149 + }, + { + "epoch": 2.4037272202523297, + "grad_norm": 1.9588803052902222, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.779028058052063, + "num_tokens": 766133238.0, + "step": 29150 + }, + { + "epoch": 2.403809680877381, + "grad_norm": 2.246725082397461, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7256612181663513, + "num_tokens": 766163078.0, + "step": 29151 + }, + { + "epoch": 2.403892141502433, + "grad_norm": 2.271914482116699, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7513258457183838, + "num_tokens": 766188101.0, + "step": 29152 + }, + { + "epoch": 2.403974602127484, + "grad_norm": 1.9561312198638916, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7681392431259155, + "num_tokens": 766223314.0, + "step": 29153 + }, + { + "epoch": 2.4040570627525355, + "grad_norm": 2.283198118209839, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.7798728942871094, + "num_tokens": 766248669.0, + "step": 29154 + }, + { + "epoch": 2.4041395233775873, + "grad_norm": 2.1279239654541016, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7770897746086121, + "num_tokens": 766276956.0, + "step": 29155 + }, + { + "epoch": 2.4042219840026386, + "grad_norm": 2.177060127258301, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7690796256065369, + "num_tokens": 766306728.0, + "step": 29156 + }, + { + "epoch": 2.4043044446276904, + "grad_norm": 2.2955429553985596, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7562518119812012, + "num_tokens": 766336996.0, + "step": 29157 + }, + { + "epoch": 2.404386905252742, + "grad_norm": 1.8990906476974487, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7726335525512695, + "num_tokens": 766370381.0, + "step": 29158 + }, + { + "epoch": 2.4044693658777936, + "grad_norm": 2.012033224105835, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7622484564781189, + "num_tokens": 766401909.0, + "step": 29159 + }, + { + "epoch": 2.404551826502845, + "grad_norm": 2.156951427459717, + "learning_rate": 1e-06, + "loss": 0.6146, + "mean_token_accuracy": 0.7854036688804626, + "num_tokens": 766431404.0, + "step": 29160 + }, + { + "epoch": 2.4046342871278963, + "grad_norm": 1.7711362838745117, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7601414918899536, + "num_tokens": 766471424.0, + "step": 29161 + }, + { + "epoch": 2.404716747752948, + "grad_norm": 2.2645556926727295, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7502714991569519, + "num_tokens": 766501234.0, + "step": 29162 + }, + { + "epoch": 2.4047992083779994, + "grad_norm": 2.444051504135132, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7564209699630737, + "num_tokens": 766524967.0, + "step": 29163 + }, + { + "epoch": 2.4048816690030512, + "grad_norm": 2.087428569793701, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7451204061508179, + "num_tokens": 766556025.0, + "step": 29164 + }, + { + "epoch": 2.4049641296281026, + "grad_norm": 2.06380033493042, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7688039541244507, + "num_tokens": 766587843.0, + "step": 29165 + }, + { + "epoch": 2.405046590253154, + "grad_norm": 2.173388719558716, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7877830266952515, + "num_tokens": 766616392.0, + "step": 29166 + }, + { + "epoch": 2.4051290508782057, + "grad_norm": 1.7868493795394897, + "learning_rate": 1e-06, + "loss": 0.5719, + "mean_token_accuracy": 0.8144170641899109, + "num_tokens": 766652241.0, + "step": 29167 + }, + { + "epoch": 2.405211511503257, + "grad_norm": 2.2023632526397705, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7436121702194214, + "num_tokens": 766679919.0, + "step": 29168 + }, + { + "epoch": 2.405293972128309, + "grad_norm": 2.198112726211548, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.774528443813324, + "num_tokens": 766705770.0, + "step": 29169 + }, + { + "epoch": 2.40537643275336, + "grad_norm": 1.9868272542953491, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7452985048294067, + "num_tokens": 766741863.0, + "step": 29170 + }, + { + "epoch": 2.4054588933784116, + "grad_norm": 2.1080589294433594, + "learning_rate": 1e-06, + "loss": 0.5597, + "mean_token_accuracy": 0.8150575160980225, + "num_tokens": 766769229.0, + "step": 29171 + }, + { + "epoch": 2.4055413540034634, + "grad_norm": 2.172700881958008, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7729787826538086, + "num_tokens": 766796445.0, + "step": 29172 + }, + { + "epoch": 2.4056238146285147, + "grad_norm": 2.1783435344696045, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7701230645179749, + "num_tokens": 766824492.0, + "step": 29173 + }, + { + "epoch": 2.4057062752535665, + "grad_norm": 2.1903398036956787, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7651475071907043, + "num_tokens": 766851835.0, + "step": 29174 + }, + { + "epoch": 2.405788735878618, + "grad_norm": 2.4396636486053467, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7724540829658508, + "num_tokens": 766876617.0, + "step": 29175 + }, + { + "epoch": 2.4058711965036697, + "grad_norm": 2.0635182857513428, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.7876745462417603, + "num_tokens": 766906298.0, + "step": 29176 + }, + { + "epoch": 2.405953657128721, + "grad_norm": 1.9830031394958496, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7803442478179932, + "num_tokens": 766942437.0, + "step": 29177 + }, + { + "epoch": 2.4060361177537724, + "grad_norm": 2.2114365100860596, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7683786749839783, + "num_tokens": 766969608.0, + "step": 29178 + }, + { + "epoch": 2.406118578378824, + "grad_norm": 2.1706290245056152, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7630589008331299, + "num_tokens": 766998738.0, + "step": 29179 + }, + { + "epoch": 2.4062010390038755, + "grad_norm": 2.2051234245300293, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7768455743789673, + "num_tokens": 767026925.0, + "step": 29180 + }, + { + "epoch": 2.4062834996289273, + "grad_norm": 2.424853563308716, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7759445309638977, + "num_tokens": 767052280.0, + "step": 29181 + }, + { + "epoch": 2.4063659602539786, + "grad_norm": 2.0490610599517822, + "learning_rate": 1e-06, + "loss": 0.522, + "mean_token_accuracy": 0.8224034309387207, + "num_tokens": 767081591.0, + "step": 29182 + }, + { + "epoch": 2.4064484208790304, + "grad_norm": 2.230745315551758, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7648555636405945, + "num_tokens": 767109487.0, + "step": 29183 + }, + { + "epoch": 2.406530881504082, + "grad_norm": 2.0098910331726074, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7507481575012207, + "num_tokens": 767142928.0, + "step": 29184 + }, + { + "epoch": 2.406613342129133, + "grad_norm": 2.047243595123291, + "learning_rate": 1e-06, + "loss": 0.5918, + "mean_token_accuracy": 0.7990546226501465, + "num_tokens": 767173950.0, + "step": 29185 + }, + { + "epoch": 2.406695802754185, + "grad_norm": 2.426596164703369, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7669434547424316, + "num_tokens": 767198840.0, + "step": 29186 + }, + { + "epoch": 2.4067782633792363, + "grad_norm": 2.0630784034729004, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7516660690307617, + "num_tokens": 767229932.0, + "step": 29187 + }, + { + "epoch": 2.406860724004288, + "grad_norm": 2.0119094848632812, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7705712914466858, + "num_tokens": 767261040.0, + "step": 29188 + }, + { + "epoch": 2.4069431846293394, + "grad_norm": 2.065790891647339, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7722392082214355, + "num_tokens": 767291527.0, + "step": 29189 + }, + { + "epoch": 2.407025645254391, + "grad_norm": 2.067246198654175, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7670441269874573, + "num_tokens": 767322009.0, + "step": 29190 + }, + { + "epoch": 2.4071081058794426, + "grad_norm": 2.2421247959136963, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.7614916563034058, + "num_tokens": 767350464.0, + "step": 29191 + }, + { + "epoch": 2.407190566504494, + "grad_norm": 2.1967198848724365, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7736979126930237, + "num_tokens": 767378794.0, + "step": 29192 + }, + { + "epoch": 2.4072730271295457, + "grad_norm": 2.124013662338257, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7603024244308472, + "num_tokens": 767410018.0, + "step": 29193 + }, + { + "epoch": 2.407355487754597, + "grad_norm": 2.166752338409424, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7487292885780334, + "num_tokens": 767441396.0, + "step": 29194 + }, + { + "epoch": 2.407437948379649, + "grad_norm": 2.0948758125305176, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7685326337814331, + "num_tokens": 767470083.0, + "step": 29195 + }, + { + "epoch": 2.4075204090047, + "grad_norm": 2.340186834335327, + "learning_rate": 1e-06, + "loss": 0.7489, + "mean_token_accuracy": 0.7557149529457092, + "num_tokens": 767495766.0, + "step": 29196 + }, + { + "epoch": 2.407602869629752, + "grad_norm": 2.093550682067871, + "learning_rate": 1e-06, + "loss": 0.734, + "mean_token_accuracy": 0.7643930315971375, + "num_tokens": 767525742.0, + "step": 29197 + }, + { + "epoch": 2.4076853302548034, + "grad_norm": 2.1052286624908447, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.7851687073707581, + "num_tokens": 767555355.0, + "step": 29198 + }, + { + "epoch": 2.4077677908798547, + "grad_norm": 1.9796695709228516, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.7927188277244568, + "num_tokens": 767584180.0, + "step": 29199 + }, + { + "epoch": 2.4078502515049065, + "grad_norm": 2.0131113529205322, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7724738121032715, + "num_tokens": 767619903.0, + "step": 29200 + }, + { + "epoch": 2.407932712129958, + "grad_norm": 1.9937820434570312, + "learning_rate": 1e-06, + "loss": 0.5603, + "mean_token_accuracy": 0.8116023540496826, + "num_tokens": 767652412.0, + "step": 29201 + }, + { + "epoch": 2.4080151727550096, + "grad_norm": 2.3223891258239746, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7581123113632202, + "num_tokens": 767677650.0, + "step": 29202 + }, + { + "epoch": 2.408097633380061, + "grad_norm": 2.3536078929901123, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7788176536560059, + "num_tokens": 767704201.0, + "step": 29203 + }, + { + "epoch": 2.408180094005113, + "grad_norm": 2.0827929973602295, + "learning_rate": 1e-06, + "loss": 0.5588, + "mean_token_accuracy": 0.8034687638282776, + "num_tokens": 767733341.0, + "step": 29204 + }, + { + "epoch": 2.408262554630164, + "grad_norm": 1.947000503540039, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7757017016410828, + "num_tokens": 767765806.0, + "step": 29205 + }, + { + "epoch": 2.4083450152552155, + "grad_norm": 1.8755069971084595, + "learning_rate": 1e-06, + "loss": 0.5515, + "mean_token_accuracy": 0.8149427175521851, + "num_tokens": 767801715.0, + "step": 29206 + }, + { + "epoch": 2.4084274758802673, + "grad_norm": 2.279207468032837, + "learning_rate": 1e-06, + "loss": 0.7688, + "mean_token_accuracy": 0.752332866191864, + "num_tokens": 767829503.0, + "step": 29207 + }, + { + "epoch": 2.4085099365053186, + "grad_norm": 2.2806265354156494, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7228105068206787, + "num_tokens": 767857139.0, + "step": 29208 + }, + { + "epoch": 2.4085923971303704, + "grad_norm": 2.1274032592773438, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7370380163192749, + "num_tokens": 767890948.0, + "step": 29209 + }, + { + "epoch": 2.408674857755422, + "grad_norm": 2.073330879211426, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.7838268280029297, + "num_tokens": 767919710.0, + "step": 29210 + }, + { + "epoch": 2.408757318380473, + "grad_norm": 2.094682455062866, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7606141567230225, + "num_tokens": 767948916.0, + "step": 29211 + }, + { + "epoch": 2.408839779005525, + "grad_norm": 2.384115695953369, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.745815634727478, + "num_tokens": 767972122.0, + "step": 29212 + }, + { + "epoch": 2.4089222396305763, + "grad_norm": 2.4942519664764404, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7686082124710083, + "num_tokens": 767994609.0, + "step": 29213 + }, + { + "epoch": 2.409004700255628, + "grad_norm": 2.2185232639312744, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7682219743728638, + "num_tokens": 768020738.0, + "step": 29214 + }, + { + "epoch": 2.4090871608806794, + "grad_norm": 2.2480151653289795, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7879178524017334, + "num_tokens": 768046324.0, + "step": 29215 + }, + { + "epoch": 2.4091696215057308, + "grad_norm": 2.08644700050354, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7753083109855652, + "num_tokens": 768075323.0, + "step": 29216 + }, + { + "epoch": 2.4092520821307826, + "grad_norm": 2.324885368347168, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7575610876083374, + "num_tokens": 768099959.0, + "step": 29217 + }, + { + "epoch": 2.409334542755834, + "grad_norm": 2.0666661262512207, + "learning_rate": 1e-06, + "loss": 0.5601, + "mean_token_accuracy": 0.8072099089622498, + "num_tokens": 768127282.0, + "step": 29218 + }, + { + "epoch": 2.4094170033808857, + "grad_norm": 2.263897657394409, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7590073347091675, + "num_tokens": 768157609.0, + "step": 29219 + }, + { + "epoch": 2.409499464005937, + "grad_norm": 2.3128886222839355, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.8000330924987793, + "num_tokens": 768185842.0, + "step": 29220 + }, + { + "epoch": 2.409581924630989, + "grad_norm": 1.883792757987976, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.7926864624023438, + "num_tokens": 768219379.0, + "step": 29221 + }, + { + "epoch": 2.40966438525604, + "grad_norm": 1.997755765914917, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7729784250259399, + "num_tokens": 768254474.0, + "step": 29222 + }, + { + "epoch": 2.4097468458810916, + "grad_norm": 1.9547325372695923, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7743901014328003, + "num_tokens": 768286048.0, + "step": 29223 + }, + { + "epoch": 2.4098293065061434, + "grad_norm": 2.096219301223755, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7699973583221436, + "num_tokens": 768315876.0, + "step": 29224 + }, + { + "epoch": 2.4099117671311947, + "grad_norm": 2.0826430320739746, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7512381672859192, + "num_tokens": 768352971.0, + "step": 29225 + }, + { + "epoch": 2.4099942277562465, + "grad_norm": 2.066816806793213, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7651566863059998, + "num_tokens": 768383882.0, + "step": 29226 + }, + { + "epoch": 2.410076688381298, + "grad_norm": 2.1802146434783936, + "learning_rate": 1e-06, + "loss": 0.6094, + "mean_token_accuracy": 0.7858656644821167, + "num_tokens": 768410686.0, + "step": 29227 + }, + { + "epoch": 2.4101591490063496, + "grad_norm": 2.1037724018096924, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7270202040672302, + "num_tokens": 768440819.0, + "step": 29228 + }, + { + "epoch": 2.410241609631401, + "grad_norm": 2.0795159339904785, + "learning_rate": 1e-06, + "loss": 0.6094, + "mean_token_accuracy": 0.7924140691757202, + "num_tokens": 768469876.0, + "step": 29229 + }, + { + "epoch": 2.4103240702564523, + "grad_norm": 1.9806503057479858, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7801333069801331, + "num_tokens": 768504819.0, + "step": 29230 + }, + { + "epoch": 2.410406530881504, + "grad_norm": 2.3407697677612305, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.7940146923065186, + "num_tokens": 768529956.0, + "step": 29231 + }, + { + "epoch": 2.4104889915065555, + "grad_norm": 2.1176605224609375, + "learning_rate": 1e-06, + "loss": 0.6389, + "mean_token_accuracy": 0.7857218980789185, + "num_tokens": 768557837.0, + "step": 29232 + }, + { + "epoch": 2.4105714521316073, + "grad_norm": 1.914738416671753, + "learning_rate": 1e-06, + "loss": 0.5569, + "mean_token_accuracy": 0.8087587356567383, + "num_tokens": 768592401.0, + "step": 29233 + }, + { + "epoch": 2.4106539127566586, + "grad_norm": 2.4706478118896484, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7622251510620117, + "num_tokens": 768617151.0, + "step": 29234 + }, + { + "epoch": 2.4107363733817104, + "grad_norm": 1.8639767169952393, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7419908046722412, + "num_tokens": 768653932.0, + "step": 29235 + }, + { + "epoch": 2.410818834006762, + "grad_norm": 1.9846192598342896, + "learning_rate": 1e-06, + "loss": 0.5798, + "mean_token_accuracy": 0.8069671392440796, + "num_tokens": 768687979.0, + "step": 29236 + }, + { + "epoch": 2.410901294631813, + "grad_norm": 2.200533390045166, + "learning_rate": 1e-06, + "loss": 0.5698, + "mean_token_accuracy": 0.8065578937530518, + "num_tokens": 768715574.0, + "step": 29237 + }, + { + "epoch": 2.410983755256865, + "grad_norm": 2.145690441131592, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7561217546463013, + "num_tokens": 768743704.0, + "step": 29238 + }, + { + "epoch": 2.4110662158819163, + "grad_norm": 1.9282785654067993, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.766315221786499, + "num_tokens": 768776469.0, + "step": 29239 + }, + { + "epoch": 2.411148676506968, + "grad_norm": 2.0478808879852295, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7586499452590942, + "num_tokens": 768809143.0, + "step": 29240 + }, + { + "epoch": 2.4112311371320194, + "grad_norm": 2.1260223388671875, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7744022607803345, + "num_tokens": 768840100.0, + "step": 29241 + }, + { + "epoch": 2.411313597757071, + "grad_norm": 1.8798587322235107, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.7958691120147705, + "num_tokens": 768875020.0, + "step": 29242 + }, + { + "epoch": 2.4113960583821226, + "grad_norm": 2.2527248859405518, + "learning_rate": 1e-06, + "loss": 0.5687, + "mean_token_accuracy": 0.7979316115379333, + "num_tokens": 768899126.0, + "step": 29243 + }, + { + "epoch": 2.411478519007174, + "grad_norm": 2.1171634197235107, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7835585474967957, + "num_tokens": 768931595.0, + "step": 29244 + }, + { + "epoch": 2.4115609796322257, + "grad_norm": 2.0743765830993652, + "learning_rate": 1e-06, + "loss": 0.6257, + "mean_token_accuracy": 0.7936796545982361, + "num_tokens": 768966212.0, + "step": 29245 + }, + { + "epoch": 2.411643440257277, + "grad_norm": 2.4166910648345947, + "learning_rate": 1e-06, + "loss": 0.5665, + "mean_token_accuracy": 0.8015030026435852, + "num_tokens": 768990831.0, + "step": 29246 + }, + { + "epoch": 2.411725900882329, + "grad_norm": 2.2541656494140625, + "learning_rate": 1e-06, + "loss": 0.7585, + "mean_token_accuracy": 0.7532778978347778, + "num_tokens": 769016613.0, + "step": 29247 + }, + { + "epoch": 2.41180836150738, + "grad_norm": 2.0839388370513916, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7548745274543762, + "num_tokens": 769047630.0, + "step": 29248 + }, + { + "epoch": 2.411890822132432, + "grad_norm": 2.330185651779175, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7765260338783264, + "num_tokens": 769074319.0, + "step": 29249 + }, + { + "epoch": 2.4119732827574834, + "grad_norm": 1.7849215269088745, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.7719008922576904, + "num_tokens": 769110832.0, + "step": 29250 + }, + { + "epoch": 2.4120557433825347, + "grad_norm": 2.0027315616607666, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.7787045836448669, + "num_tokens": 769145087.0, + "step": 29251 + }, + { + "epoch": 2.4121382040075865, + "grad_norm": 1.9546713829040527, + "learning_rate": 1e-06, + "loss": 0.7284, + "mean_token_accuracy": 0.7744137644767761, + "num_tokens": 769180800.0, + "step": 29252 + }, + { + "epoch": 2.412220664632638, + "grad_norm": 2.0588455200195312, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.774271547794342, + "num_tokens": 769212006.0, + "step": 29253 + }, + { + "epoch": 2.4123031252576896, + "grad_norm": 2.3629887104034424, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.7930222749710083, + "num_tokens": 769239418.0, + "step": 29254 + }, + { + "epoch": 2.412385585882741, + "grad_norm": 2.0229148864746094, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7303512692451477, + "num_tokens": 769275382.0, + "step": 29255 + }, + { + "epoch": 2.4124680465077923, + "grad_norm": 2.229126453399658, + "learning_rate": 1e-06, + "loss": 0.5444, + "mean_token_accuracy": 0.8054318428039551, + "num_tokens": 769301478.0, + "step": 29256 + }, + { + "epoch": 2.412550507132844, + "grad_norm": 2.119809627532959, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.758571445941925, + "num_tokens": 769329507.0, + "step": 29257 + }, + { + "epoch": 2.4126329677578955, + "grad_norm": 1.9736566543579102, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7635482549667358, + "num_tokens": 769363729.0, + "step": 29258 + }, + { + "epoch": 2.4127154283829473, + "grad_norm": 1.8421849012374878, + "learning_rate": 1e-06, + "loss": 0.6871, + "mean_token_accuracy": 0.7712243795394897, + "num_tokens": 769401108.0, + "step": 29259 + }, + { + "epoch": 2.4127978890079986, + "grad_norm": 1.903346300125122, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7766131162643433, + "num_tokens": 769433737.0, + "step": 29260 + }, + { + "epoch": 2.41288034963305, + "grad_norm": 2.3299336433410645, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7776057720184326, + "num_tokens": 769462053.0, + "step": 29261 + }, + { + "epoch": 2.4129628102581018, + "grad_norm": 2.2310545444488525, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7661597728729248, + "num_tokens": 769487687.0, + "step": 29262 + }, + { + "epoch": 2.413045270883153, + "grad_norm": 1.9922616481781006, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7644731998443604, + "num_tokens": 769520227.0, + "step": 29263 + }, + { + "epoch": 2.413127731508205, + "grad_norm": 2.221789836883545, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7619439363479614, + "num_tokens": 769548236.0, + "step": 29264 + }, + { + "epoch": 2.4132101921332563, + "grad_norm": 2.0684943199157715, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.772200345993042, + "num_tokens": 769577440.0, + "step": 29265 + }, + { + "epoch": 2.413292652758308, + "grad_norm": 2.143324613571167, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.770022451877594, + "num_tokens": 769608552.0, + "step": 29266 + }, + { + "epoch": 2.4133751133833594, + "grad_norm": 2.3967347145080566, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7345348000526428, + "num_tokens": 769634390.0, + "step": 29267 + }, + { + "epoch": 2.4134575740084108, + "grad_norm": 1.9666787385940552, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7812577486038208, + "num_tokens": 769669653.0, + "step": 29268 + }, + { + "epoch": 2.4135400346334626, + "grad_norm": 2.500946521759033, + "learning_rate": 1e-06, + "loss": 0.6526, + "mean_token_accuracy": 0.7911766767501831, + "num_tokens": 769692824.0, + "step": 29269 + }, + { + "epoch": 2.413622495258514, + "grad_norm": 1.9183330535888672, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.780072808265686, + "num_tokens": 769730057.0, + "step": 29270 + }, + { + "epoch": 2.4137049558835657, + "grad_norm": 2.0591607093811035, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7446720600128174, + "num_tokens": 769762701.0, + "step": 29271 + }, + { + "epoch": 2.413787416508617, + "grad_norm": 1.9649690389633179, + "learning_rate": 1e-06, + "loss": 0.5227, + "mean_token_accuracy": 0.8201653361320496, + "num_tokens": 769793340.0, + "step": 29272 + }, + { + "epoch": 2.413869877133669, + "grad_norm": 1.8794513940811157, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.7822714447975159, + "num_tokens": 769826840.0, + "step": 29273 + }, + { + "epoch": 2.41395233775872, + "grad_norm": 1.935837745666504, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7817992568016052, + "num_tokens": 769861217.0, + "step": 29274 + }, + { + "epoch": 2.4140347983837716, + "grad_norm": 2.07083797454834, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.720077633857727, + "num_tokens": 769893625.0, + "step": 29275 + }, + { + "epoch": 2.4141172590088233, + "grad_norm": 1.9612289667129517, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7476021647453308, + "num_tokens": 769929626.0, + "step": 29276 + }, + { + "epoch": 2.4141997196338747, + "grad_norm": 1.965229868888855, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.7926748991012573, + "num_tokens": 769963406.0, + "step": 29277 + }, + { + "epoch": 2.4142821802589265, + "grad_norm": 2.0605103969573975, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7729905247688293, + "num_tokens": 769996325.0, + "step": 29278 + }, + { + "epoch": 2.414364640883978, + "grad_norm": 2.073732852935791, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7438003420829773, + "num_tokens": 770027842.0, + "step": 29279 + }, + { + "epoch": 2.4144471015090296, + "grad_norm": 1.788916826248169, + "learning_rate": 1e-06, + "loss": 0.5191, + "mean_token_accuracy": 0.8290205597877502, + "num_tokens": 770062903.0, + "step": 29280 + }, + { + "epoch": 2.414529562134081, + "grad_norm": 2.262883186340332, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7786498069763184, + "num_tokens": 770090496.0, + "step": 29281 + }, + { + "epoch": 2.4146120227591323, + "grad_norm": 1.8264046907424927, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.7699506282806396, + "num_tokens": 770129787.0, + "step": 29282 + }, + { + "epoch": 2.414694483384184, + "grad_norm": 1.8362146615982056, + "learning_rate": 1e-06, + "loss": 0.6572, + "mean_token_accuracy": 0.78769850730896, + "num_tokens": 770169873.0, + "step": 29283 + }, + { + "epoch": 2.4147769440092355, + "grad_norm": 2.1450397968292236, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7879822254180908, + "num_tokens": 770199505.0, + "step": 29284 + }, + { + "epoch": 2.4148594046342873, + "grad_norm": 2.2672953605651855, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7618448734283447, + "num_tokens": 770226274.0, + "step": 29285 + }, + { + "epoch": 2.4149418652593386, + "grad_norm": 2.2602028846740723, + "learning_rate": 1e-06, + "loss": 0.783, + "mean_token_accuracy": 0.747981071472168, + "num_tokens": 770255014.0, + "step": 29286 + }, + { + "epoch": 2.4150243258843904, + "grad_norm": 2.0098800659179688, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7536640763282776, + "num_tokens": 770288778.0, + "step": 29287 + }, + { + "epoch": 2.4151067865094418, + "grad_norm": 2.346126079559326, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7705342173576355, + "num_tokens": 770314092.0, + "step": 29288 + }, + { + "epoch": 2.415189247134493, + "grad_norm": 1.850498914718628, + "learning_rate": 1e-06, + "loss": 0.5836, + "mean_token_accuracy": 0.8080088496208191, + "num_tokens": 770351037.0, + "step": 29289 + }, + { + "epoch": 2.415271707759545, + "grad_norm": 1.918613314628601, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7613509893417358, + "num_tokens": 770386906.0, + "step": 29290 + }, + { + "epoch": 2.4153541683845963, + "grad_norm": 2.4423112869262695, + "learning_rate": 1e-06, + "loss": 0.7966, + "mean_token_accuracy": 0.7491179704666138, + "num_tokens": 770411039.0, + "step": 29291 + }, + { + "epoch": 2.415436629009648, + "grad_norm": 1.8060394525527954, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7872201204299927, + "num_tokens": 770448538.0, + "step": 29292 + }, + { + "epoch": 2.4155190896346994, + "grad_norm": 2.2960240840911865, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7609840631484985, + "num_tokens": 770473315.0, + "step": 29293 + }, + { + "epoch": 2.415601550259751, + "grad_norm": 2.304356336593628, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7326273322105408, + "num_tokens": 770499683.0, + "step": 29294 + }, + { + "epoch": 2.4156840108848026, + "grad_norm": 2.375397205352783, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7643065452575684, + "num_tokens": 770527027.0, + "step": 29295 + }, + { + "epoch": 2.415766471509854, + "grad_norm": 2.0608067512512207, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7726565599441528, + "num_tokens": 770560964.0, + "step": 29296 + }, + { + "epoch": 2.4158489321349057, + "grad_norm": 2.259901285171509, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7387568354606628, + "num_tokens": 770587786.0, + "step": 29297 + }, + { + "epoch": 2.415931392759957, + "grad_norm": 2.131037950515747, + "learning_rate": 1e-06, + "loss": 0.7394, + "mean_token_accuracy": 0.7624657154083252, + "num_tokens": 770616330.0, + "step": 29298 + }, + { + "epoch": 2.416013853385009, + "grad_norm": 2.347771406173706, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.760297417640686, + "num_tokens": 770641271.0, + "step": 29299 + }, + { + "epoch": 2.41609631401006, + "grad_norm": 2.018071413040161, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7505204677581787, + "num_tokens": 770671296.0, + "step": 29300 + }, + { + "epoch": 2.4161787746351115, + "grad_norm": 2.158271074295044, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.7920255064964294, + "num_tokens": 770699303.0, + "step": 29301 + }, + { + "epoch": 2.4162612352601633, + "grad_norm": 2.2715539932250977, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7755153775215149, + "num_tokens": 770724660.0, + "step": 29302 + }, + { + "epoch": 2.4163436958852147, + "grad_norm": 2.177335500717163, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7786412239074707, + "num_tokens": 770754369.0, + "step": 29303 + }, + { + "epoch": 2.4164261565102665, + "grad_norm": 2.0905308723449707, + "learning_rate": 1e-06, + "loss": 0.5891, + "mean_token_accuracy": 0.8036046028137207, + "num_tokens": 770784936.0, + "step": 29304 + }, + { + "epoch": 2.416508617135318, + "grad_norm": 2.024254083633423, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7839493155479431, + "num_tokens": 770816584.0, + "step": 29305 + }, + { + "epoch": 2.416591077760369, + "grad_norm": 1.85306978225708, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7669681310653687, + "num_tokens": 770855953.0, + "step": 29306 + }, + { + "epoch": 2.416673538385421, + "grad_norm": 2.3016467094421387, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7683462500572205, + "num_tokens": 770884907.0, + "step": 29307 + }, + { + "epoch": 2.4167559990104723, + "grad_norm": 1.964951515197754, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7805403470993042, + "num_tokens": 770917658.0, + "step": 29308 + }, + { + "epoch": 2.416838459635524, + "grad_norm": 2.1860897541046143, + "learning_rate": 1e-06, + "loss": 0.6158, + "mean_token_accuracy": 0.7893537878990173, + "num_tokens": 770942177.0, + "step": 29309 + }, + { + "epoch": 2.4169209202605755, + "grad_norm": 1.9939039945602417, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7580348253250122, + "num_tokens": 770976357.0, + "step": 29310 + }, + { + "epoch": 2.4170033808856273, + "grad_norm": 1.887759804725647, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.774349570274353, + "num_tokens": 771013667.0, + "step": 29311 + }, + { + "epoch": 2.4170858415106786, + "grad_norm": 2.181767463684082, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7719528675079346, + "num_tokens": 771041920.0, + "step": 29312 + }, + { + "epoch": 2.41716830213573, + "grad_norm": 2.3234708309173584, + "learning_rate": 1e-06, + "loss": 0.6099, + "mean_token_accuracy": 0.7981782555580139, + "num_tokens": 771066018.0, + "step": 29313 + }, + { + "epoch": 2.4172507627607818, + "grad_norm": 2.062716007232666, + "learning_rate": 1e-06, + "loss": 0.6127, + "mean_token_accuracy": 0.8014182448387146, + "num_tokens": 771094732.0, + "step": 29314 + }, + { + "epoch": 2.417333223385833, + "grad_norm": 1.917512059211731, + "learning_rate": 1e-06, + "loss": 0.5785, + "mean_token_accuracy": 0.8045485615730286, + "num_tokens": 771127648.0, + "step": 29315 + }, + { + "epoch": 2.417415684010885, + "grad_norm": 2.371779680252075, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7443835735321045, + "num_tokens": 771155301.0, + "step": 29316 + }, + { + "epoch": 2.4174981446359363, + "grad_norm": 2.1921322345733643, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.7837055325508118, + "num_tokens": 771183127.0, + "step": 29317 + }, + { + "epoch": 2.417580605260988, + "grad_norm": 2.0512921810150146, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.743093729019165, + "num_tokens": 771215317.0, + "step": 29318 + }, + { + "epoch": 2.4176630658860394, + "grad_norm": 2.0007503032684326, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7642269730567932, + "num_tokens": 771247002.0, + "step": 29319 + }, + { + "epoch": 2.4177455265110908, + "grad_norm": 2.1930339336395264, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7389969229698181, + "num_tokens": 771275663.0, + "step": 29320 + }, + { + "epoch": 2.4178279871361426, + "grad_norm": 2.0044689178466797, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7696692943572998, + "num_tokens": 771310997.0, + "step": 29321 + }, + { + "epoch": 2.417910447761194, + "grad_norm": 2.2002694606781006, + "learning_rate": 1e-06, + "loss": 0.5834, + "mean_token_accuracy": 0.8089110851287842, + "num_tokens": 771337764.0, + "step": 29322 + }, + { + "epoch": 2.4179929083862457, + "grad_norm": 2.0358152389526367, + "learning_rate": 1e-06, + "loss": 0.8003, + "mean_token_accuracy": 0.742856502532959, + "num_tokens": 771372742.0, + "step": 29323 + }, + { + "epoch": 2.418075369011297, + "grad_norm": 2.328300714492798, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7607780694961548, + "num_tokens": 771405980.0, + "step": 29324 + }, + { + "epoch": 2.418157829636349, + "grad_norm": 2.2382962703704834, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7568966150283813, + "num_tokens": 771432633.0, + "step": 29325 + }, + { + "epoch": 2.4182402902614, + "grad_norm": 1.7863777875900269, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.7965694665908813, + "num_tokens": 771467785.0, + "step": 29326 + }, + { + "epoch": 2.4183227508864515, + "grad_norm": 1.9195730686187744, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.792970597743988, + "num_tokens": 771501934.0, + "step": 29327 + }, + { + "epoch": 2.4184052115115033, + "grad_norm": 2.204127788543701, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7496929168701172, + "num_tokens": 771528570.0, + "step": 29328 + }, + { + "epoch": 2.4184876721365547, + "grad_norm": 1.9861241579055786, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.797932505607605, + "num_tokens": 771559947.0, + "step": 29329 + }, + { + "epoch": 2.4185701327616065, + "grad_norm": 1.9483466148376465, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7470334768295288, + "num_tokens": 771594884.0, + "step": 29330 + }, + { + "epoch": 2.418652593386658, + "grad_norm": 2.175762891769409, + "learning_rate": 1e-06, + "loss": 0.5396, + "mean_token_accuracy": 0.8166460394859314, + "num_tokens": 771623007.0, + "step": 29331 + }, + { + "epoch": 2.4187350540117096, + "grad_norm": 2.3560893535614014, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7464727163314819, + "num_tokens": 771647975.0, + "step": 29332 + }, + { + "epoch": 2.418817514636761, + "grad_norm": 2.0491020679473877, + "learning_rate": 1e-06, + "loss": 0.6017, + "mean_token_accuracy": 0.7970088124275208, + "num_tokens": 771679333.0, + "step": 29333 + }, + { + "epoch": 2.4188999752618123, + "grad_norm": 1.905772089958191, + "learning_rate": 1e-06, + "loss": 0.4917, + "mean_token_accuracy": 0.832135021686554, + "num_tokens": 771711300.0, + "step": 29334 + }, + { + "epoch": 2.418982435886864, + "grad_norm": 2.265723705291748, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7657977938652039, + "num_tokens": 771739794.0, + "step": 29335 + }, + { + "epoch": 2.4190648965119155, + "grad_norm": 1.938833236694336, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7555197477340698, + "num_tokens": 771775533.0, + "step": 29336 + }, + { + "epoch": 2.4191473571369673, + "grad_norm": 2.014861583709717, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7684839963912964, + "num_tokens": 771809481.0, + "step": 29337 + }, + { + "epoch": 2.4192298177620186, + "grad_norm": 1.963049292564392, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7673690319061279, + "num_tokens": 771842467.0, + "step": 29338 + }, + { + "epoch": 2.4193122783870704, + "grad_norm": 2.191753625869751, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7501497864723206, + "num_tokens": 771873148.0, + "step": 29339 + }, + { + "epoch": 2.4193947390121218, + "grad_norm": 2.012169361114502, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7695638537406921, + "num_tokens": 771909046.0, + "step": 29340 + }, + { + "epoch": 2.419477199637173, + "grad_norm": 2.1089959144592285, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7423171997070312, + "num_tokens": 771940046.0, + "step": 29341 + }, + { + "epoch": 2.419559660262225, + "grad_norm": 1.9570646286010742, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7661192417144775, + "num_tokens": 771973210.0, + "step": 29342 + }, + { + "epoch": 2.4196421208872763, + "grad_norm": 1.9828746318817139, + "learning_rate": 1e-06, + "loss": 0.5932, + "mean_token_accuracy": 0.798710823059082, + "num_tokens": 772004748.0, + "step": 29343 + }, + { + "epoch": 2.419724581512328, + "grad_norm": 2.3983607292175293, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7780283093452454, + "num_tokens": 772030903.0, + "step": 29344 + }, + { + "epoch": 2.4198070421373794, + "grad_norm": 1.9938565492630005, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.762546718120575, + "num_tokens": 772065325.0, + "step": 29345 + }, + { + "epoch": 2.4198895027624308, + "grad_norm": 1.9365897178649902, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.7978829145431519, + "num_tokens": 772103750.0, + "step": 29346 + }, + { + "epoch": 2.4199719633874825, + "grad_norm": 2.161743402481079, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7695279121398926, + "num_tokens": 772131389.0, + "step": 29347 + }, + { + "epoch": 2.420054424012534, + "grad_norm": 2.43233060836792, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7142748832702637, + "num_tokens": 772156852.0, + "step": 29348 + }, + { + "epoch": 2.4201368846375857, + "grad_norm": 2.4008126258850098, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7911192774772644, + "num_tokens": 772181453.0, + "step": 29349 + }, + { + "epoch": 2.420219345262637, + "grad_norm": 2.050886631011963, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7284104824066162, + "num_tokens": 772216308.0, + "step": 29350 + }, + { + "epoch": 2.4203018058876884, + "grad_norm": 2.2456588745117188, + "learning_rate": 1e-06, + "loss": 0.5607, + "mean_token_accuracy": 0.7979824542999268, + "num_tokens": 772242486.0, + "step": 29351 + }, + { + "epoch": 2.42038426651274, + "grad_norm": 2.0558416843414307, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.7974041700363159, + "num_tokens": 772272846.0, + "step": 29352 + }, + { + "epoch": 2.4204667271377915, + "grad_norm": 1.8645023107528687, + "learning_rate": 1e-06, + "loss": 0.6486, + "mean_token_accuracy": 0.7862383127212524, + "num_tokens": 772305043.0, + "step": 29353 + }, + { + "epoch": 2.4205491877628433, + "grad_norm": 2.168766975402832, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.7815231680870056, + "num_tokens": 772335031.0, + "step": 29354 + }, + { + "epoch": 2.4206316483878947, + "grad_norm": 2.0039896965026855, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7704149484634399, + "num_tokens": 772365205.0, + "step": 29355 + }, + { + "epoch": 2.4207141090129465, + "grad_norm": 1.79314386844635, + "learning_rate": 1e-06, + "loss": 0.5782, + "mean_token_accuracy": 0.8079067468643188, + "num_tokens": 772402193.0, + "step": 29356 + }, + { + "epoch": 2.420796569637998, + "grad_norm": 2.037219762802124, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7742487788200378, + "num_tokens": 772435030.0, + "step": 29357 + }, + { + "epoch": 2.420879030263049, + "grad_norm": 2.535922050476074, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7383176684379578, + "num_tokens": 772456266.0, + "step": 29358 + }, + { + "epoch": 2.420961490888101, + "grad_norm": 2.3121628761291504, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7600297927856445, + "num_tokens": 772483640.0, + "step": 29359 + }, + { + "epoch": 2.4210439515131523, + "grad_norm": 2.278801441192627, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7454606294631958, + "num_tokens": 772511025.0, + "step": 29360 + }, + { + "epoch": 2.421126412138204, + "grad_norm": 2.6160354614257812, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7369991540908813, + "num_tokens": 772533544.0, + "step": 29361 + }, + { + "epoch": 2.4212088727632555, + "grad_norm": 2.013753652572632, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7731319665908813, + "num_tokens": 772568844.0, + "step": 29362 + }, + { + "epoch": 2.4212913333883073, + "grad_norm": 2.0203280448913574, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7409487962722778, + "num_tokens": 772601957.0, + "step": 29363 + }, + { + "epoch": 2.4213737940133586, + "grad_norm": 2.291057586669922, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7590314745903015, + "num_tokens": 772629211.0, + "step": 29364 + }, + { + "epoch": 2.42145625463841, + "grad_norm": 2.1318676471710205, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7437269687652588, + "num_tokens": 772658170.0, + "step": 29365 + }, + { + "epoch": 2.4215387152634618, + "grad_norm": 2.1589090824127197, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7177860736846924, + "num_tokens": 772689443.0, + "step": 29366 + }, + { + "epoch": 2.421621175888513, + "grad_norm": 2.0147016048431396, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7650356292724609, + "num_tokens": 772720559.0, + "step": 29367 + }, + { + "epoch": 2.421703636513565, + "grad_norm": 2.1255762577056885, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7352092266082764, + "num_tokens": 772754035.0, + "step": 29368 + }, + { + "epoch": 2.4217860971386163, + "grad_norm": 1.9429898262023926, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.7938080430030823, + "num_tokens": 772785022.0, + "step": 29369 + }, + { + "epoch": 2.421868557763668, + "grad_norm": 1.9838260412216187, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.7963172197341919, + "num_tokens": 772813605.0, + "step": 29370 + }, + { + "epoch": 2.4219510183887194, + "grad_norm": 2.1081383228302, + "learning_rate": 1e-06, + "loss": 0.5968, + "mean_token_accuracy": 0.8039906024932861, + "num_tokens": 772839754.0, + "step": 29371 + }, + { + "epoch": 2.4220334790137708, + "grad_norm": 2.0489399433135986, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7555251121520996, + "num_tokens": 772873472.0, + "step": 29372 + }, + { + "epoch": 2.4221159396388225, + "grad_norm": 2.134331226348877, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7514021992683411, + "num_tokens": 772901452.0, + "step": 29373 + }, + { + "epoch": 2.422198400263874, + "grad_norm": 1.8213611841201782, + "learning_rate": 1e-06, + "loss": 0.5677, + "mean_token_accuracy": 0.80863356590271, + "num_tokens": 772936790.0, + "step": 29374 + }, + { + "epoch": 2.4222808608889257, + "grad_norm": 1.8961557149887085, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.7740761041641235, + "num_tokens": 772967862.0, + "step": 29375 + }, + { + "epoch": 2.422363321513977, + "grad_norm": 2.4097235202789307, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.784524142742157, + "num_tokens": 772993207.0, + "step": 29376 + }, + { + "epoch": 2.422445782139029, + "grad_norm": 2.0631110668182373, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.7811708450317383, + "num_tokens": 773024866.0, + "step": 29377 + }, + { + "epoch": 2.42252824276408, + "grad_norm": 2.2460057735443115, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.784466028213501, + "num_tokens": 773050179.0, + "step": 29378 + }, + { + "epoch": 2.4226107033891315, + "grad_norm": 2.2003021240234375, + "learning_rate": 1e-06, + "loss": 0.6101, + "mean_token_accuracy": 0.780596137046814, + "num_tokens": 773078651.0, + "step": 29379 + }, + { + "epoch": 2.4226931640141833, + "grad_norm": 1.9721029996871948, + "learning_rate": 1e-06, + "loss": 0.8399, + "mean_token_accuracy": 0.7324614524841309, + "num_tokens": 773114955.0, + "step": 29380 + }, + { + "epoch": 2.4227756246392347, + "grad_norm": 1.8587210178375244, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7655191421508789, + "num_tokens": 773154593.0, + "step": 29381 + }, + { + "epoch": 2.4228580852642865, + "grad_norm": 1.7616159915924072, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7865763306617737, + "num_tokens": 773193113.0, + "step": 29382 + }, + { + "epoch": 2.422940545889338, + "grad_norm": 2.4474027156829834, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.7920440435409546, + "num_tokens": 773216997.0, + "step": 29383 + }, + { + "epoch": 2.4230230065143896, + "grad_norm": 2.0709919929504395, + "learning_rate": 1e-06, + "loss": 0.549, + "mean_token_accuracy": 0.8151368498802185, + "num_tokens": 773244781.0, + "step": 29384 + }, + { + "epoch": 2.423105467139441, + "grad_norm": 2.0747523307800293, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.7858420610427856, + "num_tokens": 773274739.0, + "step": 29385 + }, + { + "epoch": 2.4231879277644923, + "grad_norm": 2.191086769104004, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.7854365110397339, + "num_tokens": 773314256.0, + "step": 29386 + }, + { + "epoch": 2.423270388389544, + "grad_norm": 2.0606892108917236, + "learning_rate": 1e-06, + "loss": 0.5983, + "mean_token_accuracy": 0.8088328838348389, + "num_tokens": 773345194.0, + "step": 29387 + }, + { + "epoch": 2.4233528490145955, + "grad_norm": 2.507725238800049, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.7783598303794861, + "num_tokens": 773367815.0, + "step": 29388 + }, + { + "epoch": 2.4234353096396473, + "grad_norm": 2.1204826831817627, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7672117352485657, + "num_tokens": 773396727.0, + "step": 29389 + }, + { + "epoch": 2.4235177702646986, + "grad_norm": 2.1818623542785645, + "learning_rate": 1e-06, + "loss": 0.6221, + "mean_token_accuracy": 0.7927895784378052, + "num_tokens": 773422286.0, + "step": 29390 + }, + { + "epoch": 2.42360023088975, + "grad_norm": 1.8981330394744873, + "learning_rate": 1e-06, + "loss": 0.4717, + "mean_token_accuracy": 0.827349841594696, + "num_tokens": 773454767.0, + "step": 29391 + }, + { + "epoch": 2.4236826915148018, + "grad_norm": 1.9552353620529175, + "learning_rate": 1e-06, + "loss": 0.5694, + "mean_token_accuracy": 0.805314302444458, + "num_tokens": 773485741.0, + "step": 29392 + }, + { + "epoch": 2.423765152139853, + "grad_norm": 2.318307876586914, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7635369300842285, + "num_tokens": 773514933.0, + "step": 29393 + }, + { + "epoch": 2.423847612764905, + "grad_norm": 2.191908359527588, + "learning_rate": 1e-06, + "loss": 0.6, + "mean_token_accuracy": 0.8045877814292908, + "num_tokens": 773541042.0, + "step": 29394 + }, + { + "epoch": 2.4239300733899563, + "grad_norm": 2.0383400917053223, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7565801739692688, + "num_tokens": 773575995.0, + "step": 29395 + }, + { + "epoch": 2.4240125340150076, + "grad_norm": 1.9077609777450562, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.760345458984375, + "num_tokens": 773612256.0, + "step": 29396 + }, + { + "epoch": 2.4240949946400594, + "grad_norm": 2.146390438079834, + "learning_rate": 1e-06, + "loss": 0.6817, + "mean_token_accuracy": 0.7777692675590515, + "num_tokens": 773639770.0, + "step": 29397 + }, + { + "epoch": 2.4241774552651107, + "grad_norm": 2.2951161861419678, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7610117197036743, + "num_tokens": 773665474.0, + "step": 29398 + }, + { + "epoch": 2.4242599158901625, + "grad_norm": 1.8249523639678955, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.791183590888977, + "num_tokens": 773701906.0, + "step": 29399 + }, + { + "epoch": 2.424342376515214, + "grad_norm": 1.959555745124817, + "learning_rate": 1e-06, + "loss": 0.5084, + "mean_token_accuracy": 0.829176664352417, + "num_tokens": 773731788.0, + "step": 29400 + }, + { + "epoch": 2.4244248371402657, + "grad_norm": 2.3471131324768066, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7349613308906555, + "num_tokens": 773757346.0, + "step": 29401 + }, + { + "epoch": 2.424507297765317, + "grad_norm": 1.7178313732147217, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.7953563332557678, + "num_tokens": 773798816.0, + "step": 29402 + }, + { + "epoch": 2.4245897583903684, + "grad_norm": 2.023909091949463, + "learning_rate": 1e-06, + "loss": 0.5781, + "mean_token_accuracy": 0.805200457572937, + "num_tokens": 773828437.0, + "step": 29403 + }, + { + "epoch": 2.42467221901542, + "grad_norm": 1.9665299654006958, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7620570063591003, + "num_tokens": 773862304.0, + "step": 29404 + }, + { + "epoch": 2.4247546796404715, + "grad_norm": 2.0145812034606934, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7675561904907227, + "num_tokens": 773894479.0, + "step": 29405 + }, + { + "epoch": 2.4248371402655233, + "grad_norm": 2.1917693614959717, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7607454657554626, + "num_tokens": 773924678.0, + "step": 29406 + }, + { + "epoch": 2.4249196008905747, + "grad_norm": 2.0274202823638916, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7755064964294434, + "num_tokens": 773954671.0, + "step": 29407 + }, + { + "epoch": 2.4250020615156265, + "grad_norm": 2.0506629943847656, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.7919898629188538, + "num_tokens": 773985093.0, + "step": 29408 + }, + { + "epoch": 2.425084522140678, + "grad_norm": 1.9826921224594116, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7468947768211365, + "num_tokens": 774019694.0, + "step": 29409 + }, + { + "epoch": 2.425166982765729, + "grad_norm": 2.0621533393859863, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.779362678527832, + "num_tokens": 774049281.0, + "step": 29410 + }, + { + "epoch": 2.425249443390781, + "grad_norm": 2.490264654159546, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7697017788887024, + "num_tokens": 774079630.0, + "step": 29411 + }, + { + "epoch": 2.4253319040158323, + "grad_norm": 1.944103717803955, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.7870967388153076, + "num_tokens": 774114513.0, + "step": 29412 + }, + { + "epoch": 2.425414364640884, + "grad_norm": 2.2727155685424805, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.7923409938812256, + "num_tokens": 774138580.0, + "step": 29413 + }, + { + "epoch": 2.4254968252659355, + "grad_norm": 2.051797866821289, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7863637804985046, + "num_tokens": 774171815.0, + "step": 29414 + }, + { + "epoch": 2.4255792858909873, + "grad_norm": 2.043264150619507, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.788362979888916, + "num_tokens": 774201790.0, + "step": 29415 + }, + { + "epoch": 2.4256617465160386, + "grad_norm": 2.2373006343841553, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.7856522798538208, + "num_tokens": 774230617.0, + "step": 29416 + }, + { + "epoch": 2.42574420714109, + "grad_norm": 1.9804959297180176, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.78812175989151, + "num_tokens": 774266071.0, + "step": 29417 + }, + { + "epoch": 2.4258266677661418, + "grad_norm": 2.379204511642456, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.7856451869010925, + "num_tokens": 774289926.0, + "step": 29418 + }, + { + "epoch": 2.425909128391193, + "grad_norm": 2.050926685333252, + "learning_rate": 1e-06, + "loss": 0.5874, + "mean_token_accuracy": 0.8066062927246094, + "num_tokens": 774319852.0, + "step": 29419 + }, + { + "epoch": 2.425991589016245, + "grad_norm": 2.0066583156585693, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7566583156585693, + "num_tokens": 774352413.0, + "step": 29420 + }, + { + "epoch": 2.4260740496412962, + "grad_norm": 2.084446668624878, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7901186943054199, + "num_tokens": 774383376.0, + "step": 29421 + }, + { + "epoch": 2.426156510266348, + "grad_norm": 2.421041488647461, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7565423250198364, + "num_tokens": 774409592.0, + "step": 29422 + }, + { + "epoch": 2.4262389708913994, + "grad_norm": 2.0853397846221924, + "learning_rate": 1e-06, + "loss": 0.6079, + "mean_token_accuracy": 0.7913740873336792, + "num_tokens": 774443665.0, + "step": 29423 + }, + { + "epoch": 2.4263214315164507, + "grad_norm": 2.0658581256866455, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.778439998626709, + "num_tokens": 774473549.0, + "step": 29424 + }, + { + "epoch": 2.4264038921415025, + "grad_norm": 2.2524073123931885, + "learning_rate": 1e-06, + "loss": 0.486, + "mean_token_accuracy": 0.817608118057251, + "num_tokens": 774499022.0, + "step": 29425 + }, + { + "epoch": 2.426486352766554, + "grad_norm": 2.176927089691162, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7279491424560547, + "num_tokens": 774528735.0, + "step": 29426 + }, + { + "epoch": 2.4265688133916057, + "grad_norm": 2.091853618621826, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7417093515396118, + "num_tokens": 774558760.0, + "step": 29427 + }, + { + "epoch": 2.426651274016657, + "grad_norm": 2.4603748321533203, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7667832374572754, + "num_tokens": 774579513.0, + "step": 29428 + }, + { + "epoch": 2.426733734641709, + "grad_norm": 2.1890077590942383, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7511100769042969, + "num_tokens": 774607748.0, + "step": 29429 + }, + { + "epoch": 2.42681619526676, + "grad_norm": 1.816592812538147, + "learning_rate": 1e-06, + "loss": 0.496, + "mean_token_accuracy": 0.8361643552780151, + "num_tokens": 774642585.0, + "step": 29430 + }, + { + "epoch": 2.4268986558918115, + "grad_norm": 2.107164144515991, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.767379879951477, + "num_tokens": 774672487.0, + "step": 29431 + }, + { + "epoch": 2.4269811165168633, + "grad_norm": 2.3345630168914795, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7834473848342896, + "num_tokens": 774699976.0, + "step": 29432 + }, + { + "epoch": 2.4270635771419147, + "grad_norm": 2.260209560394287, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.7741750478744507, + "num_tokens": 774727827.0, + "step": 29433 + }, + { + "epoch": 2.427146037766966, + "grad_norm": 1.7562497854232788, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7794502973556519, + "num_tokens": 774766664.0, + "step": 29434 + }, + { + "epoch": 2.427228498392018, + "grad_norm": 2.516695022583008, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7697904706001282, + "num_tokens": 774790148.0, + "step": 29435 + }, + { + "epoch": 2.427310959017069, + "grad_norm": 2.186607599258423, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7620696425437927, + "num_tokens": 774822767.0, + "step": 29436 + }, + { + "epoch": 2.427393419642121, + "grad_norm": 2.4249393939971924, + "learning_rate": 1e-06, + "loss": 0.5321, + "mean_token_accuracy": 0.8155779838562012, + "num_tokens": 774846071.0, + "step": 29437 + }, + { + "epoch": 2.4274758802671723, + "grad_norm": 2.0184926986694336, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.7936052083969116, + "num_tokens": 774875127.0, + "step": 29438 + }, + { + "epoch": 2.427558340892224, + "grad_norm": 2.0723702907562256, + "learning_rate": 1e-06, + "loss": 0.5091, + "mean_token_accuracy": 0.8153076171875, + "num_tokens": 774911395.0, + "step": 29439 + }, + { + "epoch": 2.4276408015172755, + "grad_norm": 2.2845571041107178, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.7874807119369507, + "num_tokens": 774940477.0, + "step": 29440 + }, + { + "epoch": 2.427723262142327, + "grad_norm": 1.9532309770584106, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7787657380104065, + "num_tokens": 774975083.0, + "step": 29441 + }, + { + "epoch": 2.4278057227673786, + "grad_norm": 2.5675745010375977, + "learning_rate": 1e-06, + "loss": 0.6304, + "mean_token_accuracy": 0.7903797626495361, + "num_tokens": 774996194.0, + "step": 29442 + }, + { + "epoch": 2.42788818339243, + "grad_norm": 2.1175906658172607, + "learning_rate": 1e-06, + "loss": 0.502, + "mean_token_accuracy": 0.8230371475219727, + "num_tokens": 775022554.0, + "step": 29443 + }, + { + "epoch": 2.4279706440174817, + "grad_norm": 2.161543607711792, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.78357994556427, + "num_tokens": 775051235.0, + "step": 29444 + }, + { + "epoch": 2.428053104642533, + "grad_norm": 2.0643813610076904, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7806514501571655, + "num_tokens": 775079739.0, + "step": 29445 + }, + { + "epoch": 2.428135565267585, + "grad_norm": 2.272775888442993, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.766137957572937, + "num_tokens": 775108000.0, + "step": 29446 + }, + { + "epoch": 2.4282180258926362, + "grad_norm": 1.9796775579452515, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7607465386390686, + "num_tokens": 775143318.0, + "step": 29447 + }, + { + "epoch": 2.4283004865176876, + "grad_norm": 1.9539070129394531, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7923111915588379, + "num_tokens": 775176105.0, + "step": 29448 + }, + { + "epoch": 2.4283829471427394, + "grad_norm": 1.7087701559066772, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7823026776313782, + "num_tokens": 775218494.0, + "step": 29449 + }, + { + "epoch": 2.4284654077677907, + "grad_norm": 2.0582756996154785, + "learning_rate": 1e-06, + "loss": 0.7732, + "mean_token_accuracy": 0.7511228322982788, + "num_tokens": 775252400.0, + "step": 29450 + }, + { + "epoch": 2.4285478683928425, + "grad_norm": 2.2932565212249756, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.7877225875854492, + "num_tokens": 775279469.0, + "step": 29451 + }, + { + "epoch": 2.428630329017894, + "grad_norm": 2.104881525039673, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7726582884788513, + "num_tokens": 775311783.0, + "step": 29452 + }, + { + "epoch": 2.4287127896429457, + "grad_norm": 1.9788196086883545, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.7615852355957031, + "num_tokens": 775343360.0, + "step": 29453 + }, + { + "epoch": 2.428795250267997, + "grad_norm": 2.2049875259399414, + "learning_rate": 1e-06, + "loss": 0.7767, + "mean_token_accuracy": 0.7498780488967896, + "num_tokens": 775372327.0, + "step": 29454 + }, + { + "epoch": 2.4288777108930484, + "grad_norm": 2.1874070167541504, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7541570067405701, + "num_tokens": 775401252.0, + "step": 29455 + }, + { + "epoch": 2.4289601715181, + "grad_norm": 2.1152548789978027, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7813162803649902, + "num_tokens": 775428820.0, + "step": 29456 + }, + { + "epoch": 2.4290426321431515, + "grad_norm": 1.957305669784546, + "learning_rate": 1e-06, + "loss": 0.6075, + "mean_token_accuracy": 0.7884212136268616, + "num_tokens": 775462825.0, + "step": 29457 + }, + { + "epoch": 2.4291250927682033, + "grad_norm": 2.295398235321045, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.775731086730957, + "num_tokens": 775489040.0, + "step": 29458 + }, + { + "epoch": 2.4292075533932547, + "grad_norm": 1.849956750869751, + "learning_rate": 1e-06, + "loss": 0.6432, + "mean_token_accuracy": 0.7907693982124329, + "num_tokens": 775523766.0, + "step": 29459 + }, + { + "epoch": 2.4292900140183065, + "grad_norm": 2.2788074016571045, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7546979188919067, + "num_tokens": 775551959.0, + "step": 29460 + }, + { + "epoch": 2.429372474643358, + "grad_norm": 2.092987537384033, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.7973846197128296, + "num_tokens": 775582039.0, + "step": 29461 + }, + { + "epoch": 2.429454935268409, + "grad_norm": 2.5064730644226074, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.762921154499054, + "num_tokens": 775605004.0, + "step": 29462 + }, + { + "epoch": 2.429537395893461, + "grad_norm": 2.0092549324035645, + "learning_rate": 1e-06, + "loss": 0.5653, + "mean_token_accuracy": 0.8042168617248535, + "num_tokens": 775634792.0, + "step": 29463 + }, + { + "epoch": 2.4296198565185123, + "grad_norm": 2.3127245903015137, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7621322870254517, + "num_tokens": 775659280.0, + "step": 29464 + }, + { + "epoch": 2.429702317143564, + "grad_norm": 1.8926149606704712, + "learning_rate": 1e-06, + "loss": 0.5246, + "mean_token_accuracy": 0.829752504825592, + "num_tokens": 775691532.0, + "step": 29465 + }, + { + "epoch": 2.4297847777686155, + "grad_norm": 1.724182367324829, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.7893028259277344, + "num_tokens": 775731626.0, + "step": 29466 + }, + { + "epoch": 2.4298672383936673, + "grad_norm": 1.8274329900741577, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.7871500253677368, + "num_tokens": 775768766.0, + "step": 29467 + }, + { + "epoch": 2.4299496990187186, + "grad_norm": 2.295496940612793, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7493058443069458, + "num_tokens": 775795962.0, + "step": 29468 + }, + { + "epoch": 2.43003215964377, + "grad_norm": 1.971661925315857, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7840327620506287, + "num_tokens": 775829463.0, + "step": 29469 + }, + { + "epoch": 2.4301146202688217, + "grad_norm": 2.0051259994506836, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7566250562667847, + "num_tokens": 775864499.0, + "step": 29470 + }, + { + "epoch": 2.430197080893873, + "grad_norm": 2.2194058895111084, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7551321983337402, + "num_tokens": 775892483.0, + "step": 29471 + }, + { + "epoch": 2.430279541518925, + "grad_norm": 1.8932983875274658, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7715173959732056, + "num_tokens": 775926933.0, + "step": 29472 + }, + { + "epoch": 2.4303620021439762, + "grad_norm": 2.507235050201416, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7805958390235901, + "num_tokens": 775950098.0, + "step": 29473 + }, + { + "epoch": 2.430444462769028, + "grad_norm": 1.9488250017166138, + "learning_rate": 1e-06, + "loss": 0.5715, + "mean_token_accuracy": 0.8023631572723389, + "num_tokens": 775982686.0, + "step": 29474 + }, + { + "epoch": 2.4305269233940794, + "grad_norm": 2.1854000091552734, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7631704807281494, + "num_tokens": 776011002.0, + "step": 29475 + }, + { + "epoch": 2.4306093840191307, + "grad_norm": 2.2808754444122314, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7681193351745605, + "num_tokens": 776038738.0, + "step": 29476 + }, + { + "epoch": 2.4306918446441825, + "grad_norm": 2.1962790489196777, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.7835683822631836, + "num_tokens": 776067997.0, + "step": 29477 + }, + { + "epoch": 2.430774305269234, + "grad_norm": 1.9273746013641357, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7727106809616089, + "num_tokens": 776106890.0, + "step": 29478 + }, + { + "epoch": 2.4308567658942852, + "grad_norm": 2.019355535507202, + "learning_rate": 1e-06, + "loss": 0.5865, + "mean_token_accuracy": 0.8051042556762695, + "num_tokens": 776136600.0, + "step": 29479 + }, + { + "epoch": 2.430939226519337, + "grad_norm": 2.109647035598755, + "learning_rate": 1e-06, + "loss": 0.5137, + "mean_token_accuracy": 0.8246598839759827, + "num_tokens": 776164855.0, + "step": 29480 + }, + { + "epoch": 2.4310216871443884, + "grad_norm": 2.3095293045043945, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7714713215827942, + "num_tokens": 776189892.0, + "step": 29481 + }, + { + "epoch": 2.43110414776944, + "grad_norm": 2.2056524753570557, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7513209581375122, + "num_tokens": 776220419.0, + "step": 29482 + }, + { + "epoch": 2.4311866083944915, + "grad_norm": 2.6155059337615967, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7339704632759094, + "num_tokens": 776241197.0, + "step": 29483 + }, + { + "epoch": 2.4312690690195433, + "grad_norm": 2.115896463394165, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.7765929102897644, + "num_tokens": 776271910.0, + "step": 29484 + }, + { + "epoch": 2.4313515296445947, + "grad_norm": 2.2171144485473633, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7613579034805298, + "num_tokens": 776298783.0, + "step": 29485 + }, + { + "epoch": 2.431433990269646, + "grad_norm": 1.9248473644256592, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7799067497253418, + "num_tokens": 776333091.0, + "step": 29486 + }, + { + "epoch": 2.431516450894698, + "grad_norm": 1.999367594718933, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7591821551322937, + "num_tokens": 776365822.0, + "step": 29487 + }, + { + "epoch": 2.431598911519749, + "grad_norm": 2.0400657653808594, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.781203031539917, + "num_tokens": 776397910.0, + "step": 29488 + }, + { + "epoch": 2.431681372144801, + "grad_norm": 2.0826191902160645, + "learning_rate": 1e-06, + "loss": 0.7812, + "mean_token_accuracy": 0.757555365562439, + "num_tokens": 776427724.0, + "step": 29489 + }, + { + "epoch": 2.4317638327698523, + "grad_norm": 1.986823558807373, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.7934949398040771, + "num_tokens": 776459587.0, + "step": 29490 + }, + { + "epoch": 2.431846293394904, + "grad_norm": 2.324922561645508, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7433488368988037, + "num_tokens": 776486610.0, + "step": 29491 + }, + { + "epoch": 2.4319287540199555, + "grad_norm": 2.1603915691375732, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7749860286712646, + "num_tokens": 776513584.0, + "step": 29492 + }, + { + "epoch": 2.432011214645007, + "grad_norm": 2.0053551197052, + "learning_rate": 1e-06, + "loss": 0.759, + "mean_token_accuracy": 0.7406550645828247, + "num_tokens": 776547250.0, + "step": 29493 + }, + { + "epoch": 2.4320936752700586, + "grad_norm": 2.2179903984069824, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7431994080543518, + "num_tokens": 776576952.0, + "step": 29494 + }, + { + "epoch": 2.43217613589511, + "grad_norm": 2.1084213256835938, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7603315711021423, + "num_tokens": 776608107.0, + "step": 29495 + }, + { + "epoch": 2.4322585965201617, + "grad_norm": 2.2463674545288086, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7671855688095093, + "num_tokens": 776639829.0, + "step": 29496 + }, + { + "epoch": 2.432341057145213, + "grad_norm": 1.9543753862380981, + "learning_rate": 1e-06, + "loss": 0.5175, + "mean_token_accuracy": 0.8290215134620667, + "num_tokens": 776671331.0, + "step": 29497 + }, + { + "epoch": 2.432423517770265, + "grad_norm": 2.273700714111328, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.756840705871582, + "num_tokens": 776700072.0, + "step": 29498 + }, + { + "epoch": 2.4325059783953162, + "grad_norm": 1.7882022857666016, + "learning_rate": 1e-06, + "loss": 0.5803, + "mean_token_accuracy": 0.8087215423583984, + "num_tokens": 776734795.0, + "step": 29499 + }, + { + "epoch": 2.4325884390203676, + "grad_norm": 2.2218565940856934, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7384321689605713, + "num_tokens": 776763866.0, + "step": 29500 + }, + { + "epoch": 2.4326708996454194, + "grad_norm": 2.1705217361450195, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7472624778747559, + "num_tokens": 776794626.0, + "step": 29501 + }, + { + "epoch": 2.4327533602704707, + "grad_norm": 1.9070860147476196, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7676404714584351, + "num_tokens": 776832818.0, + "step": 29502 + }, + { + "epoch": 2.4328358208955225, + "grad_norm": 1.9244228601455688, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7734201550483704, + "num_tokens": 776866077.0, + "step": 29503 + }, + { + "epoch": 2.432918281520574, + "grad_norm": 2.29547119140625, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.7876278162002563, + "num_tokens": 776894032.0, + "step": 29504 + }, + { + "epoch": 2.4330007421456257, + "grad_norm": 2.0066330432891846, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7861320376396179, + "num_tokens": 776925567.0, + "step": 29505 + }, + { + "epoch": 2.433083202770677, + "grad_norm": 2.1504616737365723, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7521263957023621, + "num_tokens": 776956626.0, + "step": 29506 + }, + { + "epoch": 2.4331656633957284, + "grad_norm": 2.081984519958496, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7740434408187866, + "num_tokens": 776985599.0, + "step": 29507 + }, + { + "epoch": 2.43324812402078, + "grad_norm": 2.2110378742218018, + "learning_rate": 1e-06, + "loss": 0.5691, + "mean_token_accuracy": 0.7942980527877808, + "num_tokens": 777012951.0, + "step": 29508 + }, + { + "epoch": 2.4333305846458315, + "grad_norm": 1.923888921737671, + "learning_rate": 1e-06, + "loss": 0.807, + "mean_token_accuracy": 0.742619514465332, + "num_tokens": 777047355.0, + "step": 29509 + }, + { + "epoch": 2.4334130452708833, + "grad_norm": 2.0518572330474854, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.7855147123336792, + "num_tokens": 777081630.0, + "step": 29510 + }, + { + "epoch": 2.4334955058959347, + "grad_norm": 2.0503313541412354, + "learning_rate": 1e-06, + "loss": 0.6131, + "mean_token_accuracy": 0.7889494895935059, + "num_tokens": 777111942.0, + "step": 29511 + }, + { + "epoch": 2.4335779665209865, + "grad_norm": 2.3067808151245117, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7716729640960693, + "num_tokens": 777137728.0, + "step": 29512 + }, + { + "epoch": 2.433660427146038, + "grad_norm": 1.9056838750839233, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7346205115318298, + "num_tokens": 777170729.0, + "step": 29513 + }, + { + "epoch": 2.433742887771089, + "grad_norm": 2.5311357975006104, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.7844666242599487, + "num_tokens": 777192301.0, + "step": 29514 + }, + { + "epoch": 2.433825348396141, + "grad_norm": 2.2365334033966064, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7222165465354919, + "num_tokens": 777219244.0, + "step": 29515 + }, + { + "epoch": 2.4339078090211923, + "grad_norm": 2.07875394821167, + "learning_rate": 1e-06, + "loss": 0.7626, + "mean_token_accuracy": 0.758050799369812, + "num_tokens": 777250831.0, + "step": 29516 + }, + { + "epoch": 2.433990269646244, + "grad_norm": 2.216125965118408, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7883118987083435, + "num_tokens": 777278117.0, + "step": 29517 + }, + { + "epoch": 2.4340727302712954, + "grad_norm": 2.335559129714966, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7424042820930481, + "num_tokens": 777303941.0, + "step": 29518 + }, + { + "epoch": 2.4341551908963472, + "grad_norm": 2.1712729930877686, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7792109251022339, + "num_tokens": 777332070.0, + "step": 29519 + }, + { + "epoch": 2.4342376515213986, + "grad_norm": 2.102827787399292, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.7861671447753906, + "num_tokens": 777363678.0, + "step": 29520 + }, + { + "epoch": 2.43432011214645, + "grad_norm": 2.203019380569458, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.7804351449012756, + "num_tokens": 777392325.0, + "step": 29521 + }, + { + "epoch": 2.4344025727715017, + "grad_norm": 2.1272497177124023, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7609270811080933, + "num_tokens": 777421454.0, + "step": 29522 + }, + { + "epoch": 2.434485033396553, + "grad_norm": 2.296346664428711, + "learning_rate": 1e-06, + "loss": 0.612, + "mean_token_accuracy": 0.7959895730018616, + "num_tokens": 777447355.0, + "step": 29523 + }, + { + "epoch": 2.4345674940216044, + "grad_norm": 2.172720432281494, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.746391773223877, + "num_tokens": 777479667.0, + "step": 29524 + }, + { + "epoch": 2.4346499546466562, + "grad_norm": 1.9833123683929443, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7686513662338257, + "num_tokens": 777514241.0, + "step": 29525 + }, + { + "epoch": 2.4347324152717076, + "grad_norm": 2.0852181911468506, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7620193958282471, + "num_tokens": 777546013.0, + "step": 29526 + }, + { + "epoch": 2.4348148758967594, + "grad_norm": 1.8931372165679932, + "learning_rate": 1e-06, + "loss": 0.4816, + "mean_token_accuracy": 0.835070788860321, + "num_tokens": 777577105.0, + "step": 29527 + }, + { + "epoch": 2.4348973365218107, + "grad_norm": 2.2090694904327393, + "learning_rate": 1e-06, + "loss": 0.6349, + "mean_token_accuracy": 0.7970690131187439, + "num_tokens": 777603734.0, + "step": 29528 + }, + { + "epoch": 2.4349797971468625, + "grad_norm": 1.9444537162780762, + "learning_rate": 1e-06, + "loss": 0.5514, + "mean_token_accuracy": 0.8102879524230957, + "num_tokens": 777637563.0, + "step": 29529 + }, + { + "epoch": 2.435062257771914, + "grad_norm": 1.8220278024673462, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7800271511077881, + "num_tokens": 777673995.0, + "step": 29530 + }, + { + "epoch": 2.4351447183969652, + "grad_norm": 2.660923957824707, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.750822126865387, + "num_tokens": 777694889.0, + "step": 29531 + }, + { + "epoch": 2.435227179022017, + "grad_norm": 2.297041177749634, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.740882933139801, + "num_tokens": 777722403.0, + "step": 29532 + }, + { + "epoch": 2.4353096396470684, + "grad_norm": 1.7919286489486694, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.7911365032196045, + "num_tokens": 777757803.0, + "step": 29533 + }, + { + "epoch": 2.43539210027212, + "grad_norm": 2.091979742050171, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7585289478302002, + "num_tokens": 777787231.0, + "step": 29534 + }, + { + "epoch": 2.4354745608971715, + "grad_norm": 2.344015121459961, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7482085227966309, + "num_tokens": 777815242.0, + "step": 29535 + }, + { + "epoch": 2.4355570215222233, + "grad_norm": 2.1252667903900146, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7481657266616821, + "num_tokens": 777846112.0, + "step": 29536 + }, + { + "epoch": 2.4356394821472747, + "grad_norm": 1.8793364763259888, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7811033725738525, + "num_tokens": 777883631.0, + "step": 29537 + }, + { + "epoch": 2.435721942772326, + "grad_norm": 1.912147045135498, + "learning_rate": 1e-06, + "loss": 0.5067, + "mean_token_accuracy": 0.8265683650970459, + "num_tokens": 777917820.0, + "step": 29538 + }, + { + "epoch": 2.435804403397378, + "grad_norm": 2.257256031036377, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7550842761993408, + "num_tokens": 777944805.0, + "step": 29539 + }, + { + "epoch": 2.435886864022429, + "grad_norm": 1.9044359922409058, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.8012853264808655, + "num_tokens": 777976679.0, + "step": 29540 + }, + { + "epoch": 2.435969324647481, + "grad_norm": 2.4245240688323975, + "learning_rate": 1e-06, + "loss": 0.5234, + "mean_token_accuracy": 0.8189185261726379, + "num_tokens": 778002765.0, + "step": 29541 + }, + { + "epoch": 2.4360517852725323, + "grad_norm": 2.3045740127563477, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.7842493057250977, + "num_tokens": 778026737.0, + "step": 29542 + }, + { + "epoch": 2.436134245897584, + "grad_norm": 1.9303511381149292, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7635601758956909, + "num_tokens": 778061019.0, + "step": 29543 + }, + { + "epoch": 2.4362167065226354, + "grad_norm": 2.176325559616089, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7674009799957275, + "num_tokens": 778089728.0, + "step": 29544 + }, + { + "epoch": 2.436299167147687, + "grad_norm": 2.2372357845306396, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7239352464675903, + "num_tokens": 778121665.0, + "step": 29545 + }, + { + "epoch": 2.4363816277727386, + "grad_norm": 2.2602977752685547, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7784690856933594, + "num_tokens": 778148287.0, + "step": 29546 + }, + { + "epoch": 2.43646408839779, + "grad_norm": 2.170165777206421, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7742631435394287, + "num_tokens": 778174835.0, + "step": 29547 + }, + { + "epoch": 2.4365465490228417, + "grad_norm": 2.1168766021728516, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7654669284820557, + "num_tokens": 778204567.0, + "step": 29548 + }, + { + "epoch": 2.436629009647893, + "grad_norm": 2.129664182662964, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7595928311347961, + "num_tokens": 778233418.0, + "step": 29549 + }, + { + "epoch": 2.436711470272945, + "grad_norm": 2.078599691390991, + "learning_rate": 1e-06, + "loss": 0.7784, + "mean_token_accuracy": 0.7515970468521118, + "num_tokens": 778266473.0, + "step": 29550 + }, + { + "epoch": 2.4367939308979962, + "grad_norm": 2.1367175579071045, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.720858633518219, + "num_tokens": 778298635.0, + "step": 29551 + }, + { + "epoch": 2.4368763915230476, + "grad_norm": 2.2901201248168945, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7567819356918335, + "num_tokens": 778327632.0, + "step": 29552 + }, + { + "epoch": 2.4369588521480994, + "grad_norm": 2.115497589111328, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7766860723495483, + "num_tokens": 778356576.0, + "step": 29553 + }, + { + "epoch": 2.4370413127731507, + "grad_norm": 2.212556838989258, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.7608127593994141, + "num_tokens": 778385148.0, + "step": 29554 + }, + { + "epoch": 2.4371237733982025, + "grad_norm": 2.586686611175537, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7319236397743225, + "num_tokens": 778408242.0, + "step": 29555 + }, + { + "epoch": 2.437206234023254, + "grad_norm": 1.9603056907653809, + "learning_rate": 1e-06, + "loss": 0.6155, + "mean_token_accuracy": 0.7950416803359985, + "num_tokens": 778441958.0, + "step": 29556 + }, + { + "epoch": 2.4372886946483057, + "grad_norm": 2.184413194656372, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.7912470102310181, + "num_tokens": 778471042.0, + "step": 29557 + }, + { + "epoch": 2.437371155273357, + "grad_norm": 2.2889091968536377, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7927271723747253, + "num_tokens": 778494972.0, + "step": 29558 + }, + { + "epoch": 2.4374536158984084, + "grad_norm": 2.1382720470428467, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7424672842025757, + "num_tokens": 778525307.0, + "step": 29559 + }, + { + "epoch": 2.43753607652346, + "grad_norm": 1.8589766025543213, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.7834643125534058, + "num_tokens": 778563424.0, + "step": 29560 + }, + { + "epoch": 2.4376185371485115, + "grad_norm": 2.2513508796691895, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.775069534778595, + "num_tokens": 778589902.0, + "step": 29561 + }, + { + "epoch": 2.4377009977735633, + "grad_norm": 1.9390252828598022, + "learning_rate": 1e-06, + "loss": 0.6149, + "mean_token_accuracy": 0.789355993270874, + "num_tokens": 778623400.0, + "step": 29562 + }, + { + "epoch": 2.4377834583986147, + "grad_norm": 2.2742602825164795, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7823800444602966, + "num_tokens": 778647889.0, + "step": 29563 + }, + { + "epoch": 2.437865919023666, + "grad_norm": 1.8469172716140747, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7734360694885254, + "num_tokens": 778682876.0, + "step": 29564 + }, + { + "epoch": 2.437948379648718, + "grad_norm": 2.214287042617798, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7712386846542358, + "num_tokens": 778709715.0, + "step": 29565 + }, + { + "epoch": 2.438030840273769, + "grad_norm": 1.9482568502426147, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7587745189666748, + "num_tokens": 778744758.0, + "step": 29566 + }, + { + "epoch": 2.438113300898821, + "grad_norm": 1.721863865852356, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7779696583747864, + "num_tokens": 778786255.0, + "step": 29567 + }, + { + "epoch": 2.4381957615238723, + "grad_norm": 1.9249529838562012, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.7983829975128174, + "num_tokens": 778823701.0, + "step": 29568 + }, + { + "epoch": 2.4382782221489236, + "grad_norm": 2.100391387939453, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7856512665748596, + "num_tokens": 778850088.0, + "step": 29569 + }, + { + "epoch": 2.4383606827739754, + "grad_norm": 2.088693618774414, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7700353860855103, + "num_tokens": 778881365.0, + "step": 29570 + }, + { + "epoch": 2.438443143399027, + "grad_norm": 2.0877721309661865, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7621568441390991, + "num_tokens": 778917171.0, + "step": 29571 + }, + { + "epoch": 2.4385256040240786, + "grad_norm": 1.848449945449829, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.8031388521194458, + "num_tokens": 778955273.0, + "step": 29572 + }, + { + "epoch": 2.43860806464913, + "grad_norm": 2.12396240234375, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7470607757568359, + "num_tokens": 778987219.0, + "step": 29573 + }, + { + "epoch": 2.4386905252741817, + "grad_norm": 2.170707941055298, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.774796187877655, + "num_tokens": 779013417.0, + "step": 29574 + }, + { + "epoch": 2.438772985899233, + "grad_norm": 2.744401693344116, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7436188459396362, + "num_tokens": 779035584.0, + "step": 29575 + }, + { + "epoch": 2.4388554465242844, + "grad_norm": 2.3436474800109863, + "learning_rate": 1e-06, + "loss": 0.6068, + "mean_token_accuracy": 0.7966471314430237, + "num_tokens": 779059736.0, + "step": 29576 + }, + { + "epoch": 2.4389379071493362, + "grad_norm": 1.988254189491272, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.7834087610244751, + "num_tokens": 779092642.0, + "step": 29577 + }, + { + "epoch": 2.4390203677743876, + "grad_norm": 2.017261266708374, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.7977620363235474, + "num_tokens": 779128066.0, + "step": 29578 + }, + { + "epoch": 2.4391028283994394, + "grad_norm": 1.9745482206344604, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.7901395559310913, + "num_tokens": 779158792.0, + "step": 29579 + }, + { + "epoch": 2.4391852890244907, + "grad_norm": 2.396592617034912, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.7897492051124573, + "num_tokens": 779181504.0, + "step": 29580 + }, + { + "epoch": 2.4392677496495425, + "grad_norm": 1.9751389026641846, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.7921537160873413, + "num_tokens": 779215178.0, + "step": 29581 + }, + { + "epoch": 2.439350210274594, + "grad_norm": 1.9513930082321167, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7516312003135681, + "num_tokens": 779247725.0, + "step": 29582 + }, + { + "epoch": 2.439432670899645, + "grad_norm": 2.0972423553466797, + "learning_rate": 1e-06, + "loss": 0.5152, + "mean_token_accuracy": 0.8203679323196411, + "num_tokens": 779273719.0, + "step": 29583 + }, + { + "epoch": 2.439515131524697, + "grad_norm": 1.9940731525421143, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7429900169372559, + "num_tokens": 779308048.0, + "step": 29584 + }, + { + "epoch": 2.4395975921497484, + "grad_norm": 2.000910997390747, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7895461320877075, + "num_tokens": 779340305.0, + "step": 29585 + }, + { + "epoch": 2.4396800527748, + "grad_norm": 1.8333359956741333, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7704570293426514, + "num_tokens": 779376853.0, + "step": 29586 + }, + { + "epoch": 2.4397625133998515, + "grad_norm": 1.9791285991668701, + "learning_rate": 1e-06, + "loss": 0.5825, + "mean_token_accuracy": 0.8123127818107605, + "num_tokens": 779407489.0, + "step": 29587 + }, + { + "epoch": 2.4398449740249033, + "grad_norm": 2.042853355407715, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7299100160598755, + "num_tokens": 779439716.0, + "step": 29588 + }, + { + "epoch": 2.4399274346499547, + "grad_norm": 2.076958656311035, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7024953365325928, + "num_tokens": 779473663.0, + "step": 29589 + }, + { + "epoch": 2.440009895275006, + "grad_norm": 1.8863494396209717, + "learning_rate": 1e-06, + "loss": 0.5909, + "mean_token_accuracy": 0.8015508651733398, + "num_tokens": 779507709.0, + "step": 29590 + }, + { + "epoch": 2.440092355900058, + "grad_norm": 2.1107289791107178, + "learning_rate": 1e-06, + "loss": 0.5885, + "mean_token_accuracy": 0.8021812438964844, + "num_tokens": 779534112.0, + "step": 29591 + }, + { + "epoch": 2.440174816525109, + "grad_norm": 1.9163576364517212, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.8044096231460571, + "num_tokens": 779569428.0, + "step": 29592 + }, + { + "epoch": 2.440257277150161, + "grad_norm": 2.180445432662964, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7697516679763794, + "num_tokens": 779598492.0, + "step": 29593 + }, + { + "epoch": 2.4403397377752123, + "grad_norm": 1.8862978219985962, + "learning_rate": 1e-06, + "loss": 0.5707, + "mean_token_accuracy": 0.8051744699478149, + "num_tokens": 779630548.0, + "step": 29594 + }, + { + "epoch": 2.440422198400264, + "grad_norm": 2.2295279502868652, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7829620838165283, + "num_tokens": 779657548.0, + "step": 29595 + }, + { + "epoch": 2.4405046590253154, + "grad_norm": 2.216271162033081, + "learning_rate": 1e-06, + "loss": 0.7303, + "mean_token_accuracy": 0.7641260623931885, + "num_tokens": 779686316.0, + "step": 29596 + }, + { + "epoch": 2.440587119650367, + "grad_norm": 1.9117413759231567, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7837365865707397, + "num_tokens": 779720071.0, + "step": 29597 + }, + { + "epoch": 2.4406695802754186, + "grad_norm": 1.9801263809204102, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7777066826820374, + "num_tokens": 779752584.0, + "step": 29598 + }, + { + "epoch": 2.44075204090047, + "grad_norm": 1.9440617561340332, + "learning_rate": 1e-06, + "loss": 0.5663, + "mean_token_accuracy": 0.818457841873169, + "num_tokens": 779783185.0, + "step": 29599 + }, + { + "epoch": 2.4408345015255217, + "grad_norm": 1.853243350982666, + "learning_rate": 1e-06, + "loss": 0.5849, + "mean_token_accuracy": 0.79700767993927, + "num_tokens": 779817310.0, + "step": 29600 + }, + { + "epoch": 2.440916962150573, + "grad_norm": 2.1695499420166016, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7413178086280823, + "num_tokens": 779847306.0, + "step": 29601 + }, + { + "epoch": 2.440999422775625, + "grad_norm": 2.1723105907440186, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.7988197803497314, + "num_tokens": 779876678.0, + "step": 29602 + }, + { + "epoch": 2.4410818834006762, + "grad_norm": 2.0988950729370117, + "learning_rate": 1e-06, + "loss": 0.5653, + "mean_token_accuracy": 0.805057168006897, + "num_tokens": 779905182.0, + "step": 29603 + }, + { + "epoch": 2.4411643440257276, + "grad_norm": 2.027266502380371, + "learning_rate": 1e-06, + "loss": 0.5843, + "mean_token_accuracy": 0.7953556776046753, + "num_tokens": 779937514.0, + "step": 29604 + }, + { + "epoch": 2.4412468046507794, + "grad_norm": 2.1316003799438477, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7194616198539734, + "num_tokens": 779965983.0, + "step": 29605 + }, + { + "epoch": 2.4413292652758307, + "grad_norm": 1.851561188697815, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7596383690834045, + "num_tokens": 780002222.0, + "step": 29606 + }, + { + "epoch": 2.4414117259008825, + "grad_norm": 1.9844343662261963, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7621273398399353, + "num_tokens": 780038146.0, + "step": 29607 + }, + { + "epoch": 2.441494186525934, + "grad_norm": 2.1275572776794434, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7615771293640137, + "num_tokens": 780066315.0, + "step": 29608 + }, + { + "epoch": 2.441576647150985, + "grad_norm": 2.0338878631591797, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7782376408576965, + "num_tokens": 780096383.0, + "step": 29609 + }, + { + "epoch": 2.441659107776037, + "grad_norm": 1.9495775699615479, + "learning_rate": 1e-06, + "loss": 0.5482, + "mean_token_accuracy": 0.8189847469329834, + "num_tokens": 780130053.0, + "step": 29610 + }, + { + "epoch": 2.4417415684010884, + "grad_norm": 2.3392460346221924, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7498922348022461, + "num_tokens": 780155604.0, + "step": 29611 + }, + { + "epoch": 2.44182402902614, + "grad_norm": 2.1053426265716553, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7634860277175903, + "num_tokens": 780186377.0, + "step": 29612 + }, + { + "epoch": 2.4419064896511915, + "grad_norm": 2.0112953186035156, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7805752754211426, + "num_tokens": 780218125.0, + "step": 29613 + }, + { + "epoch": 2.441988950276243, + "grad_norm": 1.6892178058624268, + "learning_rate": 1e-06, + "loss": 0.4716, + "mean_token_accuracy": 0.8437992334365845, + "num_tokens": 780254769.0, + "step": 29614 + }, + { + "epoch": 2.4420714109012946, + "grad_norm": 1.9324880838394165, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7491695880889893, + "num_tokens": 780291554.0, + "step": 29615 + }, + { + "epoch": 2.442153871526346, + "grad_norm": 1.8812299966812134, + "learning_rate": 1e-06, + "loss": 0.4758, + "mean_token_accuracy": 0.8335798978805542, + "num_tokens": 780323687.0, + "step": 29616 + }, + { + "epoch": 2.442236332151398, + "grad_norm": 2.0618531703948975, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7439432144165039, + "num_tokens": 780355538.0, + "step": 29617 + }, + { + "epoch": 2.442318792776449, + "grad_norm": 2.0192320346832275, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7808533906936646, + "num_tokens": 780385847.0, + "step": 29618 + }, + { + "epoch": 2.442401253401501, + "grad_norm": 2.125493288040161, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7714391946792603, + "num_tokens": 780417381.0, + "step": 29619 + }, + { + "epoch": 2.4424837140265523, + "grad_norm": 2.0162432193756104, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7616533637046814, + "num_tokens": 780446781.0, + "step": 29620 + }, + { + "epoch": 2.4425661746516036, + "grad_norm": 2.273737907409668, + "learning_rate": 1e-06, + "loss": 0.6028, + "mean_token_accuracy": 0.7979394197463989, + "num_tokens": 780470897.0, + "step": 29621 + }, + { + "epoch": 2.4426486352766554, + "grad_norm": 1.8390692472457886, + "learning_rate": 1e-06, + "loss": 0.4671, + "mean_token_accuracy": 0.8299790024757385, + "num_tokens": 780502931.0, + "step": 29622 + }, + { + "epoch": 2.442731095901707, + "grad_norm": 2.168064594268799, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7592372298240662, + "num_tokens": 780530715.0, + "step": 29623 + }, + { + "epoch": 2.4428135565267586, + "grad_norm": 2.0612540245056152, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7791244387626648, + "num_tokens": 780559430.0, + "step": 29624 + }, + { + "epoch": 2.44289601715181, + "grad_norm": 1.9720468521118164, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.7860291004180908, + "num_tokens": 780591590.0, + "step": 29625 + }, + { + "epoch": 2.4429784777768617, + "grad_norm": 2.1542789936065674, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7784533500671387, + "num_tokens": 780620217.0, + "step": 29626 + }, + { + "epoch": 2.443060938401913, + "grad_norm": 2.1484506130218506, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7679831981658936, + "num_tokens": 780648713.0, + "step": 29627 + }, + { + "epoch": 2.4431433990269644, + "grad_norm": 2.0086517333984375, + "learning_rate": 1e-06, + "loss": 0.604, + "mean_token_accuracy": 0.7950528860092163, + "num_tokens": 780679256.0, + "step": 29628 + }, + { + "epoch": 2.443225859652016, + "grad_norm": 1.9288756847381592, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7806267142295837, + "num_tokens": 780713073.0, + "step": 29629 + }, + { + "epoch": 2.4433083202770676, + "grad_norm": 2.201629638671875, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.7840694189071655, + "num_tokens": 780741145.0, + "step": 29630 + }, + { + "epoch": 2.4433907809021194, + "grad_norm": 2.2353179454803467, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7522902488708496, + "num_tokens": 780771480.0, + "step": 29631 + }, + { + "epoch": 2.4434732415271707, + "grad_norm": 2.0645992755889893, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7282121181488037, + "num_tokens": 780803863.0, + "step": 29632 + }, + { + "epoch": 2.4435557021522225, + "grad_norm": 2.186098098754883, + "learning_rate": 1e-06, + "loss": 0.7529, + "mean_token_accuracy": 0.7479754090309143, + "num_tokens": 780832708.0, + "step": 29633 + }, + { + "epoch": 2.443638162777274, + "grad_norm": 2.0318596363067627, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7845573425292969, + "num_tokens": 780865458.0, + "step": 29634 + }, + { + "epoch": 2.443720623402325, + "grad_norm": 2.2944822311401367, + "learning_rate": 1e-06, + "loss": 0.6027, + "mean_token_accuracy": 0.7974343299865723, + "num_tokens": 780890343.0, + "step": 29635 + }, + { + "epoch": 2.443803084027377, + "grad_norm": 2.3833343982696533, + "learning_rate": 1e-06, + "loss": 0.7815, + "mean_token_accuracy": 0.7467289566993713, + "num_tokens": 780917298.0, + "step": 29636 + }, + { + "epoch": 2.4438855446524284, + "grad_norm": 2.0752780437469482, + "learning_rate": 1e-06, + "loss": 0.5992, + "mean_token_accuracy": 0.7983829379081726, + "num_tokens": 780947914.0, + "step": 29637 + }, + { + "epoch": 2.44396800527748, + "grad_norm": 2.2305171489715576, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7494984865188599, + "num_tokens": 780977762.0, + "step": 29638 + }, + { + "epoch": 2.4440504659025315, + "grad_norm": 2.0004470348358154, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7638235092163086, + "num_tokens": 781011451.0, + "step": 29639 + }, + { + "epoch": 2.4441329265275833, + "grad_norm": 1.9153149127960205, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7472150921821594, + "num_tokens": 781048569.0, + "step": 29640 + }, + { + "epoch": 2.4442153871526346, + "grad_norm": 2.2212390899658203, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.743046760559082, + "num_tokens": 781076204.0, + "step": 29641 + }, + { + "epoch": 2.444297847777686, + "grad_norm": 2.12414288520813, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7450903654098511, + "num_tokens": 781107640.0, + "step": 29642 + }, + { + "epoch": 2.444380308402738, + "grad_norm": 2.318732261657715, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.776368260383606, + "num_tokens": 781135736.0, + "step": 29643 + }, + { + "epoch": 2.444462769027789, + "grad_norm": 2.2987008094787598, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7775651216506958, + "num_tokens": 781163228.0, + "step": 29644 + }, + { + "epoch": 2.444545229652841, + "grad_norm": 2.2882936000823975, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7773186564445496, + "num_tokens": 781189159.0, + "step": 29645 + }, + { + "epoch": 2.4446276902778923, + "grad_norm": 2.0353128910064697, + "learning_rate": 1e-06, + "loss": 0.556, + "mean_token_accuracy": 0.8054189682006836, + "num_tokens": 781219545.0, + "step": 29646 + }, + { + "epoch": 2.444710150902944, + "grad_norm": 2.2279670238494873, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7523741722106934, + "num_tokens": 781249482.0, + "step": 29647 + }, + { + "epoch": 2.4447926115279954, + "grad_norm": 1.8079627752304077, + "learning_rate": 1e-06, + "loss": 0.5746, + "mean_token_accuracy": 0.8147483468055725, + "num_tokens": 781287125.0, + "step": 29648 + }, + { + "epoch": 2.444875072153047, + "grad_norm": 2.210975170135498, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.7850779294967651, + "num_tokens": 781313975.0, + "step": 29649 + }, + { + "epoch": 2.4449575327780986, + "grad_norm": 2.2253193855285645, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7689635753631592, + "num_tokens": 781342760.0, + "step": 29650 + }, + { + "epoch": 2.44503999340315, + "grad_norm": 2.0205767154693604, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7567299604415894, + "num_tokens": 781373957.0, + "step": 29651 + }, + { + "epoch": 2.4451224540282017, + "grad_norm": 2.6379592418670654, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7323355674743652, + "num_tokens": 781394917.0, + "step": 29652 + }, + { + "epoch": 2.445204914653253, + "grad_norm": 2.300947427749634, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7747399806976318, + "num_tokens": 781417440.0, + "step": 29653 + }, + { + "epoch": 2.4452873752783044, + "grad_norm": 2.065211057662964, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7546120285987854, + "num_tokens": 781448293.0, + "step": 29654 + }, + { + "epoch": 2.445369835903356, + "grad_norm": 2.041952133178711, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7750328779220581, + "num_tokens": 781478976.0, + "step": 29655 + }, + { + "epoch": 2.4454522965284076, + "grad_norm": 2.3776955604553223, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.7846848964691162, + "num_tokens": 781502293.0, + "step": 29656 + }, + { + "epoch": 2.4455347571534594, + "grad_norm": 2.241589069366455, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7838045358657837, + "num_tokens": 781526258.0, + "step": 29657 + }, + { + "epoch": 2.4456172177785107, + "grad_norm": 2.1021687984466553, + "learning_rate": 1e-06, + "loss": 0.5246, + "mean_token_accuracy": 0.8163975477218628, + "num_tokens": 781553425.0, + "step": 29658 + }, + { + "epoch": 2.445699678403562, + "grad_norm": 1.9450136423110962, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.7885389924049377, + "num_tokens": 781585706.0, + "step": 29659 + }, + { + "epoch": 2.445782139028614, + "grad_norm": 2.148073434829712, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7570145130157471, + "num_tokens": 781612644.0, + "step": 29660 + }, + { + "epoch": 2.445864599653665, + "grad_norm": 2.0588927268981934, + "learning_rate": 1e-06, + "loss": 0.5828, + "mean_token_accuracy": 0.8094437718391418, + "num_tokens": 781643000.0, + "step": 29661 + }, + { + "epoch": 2.445947060278717, + "grad_norm": 2.1114957332611084, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7707729339599609, + "num_tokens": 781674691.0, + "step": 29662 + }, + { + "epoch": 2.4460295209037684, + "grad_norm": 2.5620477199554443, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7206960916519165, + "num_tokens": 781697704.0, + "step": 29663 + }, + { + "epoch": 2.44611198152882, + "grad_norm": 2.2608799934387207, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7374867796897888, + "num_tokens": 781725553.0, + "step": 29664 + }, + { + "epoch": 2.4461944421538715, + "grad_norm": 1.8245445489883423, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.7896531820297241, + "num_tokens": 781763366.0, + "step": 29665 + }, + { + "epoch": 2.446276902778923, + "grad_norm": 2.02506685256958, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7723513841629028, + "num_tokens": 781797393.0, + "step": 29666 + }, + { + "epoch": 2.4463593634039746, + "grad_norm": 2.3806896209716797, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7579343318939209, + "num_tokens": 781822340.0, + "step": 29667 + }, + { + "epoch": 2.446441824029026, + "grad_norm": 2.1362788677215576, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7836968302726746, + "num_tokens": 781850037.0, + "step": 29668 + }, + { + "epoch": 2.446524284654078, + "grad_norm": 2.0506131649017334, + "learning_rate": 1e-06, + "loss": 0.5271, + "mean_token_accuracy": 0.8203670978546143, + "num_tokens": 781881649.0, + "step": 29669 + }, + { + "epoch": 2.446606745279129, + "grad_norm": 2.17751145362854, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7641019225120544, + "num_tokens": 781910789.0, + "step": 29670 + }, + { + "epoch": 2.446689205904181, + "grad_norm": 2.151902198791504, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7669374346733093, + "num_tokens": 781937981.0, + "step": 29671 + }, + { + "epoch": 2.4467716665292323, + "grad_norm": 2.0054798126220703, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.768101692199707, + "num_tokens": 781969700.0, + "step": 29672 + }, + { + "epoch": 2.4468541271542836, + "grad_norm": 1.9729101657867432, + "learning_rate": 1e-06, + "loss": 0.5711, + "mean_token_accuracy": 0.8035008311271667, + "num_tokens": 782000332.0, + "step": 29673 + }, + { + "epoch": 2.4469365877793354, + "grad_norm": 2.2441086769104004, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.7824303507804871, + "num_tokens": 782028588.0, + "step": 29674 + }, + { + "epoch": 2.4470190484043868, + "grad_norm": 2.5005438327789307, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7250577807426453, + "num_tokens": 782053172.0, + "step": 29675 + }, + { + "epoch": 2.4471015090294386, + "grad_norm": 1.9981516599655151, + "learning_rate": 1e-06, + "loss": 0.7486, + "mean_token_accuracy": 0.7599174380302429, + "num_tokens": 782087243.0, + "step": 29676 + }, + { + "epoch": 2.44718396965449, + "grad_norm": 2.298790454864502, + "learning_rate": 1e-06, + "loss": 0.5844, + "mean_token_accuracy": 0.8052488565444946, + "num_tokens": 782111481.0, + "step": 29677 + }, + { + "epoch": 2.4472664302795417, + "grad_norm": 2.0836687088012695, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7926058173179626, + "num_tokens": 782141292.0, + "step": 29678 + }, + { + "epoch": 2.447348890904593, + "grad_norm": 2.330974817276001, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7676065564155579, + "num_tokens": 782168150.0, + "step": 29679 + }, + { + "epoch": 2.4474313515296444, + "grad_norm": 2.222005605697632, + "learning_rate": 1e-06, + "loss": 0.7378, + "mean_token_accuracy": 0.7588329911231995, + "num_tokens": 782196558.0, + "step": 29680 + }, + { + "epoch": 2.447513812154696, + "grad_norm": 1.9764177799224854, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.737061619758606, + "num_tokens": 782228135.0, + "step": 29681 + }, + { + "epoch": 2.4475962727797476, + "grad_norm": 2.0400569438934326, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.7566085457801819, + "num_tokens": 782263305.0, + "step": 29682 + }, + { + "epoch": 2.4476787334047994, + "grad_norm": 2.180976152420044, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7653617858886719, + "num_tokens": 782293041.0, + "step": 29683 + }, + { + "epoch": 2.4477611940298507, + "grad_norm": 2.017454147338867, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7482458353042603, + "num_tokens": 782325657.0, + "step": 29684 + }, + { + "epoch": 2.4478436546549025, + "grad_norm": 2.5391690731048584, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7487049102783203, + "num_tokens": 782348280.0, + "step": 29685 + }, + { + "epoch": 2.447926115279954, + "grad_norm": 2.0715646743774414, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7898696660995483, + "num_tokens": 782376547.0, + "step": 29686 + }, + { + "epoch": 2.448008575905005, + "grad_norm": 2.080080270767212, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7558333873748779, + "num_tokens": 782407713.0, + "step": 29687 + }, + { + "epoch": 2.448091036530057, + "grad_norm": 1.9995932579040527, + "learning_rate": 1e-06, + "loss": 0.5704, + "mean_token_accuracy": 0.8083757162094116, + "num_tokens": 782437981.0, + "step": 29688 + }, + { + "epoch": 2.4481734971551083, + "grad_norm": 2.411362409591675, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7382228374481201, + "num_tokens": 782461206.0, + "step": 29689 + }, + { + "epoch": 2.44825595778016, + "grad_norm": 2.1147022247314453, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.725292980670929, + "num_tokens": 782493359.0, + "step": 29690 + }, + { + "epoch": 2.4483384184052115, + "grad_norm": 1.8186619281768799, + "learning_rate": 1e-06, + "loss": 0.5923, + "mean_token_accuracy": 0.7992674112319946, + "num_tokens": 782528958.0, + "step": 29691 + }, + { + "epoch": 2.4484208790302633, + "grad_norm": 2.1402573585510254, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7362973690032959, + "num_tokens": 782558813.0, + "step": 29692 + }, + { + "epoch": 2.4485033396553146, + "grad_norm": 2.2504169940948486, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.762829065322876, + "num_tokens": 782585759.0, + "step": 29693 + }, + { + "epoch": 2.448585800280366, + "grad_norm": 2.0678062438964844, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7321029901504517, + "num_tokens": 782615831.0, + "step": 29694 + }, + { + "epoch": 2.448668260905418, + "grad_norm": 2.229767322540283, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7340920567512512, + "num_tokens": 782643896.0, + "step": 29695 + }, + { + "epoch": 2.448750721530469, + "grad_norm": 2.0154457092285156, + "learning_rate": 1e-06, + "loss": 0.5323, + "mean_token_accuracy": 0.8163682222366333, + "num_tokens": 782675497.0, + "step": 29696 + }, + { + "epoch": 2.448833182155521, + "grad_norm": 2.2641074657440186, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7297977209091187, + "num_tokens": 782705632.0, + "step": 29697 + }, + { + "epoch": 2.4489156427805723, + "grad_norm": 2.3911149501800537, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.786861777305603, + "num_tokens": 782729830.0, + "step": 29698 + }, + { + "epoch": 2.4489981034056236, + "grad_norm": 2.127467393875122, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7481006383895874, + "num_tokens": 782759472.0, + "step": 29699 + }, + { + "epoch": 2.4490805640306754, + "grad_norm": 2.3982791900634766, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.7733503580093384, + "num_tokens": 782782457.0, + "step": 29700 + }, + { + "epoch": 2.4491630246557268, + "grad_norm": 2.446211338043213, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.7821287512779236, + "num_tokens": 782805854.0, + "step": 29701 + }, + { + "epoch": 2.4492454852807786, + "grad_norm": 1.970479965209961, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.730064868927002, + "num_tokens": 782843730.0, + "step": 29702 + }, + { + "epoch": 2.44932794590583, + "grad_norm": 2.271771192550659, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7696616649627686, + "num_tokens": 782870957.0, + "step": 29703 + }, + { + "epoch": 2.4494104065308813, + "grad_norm": 2.3424527645111084, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.763123095035553, + "num_tokens": 782897376.0, + "step": 29704 + }, + { + "epoch": 2.449492867155933, + "grad_norm": 2.1830179691314697, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7557978630065918, + "num_tokens": 782925970.0, + "step": 29705 + }, + { + "epoch": 2.4495753277809844, + "grad_norm": 2.328277111053467, + "learning_rate": 1e-06, + "loss": 0.5823, + "mean_token_accuracy": 0.8001061677932739, + "num_tokens": 782950541.0, + "step": 29706 + }, + { + "epoch": 2.449657788406036, + "grad_norm": 2.0419201850891113, + "learning_rate": 1e-06, + "loss": 0.5417, + "mean_token_accuracy": 0.8152775168418884, + "num_tokens": 782983476.0, + "step": 29707 + }, + { + "epoch": 2.4497402490310876, + "grad_norm": 2.138307809829712, + "learning_rate": 1e-06, + "loss": 0.6229, + "mean_token_accuracy": 0.7864041328430176, + "num_tokens": 783011166.0, + "step": 29708 + }, + { + "epoch": 2.4498227096561394, + "grad_norm": 2.0281622409820557, + "learning_rate": 1e-06, + "loss": 0.5386, + "mean_token_accuracy": 0.8208680152893066, + "num_tokens": 783043437.0, + "step": 29709 + }, + { + "epoch": 2.4499051702811907, + "grad_norm": 2.088953971862793, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.7890459895133972, + "num_tokens": 783074809.0, + "step": 29710 + }, + { + "epoch": 2.449987630906242, + "grad_norm": 1.9449396133422852, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7736903429031372, + "num_tokens": 783107486.0, + "step": 29711 + }, + { + "epoch": 2.450070091531294, + "grad_norm": 2.128425359725952, + "learning_rate": 1e-06, + "loss": 0.5994, + "mean_token_accuracy": 0.8041445016860962, + "num_tokens": 783135311.0, + "step": 29712 + }, + { + "epoch": 2.450152552156345, + "grad_norm": 2.2248103618621826, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7694121599197388, + "num_tokens": 783166233.0, + "step": 29713 + }, + { + "epoch": 2.450235012781397, + "grad_norm": 1.8814127445220947, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.744319498538971, + "num_tokens": 783202907.0, + "step": 29714 + }, + { + "epoch": 2.4503174734064483, + "grad_norm": 2.002021312713623, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.772388756275177, + "num_tokens": 783236183.0, + "step": 29715 + }, + { + "epoch": 2.4503999340315, + "grad_norm": 2.2259514331817627, + "learning_rate": 1e-06, + "loss": 0.6357, + "mean_token_accuracy": 0.799881100654602, + "num_tokens": 783267728.0, + "step": 29716 + }, + { + "epoch": 2.4504823946565515, + "grad_norm": 1.9464300870895386, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.7921185493469238, + "num_tokens": 783300379.0, + "step": 29717 + }, + { + "epoch": 2.450564855281603, + "grad_norm": 2.297386884689331, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.765060544013977, + "num_tokens": 783323247.0, + "step": 29718 + }, + { + "epoch": 2.4506473159066546, + "grad_norm": 2.052826404571533, + "learning_rate": 1e-06, + "loss": 0.5407, + "mean_token_accuracy": 0.8196285367012024, + "num_tokens": 783353043.0, + "step": 29719 + }, + { + "epoch": 2.450729776531706, + "grad_norm": 1.9096994400024414, + "learning_rate": 1e-06, + "loss": 0.5893, + "mean_token_accuracy": 0.7954246997833252, + "num_tokens": 783385881.0, + "step": 29720 + }, + { + "epoch": 2.450812237156758, + "grad_norm": 1.8405392169952393, + "learning_rate": 1e-06, + "loss": 0.7396, + "mean_token_accuracy": 0.7708919048309326, + "num_tokens": 783422169.0, + "step": 29721 + }, + { + "epoch": 2.450894697781809, + "grad_norm": 2.210631847381592, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7480427622795105, + "num_tokens": 783450414.0, + "step": 29722 + }, + { + "epoch": 2.450977158406861, + "grad_norm": 2.239528179168701, + "learning_rate": 1e-06, + "loss": 0.5412, + "mean_token_accuracy": 0.8114785552024841, + "num_tokens": 783473025.0, + "step": 29723 + }, + { + "epoch": 2.4510596190319123, + "grad_norm": 2.1249091625213623, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.764070987701416, + "num_tokens": 783502803.0, + "step": 29724 + }, + { + "epoch": 2.4511420796569636, + "grad_norm": 2.017897605895996, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.775924563407898, + "num_tokens": 783537016.0, + "step": 29725 + }, + { + "epoch": 2.4512245402820154, + "grad_norm": 2.4577746391296387, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7614015340805054, + "num_tokens": 783559505.0, + "step": 29726 + }, + { + "epoch": 2.4513070009070668, + "grad_norm": 1.9601919651031494, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.7990907430648804, + "num_tokens": 783593406.0, + "step": 29727 + }, + { + "epoch": 2.4513894615321186, + "grad_norm": 2.394325017929077, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7743713855743408, + "num_tokens": 783621050.0, + "step": 29728 + }, + { + "epoch": 2.45147192215717, + "grad_norm": 1.861912488937378, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.7913956642150879, + "num_tokens": 783654916.0, + "step": 29729 + }, + { + "epoch": 2.4515543827822217, + "grad_norm": 2.2077951431274414, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7691651582717896, + "num_tokens": 783681956.0, + "step": 29730 + }, + { + "epoch": 2.451636843407273, + "grad_norm": 1.8774094581604004, + "learning_rate": 1e-06, + "loss": 0.5152, + "mean_token_accuracy": 0.825020432472229, + "num_tokens": 783714988.0, + "step": 29731 + }, + { + "epoch": 2.4517193040323244, + "grad_norm": 1.831738829612732, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.7874297499656677, + "num_tokens": 783749582.0, + "step": 29732 + }, + { + "epoch": 2.451801764657376, + "grad_norm": 2.0477097034454346, + "learning_rate": 1e-06, + "loss": 0.6285, + "mean_token_accuracy": 0.79544597864151, + "num_tokens": 783782271.0, + "step": 29733 + }, + { + "epoch": 2.4518842252824276, + "grad_norm": 2.1643848419189453, + "learning_rate": 1e-06, + "loss": 0.7824, + "mean_token_accuracy": 0.752116322517395, + "num_tokens": 783812285.0, + "step": 29734 + }, + { + "epoch": 2.4519666859074793, + "grad_norm": 2.197584390640259, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.789035439491272, + "num_tokens": 783840294.0, + "step": 29735 + }, + { + "epoch": 2.4520491465325307, + "grad_norm": 2.170872926712036, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.744329035282135, + "num_tokens": 783868454.0, + "step": 29736 + }, + { + "epoch": 2.4521316071575825, + "grad_norm": 2.0504038333892822, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7726567983627319, + "num_tokens": 783896789.0, + "step": 29737 + }, + { + "epoch": 2.452214067782634, + "grad_norm": 2.0227465629577637, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.7771788835525513, + "num_tokens": 783929083.0, + "step": 29738 + }, + { + "epoch": 2.452296528407685, + "grad_norm": 1.9952737092971802, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7657665014266968, + "num_tokens": 783961294.0, + "step": 29739 + }, + { + "epoch": 2.452378989032737, + "grad_norm": 2.089008331298828, + "learning_rate": 1e-06, + "loss": 0.5933, + "mean_token_accuracy": 0.8029356598854065, + "num_tokens": 783987703.0, + "step": 29740 + }, + { + "epoch": 2.4524614496577883, + "grad_norm": 1.8904505968093872, + "learning_rate": 1e-06, + "loss": 0.5608, + "mean_token_accuracy": 0.8026875257492065, + "num_tokens": 784019683.0, + "step": 29741 + }, + { + "epoch": 2.45254391028284, + "grad_norm": 2.055903434753418, + "learning_rate": 1e-06, + "loss": 0.5857, + "mean_token_accuracy": 0.8021867275238037, + "num_tokens": 784051307.0, + "step": 29742 + }, + { + "epoch": 2.4526263709078915, + "grad_norm": 2.0033750534057617, + "learning_rate": 1e-06, + "loss": 0.5497, + "mean_token_accuracy": 0.8149415254592896, + "num_tokens": 784084154.0, + "step": 29743 + }, + { + "epoch": 2.452708831532943, + "grad_norm": 1.9754152297973633, + "learning_rate": 1e-06, + "loss": 0.531, + "mean_token_accuracy": 0.8160207867622375, + "num_tokens": 784117238.0, + "step": 29744 + }, + { + "epoch": 2.4527912921579946, + "grad_norm": 2.2265677452087402, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.782761812210083, + "num_tokens": 784145603.0, + "step": 29745 + }, + { + "epoch": 2.452873752783046, + "grad_norm": 2.156158447265625, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7825700044631958, + "num_tokens": 784174396.0, + "step": 29746 + }, + { + "epoch": 2.4529562134080978, + "grad_norm": 2.410975217819214, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.753459095954895, + "num_tokens": 784198357.0, + "step": 29747 + }, + { + "epoch": 2.453038674033149, + "grad_norm": 2.104541540145874, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.7848482131958008, + "num_tokens": 784227923.0, + "step": 29748 + }, + { + "epoch": 2.4531211346582005, + "grad_norm": 1.9901593923568726, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.742680013179779, + "num_tokens": 784263411.0, + "step": 29749 + }, + { + "epoch": 2.4532035952832523, + "grad_norm": 1.987285852432251, + "learning_rate": 1e-06, + "loss": 0.5609, + "mean_token_accuracy": 0.8126466274261475, + "num_tokens": 784292555.0, + "step": 29750 + }, + { + "epoch": 2.4532860559083036, + "grad_norm": 2.618773937225342, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.734350860118866, + "num_tokens": 784314721.0, + "step": 29751 + }, + { + "epoch": 2.4533685165333554, + "grad_norm": 2.2597153186798096, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7529676556587219, + "num_tokens": 784342928.0, + "step": 29752 + }, + { + "epoch": 2.4534509771584068, + "grad_norm": 2.303621530532837, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7831202149391174, + "num_tokens": 784368685.0, + "step": 29753 + }, + { + "epoch": 2.4535334377834586, + "grad_norm": 2.093297243118286, + "learning_rate": 1e-06, + "loss": 0.6794, + "mean_token_accuracy": 0.7752704620361328, + "num_tokens": 784398137.0, + "step": 29754 + }, + { + "epoch": 2.45361589840851, + "grad_norm": 2.3816840648651123, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.7750046253204346, + "num_tokens": 784420669.0, + "step": 29755 + }, + { + "epoch": 2.4536983590335613, + "grad_norm": 2.0204291343688965, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7690873742103577, + "num_tokens": 784452373.0, + "step": 29756 + }, + { + "epoch": 2.453780819658613, + "grad_norm": 2.4350643157958984, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7354609966278076, + "num_tokens": 784478450.0, + "step": 29757 + }, + { + "epoch": 2.4538632802836644, + "grad_norm": 2.1642181873321533, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7796480059623718, + "num_tokens": 784506145.0, + "step": 29758 + }, + { + "epoch": 2.453945740908716, + "grad_norm": 1.9376375675201416, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7646125555038452, + "num_tokens": 784539586.0, + "step": 29759 + }, + { + "epoch": 2.4540282015337676, + "grad_norm": 1.8302428722381592, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.792443037033081, + "num_tokens": 784577738.0, + "step": 29760 + }, + { + "epoch": 2.4541106621588193, + "grad_norm": 2.093726873397827, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7665705680847168, + "num_tokens": 784607935.0, + "step": 29761 + }, + { + "epoch": 2.4541931227838707, + "grad_norm": 2.2363169193267822, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7628682255744934, + "num_tokens": 784633936.0, + "step": 29762 + }, + { + "epoch": 2.454275583408922, + "grad_norm": 2.188325881958008, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7636663317680359, + "num_tokens": 784663537.0, + "step": 29763 + }, + { + "epoch": 2.454358044033974, + "grad_norm": 2.401906728744507, + "learning_rate": 1e-06, + "loss": 0.5992, + "mean_token_accuracy": 0.799400806427002, + "num_tokens": 784686007.0, + "step": 29764 + }, + { + "epoch": 2.454440504659025, + "grad_norm": 1.9791942834854126, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7631433606147766, + "num_tokens": 784723359.0, + "step": 29765 + }, + { + "epoch": 2.454522965284077, + "grad_norm": 1.9976067543029785, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7679983377456665, + "num_tokens": 784753687.0, + "step": 29766 + }, + { + "epoch": 2.4546054259091283, + "grad_norm": 2.5599570274353027, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7208995223045349, + "num_tokens": 784776540.0, + "step": 29767 + }, + { + "epoch": 2.45468788653418, + "grad_norm": 2.0130045413970947, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7713907361030579, + "num_tokens": 784811342.0, + "step": 29768 + }, + { + "epoch": 2.4547703471592315, + "grad_norm": 2.061100959777832, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7708487510681152, + "num_tokens": 784840515.0, + "step": 29769 + }, + { + "epoch": 2.454852807784283, + "grad_norm": 2.360602855682373, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7589268684387207, + "num_tokens": 784866971.0, + "step": 29770 + }, + { + "epoch": 2.4549352684093346, + "grad_norm": 2.4089126586914062, + "learning_rate": 1e-06, + "loss": 0.6026, + "mean_token_accuracy": 0.8019084334373474, + "num_tokens": 784891799.0, + "step": 29771 + }, + { + "epoch": 2.455017729034386, + "grad_norm": 2.256673812866211, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.7960003614425659, + "num_tokens": 784916030.0, + "step": 29772 + }, + { + "epoch": 2.4551001896594378, + "grad_norm": 2.289278507232666, + "learning_rate": 1e-06, + "loss": 0.6144, + "mean_token_accuracy": 0.795979917049408, + "num_tokens": 784939744.0, + "step": 29773 + }, + { + "epoch": 2.455182650284489, + "grad_norm": 2.0683200359344482, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7328658103942871, + "num_tokens": 784972496.0, + "step": 29774 + }, + { + "epoch": 2.455265110909541, + "grad_norm": 2.3507797718048096, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7737962007522583, + "num_tokens": 785000463.0, + "step": 29775 + }, + { + "epoch": 2.4553475715345923, + "grad_norm": 2.2589070796966553, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7323701977729797, + "num_tokens": 785025998.0, + "step": 29776 + }, + { + "epoch": 2.4554300321596436, + "grad_norm": 1.9952841997146606, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7435429096221924, + "num_tokens": 785060701.0, + "step": 29777 + }, + { + "epoch": 2.4555124927846954, + "grad_norm": 2.260589122772217, + "learning_rate": 1e-06, + "loss": 0.6262, + "mean_token_accuracy": 0.7853895425796509, + "num_tokens": 785088597.0, + "step": 29778 + }, + { + "epoch": 2.4555949534097468, + "grad_norm": 2.0468318462371826, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.7912694811820984, + "num_tokens": 785121043.0, + "step": 29779 + }, + { + "epoch": 2.4556774140347986, + "grad_norm": 2.034201145172119, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.770592212677002, + "num_tokens": 785153720.0, + "step": 29780 + }, + { + "epoch": 2.45575987465985, + "grad_norm": 2.265828847885132, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7700773477554321, + "num_tokens": 785181170.0, + "step": 29781 + }, + { + "epoch": 2.4558423352849017, + "grad_norm": 2.1837942600250244, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7449578046798706, + "num_tokens": 785210098.0, + "step": 29782 + }, + { + "epoch": 2.455924795909953, + "grad_norm": 2.050583839416504, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7654756307601929, + "num_tokens": 785241407.0, + "step": 29783 + }, + { + "epoch": 2.4560072565350044, + "grad_norm": 2.1300694942474365, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7735408544540405, + "num_tokens": 785269782.0, + "step": 29784 + }, + { + "epoch": 2.456089717160056, + "grad_norm": 2.072458505630493, + "learning_rate": 1e-06, + "loss": 0.6528, + "mean_token_accuracy": 0.7869234681129456, + "num_tokens": 785298819.0, + "step": 29785 + }, + { + "epoch": 2.4561721777851075, + "grad_norm": 2.2000255584716797, + "learning_rate": 1e-06, + "loss": 0.7475, + "mean_token_accuracy": 0.7566328048706055, + "num_tokens": 785328839.0, + "step": 29786 + }, + { + "epoch": 2.4562546384101593, + "grad_norm": 1.7313807010650635, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7638791799545288, + "num_tokens": 785369082.0, + "step": 29787 + }, + { + "epoch": 2.4563370990352107, + "grad_norm": 2.0943408012390137, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7788935303688049, + "num_tokens": 785396389.0, + "step": 29788 + }, + { + "epoch": 2.456419559660262, + "grad_norm": 2.24487566947937, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.7831608057022095, + "num_tokens": 785425023.0, + "step": 29789 + }, + { + "epoch": 2.456502020285314, + "grad_norm": 1.9754923582077026, + "learning_rate": 1e-06, + "loss": 0.6014, + "mean_token_accuracy": 0.799562394618988, + "num_tokens": 785455454.0, + "step": 29790 + }, + { + "epoch": 2.456584480910365, + "grad_norm": 2.4325950145721436, + "learning_rate": 1e-06, + "loss": 0.5971, + "mean_token_accuracy": 0.7955126166343689, + "num_tokens": 785478027.0, + "step": 29791 + }, + { + "epoch": 2.456666941535417, + "grad_norm": 1.9930065870285034, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7580801248550415, + "num_tokens": 785510028.0, + "step": 29792 + }, + { + "epoch": 2.4567494021604683, + "grad_norm": 2.0102827548980713, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.7844180464744568, + "num_tokens": 785543975.0, + "step": 29793 + }, + { + "epoch": 2.4568318627855197, + "grad_norm": 2.3181862831115723, + "learning_rate": 1e-06, + "loss": 0.5531, + "mean_token_accuracy": 0.8115546703338623, + "num_tokens": 785569487.0, + "step": 29794 + }, + { + "epoch": 2.4569143234105715, + "grad_norm": 2.2505040168762207, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.7776634693145752, + "num_tokens": 785595604.0, + "step": 29795 + }, + { + "epoch": 2.456996784035623, + "grad_norm": 2.075075149536133, + "learning_rate": 1e-06, + "loss": 0.6019, + "mean_token_accuracy": 0.8022849559783936, + "num_tokens": 785624221.0, + "step": 29796 + }, + { + "epoch": 2.4570792446606746, + "grad_norm": 2.1884310245513916, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.7818343043327332, + "num_tokens": 785653552.0, + "step": 29797 + }, + { + "epoch": 2.457161705285726, + "grad_norm": 1.9698032140731812, + "learning_rate": 1e-06, + "loss": 0.5978, + "mean_token_accuracy": 0.8041667938232422, + "num_tokens": 785687439.0, + "step": 29798 + }, + { + "epoch": 2.4572441659107778, + "grad_norm": 2.279538631439209, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7716997861862183, + "num_tokens": 785715340.0, + "step": 29799 + }, + { + "epoch": 2.457326626535829, + "grad_norm": 1.9412776231765747, + "learning_rate": 1e-06, + "loss": 0.6013, + "mean_token_accuracy": 0.795767068862915, + "num_tokens": 785745986.0, + "step": 29800 + }, + { + "epoch": 2.4574090871608805, + "grad_norm": 2.1468634605407715, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.7857362031936646, + "num_tokens": 785773501.0, + "step": 29801 + }, + { + "epoch": 2.4574915477859323, + "grad_norm": 2.3603625297546387, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7645639181137085, + "num_tokens": 785799500.0, + "step": 29802 + }, + { + "epoch": 2.4575740084109836, + "grad_norm": 2.0226521492004395, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7675524353981018, + "num_tokens": 785830925.0, + "step": 29803 + }, + { + "epoch": 2.4576564690360354, + "grad_norm": 2.35323429107666, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7572121620178223, + "num_tokens": 785857291.0, + "step": 29804 + }, + { + "epoch": 2.4577389296610868, + "grad_norm": 2.1957926750183105, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7360340356826782, + "num_tokens": 785885659.0, + "step": 29805 + }, + { + "epoch": 2.4578213902861386, + "grad_norm": 2.3594400882720947, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.7836645841598511, + "num_tokens": 785912783.0, + "step": 29806 + }, + { + "epoch": 2.45790385091119, + "grad_norm": 2.1101577281951904, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7667778730392456, + "num_tokens": 785944045.0, + "step": 29807 + }, + { + "epoch": 2.4579863115362413, + "grad_norm": 1.992122769355774, + "learning_rate": 1e-06, + "loss": 0.5821, + "mean_token_accuracy": 0.8105911612510681, + "num_tokens": 785974541.0, + "step": 29808 + }, + { + "epoch": 2.458068772161293, + "grad_norm": 2.181288003921509, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7798290848731995, + "num_tokens": 786002058.0, + "step": 29809 + }, + { + "epoch": 2.4581512327863444, + "grad_norm": 2.0114552974700928, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7250662446022034, + "num_tokens": 786035136.0, + "step": 29810 + }, + { + "epoch": 2.458233693411396, + "grad_norm": 1.8498820066452026, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7652072906494141, + "num_tokens": 786075657.0, + "step": 29811 + }, + { + "epoch": 2.4583161540364475, + "grad_norm": 2.2455050945281982, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.750133752822876, + "num_tokens": 786107468.0, + "step": 29812 + }, + { + "epoch": 2.4583986146614993, + "grad_norm": 2.2791683673858643, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7702521681785583, + "num_tokens": 786135369.0, + "step": 29813 + }, + { + "epoch": 2.4584810752865507, + "grad_norm": 1.9574620723724365, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7682977914810181, + "num_tokens": 786168959.0, + "step": 29814 + }, + { + "epoch": 2.458563535911602, + "grad_norm": 2.0783283710479736, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7714802622795105, + "num_tokens": 786199159.0, + "step": 29815 + }, + { + "epoch": 2.458645996536654, + "grad_norm": 2.0845980644226074, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7788417935371399, + "num_tokens": 786230726.0, + "step": 29816 + }, + { + "epoch": 2.458728457161705, + "grad_norm": 2.1014819145202637, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7630532383918762, + "num_tokens": 786259576.0, + "step": 29817 + }, + { + "epoch": 2.458810917786757, + "grad_norm": 2.0069189071655273, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7623914480209351, + "num_tokens": 786290035.0, + "step": 29818 + }, + { + "epoch": 2.4588933784118083, + "grad_norm": 2.4138195514678955, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7512067556381226, + "num_tokens": 786314233.0, + "step": 29819 + }, + { + "epoch": 2.45897583903686, + "grad_norm": 2.363917589187622, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7717714309692383, + "num_tokens": 786341877.0, + "step": 29820 + }, + { + "epoch": 2.4590582996619115, + "grad_norm": 2.063722848892212, + "learning_rate": 1e-06, + "loss": 0.5854, + "mean_token_accuracy": 0.809783399105072, + "num_tokens": 786369915.0, + "step": 29821 + }, + { + "epoch": 2.459140760286963, + "grad_norm": 2.2166683673858643, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7352213859558105, + "num_tokens": 786397347.0, + "step": 29822 + }, + { + "epoch": 2.4592232209120146, + "grad_norm": 1.8524479866027832, + "learning_rate": 1e-06, + "loss": 0.5375, + "mean_token_accuracy": 0.8245187401771545, + "num_tokens": 786433169.0, + "step": 29823 + }, + { + "epoch": 2.459305681537066, + "grad_norm": 1.7548431158065796, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7565245628356934, + "num_tokens": 786474274.0, + "step": 29824 + }, + { + "epoch": 2.4593881421621178, + "grad_norm": 2.2142624855041504, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7820755243301392, + "num_tokens": 786500286.0, + "step": 29825 + }, + { + "epoch": 2.459470602787169, + "grad_norm": 2.2742974758148193, + "learning_rate": 1e-06, + "loss": 0.775, + "mean_token_accuracy": 0.7485014200210571, + "num_tokens": 786529138.0, + "step": 29826 + }, + { + "epoch": 2.459553063412221, + "grad_norm": 2.4205267429351807, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.747458815574646, + "num_tokens": 786555119.0, + "step": 29827 + }, + { + "epoch": 2.4596355240372723, + "grad_norm": 2.0584845542907715, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7854626774787903, + "num_tokens": 786585565.0, + "step": 29828 + }, + { + "epoch": 2.4597179846623236, + "grad_norm": 2.2273519039154053, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.7965099811553955, + "num_tokens": 786612685.0, + "step": 29829 + }, + { + "epoch": 2.4598004452873754, + "grad_norm": 2.1239538192749023, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7793440222740173, + "num_tokens": 786643114.0, + "step": 29830 + }, + { + "epoch": 2.4598829059124268, + "grad_norm": 2.0459213256835938, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.7831041216850281, + "num_tokens": 786674087.0, + "step": 29831 + }, + { + "epoch": 2.4599653665374785, + "grad_norm": 2.350771427154541, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7759398818016052, + "num_tokens": 786698783.0, + "step": 29832 + }, + { + "epoch": 2.46004782716253, + "grad_norm": 1.966968297958374, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.777532696723938, + "num_tokens": 786735101.0, + "step": 29833 + }, + { + "epoch": 2.4601302877875812, + "grad_norm": 1.834867238998413, + "learning_rate": 1e-06, + "loss": 0.6878, + "mean_token_accuracy": 0.7757106423377991, + "num_tokens": 786771028.0, + "step": 29834 + }, + { + "epoch": 2.460212748412633, + "grad_norm": 2.107569694519043, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7551103830337524, + "num_tokens": 786802911.0, + "step": 29835 + }, + { + "epoch": 2.4602952090376844, + "grad_norm": 2.0413014888763428, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7820149660110474, + "num_tokens": 786832248.0, + "step": 29836 + }, + { + "epoch": 2.460377669662736, + "grad_norm": 1.9608091115951538, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.7902500629425049, + "num_tokens": 786863183.0, + "step": 29837 + }, + { + "epoch": 2.4604601302877875, + "grad_norm": 2.292480945587158, + "learning_rate": 1e-06, + "loss": 0.5901, + "mean_token_accuracy": 0.7912063002586365, + "num_tokens": 786888329.0, + "step": 29838 + }, + { + "epoch": 2.460542590912839, + "grad_norm": 2.2618279457092285, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.7877712845802307, + "num_tokens": 786914711.0, + "step": 29839 + }, + { + "epoch": 2.4606250515378907, + "grad_norm": 1.9072765111923218, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7489801645278931, + "num_tokens": 786951920.0, + "step": 29840 + }, + { + "epoch": 2.460707512162942, + "grad_norm": 2.0952630043029785, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.7784804105758667, + "num_tokens": 786981559.0, + "step": 29841 + }, + { + "epoch": 2.460789972787994, + "grad_norm": 2.180051565170288, + "learning_rate": 1e-06, + "loss": 0.5908, + "mean_token_accuracy": 0.8102396130561829, + "num_tokens": 787007905.0, + "step": 29842 + }, + { + "epoch": 2.460872433413045, + "grad_norm": 2.2652270793914795, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7690181732177734, + "num_tokens": 787036895.0, + "step": 29843 + }, + { + "epoch": 2.460954894038097, + "grad_norm": 2.1356658935546875, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7714273929595947, + "num_tokens": 787064171.0, + "step": 29844 + }, + { + "epoch": 2.4610373546631483, + "grad_norm": 2.054328203201294, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.7625088095664978, + "num_tokens": 787100513.0, + "step": 29845 + }, + { + "epoch": 2.4611198152881997, + "grad_norm": 2.39322566986084, + "learning_rate": 1e-06, + "loss": 0.5745, + "mean_token_accuracy": 0.801432728767395, + "num_tokens": 787124081.0, + "step": 29846 + }, + { + "epoch": 2.4612022759132515, + "grad_norm": 2.0563533306121826, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.772323489189148, + "num_tokens": 787151808.0, + "step": 29847 + }, + { + "epoch": 2.461284736538303, + "grad_norm": 2.247960329055786, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.7806553840637207, + "num_tokens": 787178870.0, + "step": 29848 + }, + { + "epoch": 2.4613671971633546, + "grad_norm": 1.9224812984466553, + "learning_rate": 1e-06, + "loss": 0.5198, + "mean_token_accuracy": 0.8279730081558228, + "num_tokens": 787214211.0, + "step": 29849 + }, + { + "epoch": 2.461449657788406, + "grad_norm": 2.156735897064209, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7853140830993652, + "num_tokens": 787242253.0, + "step": 29850 + }, + { + "epoch": 2.4615321184134578, + "grad_norm": 2.120441436767578, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.7918835878372192, + "num_tokens": 787273587.0, + "step": 29851 + }, + { + "epoch": 2.461614579038509, + "grad_norm": 2.147000551223755, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7776318788528442, + "num_tokens": 787305570.0, + "step": 29852 + }, + { + "epoch": 2.4616970396635605, + "grad_norm": 2.1868810653686523, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.7844753861427307, + "num_tokens": 787332810.0, + "step": 29853 + }, + { + "epoch": 2.4617795002886123, + "grad_norm": 1.929491639137268, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7726982831954956, + "num_tokens": 787367322.0, + "step": 29854 + }, + { + "epoch": 2.4618619609136636, + "grad_norm": 2.2210564613342285, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7465434670448303, + "num_tokens": 787393245.0, + "step": 29855 + }, + { + "epoch": 2.4619444215387154, + "grad_norm": 1.611579418182373, + "learning_rate": 1e-06, + "loss": 0.5915, + "mean_token_accuracy": 0.8017947673797607, + "num_tokens": 787436746.0, + "step": 29856 + }, + { + "epoch": 2.4620268821637668, + "grad_norm": 2.329648971557617, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7626155614852905, + "num_tokens": 787463735.0, + "step": 29857 + }, + { + "epoch": 2.4621093427888185, + "grad_norm": 2.1464476585388184, + "learning_rate": 1e-06, + "loss": 0.7128, + "mean_token_accuracy": 0.7634292840957642, + "num_tokens": 787491589.0, + "step": 29858 + }, + { + "epoch": 2.46219180341387, + "grad_norm": 2.453784227371216, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7748847007751465, + "num_tokens": 787514022.0, + "step": 29859 + }, + { + "epoch": 2.4622742640389212, + "grad_norm": 2.136887311935425, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7761863470077515, + "num_tokens": 787543416.0, + "step": 29860 + }, + { + "epoch": 2.462356724663973, + "grad_norm": 2.0793097019195557, + "learning_rate": 1e-06, + "loss": 0.7723, + "mean_token_accuracy": 0.7460557222366333, + "num_tokens": 787574233.0, + "step": 29861 + }, + { + "epoch": 2.4624391852890244, + "grad_norm": 1.9889417886734009, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7685887813568115, + "num_tokens": 787608371.0, + "step": 29862 + }, + { + "epoch": 2.462521645914076, + "grad_norm": 2.190782070159912, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.7792270183563232, + "num_tokens": 787636641.0, + "step": 29863 + }, + { + "epoch": 2.4626041065391275, + "grad_norm": 1.9169191122055054, + "learning_rate": 1e-06, + "loss": 0.5866, + "mean_token_accuracy": 0.8058875799179077, + "num_tokens": 787671052.0, + "step": 29864 + }, + { + "epoch": 2.4626865671641793, + "grad_norm": 1.9399319887161255, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.7837411761283875, + "num_tokens": 787704293.0, + "step": 29865 + }, + { + "epoch": 2.4627690277892307, + "grad_norm": 2.100745677947998, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.744328498840332, + "num_tokens": 787733372.0, + "step": 29866 + }, + { + "epoch": 2.462851488414282, + "grad_norm": 2.0902719497680664, + "learning_rate": 1e-06, + "loss": 0.5306, + "mean_token_accuracy": 0.8227813839912415, + "num_tokens": 787759107.0, + "step": 29867 + }, + { + "epoch": 2.462933949039334, + "grad_norm": 2.3557941913604736, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7704204320907593, + "num_tokens": 787783582.0, + "step": 29868 + }, + { + "epoch": 2.463016409664385, + "grad_norm": 2.0393786430358887, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7574405670166016, + "num_tokens": 787815692.0, + "step": 29869 + }, + { + "epoch": 2.463098870289437, + "grad_norm": 2.0796754360198975, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7666963934898376, + "num_tokens": 787847038.0, + "step": 29870 + }, + { + "epoch": 2.4631813309144883, + "grad_norm": 2.019212245941162, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7690621614456177, + "num_tokens": 787881358.0, + "step": 29871 + }, + { + "epoch": 2.46326379153954, + "grad_norm": 1.9891819953918457, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.7874999046325684, + "num_tokens": 787912821.0, + "step": 29872 + }, + { + "epoch": 2.4633462521645915, + "grad_norm": 2.1644248962402344, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7678755521774292, + "num_tokens": 787942881.0, + "step": 29873 + }, + { + "epoch": 2.463428712789643, + "grad_norm": 2.1316728591918945, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.7796365022659302, + "num_tokens": 787972225.0, + "step": 29874 + }, + { + "epoch": 2.4635111734146946, + "grad_norm": 2.280987024307251, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.763949990272522, + "num_tokens": 787999464.0, + "step": 29875 + }, + { + "epoch": 2.463593634039746, + "grad_norm": 2.0740041732788086, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.7755531072616577, + "num_tokens": 788032227.0, + "step": 29876 + }, + { + "epoch": 2.4636760946647973, + "grad_norm": 1.955102801322937, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7834745645523071, + "num_tokens": 788061368.0, + "step": 29877 + }, + { + "epoch": 2.463758555289849, + "grad_norm": 2.318845510482788, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.7825376987457275, + "num_tokens": 788085123.0, + "step": 29878 + }, + { + "epoch": 2.4638410159149005, + "grad_norm": 2.129122257232666, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.7897231578826904, + "num_tokens": 788113199.0, + "step": 29879 + }, + { + "epoch": 2.4639234765399523, + "grad_norm": 1.9411782026290894, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.7844103574752808, + "num_tokens": 788145847.0, + "step": 29880 + }, + { + "epoch": 2.4640059371650036, + "grad_norm": 1.9795575141906738, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7634743452072144, + "num_tokens": 788179186.0, + "step": 29881 + }, + { + "epoch": 2.4640883977900554, + "grad_norm": 2.0426201820373535, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.747245728969574, + "num_tokens": 788211165.0, + "step": 29882 + }, + { + "epoch": 2.4641708584151067, + "grad_norm": 2.066472053527832, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.7527490258216858, + "num_tokens": 788243443.0, + "step": 29883 + }, + { + "epoch": 2.464253319040158, + "grad_norm": 2.153386354446411, + "learning_rate": 1e-06, + "loss": 0.5185, + "mean_token_accuracy": 0.8190267086029053, + "num_tokens": 788270197.0, + "step": 29884 + }, + { + "epoch": 2.46433577966521, + "grad_norm": 2.150242567062378, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7656145095825195, + "num_tokens": 788301899.0, + "step": 29885 + }, + { + "epoch": 2.4644182402902612, + "grad_norm": 2.0476698875427246, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7674064636230469, + "num_tokens": 788332480.0, + "step": 29886 + }, + { + "epoch": 2.464500700915313, + "grad_norm": 1.9645335674285889, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7645207047462463, + "num_tokens": 788365107.0, + "step": 29887 + }, + { + "epoch": 2.4645831615403644, + "grad_norm": 1.916684627532959, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7912824153900146, + "num_tokens": 788398202.0, + "step": 29888 + }, + { + "epoch": 2.464665622165416, + "grad_norm": 2.2915632724761963, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.781540036201477, + "num_tokens": 788424037.0, + "step": 29889 + }, + { + "epoch": 2.4647480827904675, + "grad_norm": 1.87775456905365, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7250688076019287, + "num_tokens": 788462060.0, + "step": 29890 + }, + { + "epoch": 2.464830543415519, + "grad_norm": 2.333544969558716, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.747815728187561, + "num_tokens": 788492092.0, + "step": 29891 + }, + { + "epoch": 2.4649130040405707, + "grad_norm": 2.391380548477173, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7763367891311646, + "num_tokens": 788515378.0, + "step": 29892 + }, + { + "epoch": 2.464995464665622, + "grad_norm": 2.063673257827759, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7232677936553955, + "num_tokens": 788549006.0, + "step": 29893 + }, + { + "epoch": 2.465077925290674, + "grad_norm": 2.283381700515747, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7537562251091003, + "num_tokens": 788574824.0, + "step": 29894 + }, + { + "epoch": 2.465160385915725, + "grad_norm": 1.963295817375183, + "learning_rate": 1e-06, + "loss": 0.5615, + "mean_token_accuracy": 0.8154040575027466, + "num_tokens": 788605193.0, + "step": 29895 + }, + { + "epoch": 2.465242846540777, + "grad_norm": 2.1834444999694824, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.777518630027771, + "num_tokens": 788631611.0, + "step": 29896 + }, + { + "epoch": 2.4653253071658283, + "grad_norm": 1.9695080518722534, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.7923833131790161, + "num_tokens": 788663938.0, + "step": 29897 + }, + { + "epoch": 2.4654077677908797, + "grad_norm": 2.099703550338745, + "learning_rate": 1e-06, + "loss": 0.5077, + "mean_token_accuracy": 0.8281169533729553, + "num_tokens": 788691823.0, + "step": 29898 + }, + { + "epoch": 2.4654902284159315, + "grad_norm": 2.166942596435547, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7770580053329468, + "num_tokens": 788717430.0, + "step": 29899 + }, + { + "epoch": 2.465572689040983, + "grad_norm": 2.3188765048980713, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7570583820343018, + "num_tokens": 788745386.0, + "step": 29900 + }, + { + "epoch": 2.4656551496660346, + "grad_norm": 2.0408434867858887, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7677310109138489, + "num_tokens": 788778129.0, + "step": 29901 + }, + { + "epoch": 2.465737610291086, + "grad_norm": 2.169964551925659, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7741779088973999, + "num_tokens": 788812706.0, + "step": 29902 + }, + { + "epoch": 2.4658200709161378, + "grad_norm": 2.6084084510803223, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7508513927459717, + "num_tokens": 788834235.0, + "step": 29903 + }, + { + "epoch": 2.465902531541189, + "grad_norm": 2.0719265937805176, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7777829170227051, + "num_tokens": 788865727.0, + "step": 29904 + }, + { + "epoch": 2.4659849921662405, + "grad_norm": 2.0141093730926514, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7751691341400146, + "num_tokens": 788896192.0, + "step": 29905 + }, + { + "epoch": 2.4660674527912922, + "grad_norm": 1.9971442222595215, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7764566540718079, + "num_tokens": 788928290.0, + "step": 29906 + }, + { + "epoch": 2.4661499134163436, + "grad_norm": 1.9470428228378296, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7770903706550598, + "num_tokens": 788962096.0, + "step": 29907 + }, + { + "epoch": 2.4662323740413954, + "grad_norm": 2.0981404781341553, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7398576736450195, + "num_tokens": 788992883.0, + "step": 29908 + }, + { + "epoch": 2.4663148346664467, + "grad_norm": 2.190243721008301, + "learning_rate": 1e-06, + "loss": 0.6512, + "mean_token_accuracy": 0.7718163132667542, + "num_tokens": 789019413.0, + "step": 29909 + }, + { + "epoch": 2.4663972952914985, + "grad_norm": 1.842738389968872, + "learning_rate": 1e-06, + "loss": 0.5552, + "mean_token_accuracy": 0.8101252317428589, + "num_tokens": 789053564.0, + "step": 29910 + }, + { + "epoch": 2.46647975591655, + "grad_norm": 1.8106591701507568, + "learning_rate": 1e-06, + "loss": 0.4727, + "mean_token_accuracy": 0.8256791830062866, + "num_tokens": 789088069.0, + "step": 29911 + }, + { + "epoch": 2.4665622165416012, + "grad_norm": 2.2871150970458984, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.7905117869377136, + "num_tokens": 789116550.0, + "step": 29912 + }, + { + "epoch": 2.466644677166653, + "grad_norm": 1.782520055770874, + "learning_rate": 1e-06, + "loss": 0.5163, + "mean_token_accuracy": 0.8256661891937256, + "num_tokens": 789149737.0, + "step": 29913 + }, + { + "epoch": 2.4667271377917044, + "grad_norm": 2.0916309356689453, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7633092403411865, + "num_tokens": 789179047.0, + "step": 29914 + }, + { + "epoch": 2.466809598416756, + "grad_norm": 1.9603140354156494, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.76753830909729, + "num_tokens": 789214278.0, + "step": 29915 + }, + { + "epoch": 2.4668920590418075, + "grad_norm": 1.8967524766921997, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7398725748062134, + "num_tokens": 789251531.0, + "step": 29916 + }, + { + "epoch": 2.4669745196668593, + "grad_norm": 1.9391170740127563, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7803215980529785, + "num_tokens": 789286875.0, + "step": 29917 + }, + { + "epoch": 2.4670569802919107, + "grad_norm": 1.8546119928359985, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7849869132041931, + "num_tokens": 789322900.0, + "step": 29918 + }, + { + "epoch": 2.467139440916962, + "grad_norm": 1.9810057878494263, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7485376596450806, + "num_tokens": 789355827.0, + "step": 29919 + }, + { + "epoch": 2.467221901542014, + "grad_norm": 2.0211374759674072, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.765139639377594, + "num_tokens": 789387731.0, + "step": 29920 + }, + { + "epoch": 2.467304362167065, + "grad_norm": 1.9232717752456665, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7839490175247192, + "num_tokens": 789425926.0, + "step": 29921 + }, + { + "epoch": 2.4673868227921165, + "grad_norm": 1.9639531373977661, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7490650415420532, + "num_tokens": 789459660.0, + "step": 29922 + }, + { + "epoch": 2.4674692834171683, + "grad_norm": 2.239292621612549, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7410212159156799, + "num_tokens": 789487314.0, + "step": 29923 + }, + { + "epoch": 2.4675517440422197, + "grad_norm": 1.9519221782684326, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7709124088287354, + "num_tokens": 789520359.0, + "step": 29924 + }, + { + "epoch": 2.4676342046672715, + "grad_norm": 2.0366318225860596, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7698004841804504, + "num_tokens": 789552187.0, + "step": 29925 + }, + { + "epoch": 2.467716665292323, + "grad_norm": 2.1100172996520996, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7305570244789124, + "num_tokens": 789584887.0, + "step": 29926 + }, + { + "epoch": 2.4677991259173746, + "grad_norm": 2.1606528759002686, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.7981274127960205, + "num_tokens": 789612771.0, + "step": 29927 + }, + { + "epoch": 2.467881586542426, + "grad_norm": 2.2719528675079346, + "learning_rate": 1e-06, + "loss": 0.5852, + "mean_token_accuracy": 0.8032326698303223, + "num_tokens": 789636998.0, + "step": 29928 + }, + { + "epoch": 2.4679640471674773, + "grad_norm": 2.276803731918335, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7668222188949585, + "num_tokens": 789663286.0, + "step": 29929 + }, + { + "epoch": 2.468046507792529, + "grad_norm": 2.023777723312378, + "learning_rate": 1e-06, + "loss": 0.5589, + "mean_token_accuracy": 0.8053030967712402, + "num_tokens": 789693239.0, + "step": 29930 + }, + { + "epoch": 2.4681289684175804, + "grad_norm": 2.0458829402923584, + "learning_rate": 1e-06, + "loss": 0.5594, + "mean_token_accuracy": 0.8158147931098938, + "num_tokens": 789719574.0, + "step": 29931 + }, + { + "epoch": 2.4682114290426322, + "grad_norm": 2.5886447429656982, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7807954549789429, + "num_tokens": 789741952.0, + "step": 29932 + }, + { + "epoch": 2.4682938896676836, + "grad_norm": 2.0184245109558105, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7553355693817139, + "num_tokens": 789773387.0, + "step": 29933 + }, + { + "epoch": 2.4683763502927354, + "grad_norm": 1.9878250360488892, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7576655149459839, + "num_tokens": 789807514.0, + "step": 29934 + }, + { + "epoch": 2.4684588109177867, + "grad_norm": 2.2582125663757324, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7609299421310425, + "num_tokens": 789831311.0, + "step": 29935 + }, + { + "epoch": 2.468541271542838, + "grad_norm": 1.879980444908142, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.8023053407669067, + "num_tokens": 789863292.0, + "step": 29936 + }, + { + "epoch": 2.46862373216789, + "grad_norm": 2.2161381244659424, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7741866707801819, + "num_tokens": 789889477.0, + "step": 29937 + }, + { + "epoch": 2.4687061927929412, + "grad_norm": 2.131434679031372, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7682271003723145, + "num_tokens": 789921580.0, + "step": 29938 + }, + { + "epoch": 2.468788653417993, + "grad_norm": 2.072814464569092, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.738643229007721, + "num_tokens": 789952417.0, + "step": 29939 + }, + { + "epoch": 2.4688711140430444, + "grad_norm": 2.341395616531372, + "learning_rate": 1e-06, + "loss": 0.5102, + "mean_token_accuracy": 0.8236711621284485, + "num_tokens": 789974955.0, + "step": 29940 + }, + { + "epoch": 2.468953574668096, + "grad_norm": 2.1441855430603027, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.7857797741889954, + "num_tokens": 790004200.0, + "step": 29941 + }, + { + "epoch": 2.4690360352931475, + "grad_norm": 2.102438449859619, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7645055055618286, + "num_tokens": 790035364.0, + "step": 29942 + }, + { + "epoch": 2.469118495918199, + "grad_norm": 2.318063259124756, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.7836465835571289, + "num_tokens": 790058521.0, + "step": 29943 + }, + { + "epoch": 2.4692009565432507, + "grad_norm": 2.04408597946167, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7811687588691711, + "num_tokens": 790089490.0, + "step": 29944 + }, + { + "epoch": 2.469283417168302, + "grad_norm": 1.9593452215194702, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7680689692497253, + "num_tokens": 790125322.0, + "step": 29945 + }, + { + "epoch": 2.469365877793354, + "grad_norm": 2.1241867542266846, + "learning_rate": 1e-06, + "loss": 0.5725, + "mean_token_accuracy": 0.8008236885070801, + "num_tokens": 790152937.0, + "step": 29946 + }, + { + "epoch": 2.469448338418405, + "grad_norm": 2.2209830284118652, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7647360563278198, + "num_tokens": 790180518.0, + "step": 29947 + }, + { + "epoch": 2.469530799043457, + "grad_norm": 1.9654325246810913, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7515292167663574, + "num_tokens": 790214768.0, + "step": 29948 + }, + { + "epoch": 2.4696132596685083, + "grad_norm": 2.041543483734131, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7559738159179688, + "num_tokens": 790246935.0, + "step": 29949 + }, + { + "epoch": 2.4696957202935597, + "grad_norm": 2.079807758331299, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7515393495559692, + "num_tokens": 790276931.0, + "step": 29950 + }, + { + "epoch": 2.4697781809186115, + "grad_norm": 2.026489496231079, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.758899450302124, + "num_tokens": 790310016.0, + "step": 29951 + }, + { + "epoch": 2.469860641543663, + "grad_norm": 2.163560152053833, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.7577365636825562, + "num_tokens": 790339800.0, + "step": 29952 + }, + { + "epoch": 2.4699431021687146, + "grad_norm": 1.8581963777542114, + "learning_rate": 1e-06, + "loss": 0.5122, + "mean_token_accuracy": 0.824051022529602, + "num_tokens": 790374551.0, + "step": 29953 + }, + { + "epoch": 2.470025562793766, + "grad_norm": 2.075427532196045, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7698699235916138, + "num_tokens": 790407391.0, + "step": 29954 + }, + { + "epoch": 2.4701080234188177, + "grad_norm": 2.2522501945495605, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7657203674316406, + "num_tokens": 790437250.0, + "step": 29955 + }, + { + "epoch": 2.470190484043869, + "grad_norm": 1.8468563556671143, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7757037878036499, + "num_tokens": 790471988.0, + "step": 29956 + }, + { + "epoch": 2.4702729446689204, + "grad_norm": 1.8508517742156982, + "learning_rate": 1e-06, + "loss": 0.5829, + "mean_token_accuracy": 0.8058074116706848, + "num_tokens": 790505200.0, + "step": 29957 + }, + { + "epoch": 2.4703554052939722, + "grad_norm": 1.9761178493499756, + "learning_rate": 1e-06, + "loss": 0.5945, + "mean_token_accuracy": 0.8021804094314575, + "num_tokens": 790537146.0, + "step": 29958 + }, + { + "epoch": 2.4704378659190236, + "grad_norm": 2.184385061264038, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7788400650024414, + "num_tokens": 790566758.0, + "step": 29959 + }, + { + "epoch": 2.4705203265440754, + "grad_norm": 1.994837760925293, + "learning_rate": 1e-06, + "loss": 0.5658, + "mean_token_accuracy": 0.8102403283119202, + "num_tokens": 790598784.0, + "step": 29960 + }, + { + "epoch": 2.4706027871691267, + "grad_norm": 2.1606204509735107, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7628635168075562, + "num_tokens": 790629375.0, + "step": 29961 + }, + { + "epoch": 2.4706852477941785, + "grad_norm": 2.1320364475250244, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7577052116394043, + "num_tokens": 790659211.0, + "step": 29962 + }, + { + "epoch": 2.47076770841923, + "grad_norm": 2.1983702182769775, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7807256579399109, + "num_tokens": 790685046.0, + "step": 29963 + }, + { + "epoch": 2.4708501690442812, + "grad_norm": 2.3758175373077393, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.7825847268104553, + "num_tokens": 790710180.0, + "step": 29964 + }, + { + "epoch": 2.470932629669333, + "grad_norm": 2.2668514251708984, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7484806776046753, + "num_tokens": 790736137.0, + "step": 29965 + }, + { + "epoch": 2.4710150902943844, + "grad_norm": 2.1316275596618652, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.768108606338501, + "num_tokens": 790764693.0, + "step": 29966 + }, + { + "epoch": 2.4710975509194357, + "grad_norm": 2.4670753479003906, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7388736605644226, + "num_tokens": 790787702.0, + "step": 29967 + }, + { + "epoch": 2.4711800115444875, + "grad_norm": 2.0435562133789062, + "learning_rate": 1e-06, + "loss": 0.6089, + "mean_token_accuracy": 0.8043800592422485, + "num_tokens": 790818879.0, + "step": 29968 + }, + { + "epoch": 2.471262472169539, + "grad_norm": 2.0299713611602783, + "learning_rate": 1e-06, + "loss": 0.592, + "mean_token_accuracy": 0.8081223964691162, + "num_tokens": 790851515.0, + "step": 29969 + }, + { + "epoch": 2.4713449327945907, + "grad_norm": 1.9169211387634277, + "learning_rate": 1e-06, + "loss": 0.5639, + "mean_token_accuracy": 0.8145023584365845, + "num_tokens": 790882363.0, + "step": 29970 + }, + { + "epoch": 2.471427393419642, + "grad_norm": 2.0766234397888184, + "learning_rate": 1e-06, + "loss": 0.5181, + "mean_token_accuracy": 0.8082320690155029, + "num_tokens": 790909453.0, + "step": 29971 + }, + { + "epoch": 2.471509854044694, + "grad_norm": 2.6382410526275635, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7744902968406677, + "num_tokens": 790928822.0, + "step": 29972 + }, + { + "epoch": 2.471592314669745, + "grad_norm": 1.933011531829834, + "learning_rate": 1e-06, + "loss": 0.554, + "mean_token_accuracy": 0.8154556751251221, + "num_tokens": 790960055.0, + "step": 29973 + }, + { + "epoch": 2.4716747752947965, + "grad_norm": 2.03045654296875, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7748973369598389, + "num_tokens": 790992915.0, + "step": 29974 + }, + { + "epoch": 2.4717572359198483, + "grad_norm": 2.1646065711975098, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.7733520269393921, + "num_tokens": 791020068.0, + "step": 29975 + }, + { + "epoch": 2.4718396965448997, + "grad_norm": 2.057461738586426, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7469602823257446, + "num_tokens": 791052019.0, + "step": 29976 + }, + { + "epoch": 2.4719221571699515, + "grad_norm": 2.066521167755127, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7450933456420898, + "num_tokens": 791083504.0, + "step": 29977 + }, + { + "epoch": 2.472004617795003, + "grad_norm": 1.9277483224868774, + "learning_rate": 1e-06, + "loss": 0.5352, + "mean_token_accuracy": 0.808178186416626, + "num_tokens": 791113953.0, + "step": 29978 + }, + { + "epoch": 2.4720870784200546, + "grad_norm": 2.301820993423462, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7639423608779907, + "num_tokens": 791137762.0, + "step": 29979 + }, + { + "epoch": 2.472169539045106, + "grad_norm": 2.006321907043457, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.7858704924583435, + "num_tokens": 791169951.0, + "step": 29980 + }, + { + "epoch": 2.4722519996701573, + "grad_norm": 2.136659622192383, + "learning_rate": 1e-06, + "loss": 0.5627, + "mean_token_accuracy": 0.8037518262863159, + "num_tokens": 791195545.0, + "step": 29981 + }, + { + "epoch": 2.472334460295209, + "grad_norm": 2.3579673767089844, + "learning_rate": 1e-06, + "loss": 0.538, + "mean_token_accuracy": 0.8175864815711975, + "num_tokens": 791219298.0, + "step": 29982 + }, + { + "epoch": 2.4724169209202604, + "grad_norm": 2.1580748558044434, + "learning_rate": 1e-06, + "loss": 0.59, + "mean_token_accuracy": 0.8003185391426086, + "num_tokens": 791244275.0, + "step": 29983 + }, + { + "epoch": 2.4724993815453122, + "grad_norm": 2.4866421222686768, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7788347005844116, + "num_tokens": 791266565.0, + "step": 29984 + }, + { + "epoch": 2.4725818421703636, + "grad_norm": 2.570340156555176, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7462223768234253, + "num_tokens": 791290467.0, + "step": 29985 + }, + { + "epoch": 2.4726643027954154, + "grad_norm": 1.932568907737732, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7601950168609619, + "num_tokens": 791328494.0, + "step": 29986 + }, + { + "epoch": 2.4727467634204667, + "grad_norm": 1.9654844999313354, + "learning_rate": 1e-06, + "loss": 0.7223, + "mean_token_accuracy": 0.7656775712966919, + "num_tokens": 791362539.0, + "step": 29987 + }, + { + "epoch": 2.472829224045518, + "grad_norm": 2.4593758583068848, + "learning_rate": 1e-06, + "loss": 0.6123, + "mean_token_accuracy": 0.7938606142997742, + "num_tokens": 791384793.0, + "step": 29988 + }, + { + "epoch": 2.47291168467057, + "grad_norm": 2.131211519241333, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7789251208305359, + "num_tokens": 791414552.0, + "step": 29989 + }, + { + "epoch": 2.4729941452956212, + "grad_norm": 2.078965425491333, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7539809942245483, + "num_tokens": 791447142.0, + "step": 29990 + }, + { + "epoch": 2.473076605920673, + "grad_norm": 1.8324410915374756, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7652907371520996, + "num_tokens": 791483956.0, + "step": 29991 + }, + { + "epoch": 2.4731590665457244, + "grad_norm": 1.851775050163269, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7167739272117615, + "num_tokens": 791523099.0, + "step": 29992 + }, + { + "epoch": 2.473241527170776, + "grad_norm": 2.0731148719787598, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.7792565822601318, + "num_tokens": 791552932.0, + "step": 29993 + }, + { + "epoch": 2.4733239877958275, + "grad_norm": 2.21067476272583, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7602147459983826, + "num_tokens": 791582812.0, + "step": 29994 + }, + { + "epoch": 2.473406448420879, + "grad_norm": 1.910264253616333, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.7906649708747864, + "num_tokens": 791615848.0, + "step": 29995 + }, + { + "epoch": 2.4734889090459307, + "grad_norm": 2.3887228965759277, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7624176740646362, + "num_tokens": 791640593.0, + "step": 29996 + }, + { + "epoch": 2.473571369670982, + "grad_norm": 2.1685397624969482, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.776906430721283, + "num_tokens": 791667535.0, + "step": 29997 + }, + { + "epoch": 2.473653830296034, + "grad_norm": 1.9618604183197021, + "learning_rate": 1e-06, + "loss": 0.557, + "mean_token_accuracy": 0.8095815181732178, + "num_tokens": 791701032.0, + "step": 29998 + }, + { + "epoch": 2.473736290921085, + "grad_norm": 1.9652607440948486, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7718602418899536, + "num_tokens": 791735511.0, + "step": 29999 + }, + { + "epoch": 2.473818751546137, + "grad_norm": 2.1028249263763428, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.8034307360649109, + "num_tokens": 791765475.0, + "step": 30000 + }, + { + "epoch": 2.4739012121711883, + "grad_norm": 2.7313973903656006, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7834807634353638, + "num_tokens": 791792300.0, + "step": 30001 + }, + { + "epoch": 2.4739836727962397, + "grad_norm": 2.2049686908721924, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7688008546829224, + "num_tokens": 791818326.0, + "step": 30002 + }, + { + "epoch": 2.4740661334212914, + "grad_norm": 1.9797658920288086, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.7836230397224426, + "num_tokens": 791850138.0, + "step": 30003 + }, + { + "epoch": 2.474148594046343, + "grad_norm": 1.8635503053665161, + "learning_rate": 1e-06, + "loss": 0.6273, + "mean_token_accuracy": 0.7903481125831604, + "num_tokens": 791887533.0, + "step": 30004 + }, + { + "epoch": 2.4742310546713946, + "grad_norm": 1.982995629310608, + "learning_rate": 1e-06, + "loss": 0.6189, + "mean_token_accuracy": 0.7969370484352112, + "num_tokens": 791921335.0, + "step": 30005 + }, + { + "epoch": 2.474313515296446, + "grad_norm": 2.3327057361602783, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7719757556915283, + "num_tokens": 791946219.0, + "step": 30006 + }, + { + "epoch": 2.4743959759214973, + "grad_norm": 2.209733009338379, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7188239097595215, + "num_tokens": 791977059.0, + "step": 30007 + }, + { + "epoch": 2.474478436546549, + "grad_norm": 2.224980354309082, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.755328893661499, + "num_tokens": 792003430.0, + "step": 30008 + }, + { + "epoch": 2.4745608971716004, + "grad_norm": 2.022881269454956, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7729251384735107, + "num_tokens": 792032637.0, + "step": 30009 + }, + { + "epoch": 2.4746433577966522, + "grad_norm": 1.9643974304199219, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7801340818405151, + "num_tokens": 792063999.0, + "step": 30010 + }, + { + "epoch": 2.4747258184217036, + "grad_norm": 1.8616154193878174, + "learning_rate": 1e-06, + "loss": 0.5278, + "mean_token_accuracy": 0.8108186721801758, + "num_tokens": 792097601.0, + "step": 30011 + }, + { + "epoch": 2.474808279046755, + "grad_norm": 1.9606959819793701, + "learning_rate": 1e-06, + "loss": 0.6053, + "mean_token_accuracy": 0.7966200709342957, + "num_tokens": 792131985.0, + "step": 30012 + }, + { + "epoch": 2.4748907396718067, + "grad_norm": 2.2344775199890137, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7583566904067993, + "num_tokens": 792158279.0, + "step": 30013 + }, + { + "epoch": 2.474973200296858, + "grad_norm": 2.116623878479004, + "learning_rate": 1e-06, + "loss": 0.5006, + "mean_token_accuracy": 0.8281798362731934, + "num_tokens": 792186157.0, + "step": 30014 + }, + { + "epoch": 2.47505566092191, + "grad_norm": 2.151944160461426, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7413832545280457, + "num_tokens": 792216875.0, + "step": 30015 + }, + { + "epoch": 2.4751381215469612, + "grad_norm": 1.9702051877975464, + "learning_rate": 1e-06, + "loss": 0.5883, + "mean_token_accuracy": 0.7945514917373657, + "num_tokens": 792251337.0, + "step": 30016 + }, + { + "epoch": 2.475220582172013, + "grad_norm": 2.1541926860809326, + "learning_rate": 1e-06, + "loss": 0.6053, + "mean_token_accuracy": 0.7927490472793579, + "num_tokens": 792281550.0, + "step": 30017 + }, + { + "epoch": 2.4753030427970644, + "grad_norm": 1.9163155555725098, + "learning_rate": 1e-06, + "loss": 0.5859, + "mean_token_accuracy": 0.8088318109512329, + "num_tokens": 792311082.0, + "step": 30018 + }, + { + "epoch": 2.4753855034221157, + "grad_norm": 1.9510293006896973, + "learning_rate": 1e-06, + "loss": 0.54, + "mean_token_accuracy": 0.8117961883544922, + "num_tokens": 792344237.0, + "step": 30019 + }, + { + "epoch": 2.4754679640471675, + "grad_norm": 2.112464189529419, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.7874643802642822, + "num_tokens": 792371343.0, + "step": 30020 + }, + { + "epoch": 2.475550424672219, + "grad_norm": 2.2493884563446045, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7467660903930664, + "num_tokens": 792398911.0, + "step": 30021 + }, + { + "epoch": 2.4756328852972707, + "grad_norm": 2.239867687225342, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.726297914981842, + "num_tokens": 792429097.0, + "step": 30022 + }, + { + "epoch": 2.475715345922322, + "grad_norm": 2.3667476177215576, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7784623503684998, + "num_tokens": 792455636.0, + "step": 30023 + }, + { + "epoch": 2.475797806547374, + "grad_norm": 1.9652323722839355, + "learning_rate": 1e-06, + "loss": 0.6133, + "mean_token_accuracy": 0.7992575764656067, + "num_tokens": 792486284.0, + "step": 30024 + }, + { + "epoch": 2.475880267172425, + "grad_norm": 2.0621941089630127, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7620604634284973, + "num_tokens": 792517022.0, + "step": 30025 + }, + { + "epoch": 2.4759627277974765, + "grad_norm": 1.9836243391036987, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.7877932786941528, + "num_tokens": 792551758.0, + "step": 30026 + }, + { + "epoch": 2.4760451884225283, + "grad_norm": 2.1109485626220703, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7447072863578796, + "num_tokens": 792581612.0, + "step": 30027 + }, + { + "epoch": 2.4761276490475796, + "grad_norm": 2.1452062129974365, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7582826614379883, + "num_tokens": 792612327.0, + "step": 30028 + }, + { + "epoch": 2.4762101096726314, + "grad_norm": 1.9256762266159058, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.777178168296814, + "num_tokens": 792647830.0, + "step": 30029 + }, + { + "epoch": 2.476292570297683, + "grad_norm": 2.0997986793518066, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.74574214220047, + "num_tokens": 792676903.0, + "step": 30030 + }, + { + "epoch": 2.4763750309227346, + "grad_norm": 2.127814292907715, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7261830568313599, + "num_tokens": 792706839.0, + "step": 30031 + }, + { + "epoch": 2.476457491547786, + "grad_norm": 1.8726869821548462, + "learning_rate": 1e-06, + "loss": 0.6115, + "mean_token_accuracy": 0.7899872660636902, + "num_tokens": 792745098.0, + "step": 30032 + }, + { + "epoch": 2.4765399521728373, + "grad_norm": 2.256131410598755, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.7752951383590698, + "num_tokens": 792770890.0, + "step": 30033 + }, + { + "epoch": 2.476622412797889, + "grad_norm": 2.1140668392181396, + "learning_rate": 1e-06, + "loss": 0.646, + "mean_token_accuracy": 0.7888267040252686, + "num_tokens": 792801370.0, + "step": 30034 + }, + { + "epoch": 2.4767048734229404, + "grad_norm": 2.197218179702759, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7503678798675537, + "num_tokens": 792831039.0, + "step": 30035 + }, + { + "epoch": 2.4767873340479922, + "grad_norm": 2.1552622318267822, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.760844349861145, + "num_tokens": 792861375.0, + "step": 30036 + }, + { + "epoch": 2.4768697946730436, + "grad_norm": 2.3433330059051514, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7537709474563599, + "num_tokens": 792887854.0, + "step": 30037 + }, + { + "epoch": 2.4769522552980954, + "grad_norm": 2.0626065731048584, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7479004859924316, + "num_tokens": 792917471.0, + "step": 30038 + }, + { + "epoch": 2.4770347159231467, + "grad_norm": 2.083864688873291, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7720367312431335, + "num_tokens": 792951997.0, + "step": 30039 + }, + { + "epoch": 2.477117176548198, + "grad_norm": 1.9276220798492432, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.802930474281311, + "num_tokens": 792986276.0, + "step": 30040 + }, + { + "epoch": 2.47719963717325, + "grad_norm": 1.9866095781326294, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7669941186904907, + "num_tokens": 793020324.0, + "step": 30041 + }, + { + "epoch": 2.477282097798301, + "grad_norm": 2.2651243209838867, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7398256659507751, + "num_tokens": 793048030.0, + "step": 30042 + }, + { + "epoch": 2.477364558423353, + "grad_norm": 2.1666171550750732, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7813013195991516, + "num_tokens": 793076609.0, + "step": 30043 + }, + { + "epoch": 2.4774470190484044, + "grad_norm": 1.9361062049865723, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7575083374977112, + "num_tokens": 793111278.0, + "step": 30044 + }, + { + "epoch": 2.477529479673456, + "grad_norm": 2.171412706375122, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.788995623588562, + "num_tokens": 793137208.0, + "step": 30045 + }, + { + "epoch": 2.4776119402985075, + "grad_norm": 2.0806994438171387, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.7578445076942444, + "num_tokens": 793166392.0, + "step": 30046 + }, + { + "epoch": 2.477694400923559, + "grad_norm": 1.9527181386947632, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.7928622961044312, + "num_tokens": 793199355.0, + "step": 30047 + }, + { + "epoch": 2.4777768615486107, + "grad_norm": 1.995877981185913, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.7958437204360962, + "num_tokens": 793228171.0, + "step": 30048 + }, + { + "epoch": 2.477859322173662, + "grad_norm": 2.102404832839966, + "learning_rate": 1e-06, + "loss": 0.5899, + "mean_token_accuracy": 0.8060265183448792, + "num_tokens": 793256602.0, + "step": 30049 + }, + { + "epoch": 2.477941782798714, + "grad_norm": 2.4533815383911133, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7381923198699951, + "num_tokens": 793280653.0, + "step": 30050 + }, + { + "epoch": 2.478024243423765, + "grad_norm": 2.1729736328125, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7709922790527344, + "num_tokens": 793313426.0, + "step": 30051 + }, + { + "epoch": 2.4781067040488165, + "grad_norm": 1.9128532409667969, + "learning_rate": 1e-06, + "loss": 0.5163, + "mean_token_accuracy": 0.8200091123580933, + "num_tokens": 793348930.0, + "step": 30052 + }, + { + "epoch": 2.4781891646738683, + "grad_norm": 2.205547571182251, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7483706474304199, + "num_tokens": 793376651.0, + "step": 30053 + }, + { + "epoch": 2.4782716252989196, + "grad_norm": 1.749063491821289, + "learning_rate": 1e-06, + "loss": 0.5326, + "mean_token_accuracy": 0.8192580938339233, + "num_tokens": 793414053.0, + "step": 30054 + }, + { + "epoch": 2.4783540859239714, + "grad_norm": 2.0808093547821045, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7582402229309082, + "num_tokens": 793445058.0, + "step": 30055 + }, + { + "epoch": 2.478436546549023, + "grad_norm": 1.7254130840301514, + "learning_rate": 1e-06, + "loss": 0.5089, + "mean_token_accuracy": 0.8294475078582764, + "num_tokens": 793485234.0, + "step": 30056 + }, + { + "epoch": 2.478519007174074, + "grad_norm": 2.034337282180786, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7168146371841431, + "num_tokens": 793519684.0, + "step": 30057 + }, + { + "epoch": 2.478601467799126, + "grad_norm": 2.191423177719116, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.779329776763916, + "num_tokens": 793547686.0, + "step": 30058 + }, + { + "epoch": 2.4786839284241773, + "grad_norm": 1.8653403520584106, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7652734518051147, + "num_tokens": 793583487.0, + "step": 30059 + }, + { + "epoch": 2.478766389049229, + "grad_norm": 2.1752994060516357, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7711471915245056, + "num_tokens": 793611224.0, + "step": 30060 + }, + { + "epoch": 2.4788488496742804, + "grad_norm": 2.098494052886963, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7606877088546753, + "num_tokens": 793642079.0, + "step": 30061 + }, + { + "epoch": 2.4789313102993322, + "grad_norm": 1.9963951110839844, + "learning_rate": 1e-06, + "loss": 0.5506, + "mean_token_accuracy": 0.8069673776626587, + "num_tokens": 793671045.0, + "step": 30062 + }, + { + "epoch": 2.4790137709243836, + "grad_norm": 2.109760046005249, + "learning_rate": 1e-06, + "loss": 0.6072, + "mean_token_accuracy": 0.7814216017723083, + "num_tokens": 793699977.0, + "step": 30063 + }, + { + "epoch": 2.479096231549435, + "grad_norm": 2.337555408477783, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7515167593955994, + "num_tokens": 793726275.0, + "step": 30064 + }, + { + "epoch": 2.4791786921744867, + "grad_norm": 2.086867570877075, + "learning_rate": 1e-06, + "loss": 0.5681, + "mean_token_accuracy": 0.8102049827575684, + "num_tokens": 793753478.0, + "step": 30065 + }, + { + "epoch": 2.479261152799538, + "grad_norm": 2.22589111328125, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.773032546043396, + "num_tokens": 793781691.0, + "step": 30066 + }, + { + "epoch": 2.47934361342459, + "grad_norm": 1.81609308719635, + "learning_rate": 1e-06, + "loss": 0.6017, + "mean_token_accuracy": 0.7962878942489624, + "num_tokens": 793816889.0, + "step": 30067 + }, + { + "epoch": 2.479426074049641, + "grad_norm": 2.076521635055542, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7625068426132202, + "num_tokens": 793847449.0, + "step": 30068 + }, + { + "epoch": 2.479508534674693, + "grad_norm": 2.0798890590667725, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.765424907207489, + "num_tokens": 793878262.0, + "step": 30069 + }, + { + "epoch": 2.4795909952997444, + "grad_norm": 2.188636302947998, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7623326778411865, + "num_tokens": 793907949.0, + "step": 30070 + }, + { + "epoch": 2.4796734559247957, + "grad_norm": 1.9790074825286865, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7747570872306824, + "num_tokens": 793938449.0, + "step": 30071 + }, + { + "epoch": 2.4797559165498475, + "grad_norm": 2.1861605644226074, + "learning_rate": 1e-06, + "loss": 0.723, + "mean_token_accuracy": 0.7630136013031006, + "num_tokens": 793964911.0, + "step": 30072 + }, + { + "epoch": 2.479838377174899, + "grad_norm": 2.2491323947906494, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7855397462844849, + "num_tokens": 793994270.0, + "step": 30073 + }, + { + "epoch": 2.4799208377999507, + "grad_norm": 2.2523608207702637, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7704721689224243, + "num_tokens": 794019608.0, + "step": 30074 + }, + { + "epoch": 2.480003298425002, + "grad_norm": 1.9232439994812012, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.7955172657966614, + "num_tokens": 794051021.0, + "step": 30075 + }, + { + "epoch": 2.480085759050054, + "grad_norm": 2.0576601028442383, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.7572667598724365, + "num_tokens": 794082609.0, + "step": 30076 + }, + { + "epoch": 2.480168219675105, + "grad_norm": 2.003652572631836, + "learning_rate": 1e-06, + "loss": 0.5537, + "mean_token_accuracy": 0.8057699799537659, + "num_tokens": 794117763.0, + "step": 30077 + }, + { + "epoch": 2.4802506803001565, + "grad_norm": 2.464911937713623, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.749194324016571, + "num_tokens": 794140901.0, + "step": 30078 + }, + { + "epoch": 2.4803331409252083, + "grad_norm": 1.9014984369277954, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.7918496131896973, + "num_tokens": 794176053.0, + "step": 30079 + }, + { + "epoch": 2.4804156015502596, + "grad_norm": 2.158708095550537, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7444461584091187, + "num_tokens": 794206513.0, + "step": 30080 + }, + { + "epoch": 2.4804980621753114, + "grad_norm": 2.081345558166504, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.8027390837669373, + "num_tokens": 794234666.0, + "step": 30081 + }, + { + "epoch": 2.480580522800363, + "grad_norm": 2.0603671073913574, + "learning_rate": 1e-06, + "loss": 0.6074, + "mean_token_accuracy": 0.7913736701011658, + "num_tokens": 794264694.0, + "step": 30082 + }, + { + "epoch": 2.4806629834254146, + "grad_norm": 1.8390780687332153, + "learning_rate": 1e-06, + "loss": 0.5486, + "mean_token_accuracy": 0.8162195682525635, + "num_tokens": 794300974.0, + "step": 30083 + }, + { + "epoch": 2.480745444050466, + "grad_norm": 2.391312837600708, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7751309871673584, + "num_tokens": 794325568.0, + "step": 30084 + }, + { + "epoch": 2.4808279046755173, + "grad_norm": 2.113612651824951, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7575777173042297, + "num_tokens": 794356361.0, + "step": 30085 + }, + { + "epoch": 2.480910365300569, + "grad_norm": 1.9862116575241089, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.7856462001800537, + "num_tokens": 794389948.0, + "step": 30086 + }, + { + "epoch": 2.4809928259256204, + "grad_norm": 2.225983142852783, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7668459415435791, + "num_tokens": 794418982.0, + "step": 30087 + }, + { + "epoch": 2.481075286550672, + "grad_norm": 2.780576467514038, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.7847094535827637, + "num_tokens": 794442955.0, + "step": 30088 + }, + { + "epoch": 2.4811577471757236, + "grad_norm": 2.2180263996124268, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7511224746704102, + "num_tokens": 794470361.0, + "step": 30089 + }, + { + "epoch": 2.4812402078007754, + "grad_norm": 2.2505102157592773, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7756264209747314, + "num_tokens": 794495110.0, + "step": 30090 + }, + { + "epoch": 2.4813226684258267, + "grad_norm": 1.9172967672348022, + "learning_rate": 1e-06, + "loss": 0.5858, + "mean_token_accuracy": 0.7985287308692932, + "num_tokens": 794527863.0, + "step": 30091 + }, + { + "epoch": 2.481405129050878, + "grad_norm": 2.1894898414611816, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7485548257827759, + "num_tokens": 794557246.0, + "step": 30092 + }, + { + "epoch": 2.48148758967593, + "grad_norm": 1.6927580833435059, + "learning_rate": 1e-06, + "loss": 0.5058, + "mean_token_accuracy": 0.8225661516189575, + "num_tokens": 794595866.0, + "step": 30093 + }, + { + "epoch": 2.481570050300981, + "grad_norm": 2.3730247020721436, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.7869400382041931, + "num_tokens": 794620923.0, + "step": 30094 + }, + { + "epoch": 2.481652510926033, + "grad_norm": 2.115236520767212, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.784217894077301, + "num_tokens": 794650489.0, + "step": 30095 + }, + { + "epoch": 2.4817349715510844, + "grad_norm": 1.8508250713348389, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7680903673171997, + "num_tokens": 794684567.0, + "step": 30096 + }, + { + "epoch": 2.4818174321761357, + "grad_norm": 1.9455702304840088, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7578105926513672, + "num_tokens": 794721167.0, + "step": 30097 + }, + { + "epoch": 2.4818998928011875, + "grad_norm": 2.067268133163452, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7368468046188354, + "num_tokens": 794752670.0, + "step": 30098 + }, + { + "epoch": 2.481982353426239, + "grad_norm": 2.054492950439453, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7555739879608154, + "num_tokens": 794786194.0, + "step": 30099 + }, + { + "epoch": 2.4820648140512906, + "grad_norm": 1.919197916984558, + "learning_rate": 1e-06, + "loss": 0.5841, + "mean_token_accuracy": 0.8136190176010132, + "num_tokens": 794820242.0, + "step": 30100 + }, + { + "epoch": 2.482147274676342, + "grad_norm": 1.9512982368469238, + "learning_rate": 1e-06, + "loss": 0.4663, + "mean_token_accuracy": 0.838943362236023, + "num_tokens": 794851467.0, + "step": 30101 + }, + { + "epoch": 2.4822297353013933, + "grad_norm": 2.068438768386841, + "learning_rate": 1e-06, + "loss": 0.5474, + "mean_token_accuracy": 0.8158164024353027, + "num_tokens": 794879107.0, + "step": 30102 + }, + { + "epoch": 2.482312195926445, + "grad_norm": 2.206598997116089, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7696520090103149, + "num_tokens": 794906145.0, + "step": 30103 + }, + { + "epoch": 2.4823946565514965, + "grad_norm": 1.9453799724578857, + "learning_rate": 1e-06, + "loss": 0.5368, + "mean_token_accuracy": 0.8171236515045166, + "num_tokens": 794939384.0, + "step": 30104 + }, + { + "epoch": 2.4824771171765483, + "grad_norm": 2.237744092941284, + "learning_rate": 1e-06, + "loss": 0.56, + "mean_token_accuracy": 0.8066059947013855, + "num_tokens": 794964144.0, + "step": 30105 + }, + { + "epoch": 2.4825595778015996, + "grad_norm": 2.0915098190307617, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7573453187942505, + "num_tokens": 794997446.0, + "step": 30106 + }, + { + "epoch": 2.4826420384266514, + "grad_norm": 2.3816709518432617, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7742066383361816, + "num_tokens": 795023907.0, + "step": 30107 + }, + { + "epoch": 2.482724499051703, + "grad_norm": 2.2319552898406982, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.8050235509872437, + "num_tokens": 795048608.0, + "step": 30108 + }, + { + "epoch": 2.482806959676754, + "grad_norm": 1.9659196138381958, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7892123460769653, + "num_tokens": 795083766.0, + "step": 30109 + }, + { + "epoch": 2.482889420301806, + "grad_norm": 2.000770330429077, + "learning_rate": 1e-06, + "loss": 0.5697, + "mean_token_accuracy": 0.8111177086830139, + "num_tokens": 795118055.0, + "step": 30110 + }, + { + "epoch": 2.4829718809268573, + "grad_norm": 2.070305824279785, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7349140644073486, + "num_tokens": 795151769.0, + "step": 30111 + }, + { + "epoch": 2.483054341551909, + "grad_norm": 2.306899070739746, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7704519033432007, + "num_tokens": 795177807.0, + "step": 30112 + }, + { + "epoch": 2.4831368021769604, + "grad_norm": 2.3497962951660156, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7839439511299133, + "num_tokens": 795205431.0, + "step": 30113 + }, + { + "epoch": 2.483219262802012, + "grad_norm": 1.9810676574707031, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7547488212585449, + "num_tokens": 795240145.0, + "step": 30114 + }, + { + "epoch": 2.4833017234270636, + "grad_norm": 1.9775094985961914, + "learning_rate": 1e-06, + "loss": 0.5759, + "mean_token_accuracy": 0.8085973262786865, + "num_tokens": 795274223.0, + "step": 30115 + }, + { + "epoch": 2.483384184052115, + "grad_norm": 2.086404323577881, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7782516479492188, + "num_tokens": 795303306.0, + "step": 30116 + }, + { + "epoch": 2.4834666446771667, + "grad_norm": 2.035593271255493, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.794633150100708, + "num_tokens": 795332811.0, + "step": 30117 + }, + { + "epoch": 2.483549105302218, + "grad_norm": 2.0156238079071045, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7451707720756531, + "num_tokens": 795365001.0, + "step": 30118 + }, + { + "epoch": 2.48363156592727, + "grad_norm": 2.0988874435424805, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7893704175949097, + "num_tokens": 795392445.0, + "step": 30119 + }, + { + "epoch": 2.483714026552321, + "grad_norm": 1.8710637092590332, + "learning_rate": 1e-06, + "loss": 0.4988, + "mean_token_accuracy": 0.8222853541374207, + "num_tokens": 795428905.0, + "step": 30120 + }, + { + "epoch": 2.483796487177373, + "grad_norm": 2.3009839057922363, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7630215883255005, + "num_tokens": 795454002.0, + "step": 30121 + }, + { + "epoch": 2.4838789478024244, + "grad_norm": 1.9125539064407349, + "learning_rate": 1e-06, + "loss": 0.5883, + "mean_token_accuracy": 0.802605390548706, + "num_tokens": 795490437.0, + "step": 30122 + }, + { + "epoch": 2.4839614084274757, + "grad_norm": 1.904102087020874, + "learning_rate": 1e-06, + "loss": 0.6523, + "mean_token_accuracy": 0.7751576900482178, + "num_tokens": 795524745.0, + "step": 30123 + }, + { + "epoch": 2.4840438690525275, + "grad_norm": 2.047417402267456, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.7901695966720581, + "num_tokens": 795554980.0, + "step": 30124 + }, + { + "epoch": 2.484126329677579, + "grad_norm": 1.9116802215576172, + "learning_rate": 1e-06, + "loss": 0.5851, + "mean_token_accuracy": 0.7997417449951172, + "num_tokens": 795591007.0, + "step": 30125 + }, + { + "epoch": 2.4842087903026306, + "grad_norm": 2.25394606590271, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.7865729331970215, + "num_tokens": 795619164.0, + "step": 30126 + }, + { + "epoch": 2.484291250927682, + "grad_norm": 2.1335439682006836, + "learning_rate": 1e-06, + "loss": 0.5589, + "mean_token_accuracy": 0.8088225722312927, + "num_tokens": 795647277.0, + "step": 30127 + }, + { + "epoch": 2.484373711552734, + "grad_norm": 2.0200154781341553, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.7871167659759521, + "num_tokens": 795680361.0, + "step": 30128 + }, + { + "epoch": 2.484456172177785, + "grad_norm": 2.1058237552642822, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7316234111785889, + "num_tokens": 795712004.0, + "step": 30129 + }, + { + "epoch": 2.4845386328028365, + "grad_norm": 2.204474687576294, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7531128525733948, + "num_tokens": 795743828.0, + "step": 30130 + }, + { + "epoch": 2.4846210934278883, + "grad_norm": 2.341193437576294, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7734236717224121, + "num_tokens": 795770577.0, + "step": 30131 + }, + { + "epoch": 2.4847035540529396, + "grad_norm": 2.161771774291992, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7575449347496033, + "num_tokens": 795800329.0, + "step": 30132 + }, + { + "epoch": 2.4847860146779914, + "grad_norm": 1.9908150434494019, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.8062858581542969, + "num_tokens": 795831955.0, + "step": 30133 + }, + { + "epoch": 2.484868475303043, + "grad_norm": 2.1449198722839355, + "learning_rate": 1e-06, + "loss": 0.5874, + "mean_token_accuracy": 0.8020381927490234, + "num_tokens": 795860427.0, + "step": 30134 + }, + { + "epoch": 2.4849509359280946, + "grad_norm": 1.843112587928772, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7784121036529541, + "num_tokens": 795893460.0, + "step": 30135 + }, + { + "epoch": 2.485033396553146, + "grad_norm": 1.837829351425171, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7607842087745667, + "num_tokens": 795930707.0, + "step": 30136 + }, + { + "epoch": 2.4851158571781973, + "grad_norm": 2.2345402240753174, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7682315111160278, + "num_tokens": 795960610.0, + "step": 30137 + }, + { + "epoch": 2.485198317803249, + "grad_norm": 2.4583377838134766, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.774406373500824, + "num_tokens": 795985778.0, + "step": 30138 + }, + { + "epoch": 2.4852807784283004, + "grad_norm": 2.0241901874542236, + "learning_rate": 1e-06, + "loss": 0.566, + "mean_token_accuracy": 0.8111442923545837, + "num_tokens": 796016031.0, + "step": 30139 + }, + { + "epoch": 2.485363239053352, + "grad_norm": 1.881864070892334, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7544519901275635, + "num_tokens": 796050376.0, + "step": 30140 + }, + { + "epoch": 2.4854456996784036, + "grad_norm": 1.899960994720459, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7626690864562988, + "num_tokens": 796085579.0, + "step": 30141 + }, + { + "epoch": 2.485528160303455, + "grad_norm": 2.386082887649536, + "learning_rate": 1e-06, + "loss": 0.5924, + "mean_token_accuracy": 0.7955439686775208, + "num_tokens": 796109257.0, + "step": 30142 + }, + { + "epoch": 2.4856106209285067, + "grad_norm": 1.7859292030334473, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7690924406051636, + "num_tokens": 796149241.0, + "step": 30143 + }, + { + "epoch": 2.485693081553558, + "grad_norm": 2.120670795440674, + "learning_rate": 1e-06, + "loss": 0.707, + "mean_token_accuracy": 0.7722136378288269, + "num_tokens": 796178514.0, + "step": 30144 + }, + { + "epoch": 2.48577554217861, + "grad_norm": 1.9769420623779297, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.764985978603363, + "num_tokens": 796213078.0, + "step": 30145 + }, + { + "epoch": 2.485858002803661, + "grad_norm": 2.0066356658935547, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.8066456317901611, + "num_tokens": 796243762.0, + "step": 30146 + }, + { + "epoch": 2.4859404634287126, + "grad_norm": 1.995295763015747, + "learning_rate": 1e-06, + "loss": 0.5608, + "mean_token_accuracy": 0.808398425579071, + "num_tokens": 796275077.0, + "step": 30147 + }, + { + "epoch": 2.4860229240537643, + "grad_norm": 2.275503158569336, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7454441785812378, + "num_tokens": 796304799.0, + "step": 30148 + }, + { + "epoch": 2.4861053846788157, + "grad_norm": 1.8285455703735352, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7766172289848328, + "num_tokens": 796341608.0, + "step": 30149 + }, + { + "epoch": 2.4861878453038675, + "grad_norm": 2.0332086086273193, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.765055775642395, + "num_tokens": 796374637.0, + "step": 30150 + }, + { + "epoch": 2.486270305928919, + "grad_norm": 2.115600109100342, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7304120063781738, + "num_tokens": 796405164.0, + "step": 30151 + }, + { + "epoch": 2.4863527665539706, + "grad_norm": 1.9715205430984497, + "learning_rate": 1e-06, + "loss": 0.5719, + "mean_token_accuracy": 0.8114066123962402, + "num_tokens": 796437290.0, + "step": 30152 + }, + { + "epoch": 2.486435227179022, + "grad_norm": 1.8435723781585693, + "learning_rate": 1e-06, + "loss": 0.6048, + "mean_token_accuracy": 0.7994526624679565, + "num_tokens": 796474660.0, + "step": 30153 + }, + { + "epoch": 2.4865176878040733, + "grad_norm": 2.016218662261963, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.7816581130027771, + "num_tokens": 796504007.0, + "step": 30154 + }, + { + "epoch": 2.486600148429125, + "grad_norm": 2.2806546688079834, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7607405185699463, + "num_tokens": 796530833.0, + "step": 30155 + }, + { + "epoch": 2.4866826090541765, + "grad_norm": 2.117751359939575, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7282372713088989, + "num_tokens": 796561966.0, + "step": 30156 + }, + { + "epoch": 2.4867650696792283, + "grad_norm": 2.148263931274414, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.7881881594657898, + "num_tokens": 796588814.0, + "step": 30157 + }, + { + "epoch": 2.4868475303042796, + "grad_norm": 2.1371471881866455, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.768055260181427, + "num_tokens": 796617014.0, + "step": 30158 + }, + { + "epoch": 2.4869299909293314, + "grad_norm": 2.1964685916900635, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7407302260398865, + "num_tokens": 796643703.0, + "step": 30159 + }, + { + "epoch": 2.4870124515543828, + "grad_norm": 2.0583882331848145, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.7871263027191162, + "num_tokens": 796671774.0, + "step": 30160 + }, + { + "epoch": 2.487094912179434, + "grad_norm": 2.2344002723693848, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7236884832382202, + "num_tokens": 796700369.0, + "step": 30161 + }, + { + "epoch": 2.487177372804486, + "grad_norm": 2.15000057220459, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7693488001823425, + "num_tokens": 796730939.0, + "step": 30162 + }, + { + "epoch": 2.4872598334295373, + "grad_norm": 2.306034803390503, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7165738344192505, + "num_tokens": 796758793.0, + "step": 30163 + }, + { + "epoch": 2.487342294054589, + "grad_norm": 2.255852222442627, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7175198197364807, + "num_tokens": 796788117.0, + "step": 30164 + }, + { + "epoch": 2.4874247546796404, + "grad_norm": 2.06770396232605, + "learning_rate": 1e-06, + "loss": 0.8162, + "mean_token_accuracy": 0.7476382851600647, + "num_tokens": 796819969.0, + "step": 30165 + }, + { + "epoch": 2.487507215304692, + "grad_norm": 1.8041757345199585, + "learning_rate": 1e-06, + "loss": 0.5683, + "mean_token_accuracy": 0.8152570128440857, + "num_tokens": 796855497.0, + "step": 30166 + }, + { + "epoch": 2.4875896759297436, + "grad_norm": 2.2967796325683594, + "learning_rate": 1e-06, + "loss": 0.7868, + "mean_token_accuracy": 0.7561659812927246, + "num_tokens": 796883073.0, + "step": 30167 + }, + { + "epoch": 2.487672136554795, + "grad_norm": 2.4015750885009766, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7712969779968262, + "num_tokens": 796908138.0, + "step": 30168 + }, + { + "epoch": 2.4877545971798467, + "grad_norm": 1.8787870407104492, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7854862809181213, + "num_tokens": 796945714.0, + "step": 30169 + }, + { + "epoch": 2.487837057804898, + "grad_norm": 2.0362114906311035, + "learning_rate": 1e-06, + "loss": 0.5734, + "mean_token_accuracy": 0.8001226186752319, + "num_tokens": 796976677.0, + "step": 30170 + }, + { + "epoch": 2.48791951842995, + "grad_norm": 1.9779599905014038, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7526872158050537, + "num_tokens": 797011307.0, + "step": 30171 + }, + { + "epoch": 2.488001979055001, + "grad_norm": 2.3525054454803467, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7371461391448975, + "num_tokens": 797036939.0, + "step": 30172 + }, + { + "epoch": 2.488084439680053, + "grad_norm": 2.0856006145477295, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.7792519330978394, + "num_tokens": 797066353.0, + "step": 30173 + }, + { + "epoch": 2.4881669003051043, + "grad_norm": 2.8643369674682617, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7465575933456421, + "num_tokens": 797084798.0, + "step": 30174 + }, + { + "epoch": 2.4882493609301557, + "grad_norm": 2.2333385944366455, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7476176023483276, + "num_tokens": 797113660.0, + "step": 30175 + }, + { + "epoch": 2.4883318215552075, + "grad_norm": 1.9631030559539795, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7739628553390503, + "num_tokens": 797146684.0, + "step": 30176 + }, + { + "epoch": 2.488414282180259, + "grad_norm": 2.0474095344543457, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.76014244556427, + "num_tokens": 797178302.0, + "step": 30177 + }, + { + "epoch": 2.4884967428053106, + "grad_norm": 2.29909610748291, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7409077882766724, + "num_tokens": 797205738.0, + "step": 30178 + }, + { + "epoch": 2.488579203430362, + "grad_norm": 2.231463670730591, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.790523886680603, + "num_tokens": 797235081.0, + "step": 30179 + }, + { + "epoch": 2.488661664055414, + "grad_norm": 2.0808446407318115, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7518700957298279, + "num_tokens": 797266548.0, + "step": 30180 + }, + { + "epoch": 2.488744124680465, + "grad_norm": 1.9526435136795044, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7702136039733887, + "num_tokens": 797300296.0, + "step": 30181 + }, + { + "epoch": 2.4888265853055165, + "grad_norm": 2.028623342514038, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7590065002441406, + "num_tokens": 797332538.0, + "step": 30182 + }, + { + "epoch": 2.4889090459305683, + "grad_norm": 2.2064437866210938, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7800647020339966, + "num_tokens": 797360864.0, + "step": 30183 + }, + { + "epoch": 2.4889915065556196, + "grad_norm": 2.5501339435577393, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7733964920043945, + "num_tokens": 797383842.0, + "step": 30184 + }, + { + "epoch": 2.4890739671806714, + "grad_norm": 2.0570361614227295, + "learning_rate": 1e-06, + "loss": 0.5804, + "mean_token_accuracy": 0.8073845505714417, + "num_tokens": 797413314.0, + "step": 30185 + }, + { + "epoch": 2.4891564278057228, + "grad_norm": 2.219371795654297, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7314454317092896, + "num_tokens": 797441344.0, + "step": 30186 + }, + { + "epoch": 2.489238888430774, + "grad_norm": 2.0594751834869385, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.770180344581604, + "num_tokens": 797475528.0, + "step": 30187 + }, + { + "epoch": 2.489321349055826, + "grad_norm": 2.1197245121002197, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7356463074684143, + "num_tokens": 797506893.0, + "step": 30188 + }, + { + "epoch": 2.4894038096808773, + "grad_norm": 2.182856798171997, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7472936511039734, + "num_tokens": 797535438.0, + "step": 30189 + }, + { + "epoch": 2.489486270305929, + "grad_norm": 2.0489392280578613, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7379542589187622, + "num_tokens": 797570601.0, + "step": 30190 + }, + { + "epoch": 2.4895687309309804, + "grad_norm": 2.210963726043701, + "learning_rate": 1e-06, + "loss": 0.8038, + "mean_token_accuracy": 0.7400605082511902, + "num_tokens": 797600314.0, + "step": 30191 + }, + { + "epoch": 2.4896511915560318, + "grad_norm": 2.1950607299804688, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7549711465835571, + "num_tokens": 797629190.0, + "step": 30192 + }, + { + "epoch": 2.4897336521810836, + "grad_norm": 2.1346676349639893, + "learning_rate": 1e-06, + "loss": 0.6805, + "mean_token_accuracy": 0.7788788080215454, + "num_tokens": 797658790.0, + "step": 30193 + }, + { + "epoch": 2.489816112806135, + "grad_norm": 2.0093834400177, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7713907957077026, + "num_tokens": 797689505.0, + "step": 30194 + }, + { + "epoch": 2.4898985734311867, + "grad_norm": 2.2989203929901123, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7631524801254272, + "num_tokens": 797717567.0, + "step": 30195 + }, + { + "epoch": 2.489981034056238, + "grad_norm": 2.208871841430664, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.8011931777000427, + "num_tokens": 797749408.0, + "step": 30196 + }, + { + "epoch": 2.49006349468129, + "grad_norm": 1.991926670074463, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.7781752943992615, + "num_tokens": 797780269.0, + "step": 30197 + }, + { + "epoch": 2.490145955306341, + "grad_norm": 2.3547353744506836, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7653911113739014, + "num_tokens": 797805698.0, + "step": 30198 + }, + { + "epoch": 2.4902284159313925, + "grad_norm": 2.1424849033355713, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.7873133420944214, + "num_tokens": 797835146.0, + "step": 30199 + }, + { + "epoch": 2.4903108765564443, + "grad_norm": 2.3075783252716064, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.755646288394928, + "num_tokens": 797859669.0, + "step": 30200 + }, + { + "epoch": 2.4903933371814957, + "grad_norm": 2.2349472045898438, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7844175100326538, + "num_tokens": 797885666.0, + "step": 30201 + }, + { + "epoch": 2.4904757978065475, + "grad_norm": 2.2931883335113525, + "learning_rate": 1e-06, + "loss": 0.5981, + "mean_token_accuracy": 0.8029271364212036, + "num_tokens": 797914492.0, + "step": 30202 + }, + { + "epoch": 2.490558258431599, + "grad_norm": 2.086942434310913, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.772923469543457, + "num_tokens": 797946888.0, + "step": 30203 + }, + { + "epoch": 2.4906407190566506, + "grad_norm": 2.0183935165405273, + "learning_rate": 1e-06, + "loss": 0.6214, + "mean_token_accuracy": 0.7980328798294067, + "num_tokens": 797982664.0, + "step": 30204 + }, + { + "epoch": 2.490723179681702, + "grad_norm": 1.8183157444000244, + "learning_rate": 1e-06, + "loss": 0.602, + "mean_token_accuracy": 0.8106801509857178, + "num_tokens": 798020059.0, + "step": 30205 + }, + { + "epoch": 2.4908056403067533, + "grad_norm": 1.9460889101028442, + "learning_rate": 1e-06, + "loss": 0.5657, + "mean_token_accuracy": 0.8120152950286865, + "num_tokens": 798050931.0, + "step": 30206 + }, + { + "epoch": 2.490888100931805, + "grad_norm": 2.235629081726074, + "learning_rate": 1e-06, + "loss": 0.6217, + "mean_token_accuracy": 0.7959101796150208, + "num_tokens": 798076844.0, + "step": 30207 + }, + { + "epoch": 2.4909705615568565, + "grad_norm": 2.029914379119873, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7644292712211609, + "num_tokens": 798109558.0, + "step": 30208 + }, + { + "epoch": 2.4910530221819083, + "grad_norm": 2.006518602371216, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.7872617244720459, + "num_tokens": 798139233.0, + "step": 30209 + }, + { + "epoch": 2.4911354828069596, + "grad_norm": 1.9307663440704346, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.7879451513290405, + "num_tokens": 798169628.0, + "step": 30210 + }, + { + "epoch": 2.4912179434320114, + "grad_norm": 2.007908821105957, + "learning_rate": 1e-06, + "loss": 0.5176, + "mean_token_accuracy": 0.8088953495025635, + "num_tokens": 798197424.0, + "step": 30211 + }, + { + "epoch": 2.4913004040570628, + "grad_norm": 2.0376455783843994, + "learning_rate": 1e-06, + "loss": 0.7525, + "mean_token_accuracy": 0.7675696611404419, + "num_tokens": 798228924.0, + "step": 30212 + }, + { + "epoch": 2.491382864682114, + "grad_norm": 2.040121078491211, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.772651731967926, + "num_tokens": 798259071.0, + "step": 30213 + }, + { + "epoch": 2.491465325307166, + "grad_norm": 2.147109270095825, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7691212892532349, + "num_tokens": 798287706.0, + "step": 30214 + }, + { + "epoch": 2.4915477859322173, + "grad_norm": 2.089071273803711, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.7849581241607666, + "num_tokens": 798316772.0, + "step": 30215 + }, + { + "epoch": 2.491630246557269, + "grad_norm": 2.4694981575012207, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7382141947746277, + "num_tokens": 798341126.0, + "step": 30216 + }, + { + "epoch": 2.4917127071823204, + "grad_norm": 2.3684113025665283, + "learning_rate": 1e-06, + "loss": 0.6024, + "mean_token_accuracy": 0.790213942527771, + "num_tokens": 798364769.0, + "step": 30217 + }, + { + "epoch": 2.491795167807372, + "grad_norm": 2.1755897998809814, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7820510864257812, + "num_tokens": 798389978.0, + "step": 30218 + }, + { + "epoch": 2.4918776284324236, + "grad_norm": 2.1876492500305176, + "learning_rate": 1e-06, + "loss": 0.5698, + "mean_token_accuracy": 0.7967480421066284, + "num_tokens": 798415032.0, + "step": 30219 + }, + { + "epoch": 2.491960089057475, + "grad_norm": 2.1574296951293945, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7784425020217896, + "num_tokens": 798445918.0, + "step": 30220 + }, + { + "epoch": 2.4920425496825267, + "grad_norm": 2.072417736053467, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7437705397605896, + "num_tokens": 798477626.0, + "step": 30221 + }, + { + "epoch": 2.492125010307578, + "grad_norm": 2.0315394401550293, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7757354378700256, + "num_tokens": 798508158.0, + "step": 30222 + }, + { + "epoch": 2.49220747093263, + "grad_norm": 2.3673248291015625, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7334877252578735, + "num_tokens": 798533489.0, + "step": 30223 + }, + { + "epoch": 2.492289931557681, + "grad_norm": 1.8352069854736328, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.785554051399231, + "num_tokens": 798569659.0, + "step": 30224 + }, + { + "epoch": 2.492372392182733, + "grad_norm": 2.3125476837158203, + "learning_rate": 1e-06, + "loss": 0.5649, + "mean_token_accuracy": 0.8024104833602905, + "num_tokens": 798593988.0, + "step": 30225 + }, + { + "epoch": 2.4924548528077843, + "grad_norm": 2.2712223529815674, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7593309879302979, + "num_tokens": 798621055.0, + "step": 30226 + }, + { + "epoch": 2.4925373134328357, + "grad_norm": 1.9124462604522705, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7821171283721924, + "num_tokens": 798658043.0, + "step": 30227 + }, + { + "epoch": 2.4926197740578875, + "grad_norm": 1.951401710510254, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.7965940833091736, + "num_tokens": 798689404.0, + "step": 30228 + }, + { + "epoch": 2.492702234682939, + "grad_norm": 2.0545265674591064, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7664973139762878, + "num_tokens": 798717291.0, + "step": 30229 + }, + { + "epoch": 2.4927846953079906, + "grad_norm": 2.1812922954559326, + "learning_rate": 1e-06, + "loss": 0.5963, + "mean_token_accuracy": 0.8030538558959961, + "num_tokens": 798743016.0, + "step": 30230 + }, + { + "epoch": 2.492867155933042, + "grad_norm": 2.0881824493408203, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7732277512550354, + "num_tokens": 798773612.0, + "step": 30231 + }, + { + "epoch": 2.4929496165580933, + "grad_norm": 2.1578269004821777, + "learning_rate": 1e-06, + "loss": 0.5553, + "mean_token_accuracy": 0.8169401288032532, + "num_tokens": 798801394.0, + "step": 30232 + }, + { + "epoch": 2.493032077183145, + "grad_norm": 2.3051159381866455, + "learning_rate": 1e-06, + "loss": 0.7851, + "mean_token_accuracy": 0.7385337352752686, + "num_tokens": 798827780.0, + "step": 30233 + }, + { + "epoch": 2.4931145378081965, + "grad_norm": 2.1614463329315186, + "learning_rate": 1e-06, + "loss": 0.5449, + "mean_token_accuracy": 0.7966893315315247, + "num_tokens": 798852694.0, + "step": 30234 + }, + { + "epoch": 2.4931969984332483, + "grad_norm": 2.6428706645965576, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.7814909219741821, + "num_tokens": 798874409.0, + "step": 30235 + }, + { + "epoch": 2.4932794590582996, + "grad_norm": 2.333547353744507, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.74080491065979, + "num_tokens": 798899016.0, + "step": 30236 + }, + { + "epoch": 2.493361919683351, + "grad_norm": 2.285970449447632, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7693896293640137, + "num_tokens": 798927121.0, + "step": 30237 + }, + { + "epoch": 2.4934443803084028, + "grad_norm": 2.081291675567627, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.7784585952758789, + "num_tokens": 798958254.0, + "step": 30238 + }, + { + "epoch": 2.493526840933454, + "grad_norm": 2.3817873001098633, + "learning_rate": 1e-06, + "loss": 0.6524, + "mean_token_accuracy": 0.7830014824867249, + "num_tokens": 798981931.0, + "step": 30239 + }, + { + "epoch": 2.493609301558506, + "grad_norm": 2.040733814239502, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.783153772354126, + "num_tokens": 799010068.0, + "step": 30240 + }, + { + "epoch": 2.4936917621835573, + "grad_norm": 2.0641400814056396, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7746789455413818, + "num_tokens": 799041353.0, + "step": 30241 + }, + { + "epoch": 2.493774222808609, + "grad_norm": 2.216273069381714, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.7841066122055054, + "num_tokens": 799068205.0, + "step": 30242 + }, + { + "epoch": 2.4938566834336604, + "grad_norm": 1.941178321838379, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7723575830459595, + "num_tokens": 799103219.0, + "step": 30243 + }, + { + "epoch": 2.4939391440587118, + "grad_norm": 2.219567060470581, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7619799971580505, + "num_tokens": 799131270.0, + "step": 30244 + }, + { + "epoch": 2.4940216046837635, + "grad_norm": 2.0890297889709473, + "learning_rate": 1e-06, + "loss": 0.6858, + "mean_token_accuracy": 0.7781745195388794, + "num_tokens": 799163132.0, + "step": 30245 + }, + { + "epoch": 2.494104065308815, + "grad_norm": 2.1136579513549805, + "learning_rate": 1e-06, + "loss": 0.5531, + "mean_token_accuracy": 0.8112389445304871, + "num_tokens": 799189597.0, + "step": 30246 + }, + { + "epoch": 2.4941865259338667, + "grad_norm": 2.0224478244781494, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7787984609603882, + "num_tokens": 799220755.0, + "step": 30247 + }, + { + "epoch": 2.494268986558918, + "grad_norm": 1.954060673713684, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7793992757797241, + "num_tokens": 799255020.0, + "step": 30248 + }, + { + "epoch": 2.49435144718397, + "grad_norm": 2.0763254165649414, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7586236596107483, + "num_tokens": 799289549.0, + "step": 30249 + }, + { + "epoch": 2.494433907809021, + "grad_norm": 2.331376552581787, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7914553284645081, + "num_tokens": 799322995.0, + "step": 30250 + }, + { + "epoch": 2.4945163684340725, + "grad_norm": 2.284235954284668, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7690826654434204, + "num_tokens": 799350097.0, + "step": 30251 + }, + { + "epoch": 2.4945988290591243, + "grad_norm": 2.21476149559021, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7436288595199585, + "num_tokens": 799379945.0, + "step": 30252 + }, + { + "epoch": 2.4946812896841757, + "grad_norm": 2.3470566272735596, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7686829566955566, + "num_tokens": 799403544.0, + "step": 30253 + }, + { + "epoch": 2.4947637503092275, + "grad_norm": 2.049516439437866, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7558257579803467, + "num_tokens": 799435059.0, + "step": 30254 + }, + { + "epoch": 2.494846210934279, + "grad_norm": 2.0945751667022705, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7715977430343628, + "num_tokens": 799464410.0, + "step": 30255 + }, + { + "epoch": 2.4949286715593306, + "grad_norm": 2.1355106830596924, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7540754079818726, + "num_tokens": 799493765.0, + "step": 30256 + }, + { + "epoch": 2.495011132184382, + "grad_norm": 2.1546268463134766, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7657154202461243, + "num_tokens": 799521796.0, + "step": 30257 + }, + { + "epoch": 2.4950935928094333, + "grad_norm": 1.8188109397888184, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.797685980796814, + "num_tokens": 799560332.0, + "step": 30258 + }, + { + "epoch": 2.495176053434485, + "grad_norm": 1.968881607055664, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.7853769063949585, + "num_tokens": 799594128.0, + "step": 30259 + }, + { + "epoch": 2.4952585140595365, + "grad_norm": 2.041757345199585, + "learning_rate": 1e-06, + "loss": 0.6488, + "mean_token_accuracy": 0.7804155945777893, + "num_tokens": 799626531.0, + "step": 30260 + }, + { + "epoch": 2.4953409746845883, + "grad_norm": 2.016517162322998, + "learning_rate": 1e-06, + "loss": 0.6511, + "mean_token_accuracy": 0.7868732213973999, + "num_tokens": 799659022.0, + "step": 30261 + }, + { + "epoch": 2.4954234353096396, + "grad_norm": 2.185603380203247, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7236175537109375, + "num_tokens": 799686467.0, + "step": 30262 + }, + { + "epoch": 2.4955058959346914, + "grad_norm": 2.091343641281128, + "learning_rate": 1e-06, + "loss": 0.6322, + "mean_token_accuracy": 0.8008371591567993, + "num_tokens": 799713445.0, + "step": 30263 + }, + { + "epoch": 2.4955883565597428, + "grad_norm": 2.180269241333008, + "learning_rate": 1e-06, + "loss": 0.5986, + "mean_token_accuracy": 0.8004286885261536, + "num_tokens": 799742310.0, + "step": 30264 + }, + { + "epoch": 2.495670817184794, + "grad_norm": 2.414463996887207, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7702469825744629, + "num_tokens": 799766709.0, + "step": 30265 + }, + { + "epoch": 2.495753277809846, + "grad_norm": 2.282886505126953, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.752017080783844, + "num_tokens": 799794097.0, + "step": 30266 + }, + { + "epoch": 2.4958357384348973, + "grad_norm": 1.892810583114624, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7261923551559448, + "num_tokens": 799833150.0, + "step": 30267 + }, + { + "epoch": 2.495918199059949, + "grad_norm": 2.2541344165802, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7726731300354004, + "num_tokens": 799860997.0, + "step": 30268 + }, + { + "epoch": 2.4960006596850004, + "grad_norm": 2.014324426651001, + "learning_rate": 1e-06, + "loss": 0.6015, + "mean_token_accuracy": 0.8006418943405151, + "num_tokens": 799891880.0, + "step": 30269 + }, + { + "epoch": 2.496083120310052, + "grad_norm": 2.221014976501465, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.733820915222168, + "num_tokens": 799918758.0, + "step": 30270 + }, + { + "epoch": 2.4961655809351035, + "grad_norm": 2.128690481185913, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7694641947746277, + "num_tokens": 799948191.0, + "step": 30271 + }, + { + "epoch": 2.496248041560155, + "grad_norm": 2.0979974269866943, + "learning_rate": 1e-06, + "loss": 0.6065, + "mean_token_accuracy": 0.7953970432281494, + "num_tokens": 799980053.0, + "step": 30272 + }, + { + "epoch": 2.4963305021852067, + "grad_norm": 1.9360121488571167, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.760364294052124, + "num_tokens": 800012838.0, + "step": 30273 + }, + { + "epoch": 2.496412962810258, + "grad_norm": 1.9785141944885254, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7615652084350586, + "num_tokens": 800045754.0, + "step": 30274 + }, + { + "epoch": 2.4964954234353094, + "grad_norm": 2.171207904815674, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7448690533638, + "num_tokens": 800073978.0, + "step": 30275 + }, + { + "epoch": 2.496577884060361, + "grad_norm": 1.9630460739135742, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7684052586555481, + "num_tokens": 800109084.0, + "step": 30276 + }, + { + "epoch": 2.4966603446854125, + "grad_norm": 2.2966878414154053, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7509422898292542, + "num_tokens": 800137725.0, + "step": 30277 + }, + { + "epoch": 2.4967428053104643, + "grad_norm": 2.137012481689453, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7527621984481812, + "num_tokens": 800167236.0, + "step": 30278 + }, + { + "epoch": 2.4968252659355157, + "grad_norm": 2.1778643131256104, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7627056837081909, + "num_tokens": 800193799.0, + "step": 30279 + }, + { + "epoch": 2.4969077265605675, + "grad_norm": 1.9311617612838745, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.782250165939331, + "num_tokens": 800226189.0, + "step": 30280 + }, + { + "epoch": 2.496990187185619, + "grad_norm": 1.9147855043411255, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.7634255886077881, + "num_tokens": 800261789.0, + "step": 30281 + }, + { + "epoch": 2.49707264781067, + "grad_norm": 1.9659134149551392, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7650892734527588, + "num_tokens": 800300729.0, + "step": 30282 + }, + { + "epoch": 2.497155108435722, + "grad_norm": 2.2604830265045166, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.760052502155304, + "num_tokens": 800327622.0, + "step": 30283 + }, + { + "epoch": 2.4972375690607733, + "grad_norm": 2.170971632003784, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7604584693908691, + "num_tokens": 800356464.0, + "step": 30284 + }, + { + "epoch": 2.497320029685825, + "grad_norm": 1.9758344888687134, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.7837083339691162, + "num_tokens": 800390073.0, + "step": 30285 + }, + { + "epoch": 2.4974024903108765, + "grad_norm": 2.48109769821167, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7353739142417908, + "num_tokens": 800416686.0, + "step": 30286 + }, + { + "epoch": 2.4974849509359283, + "grad_norm": 1.9592602252960205, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.7883422374725342, + "num_tokens": 800452207.0, + "step": 30287 + }, + { + "epoch": 2.4975674115609796, + "grad_norm": 2.0495431423187256, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.7834881544113159, + "num_tokens": 800483052.0, + "step": 30288 + }, + { + "epoch": 2.497649872186031, + "grad_norm": 2.0359795093536377, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.7939096689224243, + "num_tokens": 800514142.0, + "step": 30289 + }, + { + "epoch": 2.4977323328110828, + "grad_norm": 2.211879253387451, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7650654315948486, + "num_tokens": 800542041.0, + "step": 30290 + }, + { + "epoch": 2.497814793436134, + "grad_norm": 2.429205894470215, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7498993873596191, + "num_tokens": 800566360.0, + "step": 30291 + }, + { + "epoch": 2.497897254061186, + "grad_norm": 1.8726085424423218, + "learning_rate": 1e-06, + "loss": 0.6007, + "mean_token_accuracy": 0.7951605319976807, + "num_tokens": 800597913.0, + "step": 30292 + }, + { + "epoch": 2.4979797146862373, + "grad_norm": 2.1882104873657227, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7442359924316406, + "num_tokens": 800627075.0, + "step": 30293 + }, + { + "epoch": 2.498062175311289, + "grad_norm": 2.287256956100464, + "learning_rate": 1e-06, + "loss": 0.7136, + "mean_token_accuracy": 0.7620953917503357, + "num_tokens": 800653046.0, + "step": 30294 + }, + { + "epoch": 2.4981446359363404, + "grad_norm": 2.285599708557129, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7713854908943176, + "num_tokens": 800679927.0, + "step": 30295 + }, + { + "epoch": 2.4982270965613917, + "grad_norm": 2.2217345237731934, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.7629421949386597, + "num_tokens": 800709124.0, + "step": 30296 + }, + { + "epoch": 2.4983095571864435, + "grad_norm": 2.1708457469940186, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7609442472457886, + "num_tokens": 800737681.0, + "step": 30297 + }, + { + "epoch": 2.498392017811495, + "grad_norm": 2.0480287075042725, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7954681515693665, + "num_tokens": 800767531.0, + "step": 30298 + }, + { + "epoch": 2.4984744784365467, + "grad_norm": 2.235729217529297, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7670059204101562, + "num_tokens": 800793997.0, + "step": 30299 + }, + { + "epoch": 2.498556939061598, + "grad_norm": 2.209676742553711, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7575033903121948, + "num_tokens": 800825232.0, + "step": 30300 + }, + { + "epoch": 2.49863939968665, + "grad_norm": 2.041126251220703, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7741832733154297, + "num_tokens": 800858408.0, + "step": 30301 + }, + { + "epoch": 2.498721860311701, + "grad_norm": 2.1764917373657227, + "learning_rate": 1e-06, + "loss": 0.5819, + "mean_token_accuracy": 0.8046499490737915, + "num_tokens": 800886067.0, + "step": 30302 + }, + { + "epoch": 2.4988043209367525, + "grad_norm": 1.8307557106018066, + "learning_rate": 1e-06, + "loss": 0.5032, + "mean_token_accuracy": 0.8282086253166199, + "num_tokens": 800923586.0, + "step": 30303 + }, + { + "epoch": 2.4988867815618043, + "grad_norm": 2.2093136310577393, + "learning_rate": 1e-06, + "loss": 0.4993, + "mean_token_accuracy": 0.83354651927948, + "num_tokens": 800950053.0, + "step": 30304 + }, + { + "epoch": 2.4989692421868557, + "grad_norm": 1.9593034982681274, + "learning_rate": 1e-06, + "loss": 0.5465, + "mean_token_accuracy": 0.8198460340499878, + "num_tokens": 800983611.0, + "step": 30305 + }, + { + "epoch": 2.4990517028119075, + "grad_norm": 2.1147947311401367, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7273638248443604, + "num_tokens": 801016754.0, + "step": 30306 + }, + { + "epoch": 2.499134163436959, + "grad_norm": 1.8567324876785278, + "learning_rate": 1e-06, + "loss": 0.515, + "mean_token_accuracy": 0.8223863244056702, + "num_tokens": 801051484.0, + "step": 30307 + }, + { + "epoch": 2.4992166240620106, + "grad_norm": 2.0841031074523926, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7704769372940063, + "num_tokens": 801084366.0, + "step": 30308 + }, + { + "epoch": 2.499299084687062, + "grad_norm": 2.6650328636169434, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7706611156463623, + "num_tokens": 801105099.0, + "step": 30309 + }, + { + "epoch": 2.4993815453121133, + "grad_norm": 2.169994592666626, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7721283435821533, + "num_tokens": 801136707.0, + "step": 30310 + }, + { + "epoch": 2.499464005937165, + "grad_norm": 1.999871015548706, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.7897195816040039, + "num_tokens": 801171368.0, + "step": 30311 + }, + { + "epoch": 2.4995464665622165, + "grad_norm": 2.0727784633636475, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.759640097618103, + "num_tokens": 801201433.0, + "step": 30312 + }, + { + "epoch": 2.4996289271872683, + "grad_norm": 2.195672035217285, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7606480121612549, + "num_tokens": 801228920.0, + "step": 30313 + }, + { + "epoch": 2.4997113878123196, + "grad_norm": 1.8258140087127686, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.7822677493095398, + "num_tokens": 801267112.0, + "step": 30314 + }, + { + "epoch": 2.4997938484373714, + "grad_norm": 1.8703972101211548, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.7969595789909363, + "num_tokens": 801302247.0, + "step": 30315 + }, + { + "epoch": 2.4998763090624228, + "grad_norm": 2.076819658279419, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7637912631034851, + "num_tokens": 801332792.0, + "step": 30316 + }, + { + "epoch": 2.499958769687474, + "grad_norm": 2.184145927429199, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.7804960012435913, + "num_tokens": 801360844.0, + "step": 30317 + }, + { + "epoch": 2.500041230312526, + "grad_norm": 1.8572577238082886, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7448121905326843, + "num_tokens": 801397484.0, + "step": 30318 + }, + { + "epoch": 2.5001236909375772, + "grad_norm": 2.053358554840088, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.781768798828125, + "num_tokens": 801425921.0, + "step": 30319 + }, + { + "epoch": 2.5002061515626286, + "grad_norm": 1.9806901216506958, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7553422451019287, + "num_tokens": 801456723.0, + "step": 30320 + }, + { + "epoch": 2.5002886121876804, + "grad_norm": 2.281336545944214, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7340899109840393, + "num_tokens": 801482804.0, + "step": 30321 + }, + { + "epoch": 2.500371072812732, + "grad_norm": 1.8774926662445068, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7565017938613892, + "num_tokens": 801516988.0, + "step": 30322 + }, + { + "epoch": 2.5004535334377835, + "grad_norm": 2.009718894958496, + "learning_rate": 1e-06, + "loss": 0.6114, + "mean_token_accuracy": 0.7806243896484375, + "num_tokens": 801548769.0, + "step": 30323 + }, + { + "epoch": 2.500535994062835, + "grad_norm": 2.202100992202759, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7635208368301392, + "num_tokens": 801576332.0, + "step": 30324 + }, + { + "epoch": 2.5006184546878867, + "grad_norm": 2.31671404838562, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.7781243324279785, + "num_tokens": 801600837.0, + "step": 30325 + }, + { + "epoch": 2.500700915312938, + "grad_norm": 1.96173095703125, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7767243385314941, + "num_tokens": 801632167.0, + "step": 30326 + }, + { + "epoch": 2.5007833759379894, + "grad_norm": 2.006347894668579, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7754968404769897, + "num_tokens": 801665603.0, + "step": 30327 + }, + { + "epoch": 2.500865836563041, + "grad_norm": 2.027970552444458, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.7918513417243958, + "num_tokens": 801696149.0, + "step": 30328 + }, + { + "epoch": 2.5009482971880925, + "grad_norm": 2.0747578144073486, + "learning_rate": 1e-06, + "loss": 0.7012, + "mean_token_accuracy": 0.7721973061561584, + "num_tokens": 801728172.0, + "step": 30329 + }, + { + "epoch": 2.5010307578131443, + "grad_norm": 2.1667723655700684, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7672678232192993, + "num_tokens": 801755762.0, + "step": 30330 + }, + { + "epoch": 2.5011132184381957, + "grad_norm": 2.172760248184204, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7773205041885376, + "num_tokens": 801785101.0, + "step": 30331 + }, + { + "epoch": 2.5011956790632475, + "grad_norm": 2.1979286670684814, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7580045461654663, + "num_tokens": 801814348.0, + "step": 30332 + }, + { + "epoch": 2.501278139688299, + "grad_norm": 2.0505518913269043, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.7981274127960205, + "num_tokens": 801844425.0, + "step": 30333 + }, + { + "epoch": 2.50136060031335, + "grad_norm": 2.227750301361084, + "learning_rate": 1e-06, + "loss": 0.6103, + "mean_token_accuracy": 0.788754403591156, + "num_tokens": 801870822.0, + "step": 30334 + }, + { + "epoch": 2.501443060938402, + "grad_norm": 1.8577890396118164, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7605998516082764, + "num_tokens": 801909541.0, + "step": 30335 + }, + { + "epoch": 2.5015255215634533, + "grad_norm": 2.1994354724884033, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7429000735282898, + "num_tokens": 801939049.0, + "step": 30336 + }, + { + "epoch": 2.501607982188505, + "grad_norm": 2.266388416290283, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7673825025558472, + "num_tokens": 801967676.0, + "step": 30337 + }, + { + "epoch": 2.5016904428135565, + "grad_norm": 1.9763116836547852, + "learning_rate": 1e-06, + "loss": 0.5009, + "mean_token_accuracy": 0.8239487409591675, + "num_tokens": 801995737.0, + "step": 30338 + }, + { + "epoch": 2.5017729034386083, + "grad_norm": 2.056403160095215, + "learning_rate": 1e-06, + "loss": 0.5959, + "mean_token_accuracy": 0.8026837706565857, + "num_tokens": 802024983.0, + "step": 30339 + }, + { + "epoch": 2.5018553640636596, + "grad_norm": 2.298691749572754, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7565305233001709, + "num_tokens": 802051627.0, + "step": 30340 + }, + { + "epoch": 2.501937824688711, + "grad_norm": 1.9835929870605469, + "learning_rate": 1e-06, + "loss": 0.7023, + "mean_token_accuracy": 0.7806482315063477, + "num_tokens": 802083677.0, + "step": 30341 + }, + { + "epoch": 2.5020202853137627, + "grad_norm": 2.2125799655914307, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7783490419387817, + "num_tokens": 802112634.0, + "step": 30342 + }, + { + "epoch": 2.502102745938814, + "grad_norm": 2.37538480758667, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7731703519821167, + "num_tokens": 802142894.0, + "step": 30343 + }, + { + "epoch": 2.502185206563866, + "grad_norm": 2.010952949523926, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7767797708511353, + "num_tokens": 802174992.0, + "step": 30344 + }, + { + "epoch": 2.5022676671889172, + "grad_norm": 2.3535807132720947, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7530070543289185, + "num_tokens": 802201333.0, + "step": 30345 + }, + { + "epoch": 2.502350127813969, + "grad_norm": 2.148449659347534, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7589902877807617, + "num_tokens": 802239129.0, + "step": 30346 + }, + { + "epoch": 2.5024325884390204, + "grad_norm": 2.280306577682495, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.7833743095397949, + "num_tokens": 802265819.0, + "step": 30347 + }, + { + "epoch": 2.5025150490640717, + "grad_norm": 1.937722086906433, + "learning_rate": 1e-06, + "loss": 0.5983, + "mean_token_accuracy": 0.7878044247627258, + "num_tokens": 802303447.0, + "step": 30348 + }, + { + "epoch": 2.5025975096891235, + "grad_norm": 1.9376624822616577, + "learning_rate": 1e-06, + "loss": 0.5739, + "mean_token_accuracy": 0.8079652786254883, + "num_tokens": 802338374.0, + "step": 30349 + }, + { + "epoch": 2.502679970314175, + "grad_norm": 2.2575933933258057, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7767090797424316, + "num_tokens": 802364028.0, + "step": 30350 + }, + { + "epoch": 2.5027624309392267, + "grad_norm": 2.3447539806365967, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7523375749588013, + "num_tokens": 802390242.0, + "step": 30351 + }, + { + "epoch": 2.502844891564278, + "grad_norm": 1.853736162185669, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7865146994590759, + "num_tokens": 802426230.0, + "step": 30352 + }, + { + "epoch": 2.50292735218933, + "grad_norm": 2.1578824520111084, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.7492051124572754, + "num_tokens": 802455424.0, + "step": 30353 + }, + { + "epoch": 2.503009812814381, + "grad_norm": 2.075514078140259, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.778290331363678, + "num_tokens": 802488661.0, + "step": 30354 + }, + { + "epoch": 2.5030922734394325, + "grad_norm": 2.214385986328125, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.7931068539619446, + "num_tokens": 802516150.0, + "step": 30355 + }, + { + "epoch": 2.5031747340644843, + "grad_norm": 1.9254283905029297, + "learning_rate": 1e-06, + "loss": 0.5338, + "mean_token_accuracy": 0.8197289705276489, + "num_tokens": 802546518.0, + "step": 30356 + }, + { + "epoch": 2.5032571946895357, + "grad_norm": 2.0033435821533203, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7753131985664368, + "num_tokens": 802576207.0, + "step": 30357 + }, + { + "epoch": 2.503339655314587, + "grad_norm": 1.7846685647964478, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.7894726991653442, + "num_tokens": 802611845.0, + "step": 30358 + }, + { + "epoch": 2.503422115939639, + "grad_norm": 2.491386890411377, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7504146099090576, + "num_tokens": 802636969.0, + "step": 30359 + }, + { + "epoch": 2.5035045765646906, + "grad_norm": 1.958550214767456, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7367791533470154, + "num_tokens": 802672290.0, + "step": 30360 + }, + { + "epoch": 2.503587037189742, + "grad_norm": 2.246868133544922, + "learning_rate": 1e-06, + "loss": 0.6039, + "mean_token_accuracy": 0.793706476688385, + "num_tokens": 802698889.0, + "step": 30361 + }, + { + "epoch": 2.5036694978147933, + "grad_norm": 1.9945012331008911, + "learning_rate": 1e-06, + "loss": 0.5742, + "mean_token_accuracy": 0.8109227418899536, + "num_tokens": 802733380.0, + "step": 30362 + }, + { + "epoch": 2.503751958439845, + "grad_norm": 1.8554610013961792, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.7921361923217773, + "num_tokens": 802772366.0, + "step": 30363 + }, + { + "epoch": 2.5038344190648965, + "grad_norm": 2.1038577556610107, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.736029863357544, + "num_tokens": 802803487.0, + "step": 30364 + }, + { + "epoch": 2.503916879689948, + "grad_norm": 2.089155673980713, + "learning_rate": 1e-06, + "loss": 0.5875, + "mean_token_accuracy": 0.8091332912445068, + "num_tokens": 802834680.0, + "step": 30365 + }, + { + "epoch": 2.5039993403149996, + "grad_norm": 2.0678763389587402, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7377734184265137, + "num_tokens": 802865526.0, + "step": 30366 + }, + { + "epoch": 2.5040818009400514, + "grad_norm": 2.266268014907837, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7604647874832153, + "num_tokens": 802894417.0, + "step": 30367 + }, + { + "epoch": 2.5041642615651027, + "grad_norm": 2.2043097019195557, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7832902669906616, + "num_tokens": 802922401.0, + "step": 30368 + }, + { + "epoch": 2.504246722190154, + "grad_norm": 2.1644232273101807, + "learning_rate": 1e-06, + "loss": 0.5903, + "mean_token_accuracy": 0.7952347993850708, + "num_tokens": 802951541.0, + "step": 30369 + }, + { + "epoch": 2.504329182815206, + "grad_norm": 1.9774435758590698, + "learning_rate": 1e-06, + "loss": 0.6218, + "mean_token_accuracy": 0.781682014465332, + "num_tokens": 802987878.0, + "step": 30370 + }, + { + "epoch": 2.5044116434402572, + "grad_norm": 2.069265365600586, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.7772078514099121, + "num_tokens": 803015688.0, + "step": 30371 + }, + { + "epoch": 2.5044941040653086, + "grad_norm": 2.0707125663757324, + "learning_rate": 1e-06, + "loss": 0.5745, + "mean_token_accuracy": 0.8089240789413452, + "num_tokens": 803046526.0, + "step": 30372 + }, + { + "epoch": 2.5045765646903604, + "grad_norm": 2.152421712875366, + "learning_rate": 1e-06, + "loss": 0.6875, + "mean_token_accuracy": 0.775322675704956, + "num_tokens": 803078137.0, + "step": 30373 + }, + { + "epoch": 2.5046590253154117, + "grad_norm": 2.0309078693389893, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7901666164398193, + "num_tokens": 803112560.0, + "step": 30374 + }, + { + "epoch": 2.5047414859404635, + "grad_norm": 1.8697351217269897, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.7855674624443054, + "num_tokens": 803148510.0, + "step": 30375 + }, + { + "epoch": 2.504823946565515, + "grad_norm": 2.3667993545532227, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.760506808757782, + "num_tokens": 803175709.0, + "step": 30376 + }, + { + "epoch": 2.5049064071905667, + "grad_norm": 2.0477168560028076, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7711108922958374, + "num_tokens": 803209541.0, + "step": 30377 + }, + { + "epoch": 2.504988867815618, + "grad_norm": 1.8650377988815308, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7700836658477783, + "num_tokens": 803245890.0, + "step": 30378 + }, + { + "epoch": 2.5050713284406694, + "grad_norm": 2.2234885692596436, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7724688053131104, + "num_tokens": 803272813.0, + "step": 30379 + }, + { + "epoch": 2.505153789065721, + "grad_norm": 2.0614328384399414, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7615861296653748, + "num_tokens": 803301878.0, + "step": 30380 + }, + { + "epoch": 2.5052362496907725, + "grad_norm": 2.2104620933532715, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.7854481935501099, + "num_tokens": 803325877.0, + "step": 30381 + }, + { + "epoch": 2.5053187103158243, + "grad_norm": 2.287046194076538, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7542414665222168, + "num_tokens": 803354198.0, + "step": 30382 + }, + { + "epoch": 2.5054011709408757, + "grad_norm": 2.1313836574554443, + "learning_rate": 1e-06, + "loss": 0.5989, + "mean_token_accuracy": 0.8023348450660706, + "num_tokens": 803380508.0, + "step": 30383 + }, + { + "epoch": 2.5054836315659275, + "grad_norm": 1.8847970962524414, + "learning_rate": 1e-06, + "loss": 0.5348, + "mean_token_accuracy": 0.814221978187561, + "num_tokens": 803412751.0, + "step": 30384 + }, + { + "epoch": 2.505566092190979, + "grad_norm": 2.26857852935791, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.7897454500198364, + "num_tokens": 803440330.0, + "step": 30385 + }, + { + "epoch": 2.50564855281603, + "grad_norm": 1.9475396871566772, + "learning_rate": 1e-06, + "loss": 0.5819, + "mean_token_accuracy": 0.7995457053184509, + "num_tokens": 803473044.0, + "step": 30386 + }, + { + "epoch": 2.505731013441082, + "grad_norm": 1.989661693572998, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7839862704277039, + "num_tokens": 803508661.0, + "step": 30387 + }, + { + "epoch": 2.5058134740661333, + "grad_norm": 2.138176441192627, + "learning_rate": 1e-06, + "loss": 0.5612, + "mean_token_accuracy": 0.8100374937057495, + "num_tokens": 803535602.0, + "step": 30388 + }, + { + "epoch": 2.505895934691185, + "grad_norm": 2.538273334503174, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.7907971143722534, + "num_tokens": 803564117.0, + "step": 30389 + }, + { + "epoch": 2.5059783953162365, + "grad_norm": 2.0597596168518066, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7748783826828003, + "num_tokens": 803593343.0, + "step": 30390 + }, + { + "epoch": 2.5060608559412882, + "grad_norm": 1.9441914558410645, + "learning_rate": 1e-06, + "loss": 0.6206, + "mean_token_accuracy": 0.8035234212875366, + "num_tokens": 803625523.0, + "step": 30391 + }, + { + "epoch": 2.5061433165663396, + "grad_norm": 2.155968427658081, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7629427909851074, + "num_tokens": 803656472.0, + "step": 30392 + }, + { + "epoch": 2.506225777191391, + "grad_norm": 1.9873168468475342, + "learning_rate": 1e-06, + "loss": 0.5188, + "mean_token_accuracy": 0.8225492835044861, + "num_tokens": 803689762.0, + "step": 30393 + }, + { + "epoch": 2.5063082378164427, + "grad_norm": 2.182978630065918, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.7923837900161743, + "num_tokens": 803717057.0, + "step": 30394 + }, + { + "epoch": 2.506390698441494, + "grad_norm": 2.151374340057373, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7815935611724854, + "num_tokens": 803745687.0, + "step": 30395 + }, + { + "epoch": 2.506473159066546, + "grad_norm": 2.274458408355713, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7673363089561462, + "num_tokens": 803774425.0, + "step": 30396 + }, + { + "epoch": 2.5065556196915972, + "grad_norm": 1.9039158821105957, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7577469348907471, + "num_tokens": 803811636.0, + "step": 30397 + }, + { + "epoch": 2.506638080316649, + "grad_norm": 2.001438617706299, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.7841365337371826, + "num_tokens": 803840776.0, + "step": 30398 + }, + { + "epoch": 2.5067205409417004, + "grad_norm": 2.0834128856658936, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7246158123016357, + "num_tokens": 803872132.0, + "step": 30399 + }, + { + "epoch": 2.5068030015667517, + "grad_norm": 2.195394992828369, + "learning_rate": 1e-06, + "loss": 0.5512, + "mean_token_accuracy": 0.7963657379150391, + "num_tokens": 803896979.0, + "step": 30400 + }, + { + "epoch": 2.5068854621918035, + "grad_norm": 2.3972280025482178, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7710113525390625, + "num_tokens": 803920736.0, + "step": 30401 + }, + { + "epoch": 2.506967922816855, + "grad_norm": 2.066373109817505, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7575933933258057, + "num_tokens": 803953196.0, + "step": 30402 + }, + { + "epoch": 2.5070503834419062, + "grad_norm": 1.7658112049102783, + "learning_rate": 1e-06, + "loss": 0.6225, + "mean_token_accuracy": 0.7935158014297485, + "num_tokens": 803991894.0, + "step": 30403 + }, + { + "epoch": 2.507132844066958, + "grad_norm": 2.3420746326446533, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.7857120037078857, + "num_tokens": 804020003.0, + "step": 30404 + }, + { + "epoch": 2.50721530469201, + "grad_norm": 1.884973168373108, + "learning_rate": 1e-06, + "loss": 0.6019, + "mean_token_accuracy": 0.8002258539199829, + "num_tokens": 804054567.0, + "step": 30405 + }, + { + "epoch": 2.507297765317061, + "grad_norm": 2.308067798614502, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7397844791412354, + "num_tokens": 804081119.0, + "step": 30406 + }, + { + "epoch": 2.5073802259421125, + "grad_norm": 2.106301784515381, + "learning_rate": 1e-06, + "loss": 0.6128, + "mean_token_accuracy": 0.7975793480873108, + "num_tokens": 804108586.0, + "step": 30407 + }, + { + "epoch": 2.5074626865671643, + "grad_norm": 2.46128249168396, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7683047652244568, + "num_tokens": 804132250.0, + "step": 30408 + }, + { + "epoch": 2.5075451471922157, + "grad_norm": 2.2981457710266113, + "learning_rate": 1e-06, + "loss": 0.7303, + "mean_token_accuracy": 0.7628692388534546, + "num_tokens": 804159055.0, + "step": 30409 + }, + { + "epoch": 2.507627607817267, + "grad_norm": 1.8773301839828491, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7527185678482056, + "num_tokens": 804194672.0, + "step": 30410 + }, + { + "epoch": 2.507710068442319, + "grad_norm": 2.6881754398345947, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7415993809700012, + "num_tokens": 804216126.0, + "step": 30411 + }, + { + "epoch": 2.5077925290673706, + "grad_norm": 2.353487253189087, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7621687054634094, + "num_tokens": 804241273.0, + "step": 30412 + }, + { + "epoch": 2.507874989692422, + "grad_norm": 1.9371193647384644, + "learning_rate": 1e-06, + "loss": 0.6048, + "mean_token_accuracy": 0.7935397028923035, + "num_tokens": 804277811.0, + "step": 30413 + }, + { + "epoch": 2.5079574503174733, + "grad_norm": 2.395754098892212, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7609095573425293, + "num_tokens": 804301900.0, + "step": 30414 + }, + { + "epoch": 2.508039910942525, + "grad_norm": 1.9961272478103638, + "learning_rate": 1e-06, + "loss": 0.6024, + "mean_token_accuracy": 0.8020612001419067, + "num_tokens": 804333664.0, + "step": 30415 + }, + { + "epoch": 2.5081223715675764, + "grad_norm": 2.2997970581054688, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7824995517730713, + "num_tokens": 804358805.0, + "step": 30416 + }, + { + "epoch": 2.508204832192628, + "grad_norm": 2.0709493160247803, + "learning_rate": 1e-06, + "loss": 0.6645, + "mean_token_accuracy": 0.7756227254867554, + "num_tokens": 804392369.0, + "step": 30417 + }, + { + "epoch": 2.5082872928176796, + "grad_norm": 1.8742759227752686, + "learning_rate": 1e-06, + "loss": 0.5355, + "mean_token_accuracy": 0.8218459486961365, + "num_tokens": 804429312.0, + "step": 30418 + }, + { + "epoch": 2.508369753442731, + "grad_norm": 2.519388198852539, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7643272280693054, + "num_tokens": 804452003.0, + "step": 30419 + }, + { + "epoch": 2.5084522140677827, + "grad_norm": 2.255289316177368, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7763947248458862, + "num_tokens": 804476846.0, + "step": 30420 + }, + { + "epoch": 2.508534674692834, + "grad_norm": 2.4038074016571045, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7430923581123352, + "num_tokens": 804502460.0, + "step": 30421 + }, + { + "epoch": 2.508617135317886, + "grad_norm": 2.4799225330352783, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7663308382034302, + "num_tokens": 804525052.0, + "step": 30422 + }, + { + "epoch": 2.5086995959429372, + "grad_norm": 2.0143492221832275, + "learning_rate": 1e-06, + "loss": 0.6075, + "mean_token_accuracy": 0.8059890866279602, + "num_tokens": 804556313.0, + "step": 30423 + }, + { + "epoch": 2.5087820565679886, + "grad_norm": 1.8929977416992188, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7596575617790222, + "num_tokens": 804589326.0, + "step": 30424 + }, + { + "epoch": 2.5088645171930404, + "grad_norm": 2.197296619415283, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.7679476737976074, + "num_tokens": 804616774.0, + "step": 30425 + }, + { + "epoch": 2.5089469778180917, + "grad_norm": 2.0605037212371826, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7706737518310547, + "num_tokens": 804648303.0, + "step": 30426 + }, + { + "epoch": 2.5090294384431435, + "grad_norm": 1.851070523262024, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7574383020401001, + "num_tokens": 804685185.0, + "step": 30427 + }, + { + "epoch": 2.509111899068195, + "grad_norm": 1.7888544797897339, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7718429565429688, + "num_tokens": 804725389.0, + "step": 30428 + }, + { + "epoch": 2.5091943596932467, + "grad_norm": 2.2601430416107178, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7391140460968018, + "num_tokens": 804755138.0, + "step": 30429 + }, + { + "epoch": 2.509276820318298, + "grad_norm": 2.1029539108276367, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7577793598175049, + "num_tokens": 804785668.0, + "step": 30430 + }, + { + "epoch": 2.5093592809433494, + "grad_norm": 2.3198986053466797, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7825750112533569, + "num_tokens": 804811165.0, + "step": 30431 + }, + { + "epoch": 2.509441741568401, + "grad_norm": 2.2250452041625977, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7503386735916138, + "num_tokens": 804840257.0, + "step": 30432 + }, + { + "epoch": 2.5095242021934525, + "grad_norm": 2.2775182723999023, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7466801404953003, + "num_tokens": 804869783.0, + "step": 30433 + }, + { + "epoch": 2.5096066628185043, + "grad_norm": 2.3597934246063232, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7550114393234253, + "num_tokens": 804895559.0, + "step": 30434 + }, + { + "epoch": 2.5096891234435557, + "grad_norm": 2.1586010456085205, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.7757269740104675, + "num_tokens": 804929541.0, + "step": 30435 + }, + { + "epoch": 2.5097715840686075, + "grad_norm": 2.198542356491089, + "learning_rate": 1e-06, + "loss": 0.7205, + "mean_token_accuracy": 0.7587774991989136, + "num_tokens": 804959926.0, + "step": 30436 + }, + { + "epoch": 2.509854044693659, + "grad_norm": 2.1987788677215576, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.768720805644989, + "num_tokens": 804988286.0, + "step": 30437 + }, + { + "epoch": 2.50993650531871, + "grad_norm": 2.1236326694488525, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7568643093109131, + "num_tokens": 805020370.0, + "step": 30438 + }, + { + "epoch": 2.510018965943762, + "grad_norm": 2.0502066612243652, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7538926601409912, + "num_tokens": 805053183.0, + "step": 30439 + }, + { + "epoch": 2.5101014265688133, + "grad_norm": 2.1575276851654053, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.7742222547531128, + "num_tokens": 805081321.0, + "step": 30440 + }, + { + "epoch": 2.510183887193865, + "grad_norm": 2.1601901054382324, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7766762971878052, + "num_tokens": 805111468.0, + "step": 30441 + }, + { + "epoch": 2.5102663478189164, + "grad_norm": 2.1783487796783447, + "learning_rate": 1e-06, + "loss": 0.5337, + "mean_token_accuracy": 0.8192775249481201, + "num_tokens": 805140964.0, + "step": 30442 + }, + { + "epoch": 2.5103488084439682, + "grad_norm": 2.0145437717437744, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7584313154220581, + "num_tokens": 805172448.0, + "step": 30443 + }, + { + "epoch": 2.5104312690690196, + "grad_norm": 2.0028774738311768, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7829238176345825, + "num_tokens": 805207353.0, + "step": 30444 + }, + { + "epoch": 2.510513729694071, + "grad_norm": 2.433326244354248, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7741003632545471, + "num_tokens": 805231129.0, + "step": 30445 + }, + { + "epoch": 2.5105961903191227, + "grad_norm": 1.9298650026321411, + "learning_rate": 1e-06, + "loss": 0.4781, + "mean_token_accuracy": 0.8352357745170593, + "num_tokens": 805261697.0, + "step": 30446 + }, + { + "epoch": 2.510678650944174, + "grad_norm": 1.962471842765808, + "learning_rate": 1e-06, + "loss": 0.5713, + "mean_token_accuracy": 0.8123046159744263, + "num_tokens": 805294750.0, + "step": 30447 + }, + { + "epoch": 2.5107611115692254, + "grad_norm": 1.8196231126785278, + "learning_rate": 1e-06, + "loss": 0.5858, + "mean_token_accuracy": 0.816536545753479, + "num_tokens": 805330203.0, + "step": 30448 + }, + { + "epoch": 2.5108435721942772, + "grad_norm": 2.1253912448883057, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.7863523960113525, + "num_tokens": 805361244.0, + "step": 30449 + }, + { + "epoch": 2.510926032819329, + "grad_norm": 2.106297492980957, + "learning_rate": 1e-06, + "loss": 0.8014, + "mean_token_accuracy": 0.7449266314506531, + "num_tokens": 805394477.0, + "step": 30450 + }, + { + "epoch": 2.5110084934443804, + "grad_norm": 2.1803579330444336, + "learning_rate": 1e-06, + "loss": 0.5762, + "mean_token_accuracy": 0.8122701048851013, + "num_tokens": 805422089.0, + "step": 30451 + }, + { + "epoch": 2.5110909540694317, + "grad_norm": 2.121211528778076, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.7712515592575073, + "num_tokens": 805452265.0, + "step": 30452 + }, + { + "epoch": 2.5111734146944835, + "grad_norm": 1.9609915018081665, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.7887449860572815, + "num_tokens": 805486857.0, + "step": 30453 + }, + { + "epoch": 2.511255875319535, + "grad_norm": 2.2941436767578125, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.7844330072402954, + "num_tokens": 805511816.0, + "step": 30454 + }, + { + "epoch": 2.511338335944586, + "grad_norm": 2.1218576431274414, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.7853589057922363, + "num_tokens": 805542963.0, + "step": 30455 + }, + { + "epoch": 2.511420796569638, + "grad_norm": 2.011613607406616, + "learning_rate": 1e-06, + "loss": 0.5804, + "mean_token_accuracy": 0.8216960430145264, + "num_tokens": 805571667.0, + "step": 30456 + }, + { + "epoch": 2.51150325719469, + "grad_norm": 2.207994222640991, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7370874881744385, + "num_tokens": 805603206.0, + "step": 30457 + }, + { + "epoch": 2.511585717819741, + "grad_norm": 2.0776748657226562, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7732586860656738, + "num_tokens": 805632481.0, + "step": 30458 + }, + { + "epoch": 2.5116681784447925, + "grad_norm": 1.818825364112854, + "learning_rate": 1e-06, + "loss": 0.5609, + "mean_token_accuracy": 0.8065081238746643, + "num_tokens": 805666744.0, + "step": 30459 + }, + { + "epoch": 2.5117506390698443, + "grad_norm": 1.815306305885315, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.774619460105896, + "num_tokens": 805707743.0, + "step": 30460 + }, + { + "epoch": 2.5118330996948957, + "grad_norm": 2.7260239124298096, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.75443434715271, + "num_tokens": 805729726.0, + "step": 30461 + }, + { + "epoch": 2.511915560319947, + "grad_norm": 1.9855766296386719, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.7778282761573792, + "num_tokens": 805762764.0, + "step": 30462 + }, + { + "epoch": 2.511998020944999, + "grad_norm": 2.302304983139038, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7618546485900879, + "num_tokens": 805788462.0, + "step": 30463 + }, + { + "epoch": 2.51208048157005, + "grad_norm": 2.364194631576538, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7514612078666687, + "num_tokens": 805815575.0, + "step": 30464 + }, + { + "epoch": 2.512162942195102, + "grad_norm": 2.1459038257598877, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.7893770933151245, + "num_tokens": 805846545.0, + "step": 30465 + }, + { + "epoch": 2.5122454028201533, + "grad_norm": 2.189512014389038, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7591938972473145, + "num_tokens": 805874246.0, + "step": 30466 + }, + { + "epoch": 2.512327863445205, + "grad_norm": 1.952337622642517, + "learning_rate": 1e-06, + "loss": 0.6042, + "mean_token_accuracy": 0.7975854277610779, + "num_tokens": 805904759.0, + "step": 30467 + }, + { + "epoch": 2.5124103240702564, + "grad_norm": 2.264460563659668, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7735192775726318, + "num_tokens": 805929143.0, + "step": 30468 + }, + { + "epoch": 2.512492784695308, + "grad_norm": 2.0345101356506348, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.785756528377533, + "num_tokens": 805961502.0, + "step": 30469 + }, + { + "epoch": 2.5125752453203596, + "grad_norm": 1.9027414321899414, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.7954772710800171, + "num_tokens": 805995304.0, + "step": 30470 + }, + { + "epoch": 2.512657705945411, + "grad_norm": 1.8031761646270752, + "learning_rate": 1e-06, + "loss": 0.4934, + "mean_token_accuracy": 0.8323124647140503, + "num_tokens": 806027525.0, + "step": 30471 + }, + { + "epoch": 2.5127401665704627, + "grad_norm": 1.9254683256149292, + "learning_rate": 1e-06, + "loss": 0.5773, + "mean_token_accuracy": 0.8099934458732605, + "num_tokens": 806061687.0, + "step": 30472 + }, + { + "epoch": 2.512822627195514, + "grad_norm": 2.3096415996551514, + "learning_rate": 1e-06, + "loss": 0.7738, + "mean_token_accuracy": 0.7544240951538086, + "num_tokens": 806089536.0, + "step": 30473 + }, + { + "epoch": 2.512905087820566, + "grad_norm": 2.2109601497650146, + "learning_rate": 1e-06, + "loss": 0.5354, + "mean_token_accuracy": 0.8155136108398438, + "num_tokens": 806116667.0, + "step": 30474 + }, + { + "epoch": 2.5129875484456172, + "grad_norm": 2.0391159057617188, + "learning_rate": 1e-06, + "loss": 0.622, + "mean_token_accuracy": 0.7843031287193298, + "num_tokens": 806146244.0, + "step": 30475 + }, + { + "epoch": 2.5130700090706686, + "grad_norm": 2.1890804767608643, + "learning_rate": 1e-06, + "loss": 0.7635, + "mean_token_accuracy": 0.7580305337905884, + "num_tokens": 806175512.0, + "step": 30476 + }, + { + "epoch": 2.5131524696957204, + "grad_norm": 1.916200876235962, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.7966820001602173, + "num_tokens": 806210372.0, + "step": 30477 + }, + { + "epoch": 2.5132349303207717, + "grad_norm": 1.9405733346939087, + "learning_rate": 1e-06, + "loss": 0.612, + "mean_token_accuracy": 0.7941891551017761, + "num_tokens": 806242018.0, + "step": 30478 + }, + { + "epoch": 2.5133173909458235, + "grad_norm": 2.0202531814575195, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.7827692031860352, + "num_tokens": 806272218.0, + "step": 30479 + }, + { + "epoch": 2.513399851570875, + "grad_norm": 2.13934063911438, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7626758813858032, + "num_tokens": 806301803.0, + "step": 30480 + }, + { + "epoch": 2.5134823121959267, + "grad_norm": 2.357832431793213, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7463548183441162, + "num_tokens": 806330891.0, + "step": 30481 + }, + { + "epoch": 2.513564772820978, + "grad_norm": 1.9676567316055298, + "learning_rate": 1e-06, + "loss": 0.4681, + "mean_token_accuracy": 0.8326565027236938, + "num_tokens": 806360642.0, + "step": 30482 + }, + { + "epoch": 2.5136472334460294, + "grad_norm": 2.1368966102600098, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7483699321746826, + "num_tokens": 806389580.0, + "step": 30483 + }, + { + "epoch": 2.513729694071081, + "grad_norm": 1.9186352491378784, + "learning_rate": 1e-06, + "loss": 0.704, + "mean_token_accuracy": 0.7723788022994995, + "num_tokens": 806427681.0, + "step": 30484 + }, + { + "epoch": 2.5138121546961325, + "grad_norm": 1.8406553268432617, + "learning_rate": 1e-06, + "loss": 0.5838, + "mean_token_accuracy": 0.8126353025436401, + "num_tokens": 806464237.0, + "step": 30485 + }, + { + "epoch": 2.5138946153211843, + "grad_norm": 2.6889185905456543, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7385029792785645, + "num_tokens": 806485426.0, + "step": 30486 + }, + { + "epoch": 2.5139770759462357, + "grad_norm": 2.4221761226654053, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7600706815719604, + "num_tokens": 806508572.0, + "step": 30487 + }, + { + "epoch": 2.5140595365712874, + "grad_norm": 1.9605755805969238, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7831120491027832, + "num_tokens": 806545471.0, + "step": 30488 + }, + { + "epoch": 2.514141997196339, + "grad_norm": 2.1255102157592773, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.7900327444076538, + "num_tokens": 806575907.0, + "step": 30489 + }, + { + "epoch": 2.51422445782139, + "grad_norm": 2.1982810497283936, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.774936318397522, + "num_tokens": 806604721.0, + "step": 30490 + }, + { + "epoch": 2.514306918446442, + "grad_norm": 1.803965449333191, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7750396728515625, + "num_tokens": 806644935.0, + "step": 30491 + }, + { + "epoch": 2.5143893790714933, + "grad_norm": 2.0929291248321533, + "learning_rate": 1e-06, + "loss": 0.5865, + "mean_token_accuracy": 0.8044078350067139, + "num_tokens": 806671631.0, + "step": 30492 + }, + { + "epoch": 2.5144718396965446, + "grad_norm": 1.9222285747528076, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7747514247894287, + "num_tokens": 806702074.0, + "step": 30493 + }, + { + "epoch": 2.5145543003215964, + "grad_norm": 2.0977871417999268, + "learning_rate": 1e-06, + "loss": 0.5783, + "mean_token_accuracy": 0.8002578616142273, + "num_tokens": 806732719.0, + "step": 30494 + }, + { + "epoch": 2.5146367609466482, + "grad_norm": 1.9594956636428833, + "learning_rate": 1e-06, + "loss": 0.576, + "mean_token_accuracy": 0.813759446144104, + "num_tokens": 806765569.0, + "step": 30495 + }, + { + "epoch": 2.5147192215716996, + "grad_norm": 2.1405680179595947, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7590065002441406, + "num_tokens": 806798262.0, + "step": 30496 + }, + { + "epoch": 2.514801682196751, + "grad_norm": 2.0855605602264404, + "learning_rate": 1e-06, + "loss": 0.6358, + "mean_token_accuracy": 0.7923868894577026, + "num_tokens": 806827514.0, + "step": 30497 + }, + { + "epoch": 2.5148841428218027, + "grad_norm": 2.462824821472168, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7643142938613892, + "num_tokens": 806852223.0, + "step": 30498 + }, + { + "epoch": 2.514966603446854, + "grad_norm": 1.9956951141357422, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7505118250846863, + "num_tokens": 806884529.0, + "step": 30499 + }, + { + "epoch": 2.5150490640719054, + "grad_norm": 2.3160338401794434, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7553061246871948, + "num_tokens": 806911319.0, + "step": 30500 + }, + { + "epoch": 2.515131524696957, + "grad_norm": 2.005840301513672, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.7926760911941528, + "num_tokens": 806945935.0, + "step": 30501 + }, + { + "epoch": 2.515213985322009, + "grad_norm": 1.8414617776870728, + "learning_rate": 1e-06, + "loss": 0.5616, + "mean_token_accuracy": 0.8158091902732849, + "num_tokens": 806979822.0, + "step": 30502 + }, + { + "epoch": 2.5152964459470604, + "grad_norm": 2.2828361988067627, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7507440447807312, + "num_tokens": 807006979.0, + "step": 30503 + }, + { + "epoch": 2.5153789065721117, + "grad_norm": 2.225872755050659, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7726614475250244, + "num_tokens": 807033719.0, + "step": 30504 + }, + { + "epoch": 2.5154613671971635, + "grad_norm": 1.9067227840423584, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7287410497665405, + "num_tokens": 807073157.0, + "step": 30505 + }, + { + "epoch": 2.515543827822215, + "grad_norm": 1.9785343408584595, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7519938945770264, + "num_tokens": 807107202.0, + "step": 30506 + }, + { + "epoch": 2.515626288447266, + "grad_norm": 2.4043426513671875, + "learning_rate": 1e-06, + "loss": 0.5769, + "mean_token_accuracy": 0.8083690404891968, + "num_tokens": 807128844.0, + "step": 30507 + }, + { + "epoch": 2.515708749072318, + "grad_norm": 1.9302873611450195, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.770775318145752, + "num_tokens": 807166078.0, + "step": 30508 + }, + { + "epoch": 2.5157912096973694, + "grad_norm": 2.0117292404174805, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7801252603530884, + "num_tokens": 807197087.0, + "step": 30509 + }, + { + "epoch": 2.515873670322421, + "grad_norm": 1.8716344833374023, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7716884613037109, + "num_tokens": 807231665.0, + "step": 30510 + }, + { + "epoch": 2.5159561309474725, + "grad_norm": 2.535128355026245, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7658427953720093, + "num_tokens": 807254044.0, + "step": 30511 + }, + { + "epoch": 2.5160385915725243, + "grad_norm": 2.3212318420410156, + "learning_rate": 1e-06, + "loss": 0.5879, + "mean_token_accuracy": 0.7909717559814453, + "num_tokens": 807277863.0, + "step": 30512 + }, + { + "epoch": 2.5161210521975756, + "grad_norm": 2.0860660076141357, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7793886661529541, + "num_tokens": 807305120.0, + "step": 30513 + }, + { + "epoch": 2.516203512822627, + "grad_norm": 2.2314951419830322, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7866116762161255, + "num_tokens": 807332593.0, + "step": 30514 + }, + { + "epoch": 2.516285973447679, + "grad_norm": 1.9411689043045044, + "learning_rate": 1e-06, + "loss": 0.5835, + "mean_token_accuracy": 0.799209475517273, + "num_tokens": 807366141.0, + "step": 30515 + }, + { + "epoch": 2.51636843407273, + "grad_norm": 2.3168091773986816, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.7851269245147705, + "num_tokens": 807389000.0, + "step": 30516 + }, + { + "epoch": 2.516450894697782, + "grad_norm": 2.2076942920684814, + "learning_rate": 1e-06, + "loss": 0.716, + "mean_token_accuracy": 0.7716546058654785, + "num_tokens": 807415830.0, + "step": 30517 + }, + { + "epoch": 2.5165333553228333, + "grad_norm": 2.143577814102173, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.7727909088134766, + "num_tokens": 807440856.0, + "step": 30518 + }, + { + "epoch": 2.516615815947885, + "grad_norm": 2.1611440181732178, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7364109754562378, + "num_tokens": 807472260.0, + "step": 30519 + }, + { + "epoch": 2.5166982765729364, + "grad_norm": 1.9500987529754639, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7735782861709595, + "num_tokens": 807504691.0, + "step": 30520 + }, + { + "epoch": 2.516780737197988, + "grad_norm": 2.45047664642334, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7411202788352966, + "num_tokens": 807529282.0, + "step": 30521 + }, + { + "epoch": 2.5168631978230396, + "grad_norm": 2.351299285888672, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7707151174545288, + "num_tokens": 807553952.0, + "step": 30522 + }, + { + "epoch": 2.516945658448091, + "grad_norm": 2.104379177093506, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7778273820877075, + "num_tokens": 807583824.0, + "step": 30523 + }, + { + "epoch": 2.5170281190731427, + "grad_norm": 2.0042433738708496, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7792078852653503, + "num_tokens": 807618655.0, + "step": 30524 + }, + { + "epoch": 2.517110579698194, + "grad_norm": 1.9554975032806396, + "learning_rate": 1e-06, + "loss": 0.532, + "mean_token_accuracy": 0.8123330473899841, + "num_tokens": 807653745.0, + "step": 30525 + }, + { + "epoch": 2.517193040323246, + "grad_norm": 1.814522624015808, + "learning_rate": 1e-06, + "loss": 0.6118, + "mean_token_accuracy": 0.799092173576355, + "num_tokens": 807688470.0, + "step": 30526 + }, + { + "epoch": 2.517275500948297, + "grad_norm": 2.212347984313965, + "learning_rate": 1e-06, + "loss": 0.5848, + "mean_token_accuracy": 0.796499490737915, + "num_tokens": 807716515.0, + "step": 30527 + }, + { + "epoch": 2.5173579615733486, + "grad_norm": 2.094719171524048, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7578786015510559, + "num_tokens": 807744936.0, + "step": 30528 + }, + { + "epoch": 2.5174404221984004, + "grad_norm": 1.8299144506454468, + "learning_rate": 1e-06, + "loss": 0.6225, + "mean_token_accuracy": 0.8008600473403931, + "num_tokens": 807780616.0, + "step": 30529 + }, + { + "epoch": 2.5175228828234517, + "grad_norm": 1.9864656925201416, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7666869163513184, + "num_tokens": 807815777.0, + "step": 30530 + }, + { + "epoch": 2.517605343448503, + "grad_norm": 2.0522844791412354, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.7929703593254089, + "num_tokens": 807847171.0, + "step": 30531 + }, + { + "epoch": 2.517687804073555, + "grad_norm": 1.9598942995071411, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7703432440757751, + "num_tokens": 807883820.0, + "step": 30532 + }, + { + "epoch": 2.5177702646986067, + "grad_norm": 2.165144205093384, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.7859011888504028, + "num_tokens": 807914113.0, + "step": 30533 + }, + { + "epoch": 2.517852725323658, + "grad_norm": 1.991621732711792, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7485568523406982, + "num_tokens": 807948604.0, + "step": 30534 + }, + { + "epoch": 2.5179351859487094, + "grad_norm": 1.8790580034255981, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7712481021881104, + "num_tokens": 807982509.0, + "step": 30535 + }, + { + "epoch": 2.518017646573761, + "grad_norm": 2.0001535415649414, + "learning_rate": 1e-06, + "loss": 0.6493, + "mean_token_accuracy": 0.7812677621841431, + "num_tokens": 808016165.0, + "step": 30536 + }, + { + "epoch": 2.5181001071988125, + "grad_norm": 2.1557276248931885, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7740609645843506, + "num_tokens": 808045369.0, + "step": 30537 + }, + { + "epoch": 2.518182567823864, + "grad_norm": 2.3109827041625977, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.7896100282669067, + "num_tokens": 808070714.0, + "step": 30538 + }, + { + "epoch": 2.5182650284489156, + "grad_norm": 1.914764404296875, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.795799195766449, + "num_tokens": 808103374.0, + "step": 30539 + }, + { + "epoch": 2.5183474890739674, + "grad_norm": 2.6210832595825195, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.757357656955719, + "num_tokens": 808125150.0, + "step": 30540 + }, + { + "epoch": 2.518429949699019, + "grad_norm": 2.417309045791626, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7633559703826904, + "num_tokens": 808150833.0, + "step": 30541 + }, + { + "epoch": 2.51851241032407, + "grad_norm": 2.0965261459350586, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7748054265975952, + "num_tokens": 808181390.0, + "step": 30542 + }, + { + "epoch": 2.518594870949122, + "grad_norm": 1.9851429462432861, + "learning_rate": 1e-06, + "loss": 0.6959, + "mean_token_accuracy": 0.7667210698127747, + "num_tokens": 808213544.0, + "step": 30543 + }, + { + "epoch": 2.5186773315741733, + "grad_norm": 2.100496292114258, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7495615482330322, + "num_tokens": 808243139.0, + "step": 30544 + }, + { + "epoch": 2.5187597921992246, + "grad_norm": 2.0027413368225098, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.7855325937271118, + "num_tokens": 808276221.0, + "step": 30545 + }, + { + "epoch": 2.5188422528242764, + "grad_norm": 2.193924903869629, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7452507019042969, + "num_tokens": 808306727.0, + "step": 30546 + }, + { + "epoch": 2.5189247134493282, + "grad_norm": 2.1361753940582275, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.7715175747871399, + "num_tokens": 808333912.0, + "step": 30547 + }, + { + "epoch": 2.5190071740743796, + "grad_norm": 1.7952812910079956, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7507781982421875, + "num_tokens": 808373351.0, + "step": 30548 + }, + { + "epoch": 2.519089634699431, + "grad_norm": 1.9551079273223877, + "learning_rate": 1e-06, + "loss": 0.5763, + "mean_token_accuracy": 0.806581974029541, + "num_tokens": 808405800.0, + "step": 30549 + }, + { + "epoch": 2.5191720953244827, + "grad_norm": 1.7802678346633911, + "learning_rate": 1e-06, + "loss": 0.6497, + "mean_token_accuracy": 0.7914680242538452, + "num_tokens": 808443694.0, + "step": 30550 + }, + { + "epoch": 2.519254555949534, + "grad_norm": 2.0791256427764893, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.7801084518432617, + "num_tokens": 808475776.0, + "step": 30551 + }, + { + "epoch": 2.5193370165745854, + "grad_norm": 1.986807107925415, + "learning_rate": 1e-06, + "loss": 0.4989, + "mean_token_accuracy": 0.8359639048576355, + "num_tokens": 808503604.0, + "step": 30552 + }, + { + "epoch": 2.519419477199637, + "grad_norm": 2.2882723808288574, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7626895308494568, + "num_tokens": 808530248.0, + "step": 30553 + }, + { + "epoch": 2.5195019378246886, + "grad_norm": 2.2762672901153564, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7405084371566772, + "num_tokens": 808558933.0, + "step": 30554 + }, + { + "epoch": 2.5195843984497404, + "grad_norm": 1.9713445901870728, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7877504229545593, + "num_tokens": 808591306.0, + "step": 30555 + }, + { + "epoch": 2.5196668590747917, + "grad_norm": 1.9404186010360718, + "learning_rate": 1e-06, + "loss": 0.5978, + "mean_token_accuracy": 0.8011962175369263, + "num_tokens": 808627303.0, + "step": 30556 + }, + { + "epoch": 2.5197493196998435, + "grad_norm": 2.041794776916504, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.7837923765182495, + "num_tokens": 808655385.0, + "step": 30557 + }, + { + "epoch": 2.519831780324895, + "grad_norm": 1.9964390993118286, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7720263600349426, + "num_tokens": 808687828.0, + "step": 30558 + }, + { + "epoch": 2.519914240949946, + "grad_norm": 2.147599935531616, + "learning_rate": 1e-06, + "loss": 0.7755, + "mean_token_accuracy": 0.745990514755249, + "num_tokens": 808716484.0, + "step": 30559 + }, + { + "epoch": 2.519996701574998, + "grad_norm": 2.1647791862487793, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7794226408004761, + "num_tokens": 808744005.0, + "step": 30560 + }, + { + "epoch": 2.5200791622000493, + "grad_norm": 2.2290403842926025, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7464488744735718, + "num_tokens": 808773105.0, + "step": 30561 + }, + { + "epoch": 2.520161622825101, + "grad_norm": 2.190540075302124, + "learning_rate": 1e-06, + "loss": 0.5572, + "mean_token_accuracy": 0.806857705116272, + "num_tokens": 808799291.0, + "step": 30562 + }, + { + "epoch": 2.5202440834501525, + "grad_norm": 2.189697742462158, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7734739780426025, + "num_tokens": 808828961.0, + "step": 30563 + }, + { + "epoch": 2.5203265440752043, + "grad_norm": 2.1701011657714844, + "learning_rate": 1e-06, + "loss": 0.5857, + "mean_token_accuracy": 0.8027797341346741, + "num_tokens": 808859373.0, + "step": 30564 + }, + { + "epoch": 2.5204090047002556, + "grad_norm": 1.9666872024536133, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.7850704193115234, + "num_tokens": 808892126.0, + "step": 30565 + }, + { + "epoch": 2.520491465325307, + "grad_norm": 1.8911644220352173, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.7834847569465637, + "num_tokens": 808928250.0, + "step": 30566 + }, + { + "epoch": 2.520573925950359, + "grad_norm": 2.118217706680298, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7791664600372314, + "num_tokens": 808959246.0, + "step": 30567 + }, + { + "epoch": 2.52065638657541, + "grad_norm": 2.1507415771484375, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7526304721832275, + "num_tokens": 808989563.0, + "step": 30568 + }, + { + "epoch": 2.520738847200462, + "grad_norm": 1.8984744548797607, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.780173659324646, + "num_tokens": 809024995.0, + "step": 30569 + }, + { + "epoch": 2.5208213078255133, + "grad_norm": 2.1425390243530273, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7684324979782104, + "num_tokens": 809057186.0, + "step": 30570 + }, + { + "epoch": 2.520903768450565, + "grad_norm": 2.1268513202667236, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7558076977729797, + "num_tokens": 809086547.0, + "step": 30571 + }, + { + "epoch": 2.5209862290756164, + "grad_norm": 2.0269527435302734, + "learning_rate": 1e-06, + "loss": 0.5913, + "mean_token_accuracy": 0.8024365901947021, + "num_tokens": 809117772.0, + "step": 30572 + }, + { + "epoch": 2.5210686897006678, + "grad_norm": 2.074583053588867, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7350875735282898, + "num_tokens": 809149573.0, + "step": 30573 + }, + { + "epoch": 2.5211511503257196, + "grad_norm": 1.8152145147323608, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7913008332252502, + "num_tokens": 809187773.0, + "step": 30574 + }, + { + "epoch": 2.521233610950771, + "grad_norm": 2.088020086288452, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7414942979812622, + "num_tokens": 809218315.0, + "step": 30575 + }, + { + "epoch": 2.5213160715758223, + "grad_norm": 1.7767360210418701, + "learning_rate": 1e-06, + "loss": 0.6026, + "mean_token_accuracy": 0.8016896843910217, + "num_tokens": 809255984.0, + "step": 30576 + }, + { + "epoch": 2.521398532200874, + "grad_norm": 2.091627836227417, + "learning_rate": 1e-06, + "loss": 0.6298, + "mean_token_accuracy": 0.7844228744506836, + "num_tokens": 809285680.0, + "step": 30577 + }, + { + "epoch": 2.521480992825926, + "grad_norm": 2.0123701095581055, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.7811187505722046, + "num_tokens": 809316155.0, + "step": 30578 + }, + { + "epoch": 2.521563453450977, + "grad_norm": 2.2801389694213867, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7704998254776001, + "num_tokens": 809341429.0, + "step": 30579 + }, + { + "epoch": 2.5216459140760286, + "grad_norm": 2.1241142749786377, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.78653883934021, + "num_tokens": 809370600.0, + "step": 30580 + }, + { + "epoch": 2.5217283747010804, + "grad_norm": 2.2024221420288086, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.7506110072135925, + "num_tokens": 809402180.0, + "step": 30581 + }, + { + "epoch": 2.5218108353261317, + "grad_norm": 2.2111740112304688, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7506194114685059, + "num_tokens": 809429085.0, + "step": 30582 + }, + { + "epoch": 2.521893295951183, + "grad_norm": 2.163735866546631, + "learning_rate": 1e-06, + "loss": 0.605, + "mean_token_accuracy": 0.8027331829071045, + "num_tokens": 809458242.0, + "step": 30583 + }, + { + "epoch": 2.521975756576235, + "grad_norm": 2.092714548110962, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.776375412940979, + "num_tokens": 809490214.0, + "step": 30584 + }, + { + "epoch": 2.5220582172012866, + "grad_norm": 2.1622958183288574, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7753006815910339, + "num_tokens": 809515604.0, + "step": 30585 + }, + { + "epoch": 2.522140677826338, + "grad_norm": 2.03550386428833, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7671643495559692, + "num_tokens": 809547327.0, + "step": 30586 + }, + { + "epoch": 2.5222231384513893, + "grad_norm": 2.4614131450653076, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.7768567800521851, + "num_tokens": 809569943.0, + "step": 30587 + }, + { + "epoch": 2.522305599076441, + "grad_norm": 2.088935136795044, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7623379826545715, + "num_tokens": 809599228.0, + "step": 30588 + }, + { + "epoch": 2.5223880597014925, + "grad_norm": 2.229562520980835, + "learning_rate": 1e-06, + "loss": 0.477, + "mean_token_accuracy": 0.8058022856712341, + "num_tokens": 809625088.0, + "step": 30589 + }, + { + "epoch": 2.522470520326544, + "grad_norm": 2.0832443237304688, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.7807865738868713, + "num_tokens": 809654673.0, + "step": 30590 + }, + { + "epoch": 2.5225529809515956, + "grad_norm": 1.9152843952178955, + "learning_rate": 1e-06, + "loss": 0.5536, + "mean_token_accuracy": 0.8159530758857727, + "num_tokens": 809689150.0, + "step": 30591 + }, + { + "epoch": 2.522635441576647, + "grad_norm": 2.216604232788086, + "learning_rate": 1e-06, + "loss": 0.5274, + "mean_token_accuracy": 0.8203462362289429, + "num_tokens": 809717057.0, + "step": 30592 + }, + { + "epoch": 2.522717902201699, + "grad_norm": 2.208465576171875, + "learning_rate": 1e-06, + "loss": 0.6145, + "mean_token_accuracy": 0.7870330810546875, + "num_tokens": 809745710.0, + "step": 30593 + }, + { + "epoch": 2.52280036282675, + "grad_norm": 1.9435306787490845, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7516477108001709, + "num_tokens": 809778306.0, + "step": 30594 + }, + { + "epoch": 2.522882823451802, + "grad_norm": 1.9669262170791626, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.785143256187439, + "num_tokens": 809808854.0, + "step": 30595 + }, + { + "epoch": 2.5229652840768533, + "grad_norm": 1.9354690313339233, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7645049095153809, + "num_tokens": 809841993.0, + "step": 30596 + }, + { + "epoch": 2.5230477447019046, + "grad_norm": 2.0425593852996826, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7878595590591431, + "num_tokens": 809872114.0, + "step": 30597 + }, + { + "epoch": 2.5231302053269564, + "grad_norm": 2.110943555831909, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7649730443954468, + "num_tokens": 809903161.0, + "step": 30598 + }, + { + "epoch": 2.5232126659520078, + "grad_norm": 2.4259750843048096, + "learning_rate": 1e-06, + "loss": 0.5328, + "mean_token_accuracy": 0.822246789932251, + "num_tokens": 809925258.0, + "step": 30599 + }, + { + "epoch": 2.5232951265770596, + "grad_norm": 1.9889309406280518, + "learning_rate": 1e-06, + "loss": 0.6184, + "mean_token_accuracy": 0.7994831204414368, + "num_tokens": 809957714.0, + "step": 30600 + }, + { + "epoch": 2.523377587202111, + "grad_norm": 2.1272709369659424, + "learning_rate": 1e-06, + "loss": 0.512, + "mean_token_accuracy": 0.8177621364593506, + "num_tokens": 809986055.0, + "step": 30601 + }, + { + "epoch": 2.5234600478271627, + "grad_norm": 2.222126007080078, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.7971373200416565, + "num_tokens": 810013000.0, + "step": 30602 + }, + { + "epoch": 2.523542508452214, + "grad_norm": 2.3586854934692383, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7613874673843384, + "num_tokens": 810037072.0, + "step": 30603 + }, + { + "epoch": 2.5236249690772654, + "grad_norm": 2.0141799449920654, + "learning_rate": 1e-06, + "loss": 0.6593, + "mean_token_accuracy": 0.7827932238578796, + "num_tokens": 810067877.0, + "step": 30604 + }, + { + "epoch": 2.523707429702317, + "grad_norm": 1.897737741470337, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.7947993278503418, + "num_tokens": 810103661.0, + "step": 30605 + }, + { + "epoch": 2.5237898903273686, + "grad_norm": 2.1524360179901123, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.775825023651123, + "num_tokens": 810130754.0, + "step": 30606 + }, + { + "epoch": 2.5238723509524204, + "grad_norm": 2.1589462757110596, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.791957437992096, + "num_tokens": 810157736.0, + "step": 30607 + }, + { + "epoch": 2.5239548115774717, + "grad_norm": 2.039907693862915, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7358751893043518, + "num_tokens": 810189804.0, + "step": 30608 + }, + { + "epoch": 2.5240372722025235, + "grad_norm": 2.1297049522399902, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7807926535606384, + "num_tokens": 810218925.0, + "step": 30609 + }, + { + "epoch": 2.524119732827575, + "grad_norm": 2.0242714881896973, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.7530341148376465, + "num_tokens": 810255209.0, + "step": 30610 + }, + { + "epoch": 2.524202193452626, + "grad_norm": 2.2024621963500977, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7592889070510864, + "num_tokens": 810286439.0, + "step": 30611 + }, + { + "epoch": 2.524284654077678, + "grad_norm": 2.0719010829925537, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7527593374252319, + "num_tokens": 810317841.0, + "step": 30612 + }, + { + "epoch": 2.5243671147027293, + "grad_norm": 2.0414133071899414, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.774304211139679, + "num_tokens": 810351139.0, + "step": 30613 + }, + { + "epoch": 2.524449575327781, + "grad_norm": 2.1518969535827637, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7422860860824585, + "num_tokens": 810380736.0, + "step": 30614 + }, + { + "epoch": 2.5245320359528325, + "grad_norm": 1.980683445930481, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7773295640945435, + "num_tokens": 810412391.0, + "step": 30615 + }, + { + "epoch": 2.5246144965778843, + "grad_norm": 2.0461111068725586, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7756977081298828, + "num_tokens": 810440849.0, + "step": 30616 + }, + { + "epoch": 2.5246969572029356, + "grad_norm": 2.0418927669525146, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7556255459785461, + "num_tokens": 810472871.0, + "step": 30617 + }, + { + "epoch": 2.524779417827987, + "grad_norm": 2.093435049057007, + "learning_rate": 1e-06, + "loss": 0.6193, + "mean_token_accuracy": 0.8017698526382446, + "num_tokens": 810501709.0, + "step": 30618 + }, + { + "epoch": 2.5248618784530388, + "grad_norm": 1.9909685850143433, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.7963319420814514, + "num_tokens": 810533952.0, + "step": 30619 + }, + { + "epoch": 2.52494433907809, + "grad_norm": 2.0379276275634766, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7688754200935364, + "num_tokens": 810564277.0, + "step": 30620 + }, + { + "epoch": 2.5250267997031415, + "grad_norm": 2.2531049251556396, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7852126359939575, + "num_tokens": 810591073.0, + "step": 30621 + }, + { + "epoch": 2.5251092603281933, + "grad_norm": 2.1062276363372803, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7571256160736084, + "num_tokens": 810620826.0, + "step": 30622 + }, + { + "epoch": 2.525191720953245, + "grad_norm": 2.2922518253326416, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7666833400726318, + "num_tokens": 810645880.0, + "step": 30623 + }, + { + "epoch": 2.5252741815782964, + "grad_norm": 2.4950358867645264, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7246811389923096, + "num_tokens": 810670537.0, + "step": 30624 + }, + { + "epoch": 2.5253566422033478, + "grad_norm": 2.015713930130005, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.786589503288269, + "num_tokens": 810700489.0, + "step": 30625 + }, + { + "epoch": 2.5254391028283996, + "grad_norm": 2.228626251220703, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7753046154975891, + "num_tokens": 810727821.0, + "step": 30626 + }, + { + "epoch": 2.525521563453451, + "grad_norm": 2.1894314289093018, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7486439347267151, + "num_tokens": 810755957.0, + "step": 30627 + }, + { + "epoch": 2.5256040240785023, + "grad_norm": 2.1909902095794678, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7486279010772705, + "num_tokens": 810786188.0, + "step": 30628 + }, + { + "epoch": 2.525686484703554, + "grad_norm": 2.1436002254486084, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7471972703933716, + "num_tokens": 810815521.0, + "step": 30629 + }, + { + "epoch": 2.525768945328606, + "grad_norm": 2.3835864067077637, + "learning_rate": 1e-06, + "loss": 0.7561, + "mean_token_accuracy": 0.7529631853103638, + "num_tokens": 810841124.0, + "step": 30630 + }, + { + "epoch": 2.525851405953657, + "grad_norm": 2.4041590690612793, + "learning_rate": 1e-06, + "loss": 0.6874, + "mean_token_accuracy": 0.7778927087783813, + "num_tokens": 810867765.0, + "step": 30631 + }, + { + "epoch": 2.5259338665787086, + "grad_norm": 1.8478233814239502, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7608399391174316, + "num_tokens": 810900764.0, + "step": 30632 + }, + { + "epoch": 2.5260163272037603, + "grad_norm": 2.0167934894561768, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7509968280792236, + "num_tokens": 810933421.0, + "step": 30633 + }, + { + "epoch": 2.5260987878288117, + "grad_norm": 1.9373750686645508, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.7771748900413513, + "num_tokens": 810966052.0, + "step": 30634 + }, + { + "epoch": 2.526181248453863, + "grad_norm": 2.2978897094726562, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7549940347671509, + "num_tokens": 810993257.0, + "step": 30635 + }, + { + "epoch": 2.526263709078915, + "grad_norm": 2.302156686782837, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7414859533309937, + "num_tokens": 811019672.0, + "step": 30636 + }, + { + "epoch": 2.526346169703966, + "grad_norm": 2.2384836673736572, + "learning_rate": 1e-06, + "loss": 0.5378, + "mean_token_accuracy": 0.8145217299461365, + "num_tokens": 811045936.0, + "step": 30637 + }, + { + "epoch": 2.526428630329018, + "grad_norm": 1.8695757389068604, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7636521458625793, + "num_tokens": 811081791.0, + "step": 30638 + }, + { + "epoch": 2.5265110909540693, + "grad_norm": 2.2906453609466553, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7664920091629028, + "num_tokens": 811108980.0, + "step": 30639 + }, + { + "epoch": 2.526593551579121, + "grad_norm": 2.0123627185821533, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7658672332763672, + "num_tokens": 811139560.0, + "step": 30640 + }, + { + "epoch": 2.5266760122041725, + "grad_norm": 2.300147771835327, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7373232245445251, + "num_tokens": 811169676.0, + "step": 30641 + }, + { + "epoch": 2.526758472829224, + "grad_norm": 2.182304859161377, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7107774019241333, + "num_tokens": 811200370.0, + "step": 30642 + }, + { + "epoch": 2.5268409334542756, + "grad_norm": 2.150023937225342, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.7758910655975342, + "num_tokens": 811228742.0, + "step": 30643 + }, + { + "epoch": 2.526923394079327, + "grad_norm": 2.092566728591919, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7671495676040649, + "num_tokens": 811264172.0, + "step": 30644 + }, + { + "epoch": 2.5270058547043788, + "grad_norm": 1.910933017730713, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7748842239379883, + "num_tokens": 811298651.0, + "step": 30645 + }, + { + "epoch": 2.52708831532943, + "grad_norm": 2.2909047603607178, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.7651556730270386, + "num_tokens": 811326084.0, + "step": 30646 + }, + { + "epoch": 2.527170775954482, + "grad_norm": 1.9893684387207031, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7757368683815002, + "num_tokens": 811358789.0, + "step": 30647 + }, + { + "epoch": 2.5272532365795333, + "grad_norm": 2.026581048965454, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7568379640579224, + "num_tokens": 811388979.0, + "step": 30648 + }, + { + "epoch": 2.5273356972045846, + "grad_norm": 2.1124706268310547, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7457554936408997, + "num_tokens": 811417604.0, + "step": 30649 + }, + { + "epoch": 2.5274181578296364, + "grad_norm": 1.8592149019241333, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7681249976158142, + "num_tokens": 811452132.0, + "step": 30650 + }, + { + "epoch": 2.5275006184546878, + "grad_norm": 2.1797232627868652, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.771338164806366, + "num_tokens": 811478479.0, + "step": 30651 + }, + { + "epoch": 2.5275830790797396, + "grad_norm": 2.3071248531341553, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7561830878257751, + "num_tokens": 811506365.0, + "step": 30652 + }, + { + "epoch": 2.527665539704791, + "grad_norm": 2.231600522994995, + "learning_rate": 1e-06, + "loss": 0.5956, + "mean_token_accuracy": 0.801559329032898, + "num_tokens": 811537179.0, + "step": 30653 + }, + { + "epoch": 2.5277480003298427, + "grad_norm": 1.9979130029678345, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7522019743919373, + "num_tokens": 811571740.0, + "step": 30654 + }, + { + "epoch": 2.527830460954894, + "grad_norm": 2.0784401893615723, + "learning_rate": 1e-06, + "loss": 0.7321, + "mean_token_accuracy": 0.7726615071296692, + "num_tokens": 811603442.0, + "step": 30655 + }, + { + "epoch": 2.5279129215799454, + "grad_norm": 2.0528006553649902, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.760670006275177, + "num_tokens": 811636076.0, + "step": 30656 + }, + { + "epoch": 2.527995382204997, + "grad_norm": 2.4438583850860596, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7534379363059998, + "num_tokens": 811668783.0, + "step": 30657 + }, + { + "epoch": 2.5280778428300485, + "grad_norm": 2.1818912029266357, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7678631544113159, + "num_tokens": 811697485.0, + "step": 30658 + }, + { + "epoch": 2.5281603034551003, + "grad_norm": 2.1257572174072266, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7719277739524841, + "num_tokens": 811727909.0, + "step": 30659 + }, + { + "epoch": 2.5282427640801517, + "grad_norm": 2.0841474533081055, + "learning_rate": 1e-06, + "loss": 0.5874, + "mean_token_accuracy": 0.8110543489456177, + "num_tokens": 811756899.0, + "step": 30660 + }, + { + "epoch": 2.5283252247052035, + "grad_norm": 2.441279888153076, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7497036457061768, + "num_tokens": 811781546.0, + "step": 30661 + }, + { + "epoch": 2.528407685330255, + "grad_norm": 2.1341564655303955, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.7843534350395203, + "num_tokens": 811809216.0, + "step": 30662 + }, + { + "epoch": 2.528490145955306, + "grad_norm": 2.054088592529297, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.7824935913085938, + "num_tokens": 811837986.0, + "step": 30663 + }, + { + "epoch": 2.528572606580358, + "grad_norm": 1.9070621728897095, + "learning_rate": 1e-06, + "loss": 0.5298, + "mean_token_accuracy": 0.8207554221153259, + "num_tokens": 811869152.0, + "step": 30664 + }, + { + "epoch": 2.5286550672054093, + "grad_norm": 2.009702682495117, + "learning_rate": 1e-06, + "loss": 0.6601, + "mean_token_accuracy": 0.7809829711914062, + "num_tokens": 811900717.0, + "step": 30665 + }, + { + "epoch": 2.5287375278304607, + "grad_norm": 2.1314220428466797, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7425939440727234, + "num_tokens": 811928920.0, + "step": 30666 + }, + { + "epoch": 2.5288199884555125, + "grad_norm": 2.3219354152679443, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.7976687550544739, + "num_tokens": 811953527.0, + "step": 30667 + }, + { + "epoch": 2.5289024490805643, + "grad_norm": 2.4314868450164795, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.740105152130127, + "num_tokens": 811977787.0, + "step": 30668 + }, + { + "epoch": 2.5289849097056156, + "grad_norm": 2.223695993423462, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.7865390777587891, + "num_tokens": 812004117.0, + "step": 30669 + }, + { + "epoch": 2.529067370330667, + "grad_norm": 1.865495204925537, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7724340558052063, + "num_tokens": 812040385.0, + "step": 30670 + }, + { + "epoch": 2.5291498309557188, + "grad_norm": 2.294067621231079, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7854941487312317, + "num_tokens": 812067476.0, + "step": 30671 + }, + { + "epoch": 2.52923229158077, + "grad_norm": 2.2567238807678223, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7853186130523682, + "num_tokens": 812092452.0, + "step": 30672 + }, + { + "epoch": 2.5293147522058215, + "grad_norm": 2.062570810317993, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7324563264846802, + "num_tokens": 812125745.0, + "step": 30673 + }, + { + "epoch": 2.5293972128308733, + "grad_norm": 1.9265998601913452, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.7542304992675781, + "num_tokens": 812159670.0, + "step": 30674 + }, + { + "epoch": 2.529479673455925, + "grad_norm": 2.0885562896728516, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7863849997520447, + "num_tokens": 812189111.0, + "step": 30675 + }, + { + "epoch": 2.5295621340809764, + "grad_norm": 2.1906678676605225, + "learning_rate": 1e-06, + "loss": 0.7543, + "mean_token_accuracy": 0.751200795173645, + "num_tokens": 812217198.0, + "step": 30676 + }, + { + "epoch": 2.5296445947060278, + "grad_norm": 2.50097393989563, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.76106858253479, + "num_tokens": 812240232.0, + "step": 30677 + }, + { + "epoch": 2.5297270553310796, + "grad_norm": 1.9362603425979614, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.761394202709198, + "num_tokens": 812273299.0, + "step": 30678 + }, + { + "epoch": 2.529809515956131, + "grad_norm": 2.191450357437134, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7403561472892761, + "num_tokens": 812301679.0, + "step": 30679 + }, + { + "epoch": 2.5298919765811823, + "grad_norm": 2.4609944820404053, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7303022146224976, + "num_tokens": 812328151.0, + "step": 30680 + }, + { + "epoch": 2.529974437206234, + "grad_norm": 1.997567057609558, + "learning_rate": 1e-06, + "loss": 0.5425, + "mean_token_accuracy": 0.8071366548538208, + "num_tokens": 812361458.0, + "step": 30681 + }, + { + "epoch": 2.5300568978312854, + "grad_norm": 2.1986136436462402, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.77269446849823, + "num_tokens": 812389846.0, + "step": 30682 + }, + { + "epoch": 2.530139358456337, + "grad_norm": 2.157136917114258, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.767792820930481, + "num_tokens": 812419294.0, + "step": 30683 + }, + { + "epoch": 2.5302218190813885, + "grad_norm": 1.902358055114746, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7639510035514832, + "num_tokens": 812455497.0, + "step": 30684 + }, + { + "epoch": 2.5303042797064403, + "grad_norm": 2.0830466747283936, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7753192186355591, + "num_tokens": 812485546.0, + "step": 30685 + }, + { + "epoch": 2.5303867403314917, + "grad_norm": 2.3309080600738525, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7616958618164062, + "num_tokens": 812512219.0, + "step": 30686 + }, + { + "epoch": 2.530469200956543, + "grad_norm": 1.9746248722076416, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.7940576076507568, + "num_tokens": 812542612.0, + "step": 30687 + }, + { + "epoch": 2.530551661581595, + "grad_norm": 1.979765772819519, + "learning_rate": 1e-06, + "loss": 0.5058, + "mean_token_accuracy": 0.8244819641113281, + "num_tokens": 812573161.0, + "step": 30688 + }, + { + "epoch": 2.530634122206646, + "grad_norm": 1.8898870944976807, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.784122109413147, + "num_tokens": 812606358.0, + "step": 30689 + }, + { + "epoch": 2.530716582831698, + "grad_norm": 2.058140277862549, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7775593400001526, + "num_tokens": 812638312.0, + "step": 30690 + }, + { + "epoch": 2.5307990434567493, + "grad_norm": 2.2059428691864014, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7581502199172974, + "num_tokens": 812670100.0, + "step": 30691 + }, + { + "epoch": 2.530881504081801, + "grad_norm": 2.136767625808716, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7097015380859375, + "num_tokens": 812703111.0, + "step": 30692 + }, + { + "epoch": 2.5309639647068525, + "grad_norm": 2.748460054397583, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.7437834739685059, + "num_tokens": 812736273.0, + "step": 30693 + }, + { + "epoch": 2.531046425331904, + "grad_norm": 2.094555377960205, + "learning_rate": 1e-06, + "loss": 0.5277, + "mean_token_accuracy": 0.8203847408294678, + "num_tokens": 812766340.0, + "step": 30694 + }, + { + "epoch": 2.5311288859569556, + "grad_norm": 2.2771496772766113, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7556350231170654, + "num_tokens": 812792876.0, + "step": 30695 + }, + { + "epoch": 2.531211346582007, + "grad_norm": 2.0997490882873535, + "learning_rate": 1e-06, + "loss": 0.5257, + "mean_token_accuracy": 0.82501620054245, + "num_tokens": 812821613.0, + "step": 30696 + }, + { + "epoch": 2.5312938072070588, + "grad_norm": 2.2244277000427246, + "learning_rate": 1e-06, + "loss": 0.6001, + "mean_token_accuracy": 0.8031799793243408, + "num_tokens": 812846805.0, + "step": 30697 + }, + { + "epoch": 2.53137626783211, + "grad_norm": 2.2120308876037598, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.745276927947998, + "num_tokens": 812875259.0, + "step": 30698 + }, + { + "epoch": 2.531458728457162, + "grad_norm": 1.819811224937439, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7716332674026489, + "num_tokens": 812912519.0, + "step": 30699 + }, + { + "epoch": 2.5315411890822133, + "grad_norm": 2.11761474609375, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.7842419147491455, + "num_tokens": 812943448.0, + "step": 30700 + }, + { + "epoch": 2.5316236497072646, + "grad_norm": 2.3694941997528076, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7373983860015869, + "num_tokens": 812969916.0, + "step": 30701 + }, + { + "epoch": 2.5317061103323164, + "grad_norm": 1.9271680116653442, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.764700174331665, + "num_tokens": 813004239.0, + "step": 30702 + }, + { + "epoch": 2.5317885709573678, + "grad_norm": 1.9341142177581787, + "learning_rate": 1e-06, + "loss": 0.6115, + "mean_token_accuracy": 0.7919414043426514, + "num_tokens": 813036397.0, + "step": 30703 + }, + { + "epoch": 2.5318710315824196, + "grad_norm": 2.0722243785858154, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.7893460988998413, + "num_tokens": 813067622.0, + "step": 30704 + }, + { + "epoch": 2.531953492207471, + "grad_norm": 2.128983497619629, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.7869568467140198, + "num_tokens": 813097402.0, + "step": 30705 + }, + { + "epoch": 2.5320359528325227, + "grad_norm": 1.9575080871582031, + "learning_rate": 1e-06, + "loss": 0.5958, + "mean_token_accuracy": 0.8055435419082642, + "num_tokens": 813131881.0, + "step": 30706 + }, + { + "epoch": 2.532118413457574, + "grad_norm": 2.2621102333068848, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7537226676940918, + "num_tokens": 813159327.0, + "step": 30707 + }, + { + "epoch": 2.5322008740826254, + "grad_norm": 2.1066393852233887, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7729000449180603, + "num_tokens": 813190640.0, + "step": 30708 + }, + { + "epoch": 2.532283334707677, + "grad_norm": 1.93812894821167, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7790737152099609, + "num_tokens": 813223550.0, + "step": 30709 + }, + { + "epoch": 2.5323657953327285, + "grad_norm": 2.1380655765533447, + "learning_rate": 1e-06, + "loss": 0.6023, + "mean_token_accuracy": 0.8032898902893066, + "num_tokens": 813250041.0, + "step": 30710 + }, + { + "epoch": 2.53244825595778, + "grad_norm": 1.806220293045044, + "learning_rate": 1e-06, + "loss": 0.5038, + "mean_token_accuracy": 0.8243491649627686, + "num_tokens": 813281873.0, + "step": 30711 + }, + { + "epoch": 2.5325307165828317, + "grad_norm": 2.150938034057617, + "learning_rate": 1e-06, + "loss": 0.7895, + "mean_token_accuracy": 0.7444911003112793, + "num_tokens": 813311826.0, + "step": 30712 + }, + { + "epoch": 2.5326131772078835, + "grad_norm": 1.935120940208435, + "learning_rate": 1e-06, + "loss": 0.5563, + "mean_token_accuracy": 0.8089537620544434, + "num_tokens": 813346693.0, + "step": 30713 + }, + { + "epoch": 2.532695637832935, + "grad_norm": 2.140611410140991, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7547483444213867, + "num_tokens": 813374585.0, + "step": 30714 + }, + { + "epoch": 2.532778098457986, + "grad_norm": 1.8505704402923584, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7777498960494995, + "num_tokens": 813412717.0, + "step": 30715 + }, + { + "epoch": 2.532860559083038, + "grad_norm": 2.557464838027954, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.749444842338562, + "num_tokens": 813433786.0, + "step": 30716 + }, + { + "epoch": 2.5329430197080893, + "grad_norm": 2.2288389205932617, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7558422088623047, + "num_tokens": 813460276.0, + "step": 30717 + }, + { + "epoch": 2.5330254803331407, + "grad_norm": 2.2903337478637695, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7607874870300293, + "num_tokens": 813484620.0, + "step": 30718 + }, + { + "epoch": 2.5331079409581925, + "grad_norm": 2.2499289512634277, + "learning_rate": 1e-06, + "loss": 0.5607, + "mean_token_accuracy": 0.8125196695327759, + "num_tokens": 813508792.0, + "step": 30719 + }, + { + "epoch": 2.5331904015832443, + "grad_norm": 2.1977484226226807, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.763378381729126, + "num_tokens": 813535893.0, + "step": 30720 + }, + { + "epoch": 2.5332728622082956, + "grad_norm": 2.0756046772003174, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.743460476398468, + "num_tokens": 813567356.0, + "step": 30721 + }, + { + "epoch": 2.533355322833347, + "grad_norm": 2.0193917751312256, + "learning_rate": 1e-06, + "loss": 0.7508, + "mean_token_accuracy": 0.760451078414917, + "num_tokens": 813597650.0, + "step": 30722 + }, + { + "epoch": 2.5334377834583988, + "grad_norm": 1.9090561866760254, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7367783188819885, + "num_tokens": 813632654.0, + "step": 30723 + }, + { + "epoch": 2.53352024408345, + "grad_norm": 2.4369125366210938, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.785212516784668, + "num_tokens": 813654033.0, + "step": 30724 + }, + { + "epoch": 2.5336027047085015, + "grad_norm": 1.9570263624191284, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7725178599357605, + "num_tokens": 813687984.0, + "step": 30725 + }, + { + "epoch": 2.5336851653335533, + "grad_norm": 2.0399436950683594, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.8009507060050964, + "num_tokens": 813721142.0, + "step": 30726 + }, + { + "epoch": 2.5337676259586046, + "grad_norm": 2.275938034057617, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7571346163749695, + "num_tokens": 813749125.0, + "step": 30727 + }, + { + "epoch": 2.5338500865836564, + "grad_norm": 2.2992441654205322, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.7933399677276611, + "num_tokens": 813774658.0, + "step": 30728 + }, + { + "epoch": 2.5339325472087078, + "grad_norm": 2.0509984493255615, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7616654634475708, + "num_tokens": 813803818.0, + "step": 30729 + }, + { + "epoch": 2.5340150078337595, + "grad_norm": 2.0736770629882812, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7854079008102417, + "num_tokens": 813835754.0, + "step": 30730 + }, + { + "epoch": 2.534097468458811, + "grad_norm": 2.1712496280670166, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7527166604995728, + "num_tokens": 813865957.0, + "step": 30731 + }, + { + "epoch": 2.5341799290838622, + "grad_norm": 2.112443685531616, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7264324426651001, + "num_tokens": 813897710.0, + "step": 30732 + }, + { + "epoch": 2.534262389708914, + "grad_norm": 2.1566195487976074, + "learning_rate": 1e-06, + "loss": 0.6062, + "mean_token_accuracy": 0.7967075109481812, + "num_tokens": 813926523.0, + "step": 30733 + }, + { + "epoch": 2.5343448503339654, + "grad_norm": 2.0936813354492188, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.7667491436004639, + "num_tokens": 813953455.0, + "step": 30734 + }, + { + "epoch": 2.534427310959017, + "grad_norm": 2.488011121749878, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7222503423690796, + "num_tokens": 813979403.0, + "step": 30735 + }, + { + "epoch": 2.5345097715840685, + "grad_norm": 1.9575188159942627, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.783336341381073, + "num_tokens": 814010375.0, + "step": 30736 + }, + { + "epoch": 2.5345922322091203, + "grad_norm": 2.1791670322418213, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7661340236663818, + "num_tokens": 814037748.0, + "step": 30737 + }, + { + "epoch": 2.5346746928341717, + "grad_norm": 1.9673742055892944, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.7891499996185303, + "num_tokens": 814070495.0, + "step": 30738 + }, + { + "epoch": 2.534757153459223, + "grad_norm": 2.129520893096924, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7613686323165894, + "num_tokens": 814101191.0, + "step": 30739 + }, + { + "epoch": 2.534839614084275, + "grad_norm": 2.443157196044922, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7818851470947266, + "num_tokens": 814123360.0, + "step": 30740 + }, + { + "epoch": 2.534922074709326, + "grad_norm": 1.8063281774520874, + "learning_rate": 1e-06, + "loss": 0.7152, + "mean_token_accuracy": 0.7719325423240662, + "num_tokens": 814161296.0, + "step": 30741 + }, + { + "epoch": 2.535004535334378, + "grad_norm": 2.196112632751465, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.771346926689148, + "num_tokens": 814191821.0, + "step": 30742 + }, + { + "epoch": 2.5350869959594293, + "grad_norm": 2.265803098678589, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.77589350938797, + "num_tokens": 814220051.0, + "step": 30743 + }, + { + "epoch": 2.535169456584481, + "grad_norm": 2.27740216255188, + "learning_rate": 1e-06, + "loss": 0.6355, + "mean_token_accuracy": 0.7957195043563843, + "num_tokens": 814244982.0, + "step": 30744 + }, + { + "epoch": 2.5352519172095325, + "grad_norm": 2.0300285816192627, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7720718383789062, + "num_tokens": 814278562.0, + "step": 30745 + }, + { + "epoch": 2.535334377834584, + "grad_norm": 2.148625612258911, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7504172921180725, + "num_tokens": 814308244.0, + "step": 30746 + }, + { + "epoch": 2.5354168384596356, + "grad_norm": 2.045151472091675, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.7949638366699219, + "num_tokens": 814339907.0, + "step": 30747 + }, + { + "epoch": 2.535499299084687, + "grad_norm": 2.002718448638916, + "learning_rate": 1e-06, + "loss": 0.5816, + "mean_token_accuracy": 0.7983954548835754, + "num_tokens": 814369216.0, + "step": 30748 + }, + { + "epoch": 2.5355817597097388, + "grad_norm": 2.262091875076294, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7373358011245728, + "num_tokens": 814396206.0, + "step": 30749 + }, + { + "epoch": 2.53566422033479, + "grad_norm": 1.950183629989624, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7873929142951965, + "num_tokens": 814429946.0, + "step": 30750 + }, + { + "epoch": 2.535746680959842, + "grad_norm": 1.7946982383728027, + "learning_rate": 1e-06, + "loss": 0.7108, + "mean_token_accuracy": 0.771769106388092, + "num_tokens": 814465772.0, + "step": 30751 + }, + { + "epoch": 2.5358291415848933, + "grad_norm": 2.2693424224853516, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7668448686599731, + "num_tokens": 814493282.0, + "step": 30752 + }, + { + "epoch": 2.5359116022099446, + "grad_norm": 1.96757972240448, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.757391095161438, + "num_tokens": 814531350.0, + "step": 30753 + }, + { + "epoch": 2.5359940628349964, + "grad_norm": 2.022818088531494, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7770551443099976, + "num_tokens": 814564371.0, + "step": 30754 + }, + { + "epoch": 2.5360765234600477, + "grad_norm": 2.3465726375579834, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7673273086547852, + "num_tokens": 814589264.0, + "step": 30755 + }, + { + "epoch": 2.536158984085099, + "grad_norm": 1.9971421957015991, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7404883503913879, + "num_tokens": 814620249.0, + "step": 30756 + }, + { + "epoch": 2.536241444710151, + "grad_norm": 2.2259459495544434, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7745068669319153, + "num_tokens": 814647512.0, + "step": 30757 + }, + { + "epoch": 2.5363239053352027, + "grad_norm": 2.2664270401000977, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7721203565597534, + "num_tokens": 814672013.0, + "step": 30758 + }, + { + "epoch": 2.536406365960254, + "grad_norm": 2.104959726333618, + "learning_rate": 1e-06, + "loss": 0.6476, + "mean_token_accuracy": 0.7834593653678894, + "num_tokens": 814702483.0, + "step": 30759 + }, + { + "epoch": 2.5364888265853054, + "grad_norm": 2.2497825622558594, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7507328987121582, + "num_tokens": 814730876.0, + "step": 30760 + }, + { + "epoch": 2.536571287210357, + "grad_norm": 2.216113805770874, + "learning_rate": 1e-06, + "loss": 0.5683, + "mean_token_accuracy": 0.8065588474273682, + "num_tokens": 814756537.0, + "step": 30761 + }, + { + "epoch": 2.5366537478354085, + "grad_norm": 2.08951473236084, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7358386516571045, + "num_tokens": 814786946.0, + "step": 30762 + }, + { + "epoch": 2.53673620846046, + "grad_norm": 1.992053508758545, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7636839151382446, + "num_tokens": 814821312.0, + "step": 30763 + }, + { + "epoch": 2.5368186690855117, + "grad_norm": 1.9374185800552368, + "learning_rate": 1e-06, + "loss": 0.6144, + "mean_token_accuracy": 0.8110672235488892, + "num_tokens": 814852627.0, + "step": 30764 + }, + { + "epoch": 2.5369011297105635, + "grad_norm": 1.9948091506958008, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7863293886184692, + "num_tokens": 814885269.0, + "step": 30765 + }, + { + "epoch": 2.536983590335615, + "grad_norm": 2.139857530593872, + "learning_rate": 1e-06, + "loss": 0.5855, + "mean_token_accuracy": 0.8057718276977539, + "num_tokens": 814914125.0, + "step": 30766 + }, + { + "epoch": 2.537066050960666, + "grad_norm": 1.9025487899780273, + "learning_rate": 1e-06, + "loss": 0.608, + "mean_token_accuracy": 0.7987781763076782, + "num_tokens": 814949863.0, + "step": 30767 + }, + { + "epoch": 2.537148511585718, + "grad_norm": 2.1279170513153076, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7745734453201294, + "num_tokens": 814979434.0, + "step": 30768 + }, + { + "epoch": 2.5372309722107693, + "grad_norm": 2.322861433029175, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7789890170097351, + "num_tokens": 815004884.0, + "step": 30769 + }, + { + "epoch": 2.5373134328358207, + "grad_norm": 2.02738356590271, + "learning_rate": 1e-06, + "loss": 0.5542, + "mean_token_accuracy": 0.8112572431564331, + "num_tokens": 815034536.0, + "step": 30770 + }, + { + "epoch": 2.5373958934608725, + "grad_norm": 1.9468690156936646, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7538301944732666, + "num_tokens": 815068994.0, + "step": 30771 + }, + { + "epoch": 2.537478354085924, + "grad_norm": 2.03204345703125, + "learning_rate": 1e-06, + "loss": 0.4521, + "mean_token_accuracy": 0.8437899947166443, + "num_tokens": 815095426.0, + "step": 30772 + }, + { + "epoch": 2.5375608147109756, + "grad_norm": 2.0228095054626465, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.773791491985321, + "num_tokens": 815128532.0, + "step": 30773 + }, + { + "epoch": 2.537643275336027, + "grad_norm": 2.5404908657073975, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7856602668762207, + "num_tokens": 815150839.0, + "step": 30774 + }, + { + "epoch": 2.5377257359610788, + "grad_norm": 1.8834110498428345, + "learning_rate": 1e-06, + "loss": 0.5716, + "mean_token_accuracy": 0.8091691136360168, + "num_tokens": 815183287.0, + "step": 30775 + }, + { + "epoch": 2.53780819658613, + "grad_norm": 1.9126302003860474, + "learning_rate": 1e-06, + "loss": 0.5739, + "mean_token_accuracy": 0.8065898418426514, + "num_tokens": 815215861.0, + "step": 30776 + }, + { + "epoch": 2.5378906572111815, + "grad_norm": 2.0258588790893555, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7601298093795776, + "num_tokens": 815247752.0, + "step": 30777 + }, + { + "epoch": 2.5379731178362332, + "grad_norm": 2.152578592300415, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.7530337572097778, + "num_tokens": 815279345.0, + "step": 30778 + }, + { + "epoch": 2.5380555784612846, + "grad_norm": 2.0960566997528076, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7572033405303955, + "num_tokens": 815311035.0, + "step": 30779 + }, + { + "epoch": 2.5381380390863364, + "grad_norm": 2.227971076965332, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7624498605728149, + "num_tokens": 815340838.0, + "step": 30780 + }, + { + "epoch": 2.5382204997113877, + "grad_norm": 2.1986196041107178, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7647199034690857, + "num_tokens": 815369170.0, + "step": 30781 + }, + { + "epoch": 2.5383029603364395, + "grad_norm": 2.1872503757476807, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7537407279014587, + "num_tokens": 815400207.0, + "step": 30782 + }, + { + "epoch": 2.538385420961491, + "grad_norm": 2.1354622840881348, + "learning_rate": 1e-06, + "loss": 0.6088, + "mean_token_accuracy": 0.801193118095398, + "num_tokens": 815430283.0, + "step": 30783 + }, + { + "epoch": 2.5384678815865422, + "grad_norm": 2.2332229614257812, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7445635199546814, + "num_tokens": 815460262.0, + "step": 30784 + }, + { + "epoch": 2.538550342211594, + "grad_norm": 2.0069193840026855, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7641299962997437, + "num_tokens": 815492113.0, + "step": 30785 + }, + { + "epoch": 2.5386328028366454, + "grad_norm": 2.032987117767334, + "learning_rate": 1e-06, + "loss": 0.5703, + "mean_token_accuracy": 0.8026646375656128, + "num_tokens": 815522431.0, + "step": 30786 + }, + { + "epoch": 2.538715263461697, + "grad_norm": 1.9645730257034302, + "learning_rate": 1e-06, + "loss": 0.5652, + "mean_token_accuracy": 0.8072190284729004, + "num_tokens": 815551554.0, + "step": 30787 + }, + { + "epoch": 2.5387977240867485, + "grad_norm": 2.3657917976379395, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7637088894844055, + "num_tokens": 815576453.0, + "step": 30788 + }, + { + "epoch": 2.5388801847118003, + "grad_norm": 2.4135825634002686, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7761144638061523, + "num_tokens": 815600023.0, + "step": 30789 + }, + { + "epoch": 2.5389626453368517, + "grad_norm": 2.351794719696045, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7386927604675293, + "num_tokens": 815626965.0, + "step": 30790 + }, + { + "epoch": 2.539045105961903, + "grad_norm": 2.159139633178711, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7572991251945496, + "num_tokens": 815658014.0, + "step": 30791 + }, + { + "epoch": 2.539127566586955, + "grad_norm": 2.163950204849243, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.7954792976379395, + "num_tokens": 815687582.0, + "step": 30792 + }, + { + "epoch": 2.539210027212006, + "grad_norm": 2.139402389526367, + "learning_rate": 1e-06, + "loss": 0.7563, + "mean_token_accuracy": 0.760702908039093, + "num_tokens": 815717310.0, + "step": 30793 + }, + { + "epoch": 2.539292487837058, + "grad_norm": 2.0082895755767822, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7633197903633118, + "num_tokens": 815748928.0, + "step": 30794 + }, + { + "epoch": 2.5393749484621093, + "grad_norm": 1.8828150033950806, + "learning_rate": 1e-06, + "loss": 0.7722, + "mean_token_accuracy": 0.7451435327529907, + "num_tokens": 815786132.0, + "step": 30795 + }, + { + "epoch": 2.539457409087161, + "grad_norm": 2.1122491359710693, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.7963763475418091, + "num_tokens": 815814881.0, + "step": 30796 + }, + { + "epoch": 2.5395398697122125, + "grad_norm": 1.9647644758224487, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.80777907371521, + "num_tokens": 815845683.0, + "step": 30797 + }, + { + "epoch": 2.539622330337264, + "grad_norm": 2.1112046241760254, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7630476951599121, + "num_tokens": 815876456.0, + "step": 30798 + }, + { + "epoch": 2.5397047909623156, + "grad_norm": 2.028388500213623, + "learning_rate": 1e-06, + "loss": 0.5605, + "mean_token_accuracy": 0.812457263469696, + "num_tokens": 815905745.0, + "step": 30799 + }, + { + "epoch": 2.539787251587367, + "grad_norm": 1.99858558177948, + "learning_rate": 1e-06, + "loss": 0.4854, + "mean_token_accuracy": 0.8392594456672668, + "num_tokens": 815937521.0, + "step": 30800 + }, + { + "epoch": 2.5398697122124183, + "grad_norm": 2.1305277347564697, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.7691022157669067, + "num_tokens": 815966953.0, + "step": 30801 + }, + { + "epoch": 2.53995217283747, + "grad_norm": 2.415463447570801, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7627061605453491, + "num_tokens": 815990229.0, + "step": 30802 + }, + { + "epoch": 2.540034633462522, + "grad_norm": 2.066241979598999, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7570172548294067, + "num_tokens": 816022842.0, + "step": 30803 + }, + { + "epoch": 2.5401170940875732, + "grad_norm": 2.203505277633667, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.76357102394104, + "num_tokens": 816051162.0, + "step": 30804 + }, + { + "epoch": 2.5401995547126246, + "grad_norm": 2.3743391036987305, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7559720277786255, + "num_tokens": 816076515.0, + "step": 30805 + }, + { + "epoch": 2.5402820153376764, + "grad_norm": 2.3667662143707275, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7688320875167847, + "num_tokens": 816101713.0, + "step": 30806 + }, + { + "epoch": 2.5403644759627277, + "grad_norm": 1.9866411685943604, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.773127019405365, + "num_tokens": 816135818.0, + "step": 30807 + }, + { + "epoch": 2.540446936587779, + "grad_norm": 2.0014915466308594, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7756137251853943, + "num_tokens": 816167988.0, + "step": 30808 + }, + { + "epoch": 2.540529397212831, + "grad_norm": 2.0137693881988525, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7743931412696838, + "num_tokens": 816202152.0, + "step": 30809 + }, + { + "epoch": 2.5406118578378827, + "grad_norm": 2.2137365341186523, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7391011714935303, + "num_tokens": 816228519.0, + "step": 30810 + }, + { + "epoch": 2.540694318462934, + "grad_norm": 1.9603934288024902, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7552059292793274, + "num_tokens": 816264076.0, + "step": 30811 + }, + { + "epoch": 2.5407767790879854, + "grad_norm": 2.1894359588623047, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7775253653526306, + "num_tokens": 816294453.0, + "step": 30812 + }, + { + "epoch": 2.540859239713037, + "grad_norm": 2.277719497680664, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7536512017250061, + "num_tokens": 816321587.0, + "step": 30813 + }, + { + "epoch": 2.5409417003380885, + "grad_norm": 2.1364142894744873, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7608245015144348, + "num_tokens": 816352663.0, + "step": 30814 + }, + { + "epoch": 2.54102416096314, + "grad_norm": 2.18184757232666, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7496318817138672, + "num_tokens": 816380417.0, + "step": 30815 + }, + { + "epoch": 2.5411066215881917, + "grad_norm": 2.0010581016540527, + "learning_rate": 1e-06, + "loss": 0.6306, + "mean_token_accuracy": 0.7928824424743652, + "num_tokens": 816411176.0, + "step": 30816 + }, + { + "epoch": 2.541189082213243, + "grad_norm": 2.413315534591675, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7698664665222168, + "num_tokens": 816435587.0, + "step": 30817 + }, + { + "epoch": 2.541271542838295, + "grad_norm": 2.078378677368164, + "learning_rate": 1e-06, + "loss": 0.5451, + "mean_token_accuracy": 0.8157468438148499, + "num_tokens": 816465378.0, + "step": 30818 + }, + { + "epoch": 2.541354003463346, + "grad_norm": 2.4895482063293457, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.7864403128623962, + "num_tokens": 816489598.0, + "step": 30819 + }, + { + "epoch": 2.541436464088398, + "grad_norm": 1.9746190309524536, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.7840356826782227, + "num_tokens": 816522219.0, + "step": 30820 + }, + { + "epoch": 2.5415189247134493, + "grad_norm": 2.3593404293060303, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7773076891899109, + "num_tokens": 816546592.0, + "step": 30821 + }, + { + "epoch": 2.5416013853385007, + "grad_norm": 2.2019505500793457, + "learning_rate": 1e-06, + "loss": 0.5906, + "mean_token_accuracy": 0.8096203804016113, + "num_tokens": 816571690.0, + "step": 30822 + }, + { + "epoch": 2.5416838459635525, + "grad_norm": 2.002256393432617, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.772619366645813, + "num_tokens": 816601903.0, + "step": 30823 + }, + { + "epoch": 2.541766306588604, + "grad_norm": 2.2179243564605713, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7777513265609741, + "num_tokens": 816629864.0, + "step": 30824 + }, + { + "epoch": 2.5418487672136556, + "grad_norm": 1.802842617034912, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.7930999994277954, + "num_tokens": 816668774.0, + "step": 30825 + }, + { + "epoch": 2.541931227838707, + "grad_norm": 2.3540992736816406, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7818784713745117, + "num_tokens": 816693388.0, + "step": 30826 + }, + { + "epoch": 2.5420136884637587, + "grad_norm": 2.085179090499878, + "learning_rate": 1e-06, + "loss": 0.5962, + "mean_token_accuracy": 0.7963473200798035, + "num_tokens": 816721776.0, + "step": 30827 + }, + { + "epoch": 2.54209614908881, + "grad_norm": 2.223357677459717, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7622405290603638, + "num_tokens": 816748123.0, + "step": 30828 + }, + { + "epoch": 2.5421786097138614, + "grad_norm": 2.1276350021362305, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7450423240661621, + "num_tokens": 816779464.0, + "step": 30829 + }, + { + "epoch": 2.5422610703389132, + "grad_norm": 1.9990806579589844, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.7925967574119568, + "num_tokens": 816810872.0, + "step": 30830 + }, + { + "epoch": 2.5423435309639646, + "grad_norm": 2.040653944015503, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7448837757110596, + "num_tokens": 816842443.0, + "step": 30831 + }, + { + "epoch": 2.5424259915890164, + "grad_norm": 2.293949842453003, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7625518441200256, + "num_tokens": 816866785.0, + "step": 30832 + }, + { + "epoch": 2.5425084522140677, + "grad_norm": 2.3112807273864746, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7359523773193359, + "num_tokens": 816894881.0, + "step": 30833 + }, + { + "epoch": 2.5425909128391195, + "grad_norm": 2.374152660369873, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7805849313735962, + "num_tokens": 816919577.0, + "step": 30834 + }, + { + "epoch": 2.542673373464171, + "grad_norm": 1.964208722114563, + "learning_rate": 1e-06, + "loss": 0.5829, + "mean_token_accuracy": 0.806820273399353, + "num_tokens": 816953134.0, + "step": 30835 + }, + { + "epoch": 2.5427558340892222, + "grad_norm": 1.9995723962783813, + "learning_rate": 1e-06, + "loss": 0.5811, + "mean_token_accuracy": 0.8033120632171631, + "num_tokens": 816985109.0, + "step": 30836 + }, + { + "epoch": 2.542838294714274, + "grad_norm": 2.114011526107788, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.750812828540802, + "num_tokens": 817012894.0, + "step": 30837 + }, + { + "epoch": 2.5429207553393254, + "grad_norm": 2.286846399307251, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7757669687271118, + "num_tokens": 817039255.0, + "step": 30838 + }, + { + "epoch": 2.543003215964377, + "grad_norm": 2.0967469215393066, + "learning_rate": 1e-06, + "loss": 0.5789, + "mean_token_accuracy": 0.7922630310058594, + "num_tokens": 817066557.0, + "step": 30839 + }, + { + "epoch": 2.5430856765894285, + "grad_norm": 2.1570398807525635, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7779709100723267, + "num_tokens": 817096565.0, + "step": 30840 + }, + { + "epoch": 2.5431681372144803, + "grad_norm": 1.9732999801635742, + "learning_rate": 1e-06, + "loss": 0.733, + "mean_token_accuracy": 0.7658268809318542, + "num_tokens": 817131614.0, + "step": 30841 + }, + { + "epoch": 2.5432505978395317, + "grad_norm": 1.942700982093811, + "learning_rate": 1e-06, + "loss": 0.5463, + "mean_token_accuracy": 0.8231067657470703, + "num_tokens": 817167011.0, + "step": 30842 + }, + { + "epoch": 2.543333058464583, + "grad_norm": 2.0290956497192383, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7544493675231934, + "num_tokens": 817201621.0, + "step": 30843 + }, + { + "epoch": 2.543415519089635, + "grad_norm": 2.0332283973693848, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.8012298345565796, + "num_tokens": 817231366.0, + "step": 30844 + }, + { + "epoch": 2.543497979714686, + "grad_norm": 2.274160623550415, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7704865336418152, + "num_tokens": 817258314.0, + "step": 30845 + }, + { + "epoch": 2.5435804403397375, + "grad_norm": 2.405791997909546, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7558373212814331, + "num_tokens": 817282712.0, + "step": 30846 + }, + { + "epoch": 2.5436629009647893, + "grad_norm": 1.8769315481185913, + "learning_rate": 1e-06, + "loss": 0.5355, + "mean_token_accuracy": 0.8126240968704224, + "num_tokens": 817316587.0, + "step": 30847 + }, + { + "epoch": 2.543745361589841, + "grad_norm": 2.01350736618042, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.7942694425582886, + "num_tokens": 817344962.0, + "step": 30848 + }, + { + "epoch": 2.5438278222148925, + "grad_norm": 1.849549412727356, + "learning_rate": 1e-06, + "loss": 0.5979, + "mean_token_accuracy": 0.8092215061187744, + "num_tokens": 817378877.0, + "step": 30849 + }, + { + "epoch": 2.543910282839944, + "grad_norm": 1.8133317232131958, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.7977561950683594, + "num_tokens": 817413896.0, + "step": 30850 + }, + { + "epoch": 2.5439927434649956, + "grad_norm": 1.9470975399017334, + "learning_rate": 1e-06, + "loss": 0.7538, + "mean_token_accuracy": 0.7581165432929993, + "num_tokens": 817448815.0, + "step": 30851 + }, + { + "epoch": 2.544075204090047, + "grad_norm": 1.9970828294754028, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.787421703338623, + "num_tokens": 817479253.0, + "step": 30852 + }, + { + "epoch": 2.5441576647150983, + "grad_norm": 2.181258201599121, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7567337155342102, + "num_tokens": 817508014.0, + "step": 30853 + }, + { + "epoch": 2.54424012534015, + "grad_norm": 2.236483097076416, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.756721019744873, + "num_tokens": 817536669.0, + "step": 30854 + }, + { + "epoch": 2.544322585965202, + "grad_norm": 2.2213058471679688, + "learning_rate": 1e-06, + "loss": 0.7025, + "mean_token_accuracy": 0.7695527672767639, + "num_tokens": 817564613.0, + "step": 30855 + }, + { + "epoch": 2.5444050465902532, + "grad_norm": 2.0613977909088135, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7728241682052612, + "num_tokens": 817598367.0, + "step": 30856 + }, + { + "epoch": 2.5444875072153046, + "grad_norm": 1.9315537214279175, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.7878866195678711, + "num_tokens": 817630446.0, + "step": 30857 + }, + { + "epoch": 2.5445699678403564, + "grad_norm": 1.94895601272583, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7510936260223389, + "num_tokens": 817664682.0, + "step": 30858 + }, + { + "epoch": 2.5446524284654077, + "grad_norm": 1.8845144510269165, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7728699445724487, + "num_tokens": 817700750.0, + "step": 30859 + }, + { + "epoch": 2.544734889090459, + "grad_norm": 1.8930493593215942, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7702728509902954, + "num_tokens": 817733925.0, + "step": 30860 + }, + { + "epoch": 2.544817349715511, + "grad_norm": 2.1649699211120605, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.745316743850708, + "num_tokens": 817760684.0, + "step": 30861 + }, + { + "epoch": 2.5448998103405622, + "grad_norm": 1.7221401929855347, + "learning_rate": 1e-06, + "loss": 0.4695, + "mean_token_accuracy": 0.8403716683387756, + "num_tokens": 817792660.0, + "step": 30862 + }, + { + "epoch": 2.544982270965614, + "grad_norm": 1.975894808769226, + "learning_rate": 1e-06, + "loss": 0.7744, + "mean_token_accuracy": 0.7537798285484314, + "num_tokens": 817827242.0, + "step": 30863 + }, + { + "epoch": 2.5450647315906654, + "grad_norm": 2.2589094638824463, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7653703689575195, + "num_tokens": 817854381.0, + "step": 30864 + }, + { + "epoch": 2.545147192215717, + "grad_norm": 2.0408570766448975, + "learning_rate": 1e-06, + "loss": 0.6024, + "mean_token_accuracy": 0.7966018915176392, + "num_tokens": 817886362.0, + "step": 30865 + }, + { + "epoch": 2.5452296528407685, + "grad_norm": 1.9169106483459473, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7278916835784912, + "num_tokens": 817922389.0, + "step": 30866 + }, + { + "epoch": 2.54531211346582, + "grad_norm": 2.2795422077178955, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.752187967300415, + "num_tokens": 817949577.0, + "step": 30867 + }, + { + "epoch": 2.5453945740908717, + "grad_norm": 2.064891815185547, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7754164934158325, + "num_tokens": 817981923.0, + "step": 30868 + }, + { + "epoch": 2.545477034715923, + "grad_norm": 2.2066073417663574, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7306338548660278, + "num_tokens": 818011828.0, + "step": 30869 + }, + { + "epoch": 2.545559495340975, + "grad_norm": 2.4058420658111572, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7524060010910034, + "num_tokens": 818041312.0, + "step": 30870 + }, + { + "epoch": 2.545641955966026, + "grad_norm": 2.2587392330169678, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7487440705299377, + "num_tokens": 818070708.0, + "step": 30871 + }, + { + "epoch": 2.545724416591078, + "grad_norm": 2.2246971130371094, + "learning_rate": 1e-06, + "loss": 0.7551, + "mean_token_accuracy": 0.7633795142173767, + "num_tokens": 818098650.0, + "step": 30872 + }, + { + "epoch": 2.5458068772161293, + "grad_norm": 1.865339756011963, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7617557048797607, + "num_tokens": 818133489.0, + "step": 30873 + }, + { + "epoch": 2.5458893378411807, + "grad_norm": 2.21584415435791, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.7876116037368774, + "num_tokens": 818161074.0, + "step": 30874 + }, + { + "epoch": 2.5459717984662324, + "grad_norm": 2.5211308002471924, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7549974322319031, + "num_tokens": 818192813.0, + "step": 30875 + }, + { + "epoch": 2.546054259091284, + "grad_norm": 2.3884952068328857, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7626729011535645, + "num_tokens": 818218410.0, + "step": 30876 + }, + { + "epoch": 2.5461367197163356, + "grad_norm": 2.109959840774536, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.7920126914978027, + "num_tokens": 818246431.0, + "step": 30877 + }, + { + "epoch": 2.546219180341387, + "grad_norm": 2.211205005645752, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7505821585655212, + "num_tokens": 818274924.0, + "step": 30878 + }, + { + "epoch": 2.5463016409664387, + "grad_norm": 2.2300453186035156, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7549146413803101, + "num_tokens": 818302903.0, + "step": 30879 + }, + { + "epoch": 2.54638410159149, + "grad_norm": 1.8289811611175537, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7849316000938416, + "num_tokens": 818338906.0, + "step": 30880 + }, + { + "epoch": 2.5464665622165414, + "grad_norm": 2.1228346824645996, + "learning_rate": 1e-06, + "loss": 0.7846, + "mean_token_accuracy": 0.7456201314926147, + "num_tokens": 818369971.0, + "step": 30881 + }, + { + "epoch": 2.5465490228415932, + "grad_norm": 2.2540690898895264, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7734270095825195, + "num_tokens": 818397210.0, + "step": 30882 + }, + { + "epoch": 2.5466314834666446, + "grad_norm": 1.910334825515747, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7718138694763184, + "num_tokens": 818432774.0, + "step": 30883 + }, + { + "epoch": 2.5467139440916964, + "grad_norm": 2.078078031539917, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7844998836517334, + "num_tokens": 818465009.0, + "step": 30884 + }, + { + "epoch": 2.5467964047167477, + "grad_norm": 2.22184157371521, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7643754482269287, + "num_tokens": 818493481.0, + "step": 30885 + }, + { + "epoch": 2.5468788653417995, + "grad_norm": 2.1887760162353516, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.7876394987106323, + "num_tokens": 818519130.0, + "step": 30886 + }, + { + "epoch": 2.546961325966851, + "grad_norm": 2.234520435333252, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7673394680023193, + "num_tokens": 818549838.0, + "step": 30887 + }, + { + "epoch": 2.5470437865919022, + "grad_norm": 1.8352640867233276, + "learning_rate": 1e-06, + "loss": 0.5686, + "mean_token_accuracy": 0.8211040496826172, + "num_tokens": 818583672.0, + "step": 30888 + }, + { + "epoch": 2.547126247216954, + "grad_norm": 2.1044998168945312, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.7924819588661194, + "num_tokens": 818613654.0, + "step": 30889 + }, + { + "epoch": 2.5472087078420054, + "grad_norm": 2.280609607696533, + "learning_rate": 1e-06, + "loss": 0.5973, + "mean_token_accuracy": 0.7859698534011841, + "num_tokens": 818639283.0, + "step": 30890 + }, + { + "epoch": 2.5472911684670567, + "grad_norm": 2.3115854263305664, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7557648420333862, + "num_tokens": 818663623.0, + "step": 30891 + }, + { + "epoch": 2.5473736290921085, + "grad_norm": 2.1490395069122314, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7434443831443787, + "num_tokens": 818692541.0, + "step": 30892 + }, + { + "epoch": 2.5474560897171603, + "grad_norm": 1.9356915950775146, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7809950113296509, + "num_tokens": 818725386.0, + "step": 30893 + }, + { + "epoch": 2.5475385503422117, + "grad_norm": 2.286872625350952, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7827106714248657, + "num_tokens": 818751039.0, + "step": 30894 + }, + { + "epoch": 2.547621010967263, + "grad_norm": 2.164233684539795, + "learning_rate": 1e-06, + "loss": 0.5943, + "mean_token_accuracy": 0.8006315231323242, + "num_tokens": 818779797.0, + "step": 30895 + }, + { + "epoch": 2.547703471592315, + "grad_norm": 2.151862621307373, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.78676438331604, + "num_tokens": 818810466.0, + "step": 30896 + }, + { + "epoch": 2.547785932217366, + "grad_norm": 1.9356486797332764, + "learning_rate": 1e-06, + "loss": 0.5901, + "mean_token_accuracy": 0.8003273010253906, + "num_tokens": 818840470.0, + "step": 30897 + }, + { + "epoch": 2.5478683928424175, + "grad_norm": 2.007106065750122, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7793435454368591, + "num_tokens": 818871143.0, + "step": 30898 + }, + { + "epoch": 2.5479508534674693, + "grad_norm": 2.2499496936798096, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7700387239456177, + "num_tokens": 818898314.0, + "step": 30899 + }, + { + "epoch": 2.548033314092521, + "grad_norm": 2.1392459869384766, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7416651248931885, + "num_tokens": 818928931.0, + "step": 30900 + }, + { + "epoch": 2.5481157747175724, + "grad_norm": 2.098836660385132, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7679834365844727, + "num_tokens": 818959741.0, + "step": 30901 + }, + { + "epoch": 2.548198235342624, + "grad_norm": 1.9854459762573242, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7845700979232788, + "num_tokens": 818992740.0, + "step": 30902 + }, + { + "epoch": 2.5482806959676756, + "grad_norm": 2.0812838077545166, + "learning_rate": 1e-06, + "loss": 0.5887, + "mean_token_accuracy": 0.8068903684616089, + "num_tokens": 819021750.0, + "step": 30903 + }, + { + "epoch": 2.548363156592727, + "grad_norm": 2.1669516563415527, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7202351689338684, + "num_tokens": 819049648.0, + "step": 30904 + }, + { + "epoch": 2.5484456172177783, + "grad_norm": 1.9508521556854248, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.777066707611084, + "num_tokens": 819081601.0, + "step": 30905 + }, + { + "epoch": 2.54852807784283, + "grad_norm": 1.895959496498108, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7593452334403992, + "num_tokens": 819119800.0, + "step": 30906 + }, + { + "epoch": 2.5486105384678814, + "grad_norm": 2.0495831966400146, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7649871110916138, + "num_tokens": 819151474.0, + "step": 30907 + }, + { + "epoch": 2.5486929990929332, + "grad_norm": 2.055499792098999, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.7881984710693359, + "num_tokens": 819182218.0, + "step": 30908 + }, + { + "epoch": 2.5487754597179846, + "grad_norm": 2.042983293533325, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.7775038480758667, + "num_tokens": 819213308.0, + "step": 30909 + }, + { + "epoch": 2.5488579203430364, + "grad_norm": 1.9022599458694458, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7516239881515503, + "num_tokens": 819245860.0, + "step": 30910 + }, + { + "epoch": 2.5489403809680877, + "grad_norm": 1.9994009733200073, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.760546863079071, + "num_tokens": 819277846.0, + "step": 30911 + }, + { + "epoch": 2.549022841593139, + "grad_norm": 2.1485304832458496, + "learning_rate": 1e-06, + "loss": 0.6589, + "mean_token_accuracy": 0.776141881942749, + "num_tokens": 819305901.0, + "step": 30912 + }, + { + "epoch": 2.549105302218191, + "grad_norm": 2.025236129760742, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.8044173717498779, + "num_tokens": 819332949.0, + "step": 30913 + }, + { + "epoch": 2.549187762843242, + "grad_norm": 2.166746139526367, + "learning_rate": 1e-06, + "loss": 0.6122, + "mean_token_accuracy": 0.7870889902114868, + "num_tokens": 819361862.0, + "step": 30914 + }, + { + "epoch": 2.549270223468294, + "grad_norm": 2.2039215564727783, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.7790387868881226, + "num_tokens": 819389453.0, + "step": 30915 + }, + { + "epoch": 2.5493526840933454, + "grad_norm": 2.428060293197632, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7717851400375366, + "num_tokens": 819414403.0, + "step": 30916 + }, + { + "epoch": 2.549435144718397, + "grad_norm": 1.9629918336868286, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7755914926528931, + "num_tokens": 819450947.0, + "step": 30917 + }, + { + "epoch": 2.5495176053434485, + "grad_norm": 2.0387580394744873, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.7922630310058594, + "num_tokens": 819482184.0, + "step": 30918 + }, + { + "epoch": 2.5496000659685, + "grad_norm": 2.41153621673584, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7253743410110474, + "num_tokens": 819507520.0, + "step": 30919 + }, + { + "epoch": 2.5496825265935517, + "grad_norm": 1.9854059219360352, + "learning_rate": 1e-06, + "loss": 0.706, + "mean_token_accuracy": 0.7765166759490967, + "num_tokens": 819542167.0, + "step": 30920 + }, + { + "epoch": 2.549764987218603, + "grad_norm": 1.9053725004196167, + "learning_rate": 1e-06, + "loss": 0.594, + "mean_token_accuracy": 0.8067185282707214, + "num_tokens": 819576371.0, + "step": 30921 + }, + { + "epoch": 2.549847447843655, + "grad_norm": 2.1754541397094727, + "learning_rate": 1e-06, + "loss": 0.5733, + "mean_token_accuracy": 0.796983540058136, + "num_tokens": 819606056.0, + "step": 30922 + }, + { + "epoch": 2.549929908468706, + "grad_norm": 2.4046568870544434, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7792304754257202, + "num_tokens": 819630902.0, + "step": 30923 + }, + { + "epoch": 2.550012369093758, + "grad_norm": 2.157191514968872, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.7593182325363159, + "num_tokens": 819658674.0, + "step": 30924 + }, + { + "epoch": 2.5500948297188093, + "grad_norm": 2.2431650161743164, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7316122055053711, + "num_tokens": 819687689.0, + "step": 30925 + }, + { + "epoch": 2.5501772903438606, + "grad_norm": 2.0629148483276367, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7875374555587769, + "num_tokens": 819717771.0, + "step": 30926 + }, + { + "epoch": 2.5502597509689124, + "grad_norm": 1.929960012435913, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7782635688781738, + "num_tokens": 819752429.0, + "step": 30927 + }, + { + "epoch": 2.550342211593964, + "grad_norm": 1.8890857696533203, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7636258006095886, + "num_tokens": 819786001.0, + "step": 30928 + }, + { + "epoch": 2.550424672219015, + "grad_norm": 2.133739471435547, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7579922676086426, + "num_tokens": 819815656.0, + "step": 30929 + }, + { + "epoch": 2.550507132844067, + "grad_norm": 2.0521678924560547, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.793834924697876, + "num_tokens": 819847640.0, + "step": 30930 + }, + { + "epoch": 2.5505895934691187, + "grad_norm": 2.0369319915771484, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7839335799217224, + "num_tokens": 819876614.0, + "step": 30931 + }, + { + "epoch": 2.55067205409417, + "grad_norm": 2.306992292404175, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7753956317901611, + "num_tokens": 819904365.0, + "step": 30932 + }, + { + "epoch": 2.5507545147192214, + "grad_norm": 2.0359885692596436, + "learning_rate": 1e-06, + "loss": 0.5767, + "mean_token_accuracy": 0.8102121353149414, + "num_tokens": 819935123.0, + "step": 30933 + }, + { + "epoch": 2.5508369753442732, + "grad_norm": 1.9214422702789307, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7772015333175659, + "num_tokens": 819968348.0, + "step": 30934 + }, + { + "epoch": 2.5509194359693246, + "grad_norm": 2.310370922088623, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.7388722896575928, + "num_tokens": 819996568.0, + "step": 30935 + }, + { + "epoch": 2.551001896594376, + "grad_norm": 1.8930251598358154, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7589088678359985, + "num_tokens": 820032347.0, + "step": 30936 + }, + { + "epoch": 2.5510843572194277, + "grad_norm": 2.0069451332092285, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.78144371509552, + "num_tokens": 820062289.0, + "step": 30937 + }, + { + "epoch": 2.5511668178444795, + "grad_norm": 1.9911402463912964, + "learning_rate": 1e-06, + "loss": 0.5393, + "mean_token_accuracy": 0.820026159286499, + "num_tokens": 820095870.0, + "step": 30938 + }, + { + "epoch": 2.551249278469531, + "grad_norm": 1.9855352640151978, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.792422890663147, + "num_tokens": 820128442.0, + "step": 30939 + }, + { + "epoch": 2.551331739094582, + "grad_norm": 2.483543634414673, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7520614862442017, + "num_tokens": 820152260.0, + "step": 30940 + }, + { + "epoch": 2.551414199719634, + "grad_norm": 2.282668352127075, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7708188891410828, + "num_tokens": 820179877.0, + "step": 30941 + }, + { + "epoch": 2.5514966603446854, + "grad_norm": 2.0716066360473633, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.7864623665809631, + "num_tokens": 820214413.0, + "step": 30942 + }, + { + "epoch": 2.5515791209697367, + "grad_norm": 2.34710955619812, + "learning_rate": 1e-06, + "loss": 0.6079, + "mean_token_accuracy": 0.7871738076210022, + "num_tokens": 820239277.0, + "step": 30943 + }, + { + "epoch": 2.5516615815947885, + "grad_norm": 1.9889369010925293, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7704172134399414, + "num_tokens": 820273791.0, + "step": 30944 + }, + { + "epoch": 2.5517440422198403, + "grad_norm": 2.2141051292419434, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.7810045480728149, + "num_tokens": 820297929.0, + "step": 30945 + }, + { + "epoch": 2.5518265028448917, + "grad_norm": 2.163469076156616, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7723819613456726, + "num_tokens": 820327975.0, + "step": 30946 + }, + { + "epoch": 2.551908963469943, + "grad_norm": 2.3793704509735107, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7644416093826294, + "num_tokens": 820352737.0, + "step": 30947 + }, + { + "epoch": 2.551991424094995, + "grad_norm": 2.2430920600891113, + "learning_rate": 1e-06, + "loss": 0.6034, + "mean_token_accuracy": 0.7923341393470764, + "num_tokens": 820377324.0, + "step": 30948 + }, + { + "epoch": 2.552073884720046, + "grad_norm": 1.8109095096588135, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.7898747324943542, + "num_tokens": 820414914.0, + "step": 30949 + }, + { + "epoch": 2.5521563453450975, + "grad_norm": 2.460782527923584, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.7680965662002563, + "num_tokens": 820437647.0, + "step": 30950 + }, + { + "epoch": 2.5522388059701493, + "grad_norm": 2.004077672958374, + "learning_rate": 1e-06, + "loss": 0.5627, + "mean_token_accuracy": 0.7809725999832153, + "num_tokens": 820467763.0, + "step": 30951 + }, + { + "epoch": 2.5523212665952006, + "grad_norm": 1.9911372661590576, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.7857379913330078, + "num_tokens": 820498985.0, + "step": 30952 + }, + { + "epoch": 2.5524037272202524, + "grad_norm": 2.4662182331085205, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7730094194412231, + "num_tokens": 820522567.0, + "step": 30953 + }, + { + "epoch": 2.552486187845304, + "grad_norm": 2.2137811183929443, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7421067953109741, + "num_tokens": 820551852.0, + "step": 30954 + }, + { + "epoch": 2.5525686484703556, + "grad_norm": 2.3187663555145264, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7281194925308228, + "num_tokens": 820580325.0, + "step": 30955 + }, + { + "epoch": 2.552651109095407, + "grad_norm": 2.0338966846466064, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7242861390113831, + "num_tokens": 820613487.0, + "step": 30956 + }, + { + "epoch": 2.5527335697204583, + "grad_norm": 2.139111042022705, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7632317543029785, + "num_tokens": 820644527.0, + "step": 30957 + }, + { + "epoch": 2.55281603034551, + "grad_norm": 2.1140494346618652, + "learning_rate": 1e-06, + "loss": 0.5675, + "mean_token_accuracy": 0.8093757629394531, + "num_tokens": 820674095.0, + "step": 30958 + }, + { + "epoch": 2.5528984909705614, + "grad_norm": 1.9109972715377808, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7744231224060059, + "num_tokens": 820709125.0, + "step": 30959 + }, + { + "epoch": 2.5529809515956132, + "grad_norm": 2.1356096267700195, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.749558687210083, + "num_tokens": 820741589.0, + "step": 30960 + }, + { + "epoch": 2.5530634122206646, + "grad_norm": 2.1629130840301514, + "learning_rate": 1e-06, + "loss": 0.5844, + "mean_token_accuracy": 0.8004826307296753, + "num_tokens": 820770329.0, + "step": 30961 + }, + { + "epoch": 2.5531458728457164, + "grad_norm": 2.2117137908935547, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7857362031936646, + "num_tokens": 820796460.0, + "step": 30962 + }, + { + "epoch": 2.5532283334707677, + "grad_norm": 1.996458888053894, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.7845945954322815, + "num_tokens": 820830304.0, + "step": 30963 + }, + { + "epoch": 2.553310794095819, + "grad_norm": 1.828654170036316, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.8061649799346924, + "num_tokens": 820864837.0, + "step": 30964 + }, + { + "epoch": 2.553393254720871, + "grad_norm": 1.9638055562973022, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.755238950252533, + "num_tokens": 820897446.0, + "step": 30965 + }, + { + "epoch": 2.553475715345922, + "grad_norm": 1.9005857706069946, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7416355609893799, + "num_tokens": 820933544.0, + "step": 30966 + }, + { + "epoch": 2.553558175970974, + "grad_norm": 2.0846970081329346, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7721682786941528, + "num_tokens": 820962942.0, + "step": 30967 + }, + { + "epoch": 2.5536406365960254, + "grad_norm": 2.2322521209716797, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.7881690263748169, + "num_tokens": 820988640.0, + "step": 30968 + }, + { + "epoch": 2.553723097221077, + "grad_norm": 2.2683894634246826, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.7753452062606812, + "num_tokens": 821016957.0, + "step": 30969 + }, + { + "epoch": 2.5538055578461285, + "grad_norm": 1.9434267282485962, + "learning_rate": 1e-06, + "loss": 0.5054, + "mean_token_accuracy": 0.8263494372367859, + "num_tokens": 821047926.0, + "step": 30970 + }, + { + "epoch": 2.55388801847118, + "grad_norm": 2.4279592037200928, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7705384492874146, + "num_tokens": 821073568.0, + "step": 30971 + }, + { + "epoch": 2.5539704790962316, + "grad_norm": 2.026705026626587, + "learning_rate": 1e-06, + "loss": 0.6195, + "mean_token_accuracy": 0.7932330369949341, + "num_tokens": 821103338.0, + "step": 30972 + }, + { + "epoch": 2.554052939721283, + "grad_norm": 2.1307973861694336, + "learning_rate": 1e-06, + "loss": 0.5882, + "mean_token_accuracy": 0.8037825226783752, + "num_tokens": 821132152.0, + "step": 30973 + }, + { + "epoch": 2.5541354003463344, + "grad_norm": 2.0233547687530518, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.7926226854324341, + "num_tokens": 821162663.0, + "step": 30974 + }, + { + "epoch": 2.554217860971386, + "grad_norm": 2.271366596221924, + "learning_rate": 1e-06, + "loss": 0.6485, + "mean_token_accuracy": 0.7862757444381714, + "num_tokens": 821186474.0, + "step": 30975 + }, + { + "epoch": 2.554300321596438, + "grad_norm": 2.3962626457214355, + "learning_rate": 1e-06, + "loss": 0.7772, + "mean_token_accuracy": 0.7499085664749146, + "num_tokens": 821214567.0, + "step": 30976 + }, + { + "epoch": 2.5543827822214893, + "grad_norm": 2.063222885131836, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7693988680839539, + "num_tokens": 821243499.0, + "step": 30977 + }, + { + "epoch": 2.5544652428465406, + "grad_norm": 2.1759183406829834, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7718154788017273, + "num_tokens": 821269887.0, + "step": 30978 + }, + { + "epoch": 2.5545477034715924, + "grad_norm": 2.255676507949829, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.741503119468689, + "num_tokens": 821297503.0, + "step": 30979 + }, + { + "epoch": 2.554630164096644, + "grad_norm": 1.994187831878662, + "learning_rate": 1e-06, + "loss": 0.7669, + "mean_token_accuracy": 0.7503538131713867, + "num_tokens": 821329788.0, + "step": 30980 + }, + { + "epoch": 2.554712624721695, + "grad_norm": 1.782653570175171, + "learning_rate": 1e-06, + "loss": 0.5729, + "mean_token_accuracy": 0.8028015494346619, + "num_tokens": 821367353.0, + "step": 30981 + }, + { + "epoch": 2.554795085346747, + "grad_norm": 2.292112350463867, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7735621929168701, + "num_tokens": 821393821.0, + "step": 30982 + }, + { + "epoch": 2.5548775459717987, + "grad_norm": 2.2900314331054688, + "learning_rate": 1e-06, + "loss": 0.5894, + "mean_token_accuracy": 0.8053923845291138, + "num_tokens": 821418823.0, + "step": 30983 + }, + { + "epoch": 2.55496000659685, + "grad_norm": 2.2756166458129883, + "learning_rate": 1e-06, + "loss": 0.5813, + "mean_token_accuracy": 0.8077920079231262, + "num_tokens": 821446177.0, + "step": 30984 + }, + { + "epoch": 2.5550424672219014, + "grad_norm": 1.780605435371399, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.8014287948608398, + "num_tokens": 821486736.0, + "step": 30985 + }, + { + "epoch": 2.555124927846953, + "grad_norm": 1.970737338066101, + "learning_rate": 1e-06, + "loss": 0.4907, + "mean_token_accuracy": 0.8344470262527466, + "num_tokens": 821515043.0, + "step": 30986 + }, + { + "epoch": 2.5552073884720046, + "grad_norm": 1.9683786630630493, + "learning_rate": 1e-06, + "loss": 0.5684, + "mean_token_accuracy": 0.8182827830314636, + "num_tokens": 821546027.0, + "step": 30987 + }, + { + "epoch": 2.555289849097056, + "grad_norm": 2.001936435699463, + "learning_rate": 1e-06, + "loss": 0.6114, + "mean_token_accuracy": 0.801133394241333, + "num_tokens": 821577746.0, + "step": 30988 + }, + { + "epoch": 2.5553723097221077, + "grad_norm": 2.244330883026123, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.752618670463562, + "num_tokens": 821605187.0, + "step": 30989 + }, + { + "epoch": 2.555454770347159, + "grad_norm": 2.0178463459014893, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7795768976211548, + "num_tokens": 821635812.0, + "step": 30990 + }, + { + "epoch": 2.555537230972211, + "grad_norm": 2.226442337036133, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7686879634857178, + "num_tokens": 821663293.0, + "step": 30991 + }, + { + "epoch": 2.555619691597262, + "grad_norm": 2.6952154636383057, + "learning_rate": 1e-06, + "loss": 0.7133, + "mean_token_accuracy": 0.7666864991188049, + "num_tokens": 821690952.0, + "step": 30992 + }, + { + "epoch": 2.555702152222314, + "grad_norm": 1.925156593322754, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.7912798523902893, + "num_tokens": 821728027.0, + "step": 30993 + }, + { + "epoch": 2.5557846128473654, + "grad_norm": 2.350714683532715, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.7749569416046143, + "num_tokens": 821752108.0, + "step": 30994 + }, + { + "epoch": 2.5558670734724167, + "grad_norm": 2.0849802494049072, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7472628951072693, + "num_tokens": 821781505.0, + "step": 30995 + }, + { + "epoch": 2.5559495340974685, + "grad_norm": 2.3917176723480225, + "learning_rate": 1e-06, + "loss": 0.5933, + "mean_token_accuracy": 0.7958335876464844, + "num_tokens": 821807676.0, + "step": 30996 + }, + { + "epoch": 2.55603199472252, + "grad_norm": 2.0160231590270996, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7871471047401428, + "num_tokens": 821839627.0, + "step": 30997 + }, + { + "epoch": 2.5561144553475716, + "grad_norm": 1.8634599447250366, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.7868490815162659, + "num_tokens": 821872346.0, + "step": 30998 + }, + { + "epoch": 2.556196915972623, + "grad_norm": 1.9590141773223877, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.77154940366745, + "num_tokens": 821911490.0, + "step": 30999 + }, + { + "epoch": 2.556279376597675, + "grad_norm": 2.1990342140197754, + "learning_rate": 1e-06, + "loss": 0.5766, + "mean_token_accuracy": 0.7971538305282593, + "num_tokens": 821938532.0, + "step": 31000 + }, + { + "epoch": 2.556361837222726, + "grad_norm": 2.151043176651001, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7723516821861267, + "num_tokens": 821967616.0, + "step": 31001 + }, + { + "epoch": 2.5564442978477775, + "grad_norm": 2.3052523136138916, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.7871782779693604, + "num_tokens": 821993413.0, + "step": 31002 + }, + { + "epoch": 2.5565267584728293, + "grad_norm": 2.137294292449951, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7535122632980347, + "num_tokens": 822025167.0, + "step": 31003 + }, + { + "epoch": 2.5566092190978806, + "grad_norm": 2.1084213256835938, + "learning_rate": 1e-06, + "loss": 0.6087, + "mean_token_accuracy": 0.7968209981918335, + "num_tokens": 822055228.0, + "step": 31004 + }, + { + "epoch": 2.5566916797229324, + "grad_norm": 1.941011667251587, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.7882456183433533, + "num_tokens": 822089727.0, + "step": 31005 + }, + { + "epoch": 2.556774140347984, + "grad_norm": 2.2518866062164307, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7750502824783325, + "num_tokens": 822116047.0, + "step": 31006 + }, + { + "epoch": 2.5568566009730356, + "grad_norm": 1.9872525930404663, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7855302095413208, + "num_tokens": 822148624.0, + "step": 31007 + }, + { + "epoch": 2.556939061598087, + "grad_norm": 2.6563665866851807, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7565069198608398, + "num_tokens": 822168274.0, + "step": 31008 + }, + { + "epoch": 2.5570215222231383, + "grad_norm": 2.0537092685699463, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7574836015701294, + "num_tokens": 822200841.0, + "step": 31009 + }, + { + "epoch": 2.55710398284819, + "grad_norm": 1.9580631256103516, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7700128555297852, + "num_tokens": 822233479.0, + "step": 31010 + }, + { + "epoch": 2.5571864434732414, + "grad_norm": 1.948727011680603, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7645739912986755, + "num_tokens": 822269198.0, + "step": 31011 + }, + { + "epoch": 2.557268904098293, + "grad_norm": 2.028012275695801, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7427824139595032, + "num_tokens": 822304297.0, + "step": 31012 + }, + { + "epoch": 2.5573513647233446, + "grad_norm": 1.9344643354415894, + "learning_rate": 1e-06, + "loss": 0.7047, + "mean_token_accuracy": 0.7734951972961426, + "num_tokens": 822335117.0, + "step": 31013 + }, + { + "epoch": 2.5574338253483964, + "grad_norm": 2.1745336055755615, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7438246011734009, + "num_tokens": 822364824.0, + "step": 31014 + }, + { + "epoch": 2.5575162859734477, + "grad_norm": 1.7321151494979858, + "learning_rate": 1e-06, + "loss": 0.5369, + "mean_token_accuracy": 0.8155995607376099, + "num_tokens": 822404646.0, + "step": 31015 + }, + { + "epoch": 2.557598746598499, + "grad_norm": 2.1321208477020264, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.762877345085144, + "num_tokens": 822432924.0, + "step": 31016 + }, + { + "epoch": 2.557681207223551, + "grad_norm": 2.160407543182373, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7663660049438477, + "num_tokens": 822462505.0, + "step": 31017 + }, + { + "epoch": 2.557763667848602, + "grad_norm": 2.2028024196624756, + "learning_rate": 1e-06, + "loss": 0.6271, + "mean_token_accuracy": 0.7868615984916687, + "num_tokens": 822490079.0, + "step": 31018 + }, + { + "epoch": 2.5578461284736536, + "grad_norm": 2.2852818965911865, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7400231957435608, + "num_tokens": 822518775.0, + "step": 31019 + }, + { + "epoch": 2.5579285890987054, + "grad_norm": 2.3337180614471436, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7720094919204712, + "num_tokens": 822544928.0, + "step": 31020 + }, + { + "epoch": 2.558011049723757, + "grad_norm": 2.3628523349761963, + "learning_rate": 1e-06, + "loss": 0.7045, + "mean_token_accuracy": 0.7668004035949707, + "num_tokens": 822571747.0, + "step": 31021 + }, + { + "epoch": 2.5580935103488085, + "grad_norm": 2.0476691722869873, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7623924016952515, + "num_tokens": 822602735.0, + "step": 31022 + }, + { + "epoch": 2.55817597097386, + "grad_norm": 1.9360225200653076, + "learning_rate": 1e-06, + "loss": 0.6186, + "mean_token_accuracy": 0.7866531610488892, + "num_tokens": 822636721.0, + "step": 31023 + }, + { + "epoch": 2.5582584315989116, + "grad_norm": 2.3147552013397217, + "learning_rate": 1e-06, + "loss": 0.5635, + "mean_token_accuracy": 0.8035244941711426, + "num_tokens": 822661367.0, + "step": 31024 + }, + { + "epoch": 2.558340892223963, + "grad_norm": 1.979762077331543, + "learning_rate": 1e-06, + "loss": 0.651, + "mean_token_accuracy": 0.7848184108734131, + "num_tokens": 822693998.0, + "step": 31025 + }, + { + "epoch": 2.5584233528490143, + "grad_norm": 2.030008316040039, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7773733139038086, + "num_tokens": 822728340.0, + "step": 31026 + }, + { + "epoch": 2.558505813474066, + "grad_norm": 1.981722354888916, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7524819374084473, + "num_tokens": 822762121.0, + "step": 31027 + }, + { + "epoch": 2.558588274099118, + "grad_norm": 2.1410796642303467, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7568902373313904, + "num_tokens": 822792416.0, + "step": 31028 + }, + { + "epoch": 2.5586707347241693, + "grad_norm": 1.9914674758911133, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7189595699310303, + "num_tokens": 822827993.0, + "step": 31029 + }, + { + "epoch": 2.5587531953492206, + "grad_norm": 2.1071488857269287, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7216481566429138, + "num_tokens": 822859495.0, + "step": 31030 + }, + { + "epoch": 2.5588356559742724, + "grad_norm": 2.1266980171203613, + "learning_rate": 1e-06, + "loss": 0.5209, + "mean_token_accuracy": 0.8221267461776733, + "num_tokens": 822887117.0, + "step": 31031 + }, + { + "epoch": 2.5589181165993238, + "grad_norm": 2.2926254272460938, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7592381238937378, + "num_tokens": 822914587.0, + "step": 31032 + }, + { + "epoch": 2.559000577224375, + "grad_norm": 1.9789007902145386, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7509502172470093, + "num_tokens": 822949352.0, + "step": 31033 + }, + { + "epoch": 2.559083037849427, + "grad_norm": 2.136479616165161, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7527033090591431, + "num_tokens": 822978403.0, + "step": 31034 + }, + { + "epoch": 2.5591654984744783, + "grad_norm": 1.9486128091812134, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.777572512626648, + "num_tokens": 823010010.0, + "step": 31035 + }, + { + "epoch": 2.55924795909953, + "grad_norm": 1.8903306722640991, + "learning_rate": 1e-06, + "loss": 0.5776, + "mean_token_accuracy": 0.8068766593933105, + "num_tokens": 823042323.0, + "step": 31036 + }, + { + "epoch": 2.5593304197245814, + "grad_norm": 2.2430617809295654, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.7580224275588989, + "num_tokens": 823072204.0, + "step": 31037 + }, + { + "epoch": 2.559412880349633, + "grad_norm": 2.1079792976379395, + "learning_rate": 1e-06, + "loss": 0.5636, + "mean_token_accuracy": 0.8116059303283691, + "num_tokens": 823100097.0, + "step": 31038 + }, + { + "epoch": 2.5594953409746846, + "grad_norm": 2.2220115661621094, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7746278047561646, + "num_tokens": 823125828.0, + "step": 31039 + }, + { + "epoch": 2.559577801599736, + "grad_norm": 2.020718574523926, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7798304557800293, + "num_tokens": 823157375.0, + "step": 31040 + }, + { + "epoch": 2.5596602622247877, + "grad_norm": 2.0448648929595947, + "learning_rate": 1e-06, + "loss": 0.5015, + "mean_token_accuracy": 0.8285050392150879, + "num_tokens": 823186437.0, + "step": 31041 + }, + { + "epoch": 2.559742722849839, + "grad_norm": 2.1488945484161377, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7409664392471313, + "num_tokens": 823214564.0, + "step": 31042 + }, + { + "epoch": 2.559825183474891, + "grad_norm": 2.2324910163879395, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7680356502532959, + "num_tokens": 823241667.0, + "step": 31043 + }, + { + "epoch": 2.559907644099942, + "grad_norm": 2.0385854244232178, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7777305841445923, + "num_tokens": 823272801.0, + "step": 31044 + }, + { + "epoch": 2.559990104724994, + "grad_norm": 2.146928310394287, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7628399133682251, + "num_tokens": 823303711.0, + "step": 31045 + }, + { + "epoch": 2.5600725653500453, + "grad_norm": 2.3191847801208496, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7900997996330261, + "num_tokens": 823330997.0, + "step": 31046 + }, + { + "epoch": 2.5601550259750967, + "grad_norm": 2.0964748859405518, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.7981936931610107, + "num_tokens": 823360660.0, + "step": 31047 + }, + { + "epoch": 2.5602374866001485, + "grad_norm": 2.0879061222076416, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.760592520236969, + "num_tokens": 823390912.0, + "step": 31048 + }, + { + "epoch": 2.5603199472252, + "grad_norm": 2.052283525466919, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.7806316614151001, + "num_tokens": 823420509.0, + "step": 31049 + }, + { + "epoch": 2.5604024078502516, + "grad_norm": 1.9082006216049194, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7674878239631653, + "num_tokens": 823456155.0, + "step": 31050 + }, + { + "epoch": 2.560484868475303, + "grad_norm": 2.2248873710632324, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7628780603408813, + "num_tokens": 823482666.0, + "step": 31051 + }, + { + "epoch": 2.560567329100355, + "grad_norm": 2.2227628231048584, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7322726845741272, + "num_tokens": 823511442.0, + "step": 31052 + }, + { + "epoch": 2.560649789725406, + "grad_norm": 2.0684499740600586, + "learning_rate": 1e-06, + "loss": 0.7614, + "mean_token_accuracy": 0.7541457414627075, + "num_tokens": 823544271.0, + "step": 31053 + }, + { + "epoch": 2.5607322503504575, + "grad_norm": 2.3744027614593506, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7623960375785828, + "num_tokens": 823567276.0, + "step": 31054 + }, + { + "epoch": 2.5608147109755093, + "grad_norm": 2.313076972961426, + "learning_rate": 1e-06, + "loss": 0.7563, + "mean_token_accuracy": 0.756050705909729, + "num_tokens": 823595441.0, + "step": 31055 + }, + { + "epoch": 2.5608971716005606, + "grad_norm": 2.1050937175750732, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7869435548782349, + "num_tokens": 823625925.0, + "step": 31056 + }, + { + "epoch": 2.5609796322256124, + "grad_norm": 1.983980417251587, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7851046919822693, + "num_tokens": 823655123.0, + "step": 31057 + }, + { + "epoch": 2.5610620928506638, + "grad_norm": 2.1369919776916504, + "learning_rate": 1e-06, + "loss": 0.5905, + "mean_token_accuracy": 0.7999889254570007, + "num_tokens": 823682768.0, + "step": 31058 + }, + { + "epoch": 2.5611445534757156, + "grad_norm": 2.2276089191436768, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7726222276687622, + "num_tokens": 823713049.0, + "step": 31059 + }, + { + "epoch": 2.561227014100767, + "grad_norm": 2.3640153408050537, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7600627541542053, + "num_tokens": 823738072.0, + "step": 31060 + }, + { + "epoch": 2.5613094747258183, + "grad_norm": 2.2488386631011963, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7776013612747192, + "num_tokens": 823764934.0, + "step": 31061 + }, + { + "epoch": 2.56139193535087, + "grad_norm": 2.1876251697540283, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7681753635406494, + "num_tokens": 823792747.0, + "step": 31062 + }, + { + "epoch": 2.5614743959759214, + "grad_norm": 2.015127420425415, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7859017848968506, + "num_tokens": 823823374.0, + "step": 31063 + }, + { + "epoch": 2.5615568566009728, + "grad_norm": 1.8447262048721313, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7445707321166992, + "num_tokens": 823861620.0, + "step": 31064 + }, + { + "epoch": 2.5616393172260246, + "grad_norm": 2.3230037689208984, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7262638211250305, + "num_tokens": 823889724.0, + "step": 31065 + }, + { + "epoch": 2.5617217778510764, + "grad_norm": 1.9782558679580688, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7302707433700562, + "num_tokens": 823925428.0, + "step": 31066 + }, + { + "epoch": 2.5618042384761277, + "grad_norm": 1.9711952209472656, + "learning_rate": 1e-06, + "loss": 0.5808, + "mean_token_accuracy": 0.8019370436668396, + "num_tokens": 823957732.0, + "step": 31067 + }, + { + "epoch": 2.561886699101179, + "grad_norm": 2.0326738357543945, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.7816498875617981, + "num_tokens": 823991020.0, + "step": 31068 + }, + { + "epoch": 2.561969159726231, + "grad_norm": 2.249211072921753, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7457722425460815, + "num_tokens": 824017952.0, + "step": 31069 + }, + { + "epoch": 2.562051620351282, + "grad_norm": 2.0385851860046387, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7894136905670166, + "num_tokens": 824051035.0, + "step": 31070 + }, + { + "epoch": 2.5621340809763335, + "grad_norm": 1.7441389560699463, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.7700150609016418, + "num_tokens": 824088707.0, + "step": 31071 + }, + { + "epoch": 2.5622165416013853, + "grad_norm": 1.9764573574066162, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.7950034141540527, + "num_tokens": 824119102.0, + "step": 31072 + }, + { + "epoch": 2.562299002226437, + "grad_norm": 1.9178297519683838, + "learning_rate": 1e-06, + "loss": 0.6066, + "mean_token_accuracy": 0.7989040613174438, + "num_tokens": 824151512.0, + "step": 31073 + }, + { + "epoch": 2.5623814628514885, + "grad_norm": 1.999778151512146, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7603017091751099, + "num_tokens": 824185983.0, + "step": 31074 + }, + { + "epoch": 2.56246392347654, + "grad_norm": 2.1632518768310547, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7352961301803589, + "num_tokens": 824214574.0, + "step": 31075 + }, + { + "epoch": 2.5625463841015916, + "grad_norm": 2.169603109359741, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7778542637825012, + "num_tokens": 824242720.0, + "step": 31076 + }, + { + "epoch": 2.562628844726643, + "grad_norm": 2.3456544876098633, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7419351935386658, + "num_tokens": 824269274.0, + "step": 31077 + }, + { + "epoch": 2.5627113053516943, + "grad_norm": 1.893243670463562, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.7908629179000854, + "num_tokens": 824305911.0, + "step": 31078 + }, + { + "epoch": 2.562793765976746, + "grad_norm": 1.8941500186920166, + "learning_rate": 1e-06, + "loss": 0.6986, + "mean_token_accuracy": 0.7644269466400146, + "num_tokens": 824340727.0, + "step": 31079 + }, + { + "epoch": 2.5628762266017975, + "grad_norm": 2.1157407760620117, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.7857140302658081, + "num_tokens": 824370369.0, + "step": 31080 + }, + { + "epoch": 2.5629586872268493, + "grad_norm": 2.5198981761932373, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7530349493026733, + "num_tokens": 824391929.0, + "step": 31081 + }, + { + "epoch": 2.5630411478519006, + "grad_norm": 1.9982237815856934, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7800532579421997, + "num_tokens": 824421117.0, + "step": 31082 + }, + { + "epoch": 2.5631236084769524, + "grad_norm": 2.3509206771850586, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7932195067405701, + "num_tokens": 824445501.0, + "step": 31083 + }, + { + "epoch": 2.5632060691020038, + "grad_norm": 2.4017210006713867, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.7744672894477844, + "num_tokens": 824469553.0, + "step": 31084 + }, + { + "epoch": 2.563288529727055, + "grad_norm": 1.8469032049179077, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7614059448242188, + "num_tokens": 824504501.0, + "step": 31085 + }, + { + "epoch": 2.563370990352107, + "grad_norm": 2.0671350955963135, + "learning_rate": 1e-06, + "loss": 0.7269, + "mean_token_accuracy": 0.7753914594650269, + "num_tokens": 824536385.0, + "step": 31086 + }, + { + "epoch": 2.5634534509771583, + "grad_norm": 1.9242501258850098, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7423310875892639, + "num_tokens": 824573120.0, + "step": 31087 + }, + { + "epoch": 2.56353591160221, + "grad_norm": 2.574906587600708, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7268822193145752, + "num_tokens": 824595993.0, + "step": 31088 + }, + { + "epoch": 2.5636183722272614, + "grad_norm": 2.271073341369629, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7701213955879211, + "num_tokens": 824622871.0, + "step": 31089 + }, + { + "epoch": 2.563700832852313, + "grad_norm": 2.065814733505249, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7610218524932861, + "num_tokens": 824656675.0, + "step": 31090 + }, + { + "epoch": 2.5637832934773646, + "grad_norm": 2.0787360668182373, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7715764045715332, + "num_tokens": 824689129.0, + "step": 31091 + }, + { + "epoch": 2.563865754102416, + "grad_norm": 2.048543930053711, + "learning_rate": 1e-06, + "loss": 0.6054, + "mean_token_accuracy": 0.8047739267349243, + "num_tokens": 824719118.0, + "step": 31092 + }, + { + "epoch": 2.5639482147274677, + "grad_norm": 2.2131049633026123, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.7936862707138062, + "num_tokens": 824745398.0, + "step": 31093 + }, + { + "epoch": 2.564030675352519, + "grad_norm": 1.9516055583953857, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.7966393232345581, + "num_tokens": 824778737.0, + "step": 31094 + }, + { + "epoch": 2.564113135977571, + "grad_norm": 2.1519155502319336, + "learning_rate": 1e-06, + "loss": 0.5642, + "mean_token_accuracy": 0.8038822412490845, + "num_tokens": 824804953.0, + "step": 31095 + }, + { + "epoch": 2.564195596602622, + "grad_norm": 2.1667983531951904, + "learning_rate": 1e-06, + "loss": 0.6007, + "mean_token_accuracy": 0.7980948686599731, + "num_tokens": 824831702.0, + "step": 31096 + }, + { + "epoch": 2.564278057227674, + "grad_norm": 2.034942626953125, + "learning_rate": 1e-06, + "loss": 0.4984, + "mean_token_accuracy": 0.8195720911026001, + "num_tokens": 824859951.0, + "step": 31097 + }, + { + "epoch": 2.5643605178527253, + "grad_norm": 2.153045892715454, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7623199224472046, + "num_tokens": 824888707.0, + "step": 31098 + }, + { + "epoch": 2.5644429784777767, + "grad_norm": 1.8386402130126953, + "learning_rate": 1e-06, + "loss": 0.6129, + "mean_token_accuracy": 0.8033676743507385, + "num_tokens": 824924418.0, + "step": 31099 + }, + { + "epoch": 2.5645254391028285, + "grad_norm": 2.124126672744751, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.7938094139099121, + "num_tokens": 824951263.0, + "step": 31100 + }, + { + "epoch": 2.56460789972788, + "grad_norm": 1.9958938360214233, + "learning_rate": 1e-06, + "loss": 0.6405, + "mean_token_accuracy": 0.7932819724082947, + "num_tokens": 824983548.0, + "step": 31101 + }, + { + "epoch": 2.5646903603529316, + "grad_norm": 2.353905200958252, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7296937704086304, + "num_tokens": 825013568.0, + "step": 31102 + }, + { + "epoch": 2.564772820977983, + "grad_norm": 2.2677927017211914, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.76753830909729, + "num_tokens": 825041004.0, + "step": 31103 + }, + { + "epoch": 2.5648552816030348, + "grad_norm": 2.2701401710510254, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7622226476669312, + "num_tokens": 825069059.0, + "step": 31104 + }, + { + "epoch": 2.564937742228086, + "grad_norm": 2.32800555229187, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.736261248588562, + "num_tokens": 825093767.0, + "step": 31105 + }, + { + "epoch": 2.5650202028531375, + "grad_norm": 2.1945137977600098, + "learning_rate": 1e-06, + "loss": 0.5532, + "mean_token_accuracy": 0.8132250308990479, + "num_tokens": 825118794.0, + "step": 31106 + }, + { + "epoch": 2.5651026634781893, + "grad_norm": 2.3635780811309814, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7553095817565918, + "num_tokens": 825144899.0, + "step": 31107 + }, + { + "epoch": 2.5651851241032406, + "grad_norm": 2.3245086669921875, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7678932547569275, + "num_tokens": 825169245.0, + "step": 31108 + }, + { + "epoch": 2.565267584728292, + "grad_norm": 2.2077269554138184, + "learning_rate": 1e-06, + "loss": 0.768, + "mean_token_accuracy": 0.7489778995513916, + "num_tokens": 825198210.0, + "step": 31109 + }, + { + "epoch": 2.5653500453533438, + "grad_norm": 1.9816642999649048, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7252153158187866, + "num_tokens": 825231305.0, + "step": 31110 + }, + { + "epoch": 2.5654325059783956, + "grad_norm": 1.906455397605896, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.7840495705604553, + "num_tokens": 825262135.0, + "step": 31111 + }, + { + "epoch": 2.565514966603447, + "grad_norm": 1.9098527431488037, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.787854790687561, + "num_tokens": 825299308.0, + "step": 31112 + }, + { + "epoch": 2.5655974272284983, + "grad_norm": 2.1945993900299072, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7613377571105957, + "num_tokens": 825327372.0, + "step": 31113 + }, + { + "epoch": 2.56567988785355, + "grad_norm": 2.3218183517456055, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.7872588634490967, + "num_tokens": 825352466.0, + "step": 31114 + }, + { + "epoch": 2.5657623484786014, + "grad_norm": 2.0776524543762207, + "learning_rate": 1e-06, + "loss": 0.5334, + "mean_token_accuracy": 0.8059889078140259, + "num_tokens": 825382398.0, + "step": 31115 + }, + { + "epoch": 2.5658448091036528, + "grad_norm": 1.8870643377304077, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.7973737120628357, + "num_tokens": 825414022.0, + "step": 31116 + }, + { + "epoch": 2.5659272697287046, + "grad_norm": 2.2521839141845703, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7690240740776062, + "num_tokens": 825441589.0, + "step": 31117 + }, + { + "epoch": 2.5660097303537563, + "grad_norm": 1.9268890619277954, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7753481268882751, + "num_tokens": 825475836.0, + "step": 31118 + }, + { + "epoch": 2.5660921909788077, + "grad_norm": 1.832828164100647, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.7918429374694824, + "num_tokens": 825512540.0, + "step": 31119 + }, + { + "epoch": 2.566174651603859, + "grad_norm": 2.1050169467926025, + "learning_rate": 1e-06, + "loss": 0.5656, + "mean_token_accuracy": 0.8023058176040649, + "num_tokens": 825538977.0, + "step": 31120 + }, + { + "epoch": 2.566257112228911, + "grad_norm": 2.257462501525879, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7377898693084717, + "num_tokens": 825566312.0, + "step": 31121 + }, + { + "epoch": 2.566339572853962, + "grad_norm": 1.993417501449585, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7397558689117432, + "num_tokens": 825600603.0, + "step": 31122 + }, + { + "epoch": 2.5664220334790135, + "grad_norm": 2.2760818004608154, + "learning_rate": 1e-06, + "loss": 0.6028, + "mean_token_accuracy": 0.7933154702186584, + "num_tokens": 825624225.0, + "step": 31123 + }, + { + "epoch": 2.5665044941040653, + "grad_norm": 2.168091297149658, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.772588312625885, + "num_tokens": 825654073.0, + "step": 31124 + }, + { + "epoch": 2.5665869547291167, + "grad_norm": 2.1779565811157227, + "learning_rate": 1e-06, + "loss": 0.6058, + "mean_token_accuracy": 0.7963346242904663, + "num_tokens": 825683376.0, + "step": 31125 + }, + { + "epoch": 2.5666694153541685, + "grad_norm": 1.958187222480774, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.7668060660362244, + "num_tokens": 825717133.0, + "step": 31126 + }, + { + "epoch": 2.56675187597922, + "grad_norm": 2.1843559741973877, + "learning_rate": 1e-06, + "loss": 0.5947, + "mean_token_accuracy": 0.8065441250801086, + "num_tokens": 825743163.0, + "step": 31127 + }, + { + "epoch": 2.5668343366042716, + "grad_norm": 2.390066385269165, + "learning_rate": 1e-06, + "loss": 0.5964, + "mean_token_accuracy": 0.8062669634819031, + "num_tokens": 825766316.0, + "step": 31128 + }, + { + "epoch": 2.566916797229323, + "grad_norm": 2.149580240249634, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.7572937607765198, + "num_tokens": 825795737.0, + "step": 31129 + }, + { + "epoch": 2.5669992578543743, + "grad_norm": 2.0977489948272705, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7691025137901306, + "num_tokens": 825825483.0, + "step": 31130 + }, + { + "epoch": 2.567081718479426, + "grad_norm": 1.9915059804916382, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7268896102905273, + "num_tokens": 825856271.0, + "step": 31131 + }, + { + "epoch": 2.5671641791044775, + "grad_norm": 1.9707013368606567, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.786638617515564, + "num_tokens": 825890793.0, + "step": 31132 + }, + { + "epoch": 2.5672466397295293, + "grad_norm": 2.2369306087493896, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7563419342041016, + "num_tokens": 825917857.0, + "step": 31133 + }, + { + "epoch": 2.5673291003545806, + "grad_norm": 2.2712903022766113, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7772332429885864, + "num_tokens": 825944535.0, + "step": 31134 + }, + { + "epoch": 2.5674115609796324, + "grad_norm": 2.390913963317871, + "learning_rate": 1e-06, + "loss": 0.5319, + "mean_token_accuracy": 0.8236980438232422, + "num_tokens": 825967477.0, + "step": 31135 + }, + { + "epoch": 2.5674940216046838, + "grad_norm": 1.894160509109497, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7540218830108643, + "num_tokens": 826003178.0, + "step": 31136 + }, + { + "epoch": 2.567576482229735, + "grad_norm": 1.8778473138809204, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7392311096191406, + "num_tokens": 826040778.0, + "step": 31137 + }, + { + "epoch": 2.567658942854787, + "grad_norm": 2.2438392639160156, + "learning_rate": 1e-06, + "loss": 0.5932, + "mean_token_accuracy": 0.8072485327720642, + "num_tokens": 826063667.0, + "step": 31138 + }, + { + "epoch": 2.5677414034798383, + "grad_norm": 2.2252230644226074, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.741572380065918, + "num_tokens": 826089839.0, + "step": 31139 + }, + { + "epoch": 2.56782386410489, + "grad_norm": 2.503288507461548, + "learning_rate": 1e-06, + "loss": 0.612, + "mean_token_accuracy": 0.7859671115875244, + "num_tokens": 826111923.0, + "step": 31140 + }, + { + "epoch": 2.5679063247299414, + "grad_norm": 1.9352569580078125, + "learning_rate": 1e-06, + "loss": 0.5479, + "mean_token_accuracy": 0.8076481819152832, + "num_tokens": 826144256.0, + "step": 31141 + }, + { + "epoch": 2.567988785354993, + "grad_norm": 2.250605583190918, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7703505754470825, + "num_tokens": 826170977.0, + "step": 31142 + }, + { + "epoch": 2.5680712459800445, + "grad_norm": 2.1808645725250244, + "learning_rate": 1e-06, + "loss": 0.5651, + "mean_token_accuracy": 0.8025400638580322, + "num_tokens": 826197579.0, + "step": 31143 + }, + { + "epoch": 2.568153706605096, + "grad_norm": 2.181950092315674, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.7786650657653809, + "num_tokens": 826223549.0, + "step": 31144 + }, + { + "epoch": 2.5682361672301477, + "grad_norm": 2.222177505493164, + "learning_rate": 1e-06, + "loss": 0.7821, + "mean_token_accuracy": 0.7509473562240601, + "num_tokens": 826252879.0, + "step": 31145 + }, + { + "epoch": 2.568318627855199, + "grad_norm": 2.1579432487487793, + "learning_rate": 1e-06, + "loss": 0.521, + "mean_token_accuracy": 0.8170520663261414, + "num_tokens": 826280321.0, + "step": 31146 + }, + { + "epoch": 2.568401088480251, + "grad_norm": 2.1360509395599365, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.7950488328933716, + "num_tokens": 826310076.0, + "step": 31147 + }, + { + "epoch": 2.568483549105302, + "grad_norm": 2.2696030139923096, + "learning_rate": 1e-06, + "loss": 0.6917, + "mean_token_accuracy": 0.771632194519043, + "num_tokens": 826336458.0, + "step": 31148 + }, + { + "epoch": 2.568566009730354, + "grad_norm": 2.1162643432617188, + "learning_rate": 1e-06, + "loss": 0.573, + "mean_token_accuracy": 0.8030931949615479, + "num_tokens": 826366853.0, + "step": 31149 + }, + { + "epoch": 2.5686484703554053, + "grad_norm": 2.1540277004241943, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.775054931640625, + "num_tokens": 826395957.0, + "step": 31150 + }, + { + "epoch": 2.5687309309804567, + "grad_norm": 2.0090155601501465, + "learning_rate": 1e-06, + "loss": 0.702, + "mean_token_accuracy": 0.7787654399871826, + "num_tokens": 826427940.0, + "step": 31151 + }, + { + "epoch": 2.5688133916055085, + "grad_norm": 2.0475969314575195, + "learning_rate": 1e-06, + "loss": 0.6051, + "mean_token_accuracy": 0.800519585609436, + "num_tokens": 826457042.0, + "step": 31152 + }, + { + "epoch": 2.56889585223056, + "grad_norm": 1.8368546962738037, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7619102597236633, + "num_tokens": 826493817.0, + "step": 31153 + }, + { + "epoch": 2.568978312855611, + "grad_norm": 2.7081828117370605, + "learning_rate": 1e-06, + "loss": 0.7659, + "mean_token_accuracy": 0.7562836408615112, + "num_tokens": 826514610.0, + "step": 31154 + }, + { + "epoch": 2.569060773480663, + "grad_norm": 2.24590802192688, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7849115133285522, + "num_tokens": 826539850.0, + "step": 31155 + }, + { + "epoch": 2.5691432341057148, + "grad_norm": 2.119755983352661, + "learning_rate": 1e-06, + "loss": 0.5927, + "mean_token_accuracy": 0.7948669195175171, + "num_tokens": 826568174.0, + "step": 31156 + }, + { + "epoch": 2.569225694730766, + "grad_norm": 2.2300310134887695, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7378396391868591, + "num_tokens": 826596977.0, + "step": 31157 + }, + { + "epoch": 2.5693081553558175, + "grad_norm": 2.205538749694824, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7734760046005249, + "num_tokens": 826623742.0, + "step": 31158 + }, + { + "epoch": 2.5693906159808693, + "grad_norm": 2.3025832176208496, + "learning_rate": 1e-06, + "loss": 0.7488, + "mean_token_accuracy": 0.7573397159576416, + "num_tokens": 826650376.0, + "step": 31159 + }, + { + "epoch": 2.5694730766059206, + "grad_norm": 1.8127690553665161, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.7928144931793213, + "num_tokens": 826687140.0, + "step": 31160 + }, + { + "epoch": 2.569555537230972, + "grad_norm": 2.551429510116577, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.7865086793899536, + "num_tokens": 826708174.0, + "step": 31161 + }, + { + "epoch": 2.5696379978560238, + "grad_norm": 2.3864855766296387, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7520042061805725, + "num_tokens": 826736013.0, + "step": 31162 + }, + { + "epoch": 2.5697204584810756, + "grad_norm": 1.8743975162506104, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.7979599237442017, + "num_tokens": 826770126.0, + "step": 31163 + }, + { + "epoch": 2.569802919106127, + "grad_norm": 1.959276795387268, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7388870120048523, + "num_tokens": 826807125.0, + "step": 31164 + }, + { + "epoch": 2.5698853797311783, + "grad_norm": 2.3556900024414062, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.7762285470962524, + "num_tokens": 826830590.0, + "step": 31165 + }, + { + "epoch": 2.56996784035623, + "grad_norm": 2.2170753479003906, + "learning_rate": 1e-06, + "loss": 0.6122, + "mean_token_accuracy": 0.793130099773407, + "num_tokens": 826855983.0, + "step": 31166 + }, + { + "epoch": 2.5700503009812814, + "grad_norm": 2.1602299213409424, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7782109975814819, + "num_tokens": 826886026.0, + "step": 31167 + }, + { + "epoch": 2.5701327616063327, + "grad_norm": 2.308474540710449, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7468134164810181, + "num_tokens": 826912639.0, + "step": 31168 + }, + { + "epoch": 2.5702152222313845, + "grad_norm": 1.8166536092758179, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7835354804992676, + "num_tokens": 826948417.0, + "step": 31169 + }, + { + "epoch": 2.570297682856436, + "grad_norm": 2.0839626789093018, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7229238152503967, + "num_tokens": 826980277.0, + "step": 31170 + }, + { + "epoch": 2.5703801434814877, + "grad_norm": 2.4145121574401855, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7646302580833435, + "num_tokens": 827004787.0, + "step": 31171 + }, + { + "epoch": 2.570462604106539, + "grad_norm": 2.316087007522583, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.774428129196167, + "num_tokens": 827029249.0, + "step": 31172 + }, + { + "epoch": 2.570545064731591, + "grad_norm": 2.001333475112915, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7752200365066528, + "num_tokens": 827060578.0, + "step": 31173 + }, + { + "epoch": 2.570627525356642, + "grad_norm": 1.9870061874389648, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7742130756378174, + "num_tokens": 827090313.0, + "step": 31174 + }, + { + "epoch": 2.5707099859816935, + "grad_norm": 2.1715338230133057, + "learning_rate": 1e-06, + "loss": 0.5715, + "mean_token_accuracy": 0.805167019367218, + "num_tokens": 827120856.0, + "step": 31175 + }, + { + "epoch": 2.5707924466067453, + "grad_norm": 1.8215619325637817, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7796913385391235, + "num_tokens": 827157023.0, + "step": 31176 + }, + { + "epoch": 2.5708749072317967, + "grad_norm": 2.195924758911133, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.7770435214042664, + "num_tokens": 827186769.0, + "step": 31177 + }, + { + "epoch": 2.5709573678568485, + "grad_norm": 2.16449236869812, + "learning_rate": 1e-06, + "loss": 0.5376, + "mean_token_accuracy": 0.8215888142585754, + "num_tokens": 827217359.0, + "step": 31178 + }, + { + "epoch": 2.5710398284819, + "grad_norm": 2.0574543476104736, + "learning_rate": 1e-06, + "loss": 0.5875, + "mean_token_accuracy": 0.8007688522338867, + "num_tokens": 827249791.0, + "step": 31179 + }, + { + "epoch": 2.5711222891069516, + "grad_norm": 2.183960199356079, + "learning_rate": 1e-06, + "loss": 0.6784, + "mean_token_accuracy": 0.768744945526123, + "num_tokens": 827278385.0, + "step": 31180 + }, + { + "epoch": 2.571204749732003, + "grad_norm": 2.6857547760009766, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.7929667234420776, + "num_tokens": 827302299.0, + "step": 31181 + }, + { + "epoch": 2.5712872103570543, + "grad_norm": 2.2624473571777344, + "learning_rate": 1e-06, + "loss": 0.7412, + "mean_token_accuracy": 0.7597859501838684, + "num_tokens": 827328519.0, + "step": 31182 + }, + { + "epoch": 2.571369670982106, + "grad_norm": 1.8648052215576172, + "learning_rate": 1e-06, + "loss": 0.5043, + "mean_token_accuracy": 0.8307915925979614, + "num_tokens": 827362394.0, + "step": 31183 + }, + { + "epoch": 2.5714521316071575, + "grad_norm": 2.2968122959136963, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.7816694974899292, + "num_tokens": 827387608.0, + "step": 31184 + }, + { + "epoch": 2.5715345922322093, + "grad_norm": 2.257993459701538, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7750788331031799, + "num_tokens": 827413382.0, + "step": 31185 + }, + { + "epoch": 2.5716170528572606, + "grad_norm": 2.122021436691284, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7785813212394714, + "num_tokens": 827441528.0, + "step": 31186 + }, + { + "epoch": 2.5716995134823124, + "grad_norm": 2.303431510925293, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7862179279327393, + "num_tokens": 827467186.0, + "step": 31187 + }, + { + "epoch": 2.5717819741073638, + "grad_norm": 2.0927488803863525, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.7870783805847168, + "num_tokens": 827497784.0, + "step": 31188 + }, + { + "epoch": 2.571864434732415, + "grad_norm": 2.1052441596984863, + "learning_rate": 1e-06, + "loss": 0.7765, + "mean_token_accuracy": 0.7507244944572449, + "num_tokens": 827529664.0, + "step": 31189 + }, + { + "epoch": 2.571946895357467, + "grad_norm": 1.843116283416748, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.7894428968429565, + "num_tokens": 827568126.0, + "step": 31190 + }, + { + "epoch": 2.5720293559825183, + "grad_norm": 2.1413955688476562, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.7928930521011353, + "num_tokens": 827598834.0, + "step": 31191 + }, + { + "epoch": 2.57211181660757, + "grad_norm": 2.0573861598968506, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7588057518005371, + "num_tokens": 827627972.0, + "step": 31192 + }, + { + "epoch": 2.5721942772326214, + "grad_norm": 2.3764572143554688, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7596263289451599, + "num_tokens": 827654030.0, + "step": 31193 + }, + { + "epoch": 2.572276737857673, + "grad_norm": 2.0561699867248535, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7796072363853455, + "num_tokens": 827682508.0, + "step": 31194 + }, + { + "epoch": 2.5723591984827245, + "grad_norm": 2.317397117614746, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.7868686318397522, + "num_tokens": 827708409.0, + "step": 31195 + }, + { + "epoch": 2.572441659107776, + "grad_norm": 1.7531837224960327, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.8044714331626892, + "num_tokens": 827745412.0, + "step": 31196 + }, + { + "epoch": 2.5725241197328277, + "grad_norm": 1.7439576387405396, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7569872736930847, + "num_tokens": 827787484.0, + "step": 31197 + }, + { + "epoch": 2.572606580357879, + "grad_norm": 2.0628883838653564, + "learning_rate": 1e-06, + "loss": 0.5089, + "mean_token_accuracy": 0.8228369951248169, + "num_tokens": 827817990.0, + "step": 31198 + }, + { + "epoch": 2.5726890409829304, + "grad_norm": 2.286494731903076, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.8034529089927673, + "num_tokens": 827842524.0, + "step": 31199 + }, + { + "epoch": 2.572771501607982, + "grad_norm": 1.9525492191314697, + "learning_rate": 1e-06, + "loss": 0.7024, + "mean_token_accuracy": 0.769819438457489, + "num_tokens": 827876684.0, + "step": 31200 + }, + { + "epoch": 2.572853962233034, + "grad_norm": 2.334911823272705, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.770930290222168, + "num_tokens": 827902115.0, + "step": 31201 + }, + { + "epoch": 2.5729364228580853, + "grad_norm": 2.349473237991333, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.737761378288269, + "num_tokens": 827928816.0, + "step": 31202 + }, + { + "epoch": 2.5730188834831367, + "grad_norm": 2.31125545501709, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.778039276599884, + "num_tokens": 827955506.0, + "step": 31203 + }, + { + "epoch": 2.5731013441081885, + "grad_norm": 1.965531587600708, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7726813554763794, + "num_tokens": 827989339.0, + "step": 31204 + }, + { + "epoch": 2.57318380473324, + "grad_norm": 2.0947506427764893, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7710617780685425, + "num_tokens": 828019702.0, + "step": 31205 + }, + { + "epoch": 2.573266265358291, + "grad_norm": 2.090023994445801, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7369141578674316, + "num_tokens": 828049229.0, + "step": 31206 + }, + { + "epoch": 2.573348725983343, + "grad_norm": 1.9696626663208008, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.766632080078125, + "num_tokens": 828082236.0, + "step": 31207 + }, + { + "epoch": 2.5734311866083948, + "grad_norm": 2.034189462661743, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.780855655670166, + "num_tokens": 828111336.0, + "step": 31208 + }, + { + "epoch": 2.573513647233446, + "grad_norm": 1.80376398563385, + "learning_rate": 1e-06, + "loss": 0.606, + "mean_token_accuracy": 0.8184335231781006, + "num_tokens": 828147225.0, + "step": 31209 + }, + { + "epoch": 2.5735961078584975, + "grad_norm": 2.094452142715454, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.7883687019348145, + "num_tokens": 828177497.0, + "step": 31210 + }, + { + "epoch": 2.5736785684835493, + "grad_norm": 2.1849703788757324, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7712200880050659, + "num_tokens": 828204469.0, + "step": 31211 + }, + { + "epoch": 2.5737610291086006, + "grad_norm": 2.257861375808716, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7654692530632019, + "num_tokens": 828230020.0, + "step": 31212 + }, + { + "epoch": 2.573843489733652, + "grad_norm": 1.9672662019729614, + "learning_rate": 1e-06, + "loss": 0.4808, + "mean_token_accuracy": 0.8242092132568359, + "num_tokens": 828257260.0, + "step": 31213 + }, + { + "epoch": 2.5739259503587038, + "grad_norm": 1.9463030099868774, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7779995799064636, + "num_tokens": 828291127.0, + "step": 31214 + }, + { + "epoch": 2.574008410983755, + "grad_norm": 2.061796188354492, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7738616466522217, + "num_tokens": 828323669.0, + "step": 31215 + }, + { + "epoch": 2.574090871608807, + "grad_norm": 2.1627085208892822, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7376995086669922, + "num_tokens": 828357564.0, + "step": 31216 + }, + { + "epoch": 2.5741733322338582, + "grad_norm": 2.119860887527466, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7727895975112915, + "num_tokens": 828387997.0, + "step": 31217 + }, + { + "epoch": 2.57425579285891, + "grad_norm": 2.021052122116089, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.713748037815094, + "num_tokens": 828417963.0, + "step": 31218 + }, + { + "epoch": 2.5743382534839614, + "grad_norm": 2.3861560821533203, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7717714309692383, + "num_tokens": 828442422.0, + "step": 31219 + }, + { + "epoch": 2.5744207141090127, + "grad_norm": 2.24068284034729, + "learning_rate": 1e-06, + "loss": 0.5724, + "mean_token_accuracy": 0.7968453764915466, + "num_tokens": 828466750.0, + "step": 31220 + }, + { + "epoch": 2.5745031747340645, + "grad_norm": 1.986810326576233, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.7493698596954346, + "num_tokens": 828499997.0, + "step": 31221 + }, + { + "epoch": 2.574585635359116, + "grad_norm": 2.009169101715088, + "learning_rate": 1e-06, + "loss": 0.5239, + "mean_token_accuracy": 0.8006801009178162, + "num_tokens": 828528412.0, + "step": 31222 + }, + { + "epoch": 2.5746680959841677, + "grad_norm": 2.194230556488037, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7373417019844055, + "num_tokens": 828559078.0, + "step": 31223 + }, + { + "epoch": 2.574750556609219, + "grad_norm": 2.29848575592041, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7601655721664429, + "num_tokens": 828586192.0, + "step": 31224 + }, + { + "epoch": 2.574833017234271, + "grad_norm": 2.050611734390259, + "learning_rate": 1e-06, + "loss": 0.6297, + "mean_token_accuracy": 0.7928333282470703, + "num_tokens": 828617593.0, + "step": 31225 + }, + { + "epoch": 2.574915477859322, + "grad_norm": 2.226024866104126, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.7771217823028564, + "num_tokens": 828645020.0, + "step": 31226 + }, + { + "epoch": 2.5749979384843735, + "grad_norm": 1.826673150062561, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.7609817981719971, + "num_tokens": 828686298.0, + "step": 31227 + }, + { + "epoch": 2.5750803991094253, + "grad_norm": 1.9463160037994385, + "learning_rate": 1e-06, + "loss": 0.6151, + "mean_token_accuracy": 0.7863200306892395, + "num_tokens": 828718290.0, + "step": 31228 + }, + { + "epoch": 2.5751628597344767, + "grad_norm": 1.8482061624526978, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7765860557556152, + "num_tokens": 828752938.0, + "step": 31229 + }, + { + "epoch": 2.5752453203595285, + "grad_norm": 1.9927096366882324, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.7943700551986694, + "num_tokens": 828785469.0, + "step": 31230 + }, + { + "epoch": 2.57532778098458, + "grad_norm": 2.3618581295013428, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7788517475128174, + "num_tokens": 828810298.0, + "step": 31231 + }, + { + "epoch": 2.5754102416096316, + "grad_norm": 2.392258644104004, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7430459856987, + "num_tokens": 828835966.0, + "step": 31232 + }, + { + "epoch": 2.575492702234683, + "grad_norm": 2.3944711685180664, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7836966514587402, + "num_tokens": 828859966.0, + "step": 31233 + }, + { + "epoch": 2.5755751628597343, + "grad_norm": 2.2454609870910645, + "learning_rate": 1e-06, + "loss": 0.5616, + "mean_token_accuracy": 0.8087760210037231, + "num_tokens": 828887992.0, + "step": 31234 + }, + { + "epoch": 2.575657623484786, + "grad_norm": 2.327216863632202, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.7878402471542358, + "num_tokens": 828914356.0, + "step": 31235 + }, + { + "epoch": 2.5757400841098375, + "grad_norm": 2.034797430038452, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7694377303123474, + "num_tokens": 828946646.0, + "step": 31236 + }, + { + "epoch": 2.5758225447348893, + "grad_norm": 2.1924943923950195, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7775373458862305, + "num_tokens": 828974488.0, + "step": 31237 + }, + { + "epoch": 2.5759050053599406, + "grad_norm": 1.9828447103500366, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7363036274909973, + "num_tokens": 829009382.0, + "step": 31238 + }, + { + "epoch": 2.5759874659849924, + "grad_norm": 2.5582265853881836, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7511794567108154, + "num_tokens": 829033641.0, + "step": 31239 + }, + { + "epoch": 2.5760699266100437, + "grad_norm": 2.0516152381896973, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7717136144638062, + "num_tokens": 829067234.0, + "step": 31240 + }, + { + "epoch": 2.576152387235095, + "grad_norm": 2.3791098594665527, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.7818717956542969, + "num_tokens": 829096479.0, + "step": 31241 + }, + { + "epoch": 2.576234847860147, + "grad_norm": 2.3996737003326416, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7420061230659485, + "num_tokens": 829121228.0, + "step": 31242 + }, + { + "epoch": 2.5763173084851982, + "grad_norm": 2.842377185821533, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7703787088394165, + "num_tokens": 829139186.0, + "step": 31243 + }, + { + "epoch": 2.5763997691102496, + "grad_norm": 2.2391581535339355, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7726877331733704, + "num_tokens": 829166295.0, + "step": 31244 + }, + { + "epoch": 2.5764822297353014, + "grad_norm": 1.9795573949813843, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.7899012565612793, + "num_tokens": 829198074.0, + "step": 31245 + }, + { + "epoch": 2.576564690360353, + "grad_norm": 2.18221378326416, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.7908992767333984, + "num_tokens": 829224795.0, + "step": 31246 + }, + { + "epoch": 2.5766471509854045, + "grad_norm": 1.9513503313064575, + "learning_rate": 1e-06, + "loss": 0.5275, + "mean_token_accuracy": 0.8126844167709351, + "num_tokens": 829257922.0, + "step": 31247 + }, + { + "epoch": 2.576729611610456, + "grad_norm": 2.24129056930542, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7678061723709106, + "num_tokens": 829286458.0, + "step": 31248 + }, + { + "epoch": 2.5768120722355077, + "grad_norm": 2.0445194244384766, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7687395215034485, + "num_tokens": 829316773.0, + "step": 31249 + }, + { + "epoch": 2.576894532860559, + "grad_norm": 2.3414559364318848, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7707555294036865, + "num_tokens": 829345780.0, + "step": 31250 + }, + { + "epoch": 2.5769769934856104, + "grad_norm": 2.226386070251465, + "learning_rate": 1e-06, + "loss": 0.5778, + "mean_token_accuracy": 0.799835205078125, + "num_tokens": 829372639.0, + "step": 31251 + }, + { + "epoch": 2.577059454110662, + "grad_norm": 2.236802339553833, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.8000282049179077, + "num_tokens": 829401299.0, + "step": 31252 + }, + { + "epoch": 2.577141914735714, + "grad_norm": 1.9503705501556396, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7644950151443481, + "num_tokens": 829436536.0, + "step": 31253 + }, + { + "epoch": 2.5772243753607653, + "grad_norm": 1.9322683811187744, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7835296988487244, + "num_tokens": 829467207.0, + "step": 31254 + }, + { + "epoch": 2.5773068359858167, + "grad_norm": 2.0843660831451416, + "learning_rate": 1e-06, + "loss": 0.5883, + "mean_token_accuracy": 0.7956263422966003, + "num_tokens": 829495742.0, + "step": 31255 + }, + { + "epoch": 2.5773892966108685, + "grad_norm": 2.183229446411133, + "learning_rate": 1e-06, + "loss": 0.6236, + "mean_token_accuracy": 0.7901716232299805, + "num_tokens": 829522982.0, + "step": 31256 + }, + { + "epoch": 2.57747175723592, + "grad_norm": 2.022003173828125, + "learning_rate": 1e-06, + "loss": 0.539, + "mean_token_accuracy": 0.82243812084198, + "num_tokens": 829555969.0, + "step": 31257 + }, + { + "epoch": 2.577554217860971, + "grad_norm": 2.204193353652954, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7801055312156677, + "num_tokens": 829586002.0, + "step": 31258 + }, + { + "epoch": 2.577636678486023, + "grad_norm": 2.2418859004974365, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7884332537651062, + "num_tokens": 829615902.0, + "step": 31259 + }, + { + "epoch": 2.5777191391110743, + "grad_norm": 2.117521286010742, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7508103847503662, + "num_tokens": 829646465.0, + "step": 31260 + }, + { + "epoch": 2.577801599736126, + "grad_norm": 2.133852958679199, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7325506210327148, + "num_tokens": 829676336.0, + "step": 31261 + }, + { + "epoch": 2.5778840603611775, + "grad_norm": 2.068310499191284, + "learning_rate": 1e-06, + "loss": 0.5949, + "mean_token_accuracy": 0.8025398850440979, + "num_tokens": 829706644.0, + "step": 31262 + }, + { + "epoch": 2.5779665209862292, + "grad_norm": 2.027737855911255, + "learning_rate": 1e-06, + "loss": 0.7259, + "mean_token_accuracy": 0.7626999020576477, + "num_tokens": 829739783.0, + "step": 31263 + }, + { + "epoch": 2.5780489816112806, + "grad_norm": 2.0536513328552246, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.772111177444458, + "num_tokens": 829771205.0, + "step": 31264 + }, + { + "epoch": 2.578131442236332, + "grad_norm": 2.3280158042907715, + "learning_rate": 1e-06, + "loss": 0.5586, + "mean_token_accuracy": 0.8069357872009277, + "num_tokens": 829795918.0, + "step": 31265 + }, + { + "epoch": 2.5782139028613837, + "grad_norm": 2.1900837421417236, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7199516892433167, + "num_tokens": 829826678.0, + "step": 31266 + }, + { + "epoch": 2.578296363486435, + "grad_norm": 2.2544586658477783, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7744864225387573, + "num_tokens": 829853799.0, + "step": 31267 + }, + { + "epoch": 2.578378824111487, + "grad_norm": 2.431213617324829, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.7703421115875244, + "num_tokens": 829877594.0, + "step": 31268 + }, + { + "epoch": 2.5784612847365382, + "grad_norm": 1.9280591011047363, + "learning_rate": 1e-06, + "loss": 0.5223, + "mean_token_accuracy": 0.8031129240989685, + "num_tokens": 829908118.0, + "step": 31269 + }, + { + "epoch": 2.57854374536159, + "grad_norm": 1.9799654483795166, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.7913463115692139, + "num_tokens": 829937308.0, + "step": 31270 + }, + { + "epoch": 2.5786262059866414, + "grad_norm": 2.071903705596924, + "learning_rate": 1e-06, + "loss": 0.5691, + "mean_token_accuracy": 0.8045204877853394, + "num_tokens": 829969280.0, + "step": 31271 + }, + { + "epoch": 2.5787086666116927, + "grad_norm": 1.8533880710601807, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.7881545424461365, + "num_tokens": 830004491.0, + "step": 31272 + }, + { + "epoch": 2.5787911272367445, + "grad_norm": 2.173163414001465, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.758622407913208, + "num_tokens": 830033909.0, + "step": 31273 + }, + { + "epoch": 2.578873587861796, + "grad_norm": 1.9983410835266113, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7754956483840942, + "num_tokens": 830067352.0, + "step": 31274 + }, + { + "epoch": 2.5789560484868477, + "grad_norm": 1.9743688106536865, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7764924764633179, + "num_tokens": 830100529.0, + "step": 31275 + }, + { + "epoch": 2.579038509111899, + "grad_norm": 1.8641889095306396, + "learning_rate": 1e-06, + "loss": 0.5927, + "mean_token_accuracy": 0.7977741956710815, + "num_tokens": 830139981.0, + "step": 31276 + }, + { + "epoch": 2.579120969736951, + "grad_norm": 2.2244365215301514, + "learning_rate": 1e-06, + "loss": 0.5816, + "mean_token_accuracy": 0.805101215839386, + "num_tokens": 830171039.0, + "step": 31277 + }, + { + "epoch": 2.579203430362002, + "grad_norm": 2.1035871505737305, + "learning_rate": 1e-06, + "loss": 0.5599, + "mean_token_accuracy": 0.8163889050483704, + "num_tokens": 830199828.0, + "step": 31278 + }, + { + "epoch": 2.5792858909870535, + "grad_norm": 2.1343467235565186, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7675643563270569, + "num_tokens": 830234446.0, + "step": 31279 + }, + { + "epoch": 2.5793683516121053, + "grad_norm": 2.1751821041107178, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7287291288375854, + "num_tokens": 830263211.0, + "step": 31280 + }, + { + "epoch": 2.5794508122371567, + "grad_norm": 2.160003185272217, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7277160882949829, + "num_tokens": 830294400.0, + "step": 31281 + }, + { + "epoch": 2.5795332728622085, + "grad_norm": 2.488600730895996, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.767987847328186, + "num_tokens": 830317366.0, + "step": 31282 + }, + { + "epoch": 2.57961573348726, + "grad_norm": 2.188944101333618, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7611055374145508, + "num_tokens": 830347335.0, + "step": 31283 + }, + { + "epoch": 2.5796981941123116, + "grad_norm": 1.990110158920288, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7621358633041382, + "num_tokens": 830382173.0, + "step": 31284 + }, + { + "epoch": 2.579780654737363, + "grad_norm": 1.9635511636734009, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7672677636146545, + "num_tokens": 830415727.0, + "step": 31285 + }, + { + "epoch": 2.5798631153624143, + "grad_norm": 1.9577150344848633, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.7844411730766296, + "num_tokens": 830446099.0, + "step": 31286 + }, + { + "epoch": 2.579945575987466, + "grad_norm": 2.240882635116577, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7842651009559631, + "num_tokens": 830474253.0, + "step": 31287 + }, + { + "epoch": 2.5800280366125175, + "grad_norm": 2.0988147258758545, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7610630989074707, + "num_tokens": 830505911.0, + "step": 31288 + }, + { + "epoch": 2.580110497237569, + "grad_norm": 1.8947865962982178, + "learning_rate": 1e-06, + "loss": 0.6042, + "mean_token_accuracy": 0.8026516437530518, + "num_tokens": 830538803.0, + "step": 31289 + }, + { + "epoch": 2.5801929578626206, + "grad_norm": 2.2187650203704834, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7745218873023987, + "num_tokens": 830568075.0, + "step": 31290 + }, + { + "epoch": 2.5802754184876724, + "grad_norm": 2.2166025638580322, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.7952609658241272, + "num_tokens": 830594025.0, + "step": 31291 + }, + { + "epoch": 2.5803578791127237, + "grad_norm": 1.9207546710968018, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7508142590522766, + "num_tokens": 830626608.0, + "step": 31292 + }, + { + "epoch": 2.580440339737775, + "grad_norm": 2.007549285888672, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.771397590637207, + "num_tokens": 830659276.0, + "step": 31293 + }, + { + "epoch": 2.580522800362827, + "grad_norm": 1.9532058238983154, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7748910188674927, + "num_tokens": 830691314.0, + "step": 31294 + }, + { + "epoch": 2.5806052609878782, + "grad_norm": 1.9158488512039185, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7826086282730103, + "num_tokens": 830725397.0, + "step": 31295 + }, + { + "epoch": 2.5806877216129296, + "grad_norm": 1.8271629810333252, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7746126651763916, + "num_tokens": 830760762.0, + "step": 31296 + }, + { + "epoch": 2.5807701822379814, + "grad_norm": 2.1525678634643555, + "learning_rate": 1e-06, + "loss": 0.4999, + "mean_token_accuracy": 0.8326988220214844, + "num_tokens": 830786361.0, + "step": 31297 + }, + { + "epoch": 2.580852642863033, + "grad_norm": 2.1408047676086426, + "learning_rate": 1e-06, + "loss": 0.6103, + "mean_token_accuracy": 0.7922738790512085, + "num_tokens": 830815467.0, + "step": 31298 + }, + { + "epoch": 2.5809351034880845, + "grad_norm": 1.9386186599731445, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.771470844745636, + "num_tokens": 830852035.0, + "step": 31299 + }, + { + "epoch": 2.581017564113136, + "grad_norm": 1.9516212940216064, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7662615776062012, + "num_tokens": 830887667.0, + "step": 31300 + }, + { + "epoch": 2.5811000247381877, + "grad_norm": 2.157442331314087, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7742788195610046, + "num_tokens": 830915195.0, + "step": 31301 + }, + { + "epoch": 2.581182485363239, + "grad_norm": 2.1177186965942383, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7283542156219482, + "num_tokens": 830945021.0, + "step": 31302 + }, + { + "epoch": 2.5812649459882904, + "grad_norm": 2.2158751487731934, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7734310626983643, + "num_tokens": 830975431.0, + "step": 31303 + }, + { + "epoch": 2.581347406613342, + "grad_norm": 1.779194712638855, + "learning_rate": 1e-06, + "loss": 0.5158, + "mean_token_accuracy": 0.8214961290359497, + "num_tokens": 831007788.0, + "step": 31304 + }, + { + "epoch": 2.5814298672383935, + "grad_norm": 2.2827377319335938, + "learning_rate": 1e-06, + "loss": 0.684, + "mean_token_accuracy": 0.7714775800704956, + "num_tokens": 831032788.0, + "step": 31305 + }, + { + "epoch": 2.5815123278634453, + "grad_norm": 2.1600890159606934, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7646159529685974, + "num_tokens": 831062672.0, + "step": 31306 + }, + { + "epoch": 2.5815947884884967, + "grad_norm": 1.8713512420654297, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.760872483253479, + "num_tokens": 831096970.0, + "step": 31307 + }, + { + "epoch": 2.5816772491135485, + "grad_norm": 2.20613694190979, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.767254114151001, + "num_tokens": 831124974.0, + "step": 31308 + }, + { + "epoch": 2.5817597097386, + "grad_norm": 2.11443829536438, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7786784172058105, + "num_tokens": 831153348.0, + "step": 31309 + }, + { + "epoch": 2.581842170363651, + "grad_norm": 2.7331671714782715, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7612106800079346, + "num_tokens": 831174369.0, + "step": 31310 + }, + { + "epoch": 2.581924630988703, + "grad_norm": 2.021106243133545, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7515045404434204, + "num_tokens": 831207811.0, + "step": 31311 + }, + { + "epoch": 2.5820070916137543, + "grad_norm": 2.064469575881958, + "learning_rate": 1e-06, + "loss": 0.5774, + "mean_token_accuracy": 0.7980445623397827, + "num_tokens": 831236336.0, + "step": 31312 + }, + { + "epoch": 2.582089552238806, + "grad_norm": 2.159080743789673, + "learning_rate": 1e-06, + "loss": 0.6401, + "mean_token_accuracy": 0.7954456806182861, + "num_tokens": 831262789.0, + "step": 31313 + }, + { + "epoch": 2.5821720128638574, + "grad_norm": 2.1855506896972656, + "learning_rate": 1e-06, + "loss": 0.6, + "mean_token_accuracy": 0.7997143268585205, + "num_tokens": 831288573.0, + "step": 31314 + }, + { + "epoch": 2.5822544734889092, + "grad_norm": 2.290680408477783, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.7871313691139221, + "num_tokens": 831314952.0, + "step": 31315 + }, + { + "epoch": 2.5823369341139606, + "grad_norm": 1.9721028804779053, + "learning_rate": 1e-06, + "loss": 0.6051, + "mean_token_accuracy": 0.7983834147453308, + "num_tokens": 831347379.0, + "step": 31316 + }, + { + "epoch": 2.582419394739012, + "grad_norm": 1.9658931493759155, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7749611139297485, + "num_tokens": 831382696.0, + "step": 31317 + }, + { + "epoch": 2.5825018553640637, + "grad_norm": 1.9187854528427124, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7587073445320129, + "num_tokens": 831420304.0, + "step": 31318 + }, + { + "epoch": 2.582584315989115, + "grad_norm": 2.23065185546875, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7209440469741821, + "num_tokens": 831448729.0, + "step": 31319 + }, + { + "epoch": 2.582666776614167, + "grad_norm": 1.8896676301956177, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.7918888330459595, + "num_tokens": 831483152.0, + "step": 31320 + }, + { + "epoch": 2.5827492372392182, + "grad_norm": 2.1752376556396484, + "learning_rate": 1e-06, + "loss": 0.7724, + "mean_token_accuracy": 0.7531750202178955, + "num_tokens": 831513569.0, + "step": 31321 + }, + { + "epoch": 2.58283169786427, + "grad_norm": 1.9437321424484253, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.743637204170227, + "num_tokens": 831547564.0, + "step": 31322 + }, + { + "epoch": 2.5829141584893214, + "grad_norm": 2.0347907543182373, + "learning_rate": 1e-06, + "loss": 0.6625, + "mean_token_accuracy": 0.7887484431266785, + "num_tokens": 831578798.0, + "step": 31323 + }, + { + "epoch": 2.5829966191143727, + "grad_norm": 2.027144193649292, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7679154872894287, + "num_tokens": 831611728.0, + "step": 31324 + }, + { + "epoch": 2.5830790797394245, + "grad_norm": 2.20636248588562, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.7877744436264038, + "num_tokens": 831637584.0, + "step": 31325 + }, + { + "epoch": 2.583161540364476, + "grad_norm": 2.2938292026519775, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7780827283859253, + "num_tokens": 831663233.0, + "step": 31326 + }, + { + "epoch": 2.5832440009895277, + "grad_norm": 1.9020262956619263, + "learning_rate": 1e-06, + "loss": 0.5699, + "mean_token_accuracy": 0.8097347021102905, + "num_tokens": 831695454.0, + "step": 31327 + }, + { + "epoch": 2.583326461614579, + "grad_norm": 2.0990939140319824, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7557104825973511, + "num_tokens": 831725499.0, + "step": 31328 + }, + { + "epoch": 2.583408922239631, + "grad_norm": 2.004605531692505, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.7868317365646362, + "num_tokens": 831758362.0, + "step": 31329 + }, + { + "epoch": 2.583491382864682, + "grad_norm": 2.456882953643799, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7512702345848083, + "num_tokens": 831783285.0, + "step": 31330 + }, + { + "epoch": 2.5835738434897335, + "grad_norm": 2.058485984802246, + "learning_rate": 1e-06, + "loss": 0.675, + "mean_token_accuracy": 0.789756178855896, + "num_tokens": 831814479.0, + "step": 31331 + }, + { + "epoch": 2.5836563041147853, + "grad_norm": 2.403529405593872, + "learning_rate": 1e-06, + "loss": 0.5808, + "mean_token_accuracy": 0.8055336475372314, + "num_tokens": 831838339.0, + "step": 31332 + }, + { + "epoch": 2.5837387647398367, + "grad_norm": 2.032153367996216, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.8015040755271912, + "num_tokens": 831867583.0, + "step": 31333 + }, + { + "epoch": 2.583821225364888, + "grad_norm": 2.0868241786956787, + "learning_rate": 1e-06, + "loss": 0.6046, + "mean_token_accuracy": 0.7958252429962158, + "num_tokens": 831896720.0, + "step": 31334 + }, + { + "epoch": 2.58390368598994, + "grad_norm": 1.81987464427948, + "learning_rate": 1e-06, + "loss": 0.5587, + "mean_token_accuracy": 0.8166531324386597, + "num_tokens": 831931496.0, + "step": 31335 + }, + { + "epoch": 2.5839861466149916, + "grad_norm": 2.1214005947113037, + "learning_rate": 1e-06, + "loss": 0.5992, + "mean_token_accuracy": 0.8025860786437988, + "num_tokens": 831958154.0, + "step": 31336 + }, + { + "epoch": 2.584068607240043, + "grad_norm": 1.6939873695373535, + "learning_rate": 1e-06, + "loss": 0.6026, + "mean_token_accuracy": 0.8089360594749451, + "num_tokens": 831999104.0, + "step": 31337 + }, + { + "epoch": 2.5841510678650943, + "grad_norm": 2.0469958782196045, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.7803832292556763, + "num_tokens": 832030507.0, + "step": 31338 + }, + { + "epoch": 2.584233528490146, + "grad_norm": 2.3138628005981445, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.7942564487457275, + "num_tokens": 832059121.0, + "step": 31339 + }, + { + "epoch": 2.5843159891151974, + "grad_norm": 2.3073232173919678, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7669230699539185, + "num_tokens": 832085730.0, + "step": 31340 + }, + { + "epoch": 2.584398449740249, + "grad_norm": 2.088818073272705, + "learning_rate": 1e-06, + "loss": 0.518, + "mean_token_accuracy": 0.8211366534233093, + "num_tokens": 832115153.0, + "step": 31341 + }, + { + "epoch": 2.5844809103653006, + "grad_norm": 1.8769443035125732, + "learning_rate": 1e-06, + "loss": 0.5146, + "mean_token_accuracy": 0.8266342878341675, + "num_tokens": 832148356.0, + "step": 31342 + }, + { + "epoch": 2.5845633709903524, + "grad_norm": 2.1688780784606934, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.7948096990585327, + "num_tokens": 832178462.0, + "step": 31343 + }, + { + "epoch": 2.5846458316154037, + "grad_norm": 1.9737861156463623, + "learning_rate": 1e-06, + "loss": 0.6166, + "mean_token_accuracy": 0.8047367334365845, + "num_tokens": 832211633.0, + "step": 31344 + }, + { + "epoch": 2.584728292240455, + "grad_norm": 2.154984474182129, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7656115293502808, + "num_tokens": 832243213.0, + "step": 31345 + }, + { + "epoch": 2.584810752865507, + "grad_norm": 2.165191411972046, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7841140031814575, + "num_tokens": 832268440.0, + "step": 31346 + }, + { + "epoch": 2.5848932134905582, + "grad_norm": 1.9882338047027588, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7321107387542725, + "num_tokens": 832303234.0, + "step": 31347 + }, + { + "epoch": 2.5849756741156096, + "grad_norm": 2.531402587890625, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.767970621585846, + "num_tokens": 832324860.0, + "step": 31348 + }, + { + "epoch": 2.5850581347406614, + "grad_norm": 2.06046724319458, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7682451009750366, + "num_tokens": 832356405.0, + "step": 31349 + }, + { + "epoch": 2.5851405953657127, + "grad_norm": 2.036376476287842, + "learning_rate": 1e-06, + "loss": 0.5972, + "mean_token_accuracy": 0.793865442276001, + "num_tokens": 832384205.0, + "step": 31350 + }, + { + "epoch": 2.5852230559907645, + "grad_norm": 2.092860460281372, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.760005533695221, + "num_tokens": 832417997.0, + "step": 31351 + }, + { + "epoch": 2.585305516615816, + "grad_norm": 2.0210988521575928, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7771576046943665, + "num_tokens": 832451248.0, + "step": 31352 + }, + { + "epoch": 2.5853879772408677, + "grad_norm": 2.1411592960357666, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7684201002120972, + "num_tokens": 832481769.0, + "step": 31353 + }, + { + "epoch": 2.585470437865919, + "grad_norm": 2.337843418121338, + "learning_rate": 1e-06, + "loss": 0.5368, + "mean_token_accuracy": 0.8081613183021545, + "num_tokens": 832506550.0, + "step": 31354 + }, + { + "epoch": 2.5855528984909704, + "grad_norm": 2.5259389877319336, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.7827370166778564, + "num_tokens": 832527641.0, + "step": 31355 + }, + { + "epoch": 2.585635359116022, + "grad_norm": 2.115176200866699, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.7799506783485413, + "num_tokens": 832558762.0, + "step": 31356 + }, + { + "epoch": 2.5857178197410735, + "grad_norm": 2.208331823348999, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.7758920192718506, + "num_tokens": 832584579.0, + "step": 31357 + }, + { + "epoch": 2.5858002803661253, + "grad_norm": 2.142354726791382, + "learning_rate": 1e-06, + "loss": 0.5932, + "mean_token_accuracy": 0.7950180768966675, + "num_tokens": 832613414.0, + "step": 31358 + }, + { + "epoch": 2.5858827409911767, + "grad_norm": 2.084716558456421, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7372848987579346, + "num_tokens": 832645532.0, + "step": 31359 + }, + { + "epoch": 2.5859652016162284, + "grad_norm": 1.8759958744049072, + "learning_rate": 1e-06, + "loss": 0.7135, + "mean_token_accuracy": 0.777876615524292, + "num_tokens": 832685667.0, + "step": 31360 + }, + { + "epoch": 2.58604766224128, + "grad_norm": 1.9422167539596558, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.780434250831604, + "num_tokens": 832717645.0, + "step": 31361 + }, + { + "epoch": 2.586130122866331, + "grad_norm": 2.4358088970184326, + "learning_rate": 1e-06, + "loss": 0.7254, + "mean_token_accuracy": 0.759145975112915, + "num_tokens": 832741767.0, + "step": 31362 + }, + { + "epoch": 2.586212583491383, + "grad_norm": 2.0709784030914307, + "learning_rate": 1e-06, + "loss": 0.5341, + "mean_token_accuracy": 0.8098620772361755, + "num_tokens": 832771094.0, + "step": 31363 + }, + { + "epoch": 2.5862950441164343, + "grad_norm": 1.9831379652023315, + "learning_rate": 1e-06, + "loss": 0.7642, + "mean_token_accuracy": 0.7532194256782532, + "num_tokens": 832806836.0, + "step": 31364 + }, + { + "epoch": 2.586377504741486, + "grad_norm": 2.1201326847076416, + "learning_rate": 1e-06, + "loss": 0.6123, + "mean_token_accuracy": 0.7923699617385864, + "num_tokens": 832834469.0, + "step": 31365 + }, + { + "epoch": 2.5864599653665374, + "grad_norm": 2.5869686603546143, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7591376304626465, + "num_tokens": 832855823.0, + "step": 31366 + }, + { + "epoch": 2.5865424259915892, + "grad_norm": 2.359193801879883, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7485253214836121, + "num_tokens": 832880314.0, + "step": 31367 + }, + { + "epoch": 2.5866248866166406, + "grad_norm": 2.1605923175811768, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7868553400039673, + "num_tokens": 832908571.0, + "step": 31368 + }, + { + "epoch": 2.586707347241692, + "grad_norm": 2.249155282974243, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.7883943319320679, + "num_tokens": 832938939.0, + "step": 31369 + }, + { + "epoch": 2.5867898078667437, + "grad_norm": 2.2707340717315674, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7493466138839722, + "num_tokens": 832965558.0, + "step": 31370 + }, + { + "epoch": 2.586872268491795, + "grad_norm": 2.138547658920288, + "learning_rate": 1e-06, + "loss": 0.6214, + "mean_token_accuracy": 0.8001211881637573, + "num_tokens": 832992073.0, + "step": 31371 + }, + { + "epoch": 2.5869547291168464, + "grad_norm": 2.0277438163757324, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7878276109695435, + "num_tokens": 833023702.0, + "step": 31372 + }, + { + "epoch": 2.5870371897418982, + "grad_norm": 2.002850294113159, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7839502096176147, + "num_tokens": 833056071.0, + "step": 31373 + }, + { + "epoch": 2.58711965036695, + "grad_norm": 1.9521127939224243, + "learning_rate": 1e-06, + "loss": 0.5075, + "mean_token_accuracy": 0.8215866088867188, + "num_tokens": 833086643.0, + "step": 31374 + }, + { + "epoch": 2.5872021109920014, + "grad_norm": 1.9614564180374146, + "learning_rate": 1e-06, + "loss": 0.6152, + "mean_token_accuracy": 0.7977654933929443, + "num_tokens": 833120751.0, + "step": 31375 + }, + { + "epoch": 2.5872845716170527, + "grad_norm": 1.9847586154937744, + "learning_rate": 1e-06, + "loss": 0.5661, + "mean_token_accuracy": 0.7999821305274963, + "num_tokens": 833153816.0, + "step": 31376 + }, + { + "epoch": 2.5873670322421045, + "grad_norm": 2.2852907180786133, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7308210134506226, + "num_tokens": 833182707.0, + "step": 31377 + }, + { + "epoch": 2.587449492867156, + "grad_norm": 1.9588929414749146, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7667890787124634, + "num_tokens": 833216298.0, + "step": 31378 + }, + { + "epoch": 2.587531953492207, + "grad_norm": 1.9086155891418457, + "learning_rate": 1e-06, + "loss": 0.4923, + "mean_token_accuracy": 0.825410008430481, + "num_tokens": 833245476.0, + "step": 31379 + }, + { + "epoch": 2.587614414117259, + "grad_norm": 2.1400017738342285, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.7894172668457031, + "num_tokens": 833275113.0, + "step": 31380 + }, + { + "epoch": 2.587696874742311, + "grad_norm": 2.271402597427368, + "learning_rate": 1e-06, + "loss": 0.5705, + "mean_token_accuracy": 0.8081653118133545, + "num_tokens": 833300328.0, + "step": 31381 + }, + { + "epoch": 2.587779335367362, + "grad_norm": 1.8866294622421265, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7734060883522034, + "num_tokens": 833339238.0, + "step": 31382 + }, + { + "epoch": 2.5878617959924135, + "grad_norm": 2.1343917846679688, + "learning_rate": 1e-06, + "loss": 0.6132, + "mean_token_accuracy": 0.7994115352630615, + "num_tokens": 833366858.0, + "step": 31383 + }, + { + "epoch": 2.5879442566174653, + "grad_norm": 1.999099612236023, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7882775068283081, + "num_tokens": 833398103.0, + "step": 31384 + }, + { + "epoch": 2.5880267172425166, + "grad_norm": 2.2788186073303223, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7801787853240967, + "num_tokens": 833422139.0, + "step": 31385 + }, + { + "epoch": 2.588109177867568, + "grad_norm": 2.111361503601074, + "learning_rate": 1e-06, + "loss": 0.6756, + "mean_token_accuracy": 0.7843109369277954, + "num_tokens": 833451168.0, + "step": 31386 + }, + { + "epoch": 2.58819163849262, + "grad_norm": 2.1147375106811523, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.7868276834487915, + "num_tokens": 833479747.0, + "step": 31387 + }, + { + "epoch": 2.5882740991176716, + "grad_norm": 2.0329480171203613, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.763216495513916, + "num_tokens": 833511028.0, + "step": 31388 + }, + { + "epoch": 2.588356559742723, + "grad_norm": 2.2711377143859863, + "learning_rate": 1e-06, + "loss": 0.5363, + "mean_token_accuracy": 0.8186084628105164, + "num_tokens": 833534717.0, + "step": 31389 + }, + { + "epoch": 2.5884390203677743, + "grad_norm": 2.1013171672821045, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7745704054832458, + "num_tokens": 833564132.0, + "step": 31390 + }, + { + "epoch": 2.588521480992826, + "grad_norm": 2.042630195617676, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7707820534706116, + "num_tokens": 833594995.0, + "step": 31391 + }, + { + "epoch": 2.5886039416178774, + "grad_norm": 1.8420183658599854, + "learning_rate": 1e-06, + "loss": 0.5096, + "mean_token_accuracy": 0.8281543254852295, + "num_tokens": 833629088.0, + "step": 31392 + }, + { + "epoch": 2.588686402242929, + "grad_norm": 2.006758213043213, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7659926414489746, + "num_tokens": 833662676.0, + "step": 31393 + }, + { + "epoch": 2.5887688628679806, + "grad_norm": 2.165741205215454, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.7510738968849182, + "num_tokens": 833690786.0, + "step": 31394 + }, + { + "epoch": 2.588851323493032, + "grad_norm": 2.1415669918060303, + "learning_rate": 1e-06, + "loss": 0.7439, + "mean_token_accuracy": 0.7564008235931396, + "num_tokens": 833719485.0, + "step": 31395 + }, + { + "epoch": 2.5889337841180837, + "grad_norm": 2.12886381149292, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7724995017051697, + "num_tokens": 833747676.0, + "step": 31396 + }, + { + "epoch": 2.589016244743135, + "grad_norm": 2.2440176010131836, + "learning_rate": 1e-06, + "loss": 0.6609, + "mean_token_accuracy": 0.7763487100601196, + "num_tokens": 833772260.0, + "step": 31397 + }, + { + "epoch": 2.589098705368187, + "grad_norm": 2.1216201782226562, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.8023425936698914, + "num_tokens": 833801428.0, + "step": 31398 + }, + { + "epoch": 2.589181165993238, + "grad_norm": 2.0584723949432373, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7715331315994263, + "num_tokens": 833831527.0, + "step": 31399 + }, + { + "epoch": 2.5892636266182896, + "grad_norm": 2.2169430255889893, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7657039165496826, + "num_tokens": 833860358.0, + "step": 31400 + }, + { + "epoch": 2.5893460872433414, + "grad_norm": 1.8402334451675415, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.7911422848701477, + "num_tokens": 833894986.0, + "step": 31401 + }, + { + "epoch": 2.5894285478683927, + "grad_norm": 2.2399191856384277, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7668895721435547, + "num_tokens": 833924576.0, + "step": 31402 + }, + { + "epoch": 2.5895110084934445, + "grad_norm": 2.0242040157318115, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.7873122096061707, + "num_tokens": 833957747.0, + "step": 31403 + }, + { + "epoch": 2.589593469118496, + "grad_norm": 2.3402469158172607, + "learning_rate": 1e-06, + "loss": 0.5849, + "mean_token_accuracy": 0.7994835376739502, + "num_tokens": 833982900.0, + "step": 31404 + }, + { + "epoch": 2.5896759297435477, + "grad_norm": 2.165452480316162, + "learning_rate": 1e-06, + "loss": 0.5916, + "mean_token_accuracy": 0.8010092377662659, + "num_tokens": 834013000.0, + "step": 31405 + }, + { + "epoch": 2.589758390368599, + "grad_norm": 1.9260717630386353, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7861523628234863, + "num_tokens": 834045926.0, + "step": 31406 + }, + { + "epoch": 2.5898408509936504, + "grad_norm": 2.162135362625122, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7592011094093323, + "num_tokens": 834075205.0, + "step": 31407 + }, + { + "epoch": 2.589923311618702, + "grad_norm": 2.2700576782226562, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7803247570991516, + "num_tokens": 834100768.0, + "step": 31408 + }, + { + "epoch": 2.5900057722437535, + "grad_norm": 1.9569514989852905, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7479574084281921, + "num_tokens": 834137726.0, + "step": 31409 + }, + { + "epoch": 2.5900882328688053, + "grad_norm": 2.4723968505859375, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7771457433700562, + "num_tokens": 834161784.0, + "step": 31410 + }, + { + "epoch": 2.5901706934938566, + "grad_norm": 1.9672094583511353, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7688208222389221, + "num_tokens": 834195525.0, + "step": 31411 + }, + { + "epoch": 2.5902531541189084, + "grad_norm": 2.077923059463501, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.7806833386421204, + "num_tokens": 834230311.0, + "step": 31412 + }, + { + "epoch": 2.59033561474396, + "grad_norm": 2.352179527282715, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7683623433113098, + "num_tokens": 834257317.0, + "step": 31413 + }, + { + "epoch": 2.590418075369011, + "grad_norm": 1.9809470176696777, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.763968825340271, + "num_tokens": 834289571.0, + "step": 31414 + }, + { + "epoch": 2.590500535994063, + "grad_norm": 2.0078136920928955, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.792733371257782, + "num_tokens": 834319958.0, + "step": 31415 + }, + { + "epoch": 2.5905829966191143, + "grad_norm": 2.1098854541778564, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7628734111785889, + "num_tokens": 834349113.0, + "step": 31416 + }, + { + "epoch": 2.5906654572441656, + "grad_norm": 2.153031826019287, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7454196214675903, + "num_tokens": 834379386.0, + "step": 31417 + }, + { + "epoch": 2.5907479178692174, + "grad_norm": 2.0379855632781982, + "learning_rate": 1e-06, + "loss": 0.7431, + "mean_token_accuracy": 0.771018385887146, + "num_tokens": 834412429.0, + "step": 31418 + }, + { + "epoch": 2.5908303784942692, + "grad_norm": 1.8286837339401245, + "learning_rate": 1e-06, + "loss": 0.5776, + "mean_token_accuracy": 0.8139508366584778, + "num_tokens": 834447876.0, + "step": 31419 + }, + { + "epoch": 2.5909128391193206, + "grad_norm": 2.1020331382751465, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7582358121871948, + "num_tokens": 834479830.0, + "step": 31420 + }, + { + "epoch": 2.590995299744372, + "grad_norm": 2.0637595653533936, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7605278491973877, + "num_tokens": 834511463.0, + "step": 31421 + }, + { + "epoch": 2.5910777603694237, + "grad_norm": 2.3408350944519043, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7816693782806396, + "num_tokens": 834535097.0, + "step": 31422 + }, + { + "epoch": 2.591160220994475, + "grad_norm": 2.3070569038391113, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7689778804779053, + "num_tokens": 834562104.0, + "step": 31423 + }, + { + "epoch": 2.5912426816195264, + "grad_norm": 2.1575372219085693, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7647773027420044, + "num_tokens": 834593419.0, + "step": 31424 + }, + { + "epoch": 2.591325142244578, + "grad_norm": 1.7832669019699097, + "learning_rate": 1e-06, + "loss": 0.4715, + "mean_token_accuracy": 0.8375818729400635, + "num_tokens": 834626986.0, + "step": 31425 + }, + { + "epoch": 2.59140760286963, + "grad_norm": 2.024280309677124, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7339740991592407, + "num_tokens": 834663586.0, + "step": 31426 + }, + { + "epoch": 2.5914900634946814, + "grad_norm": 1.913430094718933, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7735377550125122, + "num_tokens": 834698001.0, + "step": 31427 + }, + { + "epoch": 2.5915725241197327, + "grad_norm": 1.8490991592407227, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.7902491092681885, + "num_tokens": 834733038.0, + "step": 31428 + }, + { + "epoch": 2.5916549847447845, + "grad_norm": 1.9292324781417847, + "learning_rate": 1e-06, + "loss": 0.8106, + "mean_token_accuracy": 0.7398847937583923, + "num_tokens": 834770252.0, + "step": 31429 + }, + { + "epoch": 2.591737445369836, + "grad_norm": 1.7379060983657837, + "learning_rate": 1e-06, + "loss": 0.4378, + "mean_token_accuracy": 0.8501520752906799, + "num_tokens": 834808859.0, + "step": 31430 + }, + { + "epoch": 2.591819905994887, + "grad_norm": 2.100367546081543, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7604888081550598, + "num_tokens": 834841435.0, + "step": 31431 + }, + { + "epoch": 2.591902366619939, + "grad_norm": 2.3561689853668213, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7365962266921997, + "num_tokens": 834866616.0, + "step": 31432 + }, + { + "epoch": 2.5919848272449904, + "grad_norm": 2.281352996826172, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7377555966377258, + "num_tokens": 834893739.0, + "step": 31433 + }, + { + "epoch": 2.592067287870042, + "grad_norm": 2.3025829792022705, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7722172737121582, + "num_tokens": 834920602.0, + "step": 31434 + }, + { + "epoch": 2.5921497484950935, + "grad_norm": 2.259472608566284, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7823000550270081, + "num_tokens": 834947228.0, + "step": 31435 + }, + { + "epoch": 2.5922322091201453, + "grad_norm": 2.100994110107422, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7319457530975342, + "num_tokens": 834977371.0, + "step": 31436 + }, + { + "epoch": 2.5923146697451966, + "grad_norm": 2.0822341442108154, + "learning_rate": 1e-06, + "loss": 0.6087, + "mean_token_accuracy": 0.7935869693756104, + "num_tokens": 835005984.0, + "step": 31437 + }, + { + "epoch": 2.592397130370248, + "grad_norm": 2.4646143913269043, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7789182066917419, + "num_tokens": 835029805.0, + "step": 31438 + }, + { + "epoch": 2.5924795909953, + "grad_norm": 2.03836989402771, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7617696523666382, + "num_tokens": 835059441.0, + "step": 31439 + }, + { + "epoch": 2.592562051620351, + "grad_norm": 2.016960859298706, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.7907904982566833, + "num_tokens": 835089657.0, + "step": 31440 + }, + { + "epoch": 2.592644512245403, + "grad_norm": 1.9251171350479126, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.7893061637878418, + "num_tokens": 835120847.0, + "step": 31441 + }, + { + "epoch": 2.5927269728704543, + "grad_norm": 2.165525436401367, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7839319705963135, + "num_tokens": 835147685.0, + "step": 31442 + }, + { + "epoch": 2.592809433495506, + "grad_norm": 2.0908424854278564, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7750992774963379, + "num_tokens": 835175325.0, + "step": 31443 + }, + { + "epoch": 2.5928918941205574, + "grad_norm": 1.9757080078125, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7769071459770203, + "num_tokens": 835210049.0, + "step": 31444 + }, + { + "epoch": 2.592974354745609, + "grad_norm": 1.9614858627319336, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7701102495193481, + "num_tokens": 835245314.0, + "step": 31445 + }, + { + "epoch": 2.5930568153706606, + "grad_norm": 1.752909779548645, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7923412322998047, + "num_tokens": 835283251.0, + "step": 31446 + }, + { + "epoch": 2.593139275995712, + "grad_norm": 2.1810641288757324, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.7925640344619751, + "num_tokens": 835309491.0, + "step": 31447 + }, + { + "epoch": 2.5932217366207637, + "grad_norm": 2.0292017459869385, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.776985764503479, + "num_tokens": 835339589.0, + "step": 31448 + }, + { + "epoch": 2.593304197245815, + "grad_norm": 1.9026410579681396, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7810244560241699, + "num_tokens": 835373741.0, + "step": 31449 + }, + { + "epoch": 2.593386657870867, + "grad_norm": 2.1126766204833984, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7600661516189575, + "num_tokens": 835407986.0, + "step": 31450 + }, + { + "epoch": 2.593469118495918, + "grad_norm": 1.8612563610076904, + "learning_rate": 1e-06, + "loss": 0.5915, + "mean_token_accuracy": 0.799258291721344, + "num_tokens": 835442309.0, + "step": 31451 + }, + { + "epoch": 2.5935515791209696, + "grad_norm": 1.96503746509552, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7597582340240479, + "num_tokens": 835473713.0, + "step": 31452 + }, + { + "epoch": 2.5936340397460214, + "grad_norm": 2.328728199005127, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7647943496704102, + "num_tokens": 835499238.0, + "step": 31453 + }, + { + "epoch": 2.5937165003710727, + "grad_norm": 2.176670789718628, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7611440420150757, + "num_tokens": 835528686.0, + "step": 31454 + }, + { + "epoch": 2.5937989609961245, + "grad_norm": 1.94996976852417, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7630102634429932, + "num_tokens": 835565618.0, + "step": 31455 + }, + { + "epoch": 2.593881421621176, + "grad_norm": 2.222088098526001, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.7851163148880005, + "num_tokens": 835595278.0, + "step": 31456 + }, + { + "epoch": 2.5939638822462276, + "grad_norm": 2.550835371017456, + "learning_rate": 1e-06, + "loss": 0.6691, + "mean_token_accuracy": 0.7748501300811768, + "num_tokens": 835617868.0, + "step": 31457 + }, + { + "epoch": 2.594046342871279, + "grad_norm": 2.0238888263702393, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7639179825782776, + "num_tokens": 835652116.0, + "step": 31458 + }, + { + "epoch": 2.5941288034963303, + "grad_norm": 2.28583025932312, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7284101247787476, + "num_tokens": 835684672.0, + "step": 31459 + }, + { + "epoch": 2.594211264121382, + "grad_norm": 2.0569818019866943, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7829276919364929, + "num_tokens": 835715403.0, + "step": 31460 + }, + { + "epoch": 2.5942937247464335, + "grad_norm": 2.1426656246185303, + "learning_rate": 1e-06, + "loss": 0.6445, + "mean_token_accuracy": 0.7934306859970093, + "num_tokens": 835747039.0, + "step": 31461 + }, + { + "epoch": 2.594376185371485, + "grad_norm": 2.109656572341919, + "learning_rate": 1e-06, + "loss": 0.4692, + "mean_token_accuracy": 0.8294495344161987, + "num_tokens": 835773769.0, + "step": 31462 + }, + { + "epoch": 2.5944586459965366, + "grad_norm": 1.9768319129943848, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.7677895426750183, + "num_tokens": 835808764.0, + "step": 31463 + }, + { + "epoch": 2.5945411066215884, + "grad_norm": 2.067352056503296, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7816832065582275, + "num_tokens": 835843655.0, + "step": 31464 + }, + { + "epoch": 2.59462356724664, + "grad_norm": 2.356945276260376, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.7332702875137329, + "num_tokens": 835870490.0, + "step": 31465 + }, + { + "epoch": 2.594706027871691, + "grad_norm": 2.0664849281311035, + "learning_rate": 1e-06, + "loss": 0.7006, + "mean_token_accuracy": 0.770115852355957, + "num_tokens": 835901507.0, + "step": 31466 + }, + { + "epoch": 2.594788488496743, + "grad_norm": 2.2983689308166504, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7812679409980774, + "num_tokens": 835926613.0, + "step": 31467 + }, + { + "epoch": 2.5948709491217943, + "grad_norm": 1.8933117389678955, + "learning_rate": 1e-06, + "loss": 0.5244, + "mean_token_accuracy": 0.8244345784187317, + "num_tokens": 835960122.0, + "step": 31468 + }, + { + "epoch": 2.5949534097468456, + "grad_norm": 2.17848801612854, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7690432667732239, + "num_tokens": 835989382.0, + "step": 31469 + }, + { + "epoch": 2.5950358703718974, + "grad_norm": 2.0491631031036377, + "learning_rate": 1e-06, + "loss": 0.6585, + "mean_token_accuracy": 0.789013147354126, + "num_tokens": 836021642.0, + "step": 31470 + }, + { + "epoch": 2.595118330996949, + "grad_norm": 2.4565300941467285, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.7750344276428223, + "num_tokens": 836044394.0, + "step": 31471 + }, + { + "epoch": 2.5952007916220006, + "grad_norm": 2.108889102935791, + "learning_rate": 1e-06, + "loss": 0.6353, + "mean_token_accuracy": 0.7929286360740662, + "num_tokens": 836070971.0, + "step": 31472 + }, + { + "epoch": 2.595283252247052, + "grad_norm": 2.114670515060425, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7871192097663879, + "num_tokens": 836102653.0, + "step": 31473 + }, + { + "epoch": 2.5953657128721037, + "grad_norm": 2.1953744888305664, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.785088062286377, + "num_tokens": 836129054.0, + "step": 31474 + }, + { + "epoch": 2.595448173497155, + "grad_norm": 1.8363465070724487, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.7824288606643677, + "num_tokens": 836164920.0, + "step": 31475 + }, + { + "epoch": 2.5955306341222064, + "grad_norm": 2.1713898181915283, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7367751598358154, + "num_tokens": 836194021.0, + "step": 31476 + }, + { + "epoch": 2.595613094747258, + "grad_norm": 2.309602737426758, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.7846086025238037, + "num_tokens": 836219827.0, + "step": 31477 + }, + { + "epoch": 2.5956955553723096, + "grad_norm": 1.9468342065811157, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.7826820015907288, + "num_tokens": 836255195.0, + "step": 31478 + }, + { + "epoch": 2.5957780159973614, + "grad_norm": 1.9995759725570679, + "learning_rate": 1e-06, + "loss": 0.8037, + "mean_token_accuracy": 0.7336089611053467, + "num_tokens": 836287987.0, + "step": 31479 + }, + { + "epoch": 2.5958604766224127, + "grad_norm": 2.419928550720215, + "learning_rate": 1e-06, + "loss": 0.7278, + "mean_token_accuracy": 0.7662029266357422, + "num_tokens": 836312688.0, + "step": 31480 + }, + { + "epoch": 2.5959429372474645, + "grad_norm": 2.0442118644714355, + "learning_rate": 1e-06, + "loss": 0.7883, + "mean_token_accuracy": 0.7477237582206726, + "num_tokens": 836343785.0, + "step": 31481 + }, + { + "epoch": 2.596025397872516, + "grad_norm": 2.6079676151275635, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7626948356628418, + "num_tokens": 836364708.0, + "step": 31482 + }, + { + "epoch": 2.596107858497567, + "grad_norm": 1.959078073501587, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.7867948412895203, + "num_tokens": 836398430.0, + "step": 31483 + }, + { + "epoch": 2.596190319122619, + "grad_norm": 2.2421510219573975, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7514787912368774, + "num_tokens": 836424887.0, + "step": 31484 + }, + { + "epoch": 2.5962727797476703, + "grad_norm": 2.168461561203003, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.7689639329910278, + "num_tokens": 836453574.0, + "step": 31485 + }, + { + "epoch": 2.596355240372722, + "grad_norm": 2.122563123703003, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.7884325385093689, + "num_tokens": 836480412.0, + "step": 31486 + }, + { + "epoch": 2.5964377009977735, + "grad_norm": 2.159858226776123, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7604436278343201, + "num_tokens": 836507517.0, + "step": 31487 + }, + { + "epoch": 2.5965201616228253, + "grad_norm": 2.039483070373535, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.7809719443321228, + "num_tokens": 836539446.0, + "step": 31488 + }, + { + "epoch": 2.5966026222478766, + "grad_norm": 2.1430776119232178, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7677074670791626, + "num_tokens": 836568826.0, + "step": 31489 + }, + { + "epoch": 2.596685082872928, + "grad_norm": 2.330836534500122, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7101556658744812, + "num_tokens": 836594583.0, + "step": 31490 + }, + { + "epoch": 2.59676754349798, + "grad_norm": 1.8965510129928589, + "learning_rate": 1e-06, + "loss": 0.5325, + "mean_token_accuracy": 0.821844220161438, + "num_tokens": 836624344.0, + "step": 31491 + }, + { + "epoch": 2.596850004123031, + "grad_norm": 2.0978496074676514, + "learning_rate": 1e-06, + "loss": 0.6187, + "mean_token_accuracy": 0.7941359281539917, + "num_tokens": 836654783.0, + "step": 31492 + }, + { + "epoch": 2.596932464748083, + "grad_norm": 2.212102174758911, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7683720588684082, + "num_tokens": 836684266.0, + "step": 31493 + }, + { + "epoch": 2.5970149253731343, + "grad_norm": 2.1508307456970215, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.7798503637313843, + "num_tokens": 836712825.0, + "step": 31494 + }, + { + "epoch": 2.597097385998186, + "grad_norm": 1.9497791528701782, + "learning_rate": 1e-06, + "loss": 0.505, + "mean_token_accuracy": 0.8252646327018738, + "num_tokens": 836747067.0, + "step": 31495 + }, + { + "epoch": 2.5971798466232374, + "grad_norm": 2.0735175609588623, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7723216414451599, + "num_tokens": 836775195.0, + "step": 31496 + }, + { + "epoch": 2.5972623072482888, + "grad_norm": 1.9907201528549194, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7652663588523865, + "num_tokens": 836806580.0, + "step": 31497 + }, + { + "epoch": 2.5973447678733406, + "grad_norm": 2.519865036010742, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.7863744497299194, + "num_tokens": 836829000.0, + "step": 31498 + }, + { + "epoch": 2.597427228498392, + "grad_norm": 2.0448503494262695, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7698888778686523, + "num_tokens": 836860776.0, + "step": 31499 + }, + { + "epoch": 2.5975096891234437, + "grad_norm": 2.2211577892303467, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.773364782333374, + "num_tokens": 836887257.0, + "step": 31500 + }, + { + "epoch": 2.597592149748495, + "grad_norm": 1.9469695091247559, + "learning_rate": 1e-06, + "loss": 0.5523, + "mean_token_accuracy": 0.8246599435806274, + "num_tokens": 836922028.0, + "step": 31501 + }, + { + "epoch": 2.597674610373547, + "grad_norm": 1.8503934144973755, + "learning_rate": 1e-06, + "loss": 0.5601, + "mean_token_accuracy": 0.8161419034004211, + "num_tokens": 836955185.0, + "step": 31502 + }, + { + "epoch": 2.597757070998598, + "grad_norm": 1.9117631912231445, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.768751859664917, + "num_tokens": 836989451.0, + "step": 31503 + }, + { + "epoch": 2.5978395316236496, + "grad_norm": 2.021183729171753, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7840064764022827, + "num_tokens": 837019355.0, + "step": 31504 + }, + { + "epoch": 2.5979219922487014, + "grad_norm": 2.1596686840057373, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7812095284461975, + "num_tokens": 837047550.0, + "step": 31505 + }, + { + "epoch": 2.5980044528737527, + "grad_norm": 2.1509323120117188, + "learning_rate": 1e-06, + "loss": 0.5703, + "mean_token_accuracy": 0.8079825639724731, + "num_tokens": 837076050.0, + "step": 31506 + }, + { + "epoch": 2.598086913498804, + "grad_norm": 1.8350378274917603, + "learning_rate": 1e-06, + "loss": 0.6151, + "mean_token_accuracy": 0.8020362854003906, + "num_tokens": 837111898.0, + "step": 31507 + }, + { + "epoch": 2.598169374123856, + "grad_norm": 2.3323190212249756, + "learning_rate": 1e-06, + "loss": 0.6888, + "mean_token_accuracy": 0.7669011354446411, + "num_tokens": 837135666.0, + "step": 31508 + }, + { + "epoch": 2.5982518347489076, + "grad_norm": 2.099419355392456, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7655293941497803, + "num_tokens": 837167251.0, + "step": 31509 + }, + { + "epoch": 2.598334295373959, + "grad_norm": 1.8995702266693115, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.7867408990859985, + "num_tokens": 837199846.0, + "step": 31510 + }, + { + "epoch": 2.5984167559990103, + "grad_norm": 1.833897352218628, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7906786203384399, + "num_tokens": 837235344.0, + "step": 31511 + }, + { + "epoch": 2.598499216624062, + "grad_norm": 2.2061257362365723, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7516800165176392, + "num_tokens": 837264660.0, + "step": 31512 + }, + { + "epoch": 2.5985816772491135, + "grad_norm": 2.024157762527466, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7381361126899719, + "num_tokens": 837297886.0, + "step": 31513 + }, + { + "epoch": 2.598664137874165, + "grad_norm": 2.215428590774536, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.77824866771698, + "num_tokens": 837323440.0, + "step": 31514 + }, + { + "epoch": 2.5987465984992166, + "grad_norm": 2.204547643661499, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7571059465408325, + "num_tokens": 837350106.0, + "step": 31515 + }, + { + "epoch": 2.5988290591242684, + "grad_norm": 2.1225345134735107, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7743017673492432, + "num_tokens": 837381394.0, + "step": 31516 + }, + { + "epoch": 2.5989115197493198, + "grad_norm": 2.210921049118042, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7715291380882263, + "num_tokens": 837410234.0, + "step": 31517 + }, + { + "epoch": 2.598993980374371, + "grad_norm": 2.071068048477173, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.7563278675079346, + "num_tokens": 837440704.0, + "step": 31518 + }, + { + "epoch": 2.599076440999423, + "grad_norm": 2.258131265640259, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.726561427116394, + "num_tokens": 837472259.0, + "step": 31519 + }, + { + "epoch": 2.5991589016244743, + "grad_norm": 2.2417593002319336, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7775161862373352, + "num_tokens": 837498576.0, + "step": 31520 + }, + { + "epoch": 2.5992413622495256, + "grad_norm": 1.9931186437606812, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.7993359565734863, + "num_tokens": 837532432.0, + "step": 31521 + }, + { + "epoch": 2.5993238228745774, + "grad_norm": 1.8707937002182007, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.783387303352356, + "num_tokens": 837566834.0, + "step": 31522 + }, + { + "epoch": 2.5994062834996288, + "grad_norm": 1.8886945247650146, + "learning_rate": 1e-06, + "loss": 0.4514, + "mean_token_accuracy": 0.8463567495346069, + "num_tokens": 837597032.0, + "step": 31523 + }, + { + "epoch": 2.5994887441246806, + "grad_norm": 2.0727829933166504, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7579383850097656, + "num_tokens": 837629062.0, + "step": 31524 + }, + { + "epoch": 2.599571204749732, + "grad_norm": 2.3251259326934814, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.741275429725647, + "num_tokens": 837655114.0, + "step": 31525 + }, + { + "epoch": 2.5996536653747837, + "grad_norm": 2.0217273235321045, + "learning_rate": 1e-06, + "loss": 0.5615, + "mean_token_accuracy": 0.8083826899528503, + "num_tokens": 837682686.0, + "step": 31526 + }, + { + "epoch": 2.599736125999835, + "grad_norm": 2.005096197128296, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.6972635388374329, + "num_tokens": 837718942.0, + "step": 31527 + }, + { + "epoch": 2.5998185866248864, + "grad_norm": 2.0138535499572754, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7669500112533569, + "num_tokens": 837752601.0, + "step": 31528 + }, + { + "epoch": 2.599901047249938, + "grad_norm": 2.1050491333007812, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7634586095809937, + "num_tokens": 837781204.0, + "step": 31529 + }, + { + "epoch": 2.5999835078749896, + "grad_norm": 1.9974578619003296, + "learning_rate": 1e-06, + "loss": 0.5796, + "mean_token_accuracy": 0.7937329411506653, + "num_tokens": 837811593.0, + "step": 31530 + }, + { + "epoch": 2.6000659685000413, + "grad_norm": 2.0398709774017334, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.7913110256195068, + "num_tokens": 837842203.0, + "step": 31531 + }, + { + "epoch": 2.6001484291250927, + "grad_norm": 1.9657152891159058, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7697112560272217, + "num_tokens": 837873850.0, + "step": 31532 + }, + { + "epoch": 2.6002308897501445, + "grad_norm": 1.8178114891052246, + "learning_rate": 1e-06, + "loss": 0.5418, + "mean_token_accuracy": 0.8133915662765503, + "num_tokens": 837909692.0, + "step": 31533 + }, + { + "epoch": 2.600313350375196, + "grad_norm": 2.2463955879211426, + "learning_rate": 1e-06, + "loss": 0.6073, + "mean_token_accuracy": 0.788762629032135, + "num_tokens": 837936859.0, + "step": 31534 + }, + { + "epoch": 2.600395811000247, + "grad_norm": 2.1553380489349365, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.7852938175201416, + "num_tokens": 837964834.0, + "step": 31535 + }, + { + "epoch": 2.600478271625299, + "grad_norm": 2.1331896781921387, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7752953767776489, + "num_tokens": 837993752.0, + "step": 31536 + }, + { + "epoch": 2.6005607322503503, + "grad_norm": 2.6211259365081787, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.7769757509231567, + "num_tokens": 838015384.0, + "step": 31537 + }, + { + "epoch": 2.600643192875402, + "grad_norm": 2.21640944480896, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7326101660728455, + "num_tokens": 838042441.0, + "step": 31538 + }, + { + "epoch": 2.6007256535004535, + "grad_norm": 2.0308051109313965, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.780427098274231, + "num_tokens": 838075401.0, + "step": 31539 + }, + { + "epoch": 2.6008081141255053, + "grad_norm": 2.2638378143310547, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7715258598327637, + "num_tokens": 838103374.0, + "step": 31540 + }, + { + "epoch": 2.6008905747505566, + "grad_norm": 1.8858469724655151, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.7911427021026611, + "num_tokens": 838139828.0, + "step": 31541 + }, + { + "epoch": 2.600973035375608, + "grad_norm": 1.953730821609497, + "learning_rate": 1e-06, + "loss": 0.5498, + "mean_token_accuracy": 0.8083133697509766, + "num_tokens": 838169268.0, + "step": 31542 + }, + { + "epoch": 2.6010554960006598, + "grad_norm": 2.052173137664795, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7765864133834839, + "num_tokens": 838198808.0, + "step": 31543 + }, + { + "epoch": 2.601137956625711, + "grad_norm": 2.417426347732544, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7578073740005493, + "num_tokens": 838223394.0, + "step": 31544 + }, + { + "epoch": 2.601220417250763, + "grad_norm": 2.0625314712524414, + "learning_rate": 1e-06, + "loss": 0.7266, + "mean_token_accuracy": 0.7654125690460205, + "num_tokens": 838253558.0, + "step": 31545 + }, + { + "epoch": 2.6013028778758143, + "grad_norm": 2.1770122051239014, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.7951847314834595, + "num_tokens": 838283023.0, + "step": 31546 + }, + { + "epoch": 2.601385338500866, + "grad_norm": 2.149894952774048, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7691056728363037, + "num_tokens": 838310377.0, + "step": 31547 + }, + { + "epoch": 2.6014677991259174, + "grad_norm": 1.8395979404449463, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7842872142791748, + "num_tokens": 838348234.0, + "step": 31548 + }, + { + "epoch": 2.6015502597509688, + "grad_norm": 2.0672035217285156, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.7951385974884033, + "num_tokens": 838379576.0, + "step": 31549 + }, + { + "epoch": 2.6016327203760206, + "grad_norm": 1.8259246349334717, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7771771550178528, + "num_tokens": 838420032.0, + "step": 31550 + }, + { + "epoch": 2.601715181001072, + "grad_norm": 1.901087999343872, + "learning_rate": 1e-06, + "loss": 0.6642, + "mean_token_accuracy": 0.7674382925033569, + "num_tokens": 838455484.0, + "step": 31551 + }, + { + "epoch": 2.6017976416261233, + "grad_norm": 2.0597779750823975, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.766196072101593, + "num_tokens": 838484284.0, + "step": 31552 + }, + { + "epoch": 2.601880102251175, + "grad_norm": 2.0146546363830566, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.8046850562095642, + "num_tokens": 838514857.0, + "step": 31553 + }, + { + "epoch": 2.601962562876227, + "grad_norm": 2.121640205383301, + "learning_rate": 1e-06, + "loss": 0.7035, + "mean_token_accuracy": 0.7745200395584106, + "num_tokens": 838543915.0, + "step": 31554 + }, + { + "epoch": 2.602045023501278, + "grad_norm": 2.0055346488952637, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7461979389190674, + "num_tokens": 838576608.0, + "step": 31555 + }, + { + "epoch": 2.6021274841263295, + "grad_norm": 1.9481934309005737, + "learning_rate": 1e-06, + "loss": 0.4936, + "mean_token_accuracy": 0.8318567276000977, + "num_tokens": 838607472.0, + "step": 31556 + }, + { + "epoch": 2.6022099447513813, + "grad_norm": 1.9795993566513062, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.7899585366249084, + "num_tokens": 838639848.0, + "step": 31557 + }, + { + "epoch": 2.6022924053764327, + "grad_norm": 2.354642868041992, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7676661014556885, + "num_tokens": 838664327.0, + "step": 31558 + }, + { + "epoch": 2.602374866001484, + "grad_norm": 1.997883915901184, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7738123536109924, + "num_tokens": 838699650.0, + "step": 31559 + }, + { + "epoch": 2.602457326626536, + "grad_norm": 2.209287166595459, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7493740916252136, + "num_tokens": 838726733.0, + "step": 31560 + }, + { + "epoch": 2.6025397872515876, + "grad_norm": 2.2213797569274902, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7511876821517944, + "num_tokens": 838755798.0, + "step": 31561 + }, + { + "epoch": 2.602622247876639, + "grad_norm": 2.1028711795806885, + "learning_rate": 1e-06, + "loss": 0.6211, + "mean_token_accuracy": 0.8010305166244507, + "num_tokens": 838784012.0, + "step": 31562 + }, + { + "epoch": 2.6027047085016903, + "grad_norm": 2.0914559364318848, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7625097632408142, + "num_tokens": 838816843.0, + "step": 31563 + }, + { + "epoch": 2.602787169126742, + "grad_norm": 2.195295572280884, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7488605380058289, + "num_tokens": 838847281.0, + "step": 31564 + }, + { + "epoch": 2.6028696297517935, + "grad_norm": 2.1205756664276123, + "learning_rate": 1e-06, + "loss": 0.5784, + "mean_token_accuracy": 0.8042473196983337, + "num_tokens": 838875364.0, + "step": 31565 + }, + { + "epoch": 2.602952090376845, + "grad_norm": 2.186582565307617, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7695966958999634, + "num_tokens": 838901010.0, + "step": 31566 + }, + { + "epoch": 2.6030345510018966, + "grad_norm": 1.8650798797607422, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7719058990478516, + "num_tokens": 838938353.0, + "step": 31567 + }, + { + "epoch": 2.603117011626948, + "grad_norm": 2.3081612586975098, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7662915587425232, + "num_tokens": 838963802.0, + "step": 31568 + }, + { + "epoch": 2.6031994722519998, + "grad_norm": 2.89363169670105, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.765109121799469, + "num_tokens": 838980575.0, + "step": 31569 + }, + { + "epoch": 2.603281932877051, + "grad_norm": 2.0448403358459473, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7573115825653076, + "num_tokens": 839015339.0, + "step": 31570 + }, + { + "epoch": 2.603364393502103, + "grad_norm": 2.2508976459503174, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.799124002456665, + "num_tokens": 839042926.0, + "step": 31571 + }, + { + "epoch": 2.6034468541271543, + "grad_norm": 2.28189754486084, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7731195092201233, + "num_tokens": 839071933.0, + "step": 31572 + }, + { + "epoch": 2.6035293147522056, + "grad_norm": 2.020340919494629, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7489516139030457, + "num_tokens": 839100781.0, + "step": 31573 + }, + { + "epoch": 2.6036117753772574, + "grad_norm": 2.1320204734802246, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7276654243469238, + "num_tokens": 839130875.0, + "step": 31574 + }, + { + "epoch": 2.6036942360023088, + "grad_norm": 2.0764386653900146, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.788329005241394, + "num_tokens": 839160077.0, + "step": 31575 + }, + { + "epoch": 2.6037766966273606, + "grad_norm": 1.8899391889572144, + "learning_rate": 1e-06, + "loss": 0.4899, + "mean_token_accuracy": 0.8382538557052612, + "num_tokens": 839193369.0, + "step": 31576 + }, + { + "epoch": 2.603859157252412, + "grad_norm": 2.0469679832458496, + "learning_rate": 1e-06, + "loss": 0.5039, + "mean_token_accuracy": 0.829371452331543, + "num_tokens": 839222225.0, + "step": 31577 + }, + { + "epoch": 2.6039416178774637, + "grad_norm": 1.9811770915985107, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7603662014007568, + "num_tokens": 839255612.0, + "step": 31578 + }, + { + "epoch": 2.604024078502515, + "grad_norm": 2.1723122596740723, + "learning_rate": 1e-06, + "loss": 0.6031, + "mean_token_accuracy": 0.8002225756645203, + "num_tokens": 839281086.0, + "step": 31579 + }, + { + "epoch": 2.6041065391275664, + "grad_norm": 1.947298288345337, + "learning_rate": 1e-06, + "loss": 0.8394, + "mean_token_accuracy": 0.7430350184440613, + "num_tokens": 839316299.0, + "step": 31580 + }, + { + "epoch": 2.604188999752618, + "grad_norm": 2.1313016414642334, + "learning_rate": 1e-06, + "loss": 0.6268, + "mean_token_accuracy": 0.7919334769248962, + "num_tokens": 839345700.0, + "step": 31581 + }, + { + "epoch": 2.6042714603776695, + "grad_norm": 2.1497058868408203, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.76030433177948, + "num_tokens": 839374061.0, + "step": 31582 + }, + { + "epoch": 2.6043539210027213, + "grad_norm": 2.0419504642486572, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7791839838027954, + "num_tokens": 839406044.0, + "step": 31583 + }, + { + "epoch": 2.6044363816277727, + "grad_norm": 2.3629400730133057, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.763770341873169, + "num_tokens": 839430608.0, + "step": 31584 + }, + { + "epoch": 2.6045188422528245, + "grad_norm": 1.8486806154251099, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7364226579666138, + "num_tokens": 839470026.0, + "step": 31585 + }, + { + "epoch": 2.604601302877876, + "grad_norm": 2.2678582668304443, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7794288396835327, + "num_tokens": 839496486.0, + "step": 31586 + }, + { + "epoch": 2.604683763502927, + "grad_norm": 2.0643322467803955, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7501624822616577, + "num_tokens": 839528354.0, + "step": 31587 + }, + { + "epoch": 2.604766224127979, + "grad_norm": 2.026339292526245, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7638640403747559, + "num_tokens": 839562712.0, + "step": 31588 + }, + { + "epoch": 2.6048486847530303, + "grad_norm": 2.1829638481140137, + "learning_rate": 1e-06, + "loss": 0.5838, + "mean_token_accuracy": 0.8105071187019348, + "num_tokens": 839589300.0, + "step": 31589 + }, + { + "epoch": 2.604931145378082, + "grad_norm": 2.375410318374634, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7546820640563965, + "num_tokens": 839613965.0, + "step": 31590 + }, + { + "epoch": 2.6050136060031335, + "grad_norm": 2.2253222465515137, + "learning_rate": 1e-06, + "loss": 0.6839, + "mean_token_accuracy": 0.7807836532592773, + "num_tokens": 839640135.0, + "step": 31591 + }, + { + "epoch": 2.6050960666281853, + "grad_norm": 1.6665164232254028, + "learning_rate": 1e-06, + "loss": 0.5841, + "mean_token_accuracy": 0.8136470317840576, + "num_tokens": 839682618.0, + "step": 31592 + }, + { + "epoch": 2.6051785272532366, + "grad_norm": 1.9770277738571167, + "learning_rate": 1e-06, + "loss": 0.5558, + "mean_token_accuracy": 0.8116669058799744, + "num_tokens": 839716035.0, + "step": 31593 + }, + { + "epoch": 2.605260987878288, + "grad_norm": 2.053044319152832, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7453108429908752, + "num_tokens": 839746165.0, + "step": 31594 + }, + { + "epoch": 2.6053434485033398, + "grad_norm": 2.3402717113494873, + "learning_rate": 1e-06, + "loss": 0.6564, + "mean_token_accuracy": 0.7807353734970093, + "num_tokens": 839770521.0, + "step": 31595 + }, + { + "epoch": 2.605425909128391, + "grad_norm": 2.3694827556610107, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7609961032867432, + "num_tokens": 839796318.0, + "step": 31596 + }, + { + "epoch": 2.6055083697534425, + "grad_norm": 2.02223539352417, + "learning_rate": 1e-06, + "loss": 0.5452, + "mean_token_accuracy": 0.8124140501022339, + "num_tokens": 839823607.0, + "step": 31597 + }, + { + "epoch": 2.6055908303784943, + "grad_norm": 2.0147786140441895, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7785956859588623, + "num_tokens": 839855326.0, + "step": 31598 + }, + { + "epoch": 2.605673291003546, + "grad_norm": 2.0460355281829834, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7736586332321167, + "num_tokens": 839887913.0, + "step": 31599 + }, + { + "epoch": 2.6057557516285974, + "grad_norm": 2.3813188076019287, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7761596441268921, + "num_tokens": 839912167.0, + "step": 31600 + }, + { + "epoch": 2.6058382122536488, + "grad_norm": 1.8986831903457642, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7754244208335876, + "num_tokens": 839949634.0, + "step": 31601 + }, + { + "epoch": 2.6059206728787006, + "grad_norm": 2.161677122116089, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7458459734916687, + "num_tokens": 839978227.0, + "step": 31602 + }, + { + "epoch": 2.606003133503752, + "grad_norm": 2.400376319885254, + "learning_rate": 1e-06, + "loss": 0.61, + "mean_token_accuracy": 0.7982630729675293, + "num_tokens": 840002025.0, + "step": 31603 + }, + { + "epoch": 2.6060855941288033, + "grad_norm": 2.2652101516723633, + "learning_rate": 1e-06, + "loss": 0.5492, + "mean_token_accuracy": 0.812843918800354, + "num_tokens": 840027758.0, + "step": 31604 + }, + { + "epoch": 2.606168054753855, + "grad_norm": 2.226203680038452, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.7774008512496948, + "num_tokens": 840055927.0, + "step": 31605 + }, + { + "epoch": 2.606250515378907, + "grad_norm": 2.363067150115967, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7606282234191895, + "num_tokens": 840080258.0, + "step": 31606 + }, + { + "epoch": 2.606332976003958, + "grad_norm": 2.082733392715454, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.7945797443389893, + "num_tokens": 840109536.0, + "step": 31607 + }, + { + "epoch": 2.6064154366290095, + "grad_norm": 1.8884860277175903, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7334848046302795, + "num_tokens": 840148458.0, + "step": 31608 + }, + { + "epoch": 2.6064978972540613, + "grad_norm": 2.3772642612457275, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.7860039472579956, + "num_tokens": 840173857.0, + "step": 31609 + }, + { + "epoch": 2.6065803578791127, + "grad_norm": 1.812770128250122, + "learning_rate": 1e-06, + "loss": 0.6518, + "mean_token_accuracy": 0.7946600317955017, + "num_tokens": 840210980.0, + "step": 31610 + }, + { + "epoch": 2.606662818504164, + "grad_norm": 2.046495199203491, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7608330249786377, + "num_tokens": 840239934.0, + "step": 31611 + }, + { + "epoch": 2.606745279129216, + "grad_norm": 1.9055932760238647, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7807111740112305, + "num_tokens": 840275520.0, + "step": 31612 + }, + { + "epoch": 2.606827739754267, + "grad_norm": 2.1788277626037598, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7590121626853943, + "num_tokens": 840303340.0, + "step": 31613 + }, + { + "epoch": 2.606910200379319, + "grad_norm": 1.8440239429473877, + "learning_rate": 1e-06, + "loss": 0.5695, + "mean_token_accuracy": 0.8087184429168701, + "num_tokens": 840342087.0, + "step": 31614 + }, + { + "epoch": 2.6069926610043703, + "grad_norm": 1.9539681673049927, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.763157069683075, + "num_tokens": 840376109.0, + "step": 31615 + }, + { + "epoch": 2.607075121629422, + "grad_norm": 2.366835355758667, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7506299018859863, + "num_tokens": 840400706.0, + "step": 31616 + }, + { + "epoch": 2.6071575822544735, + "grad_norm": 1.8940861225128174, + "learning_rate": 1e-06, + "loss": 0.6464, + "mean_token_accuracy": 0.791856050491333, + "num_tokens": 840431462.0, + "step": 31617 + }, + { + "epoch": 2.607240042879525, + "grad_norm": 2.0563061237335205, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7565492987632751, + "num_tokens": 840462778.0, + "step": 31618 + }, + { + "epoch": 2.6073225035045766, + "grad_norm": 1.9094280004501343, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.7853596806526184, + "num_tokens": 840498240.0, + "step": 31619 + }, + { + "epoch": 2.607404964129628, + "grad_norm": 2.455749750137329, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7534533739089966, + "num_tokens": 840520694.0, + "step": 31620 + }, + { + "epoch": 2.6074874247546798, + "grad_norm": 2.4651432037353516, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7544577717781067, + "num_tokens": 840544969.0, + "step": 31621 + }, + { + "epoch": 2.607569885379731, + "grad_norm": 2.2831685543060303, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7636001110076904, + "num_tokens": 840571470.0, + "step": 31622 + }, + { + "epoch": 2.607652346004783, + "grad_norm": 1.9498647451400757, + "learning_rate": 1e-06, + "loss": 0.5514, + "mean_token_accuracy": 0.8175545930862427, + "num_tokens": 840600092.0, + "step": 31623 + }, + { + "epoch": 2.6077348066298343, + "grad_norm": 2.1714107990264893, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.764422595500946, + "num_tokens": 840632937.0, + "step": 31624 + }, + { + "epoch": 2.6078172672548856, + "grad_norm": 2.217867612838745, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7652336955070496, + "num_tokens": 840662192.0, + "step": 31625 + }, + { + "epoch": 2.6078997278799374, + "grad_norm": 2.162057399749756, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7696203589439392, + "num_tokens": 840689054.0, + "step": 31626 + }, + { + "epoch": 2.6079821885049888, + "grad_norm": 2.100908041000366, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7523661851882935, + "num_tokens": 840719273.0, + "step": 31627 + }, + { + "epoch": 2.6080646491300405, + "grad_norm": 2.137402057647705, + "learning_rate": 1e-06, + "loss": 0.5537, + "mean_token_accuracy": 0.8145046234130859, + "num_tokens": 840745543.0, + "step": 31628 + }, + { + "epoch": 2.608147109755092, + "grad_norm": 2.111701488494873, + "learning_rate": 1e-06, + "loss": 0.5317, + "mean_token_accuracy": 0.8157912492752075, + "num_tokens": 840772511.0, + "step": 31629 + }, + { + "epoch": 2.6082295703801437, + "grad_norm": 2.2724084854125977, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7671493291854858, + "num_tokens": 840799546.0, + "step": 31630 + }, + { + "epoch": 2.608312031005195, + "grad_norm": 2.014864206314087, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7681407332420349, + "num_tokens": 840832159.0, + "step": 31631 + }, + { + "epoch": 2.6083944916302464, + "grad_norm": 2.098698377609253, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7808230519294739, + "num_tokens": 840862217.0, + "step": 31632 + }, + { + "epoch": 2.608476952255298, + "grad_norm": 2.0594890117645264, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.7853590250015259, + "num_tokens": 840893635.0, + "step": 31633 + }, + { + "epoch": 2.6085594128803495, + "grad_norm": 2.180532693862915, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.7896000146865845, + "num_tokens": 840922007.0, + "step": 31634 + }, + { + "epoch": 2.6086418735054013, + "grad_norm": 2.1043617725372314, + "learning_rate": 1e-06, + "loss": 0.5802, + "mean_token_accuracy": 0.7967173457145691, + "num_tokens": 840948677.0, + "step": 31635 + }, + { + "epoch": 2.6087243341304527, + "grad_norm": 2.013493537902832, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7791550159454346, + "num_tokens": 840977446.0, + "step": 31636 + }, + { + "epoch": 2.6088067947555045, + "grad_norm": 1.9376816749572754, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7936204075813293, + "num_tokens": 841008090.0, + "step": 31637 + }, + { + "epoch": 2.608889255380556, + "grad_norm": 2.258652925491333, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.7791450619697571, + "num_tokens": 841035289.0, + "step": 31638 + }, + { + "epoch": 2.608971716005607, + "grad_norm": 2.1166539192199707, + "learning_rate": 1e-06, + "loss": 0.7637, + "mean_token_accuracy": 0.7494465112686157, + "num_tokens": 841065884.0, + "step": 31639 + }, + { + "epoch": 2.609054176630659, + "grad_norm": 2.41926646232605, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.782213568687439, + "num_tokens": 841089600.0, + "step": 31640 + }, + { + "epoch": 2.6091366372557103, + "grad_norm": 2.172372817993164, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7540628910064697, + "num_tokens": 841119996.0, + "step": 31641 + }, + { + "epoch": 2.6092190978807617, + "grad_norm": 1.9925978183746338, + "learning_rate": 1e-06, + "loss": 0.7509, + "mean_token_accuracy": 0.7614638209342957, + "num_tokens": 841152949.0, + "step": 31642 + }, + { + "epoch": 2.6093015585058135, + "grad_norm": 2.2112317085266113, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7527217864990234, + "num_tokens": 841182350.0, + "step": 31643 + }, + { + "epoch": 2.6093840191308653, + "grad_norm": 2.415025472640991, + "learning_rate": 1e-06, + "loss": 0.6372, + "mean_token_accuracy": 0.7778760194778442, + "num_tokens": 841205806.0, + "step": 31644 + }, + { + "epoch": 2.6094664797559166, + "grad_norm": 1.9934879541397095, + "learning_rate": 1e-06, + "loss": 0.5598, + "mean_token_accuracy": 0.8050937652587891, + "num_tokens": 841235425.0, + "step": 31645 + }, + { + "epoch": 2.609548940380968, + "grad_norm": 2.2183890342712402, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7565825581550598, + "num_tokens": 841263859.0, + "step": 31646 + }, + { + "epoch": 2.6096314010060198, + "grad_norm": 2.489678144454956, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7654833793640137, + "num_tokens": 841287034.0, + "step": 31647 + }, + { + "epoch": 2.609713861631071, + "grad_norm": 2.170278310775757, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.7857009172439575, + "num_tokens": 841312884.0, + "step": 31648 + }, + { + "epoch": 2.6097963222561225, + "grad_norm": 2.1216397285461426, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7571290135383606, + "num_tokens": 841344657.0, + "step": 31649 + }, + { + "epoch": 2.6098787828811743, + "grad_norm": 2.0986859798431396, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.755408525466919, + "num_tokens": 841375979.0, + "step": 31650 + }, + { + "epoch": 2.609961243506226, + "grad_norm": 1.9170186519622803, + "learning_rate": 1e-06, + "loss": 0.5947, + "mean_token_accuracy": 0.7928870320320129, + "num_tokens": 841409145.0, + "step": 31651 + }, + { + "epoch": 2.6100437041312774, + "grad_norm": 2.17278790473938, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.767277717590332, + "num_tokens": 841436531.0, + "step": 31652 + }, + { + "epoch": 2.6101261647563287, + "grad_norm": 2.1310062408447266, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7791650295257568, + "num_tokens": 841463885.0, + "step": 31653 + }, + { + "epoch": 2.6102086253813805, + "grad_norm": 2.0526413917541504, + "learning_rate": 1e-06, + "loss": 0.7145, + "mean_token_accuracy": 0.7640448808670044, + "num_tokens": 841497321.0, + "step": 31654 + }, + { + "epoch": 2.610291086006432, + "grad_norm": 2.1454315185546875, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.7828470468521118, + "num_tokens": 841526296.0, + "step": 31655 + }, + { + "epoch": 2.6103735466314832, + "grad_norm": 1.9460657835006714, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7708624005317688, + "num_tokens": 841559101.0, + "step": 31656 + }, + { + "epoch": 2.610456007256535, + "grad_norm": 2.7155017852783203, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7356917858123779, + "num_tokens": 841580128.0, + "step": 31657 + }, + { + "epoch": 2.6105384678815864, + "grad_norm": 2.174239158630371, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.8023970723152161, + "num_tokens": 841606318.0, + "step": 31658 + }, + { + "epoch": 2.610620928506638, + "grad_norm": 2.269286632537842, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.770995020866394, + "num_tokens": 841633134.0, + "step": 31659 + }, + { + "epoch": 2.6107033891316895, + "grad_norm": 1.9265879392623901, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.7921673059463501, + "num_tokens": 841668561.0, + "step": 31660 + }, + { + "epoch": 2.6107858497567413, + "grad_norm": 2.18636417388916, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7659487724304199, + "num_tokens": 841699959.0, + "step": 31661 + }, + { + "epoch": 2.6108683103817927, + "grad_norm": 2.3014140129089355, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7910494208335876, + "num_tokens": 841726983.0, + "step": 31662 + }, + { + "epoch": 2.610950771006844, + "grad_norm": 2.1830995082855225, + "learning_rate": 1e-06, + "loss": 0.5825, + "mean_token_accuracy": 0.8070451021194458, + "num_tokens": 841755703.0, + "step": 31663 + }, + { + "epoch": 2.611033231631896, + "grad_norm": 2.0111682415008545, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7676525712013245, + "num_tokens": 841789497.0, + "step": 31664 + }, + { + "epoch": 2.611115692256947, + "grad_norm": 2.085606813430786, + "learning_rate": 1e-06, + "loss": 0.5868, + "mean_token_accuracy": 0.7960413098335266, + "num_tokens": 841821071.0, + "step": 31665 + }, + { + "epoch": 2.611198152881999, + "grad_norm": 2.2427475452423096, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7453005909919739, + "num_tokens": 841849941.0, + "step": 31666 + }, + { + "epoch": 2.6112806135070503, + "grad_norm": 2.1547181606292725, + "learning_rate": 1e-06, + "loss": 0.6566, + "mean_token_accuracy": 0.7808554768562317, + "num_tokens": 841878821.0, + "step": 31667 + }, + { + "epoch": 2.611363074132102, + "grad_norm": 1.8630015850067139, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7744259834289551, + "num_tokens": 841916009.0, + "step": 31668 + }, + { + "epoch": 2.6114455347571535, + "grad_norm": 2.021411180496216, + "learning_rate": 1e-06, + "loss": 0.5366, + "mean_token_accuracy": 0.8046194911003113, + "num_tokens": 841944389.0, + "step": 31669 + }, + { + "epoch": 2.611527995382205, + "grad_norm": 2.314298629760742, + "learning_rate": 1e-06, + "loss": 0.8193, + "mean_token_accuracy": 0.7425740957260132, + "num_tokens": 841971544.0, + "step": 31670 + }, + { + "epoch": 2.6116104560072566, + "grad_norm": 2.2517759799957275, + "learning_rate": 1e-06, + "loss": 0.6131, + "mean_token_accuracy": 0.8028427362442017, + "num_tokens": 841999606.0, + "step": 31671 + }, + { + "epoch": 2.611692916632308, + "grad_norm": 2.0345048904418945, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7575277090072632, + "num_tokens": 842030978.0, + "step": 31672 + }, + { + "epoch": 2.6117753772573598, + "grad_norm": 2.3280324935913086, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.7943376302719116, + "num_tokens": 842056017.0, + "step": 31673 + }, + { + "epoch": 2.611857837882411, + "grad_norm": 2.1383023262023926, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7510980367660522, + "num_tokens": 842084714.0, + "step": 31674 + }, + { + "epoch": 2.611940298507463, + "grad_norm": 2.3408186435699463, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7671284079551697, + "num_tokens": 842112892.0, + "step": 31675 + }, + { + "epoch": 2.6120227591325142, + "grad_norm": 2.143739700317383, + "learning_rate": 1e-06, + "loss": 0.7132, + "mean_token_accuracy": 0.7657294869422913, + "num_tokens": 842142169.0, + "step": 31676 + }, + { + "epoch": 2.6121052197575656, + "grad_norm": 2.5397369861602783, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7275512218475342, + "num_tokens": 842164895.0, + "step": 31677 + }, + { + "epoch": 2.6121876803826174, + "grad_norm": 2.082179307937622, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7229946851730347, + "num_tokens": 842196332.0, + "step": 31678 + }, + { + "epoch": 2.6122701410076687, + "grad_norm": 1.941278338432312, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.764849066734314, + "num_tokens": 842231202.0, + "step": 31679 + }, + { + "epoch": 2.6123526016327205, + "grad_norm": 2.1865787506103516, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7565782070159912, + "num_tokens": 842259367.0, + "step": 31680 + }, + { + "epoch": 2.612435062257772, + "grad_norm": 2.1382944583892822, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.7841048240661621, + "num_tokens": 842286860.0, + "step": 31681 + }, + { + "epoch": 2.6125175228828237, + "grad_norm": 2.235463857650757, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7664605379104614, + "num_tokens": 842314793.0, + "step": 31682 + }, + { + "epoch": 2.612599983507875, + "grad_norm": 2.01349139213562, + "learning_rate": 1e-06, + "loss": 0.5488, + "mean_token_accuracy": 0.8097498416900635, + "num_tokens": 842344168.0, + "step": 31683 + }, + { + "epoch": 2.6126824441329264, + "grad_norm": 2.2649285793304443, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.7800036668777466, + "num_tokens": 842369969.0, + "step": 31684 + }, + { + "epoch": 2.612764904757978, + "grad_norm": 2.3721911907196045, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7544429302215576, + "num_tokens": 842397865.0, + "step": 31685 + }, + { + "epoch": 2.6128473653830295, + "grad_norm": 2.2154178619384766, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7753482460975647, + "num_tokens": 842427024.0, + "step": 31686 + }, + { + "epoch": 2.612929826008081, + "grad_norm": 2.5483901500701904, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.7898527383804321, + "num_tokens": 842452970.0, + "step": 31687 + }, + { + "epoch": 2.6130122866331327, + "grad_norm": 2.170092821121216, + "learning_rate": 1e-06, + "loss": 0.7575, + "mean_token_accuracy": 0.7589379549026489, + "num_tokens": 842481339.0, + "step": 31688 + }, + { + "epoch": 2.6130947472581845, + "grad_norm": 2.0376851558685303, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7615141868591309, + "num_tokens": 842514009.0, + "step": 31689 + }, + { + "epoch": 2.613177207883236, + "grad_norm": 2.054595470428467, + "learning_rate": 1e-06, + "loss": 0.673, + "mean_token_accuracy": 0.7866250276565552, + "num_tokens": 842545684.0, + "step": 31690 + }, + { + "epoch": 2.613259668508287, + "grad_norm": 2.129836320877075, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7804841995239258, + "num_tokens": 842572544.0, + "step": 31691 + }, + { + "epoch": 2.613342129133339, + "grad_norm": 2.0455992221832275, + "learning_rate": 1e-06, + "loss": 0.6951, + "mean_token_accuracy": 0.7767862677574158, + "num_tokens": 842604786.0, + "step": 31692 + }, + { + "epoch": 2.6134245897583903, + "grad_norm": 2.081587553024292, + "learning_rate": 1e-06, + "loss": 0.6958, + "mean_token_accuracy": 0.7761883735656738, + "num_tokens": 842635089.0, + "step": 31693 + }, + { + "epoch": 2.6135070503834417, + "grad_norm": 2.0846893787384033, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7580761909484863, + "num_tokens": 842666791.0, + "step": 31694 + }, + { + "epoch": 2.6135895110084935, + "grad_norm": 2.243570566177368, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7865648865699768, + "num_tokens": 842694550.0, + "step": 31695 + }, + { + "epoch": 2.6136719716335453, + "grad_norm": 2.145542860031128, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7766519784927368, + "num_tokens": 842726322.0, + "step": 31696 + }, + { + "epoch": 2.6137544322585966, + "grad_norm": 2.1375489234924316, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.7924121618270874, + "num_tokens": 842754130.0, + "step": 31697 + }, + { + "epoch": 2.613836892883648, + "grad_norm": 2.1298062801361084, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.7695732712745667, + "num_tokens": 842786398.0, + "step": 31698 + }, + { + "epoch": 2.6139193535086997, + "grad_norm": 2.0569300651550293, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.7901417016983032, + "num_tokens": 842818507.0, + "step": 31699 + }, + { + "epoch": 2.614001814133751, + "grad_norm": 2.056068181991577, + "learning_rate": 1e-06, + "loss": 0.5939, + "mean_token_accuracy": 0.80528724193573, + "num_tokens": 842846185.0, + "step": 31700 + }, + { + "epoch": 2.6140842747588025, + "grad_norm": 2.2541675567626953, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.7953746318817139, + "num_tokens": 842872872.0, + "step": 31701 + }, + { + "epoch": 2.6141667353838542, + "grad_norm": 2.5828561782836914, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7169271111488342, + "num_tokens": 842895319.0, + "step": 31702 + }, + { + "epoch": 2.6142491960089056, + "grad_norm": 1.8601256608963013, + "learning_rate": 1e-06, + "loss": 0.5567, + "mean_token_accuracy": 0.8134654760360718, + "num_tokens": 842928124.0, + "step": 31703 + }, + { + "epoch": 2.6143316566339574, + "grad_norm": 2.176920175552368, + "learning_rate": 1e-06, + "loss": 0.6309, + "mean_token_accuracy": 0.7902530431747437, + "num_tokens": 842957586.0, + "step": 31704 + }, + { + "epoch": 2.6144141172590087, + "grad_norm": 1.988247036933899, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.8049564361572266, + "num_tokens": 842989380.0, + "step": 31705 + }, + { + "epoch": 2.6144965778840605, + "grad_norm": 2.454066514968872, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7744682431221008, + "num_tokens": 843012755.0, + "step": 31706 + }, + { + "epoch": 2.614579038509112, + "grad_norm": 1.9297674894332886, + "learning_rate": 1e-06, + "loss": 0.4972, + "mean_token_accuracy": 0.8361776471138, + "num_tokens": 843046945.0, + "step": 31707 + }, + { + "epoch": 2.6146614991341632, + "grad_norm": 1.7915829420089722, + "learning_rate": 1e-06, + "loss": 0.5923, + "mean_token_accuracy": 0.7983320951461792, + "num_tokens": 843085822.0, + "step": 31708 + }, + { + "epoch": 2.614743959759215, + "grad_norm": 2.0944149494171143, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.7859379053115845, + "num_tokens": 843114885.0, + "step": 31709 + }, + { + "epoch": 2.6148264203842664, + "grad_norm": 2.2873497009277344, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7636524438858032, + "num_tokens": 843143335.0, + "step": 31710 + }, + { + "epoch": 2.614908881009318, + "grad_norm": 2.1355390548706055, + "learning_rate": 1e-06, + "loss": 0.6689, + "mean_token_accuracy": 0.7776569128036499, + "num_tokens": 843172596.0, + "step": 31711 + }, + { + "epoch": 2.6149913416343695, + "grad_norm": 2.0349178314208984, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7751451730728149, + "num_tokens": 843205735.0, + "step": 31712 + }, + { + "epoch": 2.6150738022594213, + "grad_norm": 1.984990119934082, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7753803730010986, + "num_tokens": 843239315.0, + "step": 31713 + }, + { + "epoch": 2.6151562628844727, + "grad_norm": 2.3404035568237305, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7351232767105103, + "num_tokens": 843266524.0, + "step": 31714 + }, + { + "epoch": 2.615238723509524, + "grad_norm": 1.8798094987869263, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.7596427202224731, + "num_tokens": 843304555.0, + "step": 31715 + }, + { + "epoch": 2.615321184134576, + "grad_norm": 2.393432140350342, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7334264516830444, + "num_tokens": 843328567.0, + "step": 31716 + }, + { + "epoch": 2.615403644759627, + "grad_norm": 1.8110417127609253, + "learning_rate": 1e-06, + "loss": 0.6375, + "mean_token_accuracy": 0.7940752506256104, + "num_tokens": 843366275.0, + "step": 31717 + }, + { + "epoch": 2.615486105384679, + "grad_norm": 1.9923176765441895, + "learning_rate": 1e-06, + "loss": 0.6086, + "mean_token_accuracy": 0.8022030591964722, + "num_tokens": 843397451.0, + "step": 31718 + }, + { + "epoch": 2.6155685660097303, + "grad_norm": 2.190460443496704, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7692943811416626, + "num_tokens": 843426667.0, + "step": 31719 + }, + { + "epoch": 2.615651026634782, + "grad_norm": 2.2701897621154785, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.7445151805877686, + "num_tokens": 843451809.0, + "step": 31720 + }, + { + "epoch": 2.6157334872598335, + "grad_norm": 2.284783363342285, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.7886287569999695, + "num_tokens": 843477571.0, + "step": 31721 + }, + { + "epoch": 2.615815947884885, + "grad_norm": 2.0456979274749756, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7751407623291016, + "num_tokens": 843509891.0, + "step": 31722 + }, + { + "epoch": 2.6158984085099366, + "grad_norm": 2.948076009750366, + "learning_rate": 1e-06, + "loss": 0.7038, + "mean_token_accuracy": 0.7712407112121582, + "num_tokens": 843531143.0, + "step": 31723 + }, + { + "epoch": 2.615980869134988, + "grad_norm": 2.039172887802124, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7840861082077026, + "num_tokens": 843567403.0, + "step": 31724 + }, + { + "epoch": 2.6160633297600397, + "grad_norm": 2.25807523727417, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7736660838127136, + "num_tokens": 843593018.0, + "step": 31725 + }, + { + "epoch": 2.616145790385091, + "grad_norm": 1.905534029006958, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.7862527966499329, + "num_tokens": 843625114.0, + "step": 31726 + }, + { + "epoch": 2.616228251010143, + "grad_norm": 2.2675552368164062, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7918685674667358, + "num_tokens": 843653306.0, + "step": 31727 + }, + { + "epoch": 2.6163107116351942, + "grad_norm": 1.8063846826553345, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7731921672821045, + "num_tokens": 843688891.0, + "step": 31728 + }, + { + "epoch": 2.6163931722602456, + "grad_norm": 2.2131729125976562, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7660344839096069, + "num_tokens": 843716895.0, + "step": 31729 + }, + { + "epoch": 2.6164756328852974, + "grad_norm": 2.268763542175293, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7524380683898926, + "num_tokens": 843743058.0, + "step": 31730 + }, + { + "epoch": 2.6165580935103487, + "grad_norm": 1.7784407138824463, + "learning_rate": 1e-06, + "loss": 0.6137, + "mean_token_accuracy": 0.7887442111968994, + "num_tokens": 843781002.0, + "step": 31731 + }, + { + "epoch": 2.6166405541354, + "grad_norm": 2.2486324310302734, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.753089189529419, + "num_tokens": 843809667.0, + "step": 31732 + }, + { + "epoch": 2.616723014760452, + "grad_norm": 2.0636508464813232, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7412540316581726, + "num_tokens": 843842079.0, + "step": 31733 + }, + { + "epoch": 2.6168054753855037, + "grad_norm": 1.7987868785858154, + "learning_rate": 1e-06, + "loss": 0.5269, + "mean_token_accuracy": 0.8200743794441223, + "num_tokens": 843880798.0, + "step": 31734 + }, + { + "epoch": 2.616887936010555, + "grad_norm": 2.0495762825012207, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7730712890625, + "num_tokens": 843914765.0, + "step": 31735 + }, + { + "epoch": 2.6169703966356064, + "grad_norm": 2.1094181537628174, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7748867273330688, + "num_tokens": 843944386.0, + "step": 31736 + }, + { + "epoch": 2.617052857260658, + "grad_norm": 2.2546563148498535, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7654579877853394, + "num_tokens": 843971491.0, + "step": 31737 + }, + { + "epoch": 2.6171353178857095, + "grad_norm": 1.9448472261428833, + "learning_rate": 1e-06, + "loss": 0.5988, + "mean_token_accuracy": 0.8022366762161255, + "num_tokens": 844011252.0, + "step": 31738 + }, + { + "epoch": 2.617217778510761, + "grad_norm": 2.1842758655548096, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.8039954900741577, + "num_tokens": 844036940.0, + "step": 31739 + }, + { + "epoch": 2.6173002391358127, + "grad_norm": 2.1024224758148193, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7771693468093872, + "num_tokens": 844065753.0, + "step": 31740 + }, + { + "epoch": 2.6173826997608645, + "grad_norm": 1.738752007484436, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.7902480363845825, + "num_tokens": 844108425.0, + "step": 31741 + }, + { + "epoch": 2.617465160385916, + "grad_norm": 2.1369056701660156, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.7779925465583801, + "num_tokens": 844134484.0, + "step": 31742 + }, + { + "epoch": 2.617547621010967, + "grad_norm": 2.0589964389801025, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7691407203674316, + "num_tokens": 844169135.0, + "step": 31743 + }, + { + "epoch": 2.617630081636019, + "grad_norm": 2.0042243003845215, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.7534394264221191, + "num_tokens": 844202475.0, + "step": 31744 + }, + { + "epoch": 2.6177125422610703, + "grad_norm": 1.804196834564209, + "learning_rate": 1e-06, + "loss": 0.5871, + "mean_token_accuracy": 0.8026092648506165, + "num_tokens": 844238933.0, + "step": 31745 + }, + { + "epoch": 2.6177950028861217, + "grad_norm": 2.104279041290283, + "learning_rate": 1e-06, + "loss": 0.5749, + "mean_token_accuracy": 0.8059080243110657, + "num_tokens": 844267599.0, + "step": 31746 + }, + { + "epoch": 2.6178774635111735, + "grad_norm": 2.1316049098968506, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7583317756652832, + "num_tokens": 844296301.0, + "step": 31747 + }, + { + "epoch": 2.617959924136225, + "grad_norm": 2.0504415035247803, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7709232568740845, + "num_tokens": 844326445.0, + "step": 31748 + }, + { + "epoch": 2.6180423847612766, + "grad_norm": 2.066746234893799, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7569675445556641, + "num_tokens": 844356336.0, + "step": 31749 + }, + { + "epoch": 2.618124845386328, + "grad_norm": 2.342775583267212, + "learning_rate": 1e-06, + "loss": 0.5868, + "mean_token_accuracy": 0.8016378879547119, + "num_tokens": 844379386.0, + "step": 31750 + }, + { + "epoch": 2.6182073060113797, + "grad_norm": 1.887668490409851, + "learning_rate": 1e-06, + "loss": 0.5116, + "mean_token_accuracy": 0.8205051422119141, + "num_tokens": 844411128.0, + "step": 31751 + }, + { + "epoch": 2.618289766636431, + "grad_norm": 2.2073874473571777, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7554609775543213, + "num_tokens": 844440516.0, + "step": 31752 + }, + { + "epoch": 2.6183722272614824, + "grad_norm": 1.944836139678955, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7599831223487854, + "num_tokens": 844474086.0, + "step": 31753 + }, + { + "epoch": 2.6184546878865342, + "grad_norm": 2.4562408924102783, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.7966720461845398, + "num_tokens": 844496641.0, + "step": 31754 + }, + { + "epoch": 2.6185371485115856, + "grad_norm": 2.2830259799957275, + "learning_rate": 1e-06, + "loss": 0.6649, + "mean_token_accuracy": 0.780176043510437, + "num_tokens": 844521897.0, + "step": 31755 + }, + { + "epoch": 2.6186196091366374, + "grad_norm": 2.3869988918304443, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7098326683044434, + "num_tokens": 844547765.0, + "step": 31756 + }, + { + "epoch": 2.6187020697616887, + "grad_norm": 2.1606404781341553, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7794170379638672, + "num_tokens": 844577496.0, + "step": 31757 + }, + { + "epoch": 2.6187845303867405, + "grad_norm": 1.868283748626709, + "learning_rate": 1e-06, + "loss": 0.6543, + "mean_token_accuracy": 0.7860976457595825, + "num_tokens": 844612672.0, + "step": 31758 + }, + { + "epoch": 2.618866991011792, + "grad_norm": 1.9610161781311035, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.7812384366989136, + "num_tokens": 844642108.0, + "step": 31759 + }, + { + "epoch": 2.6189494516368432, + "grad_norm": 1.8922839164733887, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.775775134563446, + "num_tokens": 844676022.0, + "step": 31760 + }, + { + "epoch": 2.619031912261895, + "grad_norm": 2.0341503620147705, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7418919205665588, + "num_tokens": 844710340.0, + "step": 31761 + }, + { + "epoch": 2.6191143728869464, + "grad_norm": 2.317939043045044, + "learning_rate": 1e-06, + "loss": 0.7056, + "mean_token_accuracy": 0.772346019744873, + "num_tokens": 844737097.0, + "step": 31762 + }, + { + "epoch": 2.619196833511998, + "grad_norm": 2.0380725860595703, + "learning_rate": 1e-06, + "loss": 0.5995, + "mean_token_accuracy": 0.7976922392845154, + "num_tokens": 844766975.0, + "step": 31763 + }, + { + "epoch": 2.6192792941370495, + "grad_norm": 1.8852633237838745, + "learning_rate": 1e-06, + "loss": 0.5595, + "mean_token_accuracy": 0.8040853142738342, + "num_tokens": 844800389.0, + "step": 31764 + }, + { + "epoch": 2.6193617547621013, + "grad_norm": 2.1483314037323, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7627166509628296, + "num_tokens": 844830575.0, + "step": 31765 + }, + { + "epoch": 2.6194442153871527, + "grad_norm": 1.7057543992996216, + "learning_rate": 1e-06, + "loss": 0.5819, + "mean_token_accuracy": 0.7965519428253174, + "num_tokens": 844870002.0, + "step": 31766 + }, + { + "epoch": 2.619526676012204, + "grad_norm": 2.078047513961792, + "learning_rate": 1e-06, + "loss": 0.6151, + "mean_token_accuracy": 0.7888846397399902, + "num_tokens": 844899241.0, + "step": 31767 + }, + { + "epoch": 2.619609136637256, + "grad_norm": 2.1074957847595215, + "learning_rate": 1e-06, + "loss": 0.7671, + "mean_token_accuracy": 0.7521411180496216, + "num_tokens": 844927015.0, + "step": 31768 + }, + { + "epoch": 2.619691597262307, + "grad_norm": 1.922516107559204, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7725762724876404, + "num_tokens": 844962613.0, + "step": 31769 + }, + { + "epoch": 2.6197740578873585, + "grad_norm": 2.183255434036255, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7786167860031128, + "num_tokens": 844990382.0, + "step": 31770 + }, + { + "epoch": 2.6198565185124103, + "grad_norm": 1.8576231002807617, + "learning_rate": 1e-06, + "loss": 0.4678, + "mean_token_accuracy": 0.8361977338790894, + "num_tokens": 845021695.0, + "step": 31771 + }, + { + "epoch": 2.619938979137462, + "grad_norm": 2.0198581218719482, + "learning_rate": 1e-06, + "loss": 0.5898, + "mean_token_accuracy": 0.8061293959617615, + "num_tokens": 845054502.0, + "step": 31772 + }, + { + "epoch": 2.6200214397625134, + "grad_norm": 2.0483973026275635, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7712624669075012, + "num_tokens": 845088052.0, + "step": 31773 + }, + { + "epoch": 2.620103900387565, + "grad_norm": 2.023020029067993, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7858484387397766, + "num_tokens": 845119436.0, + "step": 31774 + }, + { + "epoch": 2.6201863610126166, + "grad_norm": 2.0714402198791504, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.7814180254936218, + "num_tokens": 845154864.0, + "step": 31775 + }, + { + "epoch": 2.620268821637668, + "grad_norm": 1.9239450693130493, + "learning_rate": 1e-06, + "loss": 0.5661, + "mean_token_accuracy": 0.8105261921882629, + "num_tokens": 845185953.0, + "step": 31776 + }, + { + "epoch": 2.6203512822627193, + "grad_norm": 2.1713802814483643, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.7950385808944702, + "num_tokens": 845212720.0, + "step": 31777 + }, + { + "epoch": 2.620433742887771, + "grad_norm": 2.3106377124786377, + "learning_rate": 1e-06, + "loss": 0.6111, + "mean_token_accuracy": 0.7920311689376831, + "num_tokens": 845238725.0, + "step": 31778 + }, + { + "epoch": 2.620516203512823, + "grad_norm": 2.1315784454345703, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7815512418746948, + "num_tokens": 845267007.0, + "step": 31779 + }, + { + "epoch": 2.6205986641378742, + "grad_norm": 2.1125311851501465, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7385833859443665, + "num_tokens": 845297260.0, + "step": 31780 + }, + { + "epoch": 2.6206811247629256, + "grad_norm": 1.8966351747512817, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7382004261016846, + "num_tokens": 845335292.0, + "step": 31781 + }, + { + "epoch": 2.6207635853879774, + "grad_norm": 1.9124152660369873, + "learning_rate": 1e-06, + "loss": 0.6441, + "mean_token_accuracy": 0.7868481874465942, + "num_tokens": 845370313.0, + "step": 31782 + }, + { + "epoch": 2.6208460460130287, + "grad_norm": 2.301154851913452, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7444750070571899, + "num_tokens": 845396854.0, + "step": 31783 + }, + { + "epoch": 2.62092850663808, + "grad_norm": 2.0354912281036377, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7766967415809631, + "num_tokens": 845428467.0, + "step": 31784 + }, + { + "epoch": 2.621010967263132, + "grad_norm": 1.9677493572235107, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7772592306137085, + "num_tokens": 845463911.0, + "step": 31785 + }, + { + "epoch": 2.6210934278881837, + "grad_norm": 2.204482316970825, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.7862651348114014, + "num_tokens": 845489476.0, + "step": 31786 + }, + { + "epoch": 2.621175888513235, + "grad_norm": 2.577261209487915, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7377145290374756, + "num_tokens": 845512486.0, + "step": 31787 + }, + { + "epoch": 2.6212583491382864, + "grad_norm": 2.334005355834961, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7317364811897278, + "num_tokens": 845539304.0, + "step": 31788 + }, + { + "epoch": 2.621340809763338, + "grad_norm": 2.2676899433135986, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.735824465751648, + "num_tokens": 845568209.0, + "step": 31789 + }, + { + "epoch": 2.6214232703883895, + "grad_norm": 2.16188645362854, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7718377709388733, + "num_tokens": 845600347.0, + "step": 31790 + }, + { + "epoch": 2.621505731013441, + "grad_norm": 2.3022212982177734, + "learning_rate": 1e-06, + "loss": 0.7413, + "mean_token_accuracy": 0.7545608282089233, + "num_tokens": 845626023.0, + "step": 31791 + }, + { + "epoch": 2.6215881916384927, + "grad_norm": 2.043062210083008, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.7633702158927917, + "num_tokens": 845657874.0, + "step": 31792 + }, + { + "epoch": 2.621670652263544, + "grad_norm": 2.23500919342041, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7354768514633179, + "num_tokens": 845685762.0, + "step": 31793 + }, + { + "epoch": 2.621753112888596, + "grad_norm": 2.155181407928467, + "learning_rate": 1e-06, + "loss": 0.6075, + "mean_token_accuracy": 0.7986576557159424, + "num_tokens": 845715096.0, + "step": 31794 + }, + { + "epoch": 2.621835573513647, + "grad_norm": 2.1404364109039307, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7795446515083313, + "num_tokens": 845745127.0, + "step": 31795 + }, + { + "epoch": 2.621918034138699, + "grad_norm": 1.9231468439102173, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7807393074035645, + "num_tokens": 845779771.0, + "step": 31796 + }, + { + "epoch": 2.6220004947637503, + "grad_norm": 2.1506099700927734, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7451022863388062, + "num_tokens": 845813869.0, + "step": 31797 + }, + { + "epoch": 2.6220829553888017, + "grad_norm": 2.197603940963745, + "learning_rate": 1e-06, + "loss": 0.5856, + "mean_token_accuracy": 0.7994757890701294, + "num_tokens": 845841920.0, + "step": 31798 + }, + { + "epoch": 2.6221654160138534, + "grad_norm": 1.9580910205841064, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7460497617721558, + "num_tokens": 845876252.0, + "step": 31799 + }, + { + "epoch": 2.622247876638905, + "grad_norm": 2.501481294631958, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.761555552482605, + "num_tokens": 845901370.0, + "step": 31800 + }, + { + "epoch": 2.6223303372639566, + "grad_norm": 2.080622434616089, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7866957187652588, + "num_tokens": 845928339.0, + "step": 31801 + }, + { + "epoch": 2.622412797889008, + "grad_norm": 2.5410587787628174, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.75217205286026, + "num_tokens": 845952280.0, + "step": 31802 + }, + { + "epoch": 2.6224952585140597, + "grad_norm": 1.9684338569641113, + "learning_rate": 1e-06, + "loss": 0.5754, + "mean_token_accuracy": 0.7926790714263916, + "num_tokens": 845985807.0, + "step": 31803 + }, + { + "epoch": 2.622577719139111, + "grad_norm": 1.9068288803100586, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.7994120717048645, + "num_tokens": 846019307.0, + "step": 31804 + }, + { + "epoch": 2.6226601797641624, + "grad_norm": 2.0607144832611084, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7620154619216919, + "num_tokens": 846050530.0, + "step": 31805 + }, + { + "epoch": 2.6227426403892142, + "grad_norm": 2.178741693496704, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7611820101737976, + "num_tokens": 846079530.0, + "step": 31806 + }, + { + "epoch": 2.6228251010142656, + "grad_norm": 1.9220542907714844, + "learning_rate": 1e-06, + "loss": 0.5897, + "mean_token_accuracy": 0.8051872849464417, + "num_tokens": 846116710.0, + "step": 31807 + }, + { + "epoch": 2.6229075616393174, + "grad_norm": 2.1367154121398926, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.7760323286056519, + "num_tokens": 846143574.0, + "step": 31808 + }, + { + "epoch": 2.6229900222643687, + "grad_norm": 1.9723491668701172, + "learning_rate": 1e-06, + "loss": 0.5759, + "mean_token_accuracy": 0.8091408014297485, + "num_tokens": 846173995.0, + "step": 31809 + }, + { + "epoch": 2.6230724828894205, + "grad_norm": 1.993226170539856, + "learning_rate": 1e-06, + "loss": 0.5887, + "mean_token_accuracy": 0.8040049076080322, + "num_tokens": 846205462.0, + "step": 31810 + }, + { + "epoch": 2.623154943514472, + "grad_norm": 2.470327377319336, + "learning_rate": 1e-06, + "loss": 0.6169, + "mean_token_accuracy": 0.789713442325592, + "num_tokens": 846227394.0, + "step": 31811 + }, + { + "epoch": 2.623237404139523, + "grad_norm": 2.4088563919067383, + "learning_rate": 1e-06, + "loss": 0.7781, + "mean_token_accuracy": 0.7432578802108765, + "num_tokens": 846251394.0, + "step": 31812 + }, + { + "epoch": 2.623319864764575, + "grad_norm": 1.9560996294021606, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.756432056427002, + "num_tokens": 846286571.0, + "step": 31813 + }, + { + "epoch": 2.6234023253896264, + "grad_norm": 1.9885751008987427, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7596909403800964, + "num_tokens": 846321094.0, + "step": 31814 + }, + { + "epoch": 2.6234847860146777, + "grad_norm": 2.0196385383605957, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7683582901954651, + "num_tokens": 846351333.0, + "step": 31815 + }, + { + "epoch": 2.6235672466397295, + "grad_norm": 1.9452911615371704, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.7980334758758545, + "num_tokens": 846385345.0, + "step": 31816 + }, + { + "epoch": 2.6236497072647813, + "grad_norm": 1.9366576671600342, + "learning_rate": 1e-06, + "loss": 0.5278, + "mean_token_accuracy": 0.8212710022926331, + "num_tokens": 846416072.0, + "step": 31817 + }, + { + "epoch": 2.6237321678898327, + "grad_norm": 2.3654427528381348, + "learning_rate": 1e-06, + "loss": 0.7563, + "mean_token_accuracy": 0.7502977252006531, + "num_tokens": 846440644.0, + "step": 31818 + }, + { + "epoch": 2.623814628514884, + "grad_norm": 2.185842514038086, + "learning_rate": 1e-06, + "loss": 0.7245, + "mean_token_accuracy": 0.7645031213760376, + "num_tokens": 846470113.0, + "step": 31819 + }, + { + "epoch": 2.623897089139936, + "grad_norm": 2.209489345550537, + "learning_rate": 1e-06, + "loss": 0.5786, + "mean_token_accuracy": 0.8121838569641113, + "num_tokens": 846499717.0, + "step": 31820 + }, + { + "epoch": 2.623979549764987, + "grad_norm": 2.1900525093078613, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7391566038131714, + "num_tokens": 846534036.0, + "step": 31821 + }, + { + "epoch": 2.6240620103900385, + "grad_norm": 2.372032403945923, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7556985020637512, + "num_tokens": 846558073.0, + "step": 31822 + }, + { + "epoch": 2.6241444710150903, + "grad_norm": 2.091416597366333, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7702733874320984, + "num_tokens": 846588715.0, + "step": 31823 + }, + { + "epoch": 2.624226931640142, + "grad_norm": 1.9453705549240112, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.8030611872673035, + "num_tokens": 846623346.0, + "step": 31824 + }, + { + "epoch": 2.6243093922651934, + "grad_norm": 2.085918664932251, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7540162801742554, + "num_tokens": 846656058.0, + "step": 31825 + }, + { + "epoch": 2.624391852890245, + "grad_norm": 2.1025731563568115, + "learning_rate": 1e-06, + "loss": 0.5968, + "mean_token_accuracy": 0.7974763512611389, + "num_tokens": 846686324.0, + "step": 31826 + }, + { + "epoch": 2.6244743135152966, + "grad_norm": 1.994119644165039, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7528372406959534, + "num_tokens": 846719055.0, + "step": 31827 + }, + { + "epoch": 2.624556774140348, + "grad_norm": 2.04129958152771, + "learning_rate": 1e-06, + "loss": 0.5854, + "mean_token_accuracy": 0.7997987270355225, + "num_tokens": 846749175.0, + "step": 31828 + }, + { + "epoch": 2.6246392347653993, + "grad_norm": 2.0497243404388428, + "learning_rate": 1e-06, + "loss": 0.594, + "mean_token_accuracy": 0.7966029644012451, + "num_tokens": 846779912.0, + "step": 31829 + }, + { + "epoch": 2.624721695390451, + "grad_norm": 2.291475534439087, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7348023653030396, + "num_tokens": 846805820.0, + "step": 31830 + }, + { + "epoch": 2.6248041560155024, + "grad_norm": 2.0104613304138184, + "learning_rate": 1e-06, + "loss": 0.5817, + "mean_token_accuracy": 0.8133410215377808, + "num_tokens": 846836173.0, + "step": 31831 + }, + { + "epoch": 2.6248866166405542, + "grad_norm": 2.5676369667053223, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7467632293701172, + "num_tokens": 846859574.0, + "step": 31832 + }, + { + "epoch": 2.6249690772656056, + "grad_norm": 1.9410948753356934, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7410138845443726, + "num_tokens": 846896100.0, + "step": 31833 + }, + { + "epoch": 2.6250515378906574, + "grad_norm": 1.9561599493026733, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.7842575311660767, + "num_tokens": 846927849.0, + "step": 31834 + }, + { + "epoch": 2.6251339985157087, + "grad_norm": 2.2445132732391357, + "learning_rate": 1e-06, + "loss": 0.6098, + "mean_token_accuracy": 0.7975802421569824, + "num_tokens": 846951579.0, + "step": 31835 + }, + { + "epoch": 2.62521645914076, + "grad_norm": 1.9118355512619019, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.7916966676712036, + "num_tokens": 846984914.0, + "step": 31836 + }, + { + "epoch": 2.625298919765812, + "grad_norm": 2.126844882965088, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7437729239463806, + "num_tokens": 847013947.0, + "step": 31837 + }, + { + "epoch": 2.625381380390863, + "grad_norm": 2.0316078662872314, + "learning_rate": 1e-06, + "loss": 0.6345, + "mean_token_accuracy": 0.7889043688774109, + "num_tokens": 847045447.0, + "step": 31838 + }, + { + "epoch": 2.625463841015915, + "grad_norm": 1.9601448774337769, + "learning_rate": 1e-06, + "loss": 0.697, + "mean_token_accuracy": 0.7726547718048096, + "num_tokens": 847078083.0, + "step": 31839 + }, + { + "epoch": 2.6255463016409664, + "grad_norm": 2.665870428085327, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7749334573745728, + "num_tokens": 847098847.0, + "step": 31840 + }, + { + "epoch": 2.625628762266018, + "grad_norm": 1.9422200918197632, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7744361162185669, + "num_tokens": 847132766.0, + "step": 31841 + }, + { + "epoch": 2.6257112228910695, + "grad_norm": 1.9631752967834473, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7759122848510742, + "num_tokens": 847166445.0, + "step": 31842 + }, + { + "epoch": 2.625793683516121, + "grad_norm": 2.2582499980926514, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.7539105415344238, + "num_tokens": 847197853.0, + "step": 31843 + }, + { + "epoch": 2.6258761441411727, + "grad_norm": 2.1306066513061523, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7507439255714417, + "num_tokens": 847230298.0, + "step": 31844 + }, + { + "epoch": 2.625958604766224, + "grad_norm": 1.8887461423873901, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.8083419799804688, + "num_tokens": 847263916.0, + "step": 31845 + }, + { + "epoch": 2.626041065391276, + "grad_norm": 2.1302380561828613, + "learning_rate": 1e-06, + "loss": 0.7698, + "mean_token_accuracy": 0.7517496347427368, + "num_tokens": 847293941.0, + "step": 31846 + }, + { + "epoch": 2.626123526016327, + "grad_norm": 2.105686664581299, + "learning_rate": 1e-06, + "loss": 0.5968, + "mean_token_accuracy": 0.7997989654541016, + "num_tokens": 847324693.0, + "step": 31847 + }, + { + "epoch": 2.626205986641379, + "grad_norm": 2.044588804244995, + "learning_rate": 1e-06, + "loss": 0.5979, + "mean_token_accuracy": 0.8015822172164917, + "num_tokens": 847356906.0, + "step": 31848 + }, + { + "epoch": 2.6262884472664303, + "grad_norm": 2.0442419052124023, + "learning_rate": 1e-06, + "loss": 0.5479, + "mean_token_accuracy": 0.8100611567497253, + "num_tokens": 847385409.0, + "step": 31849 + }, + { + "epoch": 2.6263709078914816, + "grad_norm": 1.9966776371002197, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7759147882461548, + "num_tokens": 847419568.0, + "step": 31850 + }, + { + "epoch": 2.6264533685165334, + "grad_norm": 2.234443187713623, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.7874459624290466, + "num_tokens": 847446472.0, + "step": 31851 + }, + { + "epoch": 2.626535829141585, + "grad_norm": 2.3812756538391113, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7416802644729614, + "num_tokens": 847471964.0, + "step": 31852 + }, + { + "epoch": 2.6266182897666366, + "grad_norm": 1.9360345602035522, + "learning_rate": 1e-06, + "loss": 0.4784, + "mean_token_accuracy": 0.829954206943512, + "num_tokens": 847502010.0, + "step": 31853 + }, + { + "epoch": 2.626700750391688, + "grad_norm": 2.1510324478149414, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7784715294837952, + "num_tokens": 847529461.0, + "step": 31854 + }, + { + "epoch": 2.6267832110167397, + "grad_norm": 1.9641286134719849, + "learning_rate": 1e-06, + "loss": 0.5839, + "mean_token_accuracy": 0.8087347149848938, + "num_tokens": 847559866.0, + "step": 31855 + }, + { + "epoch": 2.626865671641791, + "grad_norm": 2.148468017578125, + "learning_rate": 1e-06, + "loss": 0.7293, + "mean_token_accuracy": 0.7645142078399658, + "num_tokens": 847591378.0, + "step": 31856 + }, + { + "epoch": 2.6269481322668424, + "grad_norm": 2.0759832859039307, + "learning_rate": 1e-06, + "loss": 0.5792, + "mean_token_accuracy": 0.8032312393188477, + "num_tokens": 847618544.0, + "step": 31857 + }, + { + "epoch": 2.6270305928918942, + "grad_norm": 2.5813300609588623, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.758316695690155, + "num_tokens": 847640147.0, + "step": 31858 + }, + { + "epoch": 2.6271130535169456, + "grad_norm": 2.5289499759674072, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.7782799005508423, + "num_tokens": 847661346.0, + "step": 31859 + }, + { + "epoch": 2.627195514141997, + "grad_norm": 1.8817633390426636, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.7903063297271729, + "num_tokens": 847694385.0, + "step": 31860 + }, + { + "epoch": 2.6272779747670487, + "grad_norm": 2.2393903732299805, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7692227959632874, + "num_tokens": 847721394.0, + "step": 31861 + }, + { + "epoch": 2.6273604353921005, + "grad_norm": 1.9594447612762451, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7875950336456299, + "num_tokens": 847755782.0, + "step": 31862 + }, + { + "epoch": 2.627442896017152, + "grad_norm": 2.040438652038574, + "learning_rate": 1e-06, + "loss": 0.5452, + "mean_token_accuracy": 0.8128452301025391, + "num_tokens": 847783436.0, + "step": 31863 + }, + { + "epoch": 2.627525356642203, + "grad_norm": 1.8593997955322266, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.783128559589386, + "num_tokens": 847819293.0, + "step": 31864 + }, + { + "epoch": 2.627607817267255, + "grad_norm": 1.8742859363555908, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7742258906364441, + "num_tokens": 847853899.0, + "step": 31865 + }, + { + "epoch": 2.6276902778923064, + "grad_norm": 2.3522183895111084, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7532081604003906, + "num_tokens": 847880961.0, + "step": 31866 + }, + { + "epoch": 2.6277727385173577, + "grad_norm": 2.5585741996765137, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7569104433059692, + "num_tokens": 847902923.0, + "step": 31867 + }, + { + "epoch": 2.6278551991424095, + "grad_norm": 2.1283199787139893, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7525798082351685, + "num_tokens": 847931685.0, + "step": 31868 + }, + { + "epoch": 2.6279376597674613, + "grad_norm": 2.0802533626556396, + "learning_rate": 1e-06, + "loss": 0.6001, + "mean_token_accuracy": 0.7962805032730103, + "num_tokens": 847961976.0, + "step": 31869 + }, + { + "epoch": 2.6280201203925126, + "grad_norm": 1.970815896987915, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.780609130859375, + "num_tokens": 847991204.0, + "step": 31870 + }, + { + "epoch": 2.628102581017564, + "grad_norm": 1.7263792753219604, + "learning_rate": 1e-06, + "loss": 0.5394, + "mean_token_accuracy": 0.8155276775360107, + "num_tokens": 848028178.0, + "step": 31871 + }, + { + "epoch": 2.628185041642616, + "grad_norm": 1.9658414125442505, + "learning_rate": 1e-06, + "loss": 0.7365, + "mean_token_accuracy": 0.7589947581291199, + "num_tokens": 848060358.0, + "step": 31872 + }, + { + "epoch": 2.628267502267667, + "grad_norm": 1.9881746768951416, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7355687618255615, + "num_tokens": 848096826.0, + "step": 31873 + }, + { + "epoch": 2.6283499628927185, + "grad_norm": 1.836842656135559, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7572982311248779, + "num_tokens": 848133059.0, + "step": 31874 + }, + { + "epoch": 2.6284324235177703, + "grad_norm": 2.042299270629883, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.8005353808403015, + "num_tokens": 848163659.0, + "step": 31875 + }, + { + "epoch": 2.6285148841428216, + "grad_norm": 2.2707457542419434, + "learning_rate": 1e-06, + "loss": 0.5471, + "mean_token_accuracy": 0.8177956938743591, + "num_tokens": 848188691.0, + "step": 31876 + }, + { + "epoch": 2.6285973447678734, + "grad_norm": 2.1546480655670166, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7579456567764282, + "num_tokens": 848218260.0, + "step": 31877 + }, + { + "epoch": 2.628679805392925, + "grad_norm": 1.9325604438781738, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.758818507194519, + "num_tokens": 848253114.0, + "step": 31878 + }, + { + "epoch": 2.6287622660179766, + "grad_norm": 2.425830125808716, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.774017870426178, + "num_tokens": 848278647.0, + "step": 31879 + }, + { + "epoch": 2.628844726643028, + "grad_norm": 2.2473902702331543, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.7846839427947998, + "num_tokens": 848306168.0, + "step": 31880 + }, + { + "epoch": 2.6289271872680793, + "grad_norm": 1.8661580085754395, + "learning_rate": 1e-06, + "loss": 0.5995, + "mean_token_accuracy": 0.8047491908073425, + "num_tokens": 848340852.0, + "step": 31881 + }, + { + "epoch": 2.629009647893131, + "grad_norm": 2.3248391151428223, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.769583523273468, + "num_tokens": 848365873.0, + "step": 31882 + }, + { + "epoch": 2.6290921085181824, + "grad_norm": 2.0789644718170166, + "learning_rate": 1e-06, + "loss": 0.6394, + "mean_token_accuracy": 0.7912980318069458, + "num_tokens": 848394622.0, + "step": 31883 + }, + { + "epoch": 2.629174569143234, + "grad_norm": 2.1688427925109863, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7770739197731018, + "num_tokens": 848422782.0, + "step": 31884 + }, + { + "epoch": 2.6292570297682856, + "grad_norm": 2.0542709827423096, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7712126970291138, + "num_tokens": 848456505.0, + "step": 31885 + }, + { + "epoch": 2.6293394903933374, + "grad_norm": 2.125680923461914, + "learning_rate": 1e-06, + "loss": 0.5447, + "mean_token_accuracy": 0.8075466752052307, + "num_tokens": 848484812.0, + "step": 31886 + }, + { + "epoch": 2.6294219510183887, + "grad_norm": 1.9714049100875854, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7683574557304382, + "num_tokens": 848519240.0, + "step": 31887 + }, + { + "epoch": 2.62950441164344, + "grad_norm": 2.0809144973754883, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.7789915800094604, + "num_tokens": 848555001.0, + "step": 31888 + }, + { + "epoch": 2.629586872268492, + "grad_norm": 2.4386425018310547, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7530345320701599, + "num_tokens": 848580970.0, + "step": 31889 + }, + { + "epoch": 2.629669332893543, + "grad_norm": 1.8987172842025757, + "learning_rate": 1e-06, + "loss": 0.5593, + "mean_token_accuracy": 0.8099241256713867, + "num_tokens": 848613126.0, + "step": 31890 + }, + { + "epoch": 2.629751793518595, + "grad_norm": 1.940835952758789, + "learning_rate": 1e-06, + "loss": 0.6057, + "mean_token_accuracy": 0.8090989589691162, + "num_tokens": 848643031.0, + "step": 31891 + }, + { + "epoch": 2.6298342541436464, + "grad_norm": 1.944106101989746, + "learning_rate": 1e-06, + "loss": 0.597, + "mean_token_accuracy": 0.7990437150001526, + "num_tokens": 848676096.0, + "step": 31892 + }, + { + "epoch": 2.629916714768698, + "grad_norm": 2.028195381164551, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7778872847557068, + "num_tokens": 848708328.0, + "step": 31893 + }, + { + "epoch": 2.6299991753937495, + "grad_norm": 2.1297781467437744, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7776652574539185, + "num_tokens": 848737992.0, + "step": 31894 + }, + { + "epoch": 2.630081636018801, + "grad_norm": 2.297595977783203, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7424096465110779, + "num_tokens": 848763711.0, + "step": 31895 + }, + { + "epoch": 2.6301640966438526, + "grad_norm": 2.0878331661224365, + "learning_rate": 1e-06, + "loss": 0.7391, + "mean_token_accuracy": 0.7552803158760071, + "num_tokens": 848796247.0, + "step": 31896 + }, + { + "epoch": 2.630246557268904, + "grad_norm": 1.957594633102417, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7781275510787964, + "num_tokens": 848827314.0, + "step": 31897 + }, + { + "epoch": 2.630329017893956, + "grad_norm": 1.987761378288269, + "learning_rate": 1e-06, + "loss": 0.7147, + "mean_token_accuracy": 0.7622047662734985, + "num_tokens": 848859712.0, + "step": 31898 + }, + { + "epoch": 2.630411478519007, + "grad_norm": 2.005563259124756, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7535027861595154, + "num_tokens": 848894540.0, + "step": 31899 + }, + { + "epoch": 2.630493939144059, + "grad_norm": 2.1330630779266357, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7081656455993652, + "num_tokens": 848925544.0, + "step": 31900 + }, + { + "epoch": 2.6305763997691103, + "grad_norm": 2.120285749435425, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.7930155992507935, + "num_tokens": 848954305.0, + "step": 31901 + }, + { + "epoch": 2.6306588603941616, + "grad_norm": 1.937127709388733, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7764807939529419, + "num_tokens": 848988755.0, + "step": 31902 + }, + { + "epoch": 2.6307413210192134, + "grad_norm": 2.0751469135284424, + "learning_rate": 1e-06, + "loss": 0.7625, + "mean_token_accuracy": 0.7591768503189087, + "num_tokens": 849017203.0, + "step": 31903 + }, + { + "epoch": 2.630823781644265, + "grad_norm": 1.9060401916503906, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7855862975120544, + "num_tokens": 849051050.0, + "step": 31904 + }, + { + "epoch": 2.630906242269316, + "grad_norm": 2.208941698074341, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7209738492965698, + "num_tokens": 849080120.0, + "step": 31905 + }, + { + "epoch": 2.630988702894368, + "grad_norm": 2.018980026245117, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7366865873336792, + "num_tokens": 849111670.0, + "step": 31906 + }, + { + "epoch": 2.6310711635194197, + "grad_norm": 2.4441351890563965, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.7763287425041199, + "num_tokens": 849135448.0, + "step": 31907 + }, + { + "epoch": 2.631153624144471, + "grad_norm": 2.3906214237213135, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7515987157821655, + "num_tokens": 849159086.0, + "step": 31908 + }, + { + "epoch": 2.6312360847695224, + "grad_norm": 2.053920030593872, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7347935438156128, + "num_tokens": 849192833.0, + "step": 31909 + }, + { + "epoch": 2.631318545394574, + "grad_norm": 2.3149428367614746, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7683584690093994, + "num_tokens": 849217072.0, + "step": 31910 + }, + { + "epoch": 2.6314010060196256, + "grad_norm": 2.2224559783935547, + "learning_rate": 1e-06, + "loss": 0.5684, + "mean_token_accuracy": 0.7977774739265442, + "num_tokens": 849243005.0, + "step": 31911 + }, + { + "epoch": 2.631483466644677, + "grad_norm": 2.083728551864624, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.793307363986969, + "num_tokens": 849271955.0, + "step": 31912 + }, + { + "epoch": 2.6315659272697287, + "grad_norm": 1.934423804283142, + "learning_rate": 1e-06, + "loss": 0.5956, + "mean_token_accuracy": 0.8041828870773315, + "num_tokens": 849302443.0, + "step": 31913 + }, + { + "epoch": 2.6316483878947805, + "grad_norm": 2.2406845092773438, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.7801433801651001, + "num_tokens": 849329715.0, + "step": 31914 + }, + { + "epoch": 2.631730848519832, + "grad_norm": 1.9379435777664185, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7747946977615356, + "num_tokens": 849364780.0, + "step": 31915 + }, + { + "epoch": 2.631813309144883, + "grad_norm": 2.0021753311157227, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7194010019302368, + "num_tokens": 849395142.0, + "step": 31916 + }, + { + "epoch": 2.631895769769935, + "grad_norm": 2.080872058868408, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.7806627750396729, + "num_tokens": 849425709.0, + "step": 31917 + }, + { + "epoch": 2.6319782303949864, + "grad_norm": 2.236344337463379, + "learning_rate": 1e-06, + "loss": 0.5354, + "mean_token_accuracy": 0.8177711963653564, + "num_tokens": 849454129.0, + "step": 31918 + }, + { + "epoch": 2.6320606910200377, + "grad_norm": 2.3277065753936768, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7772349119186401, + "num_tokens": 849482527.0, + "step": 31919 + }, + { + "epoch": 2.6321431516450895, + "grad_norm": 2.12725567817688, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7411506175994873, + "num_tokens": 849514148.0, + "step": 31920 + }, + { + "epoch": 2.632225612270141, + "grad_norm": 2.0992071628570557, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.7795649170875549, + "num_tokens": 849542929.0, + "step": 31921 + }, + { + "epoch": 2.6323080728951926, + "grad_norm": 1.951188325881958, + "learning_rate": 1e-06, + "loss": 0.6333, + "mean_token_accuracy": 0.7879318594932556, + "num_tokens": 849574767.0, + "step": 31922 + }, + { + "epoch": 2.632390533520244, + "grad_norm": 1.9661171436309814, + "learning_rate": 1e-06, + "loss": 0.6905, + "mean_token_accuracy": 0.77345871925354, + "num_tokens": 849610140.0, + "step": 31923 + }, + { + "epoch": 2.632472994145296, + "grad_norm": 2.0979158878326416, + "learning_rate": 1e-06, + "loss": 0.765, + "mean_token_accuracy": 0.7593882083892822, + "num_tokens": 849640945.0, + "step": 31924 + }, + { + "epoch": 2.632555454770347, + "grad_norm": 1.937992811203003, + "learning_rate": 1e-06, + "loss": 0.703, + "mean_token_accuracy": 0.7680985927581787, + "num_tokens": 849679442.0, + "step": 31925 + }, + { + "epoch": 2.6326379153953985, + "grad_norm": 2.2064402103424072, + "learning_rate": 1e-06, + "loss": 0.6496, + "mean_token_accuracy": 0.7823023200035095, + "num_tokens": 849706416.0, + "step": 31926 + }, + { + "epoch": 2.6327203760204503, + "grad_norm": 1.8501207828521729, + "learning_rate": 1e-06, + "loss": 0.6717, + "mean_token_accuracy": 0.7741100788116455, + "num_tokens": 849740477.0, + "step": 31927 + }, + { + "epoch": 2.6328028366455016, + "grad_norm": 2.347302198410034, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7658953070640564, + "num_tokens": 849764288.0, + "step": 31928 + }, + { + "epoch": 2.6328852972705534, + "grad_norm": 2.150893211364746, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.7831774353981018, + "num_tokens": 849793811.0, + "step": 31929 + }, + { + "epoch": 2.6329677578956048, + "grad_norm": 1.9982166290283203, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.7786605358123779, + "num_tokens": 849825438.0, + "step": 31930 + }, + { + "epoch": 2.6330502185206566, + "grad_norm": 1.8891111612319946, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7509231567382812, + "num_tokens": 849859828.0, + "step": 31931 + }, + { + "epoch": 2.633132679145708, + "grad_norm": 2.084242582321167, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7508980631828308, + "num_tokens": 849891174.0, + "step": 31932 + }, + { + "epoch": 2.6332151397707593, + "grad_norm": 2.1496520042419434, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7694194316864014, + "num_tokens": 849921219.0, + "step": 31933 + }, + { + "epoch": 2.633297600395811, + "grad_norm": 2.0606884956359863, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7444008588790894, + "num_tokens": 849952037.0, + "step": 31934 + }, + { + "epoch": 2.6333800610208624, + "grad_norm": 2.6550936698913574, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.7906707525253296, + "num_tokens": 849974233.0, + "step": 31935 + }, + { + "epoch": 2.633462521645914, + "grad_norm": 2.0879340171813965, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7924330234527588, + "num_tokens": 850005518.0, + "step": 31936 + }, + { + "epoch": 2.6335449822709656, + "grad_norm": 1.9434372186660767, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7553186416625977, + "num_tokens": 850041039.0, + "step": 31937 + }, + { + "epoch": 2.6336274428960174, + "grad_norm": 2.4628210067749023, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7429219484329224, + "num_tokens": 850067849.0, + "step": 31938 + }, + { + "epoch": 2.6337099035210687, + "grad_norm": 1.9881043434143066, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.8035085201263428, + "num_tokens": 850101359.0, + "step": 31939 + }, + { + "epoch": 2.63379236414612, + "grad_norm": 2.4308295249938965, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7747479677200317, + "num_tokens": 850124281.0, + "step": 31940 + }, + { + "epoch": 2.633874824771172, + "grad_norm": 2.154313564300537, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7700205445289612, + "num_tokens": 850154328.0, + "step": 31941 + }, + { + "epoch": 2.633957285396223, + "grad_norm": 2.1216683387756348, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7419074773788452, + "num_tokens": 850184635.0, + "step": 31942 + }, + { + "epoch": 2.634039746021275, + "grad_norm": 2.0670735836029053, + "learning_rate": 1e-06, + "loss": 0.6274, + "mean_token_accuracy": 0.7827085256576538, + "num_tokens": 850213655.0, + "step": 31943 + }, + { + "epoch": 2.6341222066463263, + "grad_norm": 2.118652582168579, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.7926251292228699, + "num_tokens": 850242839.0, + "step": 31944 + }, + { + "epoch": 2.634204667271378, + "grad_norm": 2.0280871391296387, + "learning_rate": 1e-06, + "loss": 0.6643, + "mean_token_accuracy": 0.7841486930847168, + "num_tokens": 850273118.0, + "step": 31945 + }, + { + "epoch": 2.6342871278964295, + "grad_norm": 2.1537365913391113, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7439360618591309, + "num_tokens": 850304019.0, + "step": 31946 + }, + { + "epoch": 2.634369588521481, + "grad_norm": 2.00734543800354, + "learning_rate": 1e-06, + "loss": 0.5565, + "mean_token_accuracy": 0.8072771430015564, + "num_tokens": 850338010.0, + "step": 31947 + }, + { + "epoch": 2.6344520491465326, + "grad_norm": 1.9032186269760132, + "learning_rate": 1e-06, + "loss": 0.5848, + "mean_token_accuracy": 0.8114043474197388, + "num_tokens": 850371867.0, + "step": 31948 + }, + { + "epoch": 2.634534509771584, + "grad_norm": 2.159886121749878, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7760133147239685, + "num_tokens": 850400024.0, + "step": 31949 + }, + { + "epoch": 2.6346169703966353, + "grad_norm": 2.0557093620300293, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7715017199516296, + "num_tokens": 850432963.0, + "step": 31950 + }, + { + "epoch": 2.634699431021687, + "grad_norm": 2.0401229858398438, + "learning_rate": 1e-06, + "loss": 0.5502, + "mean_token_accuracy": 0.8031454682350159, + "num_tokens": 850463182.0, + "step": 31951 + }, + { + "epoch": 2.634781891646739, + "grad_norm": 2.1244301795959473, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7395114898681641, + "num_tokens": 850494031.0, + "step": 31952 + }, + { + "epoch": 2.6348643522717903, + "grad_norm": 2.0685417652130127, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7623063325881958, + "num_tokens": 850524521.0, + "step": 31953 + }, + { + "epoch": 2.6349468128968416, + "grad_norm": 2.2130277156829834, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7387590408325195, + "num_tokens": 850551383.0, + "step": 31954 + }, + { + "epoch": 2.6350292735218934, + "grad_norm": 2.254148244857788, + "learning_rate": 1e-06, + "loss": 0.683, + "mean_token_accuracy": 0.7649307250976562, + "num_tokens": 850575910.0, + "step": 31955 + }, + { + "epoch": 2.6351117341469448, + "grad_norm": 1.9764779806137085, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7709304094314575, + "num_tokens": 850608239.0, + "step": 31956 + }, + { + "epoch": 2.635194194771996, + "grad_norm": 2.1679680347442627, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.786712646484375, + "num_tokens": 850636179.0, + "step": 31957 + }, + { + "epoch": 2.635276655397048, + "grad_norm": 2.0705530643463135, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7717472314834595, + "num_tokens": 850668776.0, + "step": 31958 + }, + { + "epoch": 2.6353591160220997, + "grad_norm": 2.211125612258911, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7478057146072388, + "num_tokens": 850700086.0, + "step": 31959 + }, + { + "epoch": 2.635441576647151, + "grad_norm": 2.244279146194458, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7722236514091492, + "num_tokens": 850731209.0, + "step": 31960 + }, + { + "epoch": 2.6355240372722024, + "grad_norm": 1.956653118133545, + "learning_rate": 1e-06, + "loss": 0.6483, + "mean_token_accuracy": 0.783424973487854, + "num_tokens": 850761873.0, + "step": 31961 + }, + { + "epoch": 2.635606497897254, + "grad_norm": 2.3217978477478027, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7438180446624756, + "num_tokens": 850788942.0, + "step": 31962 + }, + { + "epoch": 2.6356889585223056, + "grad_norm": 2.200296401977539, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7708094716072083, + "num_tokens": 850813898.0, + "step": 31963 + }, + { + "epoch": 2.635771419147357, + "grad_norm": 1.9866580963134766, + "learning_rate": 1e-06, + "loss": 0.6499, + "mean_token_accuracy": 0.7834103107452393, + "num_tokens": 850847556.0, + "step": 31964 + }, + { + "epoch": 2.6358538797724087, + "grad_norm": 1.8373931646347046, + "learning_rate": 1e-06, + "loss": 0.6797, + "mean_token_accuracy": 0.7830731868743896, + "num_tokens": 850884304.0, + "step": 31965 + }, + { + "epoch": 2.63593634039746, + "grad_norm": 2.1847856044769287, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.7669090032577515, + "num_tokens": 850910742.0, + "step": 31966 + }, + { + "epoch": 2.636018801022512, + "grad_norm": 2.178143262863159, + "learning_rate": 1e-06, + "loss": 0.7885, + "mean_token_accuracy": 0.7549242973327637, + "num_tokens": 850939461.0, + "step": 31967 + }, + { + "epoch": 2.636101261647563, + "grad_norm": 2.0506248474121094, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.7773922681808472, + "num_tokens": 850973249.0, + "step": 31968 + }, + { + "epoch": 2.636183722272615, + "grad_norm": 2.0031535625457764, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7667399048805237, + "num_tokens": 851005462.0, + "step": 31969 + }, + { + "epoch": 2.6362661828976663, + "grad_norm": 2.2785754203796387, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7429004907608032, + "num_tokens": 851033972.0, + "step": 31970 + }, + { + "epoch": 2.6363486435227177, + "grad_norm": 2.0510685443878174, + "learning_rate": 1e-06, + "loss": 0.6254, + "mean_token_accuracy": 0.7980190515518188, + "num_tokens": 851062341.0, + "step": 31971 + }, + { + "epoch": 2.6364311041477695, + "grad_norm": 2.0589582920074463, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7486650347709656, + "num_tokens": 851098036.0, + "step": 31972 + }, + { + "epoch": 2.636513564772821, + "grad_norm": 2.3492989540100098, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7762833833694458, + "num_tokens": 851122975.0, + "step": 31973 + }, + { + "epoch": 2.6365960253978726, + "grad_norm": 2.076112747192383, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7808336615562439, + "num_tokens": 851153435.0, + "step": 31974 + }, + { + "epoch": 2.636678486022924, + "grad_norm": 2.0723319053649902, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.803977370262146, + "num_tokens": 851182655.0, + "step": 31975 + }, + { + "epoch": 2.636760946647976, + "grad_norm": 1.9641149044036865, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.7949663400650024, + "num_tokens": 851213711.0, + "step": 31976 + }, + { + "epoch": 2.636843407273027, + "grad_norm": 2.1057653427124023, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.7624592781066895, + "num_tokens": 851247217.0, + "step": 31977 + }, + { + "epoch": 2.6369258678980785, + "grad_norm": 2.3639609813690186, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.7909948825836182, + "num_tokens": 851270048.0, + "step": 31978 + }, + { + "epoch": 2.6370083285231303, + "grad_norm": 2.1276750564575195, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.736038088798523, + "num_tokens": 851299265.0, + "step": 31979 + }, + { + "epoch": 2.6370907891481816, + "grad_norm": 2.2739391326904297, + "learning_rate": 1e-06, + "loss": 0.668, + "mean_token_accuracy": 0.7873177528381348, + "num_tokens": 851328032.0, + "step": 31980 + }, + { + "epoch": 2.6371732497732334, + "grad_norm": 1.9755775928497314, + "learning_rate": 1e-06, + "loss": 0.5713, + "mean_token_accuracy": 0.8003380298614502, + "num_tokens": 851360764.0, + "step": 31981 + }, + { + "epoch": 2.6372557103982848, + "grad_norm": 1.9917452335357666, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.783715009689331, + "num_tokens": 851393755.0, + "step": 31982 + }, + { + "epoch": 2.6373381710233366, + "grad_norm": 2.1859042644500732, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7529683113098145, + "num_tokens": 851423587.0, + "step": 31983 + }, + { + "epoch": 2.637420631648388, + "grad_norm": 2.089259147644043, + "learning_rate": 1e-06, + "loss": 0.5826, + "mean_token_accuracy": 0.8002191185951233, + "num_tokens": 851452457.0, + "step": 31984 + }, + { + "epoch": 2.6375030922734393, + "grad_norm": 2.2316625118255615, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7471436262130737, + "num_tokens": 851480186.0, + "step": 31985 + }, + { + "epoch": 2.637585552898491, + "grad_norm": 2.09120774269104, + "learning_rate": 1e-06, + "loss": 0.6286, + "mean_token_accuracy": 0.7902075052261353, + "num_tokens": 851513594.0, + "step": 31986 + }, + { + "epoch": 2.6376680135235424, + "grad_norm": 2.2826361656188965, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.7908231019973755, + "num_tokens": 851542110.0, + "step": 31987 + }, + { + "epoch": 2.637750474148594, + "grad_norm": 2.3225932121276855, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.7854340672492981, + "num_tokens": 851568904.0, + "step": 31988 + }, + { + "epoch": 2.6378329347736456, + "grad_norm": 2.1097216606140137, + "learning_rate": 1e-06, + "loss": 0.6796, + "mean_token_accuracy": 0.7802634835243225, + "num_tokens": 851599239.0, + "step": 31989 + }, + { + "epoch": 2.6379153953986973, + "grad_norm": 2.2973055839538574, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7507439851760864, + "num_tokens": 851624371.0, + "step": 31990 + }, + { + "epoch": 2.6379978560237487, + "grad_norm": 2.040292978286743, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.728209376335144, + "num_tokens": 851659583.0, + "step": 31991 + }, + { + "epoch": 2.6380803166488, + "grad_norm": 2.3982503414154053, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7379071712493896, + "num_tokens": 851686788.0, + "step": 31992 + }, + { + "epoch": 2.638162777273852, + "grad_norm": 2.3444137573242188, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7626543641090393, + "num_tokens": 851712194.0, + "step": 31993 + }, + { + "epoch": 2.638245237898903, + "grad_norm": 2.383702278137207, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.776679515838623, + "num_tokens": 851736226.0, + "step": 31994 + }, + { + "epoch": 2.6383276985239545, + "grad_norm": 1.8460414409637451, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7591345310211182, + "num_tokens": 851776707.0, + "step": 31995 + }, + { + "epoch": 2.6384101591490063, + "grad_norm": 2.2018392086029053, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.759106457233429, + "num_tokens": 851802552.0, + "step": 31996 + }, + { + "epoch": 2.638492619774058, + "grad_norm": 2.3146111965179443, + "learning_rate": 1e-06, + "loss": 0.5738, + "mean_token_accuracy": 0.8104839324951172, + "num_tokens": 851827301.0, + "step": 31997 + }, + { + "epoch": 2.6385750803991095, + "grad_norm": 2.0581560134887695, + "learning_rate": 1e-06, + "loss": 0.6205, + "mean_token_accuracy": 0.7847467660903931, + "num_tokens": 851859381.0, + "step": 31998 + }, + { + "epoch": 2.638657541024161, + "grad_norm": 1.8080121278762817, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.7781641483306885, + "num_tokens": 851900288.0, + "step": 31999 + }, + { + "epoch": 2.6387400016492126, + "grad_norm": 2.0719077587127686, + "learning_rate": 1e-06, + "loss": 0.5558, + "mean_token_accuracy": 0.8118699789047241, + "num_tokens": 851929217.0, + "step": 32000 + }, + { + "epoch": 2.638822462274264, + "grad_norm": 2.2588539123535156, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.7879025936126709, + "num_tokens": 851955536.0, + "step": 32001 + }, + { + "epoch": 2.6389049228993153, + "grad_norm": 2.3560867309570312, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.763690173625946, + "num_tokens": 851979975.0, + "step": 32002 + }, + { + "epoch": 2.638987383524367, + "grad_norm": 2.086028575897217, + "learning_rate": 1e-06, + "loss": 0.6348, + "mean_token_accuracy": 0.7879533171653748, + "num_tokens": 852008469.0, + "step": 32003 + }, + { + "epoch": 2.639069844149419, + "grad_norm": 2.083308458328247, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.771865725517273, + "num_tokens": 852040680.0, + "step": 32004 + }, + { + "epoch": 2.6391523047744703, + "grad_norm": 1.8970510959625244, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.8114238381385803, + "num_tokens": 852074962.0, + "step": 32005 + }, + { + "epoch": 2.6392347653995216, + "grad_norm": 2.1589486598968506, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7650386095046997, + "num_tokens": 852104973.0, + "step": 32006 + }, + { + "epoch": 2.6393172260245734, + "grad_norm": 2.0470824241638184, + "learning_rate": 1e-06, + "loss": 0.5264, + "mean_token_accuracy": 0.8209288120269775, + "num_tokens": 852133272.0, + "step": 32007 + }, + { + "epoch": 2.6393996866496248, + "grad_norm": 2.178276300430298, + "learning_rate": 1e-06, + "loss": 0.5598, + "mean_token_accuracy": 0.8124078512191772, + "num_tokens": 852161715.0, + "step": 32008 + }, + { + "epoch": 2.639482147274676, + "grad_norm": 2.011914014816284, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7718982696533203, + "num_tokens": 852194384.0, + "step": 32009 + }, + { + "epoch": 2.639564607899728, + "grad_norm": 1.9720158576965332, + "learning_rate": 1e-06, + "loss": 0.4547, + "mean_token_accuracy": 0.8429296016693115, + "num_tokens": 852222297.0, + "step": 32010 + }, + { + "epoch": 2.6396470685247793, + "grad_norm": 1.9964557886123657, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.774520754814148, + "num_tokens": 852254160.0, + "step": 32011 + }, + { + "epoch": 2.639729529149831, + "grad_norm": 2.058319568634033, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.7904685139656067, + "num_tokens": 852286017.0, + "step": 32012 + }, + { + "epoch": 2.6398119897748824, + "grad_norm": 2.2255184650421143, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7624239921569824, + "num_tokens": 852314761.0, + "step": 32013 + }, + { + "epoch": 2.639894450399934, + "grad_norm": 2.226642608642578, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7614186406135559, + "num_tokens": 852347287.0, + "step": 32014 + }, + { + "epoch": 2.6399769110249856, + "grad_norm": 2.6226611137390137, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7803165316581726, + "num_tokens": 852381596.0, + "step": 32015 + }, + { + "epoch": 2.640059371650037, + "grad_norm": 2.304854393005371, + "learning_rate": 1e-06, + "loss": 0.6956, + "mean_token_accuracy": 0.7723186612129211, + "num_tokens": 852405854.0, + "step": 32016 + }, + { + "epoch": 2.6401418322750887, + "grad_norm": 1.9631354808807373, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7248098850250244, + "num_tokens": 852439147.0, + "step": 32017 + }, + { + "epoch": 2.64022429290014, + "grad_norm": 2.1297409534454346, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.7749032974243164, + "num_tokens": 852469867.0, + "step": 32018 + }, + { + "epoch": 2.640306753525192, + "grad_norm": 2.489471435546875, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7301969528198242, + "num_tokens": 852500518.0, + "step": 32019 + }, + { + "epoch": 2.640389214150243, + "grad_norm": 1.865921974182129, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7553091049194336, + "num_tokens": 852537882.0, + "step": 32020 + }, + { + "epoch": 2.640471674775295, + "grad_norm": 2.099785327911377, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.7919791340827942, + "num_tokens": 852572927.0, + "step": 32021 + }, + { + "epoch": 2.6405541354003463, + "grad_norm": 2.093318462371826, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7738491296768188, + "num_tokens": 852604911.0, + "step": 32022 + }, + { + "epoch": 2.6406365960253977, + "grad_norm": 2.2278318405151367, + "learning_rate": 1e-06, + "loss": 0.7771, + "mean_token_accuracy": 0.7465893626213074, + "num_tokens": 852633568.0, + "step": 32023 + }, + { + "epoch": 2.6407190566504495, + "grad_norm": 2.1139955520629883, + "learning_rate": 1e-06, + "loss": 0.7768, + "mean_token_accuracy": 0.7430152297019958, + "num_tokens": 852663931.0, + "step": 32024 + }, + { + "epoch": 2.640801517275501, + "grad_norm": 2.2478091716766357, + "learning_rate": 1e-06, + "loss": 0.7637, + "mean_token_accuracy": 0.7560945153236389, + "num_tokens": 852689943.0, + "step": 32025 + }, + { + "epoch": 2.6408839779005526, + "grad_norm": 2.290369987487793, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7797883749008179, + "num_tokens": 852717658.0, + "step": 32026 + }, + { + "epoch": 2.640966438525604, + "grad_norm": 2.1441595554351807, + "learning_rate": 1e-06, + "loss": 0.5305, + "mean_token_accuracy": 0.8153302669525146, + "num_tokens": 852748882.0, + "step": 32027 + }, + { + "epoch": 2.6410488991506558, + "grad_norm": 2.121587038040161, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7673516869544983, + "num_tokens": 852776569.0, + "step": 32028 + }, + { + "epoch": 2.641131359775707, + "grad_norm": 2.3074827194213867, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7319010496139526, + "num_tokens": 852806045.0, + "step": 32029 + }, + { + "epoch": 2.6412138204007585, + "grad_norm": 2.0229146480560303, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7548487186431885, + "num_tokens": 852839362.0, + "step": 32030 + }, + { + "epoch": 2.6412962810258103, + "grad_norm": 2.2984416484832764, + "learning_rate": 1e-06, + "loss": 0.6386, + "mean_token_accuracy": 0.7922053933143616, + "num_tokens": 852870533.0, + "step": 32031 + }, + { + "epoch": 2.6413787416508616, + "grad_norm": 2.2766177654266357, + "learning_rate": 1e-06, + "loss": 0.7049, + "mean_token_accuracy": 0.7623918056488037, + "num_tokens": 852897826.0, + "step": 32032 + }, + { + "epoch": 2.6414612022759134, + "grad_norm": 2.2198569774627686, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.75469970703125, + "num_tokens": 852925741.0, + "step": 32033 + }, + { + "epoch": 2.6415436629009648, + "grad_norm": 1.9445308446884155, + "learning_rate": 1e-06, + "loss": 0.5764, + "mean_token_accuracy": 0.8000239133834839, + "num_tokens": 852957780.0, + "step": 32034 + }, + { + "epoch": 2.6416261235260166, + "grad_norm": 2.1056792736053467, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7605566382408142, + "num_tokens": 852988304.0, + "step": 32035 + }, + { + "epoch": 2.641708584151068, + "grad_norm": 1.8785866498947144, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.772969126701355, + "num_tokens": 853023705.0, + "step": 32036 + }, + { + "epoch": 2.6417910447761193, + "grad_norm": 2.0695457458496094, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.7914794683456421, + "num_tokens": 853056138.0, + "step": 32037 + }, + { + "epoch": 2.641873505401171, + "grad_norm": 1.792055606842041, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7752025127410889, + "num_tokens": 853096096.0, + "step": 32038 + }, + { + "epoch": 2.6419559660262224, + "grad_norm": 2.046539545059204, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.7935547828674316, + "num_tokens": 853125431.0, + "step": 32039 + }, + { + "epoch": 2.6420384266512738, + "grad_norm": 2.164247989654541, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7479181289672852, + "num_tokens": 853154403.0, + "step": 32040 + }, + { + "epoch": 2.6421208872763255, + "grad_norm": 2.4544148445129395, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7493948340415955, + "num_tokens": 853179135.0, + "step": 32041 + }, + { + "epoch": 2.6422033479013773, + "grad_norm": 2.3116164207458496, + "learning_rate": 1e-06, + "loss": 0.6098, + "mean_token_accuracy": 0.8081956505775452, + "num_tokens": 853205928.0, + "step": 32042 + }, + { + "epoch": 2.6422858085264287, + "grad_norm": 2.0466201305389404, + "learning_rate": 1e-06, + "loss": 0.6066, + "mean_token_accuracy": 0.8007972240447998, + "num_tokens": 853235989.0, + "step": 32043 + }, + { + "epoch": 2.64236826915148, + "grad_norm": 2.367072820663452, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7727271914482117, + "num_tokens": 853259374.0, + "step": 32044 + }, + { + "epoch": 2.642450729776532, + "grad_norm": 1.9101040363311768, + "learning_rate": 1e-06, + "loss": 0.7087, + "mean_token_accuracy": 0.7763015031814575, + "num_tokens": 853296258.0, + "step": 32045 + }, + { + "epoch": 2.642533190401583, + "grad_norm": 1.889359951019287, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7497097253799438, + "num_tokens": 853330776.0, + "step": 32046 + }, + { + "epoch": 2.6426156510266345, + "grad_norm": 1.9206873178482056, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7912927865982056, + "num_tokens": 853363806.0, + "step": 32047 + }, + { + "epoch": 2.6426981116516863, + "grad_norm": 2.1074559688568115, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.766197681427002, + "num_tokens": 853394807.0, + "step": 32048 + }, + { + "epoch": 2.642780572276738, + "grad_norm": 2.035637378692627, + "learning_rate": 1e-06, + "loss": 0.6877, + "mean_token_accuracy": 0.7716352343559265, + "num_tokens": 853423830.0, + "step": 32049 + }, + { + "epoch": 2.6428630329017895, + "grad_norm": 2.216090202331543, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.7863233089447021, + "num_tokens": 853449438.0, + "step": 32050 + }, + { + "epoch": 2.642945493526841, + "grad_norm": 2.0654237270355225, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7670482397079468, + "num_tokens": 853479599.0, + "step": 32051 + }, + { + "epoch": 2.6430279541518926, + "grad_norm": 2.156186103820801, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7658512592315674, + "num_tokens": 853510024.0, + "step": 32052 + }, + { + "epoch": 2.643110414776944, + "grad_norm": 2.24088454246521, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7782666683197021, + "num_tokens": 853537041.0, + "step": 32053 + }, + { + "epoch": 2.6431928754019953, + "grad_norm": 1.8991457223892212, + "learning_rate": 1e-06, + "loss": 0.6051, + "mean_token_accuracy": 0.8012997508049011, + "num_tokens": 853570272.0, + "step": 32054 + }, + { + "epoch": 2.643275336027047, + "grad_norm": 2.091797113418579, + "learning_rate": 1e-06, + "loss": 0.5653, + "mean_token_accuracy": 0.8115659952163696, + "num_tokens": 853596214.0, + "step": 32055 + }, + { + "epoch": 2.6433577966520985, + "grad_norm": 1.828860878944397, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7728513479232788, + "num_tokens": 853628593.0, + "step": 32056 + }, + { + "epoch": 2.6434402572771503, + "grad_norm": 1.8937444686889648, + "learning_rate": 1e-06, + "loss": 0.5687, + "mean_token_accuracy": 0.8072932958602905, + "num_tokens": 853664626.0, + "step": 32057 + }, + { + "epoch": 2.6435227179022016, + "grad_norm": 2.1976099014282227, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7488347291946411, + "num_tokens": 853694333.0, + "step": 32058 + }, + { + "epoch": 2.6436051785272534, + "grad_norm": 2.4817655086517334, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7438080310821533, + "num_tokens": 853716693.0, + "step": 32059 + }, + { + "epoch": 2.6436876391523048, + "grad_norm": 1.928731083869934, + "learning_rate": 1e-06, + "loss": 0.5487, + "mean_token_accuracy": 0.8189629316329956, + "num_tokens": 853748432.0, + "step": 32060 + }, + { + "epoch": 2.643770099777356, + "grad_norm": 2.0244393348693848, + "learning_rate": 1e-06, + "loss": 0.5839, + "mean_token_accuracy": 0.8005335330963135, + "num_tokens": 853778008.0, + "step": 32061 + }, + { + "epoch": 2.643852560402408, + "grad_norm": 2.2182676792144775, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7323071956634521, + "num_tokens": 853806278.0, + "step": 32062 + }, + { + "epoch": 2.6439350210274593, + "grad_norm": 2.0971710681915283, + "learning_rate": 1e-06, + "loss": 0.5618, + "mean_token_accuracy": 0.8142963647842407, + "num_tokens": 853832957.0, + "step": 32063 + }, + { + "epoch": 2.644017481652511, + "grad_norm": 1.9053126573562622, + "learning_rate": 1e-06, + "loss": 0.5774, + "mean_token_accuracy": 0.8169999122619629, + "num_tokens": 853864418.0, + "step": 32064 + }, + { + "epoch": 2.6440999422775624, + "grad_norm": 2.2209436893463135, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7479254007339478, + "num_tokens": 853893608.0, + "step": 32065 + }, + { + "epoch": 2.644182402902614, + "grad_norm": 2.3393993377685547, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7759310007095337, + "num_tokens": 853919380.0, + "step": 32066 + }, + { + "epoch": 2.6442648635276655, + "grad_norm": 2.048837900161743, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.7907772660255432, + "num_tokens": 853952263.0, + "step": 32067 + }, + { + "epoch": 2.644347324152717, + "grad_norm": 2.2310006618499756, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7717044949531555, + "num_tokens": 853979555.0, + "step": 32068 + }, + { + "epoch": 2.6444297847777687, + "grad_norm": 2.162155866622925, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.76265549659729, + "num_tokens": 854011251.0, + "step": 32069 + }, + { + "epoch": 2.64451224540282, + "grad_norm": 1.939686894416809, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.78110271692276, + "num_tokens": 854043958.0, + "step": 32070 + }, + { + "epoch": 2.644594706027872, + "grad_norm": 2.3418023586273193, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7456353902816772, + "num_tokens": 854072526.0, + "step": 32071 + }, + { + "epoch": 2.644677166652923, + "grad_norm": 1.9069679975509644, + "learning_rate": 1e-06, + "loss": 0.6126, + "mean_token_accuracy": 0.7856395244598389, + "num_tokens": 854105095.0, + "step": 32072 + }, + { + "epoch": 2.644759627277975, + "grad_norm": 2.0003862380981445, + "learning_rate": 1e-06, + "loss": 0.6514, + "mean_token_accuracy": 0.7889360785484314, + "num_tokens": 854138515.0, + "step": 32073 + }, + { + "epoch": 2.6448420879030263, + "grad_norm": 2.0067784786224365, + "learning_rate": 1e-06, + "loss": 0.5699, + "mean_token_accuracy": 0.8043912053108215, + "num_tokens": 854170440.0, + "step": 32074 + }, + { + "epoch": 2.6449245485280777, + "grad_norm": 2.050302267074585, + "learning_rate": 1e-06, + "loss": 0.5093, + "mean_token_accuracy": 0.8262405395507812, + "num_tokens": 854200953.0, + "step": 32075 + }, + { + "epoch": 2.6450070091531295, + "grad_norm": 1.9983594417572021, + "learning_rate": 1e-06, + "loss": 0.7483, + "mean_token_accuracy": 0.7524815201759338, + "num_tokens": 854233146.0, + "step": 32076 + }, + { + "epoch": 2.645089469778181, + "grad_norm": 2.1806063652038574, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7869819402694702, + "num_tokens": 854261347.0, + "step": 32077 + }, + { + "epoch": 2.6451719304032326, + "grad_norm": 2.185980796813965, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7456958889961243, + "num_tokens": 854289366.0, + "step": 32078 + }, + { + "epoch": 2.645254391028284, + "grad_norm": 2.0499842166900635, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7741026878356934, + "num_tokens": 854321434.0, + "step": 32079 + }, + { + "epoch": 2.6453368516533358, + "grad_norm": 2.4775335788726807, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7627167701721191, + "num_tokens": 854343836.0, + "step": 32080 + }, + { + "epoch": 2.645419312278387, + "grad_norm": 1.9618817567825317, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7805739045143127, + "num_tokens": 854379092.0, + "step": 32081 + }, + { + "epoch": 2.6455017729034385, + "grad_norm": 1.9536970853805542, + "learning_rate": 1e-06, + "loss": 0.6501, + "mean_token_accuracy": 0.7831529378890991, + "num_tokens": 854414052.0, + "step": 32082 + }, + { + "epoch": 2.6455842335284903, + "grad_norm": 2.2845046520233154, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7707902193069458, + "num_tokens": 854440511.0, + "step": 32083 + }, + { + "epoch": 2.6456666941535416, + "grad_norm": 2.2320737838745117, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.7823622226715088, + "num_tokens": 854466124.0, + "step": 32084 + }, + { + "epoch": 2.645749154778593, + "grad_norm": 2.353895664215088, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7302824854850769, + "num_tokens": 854490344.0, + "step": 32085 + }, + { + "epoch": 2.6458316154036448, + "grad_norm": 2.1079444885253906, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7630804777145386, + "num_tokens": 854520736.0, + "step": 32086 + }, + { + "epoch": 2.6459140760286965, + "grad_norm": 2.423727512359619, + "learning_rate": 1e-06, + "loss": 0.553, + "mean_token_accuracy": 0.798775315284729, + "num_tokens": 854542883.0, + "step": 32087 + }, + { + "epoch": 2.645996536653748, + "grad_norm": 1.8648985624313354, + "learning_rate": 1e-06, + "loss": 0.4697, + "mean_token_accuracy": 0.8386008739471436, + "num_tokens": 854576588.0, + "step": 32088 + }, + { + "epoch": 2.6460789972787992, + "grad_norm": 2.255967378616333, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.7841171622276306, + "num_tokens": 854602425.0, + "step": 32089 + }, + { + "epoch": 2.646161457903851, + "grad_norm": 2.207789421081543, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7606865763664246, + "num_tokens": 854630398.0, + "step": 32090 + }, + { + "epoch": 2.6462439185289024, + "grad_norm": 2.0881574153900146, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.7475235462188721, + "num_tokens": 854662826.0, + "step": 32091 + }, + { + "epoch": 2.6463263791539537, + "grad_norm": 2.0302164554595947, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7424310445785522, + "num_tokens": 854695623.0, + "step": 32092 + }, + { + "epoch": 2.6464088397790055, + "grad_norm": 2.3721537590026855, + "learning_rate": 1e-06, + "loss": 0.5749, + "mean_token_accuracy": 0.7949388027191162, + "num_tokens": 854719245.0, + "step": 32093 + }, + { + "epoch": 2.6464913004040573, + "grad_norm": 2.064913749694824, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7596555948257446, + "num_tokens": 854750162.0, + "step": 32094 + }, + { + "epoch": 2.6465737610291087, + "grad_norm": 2.2019267082214355, + "learning_rate": 1e-06, + "loss": 0.584, + "mean_token_accuracy": 0.805467963218689, + "num_tokens": 854774814.0, + "step": 32095 + }, + { + "epoch": 2.64665622165416, + "grad_norm": 2.2832210063934326, + "learning_rate": 1e-06, + "loss": 0.5388, + "mean_token_accuracy": 0.8084072470664978, + "num_tokens": 854797342.0, + "step": 32096 + }, + { + "epoch": 2.646738682279212, + "grad_norm": 2.0199499130249023, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7438051700592041, + "num_tokens": 854833340.0, + "step": 32097 + }, + { + "epoch": 2.646821142904263, + "grad_norm": 2.2619683742523193, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7272359132766724, + "num_tokens": 854860386.0, + "step": 32098 + }, + { + "epoch": 2.6469036035293145, + "grad_norm": 2.2780470848083496, + "learning_rate": 1e-06, + "loss": 0.7831, + "mean_token_accuracy": 0.7413214445114136, + "num_tokens": 854889848.0, + "step": 32099 + }, + { + "epoch": 2.6469860641543663, + "grad_norm": 2.1779234409332275, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7739135026931763, + "num_tokens": 854920199.0, + "step": 32100 + }, + { + "epoch": 2.6470685247794177, + "grad_norm": 2.2496979236602783, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7783888578414917, + "num_tokens": 854948466.0, + "step": 32101 + }, + { + "epoch": 2.6471509854044695, + "grad_norm": 2.0702829360961914, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.7936543822288513, + "num_tokens": 854976440.0, + "step": 32102 + }, + { + "epoch": 2.647233446029521, + "grad_norm": 1.8354876041412354, + "learning_rate": 1e-06, + "loss": 0.5412, + "mean_token_accuracy": 0.8182076215744019, + "num_tokens": 855010156.0, + "step": 32103 + }, + { + "epoch": 2.6473159066545726, + "grad_norm": 2.1818618774414062, + "learning_rate": 1e-06, + "loss": 0.7267, + "mean_token_accuracy": 0.7650105953216553, + "num_tokens": 855037527.0, + "step": 32104 + }, + { + "epoch": 2.647398367279624, + "grad_norm": 2.3753716945648193, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.7916260957717896, + "num_tokens": 855060377.0, + "step": 32105 + }, + { + "epoch": 2.6474808279046753, + "grad_norm": 2.184224843978882, + "learning_rate": 1e-06, + "loss": 0.6122, + "mean_token_accuracy": 0.7967857718467712, + "num_tokens": 855086198.0, + "step": 32106 + }, + { + "epoch": 2.647563288529727, + "grad_norm": 2.2433266639709473, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7501599192619324, + "num_tokens": 855114050.0, + "step": 32107 + }, + { + "epoch": 2.6476457491547785, + "grad_norm": 2.1067705154418945, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.754450798034668, + "num_tokens": 855144433.0, + "step": 32108 + }, + { + "epoch": 2.6477282097798303, + "grad_norm": 2.170508861541748, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.7886784076690674, + "num_tokens": 855173892.0, + "step": 32109 + }, + { + "epoch": 2.6478106704048816, + "grad_norm": 2.038527488708496, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.7821323871612549, + "num_tokens": 855203688.0, + "step": 32110 + }, + { + "epoch": 2.6478931310299334, + "grad_norm": 1.958730936050415, + "learning_rate": 1e-06, + "loss": 0.575, + "mean_token_accuracy": 0.8098936080932617, + "num_tokens": 855234192.0, + "step": 32111 + }, + { + "epoch": 2.6479755916549848, + "grad_norm": 2.2820236682891846, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7247700691223145, + "num_tokens": 855261413.0, + "step": 32112 + }, + { + "epoch": 2.648058052280036, + "grad_norm": 1.7882519960403442, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.7971441745758057, + "num_tokens": 855298871.0, + "step": 32113 + }, + { + "epoch": 2.648140512905088, + "grad_norm": 1.9561090469360352, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7694998979568481, + "num_tokens": 855332395.0, + "step": 32114 + }, + { + "epoch": 2.6482229735301392, + "grad_norm": 2.181165933609009, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7652608156204224, + "num_tokens": 855361035.0, + "step": 32115 + }, + { + "epoch": 2.648305434155191, + "grad_norm": 2.402379274368286, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7633061408996582, + "num_tokens": 855385697.0, + "step": 32116 + }, + { + "epoch": 2.6483878947802424, + "grad_norm": 2.252124071121216, + "learning_rate": 1e-06, + "loss": 0.5083, + "mean_token_accuracy": 0.8207067847251892, + "num_tokens": 855413705.0, + "step": 32117 + }, + { + "epoch": 2.648470355405294, + "grad_norm": 2.5479118824005127, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7300877571105957, + "num_tokens": 855436633.0, + "step": 32118 + }, + { + "epoch": 2.6485528160303455, + "grad_norm": 2.213693141937256, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7654317021369934, + "num_tokens": 855466740.0, + "step": 32119 + }, + { + "epoch": 2.648635276655397, + "grad_norm": 2.1291513442993164, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7312113046646118, + "num_tokens": 855496793.0, + "step": 32120 + }, + { + "epoch": 2.6487177372804487, + "grad_norm": 2.133089065551758, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7299613356590271, + "num_tokens": 855525227.0, + "step": 32121 + }, + { + "epoch": 2.6488001979055, + "grad_norm": 2.1903722286224365, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7726718783378601, + "num_tokens": 855555871.0, + "step": 32122 + }, + { + "epoch": 2.648882658530552, + "grad_norm": 2.136836528778076, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7715672850608826, + "num_tokens": 855586217.0, + "step": 32123 + }, + { + "epoch": 2.648965119155603, + "grad_norm": 2.0350286960601807, + "learning_rate": 1e-06, + "loss": 0.5879, + "mean_token_accuracy": 0.7865866422653198, + "num_tokens": 855617055.0, + "step": 32124 + }, + { + "epoch": 2.649047579780655, + "grad_norm": 2.191493511199951, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7457071542739868, + "num_tokens": 855646565.0, + "step": 32125 + }, + { + "epoch": 2.6491300404057063, + "grad_norm": 2.08453106880188, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7658681869506836, + "num_tokens": 855676346.0, + "step": 32126 + }, + { + "epoch": 2.6492125010307577, + "grad_norm": 2.384432792663574, + "learning_rate": 1e-06, + "loss": 0.5731, + "mean_token_accuracy": 0.8112913370132446, + "num_tokens": 855699490.0, + "step": 32127 + }, + { + "epoch": 2.6492949616558095, + "grad_norm": 1.9534108638763428, + "learning_rate": 1e-06, + "loss": 0.5892, + "mean_token_accuracy": 0.7985175848007202, + "num_tokens": 855734050.0, + "step": 32128 + }, + { + "epoch": 2.649377422280861, + "grad_norm": 1.9017928838729858, + "learning_rate": 1e-06, + "loss": 0.5007, + "mean_token_accuracy": 0.8271390199661255, + "num_tokens": 855765786.0, + "step": 32129 + }, + { + "epoch": 2.649459882905912, + "grad_norm": 2.253183364868164, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7628283500671387, + "num_tokens": 855791555.0, + "step": 32130 + }, + { + "epoch": 2.649542343530964, + "grad_norm": 2.0544822216033936, + "learning_rate": 1e-06, + "loss": 0.7684, + "mean_token_accuracy": 0.7537465691566467, + "num_tokens": 855822082.0, + "step": 32131 + }, + { + "epoch": 2.6496248041560158, + "grad_norm": 1.9626271724700928, + "learning_rate": 1e-06, + "loss": 0.7583, + "mean_token_accuracy": 0.7592663168907166, + "num_tokens": 855860121.0, + "step": 32132 + }, + { + "epoch": 2.649707264781067, + "grad_norm": 1.8688000440597534, + "learning_rate": 1e-06, + "loss": 0.7698, + "mean_token_accuracy": 0.7553212642669678, + "num_tokens": 855897408.0, + "step": 32133 + }, + { + "epoch": 2.6497897254061185, + "grad_norm": 2.2022829055786133, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7731336355209351, + "num_tokens": 855924174.0, + "step": 32134 + }, + { + "epoch": 2.6498721860311703, + "grad_norm": 2.065114974975586, + "learning_rate": 1e-06, + "loss": 0.5712, + "mean_token_accuracy": 0.8030111789703369, + "num_tokens": 855949680.0, + "step": 32135 + }, + { + "epoch": 2.6499546466562216, + "grad_norm": 1.967119812965393, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7291673421859741, + "num_tokens": 855982079.0, + "step": 32136 + }, + { + "epoch": 2.650037107281273, + "grad_norm": 2.253751039505005, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7288612127304077, + "num_tokens": 856010530.0, + "step": 32137 + }, + { + "epoch": 2.6501195679063247, + "grad_norm": 2.043358325958252, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7444671988487244, + "num_tokens": 856042891.0, + "step": 32138 + }, + { + "epoch": 2.6502020285313765, + "grad_norm": 2.1389944553375244, + "learning_rate": 1e-06, + "loss": 0.6036, + "mean_token_accuracy": 0.8029621839523315, + "num_tokens": 856069774.0, + "step": 32139 + }, + { + "epoch": 2.650284489156428, + "grad_norm": 2.0176243782043457, + "learning_rate": 1e-06, + "loss": 0.5758, + "mean_token_accuracy": 0.7906326651573181, + "num_tokens": 856101837.0, + "step": 32140 + }, + { + "epoch": 2.6503669497814792, + "grad_norm": 2.2426321506500244, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.722447395324707, + "num_tokens": 856130755.0, + "step": 32141 + }, + { + "epoch": 2.650449410406531, + "grad_norm": 2.147512912750244, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.7884328365325928, + "num_tokens": 856159374.0, + "step": 32142 + }, + { + "epoch": 2.6505318710315824, + "grad_norm": 2.0382418632507324, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7652515172958374, + "num_tokens": 856189884.0, + "step": 32143 + }, + { + "epoch": 2.6506143316566337, + "grad_norm": 2.1661272048950195, + "learning_rate": 1e-06, + "loss": 0.6393, + "mean_token_accuracy": 0.7889703512191772, + "num_tokens": 856217796.0, + "step": 32144 + }, + { + "epoch": 2.6506967922816855, + "grad_norm": 2.0724754333496094, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7756298184394836, + "num_tokens": 856249353.0, + "step": 32145 + }, + { + "epoch": 2.650779252906737, + "grad_norm": 2.193943977355957, + "learning_rate": 1e-06, + "loss": 0.63, + "mean_token_accuracy": 0.7928698658943176, + "num_tokens": 856274784.0, + "step": 32146 + }, + { + "epoch": 2.6508617135317887, + "grad_norm": 2.1239678859710693, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.7769889831542969, + "num_tokens": 856307481.0, + "step": 32147 + }, + { + "epoch": 2.65094417415684, + "grad_norm": 1.9948949813842773, + "learning_rate": 1e-06, + "loss": 0.6554, + "mean_token_accuracy": 0.7892586588859558, + "num_tokens": 856342566.0, + "step": 32148 + }, + { + "epoch": 2.651026634781892, + "grad_norm": 2.091926336288452, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7663928866386414, + "num_tokens": 856374799.0, + "step": 32149 + }, + { + "epoch": 2.651109095406943, + "grad_norm": 1.9214986562728882, + "learning_rate": 1e-06, + "loss": 0.618, + "mean_token_accuracy": 0.8053407669067383, + "num_tokens": 856409036.0, + "step": 32150 + }, + { + "epoch": 2.6511915560319945, + "grad_norm": 1.9693223237991333, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7532119750976562, + "num_tokens": 856445604.0, + "step": 32151 + }, + { + "epoch": 2.6512740166570463, + "grad_norm": 2.182682514190674, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7725237607955933, + "num_tokens": 856473071.0, + "step": 32152 + }, + { + "epoch": 2.6513564772820977, + "grad_norm": 2.348388910293579, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.7870297431945801, + "num_tokens": 856495911.0, + "step": 32153 + }, + { + "epoch": 2.6514389379071495, + "grad_norm": 2.3731939792633057, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.732719898223877, + "num_tokens": 856522832.0, + "step": 32154 + }, + { + "epoch": 2.651521398532201, + "grad_norm": 2.0530552864074707, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.7972646355628967, + "num_tokens": 856551167.0, + "step": 32155 + }, + { + "epoch": 2.6516038591572526, + "grad_norm": 1.914475917816162, + "learning_rate": 1e-06, + "loss": 0.6105, + "mean_token_accuracy": 0.7933036088943481, + "num_tokens": 856586074.0, + "step": 32156 + }, + { + "epoch": 2.651686319782304, + "grad_norm": 2.0988094806671143, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7500259876251221, + "num_tokens": 856617237.0, + "step": 32157 + }, + { + "epoch": 2.6517687804073553, + "grad_norm": 2.000795602798462, + "learning_rate": 1e-06, + "loss": 0.6214, + "mean_token_accuracy": 0.7850275635719299, + "num_tokens": 856650054.0, + "step": 32158 + }, + { + "epoch": 2.651851241032407, + "grad_norm": 2.0549678802490234, + "learning_rate": 1e-06, + "loss": 0.5664, + "mean_token_accuracy": 0.8139305114746094, + "num_tokens": 856681161.0, + "step": 32159 + }, + { + "epoch": 2.6519337016574585, + "grad_norm": 2.461270332336426, + "learning_rate": 1e-06, + "loss": 0.7566, + "mean_token_accuracy": 0.7671469449996948, + "num_tokens": 856706297.0, + "step": 32160 + }, + { + "epoch": 2.6520161622825102, + "grad_norm": 2.0589425563812256, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7771740555763245, + "num_tokens": 856740579.0, + "step": 32161 + }, + { + "epoch": 2.6520986229075616, + "grad_norm": 2.6185078620910645, + "learning_rate": 1e-06, + "loss": 0.6109, + "mean_token_accuracy": 0.7936290502548218, + "num_tokens": 856761904.0, + "step": 32162 + }, + { + "epoch": 2.6521810835326134, + "grad_norm": 2.0372605323791504, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7721707224845886, + "num_tokens": 856792562.0, + "step": 32163 + }, + { + "epoch": 2.6522635441576647, + "grad_norm": 1.9296644926071167, + "learning_rate": 1e-06, + "loss": 0.5731, + "mean_token_accuracy": 0.7959508895874023, + "num_tokens": 856822009.0, + "step": 32164 + }, + { + "epoch": 2.652346004782716, + "grad_norm": 2.1321637630462646, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7384103536605835, + "num_tokens": 856851478.0, + "step": 32165 + }, + { + "epoch": 2.652428465407768, + "grad_norm": 1.8472480773925781, + "learning_rate": 1e-06, + "loss": 0.5511, + "mean_token_accuracy": 0.8081680536270142, + "num_tokens": 856885405.0, + "step": 32166 + }, + { + "epoch": 2.6525109260328192, + "grad_norm": 2.149797201156616, + "learning_rate": 1e-06, + "loss": 0.6042, + "mean_token_accuracy": 0.7994106411933899, + "num_tokens": 856913989.0, + "step": 32167 + }, + { + "epoch": 2.652593386657871, + "grad_norm": 2.123838186264038, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7588070034980774, + "num_tokens": 856945077.0, + "step": 32168 + }, + { + "epoch": 2.6526758472829224, + "grad_norm": 2.3025026321411133, + "learning_rate": 1e-06, + "loss": 0.6881, + "mean_token_accuracy": 0.7796509861946106, + "num_tokens": 856972771.0, + "step": 32169 + }, + { + "epoch": 2.652758307907974, + "grad_norm": 2.503678321838379, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7569951415061951, + "num_tokens": 856996677.0, + "step": 32170 + }, + { + "epoch": 2.6528407685330255, + "grad_norm": 2.5246641635894775, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7662325501441956, + "num_tokens": 857020906.0, + "step": 32171 + }, + { + "epoch": 2.652923229158077, + "grad_norm": 2.0000498294830322, + "learning_rate": 1e-06, + "loss": 0.5882, + "mean_token_accuracy": 0.8035035133361816, + "num_tokens": 857049337.0, + "step": 32172 + }, + { + "epoch": 2.6530056897831287, + "grad_norm": 2.4248430728912354, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7513738870620728, + "num_tokens": 857077078.0, + "step": 32173 + }, + { + "epoch": 2.65308815040818, + "grad_norm": 2.2164700031280518, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7733519077301025, + "num_tokens": 857104854.0, + "step": 32174 + }, + { + "epoch": 2.6531706110332314, + "grad_norm": 2.0078651905059814, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.792657196521759, + "num_tokens": 857139346.0, + "step": 32175 + }, + { + "epoch": 2.653253071658283, + "grad_norm": 1.9484167098999023, + "learning_rate": 1e-06, + "loss": 0.5338, + "mean_token_accuracy": 0.8150089979171753, + "num_tokens": 857172190.0, + "step": 32176 + }, + { + "epoch": 2.653335532283335, + "grad_norm": 2.2049179077148438, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7877373695373535, + "num_tokens": 857201981.0, + "step": 32177 + }, + { + "epoch": 2.6534179929083863, + "grad_norm": 2.2975592613220215, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7613661885261536, + "num_tokens": 857228206.0, + "step": 32178 + }, + { + "epoch": 2.6535004535334377, + "grad_norm": 2.0802159309387207, + "learning_rate": 1e-06, + "loss": 0.7161, + "mean_token_accuracy": 0.7660099864006042, + "num_tokens": 857258712.0, + "step": 32179 + }, + { + "epoch": 2.6535829141584895, + "grad_norm": 2.4865176677703857, + "learning_rate": 1e-06, + "loss": 0.5998, + "mean_token_accuracy": 0.7940929532051086, + "num_tokens": 857282057.0, + "step": 32180 + }, + { + "epoch": 2.653665374783541, + "grad_norm": 2.0512478351593018, + "learning_rate": 1e-06, + "loss": 0.7159, + "mean_token_accuracy": 0.7730737924575806, + "num_tokens": 857310758.0, + "step": 32181 + }, + { + "epoch": 2.653747835408592, + "grad_norm": 2.0002875328063965, + "learning_rate": 1e-06, + "loss": 0.7665, + "mean_token_accuracy": 0.7430696487426758, + "num_tokens": 857344542.0, + "step": 32182 + }, + { + "epoch": 2.653830296033644, + "grad_norm": 2.1349384784698486, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7333942651748657, + "num_tokens": 857377384.0, + "step": 32183 + }, + { + "epoch": 2.6539127566586957, + "grad_norm": 2.046818971633911, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7791546583175659, + "num_tokens": 857406363.0, + "step": 32184 + }, + { + "epoch": 2.653995217283747, + "grad_norm": 2.109157085418701, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7413017153739929, + "num_tokens": 857438132.0, + "step": 32185 + }, + { + "epoch": 2.6540776779087984, + "grad_norm": 2.0985584259033203, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7730278372764587, + "num_tokens": 857467283.0, + "step": 32186 + }, + { + "epoch": 2.6541601385338502, + "grad_norm": 2.089637041091919, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.788438618183136, + "num_tokens": 857495479.0, + "step": 32187 + }, + { + "epoch": 2.6542425991589016, + "grad_norm": 2.045823812484741, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.7976893782615662, + "num_tokens": 857525141.0, + "step": 32188 + }, + { + "epoch": 2.654325059783953, + "grad_norm": 2.382594347000122, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7570192813873291, + "num_tokens": 857549416.0, + "step": 32189 + }, + { + "epoch": 2.6544075204090047, + "grad_norm": 1.8651902675628662, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.7967499494552612, + "num_tokens": 857586349.0, + "step": 32190 + }, + { + "epoch": 2.654489981034056, + "grad_norm": 1.8460123538970947, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.8053090572357178, + "num_tokens": 857622475.0, + "step": 32191 + }, + { + "epoch": 2.654572441659108, + "grad_norm": 2.264216661453247, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7738243341445923, + "num_tokens": 857651566.0, + "step": 32192 + }, + { + "epoch": 2.6546549022841592, + "grad_norm": 2.2086029052734375, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7521212100982666, + "num_tokens": 857680790.0, + "step": 32193 + }, + { + "epoch": 2.654737362909211, + "grad_norm": 2.2999541759490967, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7796735167503357, + "num_tokens": 857706197.0, + "step": 32194 + }, + { + "epoch": 2.6548198235342624, + "grad_norm": 1.9798189401626587, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.7619432806968689, + "num_tokens": 857741485.0, + "step": 32195 + }, + { + "epoch": 2.6549022841593137, + "grad_norm": 2.270387649536133, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7730356454849243, + "num_tokens": 857767486.0, + "step": 32196 + }, + { + "epoch": 2.6549847447843655, + "grad_norm": 1.9941842555999756, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.7929230332374573, + "num_tokens": 857799055.0, + "step": 32197 + }, + { + "epoch": 2.655067205409417, + "grad_norm": 2.207214593887329, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.7857009172439575, + "num_tokens": 857825332.0, + "step": 32198 + }, + { + "epoch": 2.6551496660344687, + "grad_norm": 1.9492758512496948, + "learning_rate": 1e-06, + "loss": 0.6781, + "mean_token_accuracy": 0.7826074957847595, + "num_tokens": 857857765.0, + "step": 32199 + }, + { + "epoch": 2.65523212665952, + "grad_norm": 2.035468816757202, + "learning_rate": 1e-06, + "loss": 0.593, + "mean_token_accuracy": 0.7999982237815857, + "num_tokens": 857889046.0, + "step": 32200 + }, + { + "epoch": 2.655314587284572, + "grad_norm": 2.2195518016815186, + "learning_rate": 1e-06, + "loss": 0.732, + "mean_token_accuracy": 0.7623700499534607, + "num_tokens": 857919044.0, + "step": 32201 + }, + { + "epoch": 2.655397047909623, + "grad_norm": 2.2572391033172607, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.739605724811554, + "num_tokens": 857945757.0, + "step": 32202 + }, + { + "epoch": 2.6554795085346745, + "grad_norm": 1.970828652381897, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.8022325038909912, + "num_tokens": 857978729.0, + "step": 32203 + }, + { + "epoch": 2.6555619691597263, + "grad_norm": 1.9118850231170654, + "learning_rate": 1e-06, + "loss": 0.5735, + "mean_token_accuracy": 0.8143924474716187, + "num_tokens": 858011006.0, + "step": 32204 + }, + { + "epoch": 2.6556444297847777, + "grad_norm": 1.9824397563934326, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7788751125335693, + "num_tokens": 858046247.0, + "step": 32205 + }, + { + "epoch": 2.6557268904098295, + "grad_norm": 2.3181819915771484, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7390875816345215, + "num_tokens": 858071253.0, + "step": 32206 + }, + { + "epoch": 2.655809351034881, + "grad_norm": 1.9921371936798096, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.7920702695846558, + "num_tokens": 858102657.0, + "step": 32207 + }, + { + "epoch": 2.6558918116599326, + "grad_norm": 2.40252947807312, + "learning_rate": 1e-06, + "loss": 0.7624, + "mean_token_accuracy": 0.7589816451072693, + "num_tokens": 858127969.0, + "step": 32208 + }, + { + "epoch": 2.655974272284984, + "grad_norm": 2.1122629642486572, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7208631038665771, + "num_tokens": 858162035.0, + "step": 32209 + }, + { + "epoch": 2.6560567329100353, + "grad_norm": 2.012073278427124, + "learning_rate": 1e-06, + "loss": 0.7074, + "mean_token_accuracy": 0.7695437073707581, + "num_tokens": 858195375.0, + "step": 32210 + }, + { + "epoch": 2.656139193535087, + "grad_norm": 1.8966063261032104, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7746946811676025, + "num_tokens": 858231323.0, + "step": 32211 + }, + { + "epoch": 2.6562216541601384, + "grad_norm": 2.260380983352661, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.7794130444526672, + "num_tokens": 858255903.0, + "step": 32212 + }, + { + "epoch": 2.65630411478519, + "grad_norm": 2.007376194000244, + "learning_rate": 1e-06, + "loss": 0.6118, + "mean_token_accuracy": 0.798680305480957, + "num_tokens": 858290577.0, + "step": 32213 + }, + { + "epoch": 2.6563865754102416, + "grad_norm": 1.741908073425293, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.789110541343689, + "num_tokens": 858332826.0, + "step": 32214 + }, + { + "epoch": 2.6564690360352934, + "grad_norm": 2.122448444366455, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.7900761365890503, + "num_tokens": 858360572.0, + "step": 32215 + }, + { + "epoch": 2.6565514966603447, + "grad_norm": 2.0159354209899902, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.802527904510498, + "num_tokens": 858393569.0, + "step": 32216 + }, + { + "epoch": 2.656633957285396, + "grad_norm": 2.0736567974090576, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.7938275933265686, + "num_tokens": 858425076.0, + "step": 32217 + }, + { + "epoch": 2.656716417910448, + "grad_norm": 2.012258529663086, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7933402061462402, + "num_tokens": 858457706.0, + "step": 32218 + }, + { + "epoch": 2.6567988785354992, + "grad_norm": 2.2510268688201904, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.76242995262146, + "num_tokens": 858485788.0, + "step": 32219 + }, + { + "epoch": 2.6568813391605506, + "grad_norm": 1.8711541891098022, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7765154242515564, + "num_tokens": 858520298.0, + "step": 32220 + }, + { + "epoch": 2.6569637997856024, + "grad_norm": 2.377527952194214, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.770655632019043, + "num_tokens": 858545189.0, + "step": 32221 + }, + { + "epoch": 2.657046260410654, + "grad_norm": 1.9159883260726929, + "learning_rate": 1e-06, + "loss": 0.6474, + "mean_token_accuracy": 0.7740546464920044, + "num_tokens": 858581089.0, + "step": 32222 + }, + { + "epoch": 2.6571287210357055, + "grad_norm": 2.4049227237701416, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7738075256347656, + "num_tokens": 858607522.0, + "step": 32223 + }, + { + "epoch": 2.657211181660757, + "grad_norm": 2.297487735748291, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7670930624008179, + "num_tokens": 858633473.0, + "step": 32224 + }, + { + "epoch": 2.6572936422858087, + "grad_norm": 2.185284376144409, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7604700922966003, + "num_tokens": 858661711.0, + "step": 32225 + }, + { + "epoch": 2.65737610291086, + "grad_norm": 2.283665418624878, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7668721675872803, + "num_tokens": 858689457.0, + "step": 32226 + }, + { + "epoch": 2.6574585635359114, + "grad_norm": 2.282259941101074, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.7819263935089111, + "num_tokens": 858719376.0, + "step": 32227 + }, + { + "epoch": 2.657541024160963, + "grad_norm": 2.5198192596435547, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7753457427024841, + "num_tokens": 858742695.0, + "step": 32228 + }, + { + "epoch": 2.6576234847860145, + "grad_norm": 2.2766175270080566, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7453079223632812, + "num_tokens": 858771904.0, + "step": 32229 + }, + { + "epoch": 2.6577059454110663, + "grad_norm": 2.03493332862854, + "learning_rate": 1e-06, + "loss": 0.6373, + "mean_token_accuracy": 0.7926709651947021, + "num_tokens": 858802039.0, + "step": 32230 + }, + { + "epoch": 2.6577884060361177, + "grad_norm": 2.0663163661956787, + "learning_rate": 1e-06, + "loss": 0.7563, + "mean_token_accuracy": 0.7548797130584717, + "num_tokens": 858833232.0, + "step": 32231 + }, + { + "epoch": 2.6578708666611695, + "grad_norm": 2.225034236907959, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7743975520133972, + "num_tokens": 858860162.0, + "step": 32232 + }, + { + "epoch": 2.657953327286221, + "grad_norm": 2.111875057220459, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.7945078611373901, + "num_tokens": 858888182.0, + "step": 32233 + }, + { + "epoch": 2.658035787911272, + "grad_norm": 1.9608036279678345, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.790908932685852, + "num_tokens": 858923179.0, + "step": 32234 + }, + { + "epoch": 2.658118248536324, + "grad_norm": 1.928521990776062, + "learning_rate": 1e-06, + "loss": 0.7543, + "mean_token_accuracy": 0.7545413374900818, + "num_tokens": 858956884.0, + "step": 32235 + }, + { + "epoch": 2.6582007091613753, + "grad_norm": 1.8502150774002075, + "learning_rate": 1e-06, + "loss": 0.5094, + "mean_token_accuracy": 0.827370285987854, + "num_tokens": 858992844.0, + "step": 32236 + }, + { + "epoch": 2.658283169786427, + "grad_norm": 2.2769572734832764, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7427782416343689, + "num_tokens": 859022062.0, + "step": 32237 + }, + { + "epoch": 2.6583656304114784, + "grad_norm": 2.0977821350097656, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7605102062225342, + "num_tokens": 859057428.0, + "step": 32238 + }, + { + "epoch": 2.6584480910365302, + "grad_norm": 2.2800633907318115, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7642377614974976, + "num_tokens": 859082848.0, + "step": 32239 + }, + { + "epoch": 2.6585305516615816, + "grad_norm": 2.23852801322937, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7555555105209351, + "num_tokens": 859109324.0, + "step": 32240 + }, + { + "epoch": 2.658613012286633, + "grad_norm": 2.0124402046203613, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.791631817817688, + "num_tokens": 859138666.0, + "step": 32241 + }, + { + "epoch": 2.6586954729116847, + "grad_norm": 2.261362314224243, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7593058347702026, + "num_tokens": 859165166.0, + "step": 32242 + }, + { + "epoch": 2.658777933536736, + "grad_norm": 2.056619167327881, + "learning_rate": 1e-06, + "loss": 0.6003, + "mean_token_accuracy": 0.8017849326133728, + "num_tokens": 859195742.0, + "step": 32243 + }, + { + "epoch": 2.658860394161788, + "grad_norm": 2.2133266925811768, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7301222085952759, + "num_tokens": 859225840.0, + "step": 32244 + }, + { + "epoch": 2.6589428547868392, + "grad_norm": 2.1987123489379883, + "learning_rate": 1e-06, + "loss": 0.5775, + "mean_token_accuracy": 0.806755542755127, + "num_tokens": 859252534.0, + "step": 32245 + }, + { + "epoch": 2.659025315411891, + "grad_norm": 1.958237648010254, + "learning_rate": 1e-06, + "loss": 0.5372, + "mean_token_accuracy": 0.8189529776573181, + "num_tokens": 859285216.0, + "step": 32246 + }, + { + "epoch": 2.6591077760369424, + "grad_norm": 2.0560436248779297, + "learning_rate": 1e-06, + "loss": 0.629, + "mean_token_accuracy": 0.7930524349212646, + "num_tokens": 859316888.0, + "step": 32247 + }, + { + "epoch": 2.6591902366619937, + "grad_norm": 1.8937158584594727, + "learning_rate": 1e-06, + "loss": 0.5386, + "mean_token_accuracy": 0.8221657276153564, + "num_tokens": 859348685.0, + "step": 32248 + }, + { + "epoch": 2.6592726972870455, + "grad_norm": 2.2354369163513184, + "learning_rate": 1e-06, + "loss": 0.5598, + "mean_token_accuracy": 0.8115963935852051, + "num_tokens": 859373708.0, + "step": 32249 + }, + { + "epoch": 2.659355157912097, + "grad_norm": 2.1054623126983643, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7758398652076721, + "num_tokens": 859403877.0, + "step": 32250 + }, + { + "epoch": 2.6594376185371487, + "grad_norm": 2.3923611640930176, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7684991955757141, + "num_tokens": 859427702.0, + "step": 32251 + }, + { + "epoch": 2.6595200791622, + "grad_norm": 2.1860976219177246, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.7475192546844482, + "num_tokens": 859459117.0, + "step": 32252 + }, + { + "epoch": 2.659602539787252, + "grad_norm": 1.9483258724212646, + "learning_rate": 1e-06, + "loss": 0.6857, + "mean_token_accuracy": 0.7749884128570557, + "num_tokens": 859492648.0, + "step": 32253 + }, + { + "epoch": 2.659685000412303, + "grad_norm": 1.98592209815979, + "learning_rate": 1e-06, + "loss": 0.5627, + "mean_token_accuracy": 0.8107576370239258, + "num_tokens": 859522475.0, + "step": 32254 + }, + { + "epoch": 2.6597674610373545, + "grad_norm": 2.4177560806274414, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.773979663848877, + "num_tokens": 859546384.0, + "step": 32255 + }, + { + "epoch": 2.6598499216624063, + "grad_norm": 2.2128567695617676, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.77707839012146, + "num_tokens": 859571491.0, + "step": 32256 + }, + { + "epoch": 2.6599323822874577, + "grad_norm": 2.106685161590576, + "learning_rate": 1e-06, + "loss": 0.5897, + "mean_token_accuracy": 0.8042741417884827, + "num_tokens": 859598465.0, + "step": 32257 + }, + { + "epoch": 2.660014842912509, + "grad_norm": 2.0961811542510986, + "learning_rate": 1e-06, + "loss": 0.5827, + "mean_token_accuracy": 0.8028934597969055, + "num_tokens": 859628151.0, + "step": 32258 + }, + { + "epoch": 2.660097303537561, + "grad_norm": 2.182591676712036, + "learning_rate": 1e-06, + "loss": 0.6042, + "mean_token_accuracy": 0.7866183519363403, + "num_tokens": 859656031.0, + "step": 32259 + }, + { + "epoch": 2.6601797641626126, + "grad_norm": 2.143892288208008, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.7627673149108887, + "num_tokens": 859685276.0, + "step": 32260 + }, + { + "epoch": 2.660262224787664, + "grad_norm": 1.9371494054794312, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7491093873977661, + "num_tokens": 859720365.0, + "step": 32261 + }, + { + "epoch": 2.6603446854127153, + "grad_norm": 2.0386884212493896, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7766169309616089, + "num_tokens": 859751785.0, + "step": 32262 + }, + { + "epoch": 2.660427146037767, + "grad_norm": 2.0392463207244873, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7729836702346802, + "num_tokens": 859784376.0, + "step": 32263 + }, + { + "epoch": 2.6605096066628184, + "grad_norm": 2.0306506156921387, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7734246850013733, + "num_tokens": 859817235.0, + "step": 32264 + }, + { + "epoch": 2.66059206728787, + "grad_norm": 2.1987597942352295, + "learning_rate": 1e-06, + "loss": 0.5878, + "mean_token_accuracy": 0.7868325710296631, + "num_tokens": 859844312.0, + "step": 32265 + }, + { + "epoch": 2.6606745279129216, + "grad_norm": 1.9264508485794067, + "learning_rate": 1e-06, + "loss": 0.6513, + "mean_token_accuracy": 0.7793688178062439, + "num_tokens": 859877862.0, + "step": 32266 + }, + { + "epoch": 2.6607569885379734, + "grad_norm": 2.1556200981140137, + "learning_rate": 1e-06, + "loss": 0.7382, + "mean_token_accuracy": 0.7731367349624634, + "num_tokens": 859905180.0, + "step": 32267 + }, + { + "epoch": 2.6608394491630247, + "grad_norm": 2.133120536804199, + "learning_rate": 1e-06, + "loss": 0.6387, + "mean_token_accuracy": 0.7767384052276611, + "num_tokens": 859932409.0, + "step": 32268 + }, + { + "epoch": 2.660921909788076, + "grad_norm": 1.9495619535446167, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7511720657348633, + "num_tokens": 859969533.0, + "step": 32269 + }, + { + "epoch": 2.661004370413128, + "grad_norm": 2.0228936672210693, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7699893116950989, + "num_tokens": 860003207.0, + "step": 32270 + }, + { + "epoch": 2.6610868310381792, + "grad_norm": 2.216517925262451, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7596142292022705, + "num_tokens": 860032492.0, + "step": 32271 + }, + { + "epoch": 2.6611692916632306, + "grad_norm": 2.0332772731781006, + "learning_rate": 1e-06, + "loss": 0.5601, + "mean_token_accuracy": 0.812309980392456, + "num_tokens": 860060385.0, + "step": 32272 + }, + { + "epoch": 2.6612517522882824, + "grad_norm": 2.1499478816986084, + "learning_rate": 1e-06, + "loss": 0.6205, + "mean_token_accuracy": 0.7896384000778198, + "num_tokens": 860087838.0, + "step": 32273 + }, + { + "epoch": 2.6613342129133337, + "grad_norm": 1.9554544687271118, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7773756980895996, + "num_tokens": 860120750.0, + "step": 32274 + }, + { + "epoch": 2.6614166735383855, + "grad_norm": 2.0382065773010254, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.7908576726913452, + "num_tokens": 860150509.0, + "step": 32275 + }, + { + "epoch": 2.661499134163437, + "grad_norm": 2.2129688262939453, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.7887210845947266, + "num_tokens": 860179196.0, + "step": 32276 + }, + { + "epoch": 2.6615815947884887, + "grad_norm": 2.215479850769043, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7731733918190002, + "num_tokens": 860207894.0, + "step": 32277 + }, + { + "epoch": 2.66166405541354, + "grad_norm": 2.4712586402893066, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7514910101890564, + "num_tokens": 860230049.0, + "step": 32278 + }, + { + "epoch": 2.6617465160385914, + "grad_norm": 2.0220062732696533, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.8069692254066467, + "num_tokens": 860260108.0, + "step": 32279 + }, + { + "epoch": 2.661828976663643, + "grad_norm": 2.2066473960876465, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7836357355117798, + "num_tokens": 860290060.0, + "step": 32280 + }, + { + "epoch": 2.6619114372886945, + "grad_norm": 2.2178955078125, + "learning_rate": 1e-06, + "loss": 0.5695, + "mean_token_accuracy": 0.809851348400116, + "num_tokens": 860315851.0, + "step": 32281 + }, + { + "epoch": 2.6619938979137463, + "grad_norm": 1.9722577333450317, + "learning_rate": 1e-06, + "loss": 0.5889, + "mean_token_accuracy": 0.782796323299408, + "num_tokens": 860347257.0, + "step": 32282 + }, + { + "epoch": 2.6620763585387976, + "grad_norm": 2.3512513637542725, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.7976157665252686, + "num_tokens": 860374866.0, + "step": 32283 + }, + { + "epoch": 2.6621588191638494, + "grad_norm": 2.298903226852417, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.7904197573661804, + "num_tokens": 860399938.0, + "step": 32284 + }, + { + "epoch": 2.662241279788901, + "grad_norm": 1.9632176160812378, + "learning_rate": 1e-06, + "loss": 0.6658, + "mean_token_accuracy": 0.7763798236846924, + "num_tokens": 860430191.0, + "step": 32285 + }, + { + "epoch": 2.662323740413952, + "grad_norm": 1.8727909326553345, + "learning_rate": 1e-06, + "loss": 0.5998, + "mean_token_accuracy": 0.7990924119949341, + "num_tokens": 860466942.0, + "step": 32286 + }, + { + "epoch": 2.662406201039004, + "grad_norm": 2.3087570667266846, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.786359429359436, + "num_tokens": 860489571.0, + "step": 32287 + }, + { + "epoch": 2.6624886616640553, + "grad_norm": 1.9202518463134766, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.8018714189529419, + "num_tokens": 860521993.0, + "step": 32288 + }, + { + "epoch": 2.662571122289107, + "grad_norm": 1.861474633216858, + "learning_rate": 1e-06, + "loss": 0.59, + "mean_token_accuracy": 0.8129124641418457, + "num_tokens": 860556685.0, + "step": 32289 + }, + { + "epoch": 2.6626535829141584, + "grad_norm": 2.268733501434326, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7715195417404175, + "num_tokens": 860586011.0, + "step": 32290 + }, + { + "epoch": 2.6627360435392102, + "grad_norm": 2.090989828109741, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7896975874900818, + "num_tokens": 860616115.0, + "step": 32291 + }, + { + "epoch": 2.6628185041642616, + "grad_norm": 2.2040154933929443, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7522768974304199, + "num_tokens": 860644725.0, + "step": 32292 + }, + { + "epoch": 2.662900964789313, + "grad_norm": 1.9392328262329102, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7549821138381958, + "num_tokens": 860680304.0, + "step": 32293 + }, + { + "epoch": 2.6629834254143647, + "grad_norm": 1.8046261072158813, + "learning_rate": 1e-06, + "loss": 0.5838, + "mean_token_accuracy": 0.8063511848449707, + "num_tokens": 860718634.0, + "step": 32294 + }, + { + "epoch": 2.663065886039416, + "grad_norm": 2.169792413711548, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.756698489189148, + "num_tokens": 860748292.0, + "step": 32295 + }, + { + "epoch": 2.663148346664468, + "grad_norm": 1.9303338527679443, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.7895570993423462, + "num_tokens": 860783264.0, + "step": 32296 + }, + { + "epoch": 2.663230807289519, + "grad_norm": 2.1759092807769775, + "learning_rate": 1e-06, + "loss": 0.5135, + "mean_token_accuracy": 0.8267327547073364, + "num_tokens": 860809493.0, + "step": 32297 + }, + { + "epoch": 2.663313267914571, + "grad_norm": 2.2709195613861084, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7712559700012207, + "num_tokens": 860838298.0, + "step": 32298 + }, + { + "epoch": 2.6633957285396224, + "grad_norm": 2.1531810760498047, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7428755164146423, + "num_tokens": 860870452.0, + "step": 32299 + }, + { + "epoch": 2.6634781891646737, + "grad_norm": 2.002631902694702, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.775997519493103, + "num_tokens": 860903424.0, + "step": 32300 + }, + { + "epoch": 2.6635606497897255, + "grad_norm": 2.0853893756866455, + "learning_rate": 1e-06, + "loss": 0.5914, + "mean_token_accuracy": 0.8012427091598511, + "num_tokens": 860934594.0, + "step": 32301 + }, + { + "epoch": 2.663643110414777, + "grad_norm": 1.9027348756790161, + "learning_rate": 1e-06, + "loss": 0.624, + "mean_token_accuracy": 0.7827807068824768, + "num_tokens": 860970584.0, + "step": 32302 + }, + { + "epoch": 2.663725571039828, + "grad_norm": 2.076664924621582, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.7896857857704163, + "num_tokens": 861002158.0, + "step": 32303 + }, + { + "epoch": 2.66380803166488, + "grad_norm": 2.0245282649993896, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7413183450698853, + "num_tokens": 861036255.0, + "step": 32304 + }, + { + "epoch": 2.663890492289932, + "grad_norm": 2.266165018081665, + "learning_rate": 1e-06, + "loss": 0.5955, + "mean_token_accuracy": 0.7973712682723999, + "num_tokens": 861063045.0, + "step": 32305 + }, + { + "epoch": 2.663972952914983, + "grad_norm": 2.1107139587402344, + "learning_rate": 1e-06, + "loss": 0.6866, + "mean_token_accuracy": 0.77392578125, + "num_tokens": 861095413.0, + "step": 32306 + }, + { + "epoch": 2.6640554135400345, + "grad_norm": 2.289407253265381, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.7546719312667847, + "num_tokens": 861121285.0, + "step": 32307 + }, + { + "epoch": 2.6641378741650863, + "grad_norm": 2.0924675464630127, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7561420202255249, + "num_tokens": 861152187.0, + "step": 32308 + }, + { + "epoch": 2.6642203347901376, + "grad_norm": 2.568744659423828, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7648729681968689, + "num_tokens": 861173397.0, + "step": 32309 + }, + { + "epoch": 2.664302795415189, + "grad_norm": 1.977864384651184, + "learning_rate": 1e-06, + "loss": 0.5575, + "mean_token_accuracy": 0.8110760450363159, + "num_tokens": 861204065.0, + "step": 32310 + }, + { + "epoch": 2.664385256040241, + "grad_norm": 1.8859094381332397, + "learning_rate": 1e-06, + "loss": 0.5793, + "mean_token_accuracy": 0.807879626750946, + "num_tokens": 861236560.0, + "step": 32311 + }, + { + "epoch": 2.6644677166652926, + "grad_norm": 2.1011641025543213, + "learning_rate": 1e-06, + "loss": 0.7401, + "mean_token_accuracy": 0.7608358860015869, + "num_tokens": 861268085.0, + "step": 32312 + }, + { + "epoch": 2.664550177290344, + "grad_norm": 2.2528533935546875, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7421529293060303, + "num_tokens": 861293832.0, + "step": 32313 + }, + { + "epoch": 2.6646326379153953, + "grad_norm": 2.0283615589141846, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7549616098403931, + "num_tokens": 861325206.0, + "step": 32314 + }, + { + "epoch": 2.664715098540447, + "grad_norm": 2.0474820137023926, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.7835471034049988, + "num_tokens": 861357854.0, + "step": 32315 + }, + { + "epoch": 2.6647975591654984, + "grad_norm": 2.120046377182007, + "learning_rate": 1e-06, + "loss": 0.6466, + "mean_token_accuracy": 0.7828382849693298, + "num_tokens": 861390864.0, + "step": 32316 + }, + { + "epoch": 2.66488001979055, + "grad_norm": 2.2913506031036377, + "learning_rate": 1e-06, + "loss": 0.7595, + "mean_token_accuracy": 0.7561392784118652, + "num_tokens": 861419744.0, + "step": 32317 + }, + { + "epoch": 2.6649624804156016, + "grad_norm": 2.1813015937805176, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7784903645515442, + "num_tokens": 861448060.0, + "step": 32318 + }, + { + "epoch": 2.665044941040653, + "grad_norm": 2.2639079093933105, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.730994462966919, + "num_tokens": 861475420.0, + "step": 32319 + }, + { + "epoch": 2.6651274016657047, + "grad_norm": 1.8996063470840454, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.778412401676178, + "num_tokens": 861507082.0, + "step": 32320 + }, + { + "epoch": 2.665209862290756, + "grad_norm": 1.8652331829071045, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.762502908706665, + "num_tokens": 861541671.0, + "step": 32321 + }, + { + "epoch": 2.665292322915808, + "grad_norm": 2.196380615234375, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.729616641998291, + "num_tokens": 861576003.0, + "step": 32322 + }, + { + "epoch": 2.665374783540859, + "grad_norm": 2.0080180168151855, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7720246315002441, + "num_tokens": 861610771.0, + "step": 32323 + }, + { + "epoch": 2.6654572441659106, + "grad_norm": 1.9548622369766235, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7358958125114441, + "num_tokens": 861643515.0, + "step": 32324 + }, + { + "epoch": 2.6655397047909624, + "grad_norm": 2.4242961406707764, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7725751399993896, + "num_tokens": 861666688.0, + "step": 32325 + }, + { + "epoch": 2.6656221654160137, + "grad_norm": 2.121382474899292, + "learning_rate": 1e-06, + "loss": 0.5617, + "mean_token_accuracy": 0.8126872181892395, + "num_tokens": 861698565.0, + "step": 32326 + }, + { + "epoch": 2.6657046260410655, + "grad_norm": 2.108696937561035, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7354618906974792, + "num_tokens": 861729431.0, + "step": 32327 + }, + { + "epoch": 2.665787086666117, + "grad_norm": 2.1096625328063965, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7667152881622314, + "num_tokens": 861759660.0, + "step": 32328 + }, + { + "epoch": 2.6658695472911687, + "grad_norm": 2.101208209991455, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.7898189425468445, + "num_tokens": 861790621.0, + "step": 32329 + }, + { + "epoch": 2.66595200791622, + "grad_norm": 2.035072088241577, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7375128269195557, + "num_tokens": 861824345.0, + "step": 32330 + }, + { + "epoch": 2.6660344685412714, + "grad_norm": 2.7902917861938477, + "learning_rate": 1e-06, + "loss": 0.8081, + "mean_token_accuracy": 0.7380006313323975, + "num_tokens": 861845121.0, + "step": 32331 + }, + { + "epoch": 2.666116929166323, + "grad_norm": 2.0490965843200684, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7733434438705444, + "num_tokens": 861874830.0, + "step": 32332 + }, + { + "epoch": 2.6661993897913745, + "grad_norm": 2.1170005798339844, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7296951413154602, + "num_tokens": 861907020.0, + "step": 32333 + }, + { + "epoch": 2.6662818504164263, + "grad_norm": 1.9665383100509644, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.7891048789024353, + "num_tokens": 861939904.0, + "step": 32334 + }, + { + "epoch": 2.6663643110414776, + "grad_norm": 1.8881802558898926, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7593355178833008, + "num_tokens": 861977241.0, + "step": 32335 + }, + { + "epoch": 2.6664467716665294, + "grad_norm": 2.1227126121520996, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7609422206878662, + "num_tokens": 862009275.0, + "step": 32336 + }, + { + "epoch": 2.666529232291581, + "grad_norm": 2.120201826095581, + "learning_rate": 1e-06, + "loss": 0.5683, + "mean_token_accuracy": 0.8065658807754517, + "num_tokens": 862037290.0, + "step": 32337 + }, + { + "epoch": 2.666611692916632, + "grad_norm": 2.0219180583953857, + "learning_rate": 1e-06, + "loss": 0.758, + "mean_token_accuracy": 0.7579606771469116, + "num_tokens": 862069642.0, + "step": 32338 + }, + { + "epoch": 2.666694153541684, + "grad_norm": 1.8996800184249878, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.7832298874855042, + "num_tokens": 862105101.0, + "step": 32339 + }, + { + "epoch": 2.6667766141667353, + "grad_norm": 2.7398123741149902, + "learning_rate": 1e-06, + "loss": 0.7637, + "mean_token_accuracy": 0.7595111131668091, + "num_tokens": 862125122.0, + "step": 32340 + }, + { + "epoch": 2.666859074791787, + "grad_norm": 2.1675214767456055, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.7905938625335693, + "num_tokens": 862153210.0, + "step": 32341 + }, + { + "epoch": 2.6669415354168384, + "grad_norm": 2.4015209674835205, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7482887506484985, + "num_tokens": 862179793.0, + "step": 32342 + }, + { + "epoch": 2.66702399604189, + "grad_norm": 2.0671818256378174, + "learning_rate": 1e-06, + "loss": 0.7627, + "mean_token_accuracy": 0.750542163848877, + "num_tokens": 862212410.0, + "step": 32343 + }, + { + "epoch": 2.6671064566669416, + "grad_norm": 2.335379123687744, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.7599788308143616, + "num_tokens": 862239253.0, + "step": 32344 + }, + { + "epoch": 2.667188917291993, + "grad_norm": 1.9785184860229492, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7565803527832031, + "num_tokens": 862273111.0, + "step": 32345 + }, + { + "epoch": 2.6672713779170447, + "grad_norm": 2.0236880779266357, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7682366967201233, + "num_tokens": 862303543.0, + "step": 32346 + }, + { + "epoch": 2.667353838542096, + "grad_norm": 2.194007158279419, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7529038190841675, + "num_tokens": 862333478.0, + "step": 32347 + }, + { + "epoch": 2.6674362991671474, + "grad_norm": 2.251375198364258, + "learning_rate": 1e-06, + "loss": 0.6116, + "mean_token_accuracy": 0.7884423732757568, + "num_tokens": 862359831.0, + "step": 32348 + }, + { + "epoch": 2.667518759792199, + "grad_norm": 2.2425997257232666, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7739959359169006, + "num_tokens": 862386002.0, + "step": 32349 + }, + { + "epoch": 2.667601220417251, + "grad_norm": 2.048645496368408, + "learning_rate": 1e-06, + "loss": 0.7272, + "mean_token_accuracy": 0.7643488645553589, + "num_tokens": 862419202.0, + "step": 32350 + }, + { + "epoch": 2.6676836810423024, + "grad_norm": 1.8838508129119873, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.8014390468597412, + "num_tokens": 862453818.0, + "step": 32351 + }, + { + "epoch": 2.6677661416673537, + "grad_norm": 2.0696544647216797, + "learning_rate": 1e-06, + "loss": 0.7921, + "mean_token_accuracy": 0.7499557733535767, + "num_tokens": 862483212.0, + "step": 32352 + }, + { + "epoch": 2.6678486022924055, + "grad_norm": 2.114243268966675, + "learning_rate": 1e-06, + "loss": 0.5788, + "mean_token_accuracy": 0.7997417449951172, + "num_tokens": 862509148.0, + "step": 32353 + }, + { + "epoch": 2.667931062917457, + "grad_norm": 2.2328832149505615, + "learning_rate": 1e-06, + "loss": 0.5429, + "mean_token_accuracy": 0.8173091411590576, + "num_tokens": 862535441.0, + "step": 32354 + }, + { + "epoch": 2.668013523542508, + "grad_norm": 1.9252421855926514, + "learning_rate": 1e-06, + "loss": 0.5703, + "mean_token_accuracy": 0.7976098656654358, + "num_tokens": 862569012.0, + "step": 32355 + }, + { + "epoch": 2.66809598416756, + "grad_norm": 1.9986650943756104, + "learning_rate": 1e-06, + "loss": 0.7747, + "mean_token_accuracy": 0.7551669478416443, + "num_tokens": 862604126.0, + "step": 32356 + }, + { + "epoch": 2.668178444792612, + "grad_norm": 1.8798863887786865, + "learning_rate": 1e-06, + "loss": 0.7316, + "mean_token_accuracy": 0.7649757266044617, + "num_tokens": 862640253.0, + "step": 32357 + }, + { + "epoch": 2.668260905417663, + "grad_norm": 1.952778935432434, + "learning_rate": 1e-06, + "loss": 0.6551, + "mean_token_accuracy": 0.7791392207145691, + "num_tokens": 862674042.0, + "step": 32358 + }, + { + "epoch": 2.6683433660427145, + "grad_norm": 2.0088796615600586, + "learning_rate": 1e-06, + "loss": 0.7328, + "mean_token_accuracy": 0.7699487209320068, + "num_tokens": 862708009.0, + "step": 32359 + }, + { + "epoch": 2.6684258266677663, + "grad_norm": 2.0465869903564453, + "learning_rate": 1e-06, + "loss": 0.6633, + "mean_token_accuracy": 0.7864934206008911, + "num_tokens": 862736212.0, + "step": 32360 + }, + { + "epoch": 2.6685082872928176, + "grad_norm": 1.9919720888137817, + "learning_rate": 1e-06, + "loss": 0.5952, + "mean_token_accuracy": 0.8026323318481445, + "num_tokens": 862769065.0, + "step": 32361 + }, + { + "epoch": 2.668590747917869, + "grad_norm": 1.8028024435043335, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.7935292720794678, + "num_tokens": 862806431.0, + "step": 32362 + }, + { + "epoch": 2.668673208542921, + "grad_norm": 2.312448501586914, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7620072960853577, + "num_tokens": 862833090.0, + "step": 32363 + }, + { + "epoch": 2.668755669167972, + "grad_norm": 1.8783811330795288, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7899734377861023, + "num_tokens": 862871814.0, + "step": 32364 + }, + { + "epoch": 2.668838129793024, + "grad_norm": 1.9665108919143677, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7787063717842102, + "num_tokens": 862906420.0, + "step": 32365 + }, + { + "epoch": 2.6689205904180753, + "grad_norm": 2.0945076942443848, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7694227695465088, + "num_tokens": 862937954.0, + "step": 32366 + }, + { + "epoch": 2.669003051043127, + "grad_norm": 2.2287328243255615, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7766959071159363, + "num_tokens": 862963428.0, + "step": 32367 + }, + { + "epoch": 2.6690855116681784, + "grad_norm": 1.9235159158706665, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7678787708282471, + "num_tokens": 862997534.0, + "step": 32368 + }, + { + "epoch": 2.6691679722932298, + "grad_norm": 2.0696206092834473, + "learning_rate": 1e-06, + "loss": 0.559, + "mean_token_accuracy": 0.8094576597213745, + "num_tokens": 863023940.0, + "step": 32369 + }, + { + "epoch": 2.6692504329182816, + "grad_norm": 2.348932981491089, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7691941261291504, + "num_tokens": 863049630.0, + "step": 32370 + }, + { + "epoch": 2.669332893543333, + "grad_norm": 2.1875181198120117, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7584396600723267, + "num_tokens": 863077694.0, + "step": 32371 + }, + { + "epoch": 2.6694153541683847, + "grad_norm": 2.0631320476531982, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7669167518615723, + "num_tokens": 863108034.0, + "step": 32372 + }, + { + "epoch": 2.669497814793436, + "grad_norm": 2.0512964725494385, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.7900471687316895, + "num_tokens": 863138856.0, + "step": 32373 + }, + { + "epoch": 2.669580275418488, + "grad_norm": 2.1270511150360107, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7517555356025696, + "num_tokens": 863170559.0, + "step": 32374 + }, + { + "epoch": 2.669662736043539, + "grad_norm": 2.216257333755493, + "learning_rate": 1e-06, + "loss": 0.6258, + "mean_token_accuracy": 0.7958452701568604, + "num_tokens": 863197231.0, + "step": 32375 + }, + { + "epoch": 2.6697451966685906, + "grad_norm": 1.903455376625061, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7759156227111816, + "num_tokens": 863231851.0, + "step": 32376 + }, + { + "epoch": 2.6698276572936424, + "grad_norm": 1.8754056692123413, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7698245644569397, + "num_tokens": 863268342.0, + "step": 32377 + }, + { + "epoch": 2.6699101179186937, + "grad_norm": 1.9605813026428223, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7792829275131226, + "num_tokens": 863299996.0, + "step": 32378 + }, + { + "epoch": 2.6699925785437455, + "grad_norm": 1.8587454557418823, + "learning_rate": 1e-06, + "loss": 0.5906, + "mean_token_accuracy": 0.8006430864334106, + "num_tokens": 863336642.0, + "step": 32379 + }, + { + "epoch": 2.670075039168797, + "grad_norm": 2.539942741394043, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7641986012458801, + "num_tokens": 863357476.0, + "step": 32380 + }, + { + "epoch": 2.6701574997938486, + "grad_norm": 2.4638712406158447, + "learning_rate": 1e-06, + "loss": 0.6988, + "mean_token_accuracy": 0.7769556045532227, + "num_tokens": 863380065.0, + "step": 32381 + }, + { + "epoch": 2.6702399604189, + "grad_norm": 1.865856409072876, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.7929568886756897, + "num_tokens": 863416639.0, + "step": 32382 + }, + { + "epoch": 2.6703224210439513, + "grad_norm": 1.9573936462402344, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7569462060928345, + "num_tokens": 863452067.0, + "step": 32383 + }, + { + "epoch": 2.670404881669003, + "grad_norm": 2.2473301887512207, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.754487931728363, + "num_tokens": 863478278.0, + "step": 32384 + }, + { + "epoch": 2.6704873422940545, + "grad_norm": 2.3971896171569824, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7423551082611084, + "num_tokens": 863502362.0, + "step": 32385 + }, + { + "epoch": 2.6705698029191063, + "grad_norm": 2.1666581630706787, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7737205028533936, + "num_tokens": 863530893.0, + "step": 32386 + }, + { + "epoch": 2.6706522635441576, + "grad_norm": 1.9657918214797974, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7718988656997681, + "num_tokens": 863562151.0, + "step": 32387 + }, + { + "epoch": 2.6707347241692094, + "grad_norm": 2.3298468589782715, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.786074697971344, + "num_tokens": 863586428.0, + "step": 32388 + }, + { + "epoch": 2.670817184794261, + "grad_norm": 2.124901294708252, + "learning_rate": 1e-06, + "loss": 0.6059, + "mean_token_accuracy": 0.7974016666412354, + "num_tokens": 863616490.0, + "step": 32389 + }, + { + "epoch": 2.670899645419312, + "grad_norm": 2.4524807929992676, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.7768813371658325, + "num_tokens": 863639028.0, + "step": 32390 + }, + { + "epoch": 2.670982106044364, + "grad_norm": 2.6726367473602295, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7436556816101074, + "num_tokens": 863662455.0, + "step": 32391 + }, + { + "epoch": 2.6710645666694153, + "grad_norm": 2.4394211769104004, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7521072626113892, + "num_tokens": 863691509.0, + "step": 32392 + }, + { + "epoch": 2.6711470272944666, + "grad_norm": 2.0627431869506836, + "learning_rate": 1e-06, + "loss": 0.5968, + "mean_token_accuracy": 0.8038641214370728, + "num_tokens": 863721084.0, + "step": 32393 + }, + { + "epoch": 2.6712294879195184, + "grad_norm": 1.823487639427185, + "learning_rate": 1e-06, + "loss": 0.8001, + "mean_token_accuracy": 0.7446750402450562, + "num_tokens": 863760924.0, + "step": 32394 + }, + { + "epoch": 2.67131194854457, + "grad_norm": 2.4484145641326904, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7535645961761475, + "num_tokens": 863785096.0, + "step": 32395 + }, + { + "epoch": 2.6713944091696216, + "grad_norm": 2.204796552658081, + "learning_rate": 1e-06, + "loss": 0.7835, + "mean_token_accuracy": 0.7515445351600647, + "num_tokens": 863813463.0, + "step": 32396 + }, + { + "epoch": 2.671476869794673, + "grad_norm": 1.9601136445999146, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.7854417562484741, + "num_tokens": 863844485.0, + "step": 32397 + }, + { + "epoch": 2.6715593304197247, + "grad_norm": 2.014052391052246, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7583938837051392, + "num_tokens": 863878093.0, + "step": 32398 + }, + { + "epoch": 2.671641791044776, + "grad_norm": 2.1082098484039307, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7611083984375, + "num_tokens": 863906154.0, + "step": 32399 + }, + { + "epoch": 2.6717242516698274, + "grad_norm": 2.2308430671691895, + "learning_rate": 1e-06, + "loss": 0.5946, + "mean_token_accuracy": 0.8059214353561401, + "num_tokens": 863936117.0, + "step": 32400 + }, + { + "epoch": 2.671806712294879, + "grad_norm": 2.309668779373169, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7160879373550415, + "num_tokens": 863965760.0, + "step": 32401 + }, + { + "epoch": 2.671889172919931, + "grad_norm": 2.295653820037842, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7536685466766357, + "num_tokens": 863992444.0, + "step": 32402 + }, + { + "epoch": 2.6719716335449823, + "grad_norm": 1.97926664352417, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.7863972187042236, + "num_tokens": 864023790.0, + "step": 32403 + }, + { + "epoch": 2.6720540941700337, + "grad_norm": 2.280064344406128, + "learning_rate": 1e-06, + "loss": 0.6846, + "mean_token_accuracy": 0.7674673199653625, + "num_tokens": 864048644.0, + "step": 32404 + }, + { + "epoch": 2.6721365547950855, + "grad_norm": 2.455770492553711, + "learning_rate": 1e-06, + "loss": 0.7449, + "mean_token_accuracy": 0.7520186901092529, + "num_tokens": 864070223.0, + "step": 32405 + }, + { + "epoch": 2.672219015420137, + "grad_norm": 1.9855172634124756, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7840458154678345, + "num_tokens": 864103546.0, + "step": 32406 + }, + { + "epoch": 2.672301476045188, + "grad_norm": 2.016655683517456, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7811992168426514, + "num_tokens": 864135881.0, + "step": 32407 + }, + { + "epoch": 2.67238393667024, + "grad_norm": 2.136422634124756, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.757286548614502, + "num_tokens": 864166094.0, + "step": 32408 + }, + { + "epoch": 2.6724663972952913, + "grad_norm": 2.05509352684021, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.7842414379119873, + "num_tokens": 864197454.0, + "step": 32409 + }, + { + "epoch": 2.672548857920343, + "grad_norm": 1.8905150890350342, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.7895145416259766, + "num_tokens": 864233581.0, + "step": 32410 + }, + { + "epoch": 2.6726313185453945, + "grad_norm": 2.1044137477874756, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7776763439178467, + "num_tokens": 864262377.0, + "step": 32411 + }, + { + "epoch": 2.6727137791704463, + "grad_norm": 1.9475610256195068, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.7960439324378967, + "num_tokens": 864293034.0, + "step": 32412 + }, + { + "epoch": 2.6727962397954976, + "grad_norm": 2.2136144638061523, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.7992633581161499, + "num_tokens": 864319550.0, + "step": 32413 + }, + { + "epoch": 2.672878700420549, + "grad_norm": 1.9962064027786255, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.7988513708114624, + "num_tokens": 864353303.0, + "step": 32414 + }, + { + "epoch": 2.6729611610456008, + "grad_norm": 2.028125524520874, + "learning_rate": 1e-06, + "loss": 0.765, + "mean_token_accuracy": 0.7479574084281921, + "num_tokens": 864386688.0, + "step": 32415 + }, + { + "epoch": 2.673043621670652, + "grad_norm": 2.3856663703918457, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.7847176194190979, + "num_tokens": 864410660.0, + "step": 32416 + }, + { + "epoch": 2.673126082295704, + "grad_norm": 1.7947261333465576, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7673760652542114, + "num_tokens": 864450706.0, + "step": 32417 + }, + { + "epoch": 2.6732085429207553, + "grad_norm": 2.2479970455169678, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7706261873245239, + "num_tokens": 864476964.0, + "step": 32418 + }, + { + "epoch": 2.673291003545807, + "grad_norm": 2.2868995666503906, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7051715850830078, + "num_tokens": 864507149.0, + "step": 32419 + }, + { + "epoch": 2.6733734641708584, + "grad_norm": 2.109799385070801, + "learning_rate": 1e-06, + "loss": 0.6025, + "mean_token_accuracy": 0.7943047285079956, + "num_tokens": 864536216.0, + "step": 32420 + }, + { + "epoch": 2.6734559247959098, + "grad_norm": 2.186549186706543, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7492175102233887, + "num_tokens": 864565951.0, + "step": 32421 + }, + { + "epoch": 2.6735383854209616, + "grad_norm": 1.9921749830245972, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.7990478277206421, + "num_tokens": 864598518.0, + "step": 32422 + }, + { + "epoch": 2.673620846046013, + "grad_norm": 2.294682264328003, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7754499912261963, + "num_tokens": 864626084.0, + "step": 32423 + }, + { + "epoch": 2.6737033066710647, + "grad_norm": 2.228971004486084, + "learning_rate": 1e-06, + "loss": 0.5061, + "mean_token_accuracy": 0.8249022960662842, + "num_tokens": 864651290.0, + "step": 32424 + }, + { + "epoch": 2.673785767296116, + "grad_norm": 1.918533444404602, + "learning_rate": 1e-06, + "loss": 0.5426, + "mean_token_accuracy": 0.8235235810279846, + "num_tokens": 864686011.0, + "step": 32425 + }, + { + "epoch": 2.673868227921168, + "grad_norm": 2.576817512512207, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.7826647758483887, + "num_tokens": 864710358.0, + "step": 32426 + }, + { + "epoch": 2.673950688546219, + "grad_norm": 2.1193220615386963, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.7824122905731201, + "num_tokens": 864741888.0, + "step": 32427 + }, + { + "epoch": 2.6740331491712706, + "grad_norm": 2.5283823013305664, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7615649700164795, + "num_tokens": 864764132.0, + "step": 32428 + }, + { + "epoch": 2.6741156097963223, + "grad_norm": 1.9308288097381592, + "learning_rate": 1e-06, + "loss": 0.6504, + "mean_token_accuracy": 0.7856251001358032, + "num_tokens": 864800004.0, + "step": 32429 + }, + { + "epoch": 2.6741980704213737, + "grad_norm": 1.8648055791854858, + "learning_rate": 1e-06, + "loss": 0.5928, + "mean_token_accuracy": 0.8049973249435425, + "num_tokens": 864834027.0, + "step": 32430 + }, + { + "epoch": 2.6742805310464255, + "grad_norm": 2.5593984127044678, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.738128662109375, + "num_tokens": 864857637.0, + "step": 32431 + }, + { + "epoch": 2.674362991671477, + "grad_norm": 1.959352970123291, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.7941739559173584, + "num_tokens": 864892123.0, + "step": 32432 + }, + { + "epoch": 2.6744454522965286, + "grad_norm": 2.050069570541382, + "learning_rate": 1e-06, + "loss": 0.6083, + "mean_token_accuracy": 0.7942466735839844, + "num_tokens": 864925219.0, + "step": 32433 + }, + { + "epoch": 2.67452791292158, + "grad_norm": 1.9743396043777466, + "learning_rate": 1e-06, + "loss": 0.5747, + "mean_token_accuracy": 0.8025844097137451, + "num_tokens": 864957881.0, + "step": 32434 + }, + { + "epoch": 2.6746103735466313, + "grad_norm": 2.1141602993011475, + "learning_rate": 1e-06, + "loss": 0.5701, + "mean_token_accuracy": 0.8006060123443604, + "num_tokens": 864987313.0, + "step": 32435 + }, + { + "epoch": 2.674692834171683, + "grad_norm": 2.120699405670166, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7720443606376648, + "num_tokens": 865017181.0, + "step": 32436 + }, + { + "epoch": 2.6747752947967345, + "grad_norm": 2.0539166927337646, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.792596697807312, + "num_tokens": 865045563.0, + "step": 32437 + }, + { + "epoch": 2.674857755421786, + "grad_norm": 2.0590555667877197, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7279095649719238, + "num_tokens": 865078209.0, + "step": 32438 + }, + { + "epoch": 2.6749402160468376, + "grad_norm": 2.346834182739258, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7644227743148804, + "num_tokens": 865103624.0, + "step": 32439 + }, + { + "epoch": 2.6750226766718894, + "grad_norm": 1.854494571685791, + "learning_rate": 1e-06, + "loss": 0.5425, + "mean_token_accuracy": 0.8192107677459717, + "num_tokens": 865137585.0, + "step": 32440 + }, + { + "epoch": 2.6751051372969408, + "grad_norm": 2.0667026042938232, + "learning_rate": 1e-06, + "loss": 0.7119, + "mean_token_accuracy": 0.7673854231834412, + "num_tokens": 865167526.0, + "step": 32441 + }, + { + "epoch": 2.675187597921992, + "grad_norm": 1.9897619485855103, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.7807997465133667, + "num_tokens": 865196471.0, + "step": 32442 + }, + { + "epoch": 2.675270058547044, + "grad_norm": 2.0955328941345215, + "learning_rate": 1e-06, + "loss": 0.5834, + "mean_token_accuracy": 0.800846517086029, + "num_tokens": 865227604.0, + "step": 32443 + }, + { + "epoch": 2.6753525191720953, + "grad_norm": 2.0579776763916016, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7755203247070312, + "num_tokens": 865257905.0, + "step": 32444 + }, + { + "epoch": 2.6754349797971466, + "grad_norm": 2.1091113090515137, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.778092086315155, + "num_tokens": 865285030.0, + "step": 32445 + }, + { + "epoch": 2.6755174404221984, + "grad_norm": 2.2062695026397705, + "learning_rate": 1e-06, + "loss": 0.5651, + "mean_token_accuracy": 0.8074543476104736, + "num_tokens": 865310441.0, + "step": 32446 + }, + { + "epoch": 2.67559990104725, + "grad_norm": 2.204421043395996, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.7947329878807068, + "num_tokens": 865334708.0, + "step": 32447 + }, + { + "epoch": 2.6756823616723016, + "grad_norm": 2.2210118770599365, + "learning_rate": 1e-06, + "loss": 0.7339, + "mean_token_accuracy": 0.7546713352203369, + "num_tokens": 865361985.0, + "step": 32448 + }, + { + "epoch": 2.675764822297353, + "grad_norm": 2.3022303581237793, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7762390375137329, + "num_tokens": 865388297.0, + "step": 32449 + }, + { + "epoch": 2.6758472829224047, + "grad_norm": 1.9099115133285522, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.776637613773346, + "num_tokens": 865420228.0, + "step": 32450 + }, + { + "epoch": 2.675929743547456, + "grad_norm": 2.319775104522705, + "learning_rate": 1e-06, + "loss": 0.6101, + "mean_token_accuracy": 0.8025544881820679, + "num_tokens": 865446070.0, + "step": 32451 + }, + { + "epoch": 2.6760122041725074, + "grad_norm": 1.788635015487671, + "learning_rate": 1e-06, + "loss": 0.5581, + "mean_token_accuracy": 0.8120144605636597, + "num_tokens": 865484646.0, + "step": 32452 + }, + { + "epoch": 2.676094664797559, + "grad_norm": 2.1098179817199707, + "learning_rate": 1e-06, + "loss": 0.7172, + "mean_token_accuracy": 0.7674127817153931, + "num_tokens": 865513651.0, + "step": 32453 + }, + { + "epoch": 2.6761771254226105, + "grad_norm": 1.8898245096206665, + "learning_rate": 1e-06, + "loss": 0.5372, + "mean_token_accuracy": 0.819727897644043, + "num_tokens": 865549118.0, + "step": 32454 + }, + { + "epoch": 2.6762595860476623, + "grad_norm": 2.1525511741638184, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7682876586914062, + "num_tokens": 865578441.0, + "step": 32455 + }, + { + "epoch": 2.6763420466727137, + "grad_norm": 2.3654122352600098, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7691195011138916, + "num_tokens": 865602813.0, + "step": 32456 + }, + { + "epoch": 2.6764245072977655, + "grad_norm": 2.4371302127838135, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7354354858398438, + "num_tokens": 865626470.0, + "step": 32457 + }, + { + "epoch": 2.676506967922817, + "grad_norm": 2.158560037612915, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7463460564613342, + "num_tokens": 865654305.0, + "step": 32458 + }, + { + "epoch": 2.676589428547868, + "grad_norm": 1.9819912910461426, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7334467172622681, + "num_tokens": 865688915.0, + "step": 32459 + }, + { + "epoch": 2.67667188917292, + "grad_norm": 2.117648124694824, + "learning_rate": 1e-06, + "loss": 0.5652, + "mean_token_accuracy": 0.8004406690597534, + "num_tokens": 865717317.0, + "step": 32460 + }, + { + "epoch": 2.6767543497979713, + "grad_norm": 2.1611433029174805, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7506166696548462, + "num_tokens": 865746637.0, + "step": 32461 + }, + { + "epoch": 2.676836810423023, + "grad_norm": 1.923986554145813, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7717984914779663, + "num_tokens": 865782205.0, + "step": 32462 + }, + { + "epoch": 2.6769192710480745, + "grad_norm": 2.1637330055236816, + "learning_rate": 1e-06, + "loss": 0.7915, + "mean_token_accuracy": 0.745922327041626, + "num_tokens": 865810446.0, + "step": 32463 + }, + { + "epoch": 2.6770017316731263, + "grad_norm": 1.9825010299682617, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7783241868019104, + "num_tokens": 865841626.0, + "step": 32464 + }, + { + "epoch": 2.6770841922981776, + "grad_norm": 2.4688007831573486, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7643648982048035, + "num_tokens": 865866492.0, + "step": 32465 + }, + { + "epoch": 2.677166652923229, + "grad_norm": 2.2448933124542236, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7651419639587402, + "num_tokens": 865893099.0, + "step": 32466 + }, + { + "epoch": 2.6772491135482808, + "grad_norm": 2.442318916320801, + "learning_rate": 1e-06, + "loss": 0.7171, + "mean_token_accuracy": 0.7673335075378418, + "num_tokens": 865916979.0, + "step": 32467 + }, + { + "epoch": 2.677331574173332, + "grad_norm": 2.0347883701324463, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.7583887577056885, + "num_tokens": 865949365.0, + "step": 32468 + }, + { + "epoch": 2.677414034798384, + "grad_norm": 2.119133234024048, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7773622870445251, + "num_tokens": 865976140.0, + "step": 32469 + }, + { + "epoch": 2.6774964954234353, + "grad_norm": 2.1705379486083984, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7776572704315186, + "num_tokens": 866005697.0, + "step": 32470 + }, + { + "epoch": 2.677578956048487, + "grad_norm": 2.245020627975464, + "learning_rate": 1e-06, + "loss": 0.6417, + "mean_token_accuracy": 0.784000813961029, + "num_tokens": 866033031.0, + "step": 32471 + }, + { + "epoch": 2.6776614166735384, + "grad_norm": 2.1053411960601807, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7556286454200745, + "num_tokens": 866063504.0, + "step": 32472 + }, + { + "epoch": 2.6777438772985898, + "grad_norm": 2.1868557929992676, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7525908946990967, + "num_tokens": 866093704.0, + "step": 32473 + }, + { + "epoch": 2.6778263379236416, + "grad_norm": 2.0526959896087646, + "learning_rate": 1e-06, + "loss": 0.5947, + "mean_token_accuracy": 0.806243896484375, + "num_tokens": 866123150.0, + "step": 32474 + }, + { + "epoch": 2.677908798548693, + "grad_norm": 1.9848219156265259, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7795191407203674, + "num_tokens": 866156458.0, + "step": 32475 + }, + { + "epoch": 2.6779912591737447, + "grad_norm": 2.247582197189331, + "learning_rate": 1e-06, + "loss": 0.6207, + "mean_token_accuracy": 0.78398597240448, + "num_tokens": 866182412.0, + "step": 32476 + }, + { + "epoch": 2.678073719798796, + "grad_norm": 1.883519172668457, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7739375233650208, + "num_tokens": 866219251.0, + "step": 32477 + }, + { + "epoch": 2.678156180423848, + "grad_norm": 2.272686719894409, + "learning_rate": 1e-06, + "loss": 0.7897, + "mean_token_accuracy": 0.7483423948287964, + "num_tokens": 866248024.0, + "step": 32478 + }, + { + "epoch": 2.678238641048899, + "grad_norm": 1.9635300636291504, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7869978547096252, + "num_tokens": 866279208.0, + "step": 32479 + }, + { + "epoch": 2.6783211016739505, + "grad_norm": 2.1772522926330566, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.7916698455810547, + "num_tokens": 866305481.0, + "step": 32480 + }, + { + "epoch": 2.6784035622990023, + "grad_norm": 2.0257582664489746, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7591638565063477, + "num_tokens": 866337852.0, + "step": 32481 + }, + { + "epoch": 2.6784860229240537, + "grad_norm": 2.682840347290039, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7627567052841187, + "num_tokens": 866356793.0, + "step": 32482 + }, + { + "epoch": 2.678568483549105, + "grad_norm": 2.28291654586792, + "learning_rate": 1e-06, + "loss": 0.6567, + "mean_token_accuracy": 0.7826738357543945, + "num_tokens": 866384874.0, + "step": 32483 + }, + { + "epoch": 2.678650944174157, + "grad_norm": 1.8988943099975586, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7633242607116699, + "num_tokens": 866419798.0, + "step": 32484 + }, + { + "epoch": 2.6787334047992086, + "grad_norm": 2.3154802322387695, + "learning_rate": 1e-06, + "loss": 0.6732, + "mean_token_accuracy": 0.7745382189750671, + "num_tokens": 866447286.0, + "step": 32485 + }, + { + "epoch": 2.67881586542426, + "grad_norm": 1.9384965896606445, + "learning_rate": 1e-06, + "loss": 0.6084, + "mean_token_accuracy": 0.8030277490615845, + "num_tokens": 866480381.0, + "step": 32486 + }, + { + "epoch": 2.6788983260493113, + "grad_norm": 1.6860294342041016, + "learning_rate": 1e-06, + "loss": 0.5503, + "mean_token_accuracy": 0.8165250420570374, + "num_tokens": 866518857.0, + "step": 32487 + }, + { + "epoch": 2.678980786674363, + "grad_norm": 2.038861036300659, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7719537019729614, + "num_tokens": 866550236.0, + "step": 32488 + }, + { + "epoch": 2.6790632472994145, + "grad_norm": 2.116503953933716, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7708468437194824, + "num_tokens": 866581722.0, + "step": 32489 + }, + { + "epoch": 2.679145707924466, + "grad_norm": 2.1624629497528076, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7728681564331055, + "num_tokens": 866611787.0, + "step": 32490 + }, + { + "epoch": 2.6792281685495176, + "grad_norm": 2.2663381099700928, + "learning_rate": 1e-06, + "loss": 0.8099, + "mean_token_accuracy": 0.7384206056594849, + "num_tokens": 866641548.0, + "step": 32491 + }, + { + "epoch": 2.6793106291745694, + "grad_norm": 2.0304768085479736, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7651654481887817, + "num_tokens": 866672542.0, + "step": 32492 + }, + { + "epoch": 2.6793930897996208, + "grad_norm": 2.1759448051452637, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7795288562774658, + "num_tokens": 866704036.0, + "step": 32493 + }, + { + "epoch": 2.679475550424672, + "grad_norm": 2.2039566040039062, + "learning_rate": 1e-06, + "loss": 0.579, + "mean_token_accuracy": 0.8072491884231567, + "num_tokens": 866733248.0, + "step": 32494 + }, + { + "epoch": 2.679558011049724, + "grad_norm": 2.149977207183838, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7547645568847656, + "num_tokens": 866762754.0, + "step": 32495 + }, + { + "epoch": 2.6796404716747753, + "grad_norm": 2.0452189445495605, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7546178102493286, + "num_tokens": 866793470.0, + "step": 32496 + }, + { + "epoch": 2.6797229322998266, + "grad_norm": 2.2263402938842773, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7683939337730408, + "num_tokens": 866822401.0, + "step": 32497 + }, + { + "epoch": 2.6798053929248784, + "grad_norm": 2.332150936126709, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7546905279159546, + "num_tokens": 866847333.0, + "step": 32498 + }, + { + "epoch": 2.6798878535499298, + "grad_norm": 2.067204713821411, + "learning_rate": 1e-06, + "loss": 0.5781, + "mean_token_accuracy": 0.7957864999771118, + "num_tokens": 866876699.0, + "step": 32499 + }, + { + "epoch": 2.6799703141749815, + "grad_norm": 1.917770504951477, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7503368854522705, + "num_tokens": 866912877.0, + "step": 32500 + }, + { + "epoch": 2.680052774800033, + "grad_norm": 1.8157073259353638, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7616943717002869, + "num_tokens": 866949150.0, + "step": 32501 + }, + { + "epoch": 2.6801352354250847, + "grad_norm": 2.323007583618164, + "learning_rate": 1e-06, + "loss": 0.7798, + "mean_token_accuracy": 0.7449614405632019, + "num_tokens": 866973070.0, + "step": 32502 + }, + { + "epoch": 2.680217696050136, + "grad_norm": 2.096139907836914, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7674544453620911, + "num_tokens": 867003693.0, + "step": 32503 + }, + { + "epoch": 2.6803001566751874, + "grad_norm": 1.8935861587524414, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.765643835067749, + "num_tokens": 867038285.0, + "step": 32504 + }, + { + "epoch": 2.680382617300239, + "grad_norm": 2.2227630615234375, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.776272177696228, + "num_tokens": 867064884.0, + "step": 32505 + }, + { + "epoch": 2.6804650779252905, + "grad_norm": 1.7960352897644043, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7801825404167175, + "num_tokens": 867102066.0, + "step": 32506 + }, + { + "epoch": 2.6805475385503423, + "grad_norm": 1.8690346479415894, + "learning_rate": 1e-06, + "loss": 0.5821, + "mean_token_accuracy": 0.7939965724945068, + "num_tokens": 867135585.0, + "step": 32507 + }, + { + "epoch": 2.6806299991753937, + "grad_norm": 2.1222853660583496, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7640886306762695, + "num_tokens": 867163651.0, + "step": 32508 + }, + { + "epoch": 2.6807124598004455, + "grad_norm": 2.2973763942718506, + "learning_rate": 1e-06, + "loss": 0.7658, + "mean_token_accuracy": 0.7593855857849121, + "num_tokens": 867191256.0, + "step": 32509 + }, + { + "epoch": 2.680794920425497, + "grad_norm": 1.8986262083053589, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7726590633392334, + "num_tokens": 867224589.0, + "step": 32510 + }, + { + "epoch": 2.680877381050548, + "grad_norm": 2.1306052207946777, + "learning_rate": 1e-06, + "loss": 0.7557, + "mean_token_accuracy": 0.7527801990509033, + "num_tokens": 867254532.0, + "step": 32511 + }, + { + "epoch": 2.6809598416756, + "grad_norm": 2.3487234115600586, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7505977153778076, + "num_tokens": 867281964.0, + "step": 32512 + }, + { + "epoch": 2.6810423023006513, + "grad_norm": 2.201564311981201, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7727981805801392, + "num_tokens": 867310910.0, + "step": 32513 + }, + { + "epoch": 2.681124762925703, + "grad_norm": 2.5042166709899902, + "learning_rate": 1e-06, + "loss": 0.7886, + "mean_token_accuracy": 0.7498767375946045, + "num_tokens": 867336217.0, + "step": 32514 + }, + { + "epoch": 2.6812072235507545, + "grad_norm": 2.388960361480713, + "learning_rate": 1e-06, + "loss": 0.6714, + "mean_token_accuracy": 0.7766939997673035, + "num_tokens": 867361745.0, + "step": 32515 + }, + { + "epoch": 2.6812896841758063, + "grad_norm": 2.1075448989868164, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.7724295258522034, + "num_tokens": 867390207.0, + "step": 32516 + }, + { + "epoch": 2.6813721448008576, + "grad_norm": 2.194209575653076, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.7826022505760193, + "num_tokens": 867418648.0, + "step": 32517 + }, + { + "epoch": 2.681454605425909, + "grad_norm": 2.1552562713623047, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7436323165893555, + "num_tokens": 867448776.0, + "step": 32518 + }, + { + "epoch": 2.6815370660509608, + "grad_norm": 2.2257394790649414, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.7965806126594543, + "num_tokens": 867475739.0, + "step": 32519 + }, + { + "epoch": 2.681619526676012, + "grad_norm": 2.311011791229248, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7620983719825745, + "num_tokens": 867504801.0, + "step": 32520 + }, + { + "epoch": 2.681701987301064, + "grad_norm": 1.842140793800354, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7864190340042114, + "num_tokens": 867540941.0, + "step": 32521 + }, + { + "epoch": 2.6817844479261153, + "grad_norm": 1.910048007965088, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.7919466495513916, + "num_tokens": 867574304.0, + "step": 32522 + }, + { + "epoch": 2.681866908551167, + "grad_norm": 2.0436975955963135, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.7934801578521729, + "num_tokens": 867608991.0, + "step": 32523 + }, + { + "epoch": 2.6819493691762184, + "grad_norm": 2.07843279838562, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7564912438392639, + "num_tokens": 867640847.0, + "step": 32524 + }, + { + "epoch": 2.6820318298012698, + "grad_norm": 2.2726879119873047, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7706315517425537, + "num_tokens": 867666579.0, + "step": 32525 + }, + { + "epoch": 2.6821142904263215, + "grad_norm": 1.9841549396514893, + "learning_rate": 1e-06, + "loss": 0.5471, + "mean_token_accuracy": 0.8125907182693481, + "num_tokens": 867697064.0, + "step": 32526 + }, + { + "epoch": 2.682196751051373, + "grad_norm": 2.036567449569702, + "learning_rate": 1e-06, + "loss": 0.7598, + "mean_token_accuracy": 0.7514203786849976, + "num_tokens": 867726836.0, + "step": 32527 + }, + { + "epoch": 2.6822792116764242, + "grad_norm": 2.1643974781036377, + "learning_rate": 1e-06, + "loss": 0.6653, + "mean_token_accuracy": 0.7808576822280884, + "num_tokens": 867755932.0, + "step": 32528 + }, + { + "epoch": 2.682361672301476, + "grad_norm": 2.1843857765197754, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7739349603652954, + "num_tokens": 867780931.0, + "step": 32529 + }, + { + "epoch": 2.682444132926528, + "grad_norm": 2.501760721206665, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7583968639373779, + "num_tokens": 867806498.0, + "step": 32530 + }, + { + "epoch": 2.682526593551579, + "grad_norm": 2.481037139892578, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.7932928800582886, + "num_tokens": 867829943.0, + "step": 32531 + }, + { + "epoch": 2.6826090541766305, + "grad_norm": 2.505051851272583, + "learning_rate": 1e-06, + "loss": 0.7447, + "mean_token_accuracy": 0.757693886756897, + "num_tokens": 867854829.0, + "step": 32532 + }, + { + "epoch": 2.6826915148016823, + "grad_norm": 2.502302408218384, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7458891868591309, + "num_tokens": 867877783.0, + "step": 32533 + }, + { + "epoch": 2.6827739754267337, + "grad_norm": 2.1188132762908936, + "learning_rate": 1e-06, + "loss": 0.5129, + "mean_token_accuracy": 0.8212953805923462, + "num_tokens": 867906554.0, + "step": 32534 + }, + { + "epoch": 2.682856436051785, + "grad_norm": 2.158540725708008, + "learning_rate": 1e-06, + "loss": 0.5726, + "mean_token_accuracy": 0.8101862668991089, + "num_tokens": 867933749.0, + "step": 32535 + }, + { + "epoch": 2.682938896676837, + "grad_norm": 2.030425548553467, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.7753624320030212, + "num_tokens": 867966284.0, + "step": 32536 + }, + { + "epoch": 2.6830213573018886, + "grad_norm": 2.1166341304779053, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7594130039215088, + "num_tokens": 867998821.0, + "step": 32537 + }, + { + "epoch": 2.68310381792694, + "grad_norm": 1.9687144756317139, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7426934242248535, + "num_tokens": 868033368.0, + "step": 32538 + }, + { + "epoch": 2.6831862785519913, + "grad_norm": 2.0526340007781982, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7680178284645081, + "num_tokens": 868063090.0, + "step": 32539 + }, + { + "epoch": 2.683268739177043, + "grad_norm": 2.133002758026123, + "learning_rate": 1e-06, + "loss": 0.5768, + "mean_token_accuracy": 0.8076654076576233, + "num_tokens": 868091610.0, + "step": 32540 + }, + { + "epoch": 2.6833511998020945, + "grad_norm": 2.1218178272247314, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.763458788394928, + "num_tokens": 868119409.0, + "step": 32541 + }, + { + "epoch": 2.683433660427146, + "grad_norm": 2.1770241260528564, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.7874535322189331, + "num_tokens": 868146189.0, + "step": 32542 + }, + { + "epoch": 2.6835161210521976, + "grad_norm": 2.199225902557373, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7767128944396973, + "num_tokens": 868174174.0, + "step": 32543 + }, + { + "epoch": 2.683598581677249, + "grad_norm": 2.159324884414673, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7609571218490601, + "num_tokens": 868204354.0, + "step": 32544 + }, + { + "epoch": 2.6836810423023008, + "grad_norm": 2.2071950435638428, + "learning_rate": 1e-06, + "loss": 0.7154, + "mean_token_accuracy": 0.7650757431983948, + "num_tokens": 868235862.0, + "step": 32545 + }, + { + "epoch": 2.683763502927352, + "grad_norm": 1.7516839504241943, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.7909162640571594, + "num_tokens": 868275394.0, + "step": 32546 + }, + { + "epoch": 2.683845963552404, + "grad_norm": 2.070864200592041, + "learning_rate": 1e-06, + "loss": 0.7398, + "mean_token_accuracy": 0.7552492618560791, + "num_tokens": 868306333.0, + "step": 32547 + }, + { + "epoch": 2.6839284241774553, + "grad_norm": 2.106945753097534, + "learning_rate": 1e-06, + "loss": 0.5599, + "mean_token_accuracy": 0.8156086206436157, + "num_tokens": 868334844.0, + "step": 32548 + }, + { + "epoch": 2.6840108848025066, + "grad_norm": 1.893287181854248, + "learning_rate": 1e-06, + "loss": 0.7574, + "mean_token_accuracy": 0.7615781426429749, + "num_tokens": 868373226.0, + "step": 32549 + }, + { + "epoch": 2.6840933454275584, + "grad_norm": 2.1232094764709473, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.7713831663131714, + "num_tokens": 868404045.0, + "step": 32550 + }, + { + "epoch": 2.6841758060526097, + "grad_norm": 2.259046792984009, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.7806257009506226, + "num_tokens": 868429875.0, + "step": 32551 + }, + { + "epoch": 2.6842582666776615, + "grad_norm": 2.061654567718506, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.771294116973877, + "num_tokens": 868462294.0, + "step": 32552 + }, + { + "epoch": 2.684340727302713, + "grad_norm": 1.8358341455459595, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7566496133804321, + "num_tokens": 868501063.0, + "step": 32553 + }, + { + "epoch": 2.6844231879277647, + "grad_norm": 2.3889849185943604, + "learning_rate": 1e-06, + "loss": 0.5846, + "mean_token_accuracy": 0.7954860925674438, + "num_tokens": 868526550.0, + "step": 32554 + }, + { + "epoch": 2.684505648552816, + "grad_norm": 2.5488948822021484, + "learning_rate": 1e-06, + "loss": 0.7808, + "mean_token_accuracy": 0.7557849884033203, + "num_tokens": 868549393.0, + "step": 32555 + }, + { + "epoch": 2.6845881091778674, + "grad_norm": 2.0827200412750244, + "learning_rate": 1e-06, + "loss": 0.5232, + "mean_token_accuracy": 0.8263302445411682, + "num_tokens": 868577177.0, + "step": 32556 + }, + { + "epoch": 2.684670569802919, + "grad_norm": 2.0119309425354004, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7422937154769897, + "num_tokens": 868612683.0, + "step": 32557 + }, + { + "epoch": 2.6847530304279705, + "grad_norm": 2.157552719116211, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7889825701713562, + "num_tokens": 868641658.0, + "step": 32558 + }, + { + "epoch": 2.6848354910530223, + "grad_norm": 1.8774372339248657, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7552368640899658, + "num_tokens": 868678953.0, + "step": 32559 + }, + { + "epoch": 2.6849179516780737, + "grad_norm": 1.9428328275680542, + "learning_rate": 1e-06, + "loss": 0.517, + "mean_token_accuracy": 0.8275578022003174, + "num_tokens": 868711617.0, + "step": 32560 + }, + { + "epoch": 2.6850004123031255, + "grad_norm": 1.882511854171753, + "learning_rate": 1e-06, + "loss": 0.6777, + "mean_token_accuracy": 0.7807148098945618, + "num_tokens": 868746712.0, + "step": 32561 + }, + { + "epoch": 2.685082872928177, + "grad_norm": 2.4626996517181396, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7680096626281738, + "num_tokens": 868768995.0, + "step": 32562 + }, + { + "epoch": 2.685165333553228, + "grad_norm": 2.0402204990386963, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7682235240936279, + "num_tokens": 868800394.0, + "step": 32563 + }, + { + "epoch": 2.68524779417828, + "grad_norm": 1.816352367401123, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7822104692459106, + "num_tokens": 868836957.0, + "step": 32564 + }, + { + "epoch": 2.6853302548033313, + "grad_norm": 1.7457393407821655, + "learning_rate": 1e-06, + "loss": 0.5703, + "mean_token_accuracy": 0.81159907579422, + "num_tokens": 868875722.0, + "step": 32565 + }, + { + "epoch": 2.685412715428383, + "grad_norm": 2.228436231613159, + "learning_rate": 1e-06, + "loss": 0.561, + "mean_token_accuracy": 0.803313672542572, + "num_tokens": 868900677.0, + "step": 32566 + }, + { + "epoch": 2.6854951760534345, + "grad_norm": 1.9364991188049316, + "learning_rate": 1e-06, + "loss": 0.6698, + "mean_token_accuracy": 0.7838473320007324, + "num_tokens": 868933848.0, + "step": 32567 + }, + { + "epoch": 2.6855776366784863, + "grad_norm": 2.4514527320861816, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7825578451156616, + "num_tokens": 868956867.0, + "step": 32568 + }, + { + "epoch": 2.6856600973035376, + "grad_norm": 2.055042028427124, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.7842699289321899, + "num_tokens": 868985901.0, + "step": 32569 + }, + { + "epoch": 2.685742557928589, + "grad_norm": 2.5311038494110107, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7847118377685547, + "num_tokens": 869007778.0, + "step": 32570 + }, + { + "epoch": 2.6858250185536408, + "grad_norm": 2.3908469676971436, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.7575984001159668, + "num_tokens": 869030538.0, + "step": 32571 + }, + { + "epoch": 2.685907479178692, + "grad_norm": 2.7587804794311523, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7568333148956299, + "num_tokens": 869050771.0, + "step": 32572 + }, + { + "epoch": 2.6859899398037435, + "grad_norm": 2.203340530395508, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7781830430030823, + "num_tokens": 869078092.0, + "step": 32573 + }, + { + "epoch": 2.6860724004287952, + "grad_norm": 2.2475850582122803, + "learning_rate": 1e-06, + "loss": 0.5954, + "mean_token_accuracy": 0.8005013465881348, + "num_tokens": 869104753.0, + "step": 32574 + }, + { + "epoch": 2.686154861053847, + "grad_norm": 2.095984697341919, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.785262942314148, + "num_tokens": 869134716.0, + "step": 32575 + }, + { + "epoch": 2.6862373216788984, + "grad_norm": 2.1451613903045654, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.772293210029602, + "num_tokens": 869165049.0, + "step": 32576 + }, + { + "epoch": 2.6863197823039497, + "grad_norm": 2.6307339668273926, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7484422326087952, + "num_tokens": 869186735.0, + "step": 32577 + }, + { + "epoch": 2.6864022429290015, + "grad_norm": 2.0091381072998047, + "learning_rate": 1e-06, + "loss": 0.6467, + "mean_token_accuracy": 0.7839844226837158, + "num_tokens": 869220647.0, + "step": 32578 + }, + { + "epoch": 2.686484703554053, + "grad_norm": 1.8510080575942993, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7591683864593506, + "num_tokens": 869257765.0, + "step": 32579 + }, + { + "epoch": 2.6865671641791042, + "grad_norm": 2.0583105087280273, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7452815771102905, + "num_tokens": 869288893.0, + "step": 32580 + }, + { + "epoch": 2.686649624804156, + "grad_norm": 2.034925937652588, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7775083780288696, + "num_tokens": 869319949.0, + "step": 32581 + }, + { + "epoch": 2.686732085429208, + "grad_norm": 1.8498446941375732, + "learning_rate": 1e-06, + "loss": 0.6712, + "mean_token_accuracy": 0.783348560333252, + "num_tokens": 869354446.0, + "step": 32582 + }, + { + "epoch": 2.686814546054259, + "grad_norm": 2.016922950744629, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.7980380058288574, + "num_tokens": 869386801.0, + "step": 32583 + }, + { + "epoch": 2.6868970066793105, + "grad_norm": 2.1087238788604736, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7819783687591553, + "num_tokens": 869416846.0, + "step": 32584 + }, + { + "epoch": 2.6869794673043623, + "grad_norm": 2.024493932723999, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7817658185958862, + "num_tokens": 869448500.0, + "step": 32585 + }, + { + "epoch": 2.6870619279294137, + "grad_norm": 2.101292133331299, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7297317981719971, + "num_tokens": 869483197.0, + "step": 32586 + }, + { + "epoch": 2.687144388554465, + "grad_norm": 2.2349085807800293, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7589124441146851, + "num_tokens": 869509681.0, + "step": 32587 + }, + { + "epoch": 2.687226849179517, + "grad_norm": 2.1062583923339844, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7328764200210571, + "num_tokens": 869542128.0, + "step": 32588 + }, + { + "epoch": 2.687309309804568, + "grad_norm": 1.921443223953247, + "learning_rate": 1e-06, + "loss": 0.5805, + "mean_token_accuracy": 0.8115314245223999, + "num_tokens": 869573533.0, + "step": 32589 + }, + { + "epoch": 2.68739177042962, + "grad_norm": 2.582995653152466, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7275211811065674, + "num_tokens": 869595584.0, + "step": 32590 + }, + { + "epoch": 2.6874742310546713, + "grad_norm": 2.342299699783325, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.7777622938156128, + "num_tokens": 869621839.0, + "step": 32591 + }, + { + "epoch": 2.687556691679723, + "grad_norm": 1.7670694589614868, + "learning_rate": 1e-06, + "loss": 0.6292, + "mean_token_accuracy": 0.7903679609298706, + "num_tokens": 869662966.0, + "step": 32592 + }, + { + "epoch": 2.6876391523047745, + "grad_norm": 2.041193723678589, + "learning_rate": 1e-06, + "loss": 0.5742, + "mean_token_accuracy": 0.8133020401000977, + "num_tokens": 869691770.0, + "step": 32593 + }, + { + "epoch": 2.687721612929826, + "grad_norm": 1.8694672584533691, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7567815780639648, + "num_tokens": 869727467.0, + "step": 32594 + }, + { + "epoch": 2.6878040735548776, + "grad_norm": 1.9356423616409302, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7675468921661377, + "num_tokens": 869762180.0, + "step": 32595 + }, + { + "epoch": 2.687886534179929, + "grad_norm": 2.3981573581695557, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7226028442382812, + "num_tokens": 869787235.0, + "step": 32596 + }, + { + "epoch": 2.6879689948049807, + "grad_norm": 2.074052572250366, + "learning_rate": 1e-06, + "loss": 0.4925, + "mean_token_accuracy": 0.8265240788459778, + "num_tokens": 869815343.0, + "step": 32597 + }, + { + "epoch": 2.688051455430032, + "grad_norm": 2.6021275520324707, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.7498174905776978, + "num_tokens": 869837597.0, + "step": 32598 + }, + { + "epoch": 2.688133916055084, + "grad_norm": 2.6138641834259033, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7483211755752563, + "num_tokens": 869862695.0, + "step": 32599 + }, + { + "epoch": 2.6882163766801352, + "grad_norm": 2.236053228378296, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7426458597183228, + "num_tokens": 869891263.0, + "step": 32600 + }, + { + "epoch": 2.6882988373051866, + "grad_norm": 2.0001111030578613, + "learning_rate": 1e-06, + "loss": 0.5224, + "mean_token_accuracy": 0.8282259106636047, + "num_tokens": 869923460.0, + "step": 32601 + }, + { + "epoch": 2.6883812979302384, + "grad_norm": 2.0155880451202393, + "learning_rate": 1e-06, + "loss": 0.6151, + "mean_token_accuracy": 0.7942313551902771, + "num_tokens": 869954449.0, + "step": 32602 + }, + { + "epoch": 2.6884637585552897, + "grad_norm": 2.029696464538574, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.7995671033859253, + "num_tokens": 869986108.0, + "step": 32603 + }, + { + "epoch": 2.6885462191803415, + "grad_norm": 2.139246940612793, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.781522274017334, + "num_tokens": 870011896.0, + "step": 32604 + }, + { + "epoch": 2.688628679805393, + "grad_norm": 2.1428489685058594, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7461124658584595, + "num_tokens": 870043403.0, + "step": 32605 + }, + { + "epoch": 2.6887111404304447, + "grad_norm": 1.9762024879455566, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7508203387260437, + "num_tokens": 870077736.0, + "step": 32606 + }, + { + "epoch": 2.688793601055496, + "grad_norm": 2.1970748901367188, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7674002647399902, + "num_tokens": 870105759.0, + "step": 32607 + }, + { + "epoch": 2.6888760616805474, + "grad_norm": 1.809636116027832, + "learning_rate": 1e-06, + "loss": 0.501, + "mean_token_accuracy": 0.8289436101913452, + "num_tokens": 870142047.0, + "step": 32608 + }, + { + "epoch": 2.688958522305599, + "grad_norm": 2.1524736881256104, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.7531046271324158, + "num_tokens": 870172485.0, + "step": 32609 + }, + { + "epoch": 2.6890409829306505, + "grad_norm": 1.8689519166946411, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7852193117141724, + "num_tokens": 870204549.0, + "step": 32610 + }, + { + "epoch": 2.689123443555702, + "grad_norm": 2.154680013656616, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7530019283294678, + "num_tokens": 870233526.0, + "step": 32611 + }, + { + "epoch": 2.6892059041807537, + "grad_norm": 2.3557193279266357, + "learning_rate": 1e-06, + "loss": 0.551, + "mean_token_accuracy": 0.7993389368057251, + "num_tokens": 870257939.0, + "step": 32612 + }, + { + "epoch": 2.6892883648058055, + "grad_norm": 2.3160018920898438, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7393501996994019, + "num_tokens": 870283159.0, + "step": 32613 + }, + { + "epoch": 2.689370825430857, + "grad_norm": 1.7584307193756104, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7727739214897156, + "num_tokens": 870323323.0, + "step": 32614 + }, + { + "epoch": 2.689453286055908, + "grad_norm": 2.022667407989502, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.767055094242096, + "num_tokens": 870355436.0, + "step": 32615 + }, + { + "epoch": 2.68953574668096, + "grad_norm": 1.9906069040298462, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.750710129737854, + "num_tokens": 870388442.0, + "step": 32616 + }, + { + "epoch": 2.6896182073060113, + "grad_norm": 1.9147388935089111, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.772131085395813, + "num_tokens": 870424503.0, + "step": 32617 + }, + { + "epoch": 2.6897006679310627, + "grad_norm": 1.9350093603134155, + "learning_rate": 1e-06, + "loss": 0.6602, + "mean_token_accuracy": 0.7785533666610718, + "num_tokens": 870458250.0, + "step": 32618 + }, + { + "epoch": 2.6897831285561145, + "grad_norm": 1.9173026084899902, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7810686826705933, + "num_tokens": 870491218.0, + "step": 32619 + }, + { + "epoch": 2.6898655891811662, + "grad_norm": 2.063223123550415, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.7931210994720459, + "num_tokens": 870520532.0, + "step": 32620 + }, + { + "epoch": 2.6899480498062176, + "grad_norm": 1.8870909214019775, + "learning_rate": 1e-06, + "loss": 0.5307, + "mean_token_accuracy": 0.8116947412490845, + "num_tokens": 870552589.0, + "step": 32621 + }, + { + "epoch": 2.690030510431269, + "grad_norm": 2.2111523151397705, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7422080039978027, + "num_tokens": 870581360.0, + "step": 32622 + }, + { + "epoch": 2.6901129710563207, + "grad_norm": 2.0390243530273438, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7519317865371704, + "num_tokens": 870612019.0, + "step": 32623 + }, + { + "epoch": 2.690195431681372, + "grad_norm": 2.0401463508605957, + "learning_rate": 1e-06, + "loss": 0.6673, + "mean_token_accuracy": 0.7824652791023254, + "num_tokens": 870642601.0, + "step": 32624 + }, + { + "epoch": 2.6902778923064234, + "grad_norm": 1.8563965559005737, + "learning_rate": 1e-06, + "loss": 0.6685, + "mean_token_accuracy": 0.7818835973739624, + "num_tokens": 870678932.0, + "step": 32625 + }, + { + "epoch": 2.6903603529314752, + "grad_norm": 2.005993366241455, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7356221675872803, + "num_tokens": 870713212.0, + "step": 32626 + }, + { + "epoch": 2.690442813556527, + "grad_norm": 2.181708335876465, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7789776921272278, + "num_tokens": 870742408.0, + "step": 32627 + }, + { + "epoch": 2.6905252741815784, + "grad_norm": 1.946577548980713, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.793698787689209, + "num_tokens": 870774854.0, + "step": 32628 + }, + { + "epoch": 2.6906077348066297, + "grad_norm": 2.119749069213867, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7646926045417786, + "num_tokens": 870804396.0, + "step": 32629 + }, + { + "epoch": 2.6906901954316815, + "grad_norm": 2.298229455947876, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7825281620025635, + "num_tokens": 870832670.0, + "step": 32630 + }, + { + "epoch": 2.690772656056733, + "grad_norm": 1.848766803741455, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.7926108837127686, + "num_tokens": 870868232.0, + "step": 32631 + }, + { + "epoch": 2.6908551166817842, + "grad_norm": 2.640449285507202, + "learning_rate": 1e-06, + "loss": 0.6447, + "mean_token_accuracy": 0.7922396659851074, + "num_tokens": 870901017.0, + "step": 32632 + }, + { + "epoch": 2.690937577306836, + "grad_norm": 1.9749951362609863, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7654975652694702, + "num_tokens": 870937165.0, + "step": 32633 + }, + { + "epoch": 2.6910200379318874, + "grad_norm": 1.8784065246582031, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.7717905044555664, + "num_tokens": 870974923.0, + "step": 32634 + }, + { + "epoch": 2.691102498556939, + "grad_norm": 1.9138330221176147, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.7924610376358032, + "num_tokens": 871012033.0, + "step": 32635 + }, + { + "epoch": 2.6911849591819905, + "grad_norm": 2.409259557723999, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7692998647689819, + "num_tokens": 871035899.0, + "step": 32636 + }, + { + "epoch": 2.6912674198070423, + "grad_norm": 2.1609530448913574, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.765543520450592, + "num_tokens": 871064214.0, + "step": 32637 + }, + { + "epoch": 2.6913498804320937, + "grad_norm": 2.2896828651428223, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7590216398239136, + "num_tokens": 871092464.0, + "step": 32638 + }, + { + "epoch": 2.691432341057145, + "grad_norm": 2.1754419803619385, + "learning_rate": 1e-06, + "loss": 0.6267, + "mean_token_accuracy": 0.7966996431350708, + "num_tokens": 871121603.0, + "step": 32639 + }, + { + "epoch": 2.691514801682197, + "grad_norm": 2.0103225708007812, + "learning_rate": 1e-06, + "loss": 0.6079, + "mean_token_accuracy": 0.8008298277854919, + "num_tokens": 871154473.0, + "step": 32640 + }, + { + "epoch": 2.691597262307248, + "grad_norm": 1.8363057374954224, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7947160601615906, + "num_tokens": 871193259.0, + "step": 32641 + }, + { + "epoch": 2.6916797229323, + "grad_norm": 1.9824167490005493, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7702752351760864, + "num_tokens": 871230643.0, + "step": 32642 + }, + { + "epoch": 2.6917621835573513, + "grad_norm": 2.043595552444458, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7579358816146851, + "num_tokens": 871260224.0, + "step": 32643 + }, + { + "epoch": 2.691844644182403, + "grad_norm": 2.0991880893707275, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7505546808242798, + "num_tokens": 871290538.0, + "step": 32644 + }, + { + "epoch": 2.6919271048074545, + "grad_norm": 1.9037717580795288, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.7883995771408081, + "num_tokens": 871324776.0, + "step": 32645 + }, + { + "epoch": 2.692009565432506, + "grad_norm": 2.4325687885284424, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7430335283279419, + "num_tokens": 871350505.0, + "step": 32646 + }, + { + "epoch": 2.6920920260575576, + "grad_norm": 1.91016685962677, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7725210189819336, + "num_tokens": 871384879.0, + "step": 32647 + }, + { + "epoch": 2.692174486682609, + "grad_norm": 2.1984167098999023, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7683811783790588, + "num_tokens": 871412138.0, + "step": 32648 + }, + { + "epoch": 2.6922569473076607, + "grad_norm": 1.7926019430160522, + "learning_rate": 1e-06, + "loss": 0.6743, + "mean_token_accuracy": 0.7809081077575684, + "num_tokens": 871451602.0, + "step": 32649 + }, + { + "epoch": 2.692339407932712, + "grad_norm": 2.0622196197509766, + "learning_rate": 1e-06, + "loss": 0.5654, + "mean_token_accuracy": 0.8100893497467041, + "num_tokens": 871482256.0, + "step": 32650 + }, + { + "epoch": 2.692421868557764, + "grad_norm": 2.2839980125427246, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7664885520935059, + "num_tokens": 871510234.0, + "step": 32651 + }, + { + "epoch": 2.6925043291828152, + "grad_norm": 2.288969039916992, + "learning_rate": 1e-06, + "loss": 0.8119, + "mean_token_accuracy": 0.743773877620697, + "num_tokens": 871538518.0, + "step": 32652 + }, + { + "epoch": 2.6925867898078666, + "grad_norm": 2.083139657974243, + "learning_rate": 1e-06, + "loss": 0.5222, + "mean_token_accuracy": 0.8196663856506348, + "num_tokens": 871568799.0, + "step": 32653 + }, + { + "epoch": 2.6926692504329184, + "grad_norm": 2.190500259399414, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.78665691614151, + "num_tokens": 871597065.0, + "step": 32654 + }, + { + "epoch": 2.6927517110579697, + "grad_norm": 1.9620412588119507, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.7685564160346985, + "num_tokens": 871632707.0, + "step": 32655 + }, + { + "epoch": 2.692834171683021, + "grad_norm": 2.244424819946289, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7746779918670654, + "num_tokens": 871658634.0, + "step": 32656 + }, + { + "epoch": 2.692916632308073, + "grad_norm": 2.2379143238067627, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7573885917663574, + "num_tokens": 871687497.0, + "step": 32657 + }, + { + "epoch": 2.6929990929331247, + "grad_norm": 1.9364893436431885, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.7886776924133301, + "num_tokens": 871719600.0, + "step": 32658 + }, + { + "epoch": 2.693081553558176, + "grad_norm": 2.182325601577759, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7809565663337708, + "num_tokens": 871746857.0, + "step": 32659 + }, + { + "epoch": 2.6931640141832274, + "grad_norm": 1.6352880001068115, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.7892230153083801, + "num_tokens": 871790865.0, + "step": 32660 + }, + { + "epoch": 2.693246474808279, + "grad_norm": 1.8817856311798096, + "learning_rate": 1e-06, + "loss": 0.5974, + "mean_token_accuracy": 0.8084436655044556, + "num_tokens": 871825625.0, + "step": 32661 + }, + { + "epoch": 2.6933289354333305, + "grad_norm": 2.111293077468872, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7404909133911133, + "num_tokens": 871856100.0, + "step": 32662 + }, + { + "epoch": 2.693411396058382, + "grad_norm": 2.300349473953247, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7502062320709229, + "num_tokens": 871884196.0, + "step": 32663 + }, + { + "epoch": 2.6934938566834337, + "grad_norm": 2.139227867126465, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7412523627281189, + "num_tokens": 871913670.0, + "step": 32664 + }, + { + "epoch": 2.6935763173084855, + "grad_norm": 1.948021650314331, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.7812385559082031, + "num_tokens": 871945299.0, + "step": 32665 + }, + { + "epoch": 2.693658777933537, + "grad_norm": 1.7791274785995483, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7431286573410034, + "num_tokens": 871984755.0, + "step": 32666 + }, + { + "epoch": 2.693741238558588, + "grad_norm": 2.015557289123535, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7299154996871948, + "num_tokens": 872019644.0, + "step": 32667 + }, + { + "epoch": 2.69382369918364, + "grad_norm": 2.222738742828369, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7330436110496521, + "num_tokens": 872047145.0, + "step": 32668 + }, + { + "epoch": 2.6939061598086913, + "grad_norm": 2.1798348426818848, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.7879780530929565, + "num_tokens": 872072607.0, + "step": 32669 + }, + { + "epoch": 2.6939886204337427, + "grad_norm": 2.565082550048828, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.744326114654541, + "num_tokens": 872094655.0, + "step": 32670 + }, + { + "epoch": 2.6940710810587944, + "grad_norm": 1.8281482458114624, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.8020545840263367, + "num_tokens": 872129694.0, + "step": 32671 + }, + { + "epoch": 2.694153541683846, + "grad_norm": 1.9522759914398193, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.7553502321243286, + "num_tokens": 872162065.0, + "step": 32672 + }, + { + "epoch": 2.6942360023088976, + "grad_norm": 1.920993685722351, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.7921977043151855, + "num_tokens": 872194312.0, + "step": 32673 + }, + { + "epoch": 2.694318462933949, + "grad_norm": 1.9095194339752197, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.7860084772109985, + "num_tokens": 872225844.0, + "step": 32674 + }, + { + "epoch": 2.6944009235590007, + "grad_norm": 2.5441653728485107, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7528055906295776, + "num_tokens": 872246453.0, + "step": 32675 + }, + { + "epoch": 2.694483384184052, + "grad_norm": 1.9944792985916138, + "learning_rate": 1e-06, + "loss": 0.5081, + "mean_token_accuracy": 0.8151916861534119, + "num_tokens": 872274607.0, + "step": 32676 + }, + { + "epoch": 2.6945658448091034, + "grad_norm": 2.276115894317627, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.759190559387207, + "num_tokens": 872302069.0, + "step": 32677 + }, + { + "epoch": 2.6946483054341552, + "grad_norm": 2.0712757110595703, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7635246515274048, + "num_tokens": 872333547.0, + "step": 32678 + }, + { + "epoch": 2.6947307660592066, + "grad_norm": 1.8586369752883911, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.748016357421875, + "num_tokens": 872374009.0, + "step": 32679 + }, + { + "epoch": 2.6948132266842584, + "grad_norm": 2.249441146850586, + "learning_rate": 1e-06, + "loss": 0.5676, + "mean_token_accuracy": 0.8132213950157166, + "num_tokens": 872398757.0, + "step": 32680 + }, + { + "epoch": 2.6948956873093097, + "grad_norm": 1.7706266641616821, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7551100254058838, + "num_tokens": 872438765.0, + "step": 32681 + }, + { + "epoch": 2.6949781479343615, + "grad_norm": 2.412233352661133, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7671019434928894, + "num_tokens": 872462819.0, + "step": 32682 + }, + { + "epoch": 2.695060608559413, + "grad_norm": 2.008384943008423, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7714967131614685, + "num_tokens": 872497248.0, + "step": 32683 + }, + { + "epoch": 2.6951430691844642, + "grad_norm": 2.156022548675537, + "learning_rate": 1e-06, + "loss": 0.5583, + "mean_token_accuracy": 0.8085950613021851, + "num_tokens": 872524488.0, + "step": 32684 + }, + { + "epoch": 2.695225529809516, + "grad_norm": 2.626229763031006, + "learning_rate": 1e-06, + "loss": 0.7014, + "mean_token_accuracy": 0.758734941482544, + "num_tokens": 872547876.0, + "step": 32685 + }, + { + "epoch": 2.6953079904345674, + "grad_norm": 1.9963570833206177, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7679654359817505, + "num_tokens": 872581997.0, + "step": 32686 + }, + { + "epoch": 2.695390451059619, + "grad_norm": 2.121237277984619, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.7980338335037231, + "num_tokens": 872606638.0, + "step": 32687 + }, + { + "epoch": 2.6954729116846705, + "grad_norm": 2.1150062084198, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7710337042808533, + "num_tokens": 872634301.0, + "step": 32688 + }, + { + "epoch": 2.6955553723097223, + "grad_norm": 2.2863099575042725, + "learning_rate": 1e-06, + "loss": 0.6301, + "mean_token_accuracy": 0.7924540042877197, + "num_tokens": 872659459.0, + "step": 32689 + }, + { + "epoch": 2.6956378329347737, + "grad_norm": 2.361846446990967, + "learning_rate": 1e-06, + "loss": 0.5053, + "mean_token_accuracy": 0.8189414143562317, + "num_tokens": 872681482.0, + "step": 32690 + }, + { + "epoch": 2.695720293559825, + "grad_norm": 2.046619415283203, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7828854918479919, + "num_tokens": 872711336.0, + "step": 32691 + }, + { + "epoch": 2.695802754184877, + "grad_norm": 2.0521388053894043, + "learning_rate": 1e-06, + "loss": 0.7325, + "mean_token_accuracy": 0.7653775215148926, + "num_tokens": 872744408.0, + "step": 32692 + }, + { + "epoch": 2.695885214809928, + "grad_norm": 2.095057964324951, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7724384069442749, + "num_tokens": 872773366.0, + "step": 32693 + }, + { + "epoch": 2.69596767543498, + "grad_norm": 2.1350386142730713, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7856045961380005, + "num_tokens": 872803137.0, + "step": 32694 + }, + { + "epoch": 2.6960501360600313, + "grad_norm": 1.934363603591919, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.7900125980377197, + "num_tokens": 872835973.0, + "step": 32695 + }, + { + "epoch": 2.696132596685083, + "grad_norm": 1.8716332912445068, + "learning_rate": 1e-06, + "loss": 0.581, + "mean_token_accuracy": 0.8031632900238037, + "num_tokens": 872871587.0, + "step": 32696 + }, + { + "epoch": 2.6962150573101344, + "grad_norm": 2.1586058139801025, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7489668130874634, + "num_tokens": 872904454.0, + "step": 32697 + }, + { + "epoch": 2.696297517935186, + "grad_norm": 2.132357120513916, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7607754468917847, + "num_tokens": 872933224.0, + "step": 32698 + }, + { + "epoch": 2.6963799785602376, + "grad_norm": 1.9323275089263916, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7640263438224792, + "num_tokens": 872967745.0, + "step": 32699 + }, + { + "epoch": 2.696462439185289, + "grad_norm": 2.257412910461426, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.779526948928833, + "num_tokens": 872992671.0, + "step": 32700 + }, + { + "epoch": 2.6965448998103403, + "grad_norm": 1.7373610734939575, + "learning_rate": 1e-06, + "loss": 0.4242, + "mean_token_accuracy": 0.8495925664901733, + "num_tokens": 873027451.0, + "step": 32701 + }, + { + "epoch": 2.696627360435392, + "grad_norm": 2.101672887802124, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7735961675643921, + "num_tokens": 873059452.0, + "step": 32702 + }, + { + "epoch": 2.696709821060444, + "grad_norm": 2.3704288005828857, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7556447982788086, + "num_tokens": 873087234.0, + "step": 32703 + }, + { + "epoch": 2.6967922816854952, + "grad_norm": 1.7996565103530884, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.7984093427658081, + "num_tokens": 873123341.0, + "step": 32704 + }, + { + "epoch": 2.6968747423105466, + "grad_norm": 2.297208547592163, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.7927976250648499, + "num_tokens": 873147576.0, + "step": 32705 + }, + { + "epoch": 2.6969572029355984, + "grad_norm": 2.0287110805511475, + "learning_rate": 1e-06, + "loss": 0.5876, + "mean_token_accuracy": 0.8042470216751099, + "num_tokens": 873178445.0, + "step": 32706 + }, + { + "epoch": 2.6970396635606497, + "grad_norm": 2.2123289108276367, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7585333585739136, + "num_tokens": 873208186.0, + "step": 32707 + }, + { + "epoch": 2.697122124185701, + "grad_norm": 2.3993759155273438, + "learning_rate": 1e-06, + "loss": 0.6142, + "mean_token_accuracy": 0.7803345918655396, + "num_tokens": 873231059.0, + "step": 32708 + }, + { + "epoch": 2.697204584810753, + "grad_norm": 2.088456869125366, + "learning_rate": 1e-06, + "loss": 0.572, + "mean_token_accuracy": 0.8099244236946106, + "num_tokens": 873259721.0, + "step": 32709 + }, + { + "epoch": 2.6972870454358047, + "grad_norm": 1.968662142753601, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.776447594165802, + "num_tokens": 873295589.0, + "step": 32710 + }, + { + "epoch": 2.697369506060856, + "grad_norm": 2.1658666133880615, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7320929765701294, + "num_tokens": 873325087.0, + "step": 32711 + }, + { + "epoch": 2.6974519666859074, + "grad_norm": 2.1256790161132812, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7732824683189392, + "num_tokens": 873355619.0, + "step": 32712 + }, + { + "epoch": 2.697534427310959, + "grad_norm": 2.0177314281463623, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7403372526168823, + "num_tokens": 873390086.0, + "step": 32713 + }, + { + "epoch": 2.6976168879360105, + "grad_norm": 2.2199411392211914, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7785050272941589, + "num_tokens": 873417678.0, + "step": 32714 + }, + { + "epoch": 2.697699348561062, + "grad_norm": 2.157735824584961, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7635051012039185, + "num_tokens": 873449690.0, + "step": 32715 + }, + { + "epoch": 2.6977818091861137, + "grad_norm": 1.9595633745193481, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7743285298347473, + "num_tokens": 873481643.0, + "step": 32716 + }, + { + "epoch": 2.697864269811165, + "grad_norm": 1.8725571632385254, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7618362903594971, + "num_tokens": 873517157.0, + "step": 32717 + }, + { + "epoch": 2.697946730436217, + "grad_norm": 2.2491888999938965, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7703826427459717, + "num_tokens": 873543378.0, + "step": 32718 + }, + { + "epoch": 2.698029191061268, + "grad_norm": 2.0457897186279297, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7769627571105957, + "num_tokens": 873575603.0, + "step": 32719 + }, + { + "epoch": 2.69811165168632, + "grad_norm": 2.038165330886841, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7719568014144897, + "num_tokens": 873607718.0, + "step": 32720 + }, + { + "epoch": 2.6981941123113713, + "grad_norm": 2.1962456703186035, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7612642645835876, + "num_tokens": 873638258.0, + "step": 32721 + }, + { + "epoch": 2.6982765729364226, + "grad_norm": 1.8393959999084473, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7844804525375366, + "num_tokens": 873678993.0, + "step": 32722 + }, + { + "epoch": 2.6983590335614744, + "grad_norm": 2.1617627143859863, + "learning_rate": 1e-06, + "loss": 0.6907, + "mean_token_accuracy": 0.7706894874572754, + "num_tokens": 873705760.0, + "step": 32723 + }, + { + "epoch": 2.698441494186526, + "grad_norm": 2.21709942817688, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7669091820716858, + "num_tokens": 873732002.0, + "step": 32724 + }, + { + "epoch": 2.6985239548115776, + "grad_norm": 2.5371108055114746, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7242928147315979, + "num_tokens": 873756095.0, + "step": 32725 + }, + { + "epoch": 2.698606415436629, + "grad_norm": 2.043414831161499, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.7883920669555664, + "num_tokens": 873786443.0, + "step": 32726 + }, + { + "epoch": 2.6986888760616807, + "grad_norm": 2.09771990776062, + "learning_rate": 1e-06, + "loss": 0.5938, + "mean_token_accuracy": 0.7989563941955566, + "num_tokens": 873814143.0, + "step": 32727 + }, + { + "epoch": 2.698771336686732, + "grad_norm": 1.7547879219055176, + "learning_rate": 1e-06, + "loss": 0.5912, + "mean_token_accuracy": 0.8100483417510986, + "num_tokens": 873852575.0, + "step": 32728 + }, + { + "epoch": 2.6988537973117834, + "grad_norm": 2.063277244567871, + "learning_rate": 1e-06, + "loss": 0.5857, + "mean_token_accuracy": 0.7881094217300415, + "num_tokens": 873879792.0, + "step": 32729 + }, + { + "epoch": 2.6989362579368352, + "grad_norm": 2.239860773086548, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7599154710769653, + "num_tokens": 873906643.0, + "step": 32730 + }, + { + "epoch": 2.6990187185618866, + "grad_norm": 2.4959044456481934, + "learning_rate": 1e-06, + "loss": 0.722, + "mean_token_accuracy": 0.7662599682807922, + "num_tokens": 873928199.0, + "step": 32731 + }, + { + "epoch": 2.6991011791869384, + "grad_norm": 2.110104560852051, + "learning_rate": 1e-06, + "loss": 0.756, + "mean_token_accuracy": 0.7577530741691589, + "num_tokens": 873959899.0, + "step": 32732 + }, + { + "epoch": 2.6991836398119897, + "grad_norm": 1.9197165966033936, + "learning_rate": 1e-06, + "loss": 0.5518, + "mean_token_accuracy": 0.8184130191802979, + "num_tokens": 873991404.0, + "step": 32733 + }, + { + "epoch": 2.6992661004370415, + "grad_norm": 2.272446870803833, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7533384561538696, + "num_tokens": 874018624.0, + "step": 32734 + }, + { + "epoch": 2.699348561062093, + "grad_norm": 2.308014154434204, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.7839145064353943, + "num_tokens": 874041846.0, + "step": 32735 + }, + { + "epoch": 2.699431021687144, + "grad_norm": 2.1207022666931152, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.8005313277244568, + "num_tokens": 874074297.0, + "step": 32736 + }, + { + "epoch": 2.699513482312196, + "grad_norm": 2.103184461593628, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7633143663406372, + "num_tokens": 874104225.0, + "step": 32737 + }, + { + "epoch": 2.6995959429372474, + "grad_norm": 2.1422245502471924, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7748634815216064, + "num_tokens": 874132281.0, + "step": 32738 + }, + { + "epoch": 2.699678403562299, + "grad_norm": 2.2176966667175293, + "learning_rate": 1e-06, + "loss": 0.7712, + "mean_token_accuracy": 0.7487273812294006, + "num_tokens": 874159989.0, + "step": 32739 + }, + { + "epoch": 2.6997608641873505, + "grad_norm": 1.9789166450500488, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7639633417129517, + "num_tokens": 874197064.0, + "step": 32740 + }, + { + "epoch": 2.6998433248124023, + "grad_norm": 2.190152883529663, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.7779618501663208, + "num_tokens": 874225748.0, + "step": 32741 + }, + { + "epoch": 2.6999257854374537, + "grad_norm": 1.7960867881774902, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.7902219295501709, + "num_tokens": 874263882.0, + "step": 32742 + }, + { + "epoch": 2.700008246062505, + "grad_norm": 2.012625217437744, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7873038053512573, + "num_tokens": 874294168.0, + "step": 32743 + }, + { + "epoch": 2.700090706687557, + "grad_norm": 1.9164018630981445, + "learning_rate": 1e-06, + "loss": 0.4991, + "mean_token_accuracy": 0.8303912878036499, + "num_tokens": 874325241.0, + "step": 32744 + }, + { + "epoch": 2.700173167312608, + "grad_norm": 2.2649362087249756, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7233667373657227, + "num_tokens": 874353438.0, + "step": 32745 + }, + { + "epoch": 2.7002556279376595, + "grad_norm": 1.963401436805725, + "learning_rate": 1e-06, + "loss": 0.6215, + "mean_token_accuracy": 0.8006365299224854, + "num_tokens": 874386155.0, + "step": 32746 + }, + { + "epoch": 2.7003380885627113, + "grad_norm": 2.2645578384399414, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.7916680574417114, + "num_tokens": 874417168.0, + "step": 32747 + }, + { + "epoch": 2.700420549187763, + "grad_norm": 2.0030477046966553, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7600237131118774, + "num_tokens": 874449694.0, + "step": 32748 + }, + { + "epoch": 2.7005030098128144, + "grad_norm": 2.1826210021972656, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7623807787895203, + "num_tokens": 874479280.0, + "step": 32749 + }, + { + "epoch": 2.700585470437866, + "grad_norm": 2.12772274017334, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7235184907913208, + "num_tokens": 874509002.0, + "step": 32750 + }, + { + "epoch": 2.7006679310629176, + "grad_norm": 2.049196481704712, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7675423622131348, + "num_tokens": 874539145.0, + "step": 32751 + }, + { + "epoch": 2.700750391687969, + "grad_norm": 2.1676483154296875, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7312115430831909, + "num_tokens": 874569181.0, + "step": 32752 + }, + { + "epoch": 2.7008328523130203, + "grad_norm": 2.157353162765503, + "learning_rate": 1e-06, + "loss": 0.5933, + "mean_token_accuracy": 0.7986770272254944, + "num_tokens": 874596635.0, + "step": 32753 + }, + { + "epoch": 2.700915312938072, + "grad_norm": 1.9614149332046509, + "learning_rate": 1e-06, + "loss": 0.5892, + "mean_token_accuracy": 0.8023128509521484, + "num_tokens": 874632367.0, + "step": 32754 + }, + { + "epoch": 2.700997773563124, + "grad_norm": 2.1895911693573, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7692493200302124, + "num_tokens": 874659270.0, + "step": 32755 + }, + { + "epoch": 2.701080234188175, + "grad_norm": 2.529611825942993, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7690989375114441, + "num_tokens": 874682553.0, + "step": 32756 + }, + { + "epoch": 2.7011626948132266, + "grad_norm": 1.8958951234817505, + "learning_rate": 1e-06, + "loss": 0.5796, + "mean_token_accuracy": 0.8116931319236755, + "num_tokens": 874714058.0, + "step": 32757 + }, + { + "epoch": 2.7012451554382784, + "grad_norm": 1.7467535734176636, + "learning_rate": 1e-06, + "loss": 0.6843, + "mean_token_accuracy": 0.7801576256752014, + "num_tokens": 874754311.0, + "step": 32758 + }, + { + "epoch": 2.7013276160633297, + "grad_norm": 2.342585563659668, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7394793629646301, + "num_tokens": 874779402.0, + "step": 32759 + }, + { + "epoch": 2.701410076688381, + "grad_norm": 2.1425118446350098, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7671843767166138, + "num_tokens": 874808578.0, + "step": 32760 + }, + { + "epoch": 2.701492537313433, + "grad_norm": 1.8301197290420532, + "learning_rate": 1e-06, + "loss": 0.6091, + "mean_token_accuracy": 0.7959834337234497, + "num_tokens": 874847514.0, + "step": 32761 + }, + { + "epoch": 2.701574997938484, + "grad_norm": 2.2637667655944824, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.75201016664505, + "num_tokens": 874875807.0, + "step": 32762 + }, + { + "epoch": 2.701657458563536, + "grad_norm": 2.0110514163970947, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.8067668676376343, + "num_tokens": 874905171.0, + "step": 32763 + }, + { + "epoch": 2.7017399191885874, + "grad_norm": 2.0035243034362793, + "learning_rate": 1e-06, + "loss": 0.5606, + "mean_token_accuracy": 0.8068329095840454, + "num_tokens": 874938100.0, + "step": 32764 + }, + { + "epoch": 2.701822379813639, + "grad_norm": 2.2758564949035645, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7795889377593994, + "num_tokens": 874962222.0, + "step": 32765 + }, + { + "epoch": 2.7019048404386905, + "grad_norm": 1.8535239696502686, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7703952789306641, + "num_tokens": 874999039.0, + "step": 32766 + }, + { + "epoch": 2.701987301063742, + "grad_norm": 2.1644551753997803, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.783395528793335, + "num_tokens": 875027655.0, + "step": 32767 + }, + { + "epoch": 2.7020697616887936, + "grad_norm": 2.202031373977661, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.7848434448242188, + "num_tokens": 875055228.0, + "step": 32768 + }, + { + "epoch": 2.702152222313845, + "grad_norm": 2.155808210372925, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7800146341323853, + "num_tokens": 875082421.0, + "step": 32769 + }, + { + "epoch": 2.702234682938897, + "grad_norm": 2.146834373474121, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7590699195861816, + "num_tokens": 875111229.0, + "step": 32770 + }, + { + "epoch": 2.702317143563948, + "grad_norm": 2.0826945304870605, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.7646657228469849, + "num_tokens": 875142356.0, + "step": 32771 + }, + { + "epoch": 2.702399604189, + "grad_norm": 2.366323232650757, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7507132291793823, + "num_tokens": 875166265.0, + "step": 32772 + }, + { + "epoch": 2.7024820648140513, + "grad_norm": 2.201780080795288, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7364684343338013, + "num_tokens": 875196231.0, + "step": 32773 + }, + { + "epoch": 2.7025645254391026, + "grad_norm": 2.3357808589935303, + "learning_rate": 1e-06, + "loss": 0.6225, + "mean_token_accuracy": 0.792510986328125, + "num_tokens": 875221118.0, + "step": 32774 + }, + { + "epoch": 2.7026469860641544, + "grad_norm": 1.9771754741668701, + "learning_rate": 1e-06, + "loss": 0.6006, + "mean_token_accuracy": 0.795988917350769, + "num_tokens": 875252695.0, + "step": 32775 + }, + { + "epoch": 2.702729446689206, + "grad_norm": 2.3373939990997314, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7762556672096252, + "num_tokens": 875278459.0, + "step": 32776 + }, + { + "epoch": 2.7028119073142576, + "grad_norm": 2.017638921737671, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.7826542258262634, + "num_tokens": 875310137.0, + "step": 32777 + }, + { + "epoch": 2.702894367939309, + "grad_norm": 2.1867363452911377, + "learning_rate": 1e-06, + "loss": 0.6272, + "mean_token_accuracy": 0.7913109064102173, + "num_tokens": 875336713.0, + "step": 32778 + }, + { + "epoch": 2.7029768285643607, + "grad_norm": 2.1869592666625977, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7745262384414673, + "num_tokens": 875365605.0, + "step": 32779 + }, + { + "epoch": 2.703059289189412, + "grad_norm": 2.0766563415527344, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7737322449684143, + "num_tokens": 875397738.0, + "step": 32780 + }, + { + "epoch": 2.7031417498144634, + "grad_norm": 2.385854721069336, + "learning_rate": 1e-06, + "loss": 0.7141, + "mean_token_accuracy": 0.7644094228744507, + "num_tokens": 875425851.0, + "step": 32781 + }, + { + "epoch": 2.703224210439515, + "grad_norm": 2.0697336196899414, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.7742434144020081, + "num_tokens": 875457480.0, + "step": 32782 + }, + { + "epoch": 2.7033066710645666, + "grad_norm": 2.2759125232696533, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7438075542449951, + "num_tokens": 875483284.0, + "step": 32783 + }, + { + "epoch": 2.7033891316896184, + "grad_norm": 2.3122284412384033, + "learning_rate": 1e-06, + "loss": 0.7647, + "mean_token_accuracy": 0.7511781454086304, + "num_tokens": 875511177.0, + "step": 32784 + }, + { + "epoch": 2.7034715923146697, + "grad_norm": 2.010918617248535, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.7801281213760376, + "num_tokens": 875544007.0, + "step": 32785 + }, + { + "epoch": 2.7035540529397215, + "grad_norm": 2.279527425765991, + "learning_rate": 1e-06, + "loss": 0.5196, + "mean_token_accuracy": 0.8327846527099609, + "num_tokens": 875570803.0, + "step": 32786 + }, + { + "epoch": 2.703636513564773, + "grad_norm": 2.453094482421875, + "learning_rate": 1e-06, + "loss": 0.7879, + "mean_token_accuracy": 0.7489888072013855, + "num_tokens": 875595126.0, + "step": 32787 + }, + { + "epoch": 2.703718974189824, + "grad_norm": 2.380119800567627, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7603179216384888, + "num_tokens": 875619853.0, + "step": 32788 + }, + { + "epoch": 2.703801434814876, + "grad_norm": 2.3754289150238037, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7747116088867188, + "num_tokens": 875644571.0, + "step": 32789 + }, + { + "epoch": 2.7038838954399274, + "grad_norm": 2.134418249130249, + "learning_rate": 1e-06, + "loss": 0.607, + "mean_token_accuracy": 0.810207188129425, + "num_tokens": 875673018.0, + "step": 32790 + }, + { + "epoch": 2.7039663560649787, + "grad_norm": 2.1484549045562744, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7667094469070435, + "num_tokens": 875703158.0, + "step": 32791 + }, + { + "epoch": 2.7040488166900305, + "grad_norm": 2.1727325916290283, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7665810585021973, + "num_tokens": 875732210.0, + "step": 32792 + }, + { + "epoch": 2.7041312773150823, + "grad_norm": 2.265848398208618, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7678626775741577, + "num_tokens": 875763253.0, + "step": 32793 + }, + { + "epoch": 2.7042137379401336, + "grad_norm": 1.9129539728164673, + "learning_rate": 1e-06, + "loss": 0.5762, + "mean_token_accuracy": 0.813596785068512, + "num_tokens": 875798107.0, + "step": 32794 + }, + { + "epoch": 2.704296198565185, + "grad_norm": 1.8915431499481201, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7682446241378784, + "num_tokens": 875833387.0, + "step": 32795 + }, + { + "epoch": 2.704378659190237, + "grad_norm": 1.8576375246047974, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.7893832325935364, + "num_tokens": 875871194.0, + "step": 32796 + }, + { + "epoch": 2.704461119815288, + "grad_norm": 2.5544631481170654, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7352167367935181, + "num_tokens": 875893399.0, + "step": 32797 + }, + { + "epoch": 2.7045435804403395, + "grad_norm": 2.0739591121673584, + "learning_rate": 1e-06, + "loss": 0.5936, + "mean_token_accuracy": 0.8038452863693237, + "num_tokens": 875923810.0, + "step": 32798 + }, + { + "epoch": 2.7046260410653913, + "grad_norm": 1.9068262577056885, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.7918379902839661, + "num_tokens": 875959000.0, + "step": 32799 + }, + { + "epoch": 2.704708501690443, + "grad_norm": 2.379796028137207, + "learning_rate": 1e-06, + "loss": 0.6527, + "mean_token_accuracy": 0.7663692235946655, + "num_tokens": 875982869.0, + "step": 32800 + }, + { + "epoch": 2.7047909623154944, + "grad_norm": 1.9281537532806396, + "learning_rate": 1e-06, + "loss": 0.5032, + "mean_token_accuracy": 0.821636974811554, + "num_tokens": 876013529.0, + "step": 32801 + }, + { + "epoch": 2.704873422940546, + "grad_norm": 2.260103940963745, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7826358079910278, + "num_tokens": 876039171.0, + "step": 32802 + }, + { + "epoch": 2.7049558835655976, + "grad_norm": 2.4606642723083496, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7511485815048218, + "num_tokens": 876061987.0, + "step": 32803 + }, + { + "epoch": 2.705038344190649, + "grad_norm": 1.9632861614227295, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.8045756816864014, + "num_tokens": 876093922.0, + "step": 32804 + }, + { + "epoch": 2.7051208048157003, + "grad_norm": 2.0625319480895996, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.7860899567604065, + "num_tokens": 876124079.0, + "step": 32805 + }, + { + "epoch": 2.705203265440752, + "grad_norm": 1.7757643461227417, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.7679606676101685, + "num_tokens": 876164264.0, + "step": 32806 + }, + { + "epoch": 2.7052857260658034, + "grad_norm": 2.937371015548706, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7661752700805664, + "num_tokens": 876182342.0, + "step": 32807 + }, + { + "epoch": 2.705368186690855, + "grad_norm": 2.218132734298706, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.707331657409668, + "num_tokens": 876214274.0, + "step": 32808 + }, + { + "epoch": 2.7054506473159066, + "grad_norm": 2.093971014022827, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.785516619682312, + "num_tokens": 876243344.0, + "step": 32809 + }, + { + "epoch": 2.7055331079409584, + "grad_norm": 2.302556276321411, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7322494983673096, + "num_tokens": 876270859.0, + "step": 32810 + }, + { + "epoch": 2.7056155685660097, + "grad_norm": 2.0829861164093018, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7583067417144775, + "num_tokens": 876301602.0, + "step": 32811 + }, + { + "epoch": 2.705698029191061, + "grad_norm": 2.069077253341675, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7261360883712769, + "num_tokens": 876332135.0, + "step": 32812 + }, + { + "epoch": 2.705780489816113, + "grad_norm": 1.9965234994888306, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.7685797214508057, + "num_tokens": 876363293.0, + "step": 32813 + }, + { + "epoch": 2.705862950441164, + "grad_norm": 1.845930814743042, + "learning_rate": 1e-06, + "loss": 0.5526, + "mean_token_accuracy": 0.8084776997566223, + "num_tokens": 876400985.0, + "step": 32814 + }, + { + "epoch": 2.705945411066216, + "grad_norm": 2.1718461513519287, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.7832370400428772, + "num_tokens": 876428002.0, + "step": 32815 + }, + { + "epoch": 2.7060278716912674, + "grad_norm": 2.0510811805725098, + "learning_rate": 1e-06, + "loss": 0.7179, + "mean_token_accuracy": 0.7764559984207153, + "num_tokens": 876456295.0, + "step": 32816 + }, + { + "epoch": 2.706110332316319, + "grad_norm": 1.992647409439087, + "learning_rate": 1e-06, + "loss": 0.5727, + "mean_token_accuracy": 0.8018608093261719, + "num_tokens": 876486944.0, + "step": 32817 + }, + { + "epoch": 2.7061927929413705, + "grad_norm": 1.9153056144714355, + "learning_rate": 1e-06, + "loss": 0.5935, + "mean_token_accuracy": 0.8037977814674377, + "num_tokens": 876517431.0, + "step": 32818 + }, + { + "epoch": 2.706275253566422, + "grad_norm": 2.2156002521514893, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7656306028366089, + "num_tokens": 876545617.0, + "step": 32819 + }, + { + "epoch": 2.7063577141914736, + "grad_norm": 1.9079521894454956, + "learning_rate": 1e-06, + "loss": 0.5468, + "mean_token_accuracy": 0.8145477771759033, + "num_tokens": 876579618.0, + "step": 32820 + }, + { + "epoch": 2.706440174816525, + "grad_norm": 1.8185468912124634, + "learning_rate": 1e-06, + "loss": 0.5649, + "mean_token_accuracy": 0.7996751070022583, + "num_tokens": 876613612.0, + "step": 32821 + }, + { + "epoch": 2.706522635441577, + "grad_norm": 2.010877847671509, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7789638042449951, + "num_tokens": 876646050.0, + "step": 32822 + }, + { + "epoch": 2.706605096066628, + "grad_norm": 2.203246831893921, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7610467076301575, + "num_tokens": 876674370.0, + "step": 32823 + }, + { + "epoch": 2.70668755669168, + "grad_norm": 2.222529888153076, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7583487629890442, + "num_tokens": 876702139.0, + "step": 32824 + }, + { + "epoch": 2.7067700173167313, + "grad_norm": 2.208035707473755, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7681576609611511, + "num_tokens": 876729854.0, + "step": 32825 + }, + { + "epoch": 2.7068524779417826, + "grad_norm": 1.9877924919128418, + "learning_rate": 1e-06, + "loss": 0.6069, + "mean_token_accuracy": 0.8023321032524109, + "num_tokens": 876761587.0, + "step": 32826 + }, + { + "epoch": 2.7069349385668344, + "grad_norm": 2.2038888931274414, + "learning_rate": 1e-06, + "loss": 0.6201, + "mean_token_accuracy": 0.7900137901306152, + "num_tokens": 876787473.0, + "step": 32827 + }, + { + "epoch": 2.7070173991918858, + "grad_norm": 2.0102384090423584, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7769303917884827, + "num_tokens": 876817628.0, + "step": 32828 + }, + { + "epoch": 2.7070998598169376, + "grad_norm": 2.33880615234375, + "learning_rate": 1e-06, + "loss": 0.7066, + "mean_token_accuracy": 0.7582753896713257, + "num_tokens": 876842415.0, + "step": 32829 + }, + { + "epoch": 2.707182320441989, + "grad_norm": 2.098051071166992, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7649974822998047, + "num_tokens": 876872681.0, + "step": 32830 + }, + { + "epoch": 2.7072647810670407, + "grad_norm": 2.0324866771698, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7685545086860657, + "num_tokens": 876903082.0, + "step": 32831 + }, + { + "epoch": 2.707347241692092, + "grad_norm": 1.9097166061401367, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.7888739109039307, + "num_tokens": 876936655.0, + "step": 32832 + }, + { + "epoch": 2.7074297023171434, + "grad_norm": 2.1684789657592773, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.7834131717681885, + "num_tokens": 876963784.0, + "step": 32833 + }, + { + "epoch": 2.707512162942195, + "grad_norm": 2.0265097618103027, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.7661304473876953, + "num_tokens": 876996006.0, + "step": 32834 + }, + { + "epoch": 2.7075946235672466, + "grad_norm": 1.9753094911575317, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7784813046455383, + "num_tokens": 877033052.0, + "step": 32835 + }, + { + "epoch": 2.707677084192298, + "grad_norm": 2.118852138519287, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.779904842376709, + "num_tokens": 877063210.0, + "step": 32836 + }, + { + "epoch": 2.7077595448173497, + "grad_norm": 2.023608922958374, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7823159694671631, + "num_tokens": 877095976.0, + "step": 32837 + }, + { + "epoch": 2.7078420054424015, + "grad_norm": 1.934134840965271, + "learning_rate": 1e-06, + "loss": 0.6775, + "mean_token_accuracy": 0.7811529636383057, + "num_tokens": 877132144.0, + "step": 32838 + }, + { + "epoch": 2.707924466067453, + "grad_norm": 2.1747982501983643, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7190068960189819, + "num_tokens": 877163030.0, + "step": 32839 + }, + { + "epoch": 2.708006926692504, + "grad_norm": 1.9959683418273926, + "learning_rate": 1e-06, + "loss": 0.6172, + "mean_token_accuracy": 0.7928746938705444, + "num_tokens": 877195404.0, + "step": 32840 + }, + { + "epoch": 2.708089387317556, + "grad_norm": 2.3456344604492188, + "learning_rate": 1e-06, + "loss": 0.5388, + "mean_token_accuracy": 0.8133442401885986, + "num_tokens": 877218871.0, + "step": 32841 + }, + { + "epoch": 2.7081718479426073, + "grad_norm": 2.2161481380462646, + "learning_rate": 1e-06, + "loss": 0.6113, + "mean_token_accuracy": 0.7892648577690125, + "num_tokens": 877243916.0, + "step": 32842 + }, + { + "epoch": 2.7082543085676587, + "grad_norm": 2.079615592956543, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.7866897583007812, + "num_tokens": 877273467.0, + "step": 32843 + }, + { + "epoch": 2.7083367691927105, + "grad_norm": 2.049330711364746, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7652192711830139, + "num_tokens": 877306684.0, + "step": 32844 + }, + { + "epoch": 2.7084192298177623, + "grad_norm": 2.3786733150482178, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7673115730285645, + "num_tokens": 877333418.0, + "step": 32845 + }, + { + "epoch": 2.7085016904428136, + "grad_norm": 1.7092050313949585, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7708754539489746, + "num_tokens": 877371003.0, + "step": 32846 + }, + { + "epoch": 2.708584151067865, + "grad_norm": 1.9599956274032593, + "learning_rate": 1e-06, + "loss": 0.6011, + "mean_token_accuracy": 0.7963634729385376, + "num_tokens": 877403312.0, + "step": 32847 + }, + { + "epoch": 2.708666611692917, + "grad_norm": 2.563505172729492, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7401372194290161, + "num_tokens": 877425915.0, + "step": 32848 + }, + { + "epoch": 2.708749072317968, + "grad_norm": 1.8666614294052124, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7781651020050049, + "num_tokens": 877458683.0, + "step": 32849 + }, + { + "epoch": 2.7088315329430195, + "grad_norm": 1.8570371866226196, + "learning_rate": 1e-06, + "loss": 0.6495, + "mean_token_accuracy": 0.778955340385437, + "num_tokens": 877498249.0, + "step": 32850 + }, + { + "epoch": 2.7089139935680713, + "grad_norm": 2.2077434062957764, + "learning_rate": 1e-06, + "loss": 0.6368, + "mean_token_accuracy": 0.7852854132652283, + "num_tokens": 877524496.0, + "step": 32851 + }, + { + "epoch": 2.7089964541931226, + "grad_norm": 2.1396331787109375, + "learning_rate": 1e-06, + "loss": 0.5638, + "mean_token_accuracy": 0.810723602771759, + "num_tokens": 877553093.0, + "step": 32852 + }, + { + "epoch": 2.7090789148181744, + "grad_norm": 2.3306684494018555, + "learning_rate": 1e-06, + "loss": 0.5819, + "mean_token_accuracy": 0.8057676553726196, + "num_tokens": 877577997.0, + "step": 32853 + }, + { + "epoch": 2.7091613754432258, + "grad_norm": 2.1738994121551514, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.744976282119751, + "num_tokens": 877606379.0, + "step": 32854 + }, + { + "epoch": 2.7092438360682776, + "grad_norm": 2.1758036613464355, + "learning_rate": 1e-06, + "loss": 0.7727, + "mean_token_accuracy": 0.75447016954422, + "num_tokens": 877636443.0, + "step": 32855 + }, + { + "epoch": 2.709326296693329, + "grad_norm": 2.178830862045288, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7512067556381226, + "num_tokens": 877665499.0, + "step": 32856 + }, + { + "epoch": 2.7094087573183803, + "grad_norm": 2.1480231285095215, + "learning_rate": 1e-06, + "loss": 0.5834, + "mean_token_accuracy": 0.8068114519119263, + "num_tokens": 877695713.0, + "step": 32857 + }, + { + "epoch": 2.709491217943432, + "grad_norm": 2.194087028503418, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7330517172813416, + "num_tokens": 877724101.0, + "step": 32858 + }, + { + "epoch": 2.7095736785684834, + "grad_norm": 2.339221715927124, + "learning_rate": 1e-06, + "loss": 0.7578, + "mean_token_accuracy": 0.7608272433280945, + "num_tokens": 877750970.0, + "step": 32859 + }, + { + "epoch": 2.709656139193535, + "grad_norm": 1.8644636869430542, + "learning_rate": 1e-06, + "loss": 0.5564, + "mean_token_accuracy": 0.8104121685028076, + "num_tokens": 877781940.0, + "step": 32860 + }, + { + "epoch": 2.7097385998185866, + "grad_norm": 1.776879072189331, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.7881459593772888, + "num_tokens": 877820342.0, + "step": 32861 + }, + { + "epoch": 2.7098210604436384, + "grad_norm": 2.343531608581543, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7763769030570984, + "num_tokens": 877844685.0, + "step": 32862 + }, + { + "epoch": 2.7099035210686897, + "grad_norm": 2.5628931522369385, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7575371861457825, + "num_tokens": 877867139.0, + "step": 32863 + }, + { + "epoch": 2.709985981693741, + "grad_norm": 2.2253129482269287, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7765906453132629, + "num_tokens": 877901262.0, + "step": 32864 + }, + { + "epoch": 2.710068442318793, + "grad_norm": 2.009382486343384, + "learning_rate": 1e-06, + "loss": 0.5906, + "mean_token_accuracy": 0.8016740083694458, + "num_tokens": 877933239.0, + "step": 32865 + }, + { + "epoch": 2.710150902943844, + "grad_norm": 2.4971959590911865, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7627604603767395, + "num_tokens": 877954865.0, + "step": 32866 + }, + { + "epoch": 2.710233363568896, + "grad_norm": 1.7810882329940796, + "learning_rate": 1e-06, + "loss": 0.5245, + "mean_token_accuracy": 0.8214863538742065, + "num_tokens": 877992584.0, + "step": 32867 + }, + { + "epoch": 2.7103158241939473, + "grad_norm": 1.9331860542297363, + "learning_rate": 1e-06, + "loss": 0.6362, + "mean_token_accuracy": 0.7997417449951172, + "num_tokens": 878024402.0, + "step": 32868 + }, + { + "epoch": 2.710398284818999, + "grad_norm": 2.190609931945801, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.759260892868042, + "num_tokens": 878054291.0, + "step": 32869 + }, + { + "epoch": 2.7104807454440505, + "grad_norm": 2.10066819190979, + "learning_rate": 1e-06, + "loss": 0.5971, + "mean_token_accuracy": 0.8064122200012207, + "num_tokens": 878080549.0, + "step": 32870 + }, + { + "epoch": 2.710563206069102, + "grad_norm": 2.139352798461914, + "learning_rate": 1e-06, + "loss": 0.776, + "mean_token_accuracy": 0.7617406249046326, + "num_tokens": 878111369.0, + "step": 32871 + }, + { + "epoch": 2.7106456666941536, + "grad_norm": 1.9136550426483154, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7888718843460083, + "num_tokens": 878145142.0, + "step": 32872 + }, + { + "epoch": 2.710728127319205, + "grad_norm": 2.1129612922668457, + "learning_rate": 1e-06, + "loss": 0.5986, + "mean_token_accuracy": 0.7945811748504639, + "num_tokens": 878172332.0, + "step": 32873 + }, + { + "epoch": 2.7108105879442568, + "grad_norm": 1.7912962436676025, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7716900706291199, + "num_tokens": 878207874.0, + "step": 32874 + }, + { + "epoch": 2.710893048569308, + "grad_norm": 2.045189142227173, + "learning_rate": 1e-06, + "loss": 0.6088, + "mean_token_accuracy": 0.8018073439598083, + "num_tokens": 878237896.0, + "step": 32875 + }, + { + "epoch": 2.71097550919436, + "grad_norm": 2.070712089538574, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7451976537704468, + "num_tokens": 878271836.0, + "step": 32876 + }, + { + "epoch": 2.7110579698194113, + "grad_norm": 2.068866491317749, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.7920812368392944, + "num_tokens": 878300160.0, + "step": 32877 + }, + { + "epoch": 2.7111404304444626, + "grad_norm": 2.13983416557312, + "learning_rate": 1e-06, + "loss": 0.7615, + "mean_token_accuracy": 0.754787802696228, + "num_tokens": 878332163.0, + "step": 32878 + }, + { + "epoch": 2.7112228910695144, + "grad_norm": 2.334352970123291, + "learning_rate": 1e-06, + "loss": 0.7721, + "mean_token_accuracy": 0.7499406337738037, + "num_tokens": 878359482.0, + "step": 32879 + }, + { + "epoch": 2.7113053516945658, + "grad_norm": 1.8771814107894897, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7752786874771118, + "num_tokens": 878394433.0, + "step": 32880 + }, + { + "epoch": 2.711387812319617, + "grad_norm": 2.0016567707061768, + "learning_rate": 1e-06, + "loss": 0.5958, + "mean_token_accuracy": 0.7981984615325928, + "num_tokens": 878423521.0, + "step": 32881 + }, + { + "epoch": 2.711470272944669, + "grad_norm": 2.20105242729187, + "learning_rate": 1e-06, + "loss": 0.6108, + "mean_token_accuracy": 0.7998225688934326, + "num_tokens": 878450256.0, + "step": 32882 + }, + { + "epoch": 2.7115527335697207, + "grad_norm": 2.0441293716430664, + "learning_rate": 1e-06, + "loss": 0.5899, + "mean_token_accuracy": 0.8025654554367065, + "num_tokens": 878480442.0, + "step": 32883 + }, + { + "epoch": 2.711635194194772, + "grad_norm": 2.3175225257873535, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7186421155929565, + "num_tokens": 878506810.0, + "step": 32884 + }, + { + "epoch": 2.7117176548198234, + "grad_norm": 1.9791979789733887, + "learning_rate": 1e-06, + "loss": 0.5818, + "mean_token_accuracy": 0.8031246066093445, + "num_tokens": 878536397.0, + "step": 32885 + }, + { + "epoch": 2.711800115444875, + "grad_norm": 2.1751365661621094, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7744759321212769, + "num_tokens": 878564057.0, + "step": 32886 + }, + { + "epoch": 2.7118825760699266, + "grad_norm": 2.1353354454040527, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7498325109481812, + "num_tokens": 878594776.0, + "step": 32887 + }, + { + "epoch": 2.711965036694978, + "grad_norm": 2.5367870330810547, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7540000081062317, + "num_tokens": 878619897.0, + "step": 32888 + }, + { + "epoch": 2.7120474973200297, + "grad_norm": 1.8945366144180298, + "learning_rate": 1e-06, + "loss": 0.6503, + "mean_token_accuracy": 0.7768524885177612, + "num_tokens": 878653493.0, + "step": 32889 + }, + { + "epoch": 2.7121299579450815, + "grad_norm": 1.9775710105895996, + "learning_rate": 1e-06, + "loss": 0.4904, + "mean_token_accuracy": 0.8257028460502625, + "num_tokens": 878683636.0, + "step": 32890 + }, + { + "epoch": 2.712212418570133, + "grad_norm": 2.5316808223724365, + "learning_rate": 1e-06, + "loss": 0.724, + "mean_token_accuracy": 0.76900315284729, + "num_tokens": 878706698.0, + "step": 32891 + }, + { + "epoch": 2.712294879195184, + "grad_norm": 2.1063263416290283, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7428632974624634, + "num_tokens": 878738096.0, + "step": 32892 + }, + { + "epoch": 2.712377339820236, + "grad_norm": 2.2788848876953125, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.742545485496521, + "num_tokens": 878765371.0, + "step": 32893 + }, + { + "epoch": 2.7124598004452873, + "grad_norm": 1.9504237174987793, + "learning_rate": 1e-06, + "loss": 0.6055, + "mean_token_accuracy": 0.8047065138816833, + "num_tokens": 878793700.0, + "step": 32894 + }, + { + "epoch": 2.7125422610703387, + "grad_norm": 2.2497658729553223, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7611255049705505, + "num_tokens": 878821432.0, + "step": 32895 + }, + { + "epoch": 2.7126247216953905, + "grad_norm": 1.9373350143432617, + "learning_rate": 1e-06, + "loss": 0.5668, + "mean_token_accuracy": 0.8078534603118896, + "num_tokens": 878856373.0, + "step": 32896 + }, + { + "epoch": 2.712707182320442, + "grad_norm": 2.135324478149414, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7827887535095215, + "num_tokens": 878886242.0, + "step": 32897 + }, + { + "epoch": 2.7127896429454936, + "grad_norm": 3.510814666748047, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7751997113227844, + "num_tokens": 878910897.0, + "step": 32898 + }, + { + "epoch": 2.712872103570545, + "grad_norm": 1.9771678447723389, + "learning_rate": 1e-06, + "loss": 0.6406, + "mean_token_accuracy": 0.7986211776733398, + "num_tokens": 878942253.0, + "step": 32899 + }, + { + "epoch": 2.7129545641955968, + "grad_norm": 2.3229990005493164, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7542738318443298, + "num_tokens": 878968082.0, + "step": 32900 + }, + { + "epoch": 2.713037024820648, + "grad_norm": 1.9423773288726807, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7582534551620483, + "num_tokens": 879005526.0, + "step": 32901 + }, + { + "epoch": 2.7131194854456995, + "grad_norm": 2.151296377182007, + "learning_rate": 1e-06, + "loss": 0.5673, + "mean_token_accuracy": 0.7988910675048828, + "num_tokens": 879030805.0, + "step": 32902 + }, + { + "epoch": 2.7132019460707513, + "grad_norm": 2.3643300533294678, + "learning_rate": 1e-06, + "loss": 0.5568, + "mean_token_accuracy": 0.8007065057754517, + "num_tokens": 879052956.0, + "step": 32903 + }, + { + "epoch": 2.7132844066958026, + "grad_norm": 2.194143533706665, + "learning_rate": 1e-06, + "loss": 0.6173, + "mean_token_accuracy": 0.7862455248832703, + "num_tokens": 879082812.0, + "step": 32904 + }, + { + "epoch": 2.7133668673208544, + "grad_norm": 2.423778533935547, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.7864472270011902, + "num_tokens": 879105476.0, + "step": 32905 + }, + { + "epoch": 2.7134493279459058, + "grad_norm": 2.388566732406616, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7483260631561279, + "num_tokens": 879131301.0, + "step": 32906 + }, + { + "epoch": 2.7135317885709576, + "grad_norm": 1.8930131196975708, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.7878732681274414, + "num_tokens": 879168512.0, + "step": 32907 + }, + { + "epoch": 2.713614249196009, + "grad_norm": 2.094693422317505, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7445436120033264, + "num_tokens": 879196238.0, + "step": 32908 + }, + { + "epoch": 2.7136967098210603, + "grad_norm": 1.86081063747406, + "learning_rate": 1e-06, + "loss": 0.5587, + "mean_token_accuracy": 0.8089964389801025, + "num_tokens": 879230608.0, + "step": 32909 + }, + { + "epoch": 2.713779170446112, + "grad_norm": 2.1408774852752686, + "learning_rate": 1e-06, + "loss": 0.6158, + "mean_token_accuracy": 0.7914578318595886, + "num_tokens": 879258767.0, + "step": 32910 + }, + { + "epoch": 2.7138616310711634, + "grad_norm": 2.1972496509552, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.7783783674240112, + "num_tokens": 879288325.0, + "step": 32911 + }, + { + "epoch": 2.713944091696215, + "grad_norm": 2.1930625438690186, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7367234230041504, + "num_tokens": 879318702.0, + "step": 32912 + }, + { + "epoch": 2.7140265523212665, + "grad_norm": 2.0706779956817627, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.785794734954834, + "num_tokens": 879350168.0, + "step": 32913 + }, + { + "epoch": 2.7141090129463183, + "grad_norm": 2.3818628787994385, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7376366853713989, + "num_tokens": 879374910.0, + "step": 32914 + }, + { + "epoch": 2.7141914735713697, + "grad_norm": 2.3994646072387695, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7274489402770996, + "num_tokens": 879401232.0, + "step": 32915 + }, + { + "epoch": 2.714273934196421, + "grad_norm": 2.023017406463623, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7674387693405151, + "num_tokens": 879432378.0, + "step": 32916 + }, + { + "epoch": 2.714356394821473, + "grad_norm": 2.1471705436706543, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7648850083351135, + "num_tokens": 879459022.0, + "step": 32917 + }, + { + "epoch": 2.714438855446524, + "grad_norm": 2.1994054317474365, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7210134267807007, + "num_tokens": 879489470.0, + "step": 32918 + }, + { + "epoch": 2.714521316071576, + "grad_norm": 2.014782428741455, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7801766991615295, + "num_tokens": 879522609.0, + "step": 32919 + }, + { + "epoch": 2.7146037766966273, + "grad_norm": 2.343613862991333, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7553389668464661, + "num_tokens": 879549334.0, + "step": 32920 + }, + { + "epoch": 2.714686237321679, + "grad_norm": 1.9051270484924316, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.7958604693412781, + "num_tokens": 879583897.0, + "step": 32921 + }, + { + "epoch": 2.7147686979467305, + "grad_norm": 1.9117372035980225, + "learning_rate": 1e-06, + "loss": 0.7354, + "mean_token_accuracy": 0.7739676237106323, + "num_tokens": 879618759.0, + "step": 32922 + }, + { + "epoch": 2.714851158571782, + "grad_norm": 2.427375316619873, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7711395621299744, + "num_tokens": 879646157.0, + "step": 32923 + }, + { + "epoch": 2.7149336191968336, + "grad_norm": 1.8317006826400757, + "learning_rate": 1e-06, + "loss": 0.6294, + "mean_token_accuracy": 0.8011054396629333, + "num_tokens": 879682509.0, + "step": 32924 + }, + { + "epoch": 2.715016079821885, + "grad_norm": 1.8626749515533447, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.7732053995132446, + "num_tokens": 879718534.0, + "step": 32925 + }, + { + "epoch": 2.7150985404469363, + "grad_norm": 2.325181245803833, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7661877274513245, + "num_tokens": 879744966.0, + "step": 32926 + }, + { + "epoch": 2.715181001071988, + "grad_norm": 2.1123363971710205, + "learning_rate": 1e-06, + "loss": 0.5701, + "mean_token_accuracy": 0.8129599094390869, + "num_tokens": 879773926.0, + "step": 32927 + }, + { + "epoch": 2.71526346169704, + "grad_norm": 1.8577278852462769, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.7947726249694824, + "num_tokens": 879807493.0, + "step": 32928 + }, + { + "epoch": 2.7153459223220913, + "grad_norm": 1.9829072952270508, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7916144728660583, + "num_tokens": 879839477.0, + "step": 32929 + }, + { + "epoch": 2.7154283829471426, + "grad_norm": 2.150155544281006, + "learning_rate": 1e-06, + "loss": 0.5639, + "mean_token_accuracy": 0.8094011545181274, + "num_tokens": 879866668.0, + "step": 32930 + }, + { + "epoch": 2.7155108435721944, + "grad_norm": 2.4362995624542236, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7649929523468018, + "num_tokens": 879888925.0, + "step": 32931 + }, + { + "epoch": 2.7155933041972458, + "grad_norm": 2.0650136470794678, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7753636240959167, + "num_tokens": 879920245.0, + "step": 32932 + }, + { + "epoch": 2.715675764822297, + "grad_norm": 2.1512506008148193, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7753292322158813, + "num_tokens": 879949132.0, + "step": 32933 + }, + { + "epoch": 2.715758225447349, + "grad_norm": 2.1528708934783936, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.8022764325141907, + "num_tokens": 879977048.0, + "step": 32934 + }, + { + "epoch": 2.7158406860724007, + "grad_norm": 2.1076719760894775, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.76815265417099, + "num_tokens": 880006683.0, + "step": 32935 + }, + { + "epoch": 2.715923146697452, + "grad_norm": 2.406428337097168, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7745687961578369, + "num_tokens": 880031043.0, + "step": 32936 + }, + { + "epoch": 2.7160056073225034, + "grad_norm": 2.331498622894287, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7601001858711243, + "num_tokens": 880057531.0, + "step": 32937 + }, + { + "epoch": 2.716088067947555, + "grad_norm": 2.008121967315674, + "learning_rate": 1e-06, + "loss": 0.5949, + "mean_token_accuracy": 0.792477548122406, + "num_tokens": 880085330.0, + "step": 32938 + }, + { + "epoch": 2.7161705285726065, + "grad_norm": 1.9286073446273804, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7700215578079224, + "num_tokens": 880120695.0, + "step": 32939 + }, + { + "epoch": 2.716252989197658, + "grad_norm": 1.8154197931289673, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.7997984886169434, + "num_tokens": 880157269.0, + "step": 32940 + }, + { + "epoch": 2.7163354498227097, + "grad_norm": 2.175715923309326, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7448526620864868, + "num_tokens": 880185960.0, + "step": 32941 + }, + { + "epoch": 2.716417910447761, + "grad_norm": 1.7093517780303955, + "learning_rate": 1e-06, + "loss": 0.5363, + "mean_token_accuracy": 0.8207052946090698, + "num_tokens": 880225380.0, + "step": 32942 + }, + { + "epoch": 2.716500371072813, + "grad_norm": 2.1197280883789062, + "learning_rate": 1e-06, + "loss": 0.5427, + "mean_token_accuracy": 0.8090633153915405, + "num_tokens": 880255049.0, + "step": 32943 + }, + { + "epoch": 2.716582831697864, + "grad_norm": 2.15462064743042, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.78269362449646, + "num_tokens": 880280959.0, + "step": 32944 + }, + { + "epoch": 2.716665292322916, + "grad_norm": 2.4243273735046387, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7334702014923096, + "num_tokens": 880304887.0, + "step": 32945 + }, + { + "epoch": 2.7167477529479673, + "grad_norm": 2.2367634773254395, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.775145947933197, + "num_tokens": 880333344.0, + "step": 32946 + }, + { + "epoch": 2.7168302135730187, + "grad_norm": 2.0460944175720215, + "learning_rate": 1e-06, + "loss": 0.7407, + "mean_token_accuracy": 0.7551541328430176, + "num_tokens": 880363985.0, + "step": 32947 + }, + { + "epoch": 2.7169126741980705, + "grad_norm": 2.2360246181488037, + "learning_rate": 1e-06, + "loss": 0.8116, + "mean_token_accuracy": 0.7556014060974121, + "num_tokens": 880392505.0, + "step": 32948 + }, + { + "epoch": 2.716995134823122, + "grad_norm": 2.3286612033843994, + "learning_rate": 1e-06, + "loss": 0.6165, + "mean_token_accuracy": 0.7967638373374939, + "num_tokens": 880416574.0, + "step": 32949 + }, + { + "epoch": 2.7170775954481736, + "grad_norm": 2.096783399581909, + "learning_rate": 1e-06, + "loss": 0.6048, + "mean_token_accuracy": 0.7947962284088135, + "num_tokens": 880446868.0, + "step": 32950 + }, + { + "epoch": 2.717160056073225, + "grad_norm": 2.1959569454193115, + "learning_rate": 1e-06, + "loss": 0.725, + "mean_token_accuracy": 0.7575221061706543, + "num_tokens": 880474428.0, + "step": 32951 + }, + { + "epoch": 2.7172425166982768, + "grad_norm": 2.367504358291626, + "learning_rate": 1e-06, + "loss": 0.6482, + "mean_token_accuracy": 0.7849032878875732, + "num_tokens": 880500635.0, + "step": 32952 + }, + { + "epoch": 2.717324977323328, + "grad_norm": 2.300919532775879, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.769318699836731, + "num_tokens": 880524995.0, + "step": 32953 + }, + { + "epoch": 2.7174074379483795, + "grad_norm": 2.004746437072754, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.789656400680542, + "num_tokens": 880557812.0, + "step": 32954 + }, + { + "epoch": 2.7174898985734313, + "grad_norm": 2.0376386642456055, + "learning_rate": 1e-06, + "loss": 0.7385, + "mean_token_accuracy": 0.7617819905281067, + "num_tokens": 880590073.0, + "step": 32955 + }, + { + "epoch": 2.7175723591984826, + "grad_norm": 2.1453537940979004, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7658147811889648, + "num_tokens": 880619103.0, + "step": 32956 + }, + { + "epoch": 2.7176548198235344, + "grad_norm": 1.9771373271942139, + "learning_rate": 1e-06, + "loss": 0.5574, + "mean_token_accuracy": 0.8164136409759521, + "num_tokens": 880651544.0, + "step": 32957 + }, + { + "epoch": 2.7177372804485858, + "grad_norm": 2.0012412071228027, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7683661580085754, + "num_tokens": 880685729.0, + "step": 32958 + }, + { + "epoch": 2.7178197410736376, + "grad_norm": 2.274684190750122, + "learning_rate": 1e-06, + "loss": 0.6202, + "mean_token_accuracy": 0.7935916185379028, + "num_tokens": 880711082.0, + "step": 32959 + }, + { + "epoch": 2.717902201698689, + "grad_norm": 2.1496801376342773, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7755371332168579, + "num_tokens": 880739155.0, + "step": 32960 + }, + { + "epoch": 2.7179846623237403, + "grad_norm": 1.981845498085022, + "learning_rate": 1e-06, + "loss": 0.6694, + "mean_token_accuracy": 0.7751365900039673, + "num_tokens": 880771647.0, + "step": 32961 + }, + { + "epoch": 2.718067122948792, + "grad_norm": 2.090327024459839, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7314591407775879, + "num_tokens": 880804959.0, + "step": 32962 + }, + { + "epoch": 2.7181495835738434, + "grad_norm": 2.0039942264556885, + "learning_rate": 1e-06, + "loss": 0.6409, + "mean_token_accuracy": 0.7901061177253723, + "num_tokens": 880833951.0, + "step": 32963 + }, + { + "epoch": 2.718232044198895, + "grad_norm": 2.0399014949798584, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7273595929145813, + "num_tokens": 880867457.0, + "step": 32964 + }, + { + "epoch": 2.7183145048239465, + "grad_norm": 2.1054961681365967, + "learning_rate": 1e-06, + "loss": 0.6337, + "mean_token_accuracy": 0.7823922634124756, + "num_tokens": 880894856.0, + "step": 32965 + }, + { + "epoch": 2.7183969654489983, + "grad_norm": 2.2908496856689453, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7344224452972412, + "num_tokens": 880921374.0, + "step": 32966 + }, + { + "epoch": 2.7184794260740497, + "grad_norm": 2.381643295288086, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.736150860786438, + "num_tokens": 880946164.0, + "step": 32967 + }, + { + "epoch": 2.718561886699101, + "grad_norm": 2.2517457008361816, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7653151750564575, + "num_tokens": 880971113.0, + "step": 32968 + }, + { + "epoch": 2.718644347324153, + "grad_norm": 2.02872371673584, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7737754583358765, + "num_tokens": 881006126.0, + "step": 32969 + }, + { + "epoch": 2.718726807949204, + "grad_norm": 1.9868108034133911, + "learning_rate": 1e-06, + "loss": 0.7109, + "mean_token_accuracy": 0.7727266550064087, + "num_tokens": 881038325.0, + "step": 32970 + }, + { + "epoch": 2.7188092685742555, + "grad_norm": 1.745818853378296, + "learning_rate": 1e-06, + "loss": 0.5389, + "mean_token_accuracy": 0.821068286895752, + "num_tokens": 881073816.0, + "step": 32971 + }, + { + "epoch": 2.7188917291993073, + "grad_norm": 2.1643731594085693, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.7600619792938232, + "num_tokens": 881103295.0, + "step": 32972 + }, + { + "epoch": 2.718974189824359, + "grad_norm": 2.278418779373169, + "learning_rate": 1e-06, + "loss": 0.7198, + "mean_token_accuracy": 0.7640812397003174, + "num_tokens": 881130200.0, + "step": 32973 + }, + { + "epoch": 2.7190566504494105, + "grad_norm": 2.1405041217803955, + "learning_rate": 1e-06, + "loss": 0.6584, + "mean_token_accuracy": 0.7780407667160034, + "num_tokens": 881161554.0, + "step": 32974 + }, + { + "epoch": 2.719139111074462, + "grad_norm": 1.9267498254776, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7822407484054565, + "num_tokens": 881193362.0, + "step": 32975 + }, + { + "epoch": 2.7192215716995136, + "grad_norm": 1.8792040348052979, + "learning_rate": 1e-06, + "loss": 0.5065, + "mean_token_accuracy": 0.8196961283683777, + "num_tokens": 881228994.0, + "step": 32976 + }, + { + "epoch": 2.719304032324565, + "grad_norm": 2.080204963684082, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7741512656211853, + "num_tokens": 881259168.0, + "step": 32977 + }, + { + "epoch": 2.7193864929496163, + "grad_norm": 1.7839114665985107, + "learning_rate": 1e-06, + "loss": 0.6696, + "mean_token_accuracy": 0.779601514339447, + "num_tokens": 881301622.0, + "step": 32978 + }, + { + "epoch": 2.719468953574668, + "grad_norm": 1.892417073249817, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.7933574318885803, + "num_tokens": 881335066.0, + "step": 32979 + }, + { + "epoch": 2.71955141419972, + "grad_norm": 2.001094341278076, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.749076783657074, + "num_tokens": 881366239.0, + "step": 32980 + }, + { + "epoch": 2.7196338748247713, + "grad_norm": 2.293977737426758, + "learning_rate": 1e-06, + "loss": 0.7026, + "mean_token_accuracy": 0.7520208358764648, + "num_tokens": 881390349.0, + "step": 32981 + }, + { + "epoch": 2.7197163354498226, + "grad_norm": 2.0536863803863525, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7724993824958801, + "num_tokens": 881422427.0, + "step": 32982 + }, + { + "epoch": 2.7197987960748744, + "grad_norm": 1.9330285787582397, + "learning_rate": 1e-06, + "loss": 0.589, + "mean_token_accuracy": 0.8038948774337769, + "num_tokens": 881453305.0, + "step": 32983 + }, + { + "epoch": 2.7198812566999258, + "grad_norm": 2.306007146835327, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7805202007293701, + "num_tokens": 881477653.0, + "step": 32984 + }, + { + "epoch": 2.719963717324977, + "grad_norm": 2.078500509262085, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.782421350479126, + "num_tokens": 881506502.0, + "step": 32985 + }, + { + "epoch": 2.720046177950029, + "grad_norm": 2.4404611587524414, + "learning_rate": 1e-06, + "loss": 0.7089, + "mean_token_accuracy": 0.7706645727157593, + "num_tokens": 881528937.0, + "step": 32986 + }, + { + "epoch": 2.7201286385750802, + "grad_norm": 2.0339531898498535, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.7897690534591675, + "num_tokens": 881558899.0, + "step": 32987 + }, + { + "epoch": 2.720211099200132, + "grad_norm": 2.287355899810791, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.7855437994003296, + "num_tokens": 881582731.0, + "step": 32988 + }, + { + "epoch": 2.7202935598251834, + "grad_norm": 2.1919710636138916, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7464099526405334, + "num_tokens": 881611912.0, + "step": 32989 + }, + { + "epoch": 2.720376020450235, + "grad_norm": 2.155487537384033, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7713236808776855, + "num_tokens": 881642598.0, + "step": 32990 + }, + { + "epoch": 2.7204584810752865, + "grad_norm": 2.1117358207702637, + "learning_rate": 1e-06, + "loss": 0.7415, + "mean_token_accuracy": 0.7596831321716309, + "num_tokens": 881673520.0, + "step": 32991 + }, + { + "epoch": 2.720540941700338, + "grad_norm": 2.2961585521698, + "learning_rate": 1e-06, + "loss": 0.6948, + "mean_token_accuracy": 0.7694897651672363, + "num_tokens": 881700187.0, + "step": 32992 + }, + { + "epoch": 2.7206234023253897, + "grad_norm": 2.2355823516845703, + "learning_rate": 1e-06, + "loss": 0.7995, + "mean_token_accuracy": 0.7433710098266602, + "num_tokens": 881732936.0, + "step": 32993 + }, + { + "epoch": 2.720705862950441, + "grad_norm": 1.8815982341766357, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7692444324493408, + "num_tokens": 881771298.0, + "step": 32994 + }, + { + "epoch": 2.720788323575493, + "grad_norm": 2.298063278198242, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7500253915786743, + "num_tokens": 881796352.0, + "step": 32995 + }, + { + "epoch": 2.720870784200544, + "grad_norm": 2.2200374603271484, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7343219518661499, + "num_tokens": 881823676.0, + "step": 32996 + }, + { + "epoch": 2.720953244825596, + "grad_norm": 2.047696828842163, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7669576406478882, + "num_tokens": 881853541.0, + "step": 32997 + }, + { + "epoch": 2.7210357054506473, + "grad_norm": 2.0552103519439697, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.7838407754898071, + "num_tokens": 881886969.0, + "step": 32998 + }, + { + "epoch": 2.7211181660756987, + "grad_norm": 2.0111093521118164, + "learning_rate": 1e-06, + "loss": 0.6275, + "mean_token_accuracy": 0.8053699731826782, + "num_tokens": 881921758.0, + "step": 32999 + }, + { + "epoch": 2.7212006267007505, + "grad_norm": 2.0850014686584473, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.773869514465332, + "num_tokens": 881954071.0, + "step": 33000 + }, + { + "epoch": 2.721283087325802, + "grad_norm": 1.8293263912200928, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7727225422859192, + "num_tokens": 881991816.0, + "step": 33001 + }, + { + "epoch": 2.7213655479508536, + "grad_norm": 2.399981737136841, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7528716921806335, + "num_tokens": 882015617.0, + "step": 33002 + }, + { + "epoch": 2.721448008575905, + "grad_norm": 2.336107015609741, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7584110498428345, + "num_tokens": 882040025.0, + "step": 33003 + }, + { + "epoch": 2.7215304692009568, + "grad_norm": 2.0387730598449707, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7752330303192139, + "num_tokens": 882073735.0, + "step": 33004 + }, + { + "epoch": 2.721612929826008, + "grad_norm": 2.16046404838562, + "learning_rate": 1e-06, + "loss": 0.8053, + "mean_token_accuracy": 0.7437095642089844, + "num_tokens": 882101364.0, + "step": 33005 + }, + { + "epoch": 2.7216953904510595, + "grad_norm": 2.082367181777954, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7480948567390442, + "num_tokens": 882130055.0, + "step": 33006 + }, + { + "epoch": 2.7217778510761113, + "grad_norm": 2.0068812370300293, + "learning_rate": 1e-06, + "loss": 0.7495, + "mean_token_accuracy": 0.7603467702865601, + "num_tokens": 882161830.0, + "step": 33007 + }, + { + "epoch": 2.7218603117011626, + "grad_norm": 2.6105334758758545, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7717040777206421, + "num_tokens": 882183105.0, + "step": 33008 + }, + { + "epoch": 2.7219427723262144, + "grad_norm": 2.412065029144287, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7650573253631592, + "num_tokens": 882211082.0, + "step": 33009 + }, + { + "epoch": 2.7220252329512657, + "grad_norm": 2.0194718837738037, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7503738403320312, + "num_tokens": 882241421.0, + "step": 33010 + }, + { + "epoch": 2.7221076935763175, + "grad_norm": 2.189085006713867, + "learning_rate": 1e-06, + "loss": 0.5727, + "mean_token_accuracy": 0.8018090724945068, + "num_tokens": 882268268.0, + "step": 33011 + }, + { + "epoch": 2.722190154201369, + "grad_norm": 1.9356865882873535, + "learning_rate": 1e-06, + "loss": 0.5486, + "mean_token_accuracy": 0.8026459217071533, + "num_tokens": 882299062.0, + "step": 33012 + }, + { + "epoch": 2.7222726148264202, + "grad_norm": 2.0258359909057617, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7486384510993958, + "num_tokens": 882333384.0, + "step": 33013 + }, + { + "epoch": 2.722355075451472, + "grad_norm": 2.0744125843048096, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.8021109700202942, + "num_tokens": 882362374.0, + "step": 33014 + }, + { + "epoch": 2.7224375360765234, + "grad_norm": 2.663827657699585, + "learning_rate": 1e-06, + "loss": 0.7741, + "mean_token_accuracy": 0.7580689191818237, + "num_tokens": 882382133.0, + "step": 33015 + }, + { + "epoch": 2.7225199967015747, + "grad_norm": 2.0120840072631836, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7705742716789246, + "num_tokens": 882416001.0, + "step": 33016 + }, + { + "epoch": 2.7226024573266265, + "grad_norm": 2.096571922302246, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.7936463952064514, + "num_tokens": 882447205.0, + "step": 33017 + }, + { + "epoch": 2.7226849179516783, + "grad_norm": 1.9339914321899414, + "learning_rate": 1e-06, + "loss": 0.6126, + "mean_token_accuracy": 0.796287477016449, + "num_tokens": 882483148.0, + "step": 33018 + }, + { + "epoch": 2.7227673785767297, + "grad_norm": 2.133899450302124, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7638419270515442, + "num_tokens": 882517958.0, + "step": 33019 + }, + { + "epoch": 2.722849839201781, + "grad_norm": 2.17415189743042, + "learning_rate": 1e-06, + "loss": 0.5938, + "mean_token_accuracy": 0.8022162914276123, + "num_tokens": 882547315.0, + "step": 33020 + }, + { + "epoch": 2.722932299826833, + "grad_norm": 2.1169981956481934, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7694034576416016, + "num_tokens": 882577968.0, + "step": 33021 + }, + { + "epoch": 2.723014760451884, + "grad_norm": 2.2837917804718018, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7291801571846008, + "num_tokens": 882606112.0, + "step": 33022 + }, + { + "epoch": 2.7230972210769355, + "grad_norm": 1.9716546535491943, + "learning_rate": 1e-06, + "loss": 0.6293, + "mean_token_accuracy": 0.8000558614730835, + "num_tokens": 882638324.0, + "step": 33023 + }, + { + "epoch": 2.7231796817019873, + "grad_norm": 2.0774989128112793, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.7919191122055054, + "num_tokens": 882669262.0, + "step": 33024 + }, + { + "epoch": 2.723262142327039, + "grad_norm": 1.9590004682540894, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7689791917800903, + "num_tokens": 882699219.0, + "step": 33025 + }, + { + "epoch": 2.7233446029520905, + "grad_norm": 2.2902989387512207, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7327470183372498, + "num_tokens": 882728926.0, + "step": 33026 + }, + { + "epoch": 2.723427063577142, + "grad_norm": 2.2726187705993652, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7397357225418091, + "num_tokens": 882754016.0, + "step": 33027 + }, + { + "epoch": 2.7235095242021936, + "grad_norm": 2.1299288272857666, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.7988232374191284, + "num_tokens": 882785784.0, + "step": 33028 + }, + { + "epoch": 2.723591984827245, + "grad_norm": 2.117373466491699, + "learning_rate": 1e-06, + "loss": 0.7585, + "mean_token_accuracy": 0.7538570165634155, + "num_tokens": 882816018.0, + "step": 33029 + }, + { + "epoch": 2.7236744454522963, + "grad_norm": 2.1433825492858887, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7666651606559753, + "num_tokens": 882846771.0, + "step": 33030 + }, + { + "epoch": 2.723756906077348, + "grad_norm": 1.985941767692566, + "learning_rate": 1e-06, + "loss": 0.5933, + "mean_token_accuracy": 0.8086403012275696, + "num_tokens": 882875747.0, + "step": 33031 + }, + { + "epoch": 2.7238393667023995, + "grad_norm": 2.6273562908172607, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7611210346221924, + "num_tokens": 882896889.0, + "step": 33032 + }, + { + "epoch": 2.7239218273274513, + "grad_norm": 2.0964572429656982, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.7829710245132446, + "num_tokens": 882927361.0, + "step": 33033 + }, + { + "epoch": 2.7240042879525026, + "grad_norm": 2.098858594894409, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7457113265991211, + "num_tokens": 882958865.0, + "step": 33034 + }, + { + "epoch": 2.7240867485775544, + "grad_norm": 2.2464396953582764, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7750499248504639, + "num_tokens": 882986163.0, + "step": 33035 + }, + { + "epoch": 2.7241692092026057, + "grad_norm": 2.395293951034546, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7467695474624634, + "num_tokens": 883010641.0, + "step": 33036 + }, + { + "epoch": 2.724251669827657, + "grad_norm": 2.045865058898926, + "learning_rate": 1e-06, + "loss": 0.5856, + "mean_token_accuracy": 0.8062783479690552, + "num_tokens": 883040992.0, + "step": 33037 + }, + { + "epoch": 2.724334130452709, + "grad_norm": 2.1750853061676025, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.7711416482925415, + "num_tokens": 883069445.0, + "step": 33038 + }, + { + "epoch": 2.7244165910777602, + "grad_norm": 2.120845079421997, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.779349684715271, + "num_tokens": 883099945.0, + "step": 33039 + }, + { + "epoch": 2.724499051702812, + "grad_norm": 1.7666022777557373, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.7847069501876831, + "num_tokens": 883140304.0, + "step": 33040 + }, + { + "epoch": 2.7245815123278634, + "grad_norm": 1.9890860319137573, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.7895649671554565, + "num_tokens": 883174831.0, + "step": 33041 + }, + { + "epoch": 2.724663972952915, + "grad_norm": 2.178734302520752, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.78408282995224, + "num_tokens": 883203388.0, + "step": 33042 + }, + { + "epoch": 2.7247464335779665, + "grad_norm": 1.975414752960205, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7545980215072632, + "num_tokens": 883235478.0, + "step": 33043 + }, + { + "epoch": 2.724828894203018, + "grad_norm": 2.18937611579895, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7731078863143921, + "num_tokens": 883262251.0, + "step": 33044 + }, + { + "epoch": 2.7249113548280697, + "grad_norm": 2.2508389949798584, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7456212043762207, + "num_tokens": 883291432.0, + "step": 33045 + }, + { + "epoch": 2.724993815453121, + "grad_norm": 2.0540504455566406, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7725902795791626, + "num_tokens": 883323672.0, + "step": 33046 + }, + { + "epoch": 2.725076276078173, + "grad_norm": 2.2607247829437256, + "learning_rate": 1e-06, + "loss": 0.7867, + "mean_token_accuracy": 0.7458012700080872, + "num_tokens": 883352739.0, + "step": 33047 + }, + { + "epoch": 2.725158736703224, + "grad_norm": 1.7307790517807007, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.7950218915939331, + "num_tokens": 883393555.0, + "step": 33048 + }, + { + "epoch": 2.725241197328276, + "grad_norm": 2.0466041564941406, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7665746212005615, + "num_tokens": 883424933.0, + "step": 33049 + }, + { + "epoch": 2.7253236579533273, + "grad_norm": 2.0535004138946533, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7884189486503601, + "num_tokens": 883454427.0, + "step": 33050 + }, + { + "epoch": 2.7254061185783787, + "grad_norm": 1.9656373262405396, + "learning_rate": 1e-06, + "loss": 0.6027, + "mean_token_accuracy": 0.802757203578949, + "num_tokens": 883486077.0, + "step": 33051 + }, + { + "epoch": 2.7254885792034305, + "grad_norm": 2.0048961639404297, + "learning_rate": 1e-06, + "loss": 0.5603, + "mean_token_accuracy": 0.8129961490631104, + "num_tokens": 883515812.0, + "step": 33052 + }, + { + "epoch": 2.725571039828482, + "grad_norm": 1.8807569742202759, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.7799737453460693, + "num_tokens": 883551055.0, + "step": 33053 + }, + { + "epoch": 2.725653500453533, + "grad_norm": 2.3945672512054443, + "learning_rate": 1e-06, + "loss": 0.7799, + "mean_token_accuracy": 0.7493230104446411, + "num_tokens": 883574154.0, + "step": 33054 + }, + { + "epoch": 2.725735961078585, + "grad_norm": 1.806681513786316, + "learning_rate": 1e-06, + "loss": 0.6122, + "mean_token_accuracy": 0.8094725608825684, + "num_tokens": 883608546.0, + "step": 33055 + }, + { + "epoch": 2.7258184217036368, + "grad_norm": 2.097975730895996, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7294973134994507, + "num_tokens": 883639794.0, + "step": 33056 + }, + { + "epoch": 2.725900882328688, + "grad_norm": 1.9354299306869507, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.7903817296028137, + "num_tokens": 883676444.0, + "step": 33057 + }, + { + "epoch": 2.7259833429537395, + "grad_norm": 1.9571858644485474, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7716070413589478, + "num_tokens": 883710166.0, + "step": 33058 + }, + { + "epoch": 2.7260658035787912, + "grad_norm": 2.2430639266967773, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7826517820358276, + "num_tokens": 883737059.0, + "step": 33059 + }, + { + "epoch": 2.7261482642038426, + "grad_norm": 2.1793594360351562, + "learning_rate": 1e-06, + "loss": 0.4752, + "mean_token_accuracy": 0.8309056758880615, + "num_tokens": 883763693.0, + "step": 33060 + }, + { + "epoch": 2.726230724828894, + "grad_norm": 2.0647058486938477, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7607282996177673, + "num_tokens": 883794629.0, + "step": 33061 + }, + { + "epoch": 2.7263131854539457, + "grad_norm": 2.1264209747314453, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7772984504699707, + "num_tokens": 883826153.0, + "step": 33062 + }, + { + "epoch": 2.7263956460789975, + "grad_norm": 1.9455695152282715, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.7930009365081787, + "num_tokens": 883860890.0, + "step": 33063 + }, + { + "epoch": 2.726478106704049, + "grad_norm": 1.9260497093200684, + "learning_rate": 1e-06, + "loss": 0.6904, + "mean_token_accuracy": 0.7780717611312866, + "num_tokens": 883896203.0, + "step": 33064 + }, + { + "epoch": 2.7265605673291002, + "grad_norm": 2.1590707302093506, + "learning_rate": 1e-06, + "loss": 0.7225, + "mean_token_accuracy": 0.7571528553962708, + "num_tokens": 883925514.0, + "step": 33065 + }, + { + "epoch": 2.726643027954152, + "grad_norm": 2.1421420574188232, + "learning_rate": 1e-06, + "loss": 0.5714, + "mean_token_accuracy": 0.8028764724731445, + "num_tokens": 883952765.0, + "step": 33066 + }, + { + "epoch": 2.7267254885792034, + "grad_norm": 1.9811598062515259, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7760912179946899, + "num_tokens": 883983856.0, + "step": 33067 + }, + { + "epoch": 2.7268079492042547, + "grad_norm": 1.9063456058502197, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.7831529378890991, + "num_tokens": 884020335.0, + "step": 33068 + }, + { + "epoch": 2.7268904098293065, + "grad_norm": 2.472327947616577, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7637374401092529, + "num_tokens": 884044305.0, + "step": 33069 + }, + { + "epoch": 2.726972870454358, + "grad_norm": 2.074808120727539, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7823038101196289, + "num_tokens": 884074118.0, + "step": 33070 + }, + { + "epoch": 2.7270553310794097, + "grad_norm": 1.9033161401748657, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.7961331009864807, + "num_tokens": 884110049.0, + "step": 33071 + }, + { + "epoch": 2.727137791704461, + "grad_norm": 2.0909807682037354, + "learning_rate": 1e-06, + "loss": 0.4995, + "mean_token_accuracy": 0.8184400200843811, + "num_tokens": 884139257.0, + "step": 33072 + }, + { + "epoch": 2.727220252329513, + "grad_norm": 1.9350532293319702, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7765451669692993, + "num_tokens": 884173180.0, + "step": 33073 + }, + { + "epoch": 2.727302712954564, + "grad_norm": 2.1776368618011475, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.7839009761810303, + "num_tokens": 884202642.0, + "step": 33074 + }, + { + "epoch": 2.7273851735796155, + "grad_norm": 2.146843910217285, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7497591376304626, + "num_tokens": 884232330.0, + "step": 33075 + }, + { + "epoch": 2.7274676342046673, + "grad_norm": 2.1828136444091797, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7488663196563721, + "num_tokens": 884262896.0, + "step": 33076 + }, + { + "epoch": 2.7275500948297187, + "grad_norm": 1.971178412437439, + "learning_rate": 1e-06, + "loss": 0.6361, + "mean_token_accuracy": 0.7776057124137878, + "num_tokens": 884294346.0, + "step": 33077 + }, + { + "epoch": 2.7276325554547705, + "grad_norm": 2.5185434818267822, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.771663248538971, + "num_tokens": 884318358.0, + "step": 33078 + }, + { + "epoch": 2.727715016079822, + "grad_norm": 2.1364102363586426, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7793461084365845, + "num_tokens": 884346671.0, + "step": 33079 + }, + { + "epoch": 2.7277974767048736, + "grad_norm": 2.1482057571411133, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.7901699542999268, + "num_tokens": 884374853.0, + "step": 33080 + }, + { + "epoch": 2.727879937329925, + "grad_norm": 2.368380069732666, + "learning_rate": 1e-06, + "loss": 0.7614, + "mean_token_accuracy": 0.752819299697876, + "num_tokens": 884400658.0, + "step": 33081 + }, + { + "epoch": 2.7279623979549763, + "grad_norm": 1.9171563386917114, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.7870984077453613, + "num_tokens": 884434653.0, + "step": 33082 + }, + { + "epoch": 2.728044858580028, + "grad_norm": 2.0127456188201904, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7540667653083801, + "num_tokens": 884468341.0, + "step": 33083 + }, + { + "epoch": 2.7281273192050794, + "grad_norm": 2.1963326930999756, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7370092272758484, + "num_tokens": 884498114.0, + "step": 33084 + }, + { + "epoch": 2.7282097798301312, + "grad_norm": 1.7888801097869873, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7717211246490479, + "num_tokens": 884538150.0, + "step": 33085 + }, + { + "epoch": 2.7282922404551826, + "grad_norm": 2.430859327316284, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.7930439710617065, + "num_tokens": 884560486.0, + "step": 33086 + }, + { + "epoch": 2.7283747010802344, + "grad_norm": 2.224588394165039, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.777854859828949, + "num_tokens": 884586058.0, + "step": 33087 + }, + { + "epoch": 2.7284571617052857, + "grad_norm": 1.7596181631088257, + "learning_rate": 1e-06, + "loss": 0.4909, + "mean_token_accuracy": 0.832016110420227, + "num_tokens": 884622160.0, + "step": 33088 + }, + { + "epoch": 2.728539622330337, + "grad_norm": 2.108229398727417, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.7889472246170044, + "num_tokens": 884650789.0, + "step": 33089 + }, + { + "epoch": 2.728622082955389, + "grad_norm": 2.0812556743621826, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7668784856796265, + "num_tokens": 884681989.0, + "step": 33090 + }, + { + "epoch": 2.7287045435804402, + "grad_norm": 2.310163974761963, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7350857257843018, + "num_tokens": 884708311.0, + "step": 33091 + }, + { + "epoch": 2.728787004205492, + "grad_norm": 2.5960564613342285, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7691471576690674, + "num_tokens": 884727353.0, + "step": 33092 + }, + { + "epoch": 2.7288694648305434, + "grad_norm": 1.9967600107192993, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7989603281021118, + "num_tokens": 884757947.0, + "step": 33093 + }, + { + "epoch": 2.728951925455595, + "grad_norm": 2.158669948577881, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.7953476905822754, + "num_tokens": 884787865.0, + "step": 33094 + }, + { + "epoch": 2.7290343860806465, + "grad_norm": 2.24753999710083, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7543169856071472, + "num_tokens": 884814713.0, + "step": 33095 + }, + { + "epoch": 2.729116846705698, + "grad_norm": 2.2626349925994873, + "learning_rate": 1e-06, + "loss": 0.786, + "mean_token_accuracy": 0.7508887648582458, + "num_tokens": 884841047.0, + "step": 33096 + }, + { + "epoch": 2.7291993073307497, + "grad_norm": 1.9785091876983643, + "learning_rate": 1e-06, + "loss": 0.6053, + "mean_token_accuracy": 0.799892008304596, + "num_tokens": 884873342.0, + "step": 33097 + }, + { + "epoch": 2.729281767955801, + "grad_norm": 2.452888250350952, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7626585960388184, + "num_tokens": 884900652.0, + "step": 33098 + }, + { + "epoch": 2.7293642285808524, + "grad_norm": 2.0955188274383545, + "learning_rate": 1e-06, + "loss": 0.4432, + "mean_token_accuracy": 0.8409262895584106, + "num_tokens": 884928112.0, + "step": 33099 + }, + { + "epoch": 2.729446689205904, + "grad_norm": 2.180433750152588, + "learning_rate": 1e-06, + "loss": 0.5754, + "mean_token_accuracy": 0.7890930771827698, + "num_tokens": 884953333.0, + "step": 33100 + }, + { + "epoch": 2.729529149830956, + "grad_norm": 1.8911972045898438, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.775981068611145, + "num_tokens": 884987664.0, + "step": 33101 + }, + { + "epoch": 2.7296116104560073, + "grad_norm": 2.044243335723877, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7469155788421631, + "num_tokens": 885022287.0, + "step": 33102 + }, + { + "epoch": 2.7296940710810587, + "grad_norm": 2.2873616218566895, + "learning_rate": 1e-06, + "loss": 0.6737, + "mean_token_accuracy": 0.7803912162780762, + "num_tokens": 885049219.0, + "step": 33103 + }, + { + "epoch": 2.7297765317061105, + "grad_norm": 2.3121516704559326, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7750152349472046, + "num_tokens": 885073240.0, + "step": 33104 + }, + { + "epoch": 2.729858992331162, + "grad_norm": 1.9531818628311157, + "learning_rate": 1e-06, + "loss": 0.7213, + "mean_token_accuracy": 0.7682814002037048, + "num_tokens": 885107499.0, + "step": 33105 + }, + { + "epoch": 2.729941452956213, + "grad_norm": 1.9142848253250122, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7602951526641846, + "num_tokens": 885142781.0, + "step": 33106 + }, + { + "epoch": 2.730023913581265, + "grad_norm": 2.189846992492676, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7302935123443604, + "num_tokens": 885173096.0, + "step": 33107 + }, + { + "epoch": 2.7301063742063167, + "grad_norm": 2.094259262084961, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7652705907821655, + "num_tokens": 885204734.0, + "step": 33108 + }, + { + "epoch": 2.730188834831368, + "grad_norm": 2.122184991836548, + "learning_rate": 1e-06, + "loss": 0.5479, + "mean_token_accuracy": 0.8193522691726685, + "num_tokens": 885231790.0, + "step": 33109 + }, + { + "epoch": 2.7302712954564194, + "grad_norm": 1.8727755546569824, + "learning_rate": 1e-06, + "loss": 0.5937, + "mean_token_accuracy": 0.7930787801742554, + "num_tokens": 885265077.0, + "step": 33110 + }, + { + "epoch": 2.7303537560814712, + "grad_norm": 2.2006962299346924, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7578975558280945, + "num_tokens": 885291318.0, + "step": 33111 + }, + { + "epoch": 2.7304362167065226, + "grad_norm": 2.0366997718811035, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.790198802947998, + "num_tokens": 885323205.0, + "step": 33112 + }, + { + "epoch": 2.730518677331574, + "grad_norm": 2.163801670074463, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7691745162010193, + "num_tokens": 885352281.0, + "step": 33113 + }, + { + "epoch": 2.7306011379566257, + "grad_norm": 2.128786325454712, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7595897912979126, + "num_tokens": 885381859.0, + "step": 33114 + }, + { + "epoch": 2.730683598581677, + "grad_norm": 1.857782244682312, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7625956535339355, + "num_tokens": 885416633.0, + "step": 33115 + }, + { + "epoch": 2.730766059206729, + "grad_norm": 2.3504514694213867, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7722387313842773, + "num_tokens": 885441824.0, + "step": 33116 + }, + { + "epoch": 2.7308485198317802, + "grad_norm": 2.428755044937134, + "learning_rate": 1e-06, + "loss": 0.7483, + "mean_token_accuracy": 0.7626690864562988, + "num_tokens": 885465617.0, + "step": 33117 + }, + { + "epoch": 2.730930980456832, + "grad_norm": 2.1548945903778076, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.7779065370559692, + "num_tokens": 885495838.0, + "step": 33118 + }, + { + "epoch": 2.7310134410818834, + "grad_norm": 1.943207859992981, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7705285549163818, + "num_tokens": 885530489.0, + "step": 33119 + }, + { + "epoch": 2.7310959017069347, + "grad_norm": 2.317899703979492, + "learning_rate": 1e-06, + "loss": 0.7255, + "mean_token_accuracy": 0.7640982866287231, + "num_tokens": 885557594.0, + "step": 33120 + }, + { + "epoch": 2.7311783623319865, + "grad_norm": 2.256591320037842, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7786449790000916, + "num_tokens": 885585673.0, + "step": 33121 + }, + { + "epoch": 2.731260822957038, + "grad_norm": 2.022651433944702, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7751807570457458, + "num_tokens": 885617828.0, + "step": 33122 + }, + { + "epoch": 2.7313432835820897, + "grad_norm": 2.217557430267334, + "learning_rate": 1e-06, + "loss": 0.5779, + "mean_token_accuracy": 0.7976659536361694, + "num_tokens": 885643508.0, + "step": 33123 + }, + { + "epoch": 2.731425744207141, + "grad_norm": 2.2399284839630127, + "learning_rate": 1e-06, + "loss": 0.577, + "mean_token_accuracy": 0.8098025918006897, + "num_tokens": 885671571.0, + "step": 33124 + }, + { + "epoch": 2.731508204832193, + "grad_norm": 2.093566656112671, + "learning_rate": 1e-06, + "loss": 0.6484, + "mean_token_accuracy": 0.7832258939743042, + "num_tokens": 885699456.0, + "step": 33125 + }, + { + "epoch": 2.731590665457244, + "grad_norm": 2.1589977741241455, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7694646716117859, + "num_tokens": 885730747.0, + "step": 33126 + }, + { + "epoch": 2.7316731260822955, + "grad_norm": 1.901166319847107, + "learning_rate": 1e-06, + "loss": 0.5611, + "mean_token_accuracy": 0.8130780458450317, + "num_tokens": 885762879.0, + "step": 33127 + }, + { + "epoch": 2.7317555867073473, + "grad_norm": 2.237809419631958, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7430351376533508, + "num_tokens": 885791655.0, + "step": 33128 + }, + { + "epoch": 2.7318380473323987, + "grad_norm": 2.2932026386260986, + "learning_rate": 1e-06, + "loss": 0.6012, + "mean_token_accuracy": 0.7957579493522644, + "num_tokens": 885820351.0, + "step": 33129 + }, + { + "epoch": 2.7319205079574505, + "grad_norm": 2.112490177154541, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7727636098861694, + "num_tokens": 885853430.0, + "step": 33130 + }, + { + "epoch": 2.732002968582502, + "grad_norm": 1.7583518028259277, + "learning_rate": 1e-06, + "loss": 0.5461, + "mean_token_accuracy": 0.8213154673576355, + "num_tokens": 885890060.0, + "step": 33131 + }, + { + "epoch": 2.7320854292075536, + "grad_norm": 2.3358664512634277, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7680662274360657, + "num_tokens": 885917824.0, + "step": 33132 + }, + { + "epoch": 2.732167889832605, + "grad_norm": 2.2120063304901123, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.779731273651123, + "num_tokens": 885945543.0, + "step": 33133 + }, + { + "epoch": 2.7322503504576563, + "grad_norm": 2.3071155548095703, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7562962174415588, + "num_tokens": 885974016.0, + "step": 33134 + }, + { + "epoch": 2.732332811082708, + "grad_norm": 2.0678794384002686, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.7840365171432495, + "num_tokens": 886005547.0, + "step": 33135 + }, + { + "epoch": 2.7324152717077594, + "grad_norm": 2.878462553024292, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.750253438949585, + "num_tokens": 886026156.0, + "step": 33136 + }, + { + "epoch": 2.7324977323328112, + "grad_norm": 2.4541239738464355, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7731660604476929, + "num_tokens": 886049513.0, + "step": 33137 + }, + { + "epoch": 2.7325801929578626, + "grad_norm": 2.239558219909668, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7779449820518494, + "num_tokens": 886076216.0, + "step": 33138 + }, + { + "epoch": 2.7326626535829144, + "grad_norm": 1.952829122543335, + "learning_rate": 1e-06, + "loss": 0.6023, + "mean_token_accuracy": 0.7969349026679993, + "num_tokens": 886107988.0, + "step": 33139 + }, + { + "epoch": 2.7327451142079657, + "grad_norm": 2.0095138549804688, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7706425189971924, + "num_tokens": 886139546.0, + "step": 33140 + }, + { + "epoch": 2.732827574833017, + "grad_norm": 2.1878340244293213, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.7918014526367188, + "num_tokens": 886166479.0, + "step": 33141 + }, + { + "epoch": 2.732910035458069, + "grad_norm": 2.1053662300109863, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7461529970169067, + "num_tokens": 886197072.0, + "step": 33142 + }, + { + "epoch": 2.7329924960831202, + "grad_norm": 2.399986743927002, + "learning_rate": 1e-06, + "loss": 0.6693, + "mean_token_accuracy": 0.778782844543457, + "num_tokens": 886221545.0, + "step": 33143 + }, + { + "epoch": 2.7330749567081716, + "grad_norm": 2.26163387298584, + "learning_rate": 1e-06, + "loss": 0.649, + "mean_token_accuracy": 0.7843267321586609, + "num_tokens": 886247998.0, + "step": 33144 + }, + { + "epoch": 2.7331574173332234, + "grad_norm": 2.2930352687835693, + "learning_rate": 1e-06, + "loss": 0.749, + "mean_token_accuracy": 0.7594104409217834, + "num_tokens": 886274405.0, + "step": 33145 + }, + { + "epoch": 2.733239877958275, + "grad_norm": 2.3917336463928223, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7721024751663208, + "num_tokens": 886299324.0, + "step": 33146 + }, + { + "epoch": 2.7333223385833265, + "grad_norm": 2.1602842807769775, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.7701461911201477, + "num_tokens": 886327820.0, + "step": 33147 + }, + { + "epoch": 2.733404799208378, + "grad_norm": 1.9261701107025146, + "learning_rate": 1e-06, + "loss": 0.7425, + "mean_token_accuracy": 0.7661536335945129, + "num_tokens": 886363300.0, + "step": 33148 + }, + { + "epoch": 2.7334872598334297, + "grad_norm": 1.9999510049819946, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7635335922241211, + "num_tokens": 886396433.0, + "step": 33149 + }, + { + "epoch": 2.733569720458481, + "grad_norm": 2.216026782989502, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7692108154296875, + "num_tokens": 886423634.0, + "step": 33150 + }, + { + "epoch": 2.7336521810835324, + "grad_norm": 2.1844828128814697, + "learning_rate": 1e-06, + "loss": 0.5928, + "mean_token_accuracy": 0.7954524755477905, + "num_tokens": 886451277.0, + "step": 33151 + }, + { + "epoch": 2.733734641708584, + "grad_norm": 1.9874610900878906, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.733600914478302, + "num_tokens": 886483079.0, + "step": 33152 + }, + { + "epoch": 2.733817102333636, + "grad_norm": 2.1688592433929443, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.7868782877922058, + "num_tokens": 886510643.0, + "step": 33153 + }, + { + "epoch": 2.7338995629586873, + "grad_norm": 1.8994760513305664, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.7835850715637207, + "num_tokens": 886545651.0, + "step": 33154 + }, + { + "epoch": 2.7339820235837387, + "grad_norm": 1.9702881574630737, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7766764163970947, + "num_tokens": 886578178.0, + "step": 33155 + }, + { + "epoch": 2.7340644842087904, + "grad_norm": 1.9905096292495728, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.7666704654693604, + "num_tokens": 886612275.0, + "step": 33156 + }, + { + "epoch": 2.734146944833842, + "grad_norm": 2.1037211418151855, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.743654727935791, + "num_tokens": 886645183.0, + "step": 33157 + }, + { + "epoch": 2.734229405458893, + "grad_norm": 2.317758560180664, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7633463740348816, + "num_tokens": 886673326.0, + "step": 33158 + }, + { + "epoch": 2.734311866083945, + "grad_norm": 2.0619630813598633, + "learning_rate": 1e-06, + "loss": 0.6085, + "mean_token_accuracy": 0.8059639930725098, + "num_tokens": 886705699.0, + "step": 33159 + }, + { + "epoch": 2.7343943267089963, + "grad_norm": 2.1492908000946045, + "learning_rate": 1e-06, + "loss": 0.7016, + "mean_token_accuracy": 0.7807621359825134, + "num_tokens": 886731939.0, + "step": 33160 + }, + { + "epoch": 2.734476787334048, + "grad_norm": 2.026324987411499, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7570275068283081, + "num_tokens": 886767528.0, + "step": 33161 + }, + { + "epoch": 2.7345592479590994, + "grad_norm": 2.0089685916900635, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.7890197038650513, + "num_tokens": 886796823.0, + "step": 33162 + }, + { + "epoch": 2.7346417085841512, + "grad_norm": 2.173231840133667, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7451496124267578, + "num_tokens": 886828636.0, + "step": 33163 + }, + { + "epoch": 2.7347241692092026, + "grad_norm": 2.1943137645721436, + "learning_rate": 1e-06, + "loss": 0.652, + "mean_token_accuracy": 0.7925412058830261, + "num_tokens": 886854958.0, + "step": 33164 + }, + { + "epoch": 2.734806629834254, + "grad_norm": 2.0972423553466797, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7756929397583008, + "num_tokens": 886884141.0, + "step": 33165 + }, + { + "epoch": 2.7348890904593057, + "grad_norm": 1.974434733390808, + "learning_rate": 1e-06, + "loss": 0.5981, + "mean_token_accuracy": 0.8020280003547668, + "num_tokens": 886912611.0, + "step": 33166 + }, + { + "epoch": 2.734971551084357, + "grad_norm": 1.8893673419952393, + "learning_rate": 1e-06, + "loss": 0.542, + "mean_token_accuracy": 0.8232333064079285, + "num_tokens": 886942810.0, + "step": 33167 + }, + { + "epoch": 2.735054011709409, + "grad_norm": 2.081336498260498, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7333999276161194, + "num_tokens": 886971985.0, + "step": 33168 + }, + { + "epoch": 2.7351364723344602, + "grad_norm": 2.0649468898773193, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7459583282470703, + "num_tokens": 887002863.0, + "step": 33169 + }, + { + "epoch": 2.735218932959512, + "grad_norm": 2.27052903175354, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7830561399459839, + "num_tokens": 887027729.0, + "step": 33170 + }, + { + "epoch": 2.7353013935845634, + "grad_norm": 2.1370601654052734, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7796099185943604, + "num_tokens": 887055218.0, + "step": 33171 + }, + { + "epoch": 2.7353838542096147, + "grad_norm": 1.961361289024353, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.7923853397369385, + "num_tokens": 887088622.0, + "step": 33172 + }, + { + "epoch": 2.7354663148346665, + "grad_norm": 2.0533807277679443, + "learning_rate": 1e-06, + "loss": 0.737, + "mean_token_accuracy": 0.763123095035553, + "num_tokens": 887119339.0, + "step": 33173 + }, + { + "epoch": 2.735548775459718, + "grad_norm": 1.981676459312439, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.8002691864967346, + "num_tokens": 887151830.0, + "step": 33174 + }, + { + "epoch": 2.7356312360847697, + "grad_norm": 2.318153142929077, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7687729597091675, + "num_tokens": 887177980.0, + "step": 33175 + }, + { + "epoch": 2.735713696709821, + "grad_norm": 2.1301980018615723, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.753547191619873, + "num_tokens": 887207512.0, + "step": 33176 + }, + { + "epoch": 2.735796157334873, + "grad_norm": 1.9506182670593262, + "learning_rate": 1e-06, + "loss": 0.5276, + "mean_token_accuracy": 0.8210862874984741, + "num_tokens": 887240547.0, + "step": 33177 + }, + { + "epoch": 2.735878617959924, + "grad_norm": 1.9821964502334595, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.7851358652114868, + "num_tokens": 887274742.0, + "step": 33178 + }, + { + "epoch": 2.7359610785849755, + "grad_norm": 2.1234500408172607, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7351858615875244, + "num_tokens": 887304399.0, + "step": 33179 + }, + { + "epoch": 2.7360435392100273, + "grad_norm": 2.0152084827423096, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7894583344459534, + "num_tokens": 887334864.0, + "step": 33180 + }, + { + "epoch": 2.7361259998350786, + "grad_norm": 2.1244378089904785, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.756077766418457, + "num_tokens": 887364118.0, + "step": 33181 + }, + { + "epoch": 2.7362084604601304, + "grad_norm": 2.001051902770996, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.761960506439209, + "num_tokens": 887399145.0, + "step": 33182 + }, + { + "epoch": 2.736290921085182, + "grad_norm": 2.3337128162384033, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7469336986541748, + "num_tokens": 887426039.0, + "step": 33183 + }, + { + "epoch": 2.7363733817102336, + "grad_norm": 1.8705507516860962, + "learning_rate": 1e-06, + "loss": 0.6392, + "mean_token_accuracy": 0.7893019318580627, + "num_tokens": 887461628.0, + "step": 33184 + }, + { + "epoch": 2.736455842335285, + "grad_norm": 1.9118348360061646, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7407803535461426, + "num_tokens": 887496660.0, + "step": 33185 + }, + { + "epoch": 2.7365383029603363, + "grad_norm": 2.305603265762329, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7506763339042664, + "num_tokens": 887523198.0, + "step": 33186 + }, + { + "epoch": 2.736620763585388, + "grad_norm": 2.338531255722046, + "learning_rate": 1e-06, + "loss": 0.6538, + "mean_token_accuracy": 0.7845132350921631, + "num_tokens": 887550355.0, + "step": 33187 + }, + { + "epoch": 2.7367032242104394, + "grad_norm": 2.4344794750213623, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7445429563522339, + "num_tokens": 887574943.0, + "step": 33188 + }, + { + "epoch": 2.736785684835491, + "grad_norm": 2.1494455337524414, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.7872499227523804, + "num_tokens": 887605544.0, + "step": 33189 + }, + { + "epoch": 2.7368681454605426, + "grad_norm": 1.9765784740447998, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7836087942123413, + "num_tokens": 887638015.0, + "step": 33190 + }, + { + "epoch": 2.7369506060855944, + "grad_norm": 2.0716640949249268, + "learning_rate": 1e-06, + "loss": 0.7105, + "mean_token_accuracy": 0.7655152082443237, + "num_tokens": 887667919.0, + "step": 33191 + }, + { + "epoch": 2.7370330667106457, + "grad_norm": 2.001659393310547, + "learning_rate": 1e-06, + "loss": 0.71, + "mean_token_accuracy": 0.7738385200500488, + "num_tokens": 887703832.0, + "step": 33192 + }, + { + "epoch": 2.737115527335697, + "grad_norm": 1.8690613508224487, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7721806168556213, + "num_tokens": 887740872.0, + "step": 33193 + }, + { + "epoch": 2.737197987960749, + "grad_norm": 1.912336826324463, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.7745833396911621, + "num_tokens": 887775134.0, + "step": 33194 + }, + { + "epoch": 2.7372804485858, + "grad_norm": 2.0479676723480225, + "learning_rate": 1e-06, + "loss": 0.7839, + "mean_token_accuracy": 0.7521913051605225, + "num_tokens": 887806936.0, + "step": 33195 + }, + { + "epoch": 2.7373629092108516, + "grad_norm": 1.9292209148406982, + "learning_rate": 1e-06, + "loss": 0.4877, + "mean_token_accuracy": 0.8369779586791992, + "num_tokens": 887837632.0, + "step": 33196 + }, + { + "epoch": 2.7374453698359034, + "grad_norm": 1.9813870191574097, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7761362791061401, + "num_tokens": 887872839.0, + "step": 33197 + }, + { + "epoch": 2.737527830460955, + "grad_norm": 2.0413389205932617, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7347474098205566, + "num_tokens": 887904593.0, + "step": 33198 + }, + { + "epoch": 2.7376102910860065, + "grad_norm": 1.9511743783950806, + "learning_rate": 1e-06, + "loss": 0.7474, + "mean_token_accuracy": 0.7615026235580444, + "num_tokens": 887940074.0, + "step": 33199 + }, + { + "epoch": 2.737692751711058, + "grad_norm": 2.1661806106567383, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7398123741149902, + "num_tokens": 887971520.0, + "step": 33200 + }, + { + "epoch": 2.7377752123361097, + "grad_norm": 2.170698881149292, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7775043249130249, + "num_tokens": 887998770.0, + "step": 33201 + }, + { + "epoch": 2.737857672961161, + "grad_norm": 2.2883262634277344, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.7641960382461548, + "num_tokens": 888026774.0, + "step": 33202 + }, + { + "epoch": 2.7379401335862124, + "grad_norm": 2.3974692821502686, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7571073770523071, + "num_tokens": 888050683.0, + "step": 33203 + }, + { + "epoch": 2.738022594211264, + "grad_norm": 2.174304723739624, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7753839492797852, + "num_tokens": 888079480.0, + "step": 33204 + }, + { + "epoch": 2.7381050548363155, + "grad_norm": 2.026975393295288, + "learning_rate": 1e-06, + "loss": 0.6173, + "mean_token_accuracy": 0.7984259128570557, + "num_tokens": 888109637.0, + "step": 33205 + }, + { + "epoch": 2.7381875154613673, + "grad_norm": 1.9309505224227905, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.774500846862793, + "num_tokens": 888146982.0, + "step": 33206 + }, + { + "epoch": 2.7382699760864186, + "grad_norm": 2.312856912612915, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7688028216362, + "num_tokens": 888174747.0, + "step": 33207 + }, + { + "epoch": 2.7383524367114704, + "grad_norm": 2.0646872520446777, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.7966132760047913, + "num_tokens": 888205220.0, + "step": 33208 + }, + { + "epoch": 2.738434897336522, + "grad_norm": 1.9875620603561401, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7812687158584595, + "num_tokens": 888237396.0, + "step": 33209 + }, + { + "epoch": 2.738517357961573, + "grad_norm": 2.1724612712860107, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7374118566513062, + "num_tokens": 888268001.0, + "step": 33210 + }, + { + "epoch": 2.738599818586625, + "grad_norm": 2.0731537342071533, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7670412063598633, + "num_tokens": 888299643.0, + "step": 33211 + }, + { + "epoch": 2.7386822792116763, + "grad_norm": 2.1616158485412598, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.7927990555763245, + "num_tokens": 888327924.0, + "step": 33212 + }, + { + "epoch": 2.738764739836728, + "grad_norm": 2.0078186988830566, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7579212188720703, + "num_tokens": 888362186.0, + "step": 33213 + }, + { + "epoch": 2.7388472004617794, + "grad_norm": 2.0129597187042236, + "learning_rate": 1e-06, + "loss": 0.7763, + "mean_token_accuracy": 0.7532812356948853, + "num_tokens": 888396701.0, + "step": 33214 + }, + { + "epoch": 2.7389296610868312, + "grad_norm": 2.0607869625091553, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7457648515701294, + "num_tokens": 888427676.0, + "step": 33215 + }, + { + "epoch": 2.7390121217118826, + "grad_norm": 2.222742795944214, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.774728536605835, + "num_tokens": 888457383.0, + "step": 33216 + }, + { + "epoch": 2.739094582336934, + "grad_norm": 2.168577194213867, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7705027461051941, + "num_tokens": 888489031.0, + "step": 33217 + }, + { + "epoch": 2.7391770429619857, + "grad_norm": 2.2000210285186768, + "learning_rate": 1e-06, + "loss": 0.5709, + "mean_token_accuracy": 0.7969604730606079, + "num_tokens": 888514763.0, + "step": 33218 + }, + { + "epoch": 2.739259503587037, + "grad_norm": 2.029258966445923, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.774904191493988, + "num_tokens": 888546202.0, + "step": 33219 + }, + { + "epoch": 2.739341964212089, + "grad_norm": 2.4269962310791016, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7596340179443359, + "num_tokens": 888570465.0, + "step": 33220 + }, + { + "epoch": 2.73942442483714, + "grad_norm": 1.961626410484314, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.7882053852081299, + "num_tokens": 888604398.0, + "step": 33221 + }, + { + "epoch": 2.739506885462192, + "grad_norm": 2.17659592628479, + "learning_rate": 1e-06, + "loss": 0.7317, + "mean_token_accuracy": 0.7552897930145264, + "num_tokens": 888635005.0, + "step": 33222 + }, + { + "epoch": 2.7395893460872434, + "grad_norm": 2.18009877204895, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7709226608276367, + "num_tokens": 888663511.0, + "step": 33223 + }, + { + "epoch": 2.7396718067122947, + "grad_norm": 1.9694267511367798, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7580083608627319, + "num_tokens": 888699140.0, + "step": 33224 + }, + { + "epoch": 2.7397542673373465, + "grad_norm": 1.9233750104904175, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.7861494421958923, + "num_tokens": 888733420.0, + "step": 33225 + }, + { + "epoch": 2.739836727962398, + "grad_norm": 2.198660135269165, + "learning_rate": 1e-06, + "loss": 0.5789, + "mean_token_accuracy": 0.8064000606536865, + "num_tokens": 888758590.0, + "step": 33226 + }, + { + "epoch": 2.7399191885874496, + "grad_norm": 1.9244816303253174, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7861306071281433, + "num_tokens": 888792403.0, + "step": 33227 + }, + { + "epoch": 2.740001649212501, + "grad_norm": 1.7549629211425781, + "learning_rate": 1e-06, + "loss": 0.5768, + "mean_token_accuracy": 0.8181254863739014, + "num_tokens": 888827975.0, + "step": 33228 + }, + { + "epoch": 2.740084109837553, + "grad_norm": 1.9831475019454956, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.746265172958374, + "num_tokens": 888861470.0, + "step": 33229 + }, + { + "epoch": 2.740166570462604, + "grad_norm": 2.0140273571014404, + "learning_rate": 1e-06, + "loss": 0.6933, + "mean_token_accuracy": 0.7708961963653564, + "num_tokens": 888893519.0, + "step": 33230 + }, + { + "epoch": 2.7402490310876555, + "grad_norm": 2.558655023574829, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.740084707736969, + "num_tokens": 888916161.0, + "step": 33231 + }, + { + "epoch": 2.7403314917127073, + "grad_norm": 1.952864646911621, + "learning_rate": 1e-06, + "loss": 0.59, + "mean_token_accuracy": 0.7994705438613892, + "num_tokens": 888950932.0, + "step": 33232 + }, + { + "epoch": 2.7404139523377586, + "grad_norm": 2.1307880878448486, + "learning_rate": 1e-06, + "loss": 0.7629, + "mean_token_accuracy": 0.7566384077072144, + "num_tokens": 888980239.0, + "step": 33233 + }, + { + "epoch": 2.74049641296281, + "grad_norm": 1.9964971542358398, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.7932875752449036, + "num_tokens": 889008613.0, + "step": 33234 + }, + { + "epoch": 2.740578873587862, + "grad_norm": 1.9707207679748535, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7658000588417053, + "num_tokens": 889040490.0, + "step": 33235 + }, + { + "epoch": 2.7406613342129136, + "grad_norm": 2.194716215133667, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7479891180992126, + "num_tokens": 889069085.0, + "step": 33236 + }, + { + "epoch": 2.740743794837965, + "grad_norm": 2.318999767303467, + "learning_rate": 1e-06, + "loss": 0.7607, + "mean_token_accuracy": 0.7544870376586914, + "num_tokens": 889096565.0, + "step": 33237 + }, + { + "epoch": 2.7408262554630163, + "grad_norm": 1.9509034156799316, + "learning_rate": 1e-06, + "loss": 0.6284, + "mean_token_accuracy": 0.7919030785560608, + "num_tokens": 889129491.0, + "step": 33238 + }, + { + "epoch": 2.740908716088068, + "grad_norm": 2.0627968311309814, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7771512269973755, + "num_tokens": 889159546.0, + "step": 33239 + }, + { + "epoch": 2.7409911767131194, + "grad_norm": 2.0502984523773193, + "learning_rate": 1e-06, + "loss": 0.7706, + "mean_token_accuracy": 0.7618530988693237, + "num_tokens": 889192448.0, + "step": 33240 + }, + { + "epoch": 2.7410736373381708, + "grad_norm": 2.143886089324951, + "learning_rate": 1e-06, + "loss": 0.7022, + "mean_token_accuracy": 0.7728813886642456, + "num_tokens": 889219009.0, + "step": 33241 + }, + { + "epoch": 2.7411560979632226, + "grad_norm": 2.0077741146087646, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7796036005020142, + "num_tokens": 889251943.0, + "step": 33242 + }, + { + "epoch": 2.7412385585882744, + "grad_norm": 2.0279324054718018, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7732057571411133, + "num_tokens": 889284768.0, + "step": 33243 + }, + { + "epoch": 2.7413210192133257, + "grad_norm": 2.549699306488037, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7486979961395264, + "num_tokens": 889310260.0, + "step": 33244 + }, + { + "epoch": 2.741403479838377, + "grad_norm": 1.8933285474777222, + "learning_rate": 1e-06, + "loss": 0.5916, + "mean_token_accuracy": 0.8003227114677429, + "num_tokens": 889344995.0, + "step": 33245 + }, + { + "epoch": 2.741485940463429, + "grad_norm": 2.229231595993042, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7786679863929749, + "num_tokens": 889370100.0, + "step": 33246 + }, + { + "epoch": 2.74156840108848, + "grad_norm": 2.2587180137634277, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7394883632659912, + "num_tokens": 889397744.0, + "step": 33247 + }, + { + "epoch": 2.7416508617135316, + "grad_norm": 2.327275037765503, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7518802881240845, + "num_tokens": 889426436.0, + "step": 33248 + }, + { + "epoch": 2.7417333223385834, + "grad_norm": 2.0954790115356445, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7715533971786499, + "num_tokens": 889460259.0, + "step": 33249 + }, + { + "epoch": 2.7418157829636347, + "grad_norm": 2.02878475189209, + "learning_rate": 1e-06, + "loss": 0.5615, + "mean_token_accuracy": 0.8066223859786987, + "num_tokens": 889492419.0, + "step": 33250 + }, + { + "epoch": 2.7418982435886865, + "grad_norm": 1.8259962797164917, + "learning_rate": 1e-06, + "loss": 0.5182, + "mean_token_accuracy": 0.8243854641914368, + "num_tokens": 889530311.0, + "step": 33251 + }, + { + "epoch": 2.741980704213738, + "grad_norm": 2.1422080993652344, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7654680609703064, + "num_tokens": 889559139.0, + "step": 33252 + }, + { + "epoch": 2.7420631648387896, + "grad_norm": 2.0911059379577637, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7832746505737305, + "num_tokens": 889587439.0, + "step": 33253 + }, + { + "epoch": 2.742145625463841, + "grad_norm": 1.9859120845794678, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7759659290313721, + "num_tokens": 889623954.0, + "step": 33254 + }, + { + "epoch": 2.7422280860888923, + "grad_norm": 1.8865779638290405, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.7786107063293457, + "num_tokens": 889659799.0, + "step": 33255 + }, + { + "epoch": 2.742310546713944, + "grad_norm": 2.3939387798309326, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7639813423156738, + "num_tokens": 889686180.0, + "step": 33256 + }, + { + "epoch": 2.7423930073389955, + "grad_norm": 2.0391576290130615, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7730478048324585, + "num_tokens": 889719325.0, + "step": 33257 + }, + { + "epoch": 2.7424754679640473, + "grad_norm": 2.220773458480835, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7723708152770996, + "num_tokens": 889744408.0, + "step": 33258 + }, + { + "epoch": 2.7425579285890986, + "grad_norm": 2.163508653640747, + "learning_rate": 1e-06, + "loss": 0.6897, + "mean_token_accuracy": 0.7634159326553345, + "num_tokens": 889774050.0, + "step": 33259 + }, + { + "epoch": 2.7426403892141504, + "grad_norm": 2.2218053340911865, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7482187151908875, + "num_tokens": 889801455.0, + "step": 33260 + }, + { + "epoch": 2.742722849839202, + "grad_norm": 2.1299266815185547, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7869665622711182, + "num_tokens": 889832958.0, + "step": 33261 + }, + { + "epoch": 2.742805310464253, + "grad_norm": 2.1326963901519775, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7767144441604614, + "num_tokens": 889862328.0, + "step": 33262 + }, + { + "epoch": 2.742887771089305, + "grad_norm": 2.081616163253784, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.733372151851654, + "num_tokens": 889893936.0, + "step": 33263 + }, + { + "epoch": 2.7429702317143563, + "grad_norm": 2.307318925857544, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7683700919151306, + "num_tokens": 889922698.0, + "step": 33264 + }, + { + "epoch": 2.743052692339408, + "grad_norm": 2.1520164012908936, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7621221542358398, + "num_tokens": 889950620.0, + "step": 33265 + }, + { + "epoch": 2.7431351529644594, + "grad_norm": 2.6291427612304688, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7183544039726257, + "num_tokens": 889972602.0, + "step": 33266 + }, + { + "epoch": 2.743217613589511, + "grad_norm": 2.023770809173584, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.7928599119186401, + "num_tokens": 890007225.0, + "step": 33267 + }, + { + "epoch": 2.7433000742145626, + "grad_norm": 2.1516950130462646, + "learning_rate": 1e-06, + "loss": 0.6413, + "mean_token_accuracy": 0.7975867986679077, + "num_tokens": 890036096.0, + "step": 33268 + }, + { + "epoch": 2.743382534839614, + "grad_norm": 2.5048303604125977, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7503923177719116, + "num_tokens": 890059824.0, + "step": 33269 + }, + { + "epoch": 2.7434649954646657, + "grad_norm": 2.142181158065796, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7846797704696655, + "num_tokens": 890087871.0, + "step": 33270 + }, + { + "epoch": 2.743547456089717, + "grad_norm": 2.2891039848327637, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7576471567153931, + "num_tokens": 890114126.0, + "step": 33271 + }, + { + "epoch": 2.743629916714769, + "grad_norm": 2.2316246032714844, + "learning_rate": 1e-06, + "loss": 0.7887, + "mean_token_accuracy": 0.7471857070922852, + "num_tokens": 890143966.0, + "step": 33272 + }, + { + "epoch": 2.74371237733982, + "grad_norm": 1.8357529640197754, + "learning_rate": 1e-06, + "loss": 0.5632, + "mean_token_accuracy": 0.8119446635246277, + "num_tokens": 890179334.0, + "step": 33273 + }, + { + "epoch": 2.743794837964872, + "grad_norm": 2.1364729404449463, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.773121178150177, + "num_tokens": 890210962.0, + "step": 33274 + }, + { + "epoch": 2.7438772985899234, + "grad_norm": 1.8651286363601685, + "learning_rate": 1e-06, + "loss": 0.5851, + "mean_token_accuracy": 0.8005242347717285, + "num_tokens": 890245819.0, + "step": 33275 + }, + { + "epoch": 2.7439597592149747, + "grad_norm": 2.0805184841156006, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.785193920135498, + "num_tokens": 890276315.0, + "step": 33276 + }, + { + "epoch": 2.7440422198400265, + "grad_norm": 2.031081199645996, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7838888168334961, + "num_tokens": 890308693.0, + "step": 33277 + }, + { + "epoch": 2.744124680465078, + "grad_norm": 1.7713450193405151, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7864964008331299, + "num_tokens": 890344928.0, + "step": 33278 + }, + { + "epoch": 2.744207141090129, + "grad_norm": 2.4823224544525146, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7631434202194214, + "num_tokens": 890368319.0, + "step": 33279 + }, + { + "epoch": 2.744289601715181, + "grad_norm": 1.9895519018173218, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.7950395941734314, + "num_tokens": 890400084.0, + "step": 33280 + }, + { + "epoch": 2.744372062340233, + "grad_norm": 2.3220415115356445, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7757652997970581, + "num_tokens": 890424886.0, + "step": 33281 + }, + { + "epoch": 2.744454522965284, + "grad_norm": 1.8436466455459595, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7671716809272766, + "num_tokens": 890463280.0, + "step": 33282 + }, + { + "epoch": 2.7445369835903355, + "grad_norm": 2.1030962467193604, + "learning_rate": 1e-06, + "loss": 0.689, + "mean_token_accuracy": 0.769443690776825, + "num_tokens": 890494984.0, + "step": 33283 + }, + { + "epoch": 2.7446194442153873, + "grad_norm": 2.0916123390197754, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7578181624412537, + "num_tokens": 890524721.0, + "step": 33284 + }, + { + "epoch": 2.7447019048404386, + "grad_norm": 1.8552029132843018, + "learning_rate": 1e-06, + "loss": 0.5562, + "mean_token_accuracy": 0.8138949871063232, + "num_tokens": 890560981.0, + "step": 33285 + }, + { + "epoch": 2.74478436546549, + "grad_norm": 1.9856877326965332, + "learning_rate": 1e-06, + "loss": 0.5605, + "mean_token_accuracy": 0.8074377775192261, + "num_tokens": 890590247.0, + "step": 33286 + }, + { + "epoch": 2.744866826090542, + "grad_norm": 1.9703354835510254, + "learning_rate": 1e-06, + "loss": 0.6144, + "mean_token_accuracy": 0.7962992191314697, + "num_tokens": 890623204.0, + "step": 33287 + }, + { + "epoch": 2.7449492867155936, + "grad_norm": 1.7685412168502808, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.7919207215309143, + "num_tokens": 890662295.0, + "step": 33288 + }, + { + "epoch": 2.745031747340645, + "grad_norm": 2.135101795196533, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7515379190444946, + "num_tokens": 890692082.0, + "step": 33289 + }, + { + "epoch": 2.7451142079656963, + "grad_norm": 2.0915563106536865, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.748866081237793, + "num_tokens": 890723776.0, + "step": 33290 + }, + { + "epoch": 2.745196668590748, + "grad_norm": 2.000063896179199, + "learning_rate": 1e-06, + "loss": 0.4943, + "mean_token_accuracy": 0.8381527662277222, + "num_tokens": 890751178.0, + "step": 33291 + }, + { + "epoch": 2.7452791292157994, + "grad_norm": 1.9925401210784912, + "learning_rate": 1e-06, + "loss": 0.6425, + "mean_token_accuracy": 0.7818894386291504, + "num_tokens": 890783771.0, + "step": 33292 + }, + { + "epoch": 2.7453615898408508, + "grad_norm": 1.9447190761566162, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7804027795791626, + "num_tokens": 890819109.0, + "step": 33293 + }, + { + "epoch": 2.7454440504659026, + "grad_norm": 1.946452021598816, + "learning_rate": 1e-06, + "loss": 0.6044, + "mean_token_accuracy": 0.806492269039154, + "num_tokens": 890851208.0, + "step": 33294 + }, + { + "epoch": 2.745526511090954, + "grad_norm": 1.9941987991333008, + "learning_rate": 1e-06, + "loss": 0.5355, + "mean_token_accuracy": 0.8102973103523254, + "num_tokens": 890884729.0, + "step": 33295 + }, + { + "epoch": 2.7456089717160057, + "grad_norm": 2.103977680206299, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.7880176901817322, + "num_tokens": 890914422.0, + "step": 33296 + }, + { + "epoch": 2.745691432341057, + "grad_norm": 2.4619431495666504, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7422462701797485, + "num_tokens": 890941277.0, + "step": 33297 + }, + { + "epoch": 2.745773892966109, + "grad_norm": 2.349599838256836, + "learning_rate": 1e-06, + "loss": 0.7994, + "mean_token_accuracy": 0.7443307638168335, + "num_tokens": 890968491.0, + "step": 33298 + }, + { + "epoch": 2.74585635359116, + "grad_norm": 2.143207550048828, + "learning_rate": 1e-06, + "loss": 0.6701, + "mean_token_accuracy": 0.7930985689163208, + "num_tokens": 890998795.0, + "step": 33299 + }, + { + "epoch": 2.7459388142162116, + "grad_norm": 2.2782599925994873, + "learning_rate": 1e-06, + "loss": 0.5997, + "mean_token_accuracy": 0.7905292510986328, + "num_tokens": 891026461.0, + "step": 33300 + }, + { + "epoch": 2.7460212748412633, + "grad_norm": 2.0298686027526855, + "learning_rate": 1e-06, + "loss": 0.5113, + "mean_token_accuracy": 0.8224043846130371, + "num_tokens": 891053247.0, + "step": 33301 + }, + { + "epoch": 2.7461037354663147, + "grad_norm": 1.943800449371338, + "learning_rate": 1e-06, + "loss": 0.5485, + "mean_token_accuracy": 0.8164938688278198, + "num_tokens": 891085040.0, + "step": 33302 + }, + { + "epoch": 2.7461861960913665, + "grad_norm": 2.205650568008423, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7754793167114258, + "num_tokens": 891111687.0, + "step": 33303 + }, + { + "epoch": 2.746268656716418, + "grad_norm": 2.1594901084899902, + "learning_rate": 1e-06, + "loss": 0.7417, + "mean_token_accuracy": 0.7648207545280457, + "num_tokens": 891141754.0, + "step": 33304 + }, + { + "epoch": 2.7463511173414696, + "grad_norm": 1.9682286977767944, + "learning_rate": 1e-06, + "loss": 0.7988, + "mean_token_accuracy": 0.7536771893501282, + "num_tokens": 891176134.0, + "step": 33305 + }, + { + "epoch": 2.746433577966521, + "grad_norm": 2.205899715423584, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7778598070144653, + "num_tokens": 891202242.0, + "step": 33306 + }, + { + "epoch": 2.7465160385915723, + "grad_norm": 1.9469730854034424, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7655174732208252, + "num_tokens": 891235900.0, + "step": 33307 + }, + { + "epoch": 2.746598499216624, + "grad_norm": 1.871841549873352, + "learning_rate": 1e-06, + "loss": 0.6985, + "mean_token_accuracy": 0.7841787338256836, + "num_tokens": 891268736.0, + "step": 33308 + }, + { + "epoch": 2.7466809598416755, + "grad_norm": 2.0929412841796875, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7815569639205933, + "num_tokens": 891300732.0, + "step": 33309 + }, + { + "epoch": 2.7467634204667273, + "grad_norm": 2.0318024158477783, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.791404128074646, + "num_tokens": 891333108.0, + "step": 33310 + }, + { + "epoch": 2.7468458810917786, + "grad_norm": 2.2161524295806885, + "learning_rate": 1e-06, + "loss": 0.7221, + "mean_token_accuracy": 0.7718702554702759, + "num_tokens": 891359323.0, + "step": 33311 + }, + { + "epoch": 2.7469283417168304, + "grad_norm": 2.045038938522339, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7257661819458008, + "num_tokens": 891394901.0, + "step": 33312 + }, + { + "epoch": 2.7470108023418818, + "grad_norm": 1.9727892875671387, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7581579089164734, + "num_tokens": 891430034.0, + "step": 33313 + }, + { + "epoch": 2.747093262966933, + "grad_norm": 2.2311296463012695, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.8091474175453186, + "num_tokens": 891458918.0, + "step": 33314 + }, + { + "epoch": 2.747175723591985, + "grad_norm": 2.173600912094116, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.7400397062301636, + "num_tokens": 891489526.0, + "step": 33315 + }, + { + "epoch": 2.7472581842170363, + "grad_norm": 2.061959743499756, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7913453578948975, + "num_tokens": 891520317.0, + "step": 33316 + }, + { + "epoch": 2.747340644842088, + "grad_norm": 2.306443691253662, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7343339323997498, + "num_tokens": 891547180.0, + "step": 33317 + }, + { + "epoch": 2.7474231054671394, + "grad_norm": 2.4852302074432373, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7734256386756897, + "num_tokens": 891571516.0, + "step": 33318 + }, + { + "epoch": 2.747505566092191, + "grad_norm": 2.206023693084717, + "learning_rate": 1e-06, + "loss": 0.6003, + "mean_token_accuracy": 0.8022201061248779, + "num_tokens": 891598642.0, + "step": 33319 + }, + { + "epoch": 2.7475880267172426, + "grad_norm": 2.5325777530670166, + "learning_rate": 1e-06, + "loss": 0.6769, + "mean_token_accuracy": 0.771186113357544, + "num_tokens": 891621628.0, + "step": 33320 + }, + { + "epoch": 2.747670487342294, + "grad_norm": 2.3567862510681152, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7569575309753418, + "num_tokens": 891645798.0, + "step": 33321 + }, + { + "epoch": 2.7477529479673457, + "grad_norm": 2.418199062347412, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7460541725158691, + "num_tokens": 891672528.0, + "step": 33322 + }, + { + "epoch": 2.747835408592397, + "grad_norm": 2.249473810195923, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7616361379623413, + "num_tokens": 891699333.0, + "step": 33323 + }, + { + "epoch": 2.7479178692174484, + "grad_norm": 2.169614553451538, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7586413621902466, + "num_tokens": 891729145.0, + "step": 33324 + }, + { + "epoch": 2.7480003298425, + "grad_norm": 1.7990473508834839, + "learning_rate": 1e-06, + "loss": 0.607, + "mean_token_accuracy": 0.7961430549621582, + "num_tokens": 891766532.0, + "step": 33325 + }, + { + "epoch": 2.748082790467552, + "grad_norm": 2.211318016052246, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7442705035209656, + "num_tokens": 891796469.0, + "step": 33326 + }, + { + "epoch": 2.7481652510926033, + "grad_norm": 2.1872689723968506, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7741657495498657, + "num_tokens": 891825738.0, + "step": 33327 + }, + { + "epoch": 2.7482477117176547, + "grad_norm": 2.0790717601776123, + "learning_rate": 1e-06, + "loss": 0.5811, + "mean_token_accuracy": 0.7971975803375244, + "num_tokens": 891855135.0, + "step": 33328 + }, + { + "epoch": 2.7483301723427065, + "grad_norm": 2.121089220046997, + "learning_rate": 1e-06, + "loss": 0.6742, + "mean_token_accuracy": 0.7752170562744141, + "num_tokens": 891886416.0, + "step": 33329 + }, + { + "epoch": 2.748412632967758, + "grad_norm": 2.1002867221832275, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.7705078125, + "num_tokens": 891917453.0, + "step": 33330 + }, + { + "epoch": 2.748495093592809, + "grad_norm": 2.7489287853240967, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7695227265357971, + "num_tokens": 891938102.0, + "step": 33331 + }, + { + "epoch": 2.748577554217861, + "grad_norm": 2.28157639503479, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7512112855911255, + "num_tokens": 891963809.0, + "step": 33332 + }, + { + "epoch": 2.748660014842913, + "grad_norm": 2.290410280227661, + "learning_rate": 1e-06, + "loss": 0.7567, + "mean_token_accuracy": 0.7560216784477234, + "num_tokens": 891990417.0, + "step": 33333 + }, + { + "epoch": 2.748742475467964, + "grad_norm": 2.2648377418518066, + "learning_rate": 1e-06, + "loss": 0.6335, + "mean_token_accuracy": 0.7966338992118835, + "num_tokens": 892015568.0, + "step": 33334 + }, + { + "epoch": 2.7488249360930155, + "grad_norm": 2.0638980865478516, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7844786047935486, + "num_tokens": 892045471.0, + "step": 33335 + }, + { + "epoch": 2.7489073967180673, + "grad_norm": 1.8981798887252808, + "learning_rate": 1e-06, + "loss": 0.5867, + "mean_token_accuracy": 0.8099244832992554, + "num_tokens": 892077505.0, + "step": 33336 + }, + { + "epoch": 2.7489898573431186, + "grad_norm": 1.9983439445495605, + "learning_rate": 1e-06, + "loss": 0.6063, + "mean_token_accuracy": 0.7998928427696228, + "num_tokens": 892110873.0, + "step": 33337 + }, + { + "epoch": 2.74907231796817, + "grad_norm": 2.058972120285034, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7843602895736694, + "num_tokens": 892139282.0, + "step": 33338 + }, + { + "epoch": 2.7491547785932218, + "grad_norm": 2.3866806030273438, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.761478066444397, + "num_tokens": 892164994.0, + "step": 33339 + }, + { + "epoch": 2.749237239218273, + "grad_norm": 2.188717842102051, + "learning_rate": 1e-06, + "loss": 0.6635, + "mean_token_accuracy": 0.775275707244873, + "num_tokens": 892192591.0, + "step": 33340 + }, + { + "epoch": 2.749319699843325, + "grad_norm": 2.189185619354248, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7593017816543579, + "num_tokens": 892222032.0, + "step": 33341 + }, + { + "epoch": 2.7494021604683763, + "grad_norm": 2.468846082687378, + "learning_rate": 1e-06, + "loss": 0.7208, + "mean_token_accuracy": 0.7639405131340027, + "num_tokens": 892244433.0, + "step": 33342 + }, + { + "epoch": 2.749484621093428, + "grad_norm": 1.8348450660705566, + "learning_rate": 1e-06, + "loss": 0.5883, + "mean_token_accuracy": 0.8097745180130005, + "num_tokens": 892277650.0, + "step": 33343 + }, + { + "epoch": 2.7495670817184794, + "grad_norm": 2.300954818725586, + "learning_rate": 1e-06, + "loss": 0.6631, + "mean_token_accuracy": 0.7881894111633301, + "num_tokens": 892303335.0, + "step": 33344 + }, + { + "epoch": 2.7496495423435308, + "grad_norm": 2.215902328491211, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7687501907348633, + "num_tokens": 892329985.0, + "step": 33345 + }, + { + "epoch": 2.7497320029685826, + "grad_norm": 2.118959903717041, + "learning_rate": 1e-06, + "loss": 0.6115, + "mean_token_accuracy": 0.7951890230178833, + "num_tokens": 892356847.0, + "step": 33346 + }, + { + "epoch": 2.749814463593634, + "grad_norm": 2.113051652908325, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.7449233531951904, + "num_tokens": 892386709.0, + "step": 33347 + }, + { + "epoch": 2.7498969242186857, + "grad_norm": 2.188145875930786, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7699854373931885, + "num_tokens": 892413768.0, + "step": 33348 + }, + { + "epoch": 2.749979384843737, + "grad_norm": 2.4776666164398193, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7580788135528564, + "num_tokens": 892438416.0, + "step": 33349 + }, + { + "epoch": 2.750061845468789, + "grad_norm": 2.3058152198791504, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.7720530033111572, + "num_tokens": 892461848.0, + "step": 33350 + }, + { + "epoch": 2.75014430609384, + "grad_norm": 2.3864898681640625, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7711184024810791, + "num_tokens": 892485703.0, + "step": 33351 + }, + { + "epoch": 2.7502267667188915, + "grad_norm": 2.15381121635437, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7353285551071167, + "num_tokens": 892514468.0, + "step": 33352 + }, + { + "epoch": 2.7503092273439433, + "grad_norm": 1.838631272315979, + "learning_rate": 1e-06, + "loss": 0.5497, + "mean_token_accuracy": 0.822263777256012, + "num_tokens": 892548474.0, + "step": 33353 + }, + { + "epoch": 2.7503916879689947, + "grad_norm": 2.307642698287964, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7511125802993774, + "num_tokens": 892575674.0, + "step": 33354 + }, + { + "epoch": 2.7504741485940465, + "grad_norm": 2.3980419635772705, + "learning_rate": 1e-06, + "loss": 0.741, + "mean_token_accuracy": 0.7533092498779297, + "num_tokens": 892602207.0, + "step": 33355 + }, + { + "epoch": 2.750556609219098, + "grad_norm": 2.213120698928833, + "learning_rate": 1e-06, + "loss": 0.6374, + "mean_token_accuracy": 0.779580295085907, + "num_tokens": 892629260.0, + "step": 33356 + }, + { + "epoch": 2.7506390698441496, + "grad_norm": 2.227890968322754, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7370882034301758, + "num_tokens": 892658867.0, + "step": 33357 + }, + { + "epoch": 2.750721530469201, + "grad_norm": 2.144463539123535, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.7774271965026855, + "num_tokens": 892687162.0, + "step": 33358 + }, + { + "epoch": 2.7508039910942523, + "grad_norm": 1.8670333623886108, + "learning_rate": 1e-06, + "loss": 0.5404, + "mean_token_accuracy": 0.8092548847198486, + "num_tokens": 892721783.0, + "step": 33359 + }, + { + "epoch": 2.750886451719304, + "grad_norm": 2.074164390563965, + "learning_rate": 1e-06, + "loss": 0.6604, + "mean_token_accuracy": 0.7699840068817139, + "num_tokens": 892748859.0, + "step": 33360 + }, + { + "epoch": 2.7509689123443555, + "grad_norm": 1.9263705015182495, + "learning_rate": 1e-06, + "loss": 0.6214, + "mean_token_accuracy": 0.791994035243988, + "num_tokens": 892783890.0, + "step": 33361 + }, + { + "epoch": 2.7510513729694073, + "grad_norm": 2.226229429244995, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.7883331775665283, + "num_tokens": 892812270.0, + "step": 33362 + }, + { + "epoch": 2.7511338335944586, + "grad_norm": 1.8455387353897095, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7724604606628418, + "num_tokens": 892848158.0, + "step": 33363 + }, + { + "epoch": 2.7512162942195104, + "grad_norm": 2.0206098556518555, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7910650372505188, + "num_tokens": 892877663.0, + "step": 33364 + }, + { + "epoch": 2.7512987548445618, + "grad_norm": 2.112428665161133, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7457884550094604, + "num_tokens": 892908823.0, + "step": 33365 + }, + { + "epoch": 2.751381215469613, + "grad_norm": 2.65891170501709, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7589136362075806, + "num_tokens": 892929640.0, + "step": 33366 + }, + { + "epoch": 2.751463676094665, + "grad_norm": 2.200106620788574, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7278506755828857, + "num_tokens": 892959618.0, + "step": 33367 + }, + { + "epoch": 2.7515461367197163, + "grad_norm": 2.026649236679077, + "learning_rate": 1e-06, + "loss": 0.5545, + "mean_token_accuracy": 0.8069602251052856, + "num_tokens": 892987744.0, + "step": 33368 + }, + { + "epoch": 2.7516285973447676, + "grad_norm": 2.4208757877349854, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7229773998260498, + "num_tokens": 893013286.0, + "step": 33369 + }, + { + "epoch": 2.7517110579698194, + "grad_norm": 2.4210429191589355, + "learning_rate": 1e-06, + "loss": 0.7844, + "mean_token_accuracy": 0.7559285759925842, + "num_tokens": 893037311.0, + "step": 33370 + }, + { + "epoch": 2.751793518594871, + "grad_norm": 2.217195510864258, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7462750673294067, + "num_tokens": 893067310.0, + "step": 33371 + }, + { + "epoch": 2.7518759792199226, + "grad_norm": 2.285161256790161, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.7917583584785461, + "num_tokens": 893091532.0, + "step": 33372 + }, + { + "epoch": 2.751958439844974, + "grad_norm": 1.9159983396530151, + "learning_rate": 1e-06, + "loss": 0.7366, + "mean_token_accuracy": 0.7628828287124634, + "num_tokens": 893124602.0, + "step": 33373 + }, + { + "epoch": 2.7520409004700257, + "grad_norm": 2.263378858566284, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7710797190666199, + "num_tokens": 893152456.0, + "step": 33374 + }, + { + "epoch": 2.752123361095077, + "grad_norm": 2.2009999752044678, + "learning_rate": 1e-06, + "loss": 0.7681, + "mean_token_accuracy": 0.7589212656021118, + "num_tokens": 893181825.0, + "step": 33375 + }, + { + "epoch": 2.7522058217201284, + "grad_norm": 2.099184274673462, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.7858399152755737, + "num_tokens": 893214412.0, + "step": 33376 + }, + { + "epoch": 2.75228828234518, + "grad_norm": 2.0671820640563965, + "learning_rate": 1e-06, + "loss": 0.6003, + "mean_token_accuracy": 0.7988049983978271, + "num_tokens": 893244152.0, + "step": 33377 + }, + { + "epoch": 2.752370742970232, + "grad_norm": 2.2115707397460938, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.7879056334495544, + "num_tokens": 893271460.0, + "step": 33378 + }, + { + "epoch": 2.7524532035952833, + "grad_norm": 2.284222364425659, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.793746829032898, + "num_tokens": 893300664.0, + "step": 33379 + }, + { + "epoch": 2.7525356642203347, + "grad_norm": 1.9296823740005493, + "learning_rate": 1e-06, + "loss": 0.6076, + "mean_token_accuracy": 0.7972471714019775, + "num_tokens": 893331531.0, + "step": 33380 + }, + { + "epoch": 2.7526181248453865, + "grad_norm": 1.980624794960022, + "learning_rate": 1e-06, + "loss": 0.5523, + "mean_token_accuracy": 0.8109161853790283, + "num_tokens": 893361892.0, + "step": 33381 + }, + { + "epoch": 2.752700585470438, + "grad_norm": 2.3302109241485596, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7473109364509583, + "num_tokens": 893390061.0, + "step": 33382 + }, + { + "epoch": 2.752783046095489, + "grad_norm": 2.0181314945220947, + "learning_rate": 1e-06, + "loss": 0.7277, + "mean_token_accuracy": 0.7639501690864563, + "num_tokens": 893424204.0, + "step": 33383 + }, + { + "epoch": 2.752865506720541, + "grad_norm": 1.9420615434646606, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7863969802856445, + "num_tokens": 893458366.0, + "step": 33384 + }, + { + "epoch": 2.7529479673455923, + "grad_norm": 2.3237130641937256, + "learning_rate": 1e-06, + "loss": 0.8169, + "mean_token_accuracy": 0.7442957162857056, + "num_tokens": 893484311.0, + "step": 33385 + }, + { + "epoch": 2.753030427970644, + "grad_norm": 2.123178720474243, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7739304304122925, + "num_tokens": 893514218.0, + "step": 33386 + }, + { + "epoch": 2.7531128885956955, + "grad_norm": 2.145120620727539, + "learning_rate": 1e-06, + "loss": 0.6863, + "mean_token_accuracy": 0.7816516160964966, + "num_tokens": 893546456.0, + "step": 33387 + }, + { + "epoch": 2.7531953492207473, + "grad_norm": 1.9689404964447021, + "learning_rate": 1e-06, + "loss": 0.6125, + "mean_token_accuracy": 0.7920242547988892, + "num_tokens": 893578390.0, + "step": 33388 + }, + { + "epoch": 2.7532778098457986, + "grad_norm": 2.115424633026123, + "learning_rate": 1e-06, + "loss": 0.6454, + "mean_token_accuracy": 0.7866415977478027, + "num_tokens": 893611997.0, + "step": 33389 + }, + { + "epoch": 2.75336027047085, + "grad_norm": 2.0347023010253906, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7596122026443481, + "num_tokens": 893646267.0, + "step": 33390 + }, + { + "epoch": 2.7534427310959018, + "grad_norm": 2.189568281173706, + "learning_rate": 1e-06, + "loss": 0.7218, + "mean_token_accuracy": 0.7770137786865234, + "num_tokens": 893676457.0, + "step": 33391 + }, + { + "epoch": 2.753525191720953, + "grad_norm": 2.056396722793579, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.7488220930099487, + "num_tokens": 893708830.0, + "step": 33392 + }, + { + "epoch": 2.753607652346005, + "grad_norm": 2.0353477001190186, + "learning_rate": 1e-06, + "loss": 0.7247, + "mean_token_accuracy": 0.7645043730735779, + "num_tokens": 893742706.0, + "step": 33393 + }, + { + "epoch": 2.7536901129710563, + "grad_norm": 2.1612627506256104, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.7826412320137024, + "num_tokens": 893775455.0, + "step": 33394 + }, + { + "epoch": 2.753772573596108, + "grad_norm": 2.2807090282440186, + "learning_rate": 1e-06, + "loss": 0.6786, + "mean_token_accuracy": 0.7799956798553467, + "num_tokens": 893801694.0, + "step": 33395 + }, + { + "epoch": 2.7538550342211594, + "grad_norm": 2.2607665061950684, + "learning_rate": 1e-06, + "loss": 0.6678, + "mean_token_accuracy": 0.7778298854827881, + "num_tokens": 893828478.0, + "step": 33396 + }, + { + "epoch": 2.7539374948462108, + "grad_norm": 1.9479317665100098, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7463228702545166, + "num_tokens": 893863908.0, + "step": 33397 + }, + { + "epoch": 2.7540199554712625, + "grad_norm": 2.220545768737793, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7636991739273071, + "num_tokens": 893893470.0, + "step": 33398 + }, + { + "epoch": 2.754102416096314, + "grad_norm": 2.1534531116485596, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7447770237922668, + "num_tokens": 893922705.0, + "step": 33399 + }, + { + "epoch": 2.7541848767213657, + "grad_norm": 2.1224775314331055, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7743653655052185, + "num_tokens": 893950727.0, + "step": 33400 + }, + { + "epoch": 2.754267337346417, + "grad_norm": 1.837251901626587, + "learning_rate": 1e-06, + "loss": 0.6295, + "mean_token_accuracy": 0.7928730249404907, + "num_tokens": 893985993.0, + "step": 33401 + }, + { + "epoch": 2.754349797971469, + "grad_norm": 2.187472105026245, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.748778223991394, + "num_tokens": 894015064.0, + "step": 33402 + }, + { + "epoch": 2.75443225859652, + "grad_norm": 2.234933376312256, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7466121912002563, + "num_tokens": 894043480.0, + "step": 33403 + }, + { + "epoch": 2.7545147192215715, + "grad_norm": 2.0890164375305176, + "learning_rate": 1e-06, + "loss": 0.7652, + "mean_token_accuracy": 0.7601057887077332, + "num_tokens": 894075194.0, + "step": 33404 + }, + { + "epoch": 2.7545971798466233, + "grad_norm": 2.2147185802459717, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.744236946105957, + "num_tokens": 894102949.0, + "step": 33405 + }, + { + "epoch": 2.7546796404716747, + "grad_norm": 2.112802505493164, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.790460467338562, + "num_tokens": 894134600.0, + "step": 33406 + }, + { + "epoch": 2.7547621010967265, + "grad_norm": 2.064483404159546, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7785024642944336, + "num_tokens": 894164596.0, + "step": 33407 + }, + { + "epoch": 2.754844561721778, + "grad_norm": 1.7506239414215088, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7778784036636353, + "num_tokens": 894206050.0, + "step": 33408 + }, + { + "epoch": 2.7549270223468296, + "grad_norm": 2.5931358337402344, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7616839408874512, + "num_tokens": 894230012.0, + "step": 33409 + }, + { + "epoch": 2.755009482971881, + "grad_norm": 1.9597375392913818, + "learning_rate": 1e-06, + "loss": 0.5449, + "mean_token_accuracy": 0.821561872959137, + "num_tokens": 894261566.0, + "step": 33410 + }, + { + "epoch": 2.7550919435969323, + "grad_norm": 2.1324594020843506, + "learning_rate": 1e-06, + "loss": 0.7584, + "mean_token_accuracy": 0.7655534744262695, + "num_tokens": 894289566.0, + "step": 33411 + }, + { + "epoch": 2.755174404221984, + "grad_norm": 2.2470598220825195, + "learning_rate": 1e-06, + "loss": 0.6363, + "mean_token_accuracy": 0.787887692451477, + "num_tokens": 894316179.0, + "step": 33412 + }, + { + "epoch": 2.7552568648470355, + "grad_norm": 1.8909531831741333, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.756971001625061, + "num_tokens": 894349537.0, + "step": 33413 + }, + { + "epoch": 2.755339325472087, + "grad_norm": 2.5106186866760254, + "learning_rate": 1e-06, + "loss": 0.6006, + "mean_token_accuracy": 0.7935150861740112, + "num_tokens": 894372028.0, + "step": 33414 + }, + { + "epoch": 2.7554217860971386, + "grad_norm": 2.1921756267547607, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7348523139953613, + "num_tokens": 894400725.0, + "step": 33415 + }, + { + "epoch": 2.7555042467221904, + "grad_norm": 2.3066771030426025, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7710040807723999, + "num_tokens": 894426018.0, + "step": 33416 + }, + { + "epoch": 2.7555867073472418, + "grad_norm": 1.9312794208526611, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7636419534683228, + "num_tokens": 894459898.0, + "step": 33417 + }, + { + "epoch": 2.755669167972293, + "grad_norm": 1.9025347232818604, + "learning_rate": 1e-06, + "loss": 0.6007, + "mean_token_accuracy": 0.795258641242981, + "num_tokens": 894493173.0, + "step": 33418 + }, + { + "epoch": 2.755751628597345, + "grad_norm": 2.137000799179077, + "learning_rate": 1e-06, + "loss": 0.6257, + "mean_token_accuracy": 0.7860637903213501, + "num_tokens": 894521319.0, + "step": 33419 + }, + { + "epoch": 2.7558340892223963, + "grad_norm": 2.2167506217956543, + "learning_rate": 1e-06, + "loss": 0.6764, + "mean_token_accuracy": 0.7775721549987793, + "num_tokens": 894548211.0, + "step": 33420 + }, + { + "epoch": 2.7559165498474476, + "grad_norm": 1.957401156425476, + "learning_rate": 1e-06, + "loss": 0.5058, + "mean_token_accuracy": 0.8262104988098145, + "num_tokens": 894575663.0, + "step": 33421 + }, + { + "epoch": 2.7559990104724994, + "grad_norm": 2.0349152088165283, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7717846632003784, + "num_tokens": 894609331.0, + "step": 33422 + }, + { + "epoch": 2.756081471097551, + "grad_norm": 2.1115968227386475, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7876281142234802, + "num_tokens": 894641980.0, + "step": 33423 + }, + { + "epoch": 2.7561639317226025, + "grad_norm": 2.273585796356201, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.770161509513855, + "num_tokens": 894669171.0, + "step": 33424 + }, + { + "epoch": 2.756246392347654, + "grad_norm": 1.881342887878418, + "learning_rate": 1e-06, + "loss": 0.5491, + "mean_token_accuracy": 0.8234161138534546, + "num_tokens": 894702909.0, + "step": 33425 + }, + { + "epoch": 2.7563288529727057, + "grad_norm": 1.9044502973556519, + "learning_rate": 1e-06, + "loss": 0.6167, + "mean_token_accuracy": 0.7997328042984009, + "num_tokens": 894740597.0, + "step": 33426 + }, + { + "epoch": 2.756411313597757, + "grad_norm": 2.2587475776672363, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7503034472465515, + "num_tokens": 894767925.0, + "step": 33427 + }, + { + "epoch": 2.7564937742228084, + "grad_norm": 2.0612940788269043, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.764970064163208, + "num_tokens": 894799954.0, + "step": 33428 + }, + { + "epoch": 2.75657623484786, + "grad_norm": 1.9612189531326294, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7376954555511475, + "num_tokens": 894833737.0, + "step": 33429 + }, + { + "epoch": 2.7566586954729115, + "grad_norm": 2.2417309284210205, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7637313604354858, + "num_tokens": 894860367.0, + "step": 33430 + }, + { + "epoch": 2.7567411560979633, + "grad_norm": 2.2450926303863525, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7391079068183899, + "num_tokens": 894887799.0, + "step": 33431 + }, + { + "epoch": 2.7568236167230147, + "grad_norm": 2.2295970916748047, + "learning_rate": 1e-06, + "loss": 0.6391, + "mean_token_accuracy": 0.7862133383750916, + "num_tokens": 894912669.0, + "step": 33432 + }, + { + "epoch": 2.7569060773480665, + "grad_norm": 2.239380359649658, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7467067241668701, + "num_tokens": 894941995.0, + "step": 33433 + }, + { + "epoch": 2.756988537973118, + "grad_norm": 2.0506255626678467, + "learning_rate": 1e-06, + "loss": 0.7738, + "mean_token_accuracy": 0.7524028420448303, + "num_tokens": 894973775.0, + "step": 33434 + }, + { + "epoch": 2.757070998598169, + "grad_norm": 1.9308362007141113, + "learning_rate": 1e-06, + "loss": 0.6007, + "mean_token_accuracy": 0.8054132461547852, + "num_tokens": 895006234.0, + "step": 33435 + }, + { + "epoch": 2.757153459223221, + "grad_norm": 2.001499652862549, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.772286057472229, + "num_tokens": 895037292.0, + "step": 33436 + }, + { + "epoch": 2.7572359198482723, + "grad_norm": 1.984297752380371, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7560762166976929, + "num_tokens": 895068950.0, + "step": 33437 + }, + { + "epoch": 2.757318380473324, + "grad_norm": 2.180485725402832, + "learning_rate": 1e-06, + "loss": 0.5174, + "mean_token_accuracy": 0.8209309577941895, + "num_tokens": 895096240.0, + "step": 33438 + }, + { + "epoch": 2.7574008410983755, + "grad_norm": 2.1372592449188232, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.7559299468994141, + "num_tokens": 895126184.0, + "step": 33439 + }, + { + "epoch": 2.7574833017234273, + "grad_norm": 2.4108211994171143, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7351381778717041, + "num_tokens": 895152714.0, + "step": 33440 + }, + { + "epoch": 2.7575657623484786, + "grad_norm": 2.09578275680542, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.7787209749221802, + "num_tokens": 895181870.0, + "step": 33441 + }, + { + "epoch": 2.75764822297353, + "grad_norm": 1.8582861423492432, + "learning_rate": 1e-06, + "loss": 0.6103, + "mean_token_accuracy": 0.7867624759674072, + "num_tokens": 895215367.0, + "step": 33442 + }, + { + "epoch": 2.7577306835985818, + "grad_norm": 2.3237874507904053, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7840998768806458, + "num_tokens": 895239508.0, + "step": 33443 + }, + { + "epoch": 2.757813144223633, + "grad_norm": 2.217662811279297, + "learning_rate": 1e-06, + "loss": 0.585, + "mean_token_accuracy": 0.8057668805122375, + "num_tokens": 895266678.0, + "step": 33444 + }, + { + "epoch": 2.757895604848685, + "grad_norm": 1.9721217155456543, + "learning_rate": 1e-06, + "loss": 0.5792, + "mean_token_accuracy": 0.7944577932357788, + "num_tokens": 895295712.0, + "step": 33445 + }, + { + "epoch": 2.7579780654737363, + "grad_norm": 2.1135408878326416, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.7977526783943176, + "num_tokens": 895327280.0, + "step": 33446 + }, + { + "epoch": 2.758060526098788, + "grad_norm": 2.1408135890960693, + "learning_rate": 1e-06, + "loss": 0.5494, + "mean_token_accuracy": 0.8052520751953125, + "num_tokens": 895355599.0, + "step": 33447 + }, + { + "epoch": 2.7581429867238394, + "grad_norm": 2.26467227935791, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.7458692193031311, + "num_tokens": 895381044.0, + "step": 33448 + }, + { + "epoch": 2.7582254473488907, + "grad_norm": 1.959250569343567, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7257319688796997, + "num_tokens": 895419621.0, + "step": 33449 + }, + { + "epoch": 2.7583079079739425, + "grad_norm": 2.102294683456421, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7787250280380249, + "num_tokens": 895450455.0, + "step": 33450 + }, + { + "epoch": 2.758390368598994, + "grad_norm": 2.1826436519622803, + "learning_rate": 1e-06, + "loss": 0.5968, + "mean_token_accuracy": 0.7977955341339111, + "num_tokens": 895476589.0, + "step": 33451 + }, + { + "epoch": 2.7584728292240452, + "grad_norm": 2.008726119995117, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7727393507957458, + "num_tokens": 895508357.0, + "step": 33452 + }, + { + "epoch": 2.758555289849097, + "grad_norm": 1.9973284006118774, + "learning_rate": 1e-06, + "loss": 0.5622, + "mean_token_accuracy": 0.8041139841079712, + "num_tokens": 895537926.0, + "step": 33453 + }, + { + "epoch": 2.758637750474149, + "grad_norm": 1.9768608808517456, + "learning_rate": 1e-06, + "loss": 0.609, + "mean_token_accuracy": 0.7936031222343445, + "num_tokens": 895569006.0, + "step": 33454 + }, + { + "epoch": 2.7587202110992, + "grad_norm": 2.200916290283203, + "learning_rate": 1e-06, + "loss": 0.5809, + "mean_token_accuracy": 0.813534677028656, + "num_tokens": 895596184.0, + "step": 33455 + }, + { + "epoch": 2.7588026717242515, + "grad_norm": 2.0497090816497803, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7576317191123962, + "num_tokens": 895626966.0, + "step": 33456 + }, + { + "epoch": 2.7588851323493033, + "grad_norm": 2.0676114559173584, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7484758496284485, + "num_tokens": 895656528.0, + "step": 33457 + }, + { + "epoch": 2.7589675929743547, + "grad_norm": 2.131373882293701, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.7877263426780701, + "num_tokens": 895684980.0, + "step": 33458 + }, + { + "epoch": 2.759050053599406, + "grad_norm": 2.1308705806732178, + "learning_rate": 1e-06, + "loss": 0.6039, + "mean_token_accuracy": 0.8025078177452087, + "num_tokens": 895711945.0, + "step": 33459 + }, + { + "epoch": 2.759132514224458, + "grad_norm": 1.9268885850906372, + "learning_rate": 1e-06, + "loss": 0.6872, + "mean_token_accuracy": 0.7720373272895813, + "num_tokens": 895744875.0, + "step": 33460 + }, + { + "epoch": 2.7592149748495096, + "grad_norm": 1.96170175075531, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.7912547588348389, + "num_tokens": 895781124.0, + "step": 33461 + }, + { + "epoch": 2.759297435474561, + "grad_norm": 2.0295956134796143, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.7887974977493286, + "num_tokens": 895809221.0, + "step": 33462 + }, + { + "epoch": 2.7593798960996123, + "grad_norm": 2.103383779525757, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.7873715162277222, + "num_tokens": 895837188.0, + "step": 33463 + }, + { + "epoch": 2.759462356724664, + "grad_norm": 2.083712100982666, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7556507587432861, + "num_tokens": 895869421.0, + "step": 33464 + }, + { + "epoch": 2.7595448173497155, + "grad_norm": 2.1391515731811523, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7816119194030762, + "num_tokens": 895898478.0, + "step": 33465 + }, + { + "epoch": 2.759627277974767, + "grad_norm": 1.751196026802063, + "learning_rate": 1e-06, + "loss": 0.5634, + "mean_token_accuracy": 0.8135930895805359, + "num_tokens": 895935478.0, + "step": 33466 + }, + { + "epoch": 2.7597097385998186, + "grad_norm": 2.1761410236358643, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7656217217445374, + "num_tokens": 895963414.0, + "step": 33467 + }, + { + "epoch": 2.7597921992248704, + "grad_norm": 2.150446891784668, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7502337694168091, + "num_tokens": 895990224.0, + "step": 33468 + }, + { + "epoch": 2.7598746598499218, + "grad_norm": 1.9971033334732056, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7485371828079224, + "num_tokens": 896023938.0, + "step": 33469 + }, + { + "epoch": 2.759957120474973, + "grad_norm": 1.9673616886138916, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7582932710647583, + "num_tokens": 896057156.0, + "step": 33470 + }, + { + "epoch": 2.760039581100025, + "grad_norm": 2.058833599090576, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7506697177886963, + "num_tokens": 896089243.0, + "step": 33471 + }, + { + "epoch": 2.7601220417250762, + "grad_norm": 2.1610684394836426, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.7945859432220459, + "num_tokens": 896117638.0, + "step": 33472 + }, + { + "epoch": 2.7602045023501276, + "grad_norm": 2.6184539794921875, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.770237922668457, + "num_tokens": 896139219.0, + "step": 33473 + }, + { + "epoch": 2.7602869629751794, + "grad_norm": 2.286085844039917, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.7892122268676758, + "num_tokens": 896165234.0, + "step": 33474 + }, + { + "epoch": 2.7603694236002307, + "grad_norm": 2.1438143253326416, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7419866919517517, + "num_tokens": 896196682.0, + "step": 33475 + }, + { + "epoch": 2.7604518842252825, + "grad_norm": 2.2008042335510254, + "learning_rate": 1e-06, + "loss": 0.7137, + "mean_token_accuracy": 0.7650689482688904, + "num_tokens": 896226546.0, + "step": 33476 + }, + { + "epoch": 2.760534344850334, + "grad_norm": 2.0497846603393555, + "learning_rate": 1e-06, + "loss": 0.6093, + "mean_token_accuracy": 0.7869575023651123, + "num_tokens": 896259073.0, + "step": 33477 + }, + { + "epoch": 2.7606168054753857, + "grad_norm": 2.1188225746154785, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7764164209365845, + "num_tokens": 896291532.0, + "step": 33478 + }, + { + "epoch": 2.760699266100437, + "grad_norm": 2.2333226203918457, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.7752224802970886, + "num_tokens": 896319482.0, + "step": 33479 + }, + { + "epoch": 2.7607817267254884, + "grad_norm": 2.105057716369629, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7408442497253418, + "num_tokens": 896350509.0, + "step": 33480 + }, + { + "epoch": 2.76086418735054, + "grad_norm": 2.246795415878296, + "learning_rate": 1e-06, + "loss": 0.6018, + "mean_token_accuracy": 0.796958327293396, + "num_tokens": 896378507.0, + "step": 33481 + }, + { + "epoch": 2.7609466479755915, + "grad_norm": 2.1038753986358643, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7453817129135132, + "num_tokens": 896410180.0, + "step": 33482 + }, + { + "epoch": 2.7610291086006433, + "grad_norm": 1.8942770957946777, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7552825212478638, + "num_tokens": 896444379.0, + "step": 33483 + }, + { + "epoch": 2.7611115692256947, + "grad_norm": 2.179295301437378, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.7797657251358032, + "num_tokens": 896475349.0, + "step": 33484 + }, + { + "epoch": 2.7611940298507465, + "grad_norm": 1.966296672821045, + "learning_rate": 1e-06, + "loss": 0.5016, + "mean_token_accuracy": 0.8256861567497253, + "num_tokens": 896505861.0, + "step": 33485 + }, + { + "epoch": 2.761276490475798, + "grad_norm": 1.990282654762268, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.7658714056015015, + "num_tokens": 896540032.0, + "step": 33486 + }, + { + "epoch": 2.761358951100849, + "grad_norm": 2.145892381668091, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7360053062438965, + "num_tokens": 896570228.0, + "step": 33487 + }, + { + "epoch": 2.761441411725901, + "grad_norm": 2.197389841079712, + "learning_rate": 1e-06, + "loss": 0.633, + "mean_token_accuracy": 0.7849103212356567, + "num_tokens": 896597591.0, + "step": 33488 + }, + { + "epoch": 2.7615238723509523, + "grad_norm": 1.9272626638412476, + "learning_rate": 1e-06, + "loss": 0.7182, + "mean_token_accuracy": 0.7677842974662781, + "num_tokens": 896633070.0, + "step": 33489 + }, + { + "epoch": 2.761606332976004, + "grad_norm": 2.2066776752471924, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.7922489643096924, + "num_tokens": 896659624.0, + "step": 33490 + }, + { + "epoch": 2.7616887936010555, + "grad_norm": 2.4096479415893555, + "learning_rate": 1e-06, + "loss": 0.6137, + "mean_token_accuracy": 0.7823233604431152, + "num_tokens": 896683943.0, + "step": 33491 + }, + { + "epoch": 2.7617712542261073, + "grad_norm": 2.1659891605377197, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7604780197143555, + "num_tokens": 896713817.0, + "step": 33492 + }, + { + "epoch": 2.7618537148511586, + "grad_norm": 2.3065743446350098, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.7891242504119873, + "num_tokens": 896744543.0, + "step": 33493 + }, + { + "epoch": 2.76193617547621, + "grad_norm": 1.7927767038345337, + "learning_rate": 1e-06, + "loss": 0.4319, + "mean_token_accuracy": 0.8519014120101929, + "num_tokens": 896778275.0, + "step": 33494 + }, + { + "epoch": 2.7620186361012617, + "grad_norm": 2.362877607345581, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7720638513565063, + "num_tokens": 896803845.0, + "step": 33495 + }, + { + "epoch": 2.762101096726313, + "grad_norm": 2.1646134853363037, + "learning_rate": 1e-06, + "loss": 0.7098, + "mean_token_accuracy": 0.7802810072898865, + "num_tokens": 896833362.0, + "step": 33496 + }, + { + "epoch": 2.7621835573513644, + "grad_norm": 2.390080690383911, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7493759393692017, + "num_tokens": 896859039.0, + "step": 33497 + }, + { + "epoch": 2.7622660179764162, + "grad_norm": 2.1436474323272705, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7641758322715759, + "num_tokens": 896886886.0, + "step": 33498 + }, + { + "epoch": 2.762348478601468, + "grad_norm": 1.9070888757705688, + "learning_rate": 1e-06, + "loss": 0.6359, + "mean_token_accuracy": 0.7962661385536194, + "num_tokens": 896918444.0, + "step": 33499 + }, + { + "epoch": 2.7624309392265194, + "grad_norm": 2.0423965454101562, + "learning_rate": 1e-06, + "loss": 0.6558, + "mean_token_accuracy": 0.775777280330658, + "num_tokens": 896949832.0, + "step": 33500 + }, + { + "epoch": 2.7625133998515707, + "grad_norm": 2.1541199684143066, + "learning_rate": 1e-06, + "loss": 0.5307, + "mean_token_accuracy": 0.8097583055496216, + "num_tokens": 896975343.0, + "step": 33501 + }, + { + "epoch": 2.7625958604766225, + "grad_norm": 1.878281593322754, + "learning_rate": 1e-06, + "loss": 0.5985, + "mean_token_accuracy": 0.8040964603424072, + "num_tokens": 897011621.0, + "step": 33502 + }, + { + "epoch": 2.762678321101674, + "grad_norm": 2.419625997543335, + "learning_rate": 1e-06, + "loss": 0.7988, + "mean_token_accuracy": 0.7516354918479919, + "num_tokens": 897035298.0, + "step": 33503 + }, + { + "epoch": 2.7627607817267252, + "grad_norm": 2.0904791355133057, + "learning_rate": 1e-06, + "loss": 0.6356, + "mean_token_accuracy": 0.7948063015937805, + "num_tokens": 897062433.0, + "step": 33504 + }, + { + "epoch": 2.762843242351777, + "grad_norm": 1.9182705879211426, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.7866326570510864, + "num_tokens": 897096954.0, + "step": 33505 + }, + { + "epoch": 2.762925702976829, + "grad_norm": 2.222700595855713, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.759734034538269, + "num_tokens": 897125415.0, + "step": 33506 + }, + { + "epoch": 2.76300816360188, + "grad_norm": 2.1048920154571533, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7427191138267517, + "num_tokens": 897157935.0, + "step": 33507 + }, + { + "epoch": 2.7630906242269315, + "grad_norm": 2.497830390930176, + "learning_rate": 1e-06, + "loss": 0.6632, + "mean_token_accuracy": 0.7805701494216919, + "num_tokens": 897180288.0, + "step": 33508 + }, + { + "epoch": 2.7631730848519833, + "grad_norm": 2.1249780654907227, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.7926541566848755, + "num_tokens": 897210328.0, + "step": 33509 + }, + { + "epoch": 2.7632555454770347, + "grad_norm": 2.1374847888946533, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7821040749549866, + "num_tokens": 897236915.0, + "step": 33510 + }, + { + "epoch": 2.763338006102086, + "grad_norm": 1.7741070985794067, + "learning_rate": 1e-06, + "loss": 0.6463, + "mean_token_accuracy": 0.7845349311828613, + "num_tokens": 897278529.0, + "step": 33511 + }, + { + "epoch": 2.763420466727138, + "grad_norm": 2.3281304836273193, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7697319984436035, + "num_tokens": 897303777.0, + "step": 33512 + }, + { + "epoch": 2.763502927352189, + "grad_norm": 2.0941953659057617, + "learning_rate": 1e-06, + "loss": 0.5547, + "mean_token_accuracy": 0.8136101961135864, + "num_tokens": 897333004.0, + "step": 33513 + }, + { + "epoch": 2.763585387977241, + "grad_norm": 2.240583658218384, + "learning_rate": 1e-06, + "loss": 0.6446, + "mean_token_accuracy": 0.7796564698219299, + "num_tokens": 897360371.0, + "step": 33514 + }, + { + "epoch": 2.7636678486022923, + "grad_norm": 2.0967774391174316, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.7853780388832092, + "num_tokens": 897390137.0, + "step": 33515 + }, + { + "epoch": 2.763750309227344, + "grad_norm": 2.093196153640747, + "learning_rate": 1e-06, + "loss": 0.6329, + "mean_token_accuracy": 0.7891298532485962, + "num_tokens": 897418437.0, + "step": 33516 + }, + { + "epoch": 2.7638327698523955, + "grad_norm": 1.994391918182373, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7566720843315125, + "num_tokens": 897450533.0, + "step": 33517 + }, + { + "epoch": 2.763915230477447, + "grad_norm": 2.2319912910461426, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.7830073833465576, + "num_tokens": 897479291.0, + "step": 33518 + }, + { + "epoch": 2.7639976911024986, + "grad_norm": 2.2351737022399902, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7784677147865295, + "num_tokens": 897507050.0, + "step": 33519 + }, + { + "epoch": 2.76408015172755, + "grad_norm": 2.3577027320861816, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.7696080207824707, + "num_tokens": 897531196.0, + "step": 33520 + }, + { + "epoch": 2.7641626123526017, + "grad_norm": 2.6629021167755127, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7556247711181641, + "num_tokens": 897552781.0, + "step": 33521 + }, + { + "epoch": 2.764245072977653, + "grad_norm": 1.9902019500732422, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7632384300231934, + "num_tokens": 897583250.0, + "step": 33522 + }, + { + "epoch": 2.764327533602705, + "grad_norm": 2.2989137172698975, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.765142023563385, + "num_tokens": 897608562.0, + "step": 33523 + }, + { + "epoch": 2.7644099942277562, + "grad_norm": 1.93726646900177, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.7606798410415649, + "num_tokens": 897642544.0, + "step": 33524 + }, + { + "epoch": 2.7644924548528076, + "grad_norm": 2.1190621852874756, + "learning_rate": 1e-06, + "loss": 0.7146, + "mean_token_accuracy": 0.762108564376831, + "num_tokens": 897673718.0, + "step": 33525 + }, + { + "epoch": 2.7645749154778594, + "grad_norm": 2.10260009765625, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.747815728187561, + "num_tokens": 897703328.0, + "step": 33526 + }, + { + "epoch": 2.7646573761029107, + "grad_norm": 2.0150258541107178, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.7890268564224243, + "num_tokens": 897733849.0, + "step": 33527 + }, + { + "epoch": 2.7647398367279625, + "grad_norm": 1.9469879865646362, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.7882982492446899, + "num_tokens": 897764144.0, + "step": 33528 + }, + { + "epoch": 2.764822297353014, + "grad_norm": 2.491607189178467, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7693971395492554, + "num_tokens": 897788377.0, + "step": 33529 + }, + { + "epoch": 2.7649047579780657, + "grad_norm": 2.145200490951538, + "learning_rate": 1e-06, + "loss": 0.641, + "mean_token_accuracy": 0.7860783338546753, + "num_tokens": 897818527.0, + "step": 33530 + }, + { + "epoch": 2.764987218603117, + "grad_norm": 1.9885683059692383, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7850770354270935, + "num_tokens": 897851660.0, + "step": 33531 + }, + { + "epoch": 2.7650696792281684, + "grad_norm": 2.021008014678955, + "learning_rate": 1e-06, + "loss": 0.7388, + "mean_token_accuracy": 0.7638390064239502, + "num_tokens": 897883288.0, + "step": 33532 + }, + { + "epoch": 2.76515213985322, + "grad_norm": 2.090967893600464, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.7444838881492615, + "num_tokens": 897916608.0, + "step": 33533 + }, + { + "epoch": 2.7652346004782715, + "grad_norm": 1.7469491958618164, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.7823277711868286, + "num_tokens": 897956362.0, + "step": 33534 + }, + { + "epoch": 2.7653170611033233, + "grad_norm": 2.000162124633789, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.8141950964927673, + "num_tokens": 897984678.0, + "step": 33535 + }, + { + "epoch": 2.7653995217283747, + "grad_norm": 1.912554383277893, + "learning_rate": 1e-06, + "loss": 0.6037, + "mean_token_accuracy": 0.7996562719345093, + "num_tokens": 898017778.0, + "step": 33536 + }, + { + "epoch": 2.7654819823534265, + "grad_norm": 1.8986414670944214, + "learning_rate": 1e-06, + "loss": 0.4525, + "mean_token_accuracy": 0.8336916565895081, + "num_tokens": 898048614.0, + "step": 33537 + }, + { + "epoch": 2.765564442978478, + "grad_norm": 2.064530372619629, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7761561870574951, + "num_tokens": 898076847.0, + "step": 33538 + }, + { + "epoch": 2.765646903603529, + "grad_norm": 2.3908534049987793, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7781442403793335, + "num_tokens": 898101734.0, + "step": 33539 + }, + { + "epoch": 2.765729364228581, + "grad_norm": 1.9529738426208496, + "learning_rate": 1e-06, + "loss": 0.5247, + "mean_token_accuracy": 0.8234696388244629, + "num_tokens": 898131817.0, + "step": 33540 + }, + { + "epoch": 2.7658118248536323, + "grad_norm": 2.2390811443328857, + "learning_rate": 1e-06, + "loss": 0.579, + "mean_token_accuracy": 0.8098866939544678, + "num_tokens": 898157926.0, + "step": 33541 + }, + { + "epoch": 2.7658942854786837, + "grad_norm": 2.1567773818969727, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7768415212631226, + "num_tokens": 898187928.0, + "step": 33542 + }, + { + "epoch": 2.7659767461037355, + "grad_norm": 2.2734737396240234, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7660598158836365, + "num_tokens": 898216005.0, + "step": 33543 + }, + { + "epoch": 2.7660592067287872, + "grad_norm": 2.104505777359009, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7450653314590454, + "num_tokens": 898250025.0, + "step": 33544 + }, + { + "epoch": 2.7661416673538386, + "grad_norm": 1.9985136985778809, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7508175373077393, + "num_tokens": 898281992.0, + "step": 33545 + }, + { + "epoch": 2.76622412797889, + "grad_norm": 2.231602668762207, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7608882188796997, + "num_tokens": 898308918.0, + "step": 33546 + }, + { + "epoch": 2.7663065886039417, + "grad_norm": 1.8797820806503296, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.7892371416091919, + "num_tokens": 898341600.0, + "step": 33547 + }, + { + "epoch": 2.766389049228993, + "grad_norm": 2.1253421306610107, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.7865948677062988, + "num_tokens": 898371664.0, + "step": 33548 + }, + { + "epoch": 2.7664715098540444, + "grad_norm": 1.9711816310882568, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7887465953826904, + "num_tokens": 898404060.0, + "step": 33549 + }, + { + "epoch": 2.7665539704790962, + "grad_norm": 1.899928092956543, + "learning_rate": 1e-06, + "loss": 0.687, + "mean_token_accuracy": 0.7883307933807373, + "num_tokens": 898438773.0, + "step": 33550 + }, + { + "epoch": 2.766636431104148, + "grad_norm": 2.235714912414551, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7780070900917053, + "num_tokens": 898465883.0, + "step": 33551 + }, + { + "epoch": 2.7667188917291994, + "grad_norm": 2.2633790969848633, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7751798629760742, + "num_tokens": 898495175.0, + "step": 33552 + }, + { + "epoch": 2.7668013523542507, + "grad_norm": 2.1429059505462646, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7637830972671509, + "num_tokens": 898525268.0, + "step": 33553 + }, + { + "epoch": 2.7668838129793025, + "grad_norm": 2.4206974506378174, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.7791818380355835, + "num_tokens": 898550151.0, + "step": 33554 + }, + { + "epoch": 2.766966273604354, + "grad_norm": 2.1295809745788574, + "learning_rate": 1e-06, + "loss": 0.7364, + "mean_token_accuracy": 0.7580310702323914, + "num_tokens": 898581587.0, + "step": 33555 + }, + { + "epoch": 2.7670487342294052, + "grad_norm": 1.9648351669311523, + "learning_rate": 1e-06, + "loss": 0.6061, + "mean_token_accuracy": 0.7963215708732605, + "num_tokens": 898610839.0, + "step": 33556 + }, + { + "epoch": 2.767131194854457, + "grad_norm": 1.9345173835754395, + "learning_rate": 1e-06, + "loss": 0.5898, + "mean_token_accuracy": 0.8066346645355225, + "num_tokens": 898643569.0, + "step": 33557 + }, + { + "epoch": 2.7672136554795084, + "grad_norm": 2.1837143898010254, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7529162764549255, + "num_tokens": 898672944.0, + "step": 33558 + }, + { + "epoch": 2.76729611610456, + "grad_norm": 2.291710138320923, + "learning_rate": 1e-06, + "loss": 0.6194, + "mean_token_accuracy": 0.7861193418502808, + "num_tokens": 898699010.0, + "step": 33559 + }, + { + "epoch": 2.7673785767296115, + "grad_norm": 2.220999240875244, + "learning_rate": 1e-06, + "loss": 0.6371, + "mean_token_accuracy": 0.7945802211761475, + "num_tokens": 898725950.0, + "step": 33560 + }, + { + "epoch": 2.7674610373546633, + "grad_norm": 2.074838399887085, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7570306062698364, + "num_tokens": 898759580.0, + "step": 33561 + }, + { + "epoch": 2.7675434979797147, + "grad_norm": 2.137582302093506, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.8006324768066406, + "num_tokens": 898786044.0, + "step": 33562 + }, + { + "epoch": 2.767625958604766, + "grad_norm": 2.027432680130005, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7849431037902832, + "num_tokens": 898816791.0, + "step": 33563 + }, + { + "epoch": 2.767708419229818, + "grad_norm": 2.074054718017578, + "learning_rate": 1e-06, + "loss": 0.547, + "mean_token_accuracy": 0.7952640652656555, + "num_tokens": 898842138.0, + "step": 33564 + }, + { + "epoch": 2.767790879854869, + "grad_norm": 2.2999284267425537, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7400031089782715, + "num_tokens": 898869298.0, + "step": 33565 + }, + { + "epoch": 2.767873340479921, + "grad_norm": 1.8442829847335815, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7560429573059082, + "num_tokens": 898906117.0, + "step": 33566 + }, + { + "epoch": 2.7679558011049723, + "grad_norm": 1.9491958618164062, + "learning_rate": 1e-06, + "loss": 0.7469, + "mean_token_accuracy": 0.7484412789344788, + "num_tokens": 898938363.0, + "step": 33567 + }, + { + "epoch": 2.768038261730024, + "grad_norm": 1.827157735824585, + "learning_rate": 1e-06, + "loss": 0.5872, + "mean_token_accuracy": 0.8041563630104065, + "num_tokens": 898974844.0, + "step": 33568 + }, + { + "epoch": 2.7681207223550754, + "grad_norm": 2.246875762939453, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7458903789520264, + "num_tokens": 899002619.0, + "step": 33569 + }, + { + "epoch": 2.768203182980127, + "grad_norm": 2.063127279281616, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.7646950483322144, + "num_tokens": 899035043.0, + "step": 33570 + }, + { + "epoch": 2.7682856436051786, + "grad_norm": 2.2833919525146484, + "learning_rate": 1e-06, + "loss": 0.6185, + "mean_token_accuracy": 0.8029621839523315, + "num_tokens": 899060480.0, + "step": 33571 + }, + { + "epoch": 2.76836810423023, + "grad_norm": 2.3004753589630127, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.7731180191040039, + "num_tokens": 899087881.0, + "step": 33572 + }, + { + "epoch": 2.7684505648552817, + "grad_norm": 1.9473241567611694, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7697550058364868, + "num_tokens": 899121918.0, + "step": 33573 + }, + { + "epoch": 2.768533025480333, + "grad_norm": 2.1939704418182373, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7699823379516602, + "num_tokens": 899149397.0, + "step": 33574 + }, + { + "epoch": 2.768615486105385, + "grad_norm": 2.0726935863494873, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.779511034488678, + "num_tokens": 899183245.0, + "step": 33575 + }, + { + "epoch": 2.7686979467304362, + "grad_norm": 1.9904369115829468, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7626668810844421, + "num_tokens": 899213556.0, + "step": 33576 + }, + { + "epoch": 2.7687804073554876, + "grad_norm": 2.089127540588379, + "learning_rate": 1e-06, + "loss": 0.5394, + "mean_token_accuracy": 0.8203192949295044, + "num_tokens": 899241735.0, + "step": 33577 + }, + { + "epoch": 2.7688628679805394, + "grad_norm": 2.1700985431671143, + "learning_rate": 1e-06, + "loss": 0.6821, + "mean_token_accuracy": 0.7767192125320435, + "num_tokens": 899269782.0, + "step": 33578 + }, + { + "epoch": 2.7689453286055907, + "grad_norm": 1.808781623840332, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7760778069496155, + "num_tokens": 899312681.0, + "step": 33579 + }, + { + "epoch": 2.7690277892306425, + "grad_norm": 2.049194812774658, + "learning_rate": 1e-06, + "loss": 0.5597, + "mean_token_accuracy": 0.8141486644744873, + "num_tokens": 899342589.0, + "step": 33580 + }, + { + "epoch": 2.769110249855694, + "grad_norm": 2.1128411293029785, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.7905278205871582, + "num_tokens": 899369638.0, + "step": 33581 + }, + { + "epoch": 2.7691927104807457, + "grad_norm": 2.1628966331481934, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7410684823989868, + "num_tokens": 899402375.0, + "step": 33582 + }, + { + "epoch": 2.769275171105797, + "grad_norm": 2.06099271774292, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7598015666007996, + "num_tokens": 899436124.0, + "step": 33583 + }, + { + "epoch": 2.7693576317308484, + "grad_norm": 1.9288467168807983, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7892504930496216, + "num_tokens": 899470514.0, + "step": 33584 + }, + { + "epoch": 2.7694400923559, + "grad_norm": 2.5856242179870605, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7364127039909363, + "num_tokens": 899491363.0, + "step": 33585 + }, + { + "epoch": 2.7695225529809515, + "grad_norm": 1.9742673635482788, + "learning_rate": 1e-06, + "loss": 0.5418, + "mean_token_accuracy": 0.8057774901390076, + "num_tokens": 899523204.0, + "step": 33586 + }, + { + "epoch": 2.769605013606003, + "grad_norm": 1.9080088138580322, + "learning_rate": 1e-06, + "loss": 0.5714, + "mean_token_accuracy": 0.8068556189537048, + "num_tokens": 899556753.0, + "step": 33587 + }, + { + "epoch": 2.7696874742310547, + "grad_norm": 2.0970449447631836, + "learning_rate": 1e-06, + "loss": 0.5582, + "mean_token_accuracy": 0.816316545009613, + "num_tokens": 899586766.0, + "step": 33588 + }, + { + "epoch": 2.7697699348561065, + "grad_norm": 2.1346359252929688, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7692574262619019, + "num_tokens": 899617443.0, + "step": 33589 + }, + { + "epoch": 2.769852395481158, + "grad_norm": 2.2745888233184814, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7347966432571411, + "num_tokens": 899645893.0, + "step": 33590 + }, + { + "epoch": 2.769934856106209, + "grad_norm": 2.2671971321105957, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7632170915603638, + "num_tokens": 899673476.0, + "step": 33591 + }, + { + "epoch": 2.770017316731261, + "grad_norm": 2.0402112007141113, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7605867981910706, + "num_tokens": 899705895.0, + "step": 33592 + }, + { + "epoch": 2.7700997773563123, + "grad_norm": 2.5732951164245605, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7582476139068604, + "num_tokens": 899730537.0, + "step": 33593 + }, + { + "epoch": 2.7701822379813636, + "grad_norm": 1.8330235481262207, + "learning_rate": 1e-06, + "loss": 0.5733, + "mean_token_accuracy": 0.8128831386566162, + "num_tokens": 899764391.0, + "step": 33594 + }, + { + "epoch": 2.7702646986064154, + "grad_norm": 2.0905449390411377, + "learning_rate": 1e-06, + "loss": 0.4904, + "mean_token_accuracy": 0.8244469165802002, + "num_tokens": 899792084.0, + "step": 33595 + }, + { + "epoch": 2.7703471592314672, + "grad_norm": 2.0665462017059326, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7809017896652222, + "num_tokens": 899823062.0, + "step": 33596 + }, + { + "epoch": 2.7704296198565186, + "grad_norm": 2.488264322280884, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7486782073974609, + "num_tokens": 899848378.0, + "step": 33597 + }, + { + "epoch": 2.77051208048157, + "grad_norm": 1.9565600156784058, + "learning_rate": 1e-06, + "loss": 0.7511, + "mean_token_accuracy": 0.7649918794631958, + "num_tokens": 899881038.0, + "step": 33598 + }, + { + "epoch": 2.7705945411066217, + "grad_norm": 2.0038726329803467, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7698845863342285, + "num_tokens": 899910444.0, + "step": 33599 + }, + { + "epoch": 2.770677001731673, + "grad_norm": 2.2010905742645264, + "learning_rate": 1e-06, + "loss": 0.6112, + "mean_token_accuracy": 0.7881628274917603, + "num_tokens": 899938178.0, + "step": 33600 + }, + { + "epoch": 2.7707594623567244, + "grad_norm": 2.1265668869018555, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7363080978393555, + "num_tokens": 899969149.0, + "step": 33601 + }, + { + "epoch": 2.7708419229817762, + "grad_norm": 2.232545852661133, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7620692849159241, + "num_tokens": 899997591.0, + "step": 33602 + }, + { + "epoch": 2.7709243836068276, + "grad_norm": 2.8131747245788574, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7674626111984253, + "num_tokens": 900017538.0, + "step": 33603 + }, + { + "epoch": 2.7710068442318794, + "grad_norm": 1.8648384809494019, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7671414613723755, + "num_tokens": 900053573.0, + "step": 33604 + }, + { + "epoch": 2.7710893048569307, + "grad_norm": 2.238919496536255, + "learning_rate": 1e-06, + "loss": 0.4773, + "mean_token_accuracy": 0.834915280342102, + "num_tokens": 900079063.0, + "step": 33605 + }, + { + "epoch": 2.7711717654819825, + "grad_norm": 2.301173210144043, + "learning_rate": 1e-06, + "loss": 0.7887, + "mean_token_accuracy": 0.742634654045105, + "num_tokens": 900105516.0, + "step": 33606 + }, + { + "epoch": 2.771254226107034, + "grad_norm": 2.072777032852173, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.7870205044746399, + "num_tokens": 900137901.0, + "step": 33607 + }, + { + "epoch": 2.771336686732085, + "grad_norm": 2.2395713329315186, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.744292140007019, + "num_tokens": 900165568.0, + "step": 33608 + }, + { + "epoch": 2.771419147357137, + "grad_norm": 1.9748449325561523, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.7800789475440979, + "num_tokens": 900197990.0, + "step": 33609 + }, + { + "epoch": 2.7715016079821884, + "grad_norm": 2.1822240352630615, + "learning_rate": 1e-06, + "loss": 0.5978, + "mean_token_accuracy": 0.7963826656341553, + "num_tokens": 900225835.0, + "step": 33610 + }, + { + "epoch": 2.77158406860724, + "grad_norm": 2.3514981269836426, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7717137336730957, + "num_tokens": 900250096.0, + "step": 33611 + }, + { + "epoch": 2.7716665292322915, + "grad_norm": 1.9453495740890503, + "learning_rate": 1e-06, + "loss": 0.6263, + "mean_token_accuracy": 0.7953044176101685, + "num_tokens": 900282410.0, + "step": 33612 + }, + { + "epoch": 2.7717489898573433, + "grad_norm": 1.9247850179672241, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7791639566421509, + "num_tokens": 900319164.0, + "step": 33613 + }, + { + "epoch": 2.7718314504823947, + "grad_norm": 2.3073055744171143, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7623266577720642, + "num_tokens": 900343266.0, + "step": 33614 + }, + { + "epoch": 2.771913911107446, + "grad_norm": 2.139322280883789, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7713894844055176, + "num_tokens": 900372252.0, + "step": 33615 + }, + { + "epoch": 2.771996371732498, + "grad_norm": 2.2132251262664795, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7497076988220215, + "num_tokens": 900400430.0, + "step": 33616 + }, + { + "epoch": 2.772078832357549, + "grad_norm": 2.298182249069214, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7466120719909668, + "num_tokens": 900427260.0, + "step": 33617 + }, + { + "epoch": 2.772161292982601, + "grad_norm": 1.916912317276001, + "learning_rate": 1e-06, + "loss": 0.6248, + "mean_token_accuracy": 0.7939382195472717, + "num_tokens": 900463152.0, + "step": 33618 + }, + { + "epoch": 2.7722437536076523, + "grad_norm": 2.095606565475464, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.7841557860374451, + "num_tokens": 900494619.0, + "step": 33619 + }, + { + "epoch": 2.772326214232704, + "grad_norm": 2.1032609939575195, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7309942245483398, + "num_tokens": 900525888.0, + "step": 33620 + }, + { + "epoch": 2.7724086748577554, + "grad_norm": 1.9927445650100708, + "learning_rate": 1e-06, + "loss": 0.7636, + "mean_token_accuracy": 0.7632731199264526, + "num_tokens": 900559381.0, + "step": 33621 + }, + { + "epoch": 2.772491135482807, + "grad_norm": 1.9307652711868286, + "learning_rate": 1e-06, + "loss": 0.6727, + "mean_token_accuracy": 0.7703886032104492, + "num_tokens": 900592246.0, + "step": 33622 + }, + { + "epoch": 2.7725735961078586, + "grad_norm": 2.155097246170044, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7642781734466553, + "num_tokens": 900620263.0, + "step": 33623 + }, + { + "epoch": 2.77265605673291, + "grad_norm": 1.9178764820098877, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.7908501625061035, + "num_tokens": 900653366.0, + "step": 33624 + }, + { + "epoch": 2.7727385173579617, + "grad_norm": 2.1895902156829834, + "learning_rate": 1e-06, + "loss": 0.5593, + "mean_token_accuracy": 0.8112586140632629, + "num_tokens": 900679339.0, + "step": 33625 + }, + { + "epoch": 2.772820977983013, + "grad_norm": 2.1858348846435547, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7805029153823853, + "num_tokens": 900706231.0, + "step": 33626 + }, + { + "epoch": 2.772903438608065, + "grad_norm": 2.232579231262207, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.7959585785865784, + "num_tokens": 900735451.0, + "step": 33627 + }, + { + "epoch": 2.7729858992331162, + "grad_norm": 2.0183515548706055, + "learning_rate": 1e-06, + "loss": 0.7422, + "mean_token_accuracy": 0.7534188032150269, + "num_tokens": 900767161.0, + "step": 33628 + }, + { + "epoch": 2.7730683598581676, + "grad_norm": 2.199178457260132, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7598363161087036, + "num_tokens": 900794569.0, + "step": 33629 + }, + { + "epoch": 2.7731508204832194, + "grad_norm": 2.065560817718506, + "learning_rate": 1e-06, + "loss": 0.5614, + "mean_token_accuracy": 0.7969090938568115, + "num_tokens": 900826138.0, + "step": 33630 + }, + { + "epoch": 2.7732332811082707, + "grad_norm": 2.238445520401001, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7628027200698853, + "num_tokens": 900852795.0, + "step": 33631 + }, + { + "epoch": 2.773315741733322, + "grad_norm": 2.039863348007202, + "learning_rate": 1e-06, + "loss": 0.6563, + "mean_token_accuracy": 0.7855060696601868, + "num_tokens": 900883908.0, + "step": 33632 + }, + { + "epoch": 2.773398202358374, + "grad_norm": 2.0830535888671875, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7308312654495239, + "num_tokens": 900916113.0, + "step": 33633 + }, + { + "epoch": 2.7734806629834257, + "grad_norm": 1.8153142929077148, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.72908616065979, + "num_tokens": 900955528.0, + "step": 33634 + }, + { + "epoch": 2.773563123608477, + "grad_norm": 2.0632457733154297, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7849056720733643, + "num_tokens": 900987260.0, + "step": 33635 + }, + { + "epoch": 2.7736455842335284, + "grad_norm": 2.1808202266693115, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.743963360786438, + "num_tokens": 901016151.0, + "step": 33636 + }, + { + "epoch": 2.77372804485858, + "grad_norm": 2.219007968902588, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7550077438354492, + "num_tokens": 901042671.0, + "step": 33637 + }, + { + "epoch": 2.7738105054836315, + "grad_norm": 2.0752294063568115, + "learning_rate": 1e-06, + "loss": 0.6443, + "mean_token_accuracy": 0.7819738388061523, + "num_tokens": 901072414.0, + "step": 33638 + }, + { + "epoch": 2.773892966108683, + "grad_norm": 2.1963448524475098, + "learning_rate": 1e-06, + "loss": 0.5596, + "mean_token_accuracy": 0.8076397180557251, + "num_tokens": 901100730.0, + "step": 33639 + }, + { + "epoch": 2.7739754267337347, + "grad_norm": 2.0565993785858154, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7681025266647339, + "num_tokens": 901134039.0, + "step": 33640 + }, + { + "epoch": 2.7740578873587864, + "grad_norm": 2.0856380462646484, + "learning_rate": 1e-06, + "loss": 0.5796, + "mean_token_accuracy": 0.8044034242630005, + "num_tokens": 901162804.0, + "step": 33641 + }, + { + "epoch": 2.774140347983838, + "grad_norm": 2.184260606765747, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7470915913581848, + "num_tokens": 901192647.0, + "step": 33642 + }, + { + "epoch": 2.774222808608889, + "grad_norm": 2.1595566272735596, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7515345215797424, + "num_tokens": 901225680.0, + "step": 33643 + }, + { + "epoch": 2.774305269233941, + "grad_norm": 2.0580849647521973, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7826899290084839, + "num_tokens": 901257788.0, + "step": 33644 + }, + { + "epoch": 2.7743877298589923, + "grad_norm": 2.414422035217285, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7427043914794922, + "num_tokens": 901286672.0, + "step": 33645 + }, + { + "epoch": 2.7744701904840436, + "grad_norm": 2.1941633224487305, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7529789209365845, + "num_tokens": 901313080.0, + "step": 33646 + }, + { + "epoch": 2.7745526511090954, + "grad_norm": 2.213224172592163, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7676820755004883, + "num_tokens": 901338959.0, + "step": 33647 + }, + { + "epoch": 2.774635111734147, + "grad_norm": 2.2445929050445557, + "learning_rate": 1e-06, + "loss": 0.728, + "mean_token_accuracy": 0.7646927833557129, + "num_tokens": 901365476.0, + "step": 33648 + }, + { + "epoch": 2.7747175723591986, + "grad_norm": 2.141747236251831, + "learning_rate": 1e-06, + "loss": 0.6169, + "mean_token_accuracy": 0.797522783279419, + "num_tokens": 901393656.0, + "step": 33649 + }, + { + "epoch": 2.77480003298425, + "grad_norm": 1.9298049211502075, + "learning_rate": 1e-06, + "loss": 0.5568, + "mean_token_accuracy": 0.8114385008811951, + "num_tokens": 901424670.0, + "step": 33650 + }, + { + "epoch": 2.7748824936093017, + "grad_norm": 2.1976988315582275, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7583721876144409, + "num_tokens": 901453729.0, + "step": 33651 + }, + { + "epoch": 2.774964954234353, + "grad_norm": 2.170370578765869, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7528179883956909, + "num_tokens": 901480881.0, + "step": 33652 + }, + { + "epoch": 2.7750474148594044, + "grad_norm": 2.1317107677459717, + "learning_rate": 1e-06, + "loss": 0.7242, + "mean_token_accuracy": 0.7794204950332642, + "num_tokens": 901509981.0, + "step": 33653 + }, + { + "epoch": 2.775129875484456, + "grad_norm": 2.2411367893218994, + "learning_rate": 1e-06, + "loss": 0.7854, + "mean_token_accuracy": 0.7469854354858398, + "num_tokens": 901536832.0, + "step": 33654 + }, + { + "epoch": 2.7752123361095076, + "grad_norm": 1.9635802507400513, + "learning_rate": 1e-06, + "loss": 0.5588, + "mean_token_accuracy": 0.8117178678512573, + "num_tokens": 901569460.0, + "step": 33655 + }, + { + "epoch": 2.7752947967345594, + "grad_norm": 2.027073860168457, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7600113153457642, + "num_tokens": 901607365.0, + "step": 33656 + }, + { + "epoch": 2.7753772573596107, + "grad_norm": 2.2538235187530518, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7587577104568481, + "num_tokens": 901637294.0, + "step": 33657 + }, + { + "epoch": 2.7754597179846625, + "grad_norm": 2.2295033931732178, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7817294597625732, + "num_tokens": 901664343.0, + "step": 33658 + }, + { + "epoch": 2.775542178609714, + "grad_norm": 1.8656187057495117, + "learning_rate": 1e-06, + "loss": 0.7079, + "mean_token_accuracy": 0.7694874405860901, + "num_tokens": 901700706.0, + "step": 33659 + }, + { + "epoch": 2.775624639234765, + "grad_norm": 2.077932357788086, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7661445140838623, + "num_tokens": 901733376.0, + "step": 33660 + }, + { + "epoch": 2.775707099859817, + "grad_norm": 2.036684036254883, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7852175235748291, + "num_tokens": 901764452.0, + "step": 33661 + }, + { + "epoch": 2.7757895604848684, + "grad_norm": 2.0521750450134277, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.7833691239356995, + "num_tokens": 901795735.0, + "step": 33662 + }, + { + "epoch": 2.77587202110992, + "grad_norm": 2.2077696323394775, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.8023477792739868, + "num_tokens": 901822783.0, + "step": 33663 + }, + { + "epoch": 2.7759544817349715, + "grad_norm": 2.0136067867279053, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.788500964641571, + "num_tokens": 901854674.0, + "step": 33664 + }, + { + "epoch": 2.7760369423600233, + "grad_norm": 2.1387457847595215, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7677611708641052, + "num_tokens": 901885099.0, + "step": 33665 + }, + { + "epoch": 2.7761194029850746, + "grad_norm": 2.467759847640991, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.7885763645172119, + "num_tokens": 901908957.0, + "step": 33666 + }, + { + "epoch": 2.776201863610126, + "grad_norm": 1.7629127502441406, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.7428886890411377, + "num_tokens": 901948813.0, + "step": 33667 + }, + { + "epoch": 2.776284324235178, + "grad_norm": 2.0580854415893555, + "learning_rate": 1e-06, + "loss": 0.5923, + "mean_token_accuracy": 0.8064258098602295, + "num_tokens": 901978532.0, + "step": 33668 + }, + { + "epoch": 2.776366784860229, + "grad_norm": 2.1882786750793457, + "learning_rate": 1e-06, + "loss": 0.69, + "mean_token_accuracy": 0.7835404872894287, + "num_tokens": 902004717.0, + "step": 33669 + }, + { + "epoch": 2.776449245485281, + "grad_norm": 2.2560641765594482, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.7822489142417908, + "num_tokens": 902033799.0, + "step": 33670 + }, + { + "epoch": 2.7765317061103323, + "grad_norm": 2.3668298721313477, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.7839590311050415, + "num_tokens": 902058135.0, + "step": 33671 + }, + { + "epoch": 2.776614166735384, + "grad_norm": 2.0430619716644287, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7451416254043579, + "num_tokens": 902091528.0, + "step": 33672 + }, + { + "epoch": 2.7766966273604354, + "grad_norm": 2.03574275970459, + "learning_rate": 1e-06, + "loss": 0.5972, + "mean_token_accuracy": 0.8025647401809692, + "num_tokens": 902123155.0, + "step": 33673 + }, + { + "epoch": 2.776779087985487, + "grad_norm": 2.2565228939056396, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.759894073009491, + "num_tokens": 902150566.0, + "step": 33674 + }, + { + "epoch": 2.7768615486105386, + "grad_norm": 1.9710969924926758, + "learning_rate": 1e-06, + "loss": 0.5658, + "mean_token_accuracy": 0.8101310133934021, + "num_tokens": 902186992.0, + "step": 33675 + }, + { + "epoch": 2.77694400923559, + "grad_norm": 2.449333906173706, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7807729244232178, + "num_tokens": 902212648.0, + "step": 33676 + }, + { + "epoch": 2.7770264698606413, + "grad_norm": 2.250666856765747, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7584021091461182, + "num_tokens": 902239582.0, + "step": 33677 + }, + { + "epoch": 2.777108930485693, + "grad_norm": 2.2706470489501953, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.780030369758606, + "num_tokens": 902269016.0, + "step": 33678 + }, + { + "epoch": 2.777191391110745, + "grad_norm": 2.244415521621704, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7474273443222046, + "num_tokens": 902297855.0, + "step": 33679 + }, + { + "epoch": 2.777273851735796, + "grad_norm": 2.030670642852783, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.7745237350463867, + "num_tokens": 902328327.0, + "step": 33680 + }, + { + "epoch": 2.7773563123608476, + "grad_norm": 2.1550605297088623, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7597923278808594, + "num_tokens": 902356824.0, + "step": 33681 + }, + { + "epoch": 2.7774387729858994, + "grad_norm": 2.108647108078003, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7853205800056458, + "num_tokens": 902387893.0, + "step": 33682 + }, + { + "epoch": 2.7775212336109507, + "grad_norm": 2.2130677700042725, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.7951563596725464, + "num_tokens": 902418524.0, + "step": 33683 + }, + { + "epoch": 2.777603694236002, + "grad_norm": 2.0579729080200195, + "learning_rate": 1e-06, + "loss": 0.6034, + "mean_token_accuracy": 0.808762788772583, + "num_tokens": 902445928.0, + "step": 33684 + }, + { + "epoch": 2.777686154861054, + "grad_norm": 2.0796096324920654, + "learning_rate": 1e-06, + "loss": 0.6946, + "mean_token_accuracy": 0.7681786417961121, + "num_tokens": 902475432.0, + "step": 33685 + }, + { + "epoch": 2.7777686154861057, + "grad_norm": 2.319220781326294, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7694514989852905, + "num_tokens": 902502451.0, + "step": 33686 + }, + { + "epoch": 2.777851076111157, + "grad_norm": 2.2234079837799072, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7725206613540649, + "num_tokens": 902529867.0, + "step": 33687 + }, + { + "epoch": 2.7779335367362084, + "grad_norm": 2.27535343170166, + "learning_rate": 1e-06, + "loss": 0.5896, + "mean_token_accuracy": 0.7978828549385071, + "num_tokens": 902555544.0, + "step": 33688 + }, + { + "epoch": 2.77801599736126, + "grad_norm": 2.2207040786743164, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7685496211051941, + "num_tokens": 902583923.0, + "step": 33689 + }, + { + "epoch": 2.7780984579863115, + "grad_norm": 2.095574378967285, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7616932988166809, + "num_tokens": 902617948.0, + "step": 33690 + }, + { + "epoch": 2.778180918611363, + "grad_norm": 1.7419358491897583, + "learning_rate": 1e-06, + "loss": 0.5646, + "mean_token_accuracy": 0.8154973983764648, + "num_tokens": 902657461.0, + "step": 33691 + }, + { + "epoch": 2.7782633792364146, + "grad_norm": 1.9583638906478882, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.7897160053253174, + "num_tokens": 902691621.0, + "step": 33692 + }, + { + "epoch": 2.778345839861466, + "grad_norm": 1.8887776136398315, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7685102224349976, + "num_tokens": 902724840.0, + "step": 33693 + }, + { + "epoch": 2.778428300486518, + "grad_norm": 2.154147148132324, + "learning_rate": 1e-06, + "loss": 0.7754, + "mean_token_accuracy": 0.7563724517822266, + "num_tokens": 902753546.0, + "step": 33694 + }, + { + "epoch": 2.778510761111569, + "grad_norm": 1.9683433771133423, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.777093231678009, + "num_tokens": 902785814.0, + "step": 33695 + }, + { + "epoch": 2.778593221736621, + "grad_norm": 1.9924545288085938, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.7887825965881348, + "num_tokens": 902818170.0, + "step": 33696 + }, + { + "epoch": 2.7786756823616723, + "grad_norm": 2.1019396781921387, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7419590950012207, + "num_tokens": 902849317.0, + "step": 33697 + }, + { + "epoch": 2.7787581429867236, + "grad_norm": 1.8594380617141724, + "learning_rate": 1e-06, + "loss": 0.6529, + "mean_token_accuracy": 0.7832737565040588, + "num_tokens": 902886055.0, + "step": 33698 + }, + { + "epoch": 2.7788406036117754, + "grad_norm": 2.1629645824432373, + "learning_rate": 1e-06, + "loss": 0.5964, + "mean_token_accuracy": 0.7927076816558838, + "num_tokens": 902914377.0, + "step": 33699 + }, + { + "epoch": 2.778923064236827, + "grad_norm": 2.023313522338867, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7338090538978577, + "num_tokens": 902947575.0, + "step": 33700 + }, + { + "epoch": 2.7790055248618786, + "grad_norm": 2.6601264476776123, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.748476505279541, + "num_tokens": 902969296.0, + "step": 33701 + }, + { + "epoch": 2.77908798548693, + "grad_norm": 2.0054070949554443, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7708930373191833, + "num_tokens": 903002904.0, + "step": 33702 + }, + { + "epoch": 2.7791704461119817, + "grad_norm": 1.8791286945343018, + "learning_rate": 1e-06, + "loss": 0.6148, + "mean_token_accuracy": 0.7887978553771973, + "num_tokens": 903041105.0, + "step": 33703 + }, + { + "epoch": 2.779252906737033, + "grad_norm": 1.9696345329284668, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7672872543334961, + "num_tokens": 903074566.0, + "step": 33704 + }, + { + "epoch": 2.7793353673620844, + "grad_norm": 2.051866292953491, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7326022982597351, + "num_tokens": 903107834.0, + "step": 33705 + }, + { + "epoch": 2.779417827987136, + "grad_norm": 2.0708000659942627, + "learning_rate": 1e-06, + "loss": 0.589, + "mean_token_accuracy": 0.7951479554176331, + "num_tokens": 903137109.0, + "step": 33706 + }, + { + "epoch": 2.7795002886121876, + "grad_norm": 2.2838563919067383, + "learning_rate": 1e-06, + "loss": 0.7356, + "mean_token_accuracy": 0.7574814558029175, + "num_tokens": 903165096.0, + "step": 33707 + }, + { + "epoch": 2.7795827492372394, + "grad_norm": 2.237837791442871, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7680984139442444, + "num_tokens": 903193517.0, + "step": 33708 + }, + { + "epoch": 2.7796652098622907, + "grad_norm": 2.17270565032959, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7613892555236816, + "num_tokens": 903225538.0, + "step": 33709 + }, + { + "epoch": 2.7797476704873425, + "grad_norm": 2.171666383743286, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7677018642425537, + "num_tokens": 903254266.0, + "step": 33710 + }, + { + "epoch": 2.779830131112394, + "grad_norm": 1.7463151216506958, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7602978944778442, + "num_tokens": 903293770.0, + "step": 33711 + }, + { + "epoch": 2.779912591737445, + "grad_norm": 2.496699571609497, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7457045316696167, + "num_tokens": 903317611.0, + "step": 33712 + }, + { + "epoch": 2.779995052362497, + "grad_norm": 2.2454733848571777, + "learning_rate": 1e-06, + "loss": 0.5644, + "mean_token_accuracy": 0.8095310926437378, + "num_tokens": 903343115.0, + "step": 33713 + }, + { + "epoch": 2.7800775129875483, + "grad_norm": 2.4706943035125732, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.752234160900116, + "num_tokens": 903366596.0, + "step": 33714 + }, + { + "epoch": 2.7801599736126, + "grad_norm": 1.9697015285491943, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.7829980254173279, + "num_tokens": 903394658.0, + "step": 33715 + }, + { + "epoch": 2.7802424342376515, + "grad_norm": 2.19472336769104, + "learning_rate": 1e-06, + "loss": 0.7104, + "mean_token_accuracy": 0.7637531161308289, + "num_tokens": 903422223.0, + "step": 33716 + }, + { + "epoch": 2.7803248948627033, + "grad_norm": 1.9252510070800781, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.7933754920959473, + "num_tokens": 903458404.0, + "step": 33717 + }, + { + "epoch": 2.7804073554877546, + "grad_norm": 2.067963123321533, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.7894853353500366, + "num_tokens": 903487824.0, + "step": 33718 + }, + { + "epoch": 2.780489816112806, + "grad_norm": 2.0315029621124268, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.766481876373291, + "num_tokens": 903519617.0, + "step": 33719 + }, + { + "epoch": 2.780572276737858, + "grad_norm": 1.8332531452178955, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7770242094993591, + "num_tokens": 903557363.0, + "step": 33720 + }, + { + "epoch": 2.780654737362909, + "grad_norm": 2.22184157371521, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7804868817329407, + "num_tokens": 903585340.0, + "step": 33721 + }, + { + "epoch": 2.7807371979879605, + "grad_norm": 2.088563919067383, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.7936612963676453, + "num_tokens": 903612500.0, + "step": 33722 + }, + { + "epoch": 2.7808196586130123, + "grad_norm": 1.7896889448165894, + "learning_rate": 1e-06, + "loss": 0.4711, + "mean_token_accuracy": 0.820484459400177, + "num_tokens": 903646778.0, + "step": 33723 + }, + { + "epoch": 2.780902119238064, + "grad_norm": 2.1030995845794678, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7611218094825745, + "num_tokens": 903677258.0, + "step": 33724 + }, + { + "epoch": 2.7809845798631154, + "grad_norm": 2.2067489624023438, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.7830613255500793, + "num_tokens": 903704319.0, + "step": 33725 + }, + { + "epoch": 2.7810670404881668, + "grad_norm": 2.1203300952911377, + "learning_rate": 1e-06, + "loss": 0.7343, + "mean_token_accuracy": 0.7674007415771484, + "num_tokens": 903735519.0, + "step": 33726 + }, + { + "epoch": 2.7811495011132186, + "grad_norm": 2.3099091053009033, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7589126229286194, + "num_tokens": 903762247.0, + "step": 33727 + }, + { + "epoch": 2.78123196173827, + "grad_norm": 2.0121824741363525, + "learning_rate": 1e-06, + "loss": 0.6921, + "mean_token_accuracy": 0.7733224630355835, + "num_tokens": 903793617.0, + "step": 33728 + }, + { + "epoch": 2.7813144223633213, + "grad_norm": 1.974622130393982, + "learning_rate": 1e-06, + "loss": 0.5586, + "mean_token_accuracy": 0.8148316740989685, + "num_tokens": 903826532.0, + "step": 33729 + }, + { + "epoch": 2.781396882988373, + "grad_norm": 1.9563621282577515, + "learning_rate": 1e-06, + "loss": 0.5592, + "mean_token_accuracy": 0.8109413385391235, + "num_tokens": 903857467.0, + "step": 33730 + }, + { + "epoch": 2.781479343613425, + "grad_norm": 1.7981356382369995, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7827444672584534, + "num_tokens": 903896288.0, + "step": 33731 + }, + { + "epoch": 2.781561804238476, + "grad_norm": 2.0305752754211426, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7780606746673584, + "num_tokens": 903929608.0, + "step": 33732 + }, + { + "epoch": 2.7816442648635276, + "grad_norm": 2.2320380210876465, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7394546270370483, + "num_tokens": 903959852.0, + "step": 33733 + }, + { + "epoch": 2.7817267254885794, + "grad_norm": 2.167997121810913, + "learning_rate": 1e-06, + "loss": 0.7011, + "mean_token_accuracy": 0.7695283889770508, + "num_tokens": 903989012.0, + "step": 33734 + }, + { + "epoch": 2.7818091861136307, + "grad_norm": 2.309574842453003, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7564947009086609, + "num_tokens": 904015866.0, + "step": 33735 + }, + { + "epoch": 2.781891646738682, + "grad_norm": 2.5404937267303467, + "learning_rate": 1e-06, + "loss": 0.6793, + "mean_token_accuracy": 0.7724616527557373, + "num_tokens": 904049438.0, + "step": 33736 + }, + { + "epoch": 2.781974107363734, + "grad_norm": 2.1906018257141113, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.783645510673523, + "num_tokens": 904078239.0, + "step": 33737 + }, + { + "epoch": 2.782056567988785, + "grad_norm": 1.988673210144043, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.7856748700141907, + "num_tokens": 904110957.0, + "step": 33738 + }, + { + "epoch": 2.782139028613837, + "grad_norm": 2.24540376663208, + "learning_rate": 1e-06, + "loss": 0.6545, + "mean_token_accuracy": 0.7872979640960693, + "num_tokens": 904136443.0, + "step": 33739 + }, + { + "epoch": 2.7822214892388883, + "grad_norm": 2.2293717861175537, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.770875096321106, + "num_tokens": 904164641.0, + "step": 33740 + }, + { + "epoch": 2.78230394986394, + "grad_norm": 2.0506069660186768, + "learning_rate": 1e-06, + "loss": 0.6598, + "mean_token_accuracy": 0.7775737047195435, + "num_tokens": 904197619.0, + "step": 33741 + }, + { + "epoch": 2.7823864104889915, + "grad_norm": 2.0457770824432373, + "learning_rate": 1e-06, + "loss": 0.7002, + "mean_token_accuracy": 0.7755096554756165, + "num_tokens": 904231502.0, + "step": 33742 + }, + { + "epoch": 2.782468871114043, + "grad_norm": 2.0144145488739014, + "learning_rate": 1e-06, + "loss": 0.6855, + "mean_token_accuracy": 0.7737994194030762, + "num_tokens": 904262970.0, + "step": 33743 + }, + { + "epoch": 2.7825513317390946, + "grad_norm": 2.0003418922424316, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7732251286506653, + "num_tokens": 904294277.0, + "step": 33744 + }, + { + "epoch": 2.782633792364146, + "grad_norm": 2.04190993309021, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7735617160797119, + "num_tokens": 904322771.0, + "step": 33745 + }, + { + "epoch": 2.782716252989198, + "grad_norm": 2.4840073585510254, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7488464713096619, + "num_tokens": 904346324.0, + "step": 33746 + }, + { + "epoch": 2.782798713614249, + "grad_norm": 1.9412280321121216, + "learning_rate": 1e-06, + "loss": 0.5758, + "mean_token_accuracy": 0.811833918094635, + "num_tokens": 904379234.0, + "step": 33747 + }, + { + "epoch": 2.782881174239301, + "grad_norm": 2.2222468852996826, + "learning_rate": 1e-06, + "loss": 0.6561, + "mean_token_accuracy": 0.7760457992553711, + "num_tokens": 904405815.0, + "step": 33748 + }, + { + "epoch": 2.7829636348643523, + "grad_norm": 2.5060694217681885, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7471541166305542, + "num_tokens": 904431524.0, + "step": 33749 + }, + { + "epoch": 2.7830460954894036, + "grad_norm": 2.118263006210327, + "learning_rate": 1e-06, + "loss": 0.6315, + "mean_token_accuracy": 0.7945277094841003, + "num_tokens": 904459614.0, + "step": 33750 + }, + { + "epoch": 2.7831285561144554, + "grad_norm": 1.9407833814620972, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7667197585105896, + "num_tokens": 904494342.0, + "step": 33751 + }, + { + "epoch": 2.7832110167395068, + "grad_norm": 2.192023277282715, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7809834480285645, + "num_tokens": 904522668.0, + "step": 33752 + }, + { + "epoch": 2.7832934773645586, + "grad_norm": 2.03409481048584, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7816652059555054, + "num_tokens": 904553143.0, + "step": 33753 + }, + { + "epoch": 2.78337593798961, + "grad_norm": 2.068220615386963, + "learning_rate": 1e-06, + "loss": 0.637, + "mean_token_accuracy": 0.7899835109710693, + "num_tokens": 904585775.0, + "step": 33754 + }, + { + "epoch": 2.7834583986146617, + "grad_norm": 2.2197325229644775, + "learning_rate": 1e-06, + "loss": 0.534, + "mean_token_accuracy": 0.8150033354759216, + "num_tokens": 904613294.0, + "step": 33755 + }, + { + "epoch": 2.783540859239713, + "grad_norm": 2.4181981086730957, + "learning_rate": 1e-06, + "loss": 0.6116, + "mean_token_accuracy": 0.7819676399230957, + "num_tokens": 904638770.0, + "step": 33756 + }, + { + "epoch": 2.7836233198647644, + "grad_norm": 2.369234561920166, + "learning_rate": 1e-06, + "loss": 0.7193, + "mean_token_accuracy": 0.7668279409408569, + "num_tokens": 904661957.0, + "step": 33757 + }, + { + "epoch": 2.783705780489816, + "grad_norm": 2.002568244934082, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7542077898979187, + "num_tokens": 904692709.0, + "step": 33758 + }, + { + "epoch": 2.7837882411148676, + "grad_norm": 2.356720447540283, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7598400712013245, + "num_tokens": 904718739.0, + "step": 33759 + }, + { + "epoch": 2.7838707017399194, + "grad_norm": 2.194375991821289, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7398571968078613, + "num_tokens": 904748081.0, + "step": 33760 + }, + { + "epoch": 2.7839531623649707, + "grad_norm": 1.8044098615646362, + "learning_rate": 1e-06, + "loss": 0.6351, + "mean_token_accuracy": 0.7946352958679199, + "num_tokens": 904784095.0, + "step": 33761 + }, + { + "epoch": 2.7840356229900225, + "grad_norm": 2.6240038871765137, + "learning_rate": 1e-06, + "loss": 0.6745, + "mean_token_accuracy": 0.7809605598449707, + "num_tokens": 904805595.0, + "step": 33762 + }, + { + "epoch": 2.784118083615074, + "grad_norm": 2.0475270748138428, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7698802947998047, + "num_tokens": 904839204.0, + "step": 33763 + }, + { + "epoch": 2.784200544240125, + "grad_norm": 1.88974928855896, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7505808472633362, + "num_tokens": 904871976.0, + "step": 33764 + }, + { + "epoch": 2.784283004865177, + "grad_norm": 2.310398817062378, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.7680807113647461, + "num_tokens": 904896733.0, + "step": 33765 + }, + { + "epoch": 2.7843654654902283, + "grad_norm": 2.328918933868408, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7585937976837158, + "num_tokens": 904924201.0, + "step": 33766 + }, + { + "epoch": 2.7844479261152797, + "grad_norm": 1.989037275314331, + "learning_rate": 1e-06, + "loss": 0.6592, + "mean_token_accuracy": 0.7753455638885498, + "num_tokens": 904953497.0, + "step": 33767 + }, + { + "epoch": 2.7845303867403315, + "grad_norm": 2.008518934249878, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.776680588722229, + "num_tokens": 904984413.0, + "step": 33768 + }, + { + "epoch": 2.7846128473653833, + "grad_norm": 2.0399138927459717, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7713261246681213, + "num_tokens": 905014233.0, + "step": 33769 + }, + { + "epoch": 2.7846953079904346, + "grad_norm": 2.2373807430267334, + "learning_rate": 1e-06, + "loss": 0.6928, + "mean_token_accuracy": 0.7819570899009705, + "num_tokens": 905042651.0, + "step": 33770 + }, + { + "epoch": 2.784777768615486, + "grad_norm": 1.9602246284484863, + "learning_rate": 1e-06, + "loss": 0.532, + "mean_token_accuracy": 0.8231555819511414, + "num_tokens": 905073206.0, + "step": 33771 + }, + { + "epoch": 2.7848602292405378, + "grad_norm": 2.4164350032806396, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7537961602210999, + "num_tokens": 905098064.0, + "step": 33772 + }, + { + "epoch": 2.784942689865589, + "grad_norm": 1.994839072227478, + "learning_rate": 1e-06, + "loss": 0.5908, + "mean_token_accuracy": 0.8006354570388794, + "num_tokens": 905130816.0, + "step": 33773 + }, + { + "epoch": 2.7850251504906405, + "grad_norm": 2.591738224029541, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.7520220279693604, + "num_tokens": 905152876.0, + "step": 33774 + }, + { + "epoch": 2.7851076111156923, + "grad_norm": 2.3202085494995117, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7555606961250305, + "num_tokens": 905179295.0, + "step": 33775 + }, + { + "epoch": 2.785190071740744, + "grad_norm": 2.1272034645080566, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.7852886319160461, + "num_tokens": 905209672.0, + "step": 33776 + }, + { + "epoch": 2.7852725323657954, + "grad_norm": 1.9754514694213867, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7850232124328613, + "num_tokens": 905242596.0, + "step": 33777 + }, + { + "epoch": 2.7853549929908468, + "grad_norm": 1.7281112670898438, + "learning_rate": 1e-06, + "loss": 0.6332, + "mean_token_accuracy": 0.7936694622039795, + "num_tokens": 905280586.0, + "step": 33778 + }, + { + "epoch": 2.7854374536158986, + "grad_norm": 2.1289193630218506, + "learning_rate": 1e-06, + "loss": 0.5672, + "mean_token_accuracy": 0.8072798252105713, + "num_tokens": 905305480.0, + "step": 33779 + }, + { + "epoch": 2.78551991424095, + "grad_norm": 2.0256357192993164, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7773562073707581, + "num_tokens": 905335193.0, + "step": 33780 + }, + { + "epoch": 2.7856023748660013, + "grad_norm": 1.9687180519104004, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7465415596961975, + "num_tokens": 905370391.0, + "step": 33781 + }, + { + "epoch": 2.785684835491053, + "grad_norm": 2.070359945297241, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7643873691558838, + "num_tokens": 905400833.0, + "step": 33782 + }, + { + "epoch": 2.7857672961161044, + "grad_norm": 2.4041945934295654, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7352886199951172, + "num_tokens": 905426400.0, + "step": 33783 + }, + { + "epoch": 2.785849756741156, + "grad_norm": 2.008829355239868, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7842193841934204, + "num_tokens": 905460210.0, + "step": 33784 + }, + { + "epoch": 2.7859322173662076, + "grad_norm": 2.0675947666168213, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.7719061374664307, + "num_tokens": 905490521.0, + "step": 33785 + }, + { + "epoch": 2.7860146779912593, + "grad_norm": 1.9425731897354126, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7803069949150085, + "num_tokens": 905527438.0, + "step": 33786 + }, + { + "epoch": 2.7860971386163107, + "grad_norm": 2.522484540939331, + "learning_rate": 1e-06, + "loss": 0.6077, + "mean_token_accuracy": 0.7919906377792358, + "num_tokens": 905548307.0, + "step": 33787 + }, + { + "epoch": 2.786179599241362, + "grad_norm": 1.9858813285827637, + "learning_rate": 1e-06, + "loss": 0.5362, + "mean_token_accuracy": 0.8177655935287476, + "num_tokens": 905577596.0, + "step": 33788 + }, + { + "epoch": 2.786262059866414, + "grad_norm": 2.148277997970581, + "learning_rate": 1e-06, + "loss": 0.5095, + "mean_token_accuracy": 0.8194558620452881, + "num_tokens": 905604449.0, + "step": 33789 + }, + { + "epoch": 2.786344520491465, + "grad_norm": 1.8746898174285889, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7628263831138611, + "num_tokens": 905641124.0, + "step": 33790 + }, + { + "epoch": 2.786426981116517, + "grad_norm": 2.139437675476074, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.7817071080207825, + "num_tokens": 905668243.0, + "step": 33791 + }, + { + "epoch": 2.7865094417415683, + "grad_norm": 2.064418315887451, + "learning_rate": 1e-06, + "loss": 0.6536, + "mean_token_accuracy": 0.7800803184509277, + "num_tokens": 905699439.0, + "step": 33792 + }, + { + "epoch": 2.78659190236662, + "grad_norm": 2.1887946128845215, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7403911352157593, + "num_tokens": 905729866.0, + "step": 33793 + }, + { + "epoch": 2.7866743629916715, + "grad_norm": 1.968029260635376, + "learning_rate": 1e-06, + "loss": 0.7471, + "mean_token_accuracy": 0.7629940509796143, + "num_tokens": 905762605.0, + "step": 33794 + }, + { + "epoch": 2.786756823616723, + "grad_norm": 1.9366484880447388, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.7858526706695557, + "num_tokens": 905795646.0, + "step": 33795 + }, + { + "epoch": 2.7868392842417746, + "grad_norm": 2.234773874282837, + "learning_rate": 1e-06, + "loss": 0.7416, + "mean_token_accuracy": 0.7577584981918335, + "num_tokens": 905821223.0, + "step": 33796 + }, + { + "epoch": 2.786921744866826, + "grad_norm": 2.0388972759246826, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.7899953126907349, + "num_tokens": 905849306.0, + "step": 33797 + }, + { + "epoch": 2.7870042054918778, + "grad_norm": 2.0531513690948486, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.7920933365821838, + "num_tokens": 905879688.0, + "step": 33798 + }, + { + "epoch": 2.787086666116929, + "grad_norm": 2.2228150367736816, + "learning_rate": 1e-06, + "loss": 0.7505, + "mean_token_accuracy": 0.7620670795440674, + "num_tokens": 905906357.0, + "step": 33799 + }, + { + "epoch": 2.787169126741981, + "grad_norm": 2.039616584777832, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7822632193565369, + "num_tokens": 905937152.0, + "step": 33800 + }, + { + "epoch": 2.7872515873670323, + "grad_norm": 1.9678544998168945, + "learning_rate": 1e-06, + "loss": 0.5228, + "mean_token_accuracy": 0.8148821592330933, + "num_tokens": 905966995.0, + "step": 33801 + }, + { + "epoch": 2.7873340479920836, + "grad_norm": 2.33756685256958, + "learning_rate": 1e-06, + "loss": 0.73, + "mean_token_accuracy": 0.7690580487251282, + "num_tokens": 905994302.0, + "step": 33802 + }, + { + "epoch": 2.7874165086171354, + "grad_norm": 1.8105857372283936, + "learning_rate": 1e-06, + "loss": 0.659, + "mean_token_accuracy": 0.7891268730163574, + "num_tokens": 906031397.0, + "step": 33803 + }, + { + "epoch": 2.7874989692421868, + "grad_norm": 1.7283376455307007, + "learning_rate": 1e-06, + "loss": 0.5268, + "mean_token_accuracy": 0.823140561580658, + "num_tokens": 906067039.0, + "step": 33804 + }, + { + "epoch": 2.7875814298672386, + "grad_norm": 1.922179102897644, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.7844361066818237, + "num_tokens": 906096762.0, + "step": 33805 + }, + { + "epoch": 2.78766389049229, + "grad_norm": 2.1437056064605713, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.7857034206390381, + "num_tokens": 906123186.0, + "step": 33806 + }, + { + "epoch": 2.7877463511173417, + "grad_norm": 2.1035642623901367, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.778232216835022, + "num_tokens": 906153599.0, + "step": 33807 + }, + { + "epoch": 2.787828811742393, + "grad_norm": 2.2218239307403564, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7348716855049133, + "num_tokens": 906181643.0, + "step": 33808 + }, + { + "epoch": 2.7879112723674444, + "grad_norm": 2.1691060066223145, + "learning_rate": 1e-06, + "loss": 0.6025, + "mean_token_accuracy": 0.7883872985839844, + "num_tokens": 906210653.0, + "step": 33809 + }, + { + "epoch": 2.787993732992496, + "grad_norm": 2.40985107421875, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7504887580871582, + "num_tokens": 906234804.0, + "step": 33810 + }, + { + "epoch": 2.7880761936175475, + "grad_norm": 2.1945626735687256, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7791796922683716, + "num_tokens": 906259762.0, + "step": 33811 + }, + { + "epoch": 2.788158654242599, + "grad_norm": 2.3604514598846436, + "learning_rate": 1e-06, + "loss": 0.7614, + "mean_token_accuracy": 0.757736086845398, + "num_tokens": 906287930.0, + "step": 33812 + }, + { + "epoch": 2.7882411148676507, + "grad_norm": 2.0362558364868164, + "learning_rate": 1e-06, + "loss": 0.6415, + "mean_token_accuracy": 0.7862378358840942, + "num_tokens": 906319595.0, + "step": 33813 + }, + { + "epoch": 2.7883235754927025, + "grad_norm": 2.453145742416382, + "learning_rate": 1e-06, + "loss": 0.7375, + "mean_token_accuracy": 0.7553982734680176, + "num_tokens": 906342563.0, + "step": 33814 + }, + { + "epoch": 2.788406036117754, + "grad_norm": 2.067612409591675, + "learning_rate": 1e-06, + "loss": 0.7559, + "mean_token_accuracy": 0.7548419237136841, + "num_tokens": 906373145.0, + "step": 33815 + }, + { + "epoch": 2.788488496742805, + "grad_norm": 2.0391733646392822, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.7857315540313721, + "num_tokens": 906406169.0, + "step": 33816 + }, + { + "epoch": 2.788570957367857, + "grad_norm": 2.2592575550079346, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.7597599625587463, + "num_tokens": 906436991.0, + "step": 33817 + }, + { + "epoch": 2.7886534179929083, + "grad_norm": 2.318345308303833, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.7807464599609375, + "num_tokens": 906462954.0, + "step": 33818 + }, + { + "epoch": 2.7887358786179597, + "grad_norm": 2.40752911567688, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7507344484329224, + "num_tokens": 906488867.0, + "step": 33819 + }, + { + "epoch": 2.7888183392430115, + "grad_norm": 2.3627166748046875, + "learning_rate": 1e-06, + "loss": 0.7963, + "mean_token_accuracy": 0.7469062209129333, + "num_tokens": 906513465.0, + "step": 33820 + }, + { + "epoch": 2.7889007998680633, + "grad_norm": 1.9470714330673218, + "learning_rate": 1e-06, + "loss": 0.7919, + "mean_token_accuracy": 0.7493982911109924, + "num_tokens": 906547150.0, + "step": 33821 + }, + { + "epoch": 2.7889832604931146, + "grad_norm": 2.3459606170654297, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7451280951499939, + "num_tokens": 906574247.0, + "step": 33822 + }, + { + "epoch": 2.789065721118166, + "grad_norm": 2.076936721801758, + "learning_rate": 1e-06, + "loss": 0.5967, + "mean_token_accuracy": 0.8037593364715576, + "num_tokens": 906602187.0, + "step": 33823 + }, + { + "epoch": 2.7891481817432178, + "grad_norm": 1.9348074197769165, + "learning_rate": 1e-06, + "loss": 0.6997, + "mean_token_accuracy": 0.7741276025772095, + "num_tokens": 906635101.0, + "step": 33824 + }, + { + "epoch": 2.789230642368269, + "grad_norm": 1.9990410804748535, + "learning_rate": 1e-06, + "loss": 0.567, + "mean_token_accuracy": 0.811211884021759, + "num_tokens": 906664802.0, + "step": 33825 + }, + { + "epoch": 2.7893131029933205, + "grad_norm": 2.0143537521362305, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7586309909820557, + "num_tokens": 906696997.0, + "step": 33826 + }, + { + "epoch": 2.7893955636183723, + "grad_norm": 2.228785753250122, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7532705068588257, + "num_tokens": 906722812.0, + "step": 33827 + }, + { + "epoch": 2.7894780242434236, + "grad_norm": 1.8034782409667969, + "learning_rate": 1e-06, + "loss": 0.525, + "mean_token_accuracy": 0.8247308731079102, + "num_tokens": 906756488.0, + "step": 33828 + }, + { + "epoch": 2.7895604848684754, + "grad_norm": 2.041377067565918, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7676063776016235, + "num_tokens": 906785794.0, + "step": 33829 + }, + { + "epoch": 2.7896429454935268, + "grad_norm": 2.07777738571167, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.752702534198761, + "num_tokens": 906816768.0, + "step": 33830 + }, + { + "epoch": 2.7897254061185786, + "grad_norm": 2.162163019180298, + "learning_rate": 1e-06, + "loss": 0.7149, + "mean_token_accuracy": 0.7691267132759094, + "num_tokens": 906847652.0, + "step": 33831 + }, + { + "epoch": 2.78980786674363, + "grad_norm": 2.341956615447998, + "learning_rate": 1e-06, + "loss": 0.7097, + "mean_token_accuracy": 0.76887047290802, + "num_tokens": 906875557.0, + "step": 33832 + }, + { + "epoch": 2.7898903273686813, + "grad_norm": 2.474910020828247, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7735961675643921, + "num_tokens": 906900913.0, + "step": 33833 + }, + { + "epoch": 2.789972787993733, + "grad_norm": 2.143852472305298, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.7508024573326111, + "num_tokens": 906927769.0, + "step": 33834 + }, + { + "epoch": 2.7900552486187844, + "grad_norm": 2.1091067790985107, + "learning_rate": 1e-06, + "loss": 0.8161, + "mean_token_accuracy": 0.7447760701179504, + "num_tokens": 906958027.0, + "step": 33835 + }, + { + "epoch": 2.790137709243836, + "grad_norm": 2.1465938091278076, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7711493968963623, + "num_tokens": 906988651.0, + "step": 33836 + }, + { + "epoch": 2.7902201698688875, + "grad_norm": 2.0939836502075195, + "learning_rate": 1e-06, + "loss": 0.8124, + "mean_token_accuracy": 0.7408374547958374, + "num_tokens": 907019155.0, + "step": 33837 + }, + { + "epoch": 2.7903026304939393, + "grad_norm": 1.9614105224609375, + "learning_rate": 1e-06, + "loss": 0.6502, + "mean_token_accuracy": 0.7812210321426392, + "num_tokens": 907055210.0, + "step": 33838 + }, + { + "epoch": 2.7903850911189907, + "grad_norm": 2.0644240379333496, + "learning_rate": 1e-06, + "loss": 0.7775, + "mean_token_accuracy": 0.7525483965873718, + "num_tokens": 907085069.0, + "step": 33839 + }, + { + "epoch": 2.790467551744042, + "grad_norm": 1.9329454898834229, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.7786073684692383, + "num_tokens": 907119650.0, + "step": 33840 + }, + { + "epoch": 2.790550012369094, + "grad_norm": 2.1782238483428955, + "learning_rate": 1e-06, + "loss": 0.7075, + "mean_token_accuracy": 0.7749475836753845, + "num_tokens": 907147277.0, + "step": 33841 + }, + { + "epoch": 2.790632472994145, + "grad_norm": 2.117363691329956, + "learning_rate": 1e-06, + "loss": 0.7637, + "mean_token_accuracy": 0.7599647045135498, + "num_tokens": 907179638.0, + "step": 33842 + }, + { + "epoch": 2.790714933619197, + "grad_norm": 2.315131664276123, + "learning_rate": 1e-06, + "loss": 0.695, + "mean_token_accuracy": 0.7699095010757446, + "num_tokens": 907206285.0, + "step": 33843 + }, + { + "epoch": 2.7907973942442483, + "grad_norm": 2.173330068588257, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.7881202697753906, + "num_tokens": 907239938.0, + "step": 33844 + }, + { + "epoch": 2.7908798548693, + "grad_norm": 1.923866629600525, + "learning_rate": 1e-06, + "loss": 0.5481, + "mean_token_accuracy": 0.820733904838562, + "num_tokens": 907271490.0, + "step": 33845 + }, + { + "epoch": 2.7909623154943515, + "grad_norm": 1.8599483966827393, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7927273511886597, + "num_tokens": 907305839.0, + "step": 33846 + }, + { + "epoch": 2.791044776119403, + "grad_norm": 2.3807783126831055, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7711589932441711, + "num_tokens": 907331210.0, + "step": 33847 + }, + { + "epoch": 2.7911272367444546, + "grad_norm": 2.0645880699157715, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.7643723487854004, + "num_tokens": 907364229.0, + "step": 33848 + }, + { + "epoch": 2.791209697369506, + "grad_norm": 2.0670926570892334, + "learning_rate": 1e-06, + "loss": 0.656, + "mean_token_accuracy": 0.7854148149490356, + "num_tokens": 907395671.0, + "step": 33849 + }, + { + "epoch": 2.7912921579945573, + "grad_norm": 1.980710744857788, + "learning_rate": 1e-06, + "loss": 0.6549, + "mean_token_accuracy": 0.7877678275108337, + "num_tokens": 907427268.0, + "step": 33850 + }, + { + "epoch": 2.791374618619609, + "grad_norm": 2.081916570663452, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.802115797996521, + "num_tokens": 907456818.0, + "step": 33851 + }, + { + "epoch": 2.791457079244661, + "grad_norm": 2.642073631286621, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7731592655181885, + "num_tokens": 907485971.0, + "step": 33852 + }, + { + "epoch": 2.7915395398697123, + "grad_norm": 2.1397705078125, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7680683135986328, + "num_tokens": 907515938.0, + "step": 33853 + }, + { + "epoch": 2.7916220004947636, + "grad_norm": 2.026071786880493, + "learning_rate": 1e-06, + "loss": 0.5839, + "mean_token_accuracy": 0.7964653968811035, + "num_tokens": 907546005.0, + "step": 33854 + }, + { + "epoch": 2.7917044611198154, + "grad_norm": 2.4101686477661133, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7777836322784424, + "num_tokens": 907569296.0, + "step": 33855 + }, + { + "epoch": 2.7917869217448668, + "grad_norm": 1.9958761930465698, + "learning_rate": 1e-06, + "loss": 0.6269, + "mean_token_accuracy": 0.7887464761734009, + "num_tokens": 907598986.0, + "step": 33856 + }, + { + "epoch": 2.791869382369918, + "grad_norm": 1.939444899559021, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7845250964164734, + "num_tokens": 907632641.0, + "step": 33857 + }, + { + "epoch": 2.79195184299497, + "grad_norm": 2.116539716720581, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.7865179181098938, + "num_tokens": 907662274.0, + "step": 33858 + }, + { + "epoch": 2.7920343036200217, + "grad_norm": 1.963176965713501, + "learning_rate": 1e-06, + "loss": 0.6398, + "mean_token_accuracy": 0.7908027172088623, + "num_tokens": 907697597.0, + "step": 33859 + }, + { + "epoch": 2.792116764245073, + "grad_norm": 2.0445806980133057, + "learning_rate": 1e-06, + "loss": 0.5914, + "mean_token_accuracy": 0.7995198965072632, + "num_tokens": 907726717.0, + "step": 33860 + }, + { + "epoch": 2.7921992248701244, + "grad_norm": 1.9491970539093018, + "learning_rate": 1e-06, + "loss": 0.5744, + "mean_token_accuracy": 0.8085602521896362, + "num_tokens": 907758364.0, + "step": 33861 + }, + { + "epoch": 2.792281685495176, + "grad_norm": 2.068444013595581, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7474662065505981, + "num_tokens": 907792520.0, + "step": 33862 + }, + { + "epoch": 2.7923641461202275, + "grad_norm": 1.8819659948349, + "learning_rate": 1e-06, + "loss": 0.5693, + "mean_token_accuracy": 0.8089656829833984, + "num_tokens": 907826589.0, + "step": 33863 + }, + { + "epoch": 2.792446606745279, + "grad_norm": 2.257388114929199, + "learning_rate": 1e-06, + "loss": 0.6259, + "mean_token_accuracy": 0.7874999046325684, + "num_tokens": 907851763.0, + "step": 33864 + }, + { + "epoch": 2.7925290673703307, + "grad_norm": 2.1106622219085693, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7491923570632935, + "num_tokens": 907879422.0, + "step": 33865 + }, + { + "epoch": 2.7926115279953825, + "grad_norm": 2.1940596103668213, + "learning_rate": 1e-06, + "loss": 0.7129, + "mean_token_accuracy": 0.777029275894165, + "num_tokens": 907907751.0, + "step": 33866 + }, + { + "epoch": 2.792693988620434, + "grad_norm": 2.216494083404541, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7593567371368408, + "num_tokens": 907935748.0, + "step": 33867 + }, + { + "epoch": 2.792776449245485, + "grad_norm": 2.5445468425750732, + "learning_rate": 1e-06, + "loss": 0.7426, + "mean_token_accuracy": 0.7558603286743164, + "num_tokens": 907957525.0, + "step": 33868 + }, + { + "epoch": 2.792858909870537, + "grad_norm": 1.855745553970337, + "learning_rate": 1e-06, + "loss": 0.5349, + "mean_token_accuracy": 0.8121312260627747, + "num_tokens": 907990185.0, + "step": 33869 + }, + { + "epoch": 2.7929413704955883, + "grad_norm": 2.1861729621887207, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7736238241195679, + "num_tokens": 908017734.0, + "step": 33870 + }, + { + "epoch": 2.7930238311206397, + "grad_norm": 1.8045685291290283, + "learning_rate": 1e-06, + "loss": 0.5881, + "mean_token_accuracy": 0.8002731800079346, + "num_tokens": 908052831.0, + "step": 33871 + }, + { + "epoch": 2.7931062917456915, + "grad_norm": 2.0496156215667725, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7680160403251648, + "num_tokens": 908083844.0, + "step": 33872 + }, + { + "epoch": 2.793188752370743, + "grad_norm": 2.0609700679779053, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7672646641731262, + "num_tokens": 908114405.0, + "step": 33873 + }, + { + "epoch": 2.7932712129957946, + "grad_norm": 2.0116517543792725, + "learning_rate": 1e-06, + "loss": 0.562, + "mean_token_accuracy": 0.8082572221755981, + "num_tokens": 908146543.0, + "step": 33874 + }, + { + "epoch": 2.793353673620846, + "grad_norm": 2.134528398513794, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.7794592380523682, + "num_tokens": 908177349.0, + "step": 33875 + }, + { + "epoch": 2.7934361342458978, + "grad_norm": 2.001089096069336, + "learning_rate": 1e-06, + "loss": 0.7313, + "mean_token_accuracy": 0.7582237720489502, + "num_tokens": 908210191.0, + "step": 33876 + }, + { + "epoch": 2.793518594870949, + "grad_norm": 1.7285293340682983, + "learning_rate": 1e-06, + "loss": 0.5795, + "mean_token_accuracy": 0.8022550344467163, + "num_tokens": 908251634.0, + "step": 33877 + }, + { + "epoch": 2.7936010554960005, + "grad_norm": 2.104940414428711, + "learning_rate": 1e-06, + "loss": 0.643, + "mean_token_accuracy": 0.780993640422821, + "num_tokens": 908279609.0, + "step": 33878 + }, + { + "epoch": 2.7936835161210523, + "grad_norm": 1.9632755517959595, + "learning_rate": 1e-06, + "loss": 0.5713, + "mean_token_accuracy": 0.8022053241729736, + "num_tokens": 908311542.0, + "step": 33879 + }, + { + "epoch": 2.7937659767461036, + "grad_norm": 2.3149547576904297, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7623662948608398, + "num_tokens": 908337067.0, + "step": 33880 + }, + { + "epoch": 2.7938484373711554, + "grad_norm": 1.9530588388442993, + "learning_rate": 1e-06, + "loss": 0.5058, + "mean_token_accuracy": 0.8257355690002441, + "num_tokens": 908367624.0, + "step": 33881 + }, + { + "epoch": 2.7939308979962068, + "grad_norm": 1.9132152795791626, + "learning_rate": 1e-06, + "loss": 0.4976, + "mean_token_accuracy": 0.8333269953727722, + "num_tokens": 908402422.0, + "step": 33882 + }, + { + "epoch": 2.7940133586212585, + "grad_norm": 2.1268744468688965, + "learning_rate": 1e-06, + "loss": 0.5442, + "mean_token_accuracy": 0.8224883079528809, + "num_tokens": 908427540.0, + "step": 33883 + }, + { + "epoch": 2.79409581924631, + "grad_norm": 2.341607093811035, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7775973081588745, + "num_tokens": 908453994.0, + "step": 33884 + }, + { + "epoch": 2.7941782798713612, + "grad_norm": 2.2137444019317627, + "learning_rate": 1e-06, + "loss": 0.5867, + "mean_token_accuracy": 0.80005943775177, + "num_tokens": 908482124.0, + "step": 33885 + }, + { + "epoch": 2.794260740496413, + "grad_norm": 2.1295573711395264, + "learning_rate": 1e-06, + "loss": 0.5487, + "mean_token_accuracy": 0.8043974041938782, + "num_tokens": 908509455.0, + "step": 33886 + }, + { + "epoch": 2.7943432011214644, + "grad_norm": 2.244934320449829, + "learning_rate": 1e-06, + "loss": 0.7853, + "mean_token_accuracy": 0.7421834468841553, + "num_tokens": 908537026.0, + "step": 33887 + }, + { + "epoch": 2.794425661746516, + "grad_norm": 1.9588872194290161, + "learning_rate": 1e-06, + "loss": 0.6343, + "mean_token_accuracy": 0.7877774834632874, + "num_tokens": 908569184.0, + "step": 33888 + }, + { + "epoch": 2.7945081223715675, + "grad_norm": 2.0597238540649414, + "learning_rate": 1e-06, + "loss": 0.6134, + "mean_token_accuracy": 0.8021317720413208, + "num_tokens": 908598988.0, + "step": 33889 + }, + { + "epoch": 2.7945905829966193, + "grad_norm": 2.084068775177002, + "learning_rate": 1e-06, + "loss": 0.6196, + "mean_token_accuracy": 0.806795597076416, + "num_tokens": 908628681.0, + "step": 33890 + }, + { + "epoch": 2.7946730436216707, + "grad_norm": 1.9455736875534058, + "learning_rate": 1e-06, + "loss": 0.7887, + "mean_token_accuracy": 0.7515103816986084, + "num_tokens": 908664366.0, + "step": 33891 + }, + { + "epoch": 2.794755504246722, + "grad_norm": 2.4000678062438965, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7688488960266113, + "num_tokens": 908688586.0, + "step": 33892 + }, + { + "epoch": 2.794837964871774, + "grad_norm": 2.022038221359253, + "learning_rate": 1e-06, + "loss": 0.6646, + "mean_token_accuracy": 0.775873064994812, + "num_tokens": 908723094.0, + "step": 33893 + }, + { + "epoch": 2.794920425496825, + "grad_norm": 2.1285691261291504, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7378232479095459, + "num_tokens": 908753940.0, + "step": 33894 + }, + { + "epoch": 2.7950028861218765, + "grad_norm": 1.9954791069030762, + "learning_rate": 1e-06, + "loss": 0.6021, + "mean_token_accuracy": 0.7944182753562927, + "num_tokens": 908785494.0, + "step": 33895 + }, + { + "epoch": 2.7950853467469283, + "grad_norm": 2.1757419109344482, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7726567387580872, + "num_tokens": 908811127.0, + "step": 33896 + }, + { + "epoch": 2.79516780737198, + "grad_norm": 1.9833476543426514, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7807209491729736, + "num_tokens": 908842873.0, + "step": 33897 + }, + { + "epoch": 2.7952502679970315, + "grad_norm": 1.9088470935821533, + "learning_rate": 1e-06, + "loss": 0.6085, + "mean_token_accuracy": 0.7807478904724121, + "num_tokens": 908876278.0, + "step": 33898 + }, + { + "epoch": 2.795332728622083, + "grad_norm": 2.0355231761932373, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7492153644561768, + "num_tokens": 908907767.0, + "step": 33899 + }, + { + "epoch": 2.7954151892471346, + "grad_norm": 2.226729393005371, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7578524351119995, + "num_tokens": 908939213.0, + "step": 33900 + }, + { + "epoch": 2.795497649872186, + "grad_norm": 2.4515321254730225, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.7857051491737366, + "num_tokens": 908962333.0, + "step": 33901 + }, + { + "epoch": 2.7955801104972373, + "grad_norm": 2.2214369773864746, + "learning_rate": 1e-06, + "loss": 0.7352, + "mean_token_accuracy": 0.7685970067977905, + "num_tokens": 908991447.0, + "step": 33902 + }, + { + "epoch": 2.795662571122289, + "grad_norm": 2.1718695163726807, + "learning_rate": 1e-06, + "loss": 0.6449, + "mean_token_accuracy": 0.76731938123703, + "num_tokens": 909021926.0, + "step": 33903 + }, + { + "epoch": 2.795745031747341, + "grad_norm": 1.8862733840942383, + "learning_rate": 1e-06, + "loss": 0.6347, + "mean_token_accuracy": 0.7941769361495972, + "num_tokens": 909055193.0, + "step": 33904 + }, + { + "epoch": 2.7958274923723923, + "grad_norm": 2.3199875354766846, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7521935105323792, + "num_tokens": 909081694.0, + "step": 33905 + }, + { + "epoch": 2.7959099529974436, + "grad_norm": 2.163630247116089, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7226327657699585, + "num_tokens": 909112749.0, + "step": 33906 + }, + { + "epoch": 2.7959924136224954, + "grad_norm": 1.9078363180160522, + "learning_rate": 1e-06, + "loss": 0.5712, + "mean_token_accuracy": 0.813524067401886, + "num_tokens": 909143956.0, + "step": 33907 + }, + { + "epoch": 2.7960748742475467, + "grad_norm": 2.6593527793884277, + "learning_rate": 1e-06, + "loss": 0.8055, + "mean_token_accuracy": 0.744003176689148, + "num_tokens": 909164911.0, + "step": 33908 + }, + { + "epoch": 2.796157334872598, + "grad_norm": 2.1837258338928223, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.76938396692276, + "num_tokens": 909193793.0, + "step": 33909 + }, + { + "epoch": 2.79623979549765, + "grad_norm": 2.0375118255615234, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.7933270931243896, + "num_tokens": 909224366.0, + "step": 33910 + }, + { + "epoch": 2.7963222561227012, + "grad_norm": 2.13213849067688, + "learning_rate": 1e-06, + "loss": 0.7552, + "mean_token_accuracy": 0.7562133073806763, + "num_tokens": 909254119.0, + "step": 33911 + }, + { + "epoch": 2.796404716747753, + "grad_norm": 2.196171998977661, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7759992480278015, + "num_tokens": 909282726.0, + "step": 33912 + }, + { + "epoch": 2.7964871773728044, + "grad_norm": 2.244392156600952, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7524216175079346, + "num_tokens": 909310536.0, + "step": 33913 + }, + { + "epoch": 2.796569637997856, + "grad_norm": 2.183900833129883, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7503869533538818, + "num_tokens": 909337444.0, + "step": 33914 + }, + { + "epoch": 2.7966520986229075, + "grad_norm": 2.174696922302246, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.7846165895462036, + "num_tokens": 909366104.0, + "step": 33915 + }, + { + "epoch": 2.796734559247959, + "grad_norm": 1.7786586284637451, + "learning_rate": 1e-06, + "loss": 0.4688, + "mean_token_accuracy": 0.8406507968902588, + "num_tokens": 909401728.0, + "step": 33916 + }, + { + "epoch": 2.7968170198730107, + "grad_norm": 2.2026774883270264, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7612749338150024, + "num_tokens": 909430067.0, + "step": 33917 + }, + { + "epoch": 2.796899480498062, + "grad_norm": 2.096041202545166, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7650479078292847, + "num_tokens": 909461884.0, + "step": 33918 + }, + { + "epoch": 2.796981941123114, + "grad_norm": 1.9609121084213257, + "learning_rate": 1e-06, + "loss": 0.6981, + "mean_token_accuracy": 0.7699925899505615, + "num_tokens": 909494963.0, + "step": 33919 + }, + { + "epoch": 2.797064401748165, + "grad_norm": 2.1060256958007812, + "learning_rate": 1e-06, + "loss": 0.7123, + "mean_token_accuracy": 0.7700120806694031, + "num_tokens": 909523035.0, + "step": 33920 + }, + { + "epoch": 2.797146862373217, + "grad_norm": 2.0145151615142822, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7609821557998657, + "num_tokens": 909556516.0, + "step": 33921 + }, + { + "epoch": 2.7972293229982683, + "grad_norm": 2.0565953254699707, + "learning_rate": 1e-06, + "loss": 0.7743, + "mean_token_accuracy": 0.758918285369873, + "num_tokens": 909586780.0, + "step": 33922 + }, + { + "epoch": 2.7973117836233197, + "grad_norm": 2.164180278778076, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7266194820404053, + "num_tokens": 909616508.0, + "step": 33923 + }, + { + "epoch": 2.7973942442483715, + "grad_norm": 2.046677589416504, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.7593089938163757, + "num_tokens": 909652546.0, + "step": 33924 + }, + { + "epoch": 2.797476704873423, + "grad_norm": 1.9332853555679321, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7690929770469666, + "num_tokens": 909687559.0, + "step": 33925 + }, + { + "epoch": 2.7975591654984746, + "grad_norm": 2.1897153854370117, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7525561451911926, + "num_tokens": 909718793.0, + "step": 33926 + }, + { + "epoch": 2.797641626123526, + "grad_norm": 2.0949950218200684, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7378244996070862, + "num_tokens": 909749661.0, + "step": 33927 + }, + { + "epoch": 2.7977240867485778, + "grad_norm": 2.3134121894836426, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.752611517906189, + "num_tokens": 909775858.0, + "step": 33928 + }, + { + "epoch": 2.797806547373629, + "grad_norm": 1.8052691221237183, + "learning_rate": 1e-06, + "loss": 0.516, + "mean_token_accuracy": 0.8284776210784912, + "num_tokens": 909813547.0, + "step": 33929 + }, + { + "epoch": 2.7978890079986805, + "grad_norm": 2.2362918853759766, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7648944854736328, + "num_tokens": 909841559.0, + "step": 33930 + }, + { + "epoch": 2.7979714686237322, + "grad_norm": 2.042475700378418, + "learning_rate": 1e-06, + "loss": 0.6828, + "mean_token_accuracy": 0.7775382995605469, + "num_tokens": 909873174.0, + "step": 33931 + }, + { + "epoch": 2.7980539292487836, + "grad_norm": 1.9227919578552246, + "learning_rate": 1e-06, + "loss": 0.6138, + "mean_token_accuracy": 0.7983971238136292, + "num_tokens": 909904968.0, + "step": 33932 + }, + { + "epoch": 2.7981363898738354, + "grad_norm": 2.2480404376983643, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.775362491607666, + "num_tokens": 909930260.0, + "step": 33933 + }, + { + "epoch": 2.7982188504988867, + "grad_norm": 2.271177053451538, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7556981444358826, + "num_tokens": 909956231.0, + "step": 33934 + }, + { + "epoch": 2.7983013111239385, + "grad_norm": 3.217294931411743, + "learning_rate": 1e-06, + "loss": 0.7404, + "mean_token_accuracy": 0.7579948306083679, + "num_tokens": 909987307.0, + "step": 33935 + }, + { + "epoch": 2.79838377174899, + "grad_norm": 2.228285789489746, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.790349006652832, + "num_tokens": 910013620.0, + "step": 33936 + }, + { + "epoch": 2.7984662323740412, + "grad_norm": 2.511110782623291, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7846869230270386, + "num_tokens": 910035364.0, + "step": 33937 + }, + { + "epoch": 2.798548692999093, + "grad_norm": 2.3299551010131836, + "learning_rate": 1e-06, + "loss": 0.6741, + "mean_token_accuracy": 0.7810410261154175, + "num_tokens": 910061043.0, + "step": 33938 + }, + { + "epoch": 2.7986311536241444, + "grad_norm": 2.1149120330810547, + "learning_rate": 1e-06, + "loss": 0.7411, + "mean_token_accuracy": 0.7553001642227173, + "num_tokens": 910093005.0, + "step": 33939 + }, + { + "epoch": 2.7987136142491957, + "grad_norm": 2.264281988143921, + "learning_rate": 1e-06, + "loss": 0.6913, + "mean_token_accuracy": 0.7668363451957703, + "num_tokens": 910119023.0, + "step": 33940 + }, + { + "epoch": 2.7987960748742475, + "grad_norm": 1.8686317205429077, + "learning_rate": 1e-06, + "loss": 0.6131, + "mean_token_accuracy": 0.7974894642829895, + "num_tokens": 910157582.0, + "step": 33941 + }, + { + "epoch": 2.7988785354992993, + "grad_norm": 2.17111873626709, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7311890125274658, + "num_tokens": 910186958.0, + "step": 33942 + }, + { + "epoch": 2.7989609961243507, + "grad_norm": 2.0541303157806396, + "learning_rate": 1e-06, + "loss": 0.5605, + "mean_token_accuracy": 0.8102924823760986, + "num_tokens": 910218288.0, + "step": 33943 + }, + { + "epoch": 2.799043456749402, + "grad_norm": 2.1938228607177734, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7525086998939514, + "num_tokens": 910250540.0, + "step": 33944 + }, + { + "epoch": 2.799125917374454, + "grad_norm": 2.0727157592773438, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7331421375274658, + "num_tokens": 910283669.0, + "step": 33945 + }, + { + "epoch": 2.799208377999505, + "grad_norm": 2.0462801456451416, + "learning_rate": 1e-06, + "loss": 0.5727, + "mean_token_accuracy": 0.7939671277999878, + "num_tokens": 910313759.0, + "step": 33946 + }, + { + "epoch": 2.7992908386245565, + "grad_norm": 2.176396608352661, + "learning_rate": 1e-06, + "loss": 0.7185, + "mean_token_accuracy": 0.7683196663856506, + "num_tokens": 910341118.0, + "step": 33947 + }, + { + "epoch": 2.7993732992496083, + "grad_norm": 2.0615196228027344, + "learning_rate": 1e-06, + "loss": 0.7634, + "mean_token_accuracy": 0.7525943517684937, + "num_tokens": 910372624.0, + "step": 33948 + }, + { + "epoch": 2.79945575987466, + "grad_norm": 1.9666779041290283, + "learning_rate": 1e-06, + "loss": 0.628, + "mean_token_accuracy": 0.7807540893554688, + "num_tokens": 910405392.0, + "step": 33949 + }, + { + "epoch": 2.7995382204997115, + "grad_norm": 2.4370529651641846, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7697396278381348, + "num_tokens": 910429990.0, + "step": 33950 + }, + { + "epoch": 2.799620681124763, + "grad_norm": 2.0848588943481445, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.7804596424102783, + "num_tokens": 910460404.0, + "step": 33951 + }, + { + "epoch": 2.7997031417498146, + "grad_norm": 2.069068431854248, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.7868518829345703, + "num_tokens": 910492035.0, + "step": 33952 + }, + { + "epoch": 2.799785602374866, + "grad_norm": 2.0382916927337646, + "learning_rate": 1e-06, + "loss": 0.5994, + "mean_token_accuracy": 0.8060786724090576, + "num_tokens": 910524611.0, + "step": 33953 + }, + { + "epoch": 2.7998680629999173, + "grad_norm": 2.312579870223999, + "learning_rate": 1e-06, + "loss": 0.6361, + "mean_token_accuracy": 0.782673716545105, + "num_tokens": 910550473.0, + "step": 33954 + }, + { + "epoch": 2.799950523624969, + "grad_norm": 2.3869545459747314, + "learning_rate": 1e-06, + "loss": 0.618, + "mean_token_accuracy": 0.7968606948852539, + "num_tokens": 910573731.0, + "step": 33955 + }, + { + "epoch": 2.8000329842500205, + "grad_norm": 2.02121639251709, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7776539325714111, + "num_tokens": 910605918.0, + "step": 33956 + }, + { + "epoch": 2.8001154448750722, + "grad_norm": 2.116706371307373, + "learning_rate": 1e-06, + "loss": 0.7512, + "mean_token_accuracy": 0.7577293515205383, + "num_tokens": 910635989.0, + "step": 33957 + }, + { + "epoch": 2.8001979055001236, + "grad_norm": 2.331533670425415, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7401471734046936, + "num_tokens": 910662401.0, + "step": 33958 + }, + { + "epoch": 2.8002803661251754, + "grad_norm": 2.3407366275787354, + "learning_rate": 1e-06, + "loss": 0.6252, + "mean_token_accuracy": 0.7925630807876587, + "num_tokens": 910684868.0, + "step": 33959 + }, + { + "epoch": 2.8003628267502267, + "grad_norm": 1.9056605100631714, + "learning_rate": 1e-06, + "loss": 0.5792, + "mean_token_accuracy": 0.8145151138305664, + "num_tokens": 910716672.0, + "step": 33960 + }, + { + "epoch": 2.800445287375278, + "grad_norm": 2.216479539871216, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7819293141365051, + "num_tokens": 910745478.0, + "step": 33961 + }, + { + "epoch": 2.80052774800033, + "grad_norm": 1.940364956855774, + "learning_rate": 1e-06, + "loss": 0.5306, + "mean_token_accuracy": 0.8013480305671692, + "num_tokens": 910776570.0, + "step": 33962 + }, + { + "epoch": 2.8006102086253812, + "grad_norm": 2.0480804443359375, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.770684540271759, + "num_tokens": 910808101.0, + "step": 33963 + }, + { + "epoch": 2.800692669250433, + "grad_norm": 2.252532720565796, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.7737830877304077, + "num_tokens": 910833246.0, + "step": 33964 + }, + { + "epoch": 2.8007751298754844, + "grad_norm": 1.9206024408340454, + "learning_rate": 1e-06, + "loss": 0.7237, + "mean_token_accuracy": 0.7750137448310852, + "num_tokens": 910869040.0, + "step": 33965 + }, + { + "epoch": 2.800857590500536, + "grad_norm": 2.045971393585205, + "learning_rate": 1e-06, + "loss": 0.6804, + "mean_token_accuracy": 0.7823044657707214, + "num_tokens": 910900153.0, + "step": 33966 + }, + { + "epoch": 2.8009400511255875, + "grad_norm": 2.042741060256958, + "learning_rate": 1e-06, + "loss": 0.5995, + "mean_token_accuracy": 0.7865578532218933, + "num_tokens": 910930258.0, + "step": 33967 + }, + { + "epoch": 2.801022511750639, + "grad_norm": 2.2601194381713867, + "learning_rate": 1e-06, + "loss": 0.6385, + "mean_token_accuracy": 0.7869887948036194, + "num_tokens": 910955992.0, + "step": 33968 + }, + { + "epoch": 2.8011049723756907, + "grad_norm": 2.263152599334717, + "learning_rate": 1e-06, + "loss": 0.6665, + "mean_token_accuracy": 0.785176157951355, + "num_tokens": 910982892.0, + "step": 33969 + }, + { + "epoch": 2.801187433000742, + "grad_norm": 1.861072301864624, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.744490385055542, + "num_tokens": 911021660.0, + "step": 33970 + }, + { + "epoch": 2.801269893625794, + "grad_norm": 2.1341283321380615, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7319896221160889, + "num_tokens": 911052330.0, + "step": 33971 + }, + { + "epoch": 2.801352354250845, + "grad_norm": 1.8366823196411133, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.7936029434204102, + "num_tokens": 911088805.0, + "step": 33972 + }, + { + "epoch": 2.801434814875897, + "grad_norm": 2.108485460281372, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7822402119636536, + "num_tokens": 911120378.0, + "step": 33973 + }, + { + "epoch": 2.8015172755009483, + "grad_norm": 2.2581751346588135, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7666305303573608, + "num_tokens": 911147762.0, + "step": 33974 + }, + { + "epoch": 2.8015997361259997, + "grad_norm": 2.0391364097595215, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7243045568466187, + "num_tokens": 911182631.0, + "step": 33975 + }, + { + "epoch": 2.8016821967510515, + "grad_norm": 2.0340371131896973, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7627174258232117, + "num_tokens": 911213571.0, + "step": 33976 + }, + { + "epoch": 2.801764657376103, + "grad_norm": 1.9732064008712769, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7726438045501709, + "num_tokens": 911246856.0, + "step": 33977 + }, + { + "epoch": 2.8018471180011546, + "grad_norm": 2.2341442108154297, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7440425157546997, + "num_tokens": 911275086.0, + "step": 33978 + }, + { + "epoch": 2.801929578626206, + "grad_norm": 2.231295108795166, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7407917380332947, + "num_tokens": 911302212.0, + "step": 33979 + }, + { + "epoch": 2.8020120392512577, + "grad_norm": 2.2020866870880127, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.7660622000694275, + "num_tokens": 911332648.0, + "step": 33980 + }, + { + "epoch": 2.802094499876309, + "grad_norm": 2.3260960578918457, + "learning_rate": 1e-06, + "loss": 0.7868, + "mean_token_accuracy": 0.7502787709236145, + "num_tokens": 911360315.0, + "step": 33981 + }, + { + "epoch": 2.8021769605013604, + "grad_norm": 2.1207008361816406, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.766512393951416, + "num_tokens": 911391806.0, + "step": 33982 + }, + { + "epoch": 2.8022594211264122, + "grad_norm": 2.117039203643799, + "learning_rate": 1e-06, + "loss": 0.5797, + "mean_token_accuracy": 0.7895199060440063, + "num_tokens": 911420327.0, + "step": 33983 + }, + { + "epoch": 2.8023418817514636, + "grad_norm": 1.9613523483276367, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.7870586514472961, + "num_tokens": 911451878.0, + "step": 33984 + }, + { + "epoch": 2.802424342376515, + "grad_norm": 2.2082817554473877, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7542639970779419, + "num_tokens": 911481726.0, + "step": 33985 + }, + { + "epoch": 2.8025068030015667, + "grad_norm": 2.081888437271118, + "learning_rate": 1e-06, + "loss": 0.5558, + "mean_token_accuracy": 0.810077428817749, + "num_tokens": 911508398.0, + "step": 33986 + }, + { + "epoch": 2.8025892636266185, + "grad_norm": 2.183168888092041, + "learning_rate": 1e-06, + "loss": 0.7972, + "mean_token_accuracy": 0.7439535856246948, + "num_tokens": 911539039.0, + "step": 33987 + }, + { + "epoch": 2.80267172425167, + "grad_norm": 2.247741937637329, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7445184588432312, + "num_tokens": 911568302.0, + "step": 33988 + }, + { + "epoch": 2.8027541848767212, + "grad_norm": 1.6590979099273682, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7670707702636719, + "num_tokens": 911610097.0, + "step": 33989 + }, + { + "epoch": 2.802836645501773, + "grad_norm": 1.992634654045105, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7491577863693237, + "num_tokens": 911644006.0, + "step": 33990 + }, + { + "epoch": 2.8029191061268244, + "grad_norm": 1.9200540781021118, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7720341682434082, + "num_tokens": 911679440.0, + "step": 33991 + }, + { + "epoch": 2.8030015667518757, + "grad_norm": 2.140500783920288, + "learning_rate": 1e-06, + "loss": 0.5202, + "mean_token_accuracy": 0.8197402358055115, + "num_tokens": 911707251.0, + "step": 33992 + }, + { + "epoch": 2.8030840273769275, + "grad_norm": 2.0448596477508545, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7311128377914429, + "num_tokens": 911741270.0, + "step": 33993 + }, + { + "epoch": 2.8031664880019793, + "grad_norm": 2.0550713539123535, + "learning_rate": 1e-06, + "loss": 0.6749, + "mean_token_accuracy": 0.7824814319610596, + "num_tokens": 911771402.0, + "step": 33994 + }, + { + "epoch": 2.8032489486270307, + "grad_norm": 1.738006353378296, + "learning_rate": 1e-06, + "loss": 0.5885, + "mean_token_accuracy": 0.809679388999939, + "num_tokens": 911811221.0, + "step": 33995 + }, + { + "epoch": 2.803331409252082, + "grad_norm": 2.088819980621338, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7467309236526489, + "num_tokens": 911844075.0, + "step": 33996 + }, + { + "epoch": 2.803413869877134, + "grad_norm": 1.9088411331176758, + "learning_rate": 1e-06, + "loss": 0.5252, + "mean_token_accuracy": 0.8251570463180542, + "num_tokens": 911878910.0, + "step": 33997 + }, + { + "epoch": 2.803496330502185, + "grad_norm": 2.1693649291992188, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7501106858253479, + "num_tokens": 911908385.0, + "step": 33998 + }, + { + "epoch": 2.8035787911272365, + "grad_norm": 2.2628977298736572, + "learning_rate": 1e-06, + "loss": 0.565, + "mean_token_accuracy": 0.8034040927886963, + "num_tokens": 911933496.0, + "step": 33999 + }, + { + "epoch": 2.8036612517522883, + "grad_norm": 2.2049241065979004, + "learning_rate": 1e-06, + "loss": 0.6419, + "mean_token_accuracy": 0.7947366237640381, + "num_tokens": 911959688.0, + "step": 34000 + }, + { + "epoch": 2.8037437123773397, + "grad_norm": 2.065260410308838, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.7858557105064392, + "num_tokens": 911988219.0, + "step": 34001 + }, + { + "epoch": 2.8038261730023915, + "grad_norm": 2.001418113708496, + "learning_rate": 1e-06, + "loss": 0.5446, + "mean_token_accuracy": 0.8250072002410889, + "num_tokens": 912021732.0, + "step": 34002 + }, + { + "epoch": 2.803908633627443, + "grad_norm": 2.0114405155181885, + "learning_rate": 1e-06, + "loss": 0.7621, + "mean_token_accuracy": 0.7585654258728027, + "num_tokens": 912055132.0, + "step": 34003 + }, + { + "epoch": 2.8039910942524946, + "grad_norm": 2.1958510875701904, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7426570653915405, + "num_tokens": 912084416.0, + "step": 34004 + }, + { + "epoch": 2.804073554877546, + "grad_norm": 2.244957685470581, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.7905774712562561, + "num_tokens": 912109249.0, + "step": 34005 + }, + { + "epoch": 2.8041560155025973, + "grad_norm": 2.087700366973877, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7699958682060242, + "num_tokens": 912139601.0, + "step": 34006 + }, + { + "epoch": 2.804238476127649, + "grad_norm": 2.285970449447632, + "learning_rate": 1e-06, + "loss": 0.6302, + "mean_token_accuracy": 0.7886712551116943, + "num_tokens": 912161959.0, + "step": 34007 + }, + { + "epoch": 2.8043209367527004, + "grad_norm": 1.9176772832870483, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7738395929336548, + "num_tokens": 912195278.0, + "step": 34008 + }, + { + "epoch": 2.8044033973777522, + "grad_norm": 1.8824690580368042, + "learning_rate": 1e-06, + "loss": 0.7461, + "mean_token_accuracy": 0.7546354532241821, + "num_tokens": 912229460.0, + "step": 34009 + }, + { + "epoch": 2.8044858580028036, + "grad_norm": 1.9805338382720947, + "learning_rate": 1e-06, + "loss": 0.7215, + "mean_token_accuracy": 0.7607760429382324, + "num_tokens": 912263686.0, + "step": 34010 + }, + { + "epoch": 2.8045683186278554, + "grad_norm": 1.9570428133010864, + "learning_rate": 1e-06, + "loss": 0.8208, + "mean_token_accuracy": 0.7471920251846313, + "num_tokens": 912299425.0, + "step": 34011 + }, + { + "epoch": 2.8046507792529067, + "grad_norm": 1.9581457376480103, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7800440192222595, + "num_tokens": 912332426.0, + "step": 34012 + }, + { + "epoch": 2.804733239877958, + "grad_norm": 1.9162027835845947, + "learning_rate": 1e-06, + "loss": 0.6818, + "mean_token_accuracy": 0.7784833908081055, + "num_tokens": 912366820.0, + "step": 34013 + }, + { + "epoch": 2.80481570050301, + "grad_norm": 2.1753251552581787, + "learning_rate": 1e-06, + "loss": 0.6826, + "mean_token_accuracy": 0.7709659337997437, + "num_tokens": 912395371.0, + "step": 34014 + }, + { + "epoch": 2.8048981611280612, + "grad_norm": 2.4320905208587646, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7607746124267578, + "num_tokens": 912418857.0, + "step": 34015 + }, + { + "epoch": 2.804980621753113, + "grad_norm": 2.174553871154785, + "learning_rate": 1e-06, + "loss": 0.6128, + "mean_token_accuracy": 0.790030300617218, + "num_tokens": 912444913.0, + "step": 34016 + }, + { + "epoch": 2.8050630823781644, + "grad_norm": 2.0339863300323486, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.7899237275123596, + "num_tokens": 912475221.0, + "step": 34017 + }, + { + "epoch": 2.805145543003216, + "grad_norm": 2.212651252746582, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7851629257202148, + "num_tokens": 912505478.0, + "step": 34018 + }, + { + "epoch": 2.8052280036282675, + "grad_norm": 2.01835560798645, + "learning_rate": 1e-06, + "loss": 0.7306, + "mean_token_accuracy": 0.7633789777755737, + "num_tokens": 912538463.0, + "step": 34019 + }, + { + "epoch": 2.805310464253319, + "grad_norm": 2.2806899547576904, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7679808139801025, + "num_tokens": 912565390.0, + "step": 34020 + }, + { + "epoch": 2.8053929248783707, + "grad_norm": 2.500347137451172, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7581145763397217, + "num_tokens": 912588922.0, + "step": 34021 + }, + { + "epoch": 2.805475385503422, + "grad_norm": 2.0606813430786133, + "learning_rate": 1e-06, + "loss": 0.5676, + "mean_token_accuracy": 0.8031002283096313, + "num_tokens": 912620410.0, + "step": 34022 + }, + { + "epoch": 2.805557846128474, + "grad_norm": 2.199394941329956, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7800763249397278, + "num_tokens": 912646582.0, + "step": 34023 + }, + { + "epoch": 2.805640306753525, + "grad_norm": 2.1522412300109863, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.778399646282196, + "num_tokens": 912673309.0, + "step": 34024 + }, + { + "epoch": 2.805722767378577, + "grad_norm": 2.3368306159973145, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.7417712211608887, + "num_tokens": 912697880.0, + "step": 34025 + }, + { + "epoch": 2.8058052280036283, + "grad_norm": 2.204096794128418, + "learning_rate": 1e-06, + "loss": 0.603, + "mean_token_accuracy": 0.790934681892395, + "num_tokens": 912723708.0, + "step": 34026 + }, + { + "epoch": 2.8058876886286797, + "grad_norm": 2.4079222679138184, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7537804841995239, + "num_tokens": 912747930.0, + "step": 34027 + }, + { + "epoch": 2.8059701492537314, + "grad_norm": 1.8709355592727661, + "learning_rate": 1e-06, + "loss": 0.5713, + "mean_token_accuracy": 0.8066015243530273, + "num_tokens": 912785164.0, + "step": 34028 + }, + { + "epoch": 2.806052609878783, + "grad_norm": 2.2464115619659424, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7266578674316406, + "num_tokens": 912814364.0, + "step": 34029 + }, + { + "epoch": 2.806135070503834, + "grad_norm": 2.0249814987182617, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7258526086807251, + "num_tokens": 912847526.0, + "step": 34030 + }, + { + "epoch": 2.806217531128886, + "grad_norm": 2.0417680740356445, + "learning_rate": 1e-06, + "loss": 0.763, + "mean_token_accuracy": 0.7583274841308594, + "num_tokens": 912879845.0, + "step": 34031 + }, + { + "epoch": 2.8062999917539377, + "grad_norm": 2.0589163303375244, + "learning_rate": 1e-06, + "loss": 0.7773, + "mean_token_accuracy": 0.7506197690963745, + "num_tokens": 912912196.0, + "step": 34032 + }, + { + "epoch": 2.806382452378989, + "grad_norm": 2.0169286727905273, + "learning_rate": 1e-06, + "loss": 0.6578, + "mean_token_accuracy": 0.7758450508117676, + "num_tokens": 912943472.0, + "step": 34033 + }, + { + "epoch": 2.8064649130040404, + "grad_norm": 2.3220126628875732, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.785530149936676, + "num_tokens": 912968118.0, + "step": 34034 + }, + { + "epoch": 2.8065473736290922, + "grad_norm": 2.0313782691955566, + "learning_rate": 1e-06, + "loss": 0.5881, + "mean_token_accuracy": 0.8121914267539978, + "num_tokens": 913001027.0, + "step": 34035 + }, + { + "epoch": 2.8066298342541436, + "grad_norm": 1.8865132331848145, + "learning_rate": 1e-06, + "loss": 0.666, + "mean_token_accuracy": 0.7813749313354492, + "num_tokens": 913035215.0, + "step": 34036 + }, + { + "epoch": 2.806712294879195, + "grad_norm": 2.3208420276641846, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7622188329696655, + "num_tokens": 913061411.0, + "step": 34037 + }, + { + "epoch": 2.8067947555042467, + "grad_norm": 1.9002522230148315, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7715643644332886, + "num_tokens": 913097551.0, + "step": 34038 + }, + { + "epoch": 2.8068772161292985, + "grad_norm": 2.286129951477051, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7507162094116211, + "num_tokens": 913125114.0, + "step": 34039 + }, + { + "epoch": 2.80695967675435, + "grad_norm": 2.0422415733337402, + "learning_rate": 1e-06, + "loss": 0.5768, + "mean_token_accuracy": 0.8002637624740601, + "num_tokens": 913153596.0, + "step": 34040 + }, + { + "epoch": 2.8070421373794012, + "grad_norm": 1.8472918272018433, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.7950574159622192, + "num_tokens": 913189082.0, + "step": 34041 + }, + { + "epoch": 2.807124598004453, + "grad_norm": 2.2067604064941406, + "learning_rate": 1e-06, + "loss": 0.6242, + "mean_token_accuracy": 0.7986920475959778, + "num_tokens": 913216545.0, + "step": 34042 + }, + { + "epoch": 2.8072070586295044, + "grad_norm": 2.068551778793335, + "learning_rate": 1e-06, + "loss": 0.6509, + "mean_token_accuracy": 0.7847393751144409, + "num_tokens": 913245673.0, + "step": 34043 + }, + { + "epoch": 2.8072895192545557, + "grad_norm": 2.583648920059204, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7303327322006226, + "num_tokens": 913271590.0, + "step": 34044 + }, + { + "epoch": 2.8073719798796075, + "grad_norm": 1.990794062614441, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.761701762676239, + "num_tokens": 913306202.0, + "step": 34045 + }, + { + "epoch": 2.807454440504659, + "grad_norm": 2.1551010608673096, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.8010892271995544, + "num_tokens": 913332693.0, + "step": 34046 + }, + { + "epoch": 2.8075369011297107, + "grad_norm": 2.007207155227661, + "learning_rate": 1e-06, + "loss": 0.7519, + "mean_token_accuracy": 0.7670407295227051, + "num_tokens": 913365099.0, + "step": 34047 + }, + { + "epoch": 2.807619361754762, + "grad_norm": 2.0597143173217773, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.7910876870155334, + "num_tokens": 913394898.0, + "step": 34048 + }, + { + "epoch": 2.807701822379814, + "grad_norm": 2.15898060798645, + "learning_rate": 1e-06, + "loss": 0.6118, + "mean_token_accuracy": 0.7956167459487915, + "num_tokens": 913422759.0, + "step": 34049 + }, + { + "epoch": 2.807784283004865, + "grad_norm": 2.406393051147461, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7581170201301575, + "num_tokens": 913447545.0, + "step": 34050 + }, + { + "epoch": 2.8078667436299165, + "grad_norm": 1.7555522918701172, + "learning_rate": 1e-06, + "loss": 0.58, + "mean_token_accuracy": 0.8147352337837219, + "num_tokens": 913483986.0, + "step": 34051 + }, + { + "epoch": 2.8079492042549683, + "grad_norm": 2.3077032566070557, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7575788497924805, + "num_tokens": 913509113.0, + "step": 34052 + }, + { + "epoch": 2.8080316648800197, + "grad_norm": 2.1322107315063477, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.7756000757217407, + "num_tokens": 913534940.0, + "step": 34053 + }, + { + "epoch": 2.8081141255050714, + "grad_norm": 1.9628134965896606, + "learning_rate": 1e-06, + "loss": 0.6193, + "mean_token_accuracy": 0.8018233776092529, + "num_tokens": 913565592.0, + "step": 34054 + }, + { + "epoch": 2.808196586130123, + "grad_norm": 1.9163521528244019, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7668290734291077, + "num_tokens": 913600023.0, + "step": 34055 + }, + { + "epoch": 2.8082790467551746, + "grad_norm": 1.9263910055160522, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.7949840426445007, + "num_tokens": 913631334.0, + "step": 34056 + }, + { + "epoch": 2.808361507380226, + "grad_norm": 2.2547240257263184, + "learning_rate": 1e-06, + "loss": 0.5028, + "mean_token_accuracy": 0.8213158845901489, + "num_tokens": 913655488.0, + "step": 34057 + }, + { + "epoch": 2.8084439680052773, + "grad_norm": 1.985311508178711, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7371731996536255, + "num_tokens": 913689586.0, + "step": 34058 + }, + { + "epoch": 2.808526428630329, + "grad_norm": 2.7195119857788086, + "learning_rate": 1e-06, + "loss": 0.7353, + "mean_token_accuracy": 0.7573353052139282, + "num_tokens": 913709828.0, + "step": 34059 + }, + { + "epoch": 2.8086088892553804, + "grad_norm": 2.2624189853668213, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.7468434572219849, + "num_tokens": 913736035.0, + "step": 34060 + }, + { + "epoch": 2.8086913498804322, + "grad_norm": 2.359550714492798, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7742074131965637, + "num_tokens": 913759200.0, + "step": 34061 + }, + { + "epoch": 2.8087738105054836, + "grad_norm": 2.411593198776245, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7716724872589111, + "num_tokens": 913785807.0, + "step": 34062 + }, + { + "epoch": 2.8088562711305354, + "grad_norm": 2.006657838821411, + "learning_rate": 1e-06, + "loss": 0.6037, + "mean_token_accuracy": 0.8055394291877747, + "num_tokens": 913817235.0, + "step": 34063 + }, + { + "epoch": 2.8089387317555867, + "grad_norm": 2.32462215423584, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7499243021011353, + "num_tokens": 913843672.0, + "step": 34064 + }, + { + "epoch": 2.809021192380638, + "grad_norm": 2.0236401557922363, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7722872495651245, + "num_tokens": 913876761.0, + "step": 34065 + }, + { + "epoch": 2.80910365300569, + "grad_norm": 2.3670074939727783, + "learning_rate": 1e-06, + "loss": 0.7238, + "mean_token_accuracy": 0.7618106007575989, + "num_tokens": 913902421.0, + "step": 34066 + }, + { + "epoch": 2.809186113630741, + "grad_norm": 2.210806369781494, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.767339825630188, + "num_tokens": 913931637.0, + "step": 34067 + }, + { + "epoch": 2.809268574255793, + "grad_norm": 2.1597161293029785, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7793257236480713, + "num_tokens": 913959314.0, + "step": 34068 + }, + { + "epoch": 2.8093510348808444, + "grad_norm": 2.058885097503662, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7741138339042664, + "num_tokens": 913989211.0, + "step": 34069 + }, + { + "epoch": 2.809433495505896, + "grad_norm": 1.8085997104644775, + "learning_rate": 1e-06, + "loss": 0.6044, + "mean_token_accuracy": 0.7988132238388062, + "num_tokens": 914024981.0, + "step": 34070 + }, + { + "epoch": 2.8095159561309475, + "grad_norm": 2.173431158065796, + "learning_rate": 1e-06, + "loss": 0.6416, + "mean_token_accuracy": 0.7856516242027283, + "num_tokens": 914052603.0, + "step": 34071 + }, + { + "epoch": 2.809598416755999, + "grad_norm": 2.0679378509521484, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.7977745532989502, + "num_tokens": 914081949.0, + "step": 34072 + }, + { + "epoch": 2.8096808773810507, + "grad_norm": 2.2674801349639893, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7663183212280273, + "num_tokens": 914108131.0, + "step": 34073 + }, + { + "epoch": 2.809763338006102, + "grad_norm": 2.108677625656128, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7400037050247192, + "num_tokens": 914139070.0, + "step": 34074 + }, + { + "epoch": 2.8098457986311534, + "grad_norm": 2.1457419395446777, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7609466314315796, + "num_tokens": 914169790.0, + "step": 34075 + }, + { + "epoch": 2.809928259256205, + "grad_norm": 1.917445182800293, + "learning_rate": 1e-06, + "loss": 0.6606, + "mean_token_accuracy": 0.7833566665649414, + "num_tokens": 914203154.0, + "step": 34076 + }, + { + "epoch": 2.810010719881257, + "grad_norm": 2.1103837490081787, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7532747387886047, + "num_tokens": 914232804.0, + "step": 34077 + }, + { + "epoch": 2.8100931805063083, + "grad_norm": 2.5760204792022705, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.7457094192504883, + "num_tokens": 914254778.0, + "step": 34078 + }, + { + "epoch": 2.8101756411313596, + "grad_norm": 2.4453346729278564, + "learning_rate": 1e-06, + "loss": 0.6438, + "mean_token_accuracy": 0.7924376726150513, + "num_tokens": 914277544.0, + "step": 34079 + }, + { + "epoch": 2.8102581017564114, + "grad_norm": 1.870949387550354, + "learning_rate": 1e-06, + "loss": 0.6957, + "mean_token_accuracy": 0.7722591161727905, + "num_tokens": 914313744.0, + "step": 34080 + }, + { + "epoch": 2.810340562381463, + "grad_norm": 2.194453001022339, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7675716876983643, + "num_tokens": 914342580.0, + "step": 34081 + }, + { + "epoch": 2.810423023006514, + "grad_norm": 1.871503472328186, + "learning_rate": 1e-06, + "loss": 0.5748, + "mean_token_accuracy": 0.8019661903381348, + "num_tokens": 914377179.0, + "step": 34082 + }, + { + "epoch": 2.810505483631566, + "grad_norm": 2.140756607055664, + "learning_rate": 1e-06, + "loss": 0.5553, + "mean_token_accuracy": 0.8045831918716431, + "num_tokens": 914403466.0, + "step": 34083 + }, + { + "epoch": 2.8105879442566177, + "grad_norm": 2.2286665439605713, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7540670037269592, + "num_tokens": 914428494.0, + "step": 34084 + }, + { + "epoch": 2.810670404881669, + "grad_norm": 2.309962272644043, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7715014815330505, + "num_tokens": 914453833.0, + "step": 34085 + }, + { + "epoch": 2.8107528655067204, + "grad_norm": 2.164933919906616, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7794251441955566, + "num_tokens": 914482256.0, + "step": 34086 + }, + { + "epoch": 2.8108353261317722, + "grad_norm": 2.5962326526641846, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7571687698364258, + "num_tokens": 914503664.0, + "step": 34087 + }, + { + "epoch": 2.8109177867568236, + "grad_norm": 2.1521544456481934, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7311567068099976, + "num_tokens": 914533228.0, + "step": 34088 + }, + { + "epoch": 2.811000247381875, + "grad_norm": 2.542693614959717, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7740651369094849, + "num_tokens": 914554042.0, + "step": 34089 + }, + { + "epoch": 2.8110827080069267, + "grad_norm": 2.1474239826202393, + "learning_rate": 1e-06, + "loss": 0.5961, + "mean_token_accuracy": 0.8065105676651001, + "num_tokens": 914580559.0, + "step": 34090 + }, + { + "epoch": 2.811165168631978, + "grad_norm": 2.2989227771759033, + "learning_rate": 1e-06, + "loss": 0.6121, + "mean_token_accuracy": 0.7970364093780518, + "num_tokens": 914604883.0, + "step": 34091 + }, + { + "epoch": 2.81124762925703, + "grad_norm": 1.990267038345337, + "learning_rate": 1e-06, + "loss": 0.6747, + "mean_token_accuracy": 0.7840849161148071, + "num_tokens": 914637937.0, + "step": 34092 + }, + { + "epoch": 2.811330089882081, + "grad_norm": 2.238429069519043, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7413806915283203, + "num_tokens": 914667076.0, + "step": 34093 + }, + { + "epoch": 2.811412550507133, + "grad_norm": 1.943282961845398, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7506048083305359, + "num_tokens": 914704317.0, + "step": 34094 + }, + { + "epoch": 2.8114950111321844, + "grad_norm": 1.8986468315124512, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7770960330963135, + "num_tokens": 914739290.0, + "step": 34095 + }, + { + "epoch": 2.8115774717572357, + "grad_norm": 2.141781806945801, + "learning_rate": 1e-06, + "loss": 0.5907, + "mean_token_accuracy": 0.8017395734786987, + "num_tokens": 914766354.0, + "step": 34096 + }, + { + "epoch": 2.8116599323822875, + "grad_norm": 2.1712610721588135, + "learning_rate": 1e-06, + "loss": 0.6719, + "mean_token_accuracy": 0.7796107530593872, + "num_tokens": 914795567.0, + "step": 34097 + }, + { + "epoch": 2.811742393007339, + "grad_norm": 2.191265344619751, + "learning_rate": 1e-06, + "loss": 0.7513, + "mean_token_accuracy": 0.7640615701675415, + "num_tokens": 914825143.0, + "step": 34098 + }, + { + "epoch": 2.8118248536323907, + "grad_norm": 2.1310341358184814, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.7948599457740784, + "num_tokens": 914853508.0, + "step": 34099 + }, + { + "epoch": 2.811907314257442, + "grad_norm": 2.2840960025787354, + "learning_rate": 1e-06, + "loss": 0.5112, + "mean_token_accuracy": 0.8239707946777344, + "num_tokens": 914876202.0, + "step": 34100 + }, + { + "epoch": 2.811989774882494, + "grad_norm": 1.9462641477584839, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.7906829714775085, + "num_tokens": 914909518.0, + "step": 34101 + }, + { + "epoch": 2.812072235507545, + "grad_norm": 1.974491000175476, + "learning_rate": 1e-06, + "loss": 0.6713, + "mean_token_accuracy": 0.7837911248207092, + "num_tokens": 914942879.0, + "step": 34102 + }, + { + "epoch": 2.8121546961325965, + "grad_norm": 2.3071787357330322, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.7872891426086426, + "num_tokens": 914967293.0, + "step": 34103 + }, + { + "epoch": 2.8122371567576483, + "grad_norm": 2.1386849880218506, + "learning_rate": 1e-06, + "loss": 0.6851, + "mean_token_accuracy": 0.7713595628738403, + "num_tokens": 914995277.0, + "step": 34104 + }, + { + "epoch": 2.8123196173826996, + "grad_norm": 2.2257258892059326, + "learning_rate": 1e-06, + "loss": 0.7072, + "mean_token_accuracy": 0.7668493986129761, + "num_tokens": 915024045.0, + "step": 34105 + }, + { + "epoch": 2.8124020780077514, + "grad_norm": 1.9983102083206177, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7378120422363281, + "num_tokens": 915060307.0, + "step": 34106 + }, + { + "epoch": 2.812484538632803, + "grad_norm": 2.080528974533081, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7760659456253052, + "num_tokens": 915090542.0, + "step": 34107 + }, + { + "epoch": 2.8125669992578546, + "grad_norm": 1.7466108798980713, + "learning_rate": 1e-06, + "loss": 0.605, + "mean_token_accuracy": 0.8055901527404785, + "num_tokens": 915132522.0, + "step": 34108 + }, + { + "epoch": 2.812649459882906, + "grad_norm": 2.4630415439605713, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.7877578139305115, + "num_tokens": 915153596.0, + "step": 34109 + }, + { + "epoch": 2.8127319205079573, + "grad_norm": 2.243086814880371, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7545667886734009, + "num_tokens": 915180368.0, + "step": 34110 + }, + { + "epoch": 2.812814381133009, + "grad_norm": 2.0146806240081787, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7803236246109009, + "num_tokens": 915213488.0, + "step": 34111 + }, + { + "epoch": 2.8128968417580604, + "grad_norm": 2.101473808288574, + "learning_rate": 1e-06, + "loss": 0.7153, + "mean_token_accuracy": 0.7721302509307861, + "num_tokens": 915242836.0, + "step": 34112 + }, + { + "epoch": 2.8129793023831122, + "grad_norm": 2.2174203395843506, + "learning_rate": 1e-06, + "loss": 0.6003, + "mean_token_accuracy": 0.7922049164772034, + "num_tokens": 915267764.0, + "step": 34113 + }, + { + "epoch": 2.8130617630081636, + "grad_norm": 2.0797698497772217, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7593506574630737, + "num_tokens": 915300433.0, + "step": 34114 + }, + { + "epoch": 2.8131442236332154, + "grad_norm": 2.058999538421631, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.7899559140205383, + "num_tokens": 915329957.0, + "step": 34115 + }, + { + "epoch": 2.8132266842582667, + "grad_norm": 2.589097023010254, + "learning_rate": 1e-06, + "loss": 0.6291, + "mean_token_accuracy": 0.7873080968856812, + "num_tokens": 915356452.0, + "step": 34116 + }, + { + "epoch": 2.813309144883318, + "grad_norm": 2.112525224685669, + "learning_rate": 1e-06, + "loss": 0.6789, + "mean_token_accuracy": 0.7743626236915588, + "num_tokens": 915387408.0, + "step": 34117 + }, + { + "epoch": 2.81339160550837, + "grad_norm": 1.999563455581665, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7657082080841064, + "num_tokens": 915419719.0, + "step": 34118 + }, + { + "epoch": 2.813474066133421, + "grad_norm": 2.093132495880127, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7192875742912292, + "num_tokens": 915451197.0, + "step": 34119 + }, + { + "epoch": 2.8135565267584726, + "grad_norm": 2.0944955348968506, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7711416482925415, + "num_tokens": 915478957.0, + "step": 34120 + }, + { + "epoch": 2.8136389873835244, + "grad_norm": 2.0283303260803223, + "learning_rate": 1e-06, + "loss": 0.7163, + "mean_token_accuracy": 0.7598459720611572, + "num_tokens": 915510169.0, + "step": 34121 + }, + { + "epoch": 2.813721448008576, + "grad_norm": 2.0046451091766357, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7745743989944458, + "num_tokens": 915544137.0, + "step": 34122 + }, + { + "epoch": 2.8138039086336275, + "grad_norm": 2.457613229751587, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7615430355072021, + "num_tokens": 915566688.0, + "step": 34123 + }, + { + "epoch": 2.813886369258679, + "grad_norm": 2.105959892272949, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7420228719711304, + "num_tokens": 915597240.0, + "step": 34124 + }, + { + "epoch": 2.8139688298837306, + "grad_norm": 2.3626465797424316, + "learning_rate": 1e-06, + "loss": 0.663, + "mean_token_accuracy": 0.779144287109375, + "num_tokens": 915623540.0, + "step": 34125 + }, + { + "epoch": 2.814051290508782, + "grad_norm": 2.0670886039733887, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7737693786621094, + "num_tokens": 915654684.0, + "step": 34126 + }, + { + "epoch": 2.8141337511338333, + "grad_norm": 2.016972303390503, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7838144302368164, + "num_tokens": 915686066.0, + "step": 34127 + }, + { + "epoch": 2.814216211758885, + "grad_norm": 2.4089226722717285, + "learning_rate": 1e-06, + "loss": 0.7424, + "mean_token_accuracy": 0.7650774717330933, + "num_tokens": 915713871.0, + "step": 34128 + }, + { + "epoch": 2.814298672383937, + "grad_norm": 2.239335775375366, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.7869402170181274, + "num_tokens": 915740494.0, + "step": 34129 + }, + { + "epoch": 2.8143811330089883, + "grad_norm": 2.0247280597686768, + "learning_rate": 1e-06, + "loss": 0.6802, + "mean_token_accuracy": 0.7866809964179993, + "num_tokens": 915774742.0, + "step": 34130 + }, + { + "epoch": 2.8144635936340396, + "grad_norm": 1.9514738321304321, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7467514872550964, + "num_tokens": 915809807.0, + "step": 34131 + }, + { + "epoch": 2.8145460542590914, + "grad_norm": 1.941120982170105, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7748720645904541, + "num_tokens": 915845529.0, + "step": 34132 + }, + { + "epoch": 2.814628514884143, + "grad_norm": 2.0951969623565674, + "learning_rate": 1e-06, + "loss": 0.7233, + "mean_token_accuracy": 0.7587871551513672, + "num_tokens": 915876302.0, + "step": 34133 + }, + { + "epoch": 2.814710975509194, + "grad_norm": 1.9543473720550537, + "learning_rate": 1e-06, + "loss": 0.6163, + "mean_token_accuracy": 0.7992757558822632, + "num_tokens": 915909430.0, + "step": 34134 + }, + { + "epoch": 2.814793436134246, + "grad_norm": 1.841912031173706, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7938904762268066, + "num_tokens": 915947232.0, + "step": 34135 + }, + { + "epoch": 2.8148758967592973, + "grad_norm": 2.0822322368621826, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7444320917129517, + "num_tokens": 915980447.0, + "step": 34136 + }, + { + "epoch": 2.814958357384349, + "grad_norm": 2.166111707687378, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.7986366748809814, + "num_tokens": 916007154.0, + "step": 34137 + }, + { + "epoch": 2.8150408180094004, + "grad_norm": 1.8544946908950806, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7793477773666382, + "num_tokens": 916043502.0, + "step": 34138 + }, + { + "epoch": 2.815123278634452, + "grad_norm": 2.1641151905059814, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7304770350456238, + "num_tokens": 916074599.0, + "step": 34139 + }, + { + "epoch": 2.8152057392595036, + "grad_norm": 2.218768835067749, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7785448431968689, + "num_tokens": 916101504.0, + "step": 34140 + }, + { + "epoch": 2.815288199884555, + "grad_norm": 2.24060320854187, + "learning_rate": 1e-06, + "loss": 0.5543, + "mean_token_accuracy": 0.8118212223052979, + "num_tokens": 916124812.0, + "step": 34141 + }, + { + "epoch": 2.8153706605096067, + "grad_norm": 2.396130084991455, + "learning_rate": 1e-06, + "loss": 0.6716, + "mean_token_accuracy": 0.7703810930252075, + "num_tokens": 916148924.0, + "step": 34142 + }, + { + "epoch": 2.815453121134658, + "grad_norm": 1.9973454475402832, + "learning_rate": 1e-06, + "loss": 0.5368, + "mean_token_accuracy": 0.8223011493682861, + "num_tokens": 916177477.0, + "step": 34143 + }, + { + "epoch": 2.81553558175971, + "grad_norm": 1.8654603958129883, + "learning_rate": 1e-06, + "loss": 0.5198, + "mean_token_accuracy": 0.8226034641265869, + "num_tokens": 916214181.0, + "step": 34144 + }, + { + "epoch": 2.815618042384761, + "grad_norm": 2.121847152709961, + "learning_rate": 1e-06, + "loss": 0.7402, + "mean_token_accuracy": 0.7684464454650879, + "num_tokens": 916242267.0, + "step": 34145 + }, + { + "epoch": 2.815700503009813, + "grad_norm": 2.1060359477996826, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.745493471622467, + "num_tokens": 916271282.0, + "step": 34146 + }, + { + "epoch": 2.8157829636348644, + "grad_norm": 2.1511266231536865, + "learning_rate": 1e-06, + "loss": 0.7733, + "mean_token_accuracy": 0.7509453296661377, + "num_tokens": 916300866.0, + "step": 34147 + }, + { + "epoch": 2.8158654242599157, + "grad_norm": 2.0075156688690186, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7768738269805908, + "num_tokens": 916332530.0, + "step": 34148 + }, + { + "epoch": 2.8159478848849675, + "grad_norm": 2.19740891456604, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7482709884643555, + "num_tokens": 916362287.0, + "step": 34149 + }, + { + "epoch": 2.816030345510019, + "grad_norm": 1.787731647491455, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.7877147197723389, + "num_tokens": 916398934.0, + "step": 34150 + }, + { + "epoch": 2.8161128061350706, + "grad_norm": 2.164216995239258, + "learning_rate": 1e-06, + "loss": 0.5843, + "mean_token_accuracy": 0.8113723397254944, + "num_tokens": 916425322.0, + "step": 34151 + }, + { + "epoch": 2.816195266760122, + "grad_norm": 2.0772879123687744, + "learning_rate": 1e-06, + "loss": 0.5958, + "mean_token_accuracy": 0.7933650612831116, + "num_tokens": 916454300.0, + "step": 34152 + }, + { + "epoch": 2.816277727385174, + "grad_norm": 1.9686870574951172, + "learning_rate": 1e-06, + "loss": 0.6539, + "mean_token_accuracy": 0.7865905165672302, + "num_tokens": 916485279.0, + "step": 34153 + }, + { + "epoch": 2.816360188010225, + "grad_norm": 1.8167153596878052, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7823881506919861, + "num_tokens": 916524032.0, + "step": 34154 + }, + { + "epoch": 2.8164426486352765, + "grad_norm": 2.0298852920532227, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7829744815826416, + "num_tokens": 916554075.0, + "step": 34155 + }, + { + "epoch": 2.8165251092603283, + "grad_norm": 1.9982008934020996, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7667219638824463, + "num_tokens": 916588321.0, + "step": 34156 + }, + { + "epoch": 2.8166075698853796, + "grad_norm": 2.1926002502441406, + "learning_rate": 1e-06, + "loss": 0.7121, + "mean_token_accuracy": 0.7722665667533875, + "num_tokens": 916619754.0, + "step": 34157 + }, + { + "epoch": 2.8166900305104314, + "grad_norm": 2.1209428310394287, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7703724503517151, + "num_tokens": 916650527.0, + "step": 34158 + }, + { + "epoch": 2.816772491135483, + "grad_norm": 2.2254135608673096, + "learning_rate": 1e-06, + "loss": 0.6193, + "mean_token_accuracy": 0.7879271507263184, + "num_tokens": 916677889.0, + "step": 34159 + }, + { + "epoch": 2.8168549517605346, + "grad_norm": 2.308971405029297, + "learning_rate": 1e-06, + "loss": 0.6565, + "mean_token_accuracy": 0.7872692942619324, + "num_tokens": 916703442.0, + "step": 34160 + }, + { + "epoch": 2.816937412385586, + "grad_norm": 2.2919256687164307, + "learning_rate": 1e-06, + "loss": 0.686, + "mean_token_accuracy": 0.759783923625946, + "num_tokens": 916729123.0, + "step": 34161 + }, + { + "epoch": 2.8170198730106373, + "grad_norm": 2.1140756607055664, + "learning_rate": 1e-06, + "loss": 0.7292, + "mean_token_accuracy": 0.7582719922065735, + "num_tokens": 916758217.0, + "step": 34162 + }, + { + "epoch": 2.817102333635689, + "grad_norm": 1.9854191541671753, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7724180221557617, + "num_tokens": 916792977.0, + "step": 34163 + }, + { + "epoch": 2.8171847942607404, + "grad_norm": 2.236241340637207, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7762285470962524, + "num_tokens": 916819206.0, + "step": 34164 + }, + { + "epoch": 2.8172672548857918, + "grad_norm": 1.8387176990509033, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.7851407527923584, + "num_tokens": 916858656.0, + "step": 34165 + }, + { + "epoch": 2.8173497155108436, + "grad_norm": 1.9513089656829834, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.7782241702079773, + "num_tokens": 916892271.0, + "step": 34166 + }, + { + "epoch": 2.8174321761358954, + "grad_norm": 2.318176507949829, + "learning_rate": 1e-06, + "loss": 0.6026, + "mean_token_accuracy": 0.7978429794311523, + "num_tokens": 916918484.0, + "step": 34167 + }, + { + "epoch": 2.8175146367609467, + "grad_norm": 2.0979082584381104, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7642343044281006, + "num_tokens": 916949290.0, + "step": 34168 + }, + { + "epoch": 2.817597097385998, + "grad_norm": 2.1237616539001465, + "learning_rate": 1e-06, + "loss": 0.7596, + "mean_token_accuracy": 0.7669100165367126, + "num_tokens": 916977329.0, + "step": 34169 + }, + { + "epoch": 2.81767955801105, + "grad_norm": 2.419518232345581, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7769896984100342, + "num_tokens": 917001947.0, + "step": 34170 + }, + { + "epoch": 2.817762018636101, + "grad_norm": 2.322842597961426, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7564438581466675, + "num_tokens": 917029201.0, + "step": 34171 + }, + { + "epoch": 2.8178444792611526, + "grad_norm": 2.112180709838867, + "learning_rate": 1e-06, + "loss": 0.7913, + "mean_token_accuracy": 0.7461947798728943, + "num_tokens": 917059466.0, + "step": 34172 + }, + { + "epoch": 2.8179269398862044, + "grad_norm": 2.2648890018463135, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.7340936064720154, + "num_tokens": 917087233.0, + "step": 34173 + }, + { + "epoch": 2.818009400511256, + "grad_norm": 2.016680955886841, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7476219534873962, + "num_tokens": 917119906.0, + "step": 34174 + }, + { + "epoch": 2.8180918611363075, + "grad_norm": 1.8550710678100586, + "learning_rate": 1e-06, + "loss": 0.6465, + "mean_token_accuracy": 0.7891597747802734, + "num_tokens": 917156091.0, + "step": 34175 + }, + { + "epoch": 2.818174321761359, + "grad_norm": 1.9901599884033203, + "learning_rate": 1e-06, + "loss": 0.6992, + "mean_token_accuracy": 0.7696113586425781, + "num_tokens": 917188661.0, + "step": 34176 + }, + { + "epoch": 2.8182567823864106, + "grad_norm": 2.1907999515533447, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.736460268497467, + "num_tokens": 917218745.0, + "step": 34177 + }, + { + "epoch": 2.818339243011462, + "grad_norm": 1.8049026727676392, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.7989175319671631, + "num_tokens": 917255749.0, + "step": 34178 + }, + { + "epoch": 2.8184217036365133, + "grad_norm": 2.49141526222229, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7218331694602966, + "num_tokens": 917278474.0, + "step": 34179 + }, + { + "epoch": 2.818504164261565, + "grad_norm": 2.1077215671539307, + "learning_rate": 1e-06, + "loss": 0.6618, + "mean_token_accuracy": 0.7883751392364502, + "num_tokens": 917313994.0, + "step": 34180 + }, + { + "epoch": 2.8185866248866165, + "grad_norm": 2.1025407314300537, + "learning_rate": 1e-06, + "loss": 0.5293, + "mean_token_accuracy": 0.8156416416168213, + "num_tokens": 917342256.0, + "step": 34181 + }, + { + "epoch": 2.8186690855116683, + "grad_norm": 2.10382080078125, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7627765536308289, + "num_tokens": 917370242.0, + "step": 34182 + }, + { + "epoch": 2.8187515461367196, + "grad_norm": 1.9395235776901245, + "learning_rate": 1e-06, + "loss": 0.5866, + "mean_token_accuracy": 0.8074016571044922, + "num_tokens": 917400163.0, + "step": 34183 + }, + { + "epoch": 2.8188340067617714, + "grad_norm": 2.1424453258514404, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7941817045211792, + "num_tokens": 917427830.0, + "step": 34184 + }, + { + "epoch": 2.8189164673868228, + "grad_norm": 2.0906214714050293, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.7971726655960083, + "num_tokens": 917457734.0, + "step": 34185 + }, + { + "epoch": 2.818998928011874, + "grad_norm": 2.1420040130615234, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7637102007865906, + "num_tokens": 917490083.0, + "step": 34186 + }, + { + "epoch": 2.819081388636926, + "grad_norm": 2.0818755626678467, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.738182008266449, + "num_tokens": 917520126.0, + "step": 34187 + }, + { + "epoch": 2.8191638492619773, + "grad_norm": 2.0498642921447754, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7448094487190247, + "num_tokens": 917551277.0, + "step": 34188 + }, + { + "epoch": 2.819246309887029, + "grad_norm": 2.153353214263916, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.755839467048645, + "num_tokens": 917580376.0, + "step": 34189 + }, + { + "epoch": 2.8193287705120804, + "grad_norm": 1.9688962697982788, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7660938501358032, + "num_tokens": 917613002.0, + "step": 34190 + }, + { + "epoch": 2.819411231137132, + "grad_norm": 2.4153311252593994, + "learning_rate": 1e-06, + "loss": 0.6452, + "mean_token_accuracy": 0.7854102849960327, + "num_tokens": 917636110.0, + "step": 34191 + }, + { + "epoch": 2.8194936917621836, + "grad_norm": 2.0060150623321533, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7800421118736267, + "num_tokens": 917668153.0, + "step": 34192 + }, + { + "epoch": 2.819576152387235, + "grad_norm": 2.0312280654907227, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7451068162918091, + "num_tokens": 917704255.0, + "step": 34193 + }, + { + "epoch": 2.8196586130122867, + "grad_norm": 2.349623918533325, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7827262878417969, + "num_tokens": 917728015.0, + "step": 34194 + }, + { + "epoch": 2.819741073637338, + "grad_norm": 2.1550633907318115, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.7923421859741211, + "num_tokens": 917757191.0, + "step": 34195 + }, + { + "epoch": 2.81982353426239, + "grad_norm": 1.96412992477417, + "learning_rate": 1e-06, + "loss": 0.7978, + "mean_token_accuracy": 0.7437095642089844, + "num_tokens": 917793156.0, + "step": 34196 + }, + { + "epoch": 2.819905994887441, + "grad_norm": 2.3944075107574463, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7279472947120667, + "num_tokens": 917818426.0, + "step": 34197 + }, + { + "epoch": 2.819988455512493, + "grad_norm": 2.0817513465881348, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7611187696456909, + "num_tokens": 917848207.0, + "step": 34198 + }, + { + "epoch": 2.8200709161375443, + "grad_norm": 2.111809253692627, + "learning_rate": 1e-06, + "loss": 0.7265, + "mean_token_accuracy": 0.7674263715744019, + "num_tokens": 917877204.0, + "step": 34199 + }, + { + "epoch": 2.8201533767625957, + "grad_norm": 2.1606698036193848, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.7835143804550171, + "num_tokens": 917903260.0, + "step": 34200 + }, + { + "epoch": 2.8202358373876475, + "grad_norm": 1.9413557052612305, + "learning_rate": 1e-06, + "loss": 0.669, + "mean_token_accuracy": 0.7814815044403076, + "num_tokens": 917935887.0, + "step": 34201 + }, + { + "epoch": 2.820318298012699, + "grad_norm": 2.039254903793335, + "learning_rate": 1e-06, + "loss": 0.6156, + "mean_token_accuracy": 0.7904098629951477, + "num_tokens": 917968407.0, + "step": 34202 + }, + { + "epoch": 2.8204007586377506, + "grad_norm": 2.0831940174102783, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.8041262626647949, + "num_tokens": 917996544.0, + "step": 34203 + }, + { + "epoch": 2.820483219262802, + "grad_norm": 2.259326934814453, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7006752490997314, + "num_tokens": 918026061.0, + "step": 34204 + }, + { + "epoch": 2.820565679887854, + "grad_norm": 2.03255033493042, + "learning_rate": 1e-06, + "loss": 0.5642, + "mean_token_accuracy": 0.8038026690483093, + "num_tokens": 918057541.0, + "step": 34205 + }, + { + "epoch": 2.820648140512905, + "grad_norm": 1.8575191497802734, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7464138865470886, + "num_tokens": 918096782.0, + "step": 34206 + }, + { + "epoch": 2.8207306011379565, + "grad_norm": 2.000697135925293, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7742248773574829, + "num_tokens": 918127728.0, + "step": 34207 + }, + { + "epoch": 2.8208130617630083, + "grad_norm": 2.0778398513793945, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7504892349243164, + "num_tokens": 918158566.0, + "step": 34208 + }, + { + "epoch": 2.8208955223880596, + "grad_norm": 1.968052625656128, + "learning_rate": 1e-06, + "loss": 0.5538, + "mean_token_accuracy": 0.8180462718009949, + "num_tokens": 918189468.0, + "step": 34209 + }, + { + "epoch": 2.820977983013111, + "grad_norm": 2.795776605606079, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.7585280537605286, + "num_tokens": 918208910.0, + "step": 34210 + }, + { + "epoch": 2.8210604436381628, + "grad_norm": 2.0891854763031006, + "learning_rate": 1e-06, + "loss": 0.6648, + "mean_token_accuracy": 0.7896387577056885, + "num_tokens": 918241038.0, + "step": 34211 + }, + { + "epoch": 2.8211429042632146, + "grad_norm": 2.188812732696533, + "learning_rate": 1e-06, + "loss": 0.7263, + "mean_token_accuracy": 0.7537577152252197, + "num_tokens": 918271344.0, + "step": 34212 + }, + { + "epoch": 2.821225364888266, + "grad_norm": 2.197103261947632, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.744799017906189, + "num_tokens": 918299468.0, + "step": 34213 + }, + { + "epoch": 2.8213078255133173, + "grad_norm": 2.007732391357422, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7659133672714233, + "num_tokens": 918330202.0, + "step": 34214 + }, + { + "epoch": 2.821390286138369, + "grad_norm": 2.075514554977417, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.727170467376709, + "num_tokens": 918364494.0, + "step": 34215 + }, + { + "epoch": 2.8214727467634204, + "grad_norm": 1.8664249181747437, + "learning_rate": 1e-06, + "loss": 0.6505, + "mean_token_accuracy": 0.7874115109443665, + "num_tokens": 918399630.0, + "step": 34216 + }, + { + "epoch": 2.8215552073884718, + "grad_norm": 2.1042087078094482, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.7995928525924683, + "num_tokens": 918429810.0, + "step": 34217 + }, + { + "epoch": 2.8216376680135236, + "grad_norm": 2.1355981826782227, + "learning_rate": 1e-06, + "loss": 0.5943, + "mean_token_accuracy": 0.8005298972129822, + "num_tokens": 918457381.0, + "step": 34218 + }, + { + "epoch": 2.8217201286385754, + "grad_norm": 2.2810404300689697, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7547651529312134, + "num_tokens": 918486217.0, + "step": 34219 + }, + { + "epoch": 2.8218025892636267, + "grad_norm": 2.133788824081421, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7417171001434326, + "num_tokens": 918515128.0, + "step": 34220 + }, + { + "epoch": 2.821885049888678, + "grad_norm": 2.386861801147461, + "learning_rate": 1e-06, + "loss": 0.5417, + "mean_token_accuracy": 0.8124123811721802, + "num_tokens": 918535907.0, + "step": 34221 + }, + { + "epoch": 2.82196751051373, + "grad_norm": 2.255758762359619, + "learning_rate": 1e-06, + "loss": 0.661, + "mean_token_accuracy": 0.7763369083404541, + "num_tokens": 918562432.0, + "step": 34222 + }, + { + "epoch": 2.822049971138781, + "grad_norm": 1.8851889371871948, + "learning_rate": 1e-06, + "loss": 0.5779, + "mean_token_accuracy": 0.8018202781677246, + "num_tokens": 918599727.0, + "step": 34223 + }, + { + "epoch": 2.8221324317638325, + "grad_norm": 2.136685371398926, + "learning_rate": 1e-06, + "loss": 0.7207, + "mean_token_accuracy": 0.7576730251312256, + "num_tokens": 918629124.0, + "step": 34224 + }, + { + "epoch": 2.8222148923888843, + "grad_norm": 1.9376788139343262, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.8027658462524414, + "num_tokens": 918660198.0, + "step": 34225 + }, + { + "epoch": 2.8222973530139357, + "grad_norm": 1.9970229864120483, + "learning_rate": 1e-06, + "loss": 0.7899, + "mean_token_accuracy": 0.7475634813308716, + "num_tokens": 918693460.0, + "step": 34226 + }, + { + "epoch": 2.8223798136389875, + "grad_norm": 2.0446817874908447, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7338898777961731, + "num_tokens": 918729449.0, + "step": 34227 + }, + { + "epoch": 2.822462274264039, + "grad_norm": 2.082411527633667, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7651083469390869, + "num_tokens": 918765234.0, + "step": 34228 + }, + { + "epoch": 2.8225447348890906, + "grad_norm": 2.1585681438446045, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7563614845275879, + "num_tokens": 918794737.0, + "step": 34229 + }, + { + "epoch": 2.822627195514142, + "grad_norm": 2.1156537532806396, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7668986320495605, + "num_tokens": 918823475.0, + "step": 34230 + }, + { + "epoch": 2.8227096561391933, + "grad_norm": 2.3096883296966553, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.748763382434845, + "num_tokens": 918852665.0, + "step": 34231 + }, + { + "epoch": 2.822792116764245, + "grad_norm": 1.9300678968429565, + "learning_rate": 1e-06, + "loss": 0.6194, + "mean_token_accuracy": 0.8028818368911743, + "num_tokens": 918888569.0, + "step": 34232 + }, + { + "epoch": 2.8228745773892965, + "grad_norm": 1.9165831804275513, + "learning_rate": 1e-06, + "loss": 0.5722, + "mean_token_accuracy": 0.8123726844787598, + "num_tokens": 918920249.0, + "step": 34233 + }, + { + "epoch": 2.8229570380143483, + "grad_norm": 2.174961805343628, + "learning_rate": 1e-06, + "loss": 0.7348, + "mean_token_accuracy": 0.7617889642715454, + "num_tokens": 918949321.0, + "step": 34234 + }, + { + "epoch": 2.8230394986393996, + "grad_norm": 1.9703948497772217, + "learning_rate": 1e-06, + "loss": 0.6377, + "mean_token_accuracy": 0.7926421165466309, + "num_tokens": 918980311.0, + "step": 34235 + }, + { + "epoch": 2.8231219592644514, + "grad_norm": 1.9514522552490234, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7653523683547974, + "num_tokens": 919014046.0, + "step": 34236 + }, + { + "epoch": 2.8232044198895028, + "grad_norm": 2.3451695442199707, + "learning_rate": 1e-06, + "loss": 0.6434, + "mean_token_accuracy": 0.788795530796051, + "num_tokens": 919037497.0, + "step": 34237 + }, + { + "epoch": 2.823286880514554, + "grad_norm": 2.0927693843841553, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7818125486373901, + "num_tokens": 919067469.0, + "step": 34238 + }, + { + "epoch": 2.823369341139606, + "grad_norm": 2.2577404975891113, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7638319134712219, + "num_tokens": 919096657.0, + "step": 34239 + }, + { + "epoch": 2.8234518017646573, + "grad_norm": 2.274371862411499, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7364722490310669, + "num_tokens": 919124439.0, + "step": 34240 + }, + { + "epoch": 2.823534262389709, + "grad_norm": 2.0787577629089355, + "learning_rate": 1e-06, + "loss": 0.792, + "mean_token_accuracy": 0.7506097555160522, + "num_tokens": 919156551.0, + "step": 34241 + }, + { + "epoch": 2.8236167230147604, + "grad_norm": 2.110863447189331, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7427494525909424, + "num_tokens": 919185255.0, + "step": 34242 + }, + { + "epoch": 2.823699183639812, + "grad_norm": 2.127434015274048, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.7814740538597107, + "num_tokens": 919211400.0, + "step": 34243 + }, + { + "epoch": 2.8237816442648636, + "grad_norm": 2.209378242492676, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7539955973625183, + "num_tokens": 919241866.0, + "step": 34244 + }, + { + "epoch": 2.823864104889915, + "grad_norm": 2.1223087310791016, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7319017648696899, + "num_tokens": 919272161.0, + "step": 34245 + }, + { + "epoch": 2.8239465655149667, + "grad_norm": 2.2917678356170654, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7641696333885193, + "num_tokens": 919297447.0, + "step": 34246 + }, + { + "epoch": 2.824029026140018, + "grad_norm": 2.05639386177063, + "learning_rate": 1e-06, + "loss": 0.5897, + "mean_token_accuracy": 0.8164281845092773, + "num_tokens": 919328734.0, + "step": 34247 + }, + { + "epoch": 2.82411148676507, + "grad_norm": 2.1584229469299316, + "learning_rate": 1e-06, + "loss": 0.5758, + "mean_token_accuracy": 0.8051355481147766, + "num_tokens": 919354126.0, + "step": 34248 + }, + { + "epoch": 2.824193947390121, + "grad_norm": 2.190859317779541, + "learning_rate": 1e-06, + "loss": 0.7433, + "mean_token_accuracy": 0.760672390460968, + "num_tokens": 919382843.0, + "step": 34249 + }, + { + "epoch": 2.824276408015173, + "grad_norm": 2.1379196643829346, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7830688953399658, + "num_tokens": 919411076.0, + "step": 34250 + }, + { + "epoch": 2.8243588686402243, + "grad_norm": 2.260883331298828, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7706162929534912, + "num_tokens": 919438279.0, + "step": 34251 + }, + { + "epoch": 2.8244413292652757, + "grad_norm": 2.6243350505828857, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.750940203666687, + "num_tokens": 919458871.0, + "step": 34252 + }, + { + "epoch": 2.8245237898903275, + "grad_norm": 2.1310231685638428, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7645611763000488, + "num_tokens": 919488400.0, + "step": 34253 + }, + { + "epoch": 2.824606250515379, + "grad_norm": 2.1720774173736572, + "learning_rate": 1e-06, + "loss": 0.6139, + "mean_token_accuracy": 0.7956454753875732, + "num_tokens": 919517437.0, + "step": 34254 + }, + { + "epoch": 2.82468871114043, + "grad_norm": 2.66715145111084, + "learning_rate": 1e-06, + "loss": 0.742, + "mean_token_accuracy": 0.755832314491272, + "num_tokens": 919541749.0, + "step": 34255 + }, + { + "epoch": 2.824771171765482, + "grad_norm": 2.250126361846924, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7430380582809448, + "num_tokens": 919570847.0, + "step": 34256 + }, + { + "epoch": 2.8248536323905338, + "grad_norm": 2.2967045307159424, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7679746150970459, + "num_tokens": 919598632.0, + "step": 34257 + }, + { + "epoch": 2.824936093015585, + "grad_norm": 2.3647079467773438, + "learning_rate": 1e-06, + "loss": 0.7051, + "mean_token_accuracy": 0.7758951187133789, + "num_tokens": 919623222.0, + "step": 34258 + }, + { + "epoch": 2.8250185536406365, + "grad_norm": 2.0398411750793457, + "learning_rate": 1e-06, + "loss": 0.6995, + "mean_token_accuracy": 0.7641286849975586, + "num_tokens": 919654301.0, + "step": 34259 + }, + { + "epoch": 2.8251010142656883, + "grad_norm": 2.2259163856506348, + "learning_rate": 1e-06, + "loss": 0.693, + "mean_token_accuracy": 0.772456705570221, + "num_tokens": 919680641.0, + "step": 34260 + }, + { + "epoch": 2.8251834748907396, + "grad_norm": 1.9134706258773804, + "learning_rate": 1e-06, + "loss": 0.5413, + "mean_token_accuracy": 0.8109864592552185, + "num_tokens": 919713930.0, + "step": 34261 + }, + { + "epoch": 2.825265935515791, + "grad_norm": 2.4273529052734375, + "learning_rate": 1e-06, + "loss": 0.7054, + "mean_token_accuracy": 0.7757565975189209, + "num_tokens": 919736777.0, + "step": 34262 + }, + { + "epoch": 2.8253483961408428, + "grad_norm": 2.2703001499176025, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7786130905151367, + "num_tokens": 919763794.0, + "step": 34263 + }, + { + "epoch": 2.8254308567658946, + "grad_norm": 2.361588716506958, + "learning_rate": 1e-06, + "loss": 0.8078, + "mean_token_accuracy": 0.7451795935630798, + "num_tokens": 919789906.0, + "step": 34264 + }, + { + "epoch": 2.825513317390946, + "grad_norm": 2.2053990364074707, + "learning_rate": 1e-06, + "loss": 0.7457, + "mean_token_accuracy": 0.7582513689994812, + "num_tokens": 919816972.0, + "step": 34265 + }, + { + "epoch": 2.8255957780159973, + "grad_norm": 2.152879476547241, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7596793174743652, + "num_tokens": 919847757.0, + "step": 34266 + }, + { + "epoch": 2.825678238641049, + "grad_norm": 2.2068700790405273, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.7878111600875854, + "num_tokens": 919872921.0, + "step": 34267 + }, + { + "epoch": 2.8257606992661004, + "grad_norm": 2.1621298789978027, + "learning_rate": 1e-06, + "loss": 0.5389, + "mean_token_accuracy": 0.8087055683135986, + "num_tokens": 919901321.0, + "step": 34268 + }, + { + "epoch": 2.8258431598911518, + "grad_norm": 2.061079978942871, + "learning_rate": 1e-06, + "loss": 0.6546, + "mean_token_accuracy": 0.7818636298179626, + "num_tokens": 919929519.0, + "step": 34269 + }, + { + "epoch": 2.8259256205162036, + "grad_norm": 2.960775136947632, + "learning_rate": 1e-06, + "loss": 0.7257, + "mean_token_accuracy": 0.7612799406051636, + "num_tokens": 919948039.0, + "step": 34270 + }, + { + "epoch": 2.826008081141255, + "grad_norm": 2.1487183570861816, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7420139312744141, + "num_tokens": 919977026.0, + "step": 34271 + }, + { + "epoch": 2.8260905417663067, + "grad_norm": 2.1228063106536865, + "learning_rate": 1e-06, + "loss": 0.7841, + "mean_token_accuracy": 0.7479956150054932, + "num_tokens": 920008533.0, + "step": 34272 + }, + { + "epoch": 2.826173002391358, + "grad_norm": 2.1464719772338867, + "learning_rate": 1e-06, + "loss": 0.6439, + "mean_token_accuracy": 0.7890915870666504, + "num_tokens": 920036050.0, + "step": 34273 + }, + { + "epoch": 2.82625546301641, + "grad_norm": 2.2247061729431152, + "learning_rate": 1e-06, + "loss": 0.5966, + "mean_token_accuracy": 0.80156409740448, + "num_tokens": 920061571.0, + "step": 34274 + }, + { + "epoch": 2.826337923641461, + "grad_norm": 1.8634482622146606, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7680361866950989, + "num_tokens": 920098192.0, + "step": 34275 + }, + { + "epoch": 2.8264203842665125, + "grad_norm": 2.148876428604126, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7745022177696228, + "num_tokens": 920125833.0, + "step": 34276 + }, + { + "epoch": 2.8265028448915643, + "grad_norm": 2.283172369003296, + "learning_rate": 1e-06, + "loss": 0.642, + "mean_token_accuracy": 0.7775241136550903, + "num_tokens": 920154114.0, + "step": 34277 + }, + { + "epoch": 2.8265853055166157, + "grad_norm": 2.122415065765381, + "learning_rate": 1e-06, + "loss": 0.7167, + "mean_token_accuracy": 0.7667232155799866, + "num_tokens": 920187456.0, + "step": 34278 + }, + { + "epoch": 2.8266677661416675, + "grad_norm": 2.3860208988189697, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7681594491004944, + "num_tokens": 920211395.0, + "step": 34279 + }, + { + "epoch": 2.826750226766719, + "grad_norm": 2.0610475540161133, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7183400392532349, + "num_tokens": 920244682.0, + "step": 34280 + }, + { + "epoch": 2.8268326873917706, + "grad_norm": 2.4510128498077393, + "learning_rate": 1e-06, + "loss": 0.4992, + "mean_token_accuracy": 0.8256646394729614, + "num_tokens": 920267777.0, + "step": 34281 + }, + { + "epoch": 2.826915148016822, + "grad_norm": 2.0976922512054443, + "learning_rate": 1e-06, + "loss": 0.6519, + "mean_token_accuracy": 0.7778783440589905, + "num_tokens": 920296668.0, + "step": 34282 + }, + { + "epoch": 2.8269976086418733, + "grad_norm": 2.615842819213867, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7432472109794617, + "num_tokens": 920322643.0, + "step": 34283 + }, + { + "epoch": 2.827080069266925, + "grad_norm": 2.142274856567383, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.7690391540527344, + "num_tokens": 920351491.0, + "step": 34284 + }, + { + "epoch": 2.8271625298919765, + "grad_norm": 2.029344081878662, + "learning_rate": 1e-06, + "loss": 0.5979, + "mean_token_accuracy": 0.8022679090499878, + "num_tokens": 920382497.0, + "step": 34285 + }, + { + "epoch": 2.8272449905170283, + "grad_norm": 2.0178425312042236, + "learning_rate": 1e-06, + "loss": 0.681, + "mean_token_accuracy": 0.7750053405761719, + "num_tokens": 920416143.0, + "step": 34286 + }, + { + "epoch": 2.8273274511420796, + "grad_norm": 2.4348886013031006, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7420212626457214, + "num_tokens": 920442654.0, + "step": 34287 + }, + { + "epoch": 2.8274099117671314, + "grad_norm": 2.1287131309509277, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.8034155964851379, + "num_tokens": 920470029.0, + "step": 34288 + }, + { + "epoch": 2.8274923723921828, + "grad_norm": 1.8847297430038452, + "learning_rate": 1e-06, + "loss": 0.6728, + "mean_token_accuracy": 0.7766267657279968, + "num_tokens": 920505933.0, + "step": 34289 + }, + { + "epoch": 2.827574833017234, + "grad_norm": 1.9067184925079346, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.776032030582428, + "num_tokens": 920538252.0, + "step": 34290 + }, + { + "epoch": 2.827657293642286, + "grad_norm": 2.1614980697631836, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7612730264663696, + "num_tokens": 920567793.0, + "step": 34291 + }, + { + "epoch": 2.8277397542673373, + "grad_norm": 2.3301777839660645, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.766794741153717, + "num_tokens": 920596200.0, + "step": 34292 + }, + { + "epoch": 2.8278222148923886, + "grad_norm": 2.2035796642303467, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7191317081451416, + "num_tokens": 920624483.0, + "step": 34293 + }, + { + "epoch": 2.8279046755174404, + "grad_norm": 2.5204904079437256, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.7658523321151733, + "num_tokens": 920646834.0, + "step": 34294 + }, + { + "epoch": 2.827987136142492, + "grad_norm": 1.8264213800430298, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7860761880874634, + "num_tokens": 920684196.0, + "step": 34295 + }, + { + "epoch": 2.8280695967675435, + "grad_norm": 2.3551599979400635, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7366104125976562, + "num_tokens": 920711959.0, + "step": 34296 + }, + { + "epoch": 2.828152057392595, + "grad_norm": 2.078537702560425, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.7477888464927673, + "num_tokens": 920743170.0, + "step": 34297 + }, + { + "epoch": 2.8282345180176467, + "grad_norm": 2.1918084621429443, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7365667819976807, + "num_tokens": 920770529.0, + "step": 34298 + }, + { + "epoch": 2.828316978642698, + "grad_norm": 2.0036423206329346, + "learning_rate": 1e-06, + "loss": 0.5461, + "mean_token_accuracy": 0.8148983716964722, + "num_tokens": 920801024.0, + "step": 34299 + }, + { + "epoch": 2.8283994392677494, + "grad_norm": 1.9807546138763428, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7618206739425659, + "num_tokens": 920835851.0, + "step": 34300 + }, + { + "epoch": 2.828481899892801, + "grad_norm": 2.087367057800293, + "learning_rate": 1e-06, + "loss": 0.734, + "mean_token_accuracy": 0.7588177919387817, + "num_tokens": 920865594.0, + "step": 34301 + }, + { + "epoch": 2.828564360517853, + "grad_norm": 2.3356730937957764, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7842684984207153, + "num_tokens": 920894911.0, + "step": 34302 + }, + { + "epoch": 2.8286468211429043, + "grad_norm": 2.0176615715026855, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7553366422653198, + "num_tokens": 920926775.0, + "step": 34303 + }, + { + "epoch": 2.8287292817679557, + "grad_norm": 1.998421311378479, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.7906953692436218, + "num_tokens": 920957133.0, + "step": 34304 + }, + { + "epoch": 2.8288117423930075, + "grad_norm": 2.2897422313690186, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7580767869949341, + "num_tokens": 920983499.0, + "step": 34305 + }, + { + "epoch": 2.828894203018059, + "grad_norm": 2.0460879802703857, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7706170678138733, + "num_tokens": 921019323.0, + "step": 34306 + }, + { + "epoch": 2.82897666364311, + "grad_norm": 1.8598276376724243, + "learning_rate": 1e-06, + "loss": 0.5994, + "mean_token_accuracy": 0.8047472238540649, + "num_tokens": 921055583.0, + "step": 34307 + }, + { + "epoch": 2.829059124268162, + "grad_norm": 1.9847280979156494, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.7609599828720093, + "num_tokens": 921091214.0, + "step": 34308 + }, + { + "epoch": 2.8291415848932138, + "grad_norm": 2.0344510078430176, + "learning_rate": 1e-06, + "loss": 0.6061, + "mean_token_accuracy": 0.799247145652771, + "num_tokens": 921122971.0, + "step": 34309 + }, + { + "epoch": 2.829224045518265, + "grad_norm": 2.139002561569214, + "learning_rate": 1e-06, + "loss": 0.7096, + "mean_token_accuracy": 0.7662195563316345, + "num_tokens": 921149988.0, + "step": 34310 + }, + { + "epoch": 2.8293065061433165, + "grad_norm": 2.233140707015991, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7376293540000916, + "num_tokens": 921179582.0, + "step": 34311 + }, + { + "epoch": 2.8293889667683683, + "grad_norm": 1.9704484939575195, + "learning_rate": 1e-06, + "loss": 0.5942, + "mean_token_accuracy": 0.7920452356338501, + "num_tokens": 921210663.0, + "step": 34312 + }, + { + "epoch": 2.8294714273934196, + "grad_norm": 1.9678651094436646, + "learning_rate": 1e-06, + "loss": 0.6079, + "mean_token_accuracy": 0.8013754487037659, + "num_tokens": 921242983.0, + "step": 34313 + }, + { + "epoch": 2.829553888018471, + "grad_norm": 1.9941641092300415, + "learning_rate": 1e-06, + "loss": 0.6228, + "mean_token_accuracy": 0.7876626253128052, + "num_tokens": 921276847.0, + "step": 34314 + }, + { + "epoch": 2.8296363486435228, + "grad_norm": 2.170278549194336, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7515025734901428, + "num_tokens": 921305402.0, + "step": 34315 + }, + { + "epoch": 2.829718809268574, + "grad_norm": 1.9944361448287964, + "learning_rate": 1e-06, + "loss": 0.7532, + "mean_token_accuracy": 0.7580141425132751, + "num_tokens": 921337112.0, + "step": 34316 + }, + { + "epoch": 2.829801269893626, + "grad_norm": 1.832457423210144, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.7686892747879028, + "num_tokens": 921380169.0, + "step": 34317 + }, + { + "epoch": 2.8298837305186773, + "grad_norm": 1.8907368183135986, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.7931016683578491, + "num_tokens": 921416209.0, + "step": 34318 + }, + { + "epoch": 2.829966191143729, + "grad_norm": 1.9903244972229004, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7351149320602417, + "num_tokens": 921451146.0, + "step": 34319 + }, + { + "epoch": 2.8300486517687804, + "grad_norm": 1.7534635066986084, + "learning_rate": 1e-06, + "loss": 0.5342, + "mean_token_accuracy": 0.8142595291137695, + "num_tokens": 921491501.0, + "step": 34320 + }, + { + "epoch": 2.8301311123938317, + "grad_norm": 2.3542191982269287, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7351855039596558, + "num_tokens": 921520465.0, + "step": 34321 + }, + { + "epoch": 2.8302135730188835, + "grad_norm": 2.1981468200683594, + "learning_rate": 1e-06, + "loss": 0.7004, + "mean_token_accuracy": 0.7690364718437195, + "num_tokens": 921552264.0, + "step": 34322 + }, + { + "epoch": 2.830296033643935, + "grad_norm": 2.0164101123809814, + "learning_rate": 1e-06, + "loss": 0.7834, + "mean_token_accuracy": 0.7504266500473022, + "num_tokens": 921587413.0, + "step": 34323 + }, + { + "epoch": 2.8303784942689867, + "grad_norm": 2.319268226623535, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7620800137519836, + "num_tokens": 921616744.0, + "step": 34324 + }, + { + "epoch": 2.830460954894038, + "grad_norm": 2.201425075531006, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7512238025665283, + "num_tokens": 921647268.0, + "step": 34325 + }, + { + "epoch": 2.83054341551909, + "grad_norm": 1.979010820388794, + "learning_rate": 1e-06, + "loss": 0.6966, + "mean_token_accuracy": 0.773328423500061, + "num_tokens": 921676657.0, + "step": 34326 + }, + { + "epoch": 2.830625876144141, + "grad_norm": 2.230149030685425, + "learning_rate": 1e-06, + "loss": 0.5649, + "mean_token_accuracy": 0.805698573589325, + "num_tokens": 921702915.0, + "step": 34327 + }, + { + "epoch": 2.8307083367691925, + "grad_norm": 2.8273956775665283, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7554264068603516, + "num_tokens": 921722830.0, + "step": 34328 + }, + { + "epoch": 2.8307907973942443, + "grad_norm": 1.9615236520767212, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7819482684135437, + "num_tokens": 921755270.0, + "step": 34329 + }, + { + "epoch": 2.8308732580192957, + "grad_norm": 2.1901028156280518, + "learning_rate": 1e-06, + "loss": 0.7318, + "mean_token_accuracy": 0.756803572177887, + "num_tokens": 921787203.0, + "step": 34330 + }, + { + "epoch": 2.8309557186443475, + "grad_norm": 2.2502224445343018, + "learning_rate": 1e-06, + "loss": 0.5942, + "mean_token_accuracy": 0.7991646528244019, + "num_tokens": 921812906.0, + "step": 34331 + }, + { + "epoch": 2.831038179269399, + "grad_norm": 1.9974294900894165, + "learning_rate": 1e-06, + "loss": 0.5201, + "mean_token_accuracy": 0.8121083974838257, + "num_tokens": 921841932.0, + "step": 34332 + }, + { + "epoch": 2.8311206398944506, + "grad_norm": 1.9136196374893188, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7629640698432922, + "num_tokens": 921878522.0, + "step": 34333 + }, + { + "epoch": 2.831203100519502, + "grad_norm": 2.3896687030792236, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7365223169326782, + "num_tokens": 921904019.0, + "step": 34334 + }, + { + "epoch": 2.8312855611445533, + "grad_norm": 2.087063789367676, + "learning_rate": 1e-06, + "loss": 0.7668, + "mean_token_accuracy": 0.7561315298080444, + "num_tokens": 921933515.0, + "step": 34335 + }, + { + "epoch": 2.831368021769605, + "grad_norm": 2.1474008560180664, + "learning_rate": 1e-06, + "loss": 0.6591, + "mean_token_accuracy": 0.7749160528182983, + "num_tokens": 921962052.0, + "step": 34336 + }, + { + "epoch": 2.8314504823946565, + "grad_norm": 2.3162355422973633, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7458242177963257, + "num_tokens": 921987095.0, + "step": 34337 + }, + { + "epoch": 2.831532943019708, + "grad_norm": 2.455322742462158, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7323461771011353, + "num_tokens": 922011485.0, + "step": 34338 + }, + { + "epoch": 2.8316154036447596, + "grad_norm": 2.0865657329559326, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.7941648960113525, + "num_tokens": 922041060.0, + "step": 34339 + }, + { + "epoch": 2.8316978642698114, + "grad_norm": 2.281719207763672, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7560442686080933, + "num_tokens": 922066386.0, + "step": 34340 + }, + { + "epoch": 2.8317803248948628, + "grad_norm": 2.0885560512542725, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7784779071807861, + "num_tokens": 922095824.0, + "step": 34341 + }, + { + "epoch": 2.831862785519914, + "grad_norm": 1.9309152364730835, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7825654149055481, + "num_tokens": 922130479.0, + "step": 34342 + }, + { + "epoch": 2.831945246144966, + "grad_norm": 1.7910282611846924, + "learning_rate": 1e-06, + "loss": 0.5724, + "mean_token_accuracy": 0.8026984930038452, + "num_tokens": 922167523.0, + "step": 34343 + }, + { + "epoch": 2.8320277067700173, + "grad_norm": 2.0807018280029297, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7717648148536682, + "num_tokens": 922201460.0, + "step": 34344 + }, + { + "epoch": 2.8321101673950686, + "grad_norm": 2.2534375190734863, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.7833742499351501, + "num_tokens": 922228204.0, + "step": 34345 + }, + { + "epoch": 2.8321926280201204, + "grad_norm": 2.208606719970703, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.7948064804077148, + "num_tokens": 922257714.0, + "step": 34346 + }, + { + "epoch": 2.832275088645172, + "grad_norm": 2.177474021911621, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7717634439468384, + "num_tokens": 922286911.0, + "step": 34347 + }, + { + "epoch": 2.8323575492702235, + "grad_norm": 2.0372278690338135, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7198746204376221, + "num_tokens": 922319473.0, + "step": 34348 + }, + { + "epoch": 2.832440009895275, + "grad_norm": 1.9913790225982666, + "learning_rate": 1e-06, + "loss": 0.6019, + "mean_token_accuracy": 0.8007432222366333, + "num_tokens": 922351619.0, + "step": 34349 + }, + { + "epoch": 2.8325224705203267, + "grad_norm": 2.177271604537964, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7648584842681885, + "num_tokens": 922381803.0, + "step": 34350 + }, + { + "epoch": 2.832604931145378, + "grad_norm": 1.9590046405792236, + "learning_rate": 1e-06, + "loss": 0.5243, + "mean_token_accuracy": 0.8217076063156128, + "num_tokens": 922415285.0, + "step": 34351 + }, + { + "epoch": 2.8326873917704294, + "grad_norm": 1.9930559396743774, + "learning_rate": 1e-06, + "loss": 0.6256, + "mean_token_accuracy": 0.7902357578277588, + "num_tokens": 922454155.0, + "step": 34352 + }, + { + "epoch": 2.832769852395481, + "grad_norm": 2.2487754821777344, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7480226159095764, + "num_tokens": 922482299.0, + "step": 34353 + }, + { + "epoch": 2.8328523130205325, + "grad_norm": 2.0390772819519043, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7777988314628601, + "num_tokens": 922516291.0, + "step": 34354 + }, + { + "epoch": 2.8329347736455843, + "grad_norm": 1.976718544960022, + "learning_rate": 1e-06, + "loss": 0.5966, + "mean_token_accuracy": 0.8036418557167053, + "num_tokens": 922549222.0, + "step": 34355 + }, + { + "epoch": 2.8330172342706357, + "grad_norm": 2.0205492973327637, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7476432919502258, + "num_tokens": 922579780.0, + "step": 34356 + }, + { + "epoch": 2.8330996948956875, + "grad_norm": 2.068397283554077, + "learning_rate": 1e-06, + "loss": 0.5578, + "mean_token_accuracy": 0.8086836338043213, + "num_tokens": 922606841.0, + "step": 34357 + }, + { + "epoch": 2.833182155520739, + "grad_norm": 2.2568843364715576, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.7852819561958313, + "num_tokens": 922630207.0, + "step": 34358 + }, + { + "epoch": 2.83326461614579, + "grad_norm": 1.9860575199127197, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7560024261474609, + "num_tokens": 922663584.0, + "step": 34359 + }, + { + "epoch": 2.833347076770842, + "grad_norm": 1.9876192808151245, + "learning_rate": 1e-06, + "loss": 0.6122, + "mean_token_accuracy": 0.7970151305198669, + "num_tokens": 922693989.0, + "step": 34360 + }, + { + "epoch": 2.8334295373958933, + "grad_norm": 2.0565035343170166, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.7989685535430908, + "num_tokens": 922722777.0, + "step": 34361 + }, + { + "epoch": 2.833511998020945, + "grad_norm": 1.9247932434082031, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7540682554244995, + "num_tokens": 922756349.0, + "step": 34362 + }, + { + "epoch": 2.8335944586459965, + "grad_norm": 2.2653048038482666, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7883946299552917, + "num_tokens": 922783171.0, + "step": 34363 + }, + { + "epoch": 2.8336769192710483, + "grad_norm": 2.132230758666992, + "learning_rate": 1e-06, + "loss": 0.6001, + "mean_token_accuracy": 0.7960535287857056, + "num_tokens": 922812231.0, + "step": 34364 + }, + { + "epoch": 2.8337593798960996, + "grad_norm": 1.879897117614746, + "learning_rate": 1e-06, + "loss": 0.7816, + "mean_token_accuracy": 0.7471110820770264, + "num_tokens": 922850294.0, + "step": 34365 + }, + { + "epoch": 2.833841840521151, + "grad_norm": 1.980278491973877, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7777323722839355, + "num_tokens": 922882038.0, + "step": 34366 + }, + { + "epoch": 2.8339243011462028, + "grad_norm": 2.3139829635620117, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7530361413955688, + "num_tokens": 922909222.0, + "step": 34367 + }, + { + "epoch": 2.834006761771254, + "grad_norm": 1.8741672039031982, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7684494853019714, + "num_tokens": 922945262.0, + "step": 34368 + }, + { + "epoch": 2.834089222396306, + "grad_norm": 1.8799408674240112, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.780713677406311, + "num_tokens": 922980226.0, + "step": 34369 + }, + { + "epoch": 2.8341716830213572, + "grad_norm": 2.0769734382629395, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7424332499504089, + "num_tokens": 923011731.0, + "step": 34370 + }, + { + "epoch": 2.834254143646409, + "grad_norm": 1.948538064956665, + "learning_rate": 1e-06, + "loss": 0.7094, + "mean_token_accuracy": 0.7591975927352905, + "num_tokens": 923045117.0, + "step": 34371 + }, + { + "epoch": 2.8343366042714604, + "grad_norm": 2.0680699348449707, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7677011489868164, + "num_tokens": 923075995.0, + "step": 34372 + }, + { + "epoch": 2.8344190648965117, + "grad_norm": 1.9355133771896362, + "learning_rate": 1e-06, + "loss": 0.736, + "mean_token_accuracy": 0.7722663879394531, + "num_tokens": 923112664.0, + "step": 34373 + }, + { + "epoch": 2.8345015255215635, + "grad_norm": 1.9177604913711548, + "learning_rate": 1e-06, + "loss": 0.7073, + "mean_token_accuracy": 0.776953935623169, + "num_tokens": 923147995.0, + "step": 34374 + }, + { + "epoch": 2.834583986146615, + "grad_norm": 2.2778992652893066, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.7592862844467163, + "num_tokens": 923172984.0, + "step": 34375 + }, + { + "epoch": 2.8346664467716667, + "grad_norm": 2.04453182220459, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.7510465383529663, + "num_tokens": 923204095.0, + "step": 34376 + }, + { + "epoch": 2.834748907396718, + "grad_norm": 1.9664286375045776, + "learning_rate": 1e-06, + "loss": 0.5962, + "mean_token_accuracy": 0.804081916809082, + "num_tokens": 923237064.0, + "step": 34377 + }, + { + "epoch": 2.83483136802177, + "grad_norm": 2.0314276218414307, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7576460242271423, + "num_tokens": 923266626.0, + "step": 34378 + }, + { + "epoch": 2.834913828646821, + "grad_norm": 2.1888267993927, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7758857011795044, + "num_tokens": 923295863.0, + "step": 34379 + }, + { + "epoch": 2.8349962892718725, + "grad_norm": 1.9379924535751343, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7718842029571533, + "num_tokens": 923328252.0, + "step": 34380 + }, + { + "epoch": 2.8350787498969243, + "grad_norm": 2.04560923576355, + "learning_rate": 1e-06, + "loss": 0.6458, + "mean_token_accuracy": 0.7835914492607117, + "num_tokens": 923356789.0, + "step": 34381 + }, + { + "epoch": 2.8351612105219757, + "grad_norm": 2.057093620300293, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7699342966079712, + "num_tokens": 923387164.0, + "step": 34382 + }, + { + "epoch": 2.835243671147027, + "grad_norm": 2.7358367443084717, + "learning_rate": 1e-06, + "loss": 0.6411, + "mean_token_accuracy": 0.7843297719955444, + "num_tokens": 923405246.0, + "step": 34383 + }, + { + "epoch": 2.835326131772079, + "grad_norm": 2.059363842010498, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.8030146360397339, + "num_tokens": 923436662.0, + "step": 34384 + }, + { + "epoch": 2.8354085923971306, + "grad_norm": 1.9735705852508545, + "learning_rate": 1e-06, + "loss": 0.5513, + "mean_token_accuracy": 0.8150426149368286, + "num_tokens": 923469728.0, + "step": 34385 + }, + { + "epoch": 2.835491053022182, + "grad_norm": 2.589392900466919, + "learning_rate": 1e-06, + "loss": 0.7189, + "mean_token_accuracy": 0.761139988899231, + "num_tokens": 923493864.0, + "step": 34386 + }, + { + "epoch": 2.8355735136472333, + "grad_norm": 1.987046480178833, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.7577841281890869, + "num_tokens": 923526566.0, + "step": 34387 + }, + { + "epoch": 2.835655974272285, + "grad_norm": 2.5875723361968994, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7456092238426208, + "num_tokens": 923549323.0, + "step": 34388 + }, + { + "epoch": 2.8357384348973365, + "grad_norm": 1.9817280769348145, + "learning_rate": 1e-06, + "loss": 0.6199, + "mean_token_accuracy": 0.8122701644897461, + "num_tokens": 923581262.0, + "step": 34389 + }, + { + "epoch": 2.835820895522388, + "grad_norm": 2.096557855606079, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.75029456615448, + "num_tokens": 923612264.0, + "step": 34390 + }, + { + "epoch": 2.8359033561474396, + "grad_norm": 2.184288501739502, + "learning_rate": 1e-06, + "loss": 0.7178, + "mean_token_accuracy": 0.7636401653289795, + "num_tokens": 923641354.0, + "step": 34391 + }, + { + "epoch": 2.8359858167724914, + "grad_norm": 2.3235700130462646, + "learning_rate": 1e-06, + "loss": 0.7462, + "mean_token_accuracy": 0.7702221870422363, + "num_tokens": 923669108.0, + "step": 34392 + }, + { + "epoch": 2.8360682773975427, + "grad_norm": 1.9389960765838623, + "learning_rate": 1e-06, + "loss": 0.7666, + "mean_token_accuracy": 0.7590512633323669, + "num_tokens": 923701408.0, + "step": 34393 + }, + { + "epoch": 2.836150738022594, + "grad_norm": 2.011904239654541, + "learning_rate": 1e-06, + "loss": 0.6161, + "mean_token_accuracy": 0.7967990636825562, + "num_tokens": 923733705.0, + "step": 34394 + }, + { + "epoch": 2.836233198647646, + "grad_norm": 1.899065613746643, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7674853205680847, + "num_tokens": 923767793.0, + "step": 34395 + }, + { + "epoch": 2.8363156592726972, + "grad_norm": 1.7716318368911743, + "learning_rate": 1e-06, + "loss": 0.5189, + "mean_token_accuracy": 0.8257925510406494, + "num_tokens": 923804531.0, + "step": 34396 + }, + { + "epoch": 2.8363981198977486, + "grad_norm": 1.9008331298828125, + "learning_rate": 1e-06, + "loss": 0.7948, + "mean_token_accuracy": 0.7451222538948059, + "num_tokens": 923841513.0, + "step": 34397 + }, + { + "epoch": 2.8364805805228004, + "grad_norm": 2.187962055206299, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.7857108116149902, + "num_tokens": 923870057.0, + "step": 34398 + }, + { + "epoch": 2.8365630411478517, + "grad_norm": 2.2834184169769287, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7322063446044922, + "num_tokens": 923895647.0, + "step": 34399 + }, + { + "epoch": 2.8366455017729035, + "grad_norm": 1.9132901430130005, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.773334264755249, + "num_tokens": 923930776.0, + "step": 34400 + }, + { + "epoch": 2.836727962397955, + "grad_norm": 2.26509690284729, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7880885601043701, + "num_tokens": 923956232.0, + "step": 34401 + }, + { + "epoch": 2.8368104230230067, + "grad_norm": 2.173851728439331, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7521985769271851, + "num_tokens": 923986937.0, + "step": 34402 + }, + { + "epoch": 2.836892883648058, + "grad_norm": 2.191641092300415, + "learning_rate": 1e-06, + "loss": 0.7621, + "mean_token_accuracy": 0.7623609304428101, + "num_tokens": 924015126.0, + "step": 34403 + }, + { + "epoch": 2.8369753442731094, + "grad_norm": 2.006335496902466, + "learning_rate": 1e-06, + "loss": 0.6017, + "mean_token_accuracy": 0.8041303753852844, + "num_tokens": 924045001.0, + "step": 34404 + }, + { + "epoch": 2.837057804898161, + "grad_norm": 2.1286377906799316, + "learning_rate": 1e-06, + "loss": 0.6576, + "mean_token_accuracy": 0.7810778617858887, + "num_tokens": 924074345.0, + "step": 34405 + }, + { + "epoch": 2.8371402655232125, + "grad_norm": 2.466654062271118, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7389912605285645, + "num_tokens": 924097156.0, + "step": 34406 + }, + { + "epoch": 2.8372227261482643, + "grad_norm": 1.9582973718643188, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7645791172981262, + "num_tokens": 924134080.0, + "step": 34407 + }, + { + "epoch": 2.8373051867733157, + "grad_norm": 1.9182273149490356, + "learning_rate": 1e-06, + "loss": 0.5567, + "mean_token_accuracy": 0.8217400312423706, + "num_tokens": 924165468.0, + "step": 34408 + }, + { + "epoch": 2.8373876473983675, + "grad_norm": 1.8173202276229858, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7728786468505859, + "num_tokens": 924203572.0, + "step": 34409 + }, + { + "epoch": 2.837470108023419, + "grad_norm": 2.48868465423584, + "learning_rate": 1e-06, + "loss": 0.7516, + "mean_token_accuracy": 0.7579762935638428, + "num_tokens": 924229965.0, + "step": 34410 + }, + { + "epoch": 2.83755256864847, + "grad_norm": 2.014538049697876, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7560631036758423, + "num_tokens": 924262764.0, + "step": 34411 + }, + { + "epoch": 2.837635029273522, + "grad_norm": 2.1484971046447754, + "learning_rate": 1e-06, + "loss": 0.7007, + "mean_token_accuracy": 0.774012565612793, + "num_tokens": 924291648.0, + "step": 34412 + }, + { + "epoch": 2.8377174898985733, + "grad_norm": 2.1270861625671387, + "learning_rate": 1e-06, + "loss": 0.802, + "mean_token_accuracy": 0.732860803604126, + "num_tokens": 924323759.0, + "step": 34413 + }, + { + "epoch": 2.837799950523625, + "grad_norm": 2.032405138015747, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7681688070297241, + "num_tokens": 924355858.0, + "step": 34414 + }, + { + "epoch": 2.8378824111486765, + "grad_norm": 2.171248197555542, + "learning_rate": 1e-06, + "loss": 0.587, + "mean_token_accuracy": 0.8015796542167664, + "num_tokens": 924383741.0, + "step": 34415 + }, + { + "epoch": 2.8379648717737282, + "grad_norm": 1.7772526741027832, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.755426824092865, + "num_tokens": 924423563.0, + "step": 34416 + }, + { + "epoch": 2.8380473323987796, + "grad_norm": 2.4477717876434326, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7396015524864197, + "num_tokens": 924449840.0, + "step": 34417 + }, + { + "epoch": 2.838129793023831, + "grad_norm": 2.2857935428619385, + "learning_rate": 1e-06, + "loss": 0.7169, + "mean_token_accuracy": 0.7669745683670044, + "num_tokens": 924474804.0, + "step": 34418 + }, + { + "epoch": 2.8382122536488827, + "grad_norm": 2.2477622032165527, + "learning_rate": 1e-06, + "loss": 0.6533, + "mean_token_accuracy": 0.7807357311248779, + "num_tokens": 924501694.0, + "step": 34419 + }, + { + "epoch": 2.838294714273934, + "grad_norm": 1.9423962831497192, + "learning_rate": 1e-06, + "loss": 0.551, + "mean_token_accuracy": 0.8202192187309265, + "num_tokens": 924533356.0, + "step": 34420 + }, + { + "epoch": 2.838377174898986, + "grad_norm": 2.0960512161254883, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.7870221138000488, + "num_tokens": 924563803.0, + "step": 34421 + }, + { + "epoch": 2.8384596355240372, + "grad_norm": 2.2003238201141357, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7763597369194031, + "num_tokens": 924590341.0, + "step": 34422 + }, + { + "epoch": 2.838542096149089, + "grad_norm": 2.0225231647491455, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7508236169815063, + "num_tokens": 924623071.0, + "step": 34423 + }, + { + "epoch": 2.8386245567741404, + "grad_norm": 1.9805669784545898, + "learning_rate": 1e-06, + "loss": 0.5774, + "mean_token_accuracy": 0.8082106113433838, + "num_tokens": 924657222.0, + "step": 34424 + }, + { + "epoch": 2.8387070173991917, + "grad_norm": 1.9721211194992065, + "learning_rate": 1e-06, + "loss": 0.5929, + "mean_token_accuracy": 0.7933293581008911, + "num_tokens": 924688307.0, + "step": 34425 + }, + { + "epoch": 2.8387894780242435, + "grad_norm": 1.9214084148406982, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7470459342002869, + "num_tokens": 924721883.0, + "step": 34426 + }, + { + "epoch": 2.838871938649295, + "grad_norm": 2.78385591506958, + "learning_rate": 1e-06, + "loss": 0.6918, + "mean_token_accuracy": 0.768216609954834, + "num_tokens": 924746088.0, + "step": 34427 + }, + { + "epoch": 2.8389543992743462, + "grad_norm": 2.0465095043182373, + "learning_rate": 1e-06, + "loss": 0.7832, + "mean_token_accuracy": 0.7421443462371826, + "num_tokens": 924777331.0, + "step": 34428 + }, + { + "epoch": 2.839036859899398, + "grad_norm": 2.0477285385131836, + "learning_rate": 1e-06, + "loss": 0.5998, + "mean_token_accuracy": 0.7921286821365356, + "num_tokens": 924807190.0, + "step": 34429 + }, + { + "epoch": 2.83911932052445, + "grad_norm": 2.0779380798339844, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7844158411026001, + "num_tokens": 924835949.0, + "step": 34430 + }, + { + "epoch": 2.839201781149501, + "grad_norm": 2.1196484565734863, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7313907146453857, + "num_tokens": 924868502.0, + "step": 34431 + }, + { + "epoch": 2.8392842417745525, + "grad_norm": 2.4753811359405518, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7335189580917358, + "num_tokens": 924894694.0, + "step": 34432 + }, + { + "epoch": 2.8393667023996043, + "grad_norm": 1.9503461122512817, + "learning_rate": 1e-06, + "loss": 0.6783, + "mean_token_accuracy": 0.7822278738021851, + "num_tokens": 924928293.0, + "step": 34433 + }, + { + "epoch": 2.8394491630246557, + "grad_norm": 2.2163023948669434, + "learning_rate": 1e-06, + "loss": 0.6912, + "mean_token_accuracy": 0.7743431329727173, + "num_tokens": 924957728.0, + "step": 34434 + }, + { + "epoch": 2.839531623649707, + "grad_norm": 1.9575467109680176, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7837958335876465, + "num_tokens": 924988663.0, + "step": 34435 + }, + { + "epoch": 2.839614084274759, + "grad_norm": 2.070328712463379, + "learning_rate": 1e-06, + "loss": 0.644, + "mean_token_accuracy": 0.786116361618042, + "num_tokens": 925019283.0, + "step": 34436 + }, + { + "epoch": 2.8396965448998106, + "grad_norm": 2.2932398319244385, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7484480142593384, + "num_tokens": 925047570.0, + "step": 34437 + }, + { + "epoch": 2.839779005524862, + "grad_norm": 2.7519309520721436, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7768605947494507, + "num_tokens": 925066993.0, + "step": 34438 + }, + { + "epoch": 2.8398614661499133, + "grad_norm": 2.440338373184204, + "learning_rate": 1e-06, + "loss": 0.7162, + "mean_token_accuracy": 0.7721421718597412, + "num_tokens": 925090454.0, + "step": 34439 + }, + { + "epoch": 2.839943926774965, + "grad_norm": 2.3650200366973877, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7618963718414307, + "num_tokens": 925116016.0, + "step": 34440 + }, + { + "epoch": 2.8400263874000164, + "grad_norm": 1.9548072814941406, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7832965850830078, + "num_tokens": 925150541.0, + "step": 34441 + }, + { + "epoch": 2.840108848025068, + "grad_norm": 1.8675925731658936, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.7995407581329346, + "num_tokens": 925184663.0, + "step": 34442 + }, + { + "epoch": 2.8401913086501196, + "grad_norm": 2.1426570415496826, + "learning_rate": 1e-06, + "loss": 0.6011, + "mean_token_accuracy": 0.7966092824935913, + "num_tokens": 925215538.0, + "step": 34443 + }, + { + "epoch": 2.840273769275171, + "grad_norm": 1.9993945360183716, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.7815276384353638, + "num_tokens": 925245739.0, + "step": 34444 + }, + { + "epoch": 2.8403562299002227, + "grad_norm": 2.0792551040649414, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7675131559371948, + "num_tokens": 925273393.0, + "step": 34445 + }, + { + "epoch": 2.840438690525274, + "grad_norm": 2.1414191722869873, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7797659039497375, + "num_tokens": 925301710.0, + "step": 34446 + }, + { + "epoch": 2.840521151150326, + "grad_norm": 2.3704092502593994, + "learning_rate": 1e-06, + "loss": 0.625, + "mean_token_accuracy": 0.7862948179244995, + "num_tokens": 925325553.0, + "step": 34447 + }, + { + "epoch": 2.8406036117753772, + "grad_norm": 1.79023277759552, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7505558729171753, + "num_tokens": 925362522.0, + "step": 34448 + }, + { + "epoch": 2.8406860724004286, + "grad_norm": 1.900945782661438, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7714080214500427, + "num_tokens": 925397109.0, + "step": 34449 + }, + { + "epoch": 2.8407685330254804, + "grad_norm": 2.642526388168335, + "learning_rate": 1e-06, + "loss": 0.7737, + "mean_token_accuracy": 0.7584431767463684, + "num_tokens": 925419599.0, + "step": 34450 + }, + { + "epoch": 2.8408509936505317, + "grad_norm": 1.8057003021240234, + "learning_rate": 1e-06, + "loss": 0.6813, + "mean_token_accuracy": 0.7791690826416016, + "num_tokens": 925457504.0, + "step": 34451 + }, + { + "epoch": 2.8409334542755835, + "grad_norm": 2.2276930809020996, + "learning_rate": 1e-06, + "loss": 0.7517, + "mean_token_accuracy": 0.7515988349914551, + "num_tokens": 925484701.0, + "step": 34452 + }, + { + "epoch": 2.841015914900635, + "grad_norm": 2.532863140106201, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7331981062889099, + "num_tokens": 925507538.0, + "step": 34453 + }, + { + "epoch": 2.8410983755256867, + "grad_norm": 2.0684945583343506, + "learning_rate": 1e-06, + "loss": 0.5393, + "mean_token_accuracy": 0.7976884841918945, + "num_tokens": 925535998.0, + "step": 34454 + }, + { + "epoch": 2.841180836150738, + "grad_norm": 2.2815284729003906, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.7723280191421509, + "num_tokens": 925562881.0, + "step": 34455 + }, + { + "epoch": 2.8412632967757894, + "grad_norm": 2.0785279273986816, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7579777240753174, + "num_tokens": 925591722.0, + "step": 34456 + }, + { + "epoch": 2.841345757400841, + "grad_norm": 2.0352046489715576, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.7831713557243347, + "num_tokens": 925624794.0, + "step": 34457 + }, + { + "epoch": 2.8414282180258925, + "grad_norm": 2.3251445293426514, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.780919075012207, + "num_tokens": 925649733.0, + "step": 34458 + }, + { + "epoch": 2.8415106786509443, + "grad_norm": 1.9664393663406372, + "learning_rate": 1e-06, + "loss": 0.7499, + "mean_token_accuracy": 0.7635458111763, + "num_tokens": 925683384.0, + "step": 34459 + }, + { + "epoch": 2.8415931392759957, + "grad_norm": 2.0316920280456543, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.7795783877372742, + "num_tokens": 925716032.0, + "step": 34460 + }, + { + "epoch": 2.8416755999010475, + "grad_norm": 2.1659841537475586, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.741245687007904, + "num_tokens": 925746023.0, + "step": 34461 + }, + { + "epoch": 2.841758060526099, + "grad_norm": 2.4353370666503906, + "learning_rate": 1e-06, + "loss": 0.6915, + "mean_token_accuracy": 0.7734155654907227, + "num_tokens": 925768807.0, + "step": 34462 + }, + { + "epoch": 2.84184052115115, + "grad_norm": 2.3502912521362305, + "learning_rate": 1e-06, + "loss": 0.7484, + "mean_token_accuracy": 0.7669450640678406, + "num_tokens": 925794226.0, + "step": 34463 + }, + { + "epoch": 2.841922981776202, + "grad_norm": 2.488023042678833, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7399671673774719, + "num_tokens": 925815466.0, + "step": 34464 + }, + { + "epoch": 2.8420054424012533, + "grad_norm": 1.7966222763061523, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7440732717514038, + "num_tokens": 925856511.0, + "step": 34465 + }, + { + "epoch": 2.842087903026305, + "grad_norm": 2.1632444858551025, + "learning_rate": 1e-06, + "loss": 0.5801, + "mean_token_accuracy": 0.8093001842498779, + "num_tokens": 925885288.0, + "step": 34466 + }, + { + "epoch": 2.8421703636513564, + "grad_norm": 2.033778667449951, + "learning_rate": 1e-06, + "loss": 0.6896, + "mean_token_accuracy": 0.7710681557655334, + "num_tokens": 925918631.0, + "step": 34467 + }, + { + "epoch": 2.8422528242764082, + "grad_norm": 1.998103380203247, + "learning_rate": 1e-06, + "loss": 0.5791, + "mean_token_accuracy": 0.7985658645629883, + "num_tokens": 925951338.0, + "step": 34468 + }, + { + "epoch": 2.8423352849014596, + "grad_norm": 2.219561815261841, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7726938724517822, + "num_tokens": 925979316.0, + "step": 34469 + }, + { + "epoch": 2.842417745526511, + "grad_norm": 2.291405200958252, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7404210567474365, + "num_tokens": 926007452.0, + "step": 34470 + }, + { + "epoch": 2.8425002061515627, + "grad_norm": 1.9730758666992188, + "learning_rate": 1e-06, + "loss": 0.6326, + "mean_token_accuracy": 0.7857179641723633, + "num_tokens": 926039038.0, + "step": 34471 + }, + { + "epoch": 2.842582666776614, + "grad_norm": 1.8356471061706543, + "learning_rate": 1e-06, + "loss": 0.6433, + "mean_token_accuracy": 0.7977066040039062, + "num_tokens": 926074187.0, + "step": 34472 + }, + { + "epoch": 2.8426651274016654, + "grad_norm": 2.257338285446167, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7371375560760498, + "num_tokens": 926105003.0, + "step": 34473 + }, + { + "epoch": 2.8427475880267172, + "grad_norm": 2.6104202270507812, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7715929746627808, + "num_tokens": 926126030.0, + "step": 34474 + }, + { + "epoch": 2.842830048651769, + "grad_norm": 2.2570931911468506, + "learning_rate": 1e-06, + "loss": 0.7582, + "mean_token_accuracy": 0.7654861211776733, + "num_tokens": 926152909.0, + "step": 34475 + }, + { + "epoch": 2.8429125092768204, + "grad_norm": 2.2531228065490723, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7341113090515137, + "num_tokens": 926179493.0, + "step": 34476 + }, + { + "epoch": 2.8429949699018717, + "grad_norm": 2.3028745651245117, + "learning_rate": 1e-06, + "loss": 0.7371, + "mean_token_accuracy": 0.7631049752235413, + "num_tokens": 926208229.0, + "step": 34477 + }, + { + "epoch": 2.8430774305269235, + "grad_norm": 2.0943918228149414, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.7924081087112427, + "num_tokens": 926240013.0, + "step": 34478 + }, + { + "epoch": 2.843159891151975, + "grad_norm": 2.0623457431793213, + "learning_rate": 1e-06, + "loss": 0.7481, + "mean_token_accuracy": 0.7581331729888916, + "num_tokens": 926269314.0, + "step": 34479 + }, + { + "epoch": 2.8432423517770262, + "grad_norm": 1.9319721460342407, + "learning_rate": 1e-06, + "loss": 0.5663, + "mean_token_accuracy": 0.8092607259750366, + "num_tokens": 926306664.0, + "step": 34480 + }, + { + "epoch": 2.843324812402078, + "grad_norm": 2.028141498565674, + "learning_rate": 1e-06, + "loss": 0.5743, + "mean_token_accuracy": 0.7935960292816162, + "num_tokens": 926336246.0, + "step": 34481 + }, + { + "epoch": 2.84340727302713, + "grad_norm": 2.1392879486083984, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7814569473266602, + "num_tokens": 926363281.0, + "step": 34482 + }, + { + "epoch": 2.843489733652181, + "grad_norm": 2.0506670475006104, + "learning_rate": 1e-06, + "loss": 0.6153, + "mean_token_accuracy": 0.7986500263214111, + "num_tokens": 926395641.0, + "step": 34483 + }, + { + "epoch": 2.8435721942772325, + "grad_norm": 2.6264255046844482, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7339791059494019, + "num_tokens": 926418325.0, + "step": 34484 + }, + { + "epoch": 2.8436546549022843, + "grad_norm": 1.6868547201156616, + "learning_rate": 1e-06, + "loss": 0.5691, + "mean_token_accuracy": 0.8142616748809814, + "num_tokens": 926460669.0, + "step": 34485 + }, + { + "epoch": 2.8437371155273357, + "grad_norm": 2.105344295501709, + "learning_rate": 1e-06, + "loss": 0.566, + "mean_token_accuracy": 0.8079084157943726, + "num_tokens": 926487304.0, + "step": 34486 + }, + { + "epoch": 2.843819576152387, + "grad_norm": 2.2289929389953613, + "learning_rate": 1e-06, + "loss": 0.7231, + "mean_token_accuracy": 0.7566974759101868, + "num_tokens": 926515189.0, + "step": 34487 + }, + { + "epoch": 2.843902036777439, + "grad_norm": 2.0488061904907227, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.7820633053779602, + "num_tokens": 926550125.0, + "step": 34488 + }, + { + "epoch": 2.84398449740249, + "grad_norm": 2.004255533218384, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7461413145065308, + "num_tokens": 926584321.0, + "step": 34489 + }, + { + "epoch": 2.844066958027542, + "grad_norm": 2.4880847930908203, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7855422496795654, + "num_tokens": 926608994.0, + "step": 34490 + }, + { + "epoch": 2.8441494186525933, + "grad_norm": 2.0228426456451416, + "learning_rate": 1e-06, + "loss": 0.568, + "mean_token_accuracy": 0.8037022352218628, + "num_tokens": 926638802.0, + "step": 34491 + }, + { + "epoch": 2.844231879277645, + "grad_norm": 2.2990362644195557, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.7379516363143921, + "num_tokens": 926667661.0, + "step": 34492 + }, + { + "epoch": 2.8443143399026964, + "grad_norm": 2.2056310176849365, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7503687143325806, + "num_tokens": 926699797.0, + "step": 34493 + }, + { + "epoch": 2.844396800527748, + "grad_norm": 2.160456418991089, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7708089351654053, + "num_tokens": 926729463.0, + "step": 34494 + }, + { + "epoch": 2.8444792611527996, + "grad_norm": 1.8181146383285522, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7761995792388916, + "num_tokens": 926764879.0, + "step": 34495 + }, + { + "epoch": 2.844561721777851, + "grad_norm": 2.2325599193573, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7873347997665405, + "num_tokens": 926792145.0, + "step": 34496 + }, + { + "epoch": 2.8446441824029027, + "grad_norm": 2.2341387271881104, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.7871296405792236, + "num_tokens": 926818497.0, + "step": 34497 + }, + { + "epoch": 2.844726643027954, + "grad_norm": 2.29826283454895, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7314871549606323, + "num_tokens": 926845961.0, + "step": 34498 + }, + { + "epoch": 2.844809103653006, + "grad_norm": 2.027256727218628, + "learning_rate": 1e-06, + "loss": 0.6117, + "mean_token_accuracy": 0.7997763156890869, + "num_tokens": 926874598.0, + "step": 34499 + }, + { + "epoch": 2.8448915642780572, + "grad_norm": 2.190875768661499, + "learning_rate": 1e-06, + "loss": 0.6506, + "mean_token_accuracy": 0.7946009635925293, + "num_tokens": 926900743.0, + "step": 34500 + }, + { + "epoch": 2.8449740249031086, + "grad_norm": 2.429550886154175, + "learning_rate": 1e-06, + "loss": 0.6315, + "mean_token_accuracy": 0.795532763004303, + "num_tokens": 926923323.0, + "step": 34501 + }, + { + "epoch": 2.8450564855281604, + "grad_norm": 2.001279354095459, + "learning_rate": 1e-06, + "loss": 0.6613, + "mean_token_accuracy": 0.7810060381889343, + "num_tokens": 926954499.0, + "step": 34502 + }, + { + "epoch": 2.8451389461532117, + "grad_norm": 2.0274627208709717, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.7952654957771301, + "num_tokens": 926984764.0, + "step": 34503 + }, + { + "epoch": 2.8452214067782635, + "grad_norm": 2.355567693710327, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7533876895904541, + "num_tokens": 927008188.0, + "step": 34504 + }, + { + "epoch": 2.845303867403315, + "grad_norm": 1.9829179048538208, + "learning_rate": 1e-06, + "loss": 0.6104, + "mean_token_accuracy": 0.8014142513275146, + "num_tokens": 927040876.0, + "step": 34505 + }, + { + "epoch": 2.8453863280283667, + "grad_norm": 2.2541470527648926, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7322417497634888, + "num_tokens": 927067620.0, + "step": 34506 + }, + { + "epoch": 2.845468788653418, + "grad_norm": 1.9715311527252197, + "learning_rate": 1e-06, + "loss": 0.7719, + "mean_token_accuracy": 0.758660078048706, + "num_tokens": 927103342.0, + "step": 34507 + }, + { + "epoch": 2.8455512492784694, + "grad_norm": 2.2782206535339355, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7464567422866821, + "num_tokens": 927131245.0, + "step": 34508 + }, + { + "epoch": 2.845633709903521, + "grad_norm": 2.160820484161377, + "learning_rate": 1e-06, + "loss": 0.6231, + "mean_token_accuracy": 0.8041164875030518, + "num_tokens": 927160991.0, + "step": 34509 + }, + { + "epoch": 2.8457161705285725, + "grad_norm": 2.219935894012451, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7358712553977966, + "num_tokens": 927189596.0, + "step": 34510 + }, + { + "epoch": 2.8457986311536243, + "grad_norm": 2.265162706375122, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7523142099380493, + "num_tokens": 927215246.0, + "step": 34511 + }, + { + "epoch": 2.8458810917786757, + "grad_norm": 2.223926305770874, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7736140489578247, + "num_tokens": 927246002.0, + "step": 34512 + }, + { + "epoch": 2.8459635524037274, + "grad_norm": 2.1030642986297607, + "learning_rate": 1e-06, + "loss": 0.6118, + "mean_token_accuracy": 0.805099368095398, + "num_tokens": 927276372.0, + "step": 34513 + }, + { + "epoch": 2.846046013028779, + "grad_norm": 1.809118390083313, + "learning_rate": 1e-06, + "loss": 0.6059, + "mean_token_accuracy": 0.7991303205490112, + "num_tokens": 927310938.0, + "step": 34514 + }, + { + "epoch": 2.84612847365383, + "grad_norm": 2.0155439376831055, + "learning_rate": 1e-06, + "loss": 0.5633, + "mean_token_accuracy": 0.8133447170257568, + "num_tokens": 927339342.0, + "step": 34515 + }, + { + "epoch": 2.846210934278882, + "grad_norm": 2.592272996902466, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.7738927006721497, + "num_tokens": 927360652.0, + "step": 34516 + }, + { + "epoch": 2.8462933949039333, + "grad_norm": 2.1093294620513916, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.7786412239074707, + "num_tokens": 927387017.0, + "step": 34517 + }, + { + "epoch": 2.8463758555289846, + "grad_norm": 2.2072556018829346, + "learning_rate": 1e-06, + "loss": 0.671, + "mean_token_accuracy": 0.7693870067596436, + "num_tokens": 927414152.0, + "step": 34518 + }, + { + "epoch": 2.8464583161540364, + "grad_norm": 2.132802963256836, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7763636112213135, + "num_tokens": 927443309.0, + "step": 34519 + }, + { + "epoch": 2.8465407767790882, + "grad_norm": 2.029900550842285, + "learning_rate": 1e-06, + "loss": 0.6972, + "mean_token_accuracy": 0.7706320285797119, + "num_tokens": 927475016.0, + "step": 34520 + }, + { + "epoch": 2.8466232374041396, + "grad_norm": 2.146610975265503, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7518888711929321, + "num_tokens": 927504267.0, + "step": 34521 + }, + { + "epoch": 2.846705698029191, + "grad_norm": 2.0141761302948, + "learning_rate": 1e-06, + "loss": 0.5733, + "mean_token_accuracy": 0.7985784411430359, + "num_tokens": 927534593.0, + "step": 34522 + }, + { + "epoch": 2.8467881586542427, + "grad_norm": 2.1007027626037598, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7529054880142212, + "num_tokens": 927568475.0, + "step": 34523 + }, + { + "epoch": 2.846870619279294, + "grad_norm": 2.0288145542144775, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7503154277801514, + "num_tokens": 927600889.0, + "step": 34524 + }, + { + "epoch": 2.8469530799043454, + "grad_norm": 2.2016873359680176, + "learning_rate": 1e-06, + "loss": 0.6086, + "mean_token_accuracy": 0.789447546005249, + "num_tokens": 927625560.0, + "step": 34525 + }, + { + "epoch": 2.8470355405293972, + "grad_norm": 2.055189609527588, + "learning_rate": 1e-06, + "loss": 0.7715, + "mean_token_accuracy": 0.7514513731002808, + "num_tokens": 927655445.0, + "step": 34526 + }, + { + "epoch": 2.847118001154449, + "grad_norm": 2.2671732902526855, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7703677415847778, + "num_tokens": 927682790.0, + "step": 34527 + }, + { + "epoch": 2.8472004617795004, + "grad_norm": 1.8822497129440308, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.8001694679260254, + "num_tokens": 927719311.0, + "step": 34528 + }, + { + "epoch": 2.8472829224045517, + "grad_norm": 2.353715181350708, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7824488878250122, + "num_tokens": 927744296.0, + "step": 34529 + }, + { + "epoch": 2.8473653830296035, + "grad_norm": 2.2440643310546875, + "learning_rate": 1e-06, + "loss": 0.6778, + "mean_token_accuracy": 0.7777655124664307, + "num_tokens": 927770734.0, + "step": 34530 + }, + { + "epoch": 2.847447843654655, + "grad_norm": 2.236433982849121, + "learning_rate": 1e-06, + "loss": 0.5902, + "mean_token_accuracy": 0.7994921207427979, + "num_tokens": 927796124.0, + "step": 34531 + }, + { + "epoch": 2.847530304279706, + "grad_norm": 1.94065260887146, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7695457339286804, + "num_tokens": 927828851.0, + "step": 34532 + }, + { + "epoch": 2.847612764904758, + "grad_norm": 2.004167318344116, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.7602484226226807, + "num_tokens": 927864999.0, + "step": 34533 + }, + { + "epoch": 2.8476952255298094, + "grad_norm": 2.4153594970703125, + "learning_rate": 1e-06, + "loss": 0.7052, + "mean_token_accuracy": 0.7647015452384949, + "num_tokens": 927888612.0, + "step": 34534 + }, + { + "epoch": 2.847777686154861, + "grad_norm": 2.210768938064575, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.76419997215271, + "num_tokens": 927917013.0, + "step": 34535 + }, + { + "epoch": 2.8478601467799125, + "grad_norm": 1.9018654823303223, + "learning_rate": 1e-06, + "loss": 0.6521, + "mean_token_accuracy": 0.791120171546936, + "num_tokens": 927950741.0, + "step": 34536 + }, + { + "epoch": 2.8479426074049643, + "grad_norm": 2.004805326461792, + "learning_rate": 1e-06, + "loss": 0.6795, + "mean_token_accuracy": 0.7874772548675537, + "num_tokens": 927984475.0, + "step": 34537 + }, + { + "epoch": 2.8480250680300156, + "grad_norm": 2.33441424369812, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7772231698036194, + "num_tokens": 928010808.0, + "step": 34538 + }, + { + "epoch": 2.848107528655067, + "grad_norm": 2.3565354347229004, + "learning_rate": 1e-06, + "loss": 0.6427, + "mean_token_accuracy": 0.7899901866912842, + "num_tokens": 928036337.0, + "step": 34539 + }, + { + "epoch": 2.848189989280119, + "grad_norm": 1.86480712890625, + "learning_rate": 1e-06, + "loss": 0.5989, + "mean_token_accuracy": 0.7985553741455078, + "num_tokens": 928069474.0, + "step": 34540 + }, + { + "epoch": 2.84827244990517, + "grad_norm": 2.0826563835144043, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.775547981262207, + "num_tokens": 928095444.0, + "step": 34541 + }, + { + "epoch": 2.848354910530222, + "grad_norm": 1.9534136056900024, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7540399432182312, + "num_tokens": 928133079.0, + "step": 34542 + }, + { + "epoch": 2.8484373711552733, + "grad_norm": 1.8335866928100586, + "learning_rate": 1e-06, + "loss": 0.5479, + "mean_token_accuracy": 0.817577600479126, + "num_tokens": 928166201.0, + "step": 34543 + }, + { + "epoch": 2.848519831780325, + "grad_norm": 2.19454288482666, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.7597792744636536, + "num_tokens": 928196280.0, + "step": 34544 + }, + { + "epoch": 2.8486022924053764, + "grad_norm": 2.175426959991455, + "learning_rate": 1e-06, + "loss": 0.6616, + "mean_token_accuracy": 0.7740694284439087, + "num_tokens": 928227795.0, + "step": 34545 + }, + { + "epoch": 2.848684753030428, + "grad_norm": 2.105950117111206, + "learning_rate": 1e-06, + "loss": 0.5197, + "mean_token_accuracy": 0.8208067417144775, + "num_tokens": 928253192.0, + "step": 34546 + }, + { + "epoch": 2.8487672136554796, + "grad_norm": 2.325922966003418, + "learning_rate": 1e-06, + "loss": 0.574, + "mean_token_accuracy": 0.8011096715927124, + "num_tokens": 928275838.0, + "step": 34547 + }, + { + "epoch": 2.848849674280531, + "grad_norm": 2.239792823791504, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.736092209815979, + "num_tokens": 928307386.0, + "step": 34548 + }, + { + "epoch": 2.8489321349055827, + "grad_norm": 2.0455539226531982, + "learning_rate": 1e-06, + "loss": 0.6873, + "mean_token_accuracy": 0.7769773006439209, + "num_tokens": 928336647.0, + "step": 34549 + }, + { + "epoch": 2.849014595530634, + "grad_norm": 1.9003506898880005, + "learning_rate": 1e-06, + "loss": 0.6462, + "mean_token_accuracy": 0.7846238613128662, + "num_tokens": 928371211.0, + "step": 34550 + }, + { + "epoch": 2.849097056155686, + "grad_norm": 2.2630419731140137, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7465206384658813, + "num_tokens": 928398749.0, + "step": 34551 + }, + { + "epoch": 2.849179516780737, + "grad_norm": 2.17602801322937, + "learning_rate": 1e-06, + "loss": 0.7027, + "mean_token_accuracy": 0.7793394923210144, + "num_tokens": 928426898.0, + "step": 34552 + }, + { + "epoch": 2.8492619774057886, + "grad_norm": 1.9862369298934937, + "learning_rate": 1e-06, + "loss": 0.6233, + "mean_token_accuracy": 0.7939614057540894, + "num_tokens": 928457983.0, + "step": 34553 + }, + { + "epoch": 2.8493444380308404, + "grad_norm": 1.9353142976760864, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7627200484275818, + "num_tokens": 928491157.0, + "step": 34554 + }, + { + "epoch": 2.8494268986558917, + "grad_norm": 2.024449110031128, + "learning_rate": 1e-06, + "loss": 0.6656, + "mean_token_accuracy": 0.7786245942115784, + "num_tokens": 928523383.0, + "step": 34555 + }, + { + "epoch": 2.8495093592809435, + "grad_norm": 2.2437052726745605, + "learning_rate": 1e-06, + "loss": 0.5863, + "mean_token_accuracy": 0.8088158369064331, + "num_tokens": 928546616.0, + "step": 34556 + }, + { + "epoch": 2.849591819905995, + "grad_norm": 1.962751030921936, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7561006546020508, + "num_tokens": 928579394.0, + "step": 34557 + }, + { + "epoch": 2.8496742805310467, + "grad_norm": 2.139521837234497, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7865530848503113, + "num_tokens": 928608745.0, + "step": 34558 + }, + { + "epoch": 2.849756741156098, + "grad_norm": 2.2733800411224365, + "learning_rate": 1e-06, + "loss": 0.6334, + "mean_token_accuracy": 0.7792083621025085, + "num_tokens": 928634214.0, + "step": 34559 + }, + { + "epoch": 2.8498392017811494, + "grad_norm": 2.0861222743988037, + "learning_rate": 1e-06, + "loss": 0.62, + "mean_token_accuracy": 0.802286684513092, + "num_tokens": 928662211.0, + "step": 34560 + }, + { + "epoch": 2.849921662406201, + "grad_norm": 1.7871276140213013, + "learning_rate": 1e-06, + "loss": 0.618, + "mean_token_accuracy": 0.7992044687271118, + "num_tokens": 928701817.0, + "step": 34561 + }, + { + "epoch": 2.8500041230312525, + "grad_norm": 2.258561134338379, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7407240867614746, + "num_tokens": 928729172.0, + "step": 34562 + }, + { + "epoch": 2.850086583656304, + "grad_norm": 1.9941389560699463, + "learning_rate": 1e-06, + "loss": 0.5422, + "mean_token_accuracy": 0.8108973503112793, + "num_tokens": 928758943.0, + "step": 34563 + }, + { + "epoch": 2.8501690442813556, + "grad_norm": 2.1762139797210693, + "learning_rate": 1e-06, + "loss": 0.717, + "mean_token_accuracy": 0.7657926082611084, + "num_tokens": 928786406.0, + "step": 34564 + }, + { + "epoch": 2.8502515049064074, + "grad_norm": 2.269059181213379, + "learning_rate": 1e-06, + "loss": 0.6232, + "mean_token_accuracy": 0.7918741703033447, + "num_tokens": 928812059.0, + "step": 34565 + }, + { + "epoch": 2.850333965531459, + "grad_norm": 1.9761329889297485, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7442512512207031, + "num_tokens": 928846294.0, + "step": 34566 + }, + { + "epoch": 2.85041642615651, + "grad_norm": 1.9626394510269165, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7584757804870605, + "num_tokens": 928879912.0, + "step": 34567 + }, + { + "epoch": 2.850498886781562, + "grad_norm": 1.9686179161071777, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7724241018295288, + "num_tokens": 928912730.0, + "step": 34568 + }, + { + "epoch": 2.8505813474066133, + "grad_norm": 1.9252920150756836, + "learning_rate": 1e-06, + "loss": 0.5874, + "mean_token_accuracy": 0.8072261214256287, + "num_tokens": 928943583.0, + "step": 34569 + }, + { + "epoch": 2.8506638080316646, + "grad_norm": 2.134760618209839, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7329769134521484, + "num_tokens": 928973548.0, + "step": 34570 + }, + { + "epoch": 2.8507462686567164, + "grad_norm": 2.408203601837158, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7234371304512024, + "num_tokens": 929000163.0, + "step": 34571 + }, + { + "epoch": 2.8508287292817682, + "grad_norm": 2.157761573791504, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7712458372116089, + "num_tokens": 929028839.0, + "step": 34572 + }, + { + "epoch": 2.8509111899068196, + "grad_norm": 2.1949806213378906, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.7839050889015198, + "num_tokens": 929055099.0, + "step": 34573 + }, + { + "epoch": 2.850993650531871, + "grad_norm": 1.854413390159607, + "learning_rate": 1e-06, + "loss": 0.6849, + "mean_token_accuracy": 0.7870807647705078, + "num_tokens": 929096493.0, + "step": 34574 + }, + { + "epoch": 2.8510761111569227, + "grad_norm": 1.975873589515686, + "learning_rate": 1e-06, + "loss": 0.6313, + "mean_token_accuracy": 0.7859028577804565, + "num_tokens": 929129342.0, + "step": 34575 + }, + { + "epoch": 2.851158571781974, + "grad_norm": 2.2786195278167725, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7687005996704102, + "num_tokens": 929155551.0, + "step": 34576 + }, + { + "epoch": 2.8512410324070254, + "grad_norm": 2.09494686126709, + "learning_rate": 1e-06, + "loss": 0.7673, + "mean_token_accuracy": 0.7463621497154236, + "num_tokens": 929187047.0, + "step": 34577 + }, + { + "epoch": 2.851323493032077, + "grad_norm": 1.9586212635040283, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.750524640083313, + "num_tokens": 929222014.0, + "step": 34578 + }, + { + "epoch": 2.8514059536571286, + "grad_norm": 2.0043163299560547, + "learning_rate": 1e-06, + "loss": 0.6885, + "mean_token_accuracy": 0.774978756904602, + "num_tokens": 929254538.0, + "step": 34579 + }, + { + "epoch": 2.8514884142821804, + "grad_norm": 2.1907739639282227, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7659738063812256, + "num_tokens": 929282174.0, + "step": 34580 + }, + { + "epoch": 2.8515708749072317, + "grad_norm": 2.4259634017944336, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7623624801635742, + "num_tokens": 929309761.0, + "step": 34581 + }, + { + "epoch": 2.8516533355322835, + "grad_norm": 2.268807888031006, + "learning_rate": 1e-06, + "loss": 0.5976, + "mean_token_accuracy": 0.8023664951324463, + "num_tokens": 929334970.0, + "step": 34582 + }, + { + "epoch": 2.851735796157335, + "grad_norm": 2.216230630874634, + "learning_rate": 1e-06, + "loss": 0.6757, + "mean_token_accuracy": 0.7755930423736572, + "num_tokens": 929362581.0, + "step": 34583 + }, + { + "epoch": 2.851818256782386, + "grad_norm": 2.174307346343994, + "learning_rate": 1e-06, + "loss": 0.5437, + "mean_token_accuracy": 0.8157339692115784, + "num_tokens": 929390717.0, + "step": 34584 + }, + { + "epoch": 2.851900717407438, + "grad_norm": 2.084383010864258, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7357785701751709, + "num_tokens": 929423065.0, + "step": 34585 + }, + { + "epoch": 2.8519831780324894, + "grad_norm": 2.193531036376953, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7404129505157471, + "num_tokens": 929453125.0, + "step": 34586 + }, + { + "epoch": 2.852065638657541, + "grad_norm": 2.0269341468811035, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7628628015518188, + "num_tokens": 929488108.0, + "step": 34587 + }, + { + "epoch": 2.8521480992825925, + "grad_norm": 1.7000640630722046, + "learning_rate": 1e-06, + "loss": 0.6154, + "mean_token_accuracy": 0.792957067489624, + "num_tokens": 929529094.0, + "step": 34588 + }, + { + "epoch": 2.8522305599076443, + "grad_norm": 2.153390645980835, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7767549753189087, + "num_tokens": 929558344.0, + "step": 34589 + }, + { + "epoch": 2.8523130205326956, + "grad_norm": 2.196326732635498, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7509511113166809, + "num_tokens": 929583810.0, + "step": 34590 + }, + { + "epoch": 2.852395481157747, + "grad_norm": 2.2333130836486816, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.7873114347457886, + "num_tokens": 929609175.0, + "step": 34591 + }, + { + "epoch": 2.852477941782799, + "grad_norm": 1.8076547384262085, + "learning_rate": 1e-06, + "loss": 0.5637, + "mean_token_accuracy": 0.8161290884017944, + "num_tokens": 929646513.0, + "step": 34592 + }, + { + "epoch": 2.85256040240785, + "grad_norm": 2.0665836334228516, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7689526677131653, + "num_tokens": 929677870.0, + "step": 34593 + }, + { + "epoch": 2.852642863032902, + "grad_norm": 2.3263962268829346, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7536166906356812, + "num_tokens": 929704636.0, + "step": 34594 + }, + { + "epoch": 2.8527253236579533, + "grad_norm": 2.3815765380859375, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7647608518600464, + "num_tokens": 929730323.0, + "step": 34595 + }, + { + "epoch": 2.852807784283005, + "grad_norm": 2.3791608810424805, + "learning_rate": 1e-06, + "loss": 0.6253, + "mean_token_accuracy": 0.7892874479293823, + "num_tokens": 929755707.0, + "step": 34596 + }, + { + "epoch": 2.8528902449080564, + "grad_norm": 2.050379991531372, + "learning_rate": 1e-06, + "loss": 0.6141, + "mean_token_accuracy": 0.7993083596229553, + "num_tokens": 929784951.0, + "step": 34597 + }, + { + "epoch": 2.8529727055331078, + "grad_norm": 2.011505365371704, + "learning_rate": 1e-06, + "loss": 0.7166, + "mean_token_accuracy": 0.7719616889953613, + "num_tokens": 929818653.0, + "step": 34598 + }, + { + "epoch": 2.8530551661581596, + "grad_norm": 1.796697735786438, + "learning_rate": 1e-06, + "loss": 0.4769, + "mean_token_accuracy": 0.8299856185913086, + "num_tokens": 929852781.0, + "step": 34599 + }, + { + "epoch": 2.853137626783211, + "grad_norm": 2.0108513832092285, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7613462805747986, + "num_tokens": 929883542.0, + "step": 34600 + }, + { + "epoch": 2.8532200874082627, + "grad_norm": 2.074564218521118, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.7875511646270752, + "num_tokens": 929913952.0, + "step": 34601 + }, + { + "epoch": 2.853302548033314, + "grad_norm": 2.3144190311431885, + "learning_rate": 1e-06, + "loss": 0.7603, + "mean_token_accuracy": 0.7501649856567383, + "num_tokens": 929940206.0, + "step": 34602 + }, + { + "epoch": 2.853385008658366, + "grad_norm": 1.8028978109359741, + "learning_rate": 1e-06, + "loss": 0.5487, + "mean_token_accuracy": 0.8121045827865601, + "num_tokens": 929976180.0, + "step": 34603 + }, + { + "epoch": 2.853467469283417, + "grad_norm": 2.032978057861328, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.7934055328369141, + "num_tokens": 930007082.0, + "step": 34604 + }, + { + "epoch": 2.8535499299084686, + "grad_norm": 2.2092230319976807, + "learning_rate": 1e-06, + "loss": 0.7455, + "mean_token_accuracy": 0.7626790404319763, + "num_tokens": 930031613.0, + "step": 34605 + }, + { + "epoch": 2.8536323905335204, + "grad_norm": 2.2448573112487793, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7284963726997375, + "num_tokens": 930060951.0, + "step": 34606 + }, + { + "epoch": 2.8537148511585717, + "grad_norm": 1.9943759441375732, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7798851728439331, + "num_tokens": 930094053.0, + "step": 34607 + }, + { + "epoch": 2.853797311783623, + "grad_norm": 1.9049981832504272, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.7989951968193054, + "num_tokens": 930130018.0, + "step": 34608 + }, + { + "epoch": 2.853879772408675, + "grad_norm": 2.2542548179626465, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7433010339736938, + "num_tokens": 930157992.0, + "step": 34609 + }, + { + "epoch": 2.8539622330337266, + "grad_norm": 2.0555179119110107, + "learning_rate": 1e-06, + "loss": 0.6842, + "mean_token_accuracy": 0.78409743309021, + "num_tokens": 930187863.0, + "step": 34610 + }, + { + "epoch": 2.854044693658778, + "grad_norm": 2.094658851623535, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7700769901275635, + "num_tokens": 930217277.0, + "step": 34611 + }, + { + "epoch": 2.8541271542838293, + "grad_norm": 2.0390000343322754, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7816194295883179, + "num_tokens": 930247917.0, + "step": 34612 + }, + { + "epoch": 2.854209614908881, + "grad_norm": 1.912602424621582, + "learning_rate": 1e-06, + "loss": 0.6661, + "mean_token_accuracy": 0.7812165021896362, + "num_tokens": 930285658.0, + "step": 34613 + }, + { + "epoch": 2.8542920755339325, + "grad_norm": 2.935997247695923, + "learning_rate": 1e-06, + "loss": 0.8084, + "mean_token_accuracy": 0.7445744276046753, + "num_tokens": 930305390.0, + "step": 34614 + }, + { + "epoch": 2.854374536158984, + "grad_norm": 2.287546396255493, + "learning_rate": 1e-06, + "loss": 0.7518, + "mean_token_accuracy": 0.7693915367126465, + "num_tokens": 930329970.0, + "step": 34615 + }, + { + "epoch": 2.8544569967840356, + "grad_norm": 1.993180751800537, + "learning_rate": 1e-06, + "loss": 0.7438, + "mean_token_accuracy": 0.7584536075592041, + "num_tokens": 930363340.0, + "step": 34616 + }, + { + "epoch": 2.8545394574090874, + "grad_norm": 1.977123498916626, + "learning_rate": 1e-06, + "loss": 0.8033, + "mean_token_accuracy": 0.7352418899536133, + "num_tokens": 930397591.0, + "step": 34617 + }, + { + "epoch": 2.854621918034139, + "grad_norm": 2.106125593185425, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7775179147720337, + "num_tokens": 930426662.0, + "step": 34618 + }, + { + "epoch": 2.85470437865919, + "grad_norm": 1.8758236169815063, + "learning_rate": 1e-06, + "loss": 0.5259, + "mean_token_accuracy": 0.8229015469551086, + "num_tokens": 930457607.0, + "step": 34619 + }, + { + "epoch": 2.854786839284242, + "grad_norm": 2.1372318267822266, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.7849425673484802, + "num_tokens": 930487321.0, + "step": 34620 + }, + { + "epoch": 2.8548692999092933, + "grad_norm": 2.094712257385254, + "learning_rate": 1e-06, + "loss": 0.6277, + "mean_token_accuracy": 0.7865056395530701, + "num_tokens": 930514763.0, + "step": 34621 + }, + { + "epoch": 2.8549517605343446, + "grad_norm": 2.2459752559661865, + "learning_rate": 1e-06, + "loss": 0.6189, + "mean_token_accuracy": 0.788996160030365, + "num_tokens": 930541425.0, + "step": 34622 + }, + { + "epoch": 2.8550342211593964, + "grad_norm": 2.083336114883423, + "learning_rate": 1e-06, + "loss": 0.7357, + "mean_token_accuracy": 0.7673034071922302, + "num_tokens": 930571328.0, + "step": 34623 + }, + { + "epoch": 2.8551166817844478, + "grad_norm": 2.036787271499634, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.790401816368103, + "num_tokens": 930601721.0, + "step": 34624 + }, + { + "epoch": 2.8551991424094996, + "grad_norm": 2.4023807048797607, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7391230463981628, + "num_tokens": 930627534.0, + "step": 34625 + }, + { + "epoch": 2.855281603034551, + "grad_norm": 2.305624485015869, + "learning_rate": 1e-06, + "loss": 0.7707, + "mean_token_accuracy": 0.7485226988792419, + "num_tokens": 930654987.0, + "step": 34626 + }, + { + "epoch": 2.8553640636596027, + "grad_norm": 2.0874667167663574, + "learning_rate": 1e-06, + "loss": 0.6081, + "mean_token_accuracy": 0.7942476272583008, + "num_tokens": 930682677.0, + "step": 34627 + }, + { + "epoch": 2.855446524284654, + "grad_norm": 2.029902219772339, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7737032175064087, + "num_tokens": 930715139.0, + "step": 34628 + }, + { + "epoch": 2.8555289849097054, + "grad_norm": 2.3356828689575195, + "learning_rate": 1e-06, + "loss": 0.6222, + "mean_token_accuracy": 0.7934926748275757, + "num_tokens": 930740488.0, + "step": 34629 + }, + { + "epoch": 2.855611445534757, + "grad_norm": 2.241586208343506, + "learning_rate": 1e-06, + "loss": 0.6238, + "mean_token_accuracy": 0.7988527417182922, + "num_tokens": 930765200.0, + "step": 34630 + }, + { + "epoch": 2.8556939061598086, + "grad_norm": 2.2080817222595215, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7659752368927002, + "num_tokens": 930794383.0, + "step": 34631 + }, + { + "epoch": 2.8557763667848604, + "grad_norm": 1.9206469058990479, + "learning_rate": 1e-06, + "loss": 0.5227, + "mean_token_accuracy": 0.8162726163864136, + "num_tokens": 930829764.0, + "step": 34632 + }, + { + "epoch": 2.8558588274099117, + "grad_norm": 2.243438959121704, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.7779484987258911, + "num_tokens": 930858141.0, + "step": 34633 + }, + { + "epoch": 2.8559412880349635, + "grad_norm": 2.039324998855591, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7779979705810547, + "num_tokens": 930890415.0, + "step": 34634 + }, + { + "epoch": 2.856023748660015, + "grad_norm": 2.034651517868042, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7721028327941895, + "num_tokens": 930919808.0, + "step": 34635 + }, + { + "epoch": 2.856106209285066, + "grad_norm": 2.358274459838867, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7724872827529907, + "num_tokens": 930944278.0, + "step": 34636 + }, + { + "epoch": 2.856188669910118, + "grad_norm": 2.109952449798584, + "learning_rate": 1e-06, + "loss": 0.7044, + "mean_token_accuracy": 0.7673506736755371, + "num_tokens": 930974627.0, + "step": 34637 + }, + { + "epoch": 2.8562711305351693, + "grad_norm": 2.080855131149292, + "learning_rate": 1e-06, + "loss": 0.7281, + "mean_token_accuracy": 0.7631394267082214, + "num_tokens": 931005455.0, + "step": 34638 + }, + { + "epoch": 2.856353591160221, + "grad_norm": 2.3482308387756348, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7366893887519836, + "num_tokens": 931032992.0, + "step": 34639 + }, + { + "epoch": 2.8564360517852725, + "grad_norm": 1.9758201837539673, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7333208322525024, + "num_tokens": 931064748.0, + "step": 34640 + }, + { + "epoch": 2.8565185124103243, + "grad_norm": 1.9782153367996216, + "learning_rate": 1e-06, + "loss": 0.7546, + "mean_token_accuracy": 0.7635315656661987, + "num_tokens": 931097332.0, + "step": 34641 + }, + { + "epoch": 2.8566009730353756, + "grad_norm": 2.388375997543335, + "learning_rate": 1e-06, + "loss": 0.65, + "mean_token_accuracy": 0.7728328108787537, + "num_tokens": 931121987.0, + "step": 34642 + }, + { + "epoch": 2.856683433660427, + "grad_norm": 2.2557475566864014, + "learning_rate": 1e-06, + "loss": 0.5881, + "mean_token_accuracy": 0.7985715270042419, + "num_tokens": 931146512.0, + "step": 34643 + }, + { + "epoch": 2.856765894285479, + "grad_norm": 2.32656192779541, + "learning_rate": 1e-06, + "loss": 0.7374, + "mean_token_accuracy": 0.7609157562255859, + "num_tokens": 931172803.0, + "step": 34644 + }, + { + "epoch": 2.85684835491053, + "grad_norm": 2.2280945777893066, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7538110613822937, + "num_tokens": 931202668.0, + "step": 34645 + }, + { + "epoch": 2.856930815535582, + "grad_norm": 2.2482681274414062, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.7863652110099792, + "num_tokens": 931228618.0, + "step": 34646 + }, + { + "epoch": 2.8570132761606333, + "grad_norm": 1.9730756282806396, + "learning_rate": 1e-06, + "loss": 0.5824, + "mean_token_accuracy": 0.8032784461975098, + "num_tokens": 931259973.0, + "step": 34647 + }, + { + "epoch": 2.857095736785685, + "grad_norm": 2.233456611633301, + "learning_rate": 1e-06, + "loss": 0.7654, + "mean_token_accuracy": 0.7575666904449463, + "num_tokens": 931287047.0, + "step": 34648 + }, + { + "epoch": 2.8571781974107364, + "grad_norm": 2.1700828075408936, + "learning_rate": 1e-06, + "loss": 0.5363, + "mean_token_accuracy": 0.8162360191345215, + "num_tokens": 931316863.0, + "step": 34649 + }, + { + "epoch": 2.8572606580357878, + "grad_norm": 2.294262170791626, + "learning_rate": 1e-06, + "loss": 0.7046, + "mean_token_accuracy": 0.7719497680664062, + "num_tokens": 931342915.0, + "step": 34650 + }, + { + "epoch": 2.8573431186608396, + "grad_norm": 2.2098875045776367, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7381653785705566, + "num_tokens": 931371497.0, + "step": 34651 + }, + { + "epoch": 2.857425579285891, + "grad_norm": 1.9012398719787598, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.7877762913703918, + "num_tokens": 931406198.0, + "step": 34652 + }, + { + "epoch": 2.8575080399109423, + "grad_norm": 2.0842273235321045, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7710980176925659, + "num_tokens": 931437074.0, + "step": 34653 + }, + { + "epoch": 2.857590500535994, + "grad_norm": 2.1458921432495117, + "learning_rate": 1e-06, + "loss": 0.7309, + "mean_token_accuracy": 0.766372799873352, + "num_tokens": 931468704.0, + "step": 34654 + }, + { + "epoch": 2.857672961161046, + "grad_norm": 1.8530381917953491, + "learning_rate": 1e-06, + "loss": 0.6174, + "mean_token_accuracy": 0.8011733889579773, + "num_tokens": 931504282.0, + "step": 34655 + }, + { + "epoch": 2.857755421786097, + "grad_norm": 1.8683865070343018, + "learning_rate": 1e-06, + "loss": 0.5551, + "mean_token_accuracy": 0.8002319931983948, + "num_tokens": 931538091.0, + "step": 34656 + }, + { + "epoch": 2.8578378824111486, + "grad_norm": 2.0627360343933105, + "learning_rate": 1e-06, + "loss": 0.6319, + "mean_token_accuracy": 0.7907192707061768, + "num_tokens": 931567541.0, + "step": 34657 + }, + { + "epoch": 2.8579203430362004, + "grad_norm": 1.8916428089141846, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.7881349325180054, + "num_tokens": 931602512.0, + "step": 34658 + }, + { + "epoch": 2.8580028036612517, + "grad_norm": 2.1389312744140625, + "learning_rate": 1e-06, + "loss": 0.7151, + "mean_token_accuracy": 0.7692253589630127, + "num_tokens": 931633184.0, + "step": 34659 + }, + { + "epoch": 2.858085264286303, + "grad_norm": 1.8659179210662842, + "learning_rate": 1e-06, + "loss": 0.6418, + "mean_token_accuracy": 0.780786395072937, + "num_tokens": 931668287.0, + "step": 34660 + }, + { + "epoch": 2.858167724911355, + "grad_norm": 2.1248950958251953, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7666910886764526, + "num_tokens": 931697733.0, + "step": 34661 + }, + { + "epoch": 2.8582501855364066, + "grad_norm": 2.0178534984588623, + "learning_rate": 1e-06, + "loss": 0.6971, + "mean_token_accuracy": 0.7752196788787842, + "num_tokens": 931726810.0, + "step": 34662 + }, + { + "epoch": 2.858332646161458, + "grad_norm": 1.9625520706176758, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7778244018554688, + "num_tokens": 931759226.0, + "step": 34663 + }, + { + "epoch": 2.8584151067865093, + "grad_norm": 2.2292563915252686, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7440873384475708, + "num_tokens": 931789342.0, + "step": 34664 + }, + { + "epoch": 2.858497567411561, + "grad_norm": 2.1142828464508057, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7805180549621582, + "num_tokens": 931818637.0, + "step": 34665 + }, + { + "epoch": 2.8585800280366125, + "grad_norm": 2.2049601078033447, + "learning_rate": 1e-06, + "loss": 0.6996, + "mean_token_accuracy": 0.7668108344078064, + "num_tokens": 931848826.0, + "step": 34666 + }, + { + "epoch": 2.858662488661664, + "grad_norm": 2.301344633102417, + "learning_rate": 1e-06, + "loss": 0.6006, + "mean_token_accuracy": 0.8029130697250366, + "num_tokens": 931872438.0, + "step": 34667 + }, + { + "epoch": 2.8587449492867156, + "grad_norm": 2.3223350048065186, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7574132084846497, + "num_tokens": 931897952.0, + "step": 34668 + }, + { + "epoch": 2.858827409911767, + "grad_norm": 2.4069457054138184, + "learning_rate": 1e-06, + "loss": 0.7477, + "mean_token_accuracy": 0.7727770209312439, + "num_tokens": 931922148.0, + "step": 34669 + }, + { + "epoch": 2.8589098705368188, + "grad_norm": 2.0807101726531982, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7762627601623535, + "num_tokens": 931953057.0, + "step": 34670 + }, + { + "epoch": 2.85899233116187, + "grad_norm": 2.207448720932007, + "learning_rate": 1e-06, + "loss": 0.6376, + "mean_token_accuracy": 0.7905372381210327, + "num_tokens": 931979966.0, + "step": 34671 + }, + { + "epoch": 2.859074791786922, + "grad_norm": 2.262162685394287, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7470537424087524, + "num_tokens": 932006732.0, + "step": 34672 + }, + { + "epoch": 2.8591572524119733, + "grad_norm": 1.9762951135635376, + "learning_rate": 1e-06, + "loss": 0.6837, + "mean_token_accuracy": 0.7808824181556702, + "num_tokens": 932040496.0, + "step": 34673 + }, + { + "epoch": 2.8592397130370246, + "grad_norm": 1.9621903896331787, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.7963467836380005, + "num_tokens": 932071716.0, + "step": 34674 + }, + { + "epoch": 2.8593221736620764, + "grad_norm": 1.9864718914031982, + "learning_rate": 1e-06, + "loss": 0.6895, + "mean_token_accuracy": 0.766700267791748, + "num_tokens": 932107302.0, + "step": 34675 + }, + { + "epoch": 2.8594046342871278, + "grad_norm": 1.9743670225143433, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7685294151306152, + "num_tokens": 932142090.0, + "step": 34676 + }, + { + "epoch": 2.8594870949121796, + "grad_norm": 2.211833953857422, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7400309443473816, + "num_tokens": 932174819.0, + "step": 34677 + }, + { + "epoch": 2.859569555537231, + "grad_norm": 2.2687766551971436, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.7644466161727905, + "num_tokens": 932202076.0, + "step": 34678 + }, + { + "epoch": 2.8596520161622827, + "grad_norm": 2.1225461959838867, + "learning_rate": 1e-06, + "loss": 0.7282, + "mean_token_accuracy": 0.7600336670875549, + "num_tokens": 932231753.0, + "step": 34679 + }, + { + "epoch": 2.859734476787334, + "grad_norm": 1.83037531375885, + "learning_rate": 1e-06, + "loss": 0.6811, + "mean_token_accuracy": 0.7708114385604858, + "num_tokens": 932268218.0, + "step": 34680 + }, + { + "epoch": 2.8598169374123854, + "grad_norm": 2.2220470905303955, + "learning_rate": 1e-06, + "loss": 0.6224, + "mean_token_accuracy": 0.78252774477005, + "num_tokens": 932299167.0, + "step": 34681 + }, + { + "epoch": 2.859899398037437, + "grad_norm": 2.27815842628479, + "learning_rate": 1e-06, + "loss": 0.7085, + "mean_token_accuracy": 0.7747124433517456, + "num_tokens": 932327741.0, + "step": 34682 + }, + { + "epoch": 2.8599818586624886, + "grad_norm": 1.9190834760665894, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.7678197026252747, + "num_tokens": 932364752.0, + "step": 34683 + }, + { + "epoch": 2.8600643192875403, + "grad_norm": 2.006753921508789, + "learning_rate": 1e-06, + "loss": 0.7363, + "mean_token_accuracy": 0.7738368511199951, + "num_tokens": 932397562.0, + "step": 34684 + }, + { + "epoch": 2.8601467799125917, + "grad_norm": 2.100663423538208, + "learning_rate": 1e-06, + "loss": 0.7048, + "mean_token_accuracy": 0.7691602110862732, + "num_tokens": 932429025.0, + "step": 34685 + }, + { + "epoch": 2.8602292405376435, + "grad_norm": 2.0684866905212402, + "learning_rate": 1e-06, + "loss": 0.6639, + "mean_token_accuracy": 0.7866784930229187, + "num_tokens": 932462143.0, + "step": 34686 + }, + { + "epoch": 2.860311701162695, + "grad_norm": 1.9605906009674072, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7573079466819763, + "num_tokens": 932494842.0, + "step": 34687 + }, + { + "epoch": 2.860394161787746, + "grad_norm": 2.1240694522857666, + "learning_rate": 1e-06, + "loss": 0.6891, + "mean_token_accuracy": 0.7727019786834717, + "num_tokens": 932524090.0, + "step": 34688 + }, + { + "epoch": 2.860476622412798, + "grad_norm": 2.0248947143554688, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7547316551208496, + "num_tokens": 932557441.0, + "step": 34689 + }, + { + "epoch": 2.8605590830378493, + "grad_norm": 2.1291308403015137, + "learning_rate": 1e-06, + "loss": 0.7058, + "mean_token_accuracy": 0.7654066681861877, + "num_tokens": 932583058.0, + "step": 34690 + }, + { + "epoch": 2.8606415436629007, + "grad_norm": 2.1372368335723877, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.757628321647644, + "num_tokens": 932611823.0, + "step": 34691 + }, + { + "epoch": 2.8607240042879525, + "grad_norm": 1.9931455850601196, + "learning_rate": 1e-06, + "loss": 0.5455, + "mean_token_accuracy": 0.8054923415184021, + "num_tokens": 932644003.0, + "step": 34692 + }, + { + "epoch": 2.8608064649130043, + "grad_norm": 2.0106213092803955, + "learning_rate": 1e-06, + "loss": 0.6954, + "mean_token_accuracy": 0.7729849815368652, + "num_tokens": 932675326.0, + "step": 34693 + }, + { + "epoch": 2.8608889255380556, + "grad_norm": 2.0233399868011475, + "learning_rate": 1e-06, + "loss": 0.6708, + "mean_token_accuracy": 0.7744917869567871, + "num_tokens": 932710229.0, + "step": 34694 + }, + { + "epoch": 2.860971386163107, + "grad_norm": 2.2198448181152344, + "learning_rate": 1e-06, + "loss": 0.7974, + "mean_token_accuracy": 0.752473771572113, + "num_tokens": 932738741.0, + "step": 34695 + }, + { + "epoch": 2.8610538467881588, + "grad_norm": 2.5208253860473633, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.7766016721725464, + "num_tokens": 932760232.0, + "step": 34696 + }, + { + "epoch": 2.86113630741321, + "grad_norm": 2.1164355278015137, + "learning_rate": 1e-06, + "loss": 0.6929, + "mean_token_accuracy": 0.789982795715332, + "num_tokens": 932788082.0, + "step": 34697 + }, + { + "epoch": 2.8612187680382615, + "grad_norm": 2.2358450889587402, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7421363592147827, + "num_tokens": 932815231.0, + "step": 34698 + }, + { + "epoch": 2.8613012286633133, + "grad_norm": 1.951319694519043, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.7680366039276123, + "num_tokens": 932847768.0, + "step": 34699 + }, + { + "epoch": 2.861383689288365, + "grad_norm": 1.9026198387145996, + "learning_rate": 1e-06, + "loss": 0.5538, + "mean_token_accuracy": 0.8055422902107239, + "num_tokens": 932878839.0, + "step": 34700 + }, + { + "epoch": 2.8614661499134164, + "grad_norm": 2.1721067428588867, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7799650430679321, + "num_tokens": 932907816.0, + "step": 34701 + }, + { + "epoch": 2.8615486105384678, + "grad_norm": 2.058225631713867, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.7822401523590088, + "num_tokens": 932937884.0, + "step": 34702 + }, + { + "epoch": 2.8616310711635196, + "grad_norm": 2.246438980102539, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.7689089775085449, + "num_tokens": 932967895.0, + "step": 34703 + }, + { + "epoch": 2.861713531788571, + "grad_norm": 2.248641014099121, + "learning_rate": 1e-06, + "loss": 0.7376, + "mean_token_accuracy": 0.7568044066429138, + "num_tokens": 932993881.0, + "step": 34704 + }, + { + "epoch": 2.8617959924136223, + "grad_norm": 1.948262333869934, + "learning_rate": 1e-06, + "loss": 0.5124, + "mean_token_accuracy": 0.8253791332244873, + "num_tokens": 933025085.0, + "step": 34705 + }, + { + "epoch": 2.861878453038674, + "grad_norm": 1.9089864492416382, + "learning_rate": 1e-06, + "loss": 0.6963, + "mean_token_accuracy": 0.7750964760780334, + "num_tokens": 933059980.0, + "step": 34706 + }, + { + "epoch": 2.861960913663726, + "grad_norm": 2.4242749214172363, + "learning_rate": 1e-06, + "loss": 0.5896, + "mean_token_accuracy": 0.8006787896156311, + "num_tokens": 933082624.0, + "step": 34707 + }, + { + "epoch": 2.862043374288777, + "grad_norm": 2.1027956008911133, + "learning_rate": 1e-06, + "loss": 0.7464, + "mean_token_accuracy": 0.7595697045326233, + "num_tokens": 933114912.0, + "step": 34708 + }, + { + "epoch": 2.8621258349138285, + "grad_norm": 2.1952896118164062, + "learning_rate": 1e-06, + "loss": 0.7992, + "mean_token_accuracy": 0.759148895740509, + "num_tokens": 933142846.0, + "step": 34709 + }, + { + "epoch": 2.8622082955388803, + "grad_norm": 2.2611560821533203, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7654845118522644, + "num_tokens": 933167661.0, + "step": 34710 + }, + { + "epoch": 2.8622907561639317, + "grad_norm": 2.3223588466644287, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.784416675567627, + "num_tokens": 933191326.0, + "step": 34711 + }, + { + "epoch": 2.862373216788983, + "grad_norm": 1.839612603187561, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7725849151611328, + "num_tokens": 933230264.0, + "step": 34712 + }, + { + "epoch": 2.862455677414035, + "grad_norm": 2.0311057567596436, + "learning_rate": 1e-06, + "loss": 0.6007, + "mean_token_accuracy": 0.8009092807769775, + "num_tokens": 933259875.0, + "step": 34713 + }, + { + "epoch": 2.862538138039086, + "grad_norm": 2.0375373363494873, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7658705115318298, + "num_tokens": 933292290.0, + "step": 34714 + }, + { + "epoch": 2.862620598664138, + "grad_norm": 2.123526096343994, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7690144181251526, + "num_tokens": 933321069.0, + "step": 34715 + }, + { + "epoch": 2.8627030592891893, + "grad_norm": 2.2555553913116455, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7354638576507568, + "num_tokens": 933352975.0, + "step": 34716 + }, + { + "epoch": 2.862785519914241, + "grad_norm": 2.361558198928833, + "learning_rate": 1e-06, + "loss": 0.6949, + "mean_token_accuracy": 0.7729377746582031, + "num_tokens": 933376941.0, + "step": 34717 + }, + { + "epoch": 2.8628679805392925, + "grad_norm": 1.816803216934204, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7802548408508301, + "num_tokens": 933415648.0, + "step": 34718 + }, + { + "epoch": 2.862950441164344, + "grad_norm": 2.205299139022827, + "learning_rate": 1e-06, + "loss": 0.6264, + "mean_token_accuracy": 0.7954865097999573, + "num_tokens": 933445039.0, + "step": 34719 + }, + { + "epoch": 2.8630329017893956, + "grad_norm": 1.9259706735610962, + "learning_rate": 1e-06, + "loss": 0.676, + "mean_token_accuracy": 0.7843135595321655, + "num_tokens": 933478236.0, + "step": 34720 + }, + { + "epoch": 2.863115362414447, + "grad_norm": 2.0551626682281494, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.7872886061668396, + "num_tokens": 933510582.0, + "step": 34721 + }, + { + "epoch": 2.8631978230394988, + "grad_norm": 2.407014846801758, + "learning_rate": 1e-06, + "loss": 0.6964, + "mean_token_accuracy": 0.7709420919418335, + "num_tokens": 933533474.0, + "step": 34722 + }, + { + "epoch": 2.86328028366455, + "grad_norm": 1.776733160018921, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7737183570861816, + "num_tokens": 933572073.0, + "step": 34723 + }, + { + "epoch": 2.863362744289602, + "grad_norm": 2.497732162475586, + "learning_rate": 1e-06, + "loss": 0.5706, + "mean_token_accuracy": 0.8029416799545288, + "num_tokens": 933594016.0, + "step": 34724 + }, + { + "epoch": 2.8634452049146533, + "grad_norm": 2.0950663089752197, + "learning_rate": 1e-06, + "loss": 0.6579, + "mean_token_accuracy": 0.7877718210220337, + "num_tokens": 933621665.0, + "step": 34725 + }, + { + "epoch": 2.8635276655397046, + "grad_norm": 2.0890305042266846, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.7901384830474854, + "num_tokens": 933652651.0, + "step": 34726 + }, + { + "epoch": 2.8636101261647564, + "grad_norm": 2.063905715942383, + "learning_rate": 1e-06, + "loss": 0.7372, + "mean_token_accuracy": 0.7606139183044434, + "num_tokens": 933683926.0, + "step": 34727 + }, + { + "epoch": 2.8636925867898078, + "grad_norm": 2.1497714519500732, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7407844066619873, + "num_tokens": 933712630.0, + "step": 34728 + }, + { + "epoch": 2.8637750474148596, + "grad_norm": 2.047821044921875, + "learning_rate": 1e-06, + "loss": 0.7373, + "mean_token_accuracy": 0.7699573040008545, + "num_tokens": 933742255.0, + "step": 34729 + }, + { + "epoch": 2.863857508039911, + "grad_norm": 2.0104010105133057, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.7844093441963196, + "num_tokens": 933772033.0, + "step": 34730 + }, + { + "epoch": 2.8639399686649627, + "grad_norm": 2.0081019401550293, + "learning_rate": 1e-06, + "loss": 0.6941, + "mean_token_accuracy": 0.7775905728340149, + "num_tokens": 933803367.0, + "step": 34731 + }, + { + "epoch": 2.864022429290014, + "grad_norm": 2.06945538520813, + "learning_rate": 1e-06, + "loss": 0.677, + "mean_token_accuracy": 0.7770313024520874, + "num_tokens": 933833302.0, + "step": 34732 + }, + { + "epoch": 2.8641048899150654, + "grad_norm": 2.0301873683929443, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7488646507263184, + "num_tokens": 933866535.0, + "step": 34733 + }, + { + "epoch": 2.864187350540117, + "grad_norm": 1.9525810480117798, + "learning_rate": 1e-06, + "loss": 0.4695, + "mean_token_accuracy": 0.831957221031189, + "num_tokens": 933897868.0, + "step": 34734 + }, + { + "epoch": 2.8642698111651685, + "grad_norm": 1.8636444807052612, + "learning_rate": 1e-06, + "loss": 0.6676, + "mean_token_accuracy": 0.7692314386367798, + "num_tokens": 933936603.0, + "step": 34735 + }, + { + "epoch": 2.86435227179022, + "grad_norm": 2.247872829437256, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7599064111709595, + "num_tokens": 933965402.0, + "step": 34736 + }, + { + "epoch": 2.8644347324152717, + "grad_norm": 2.049083709716797, + "learning_rate": 1e-06, + "loss": 0.6047, + "mean_token_accuracy": 0.8005584478378296, + "num_tokens": 933996068.0, + "step": 34737 + }, + { + "epoch": 2.8645171930403235, + "grad_norm": 2.0160603523254395, + "learning_rate": 1e-06, + "loss": 0.4678, + "mean_token_accuracy": 0.8341591954231262, + "num_tokens": 934023866.0, + "step": 34738 + }, + { + "epoch": 2.864599653665375, + "grad_norm": 1.905651330947876, + "learning_rate": 1e-06, + "loss": 0.6122, + "mean_token_accuracy": 0.7936578989028931, + "num_tokens": 934057767.0, + "step": 34739 + }, + { + "epoch": 2.864682114290426, + "grad_norm": 2.040269374847412, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.7854526042938232, + "num_tokens": 934086076.0, + "step": 34740 + }, + { + "epoch": 2.864764574915478, + "grad_norm": 1.9847620725631714, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7690712809562683, + "num_tokens": 934118709.0, + "step": 34741 + }, + { + "epoch": 2.8648470355405293, + "grad_norm": 2.3698272705078125, + "learning_rate": 1e-06, + "loss": 0.6289, + "mean_token_accuracy": 0.7843800783157349, + "num_tokens": 934142800.0, + "step": 34742 + }, + { + "epoch": 2.8649294961655807, + "grad_norm": 1.9687433242797852, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7485460638999939, + "num_tokens": 934176640.0, + "step": 34743 + }, + { + "epoch": 2.8650119567906325, + "grad_norm": 2.22929048538208, + "learning_rate": 1e-06, + "loss": 0.7156, + "mean_token_accuracy": 0.7711330056190491, + "num_tokens": 934206220.0, + "step": 34744 + }, + { + "epoch": 2.8650944174156843, + "grad_norm": 1.9812499284744263, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7591871023178101, + "num_tokens": 934239519.0, + "step": 34745 + }, + { + "epoch": 2.8651768780407356, + "grad_norm": 2.667731285095215, + "learning_rate": 1e-06, + "loss": 0.7428, + "mean_token_accuracy": 0.7556319236755371, + "num_tokens": 934259885.0, + "step": 34746 + }, + { + "epoch": 2.865259338665787, + "grad_norm": 1.8713819980621338, + "learning_rate": 1e-06, + "loss": 0.6247, + "mean_token_accuracy": 0.789457380771637, + "num_tokens": 934294740.0, + "step": 34747 + }, + { + "epoch": 2.8653417992908388, + "grad_norm": 2.0045464038848877, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7810783982276917, + "num_tokens": 934326217.0, + "step": 34748 + }, + { + "epoch": 2.86542425991589, + "grad_norm": 1.979535460472107, + "learning_rate": 1e-06, + "loss": 0.7312, + "mean_token_accuracy": 0.7647271752357483, + "num_tokens": 934358014.0, + "step": 34749 + }, + { + "epoch": 2.8655067205409415, + "grad_norm": 2.751929998397827, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7630831003189087, + "num_tokens": 934377959.0, + "step": 34750 + }, + { + "epoch": 2.8655891811659933, + "grad_norm": 1.9086614847183228, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7617770433425903, + "num_tokens": 934413738.0, + "step": 34751 + }, + { + "epoch": 2.8656716417910446, + "grad_norm": 2.274590253829956, + "learning_rate": 1e-06, + "loss": 0.6909, + "mean_token_accuracy": 0.7617924809455872, + "num_tokens": 934440450.0, + "step": 34752 + }, + { + "epoch": 2.8657541024160964, + "grad_norm": 2.343549966812134, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7405557632446289, + "num_tokens": 934464877.0, + "step": 34753 + }, + { + "epoch": 2.8658365630411478, + "grad_norm": 2.338655471801758, + "learning_rate": 1e-06, + "loss": 0.7155, + "mean_token_accuracy": 0.7691943645477295, + "num_tokens": 934490158.0, + "step": 34754 + }, + { + "epoch": 2.8659190236661995, + "grad_norm": 2.1496777534484863, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.756895124912262, + "num_tokens": 934519443.0, + "step": 34755 + }, + { + "epoch": 2.866001484291251, + "grad_norm": 2.1529345512390137, + "learning_rate": 1e-06, + "loss": 0.7706, + "mean_token_accuracy": 0.7494280338287354, + "num_tokens": 934548069.0, + "step": 34756 + }, + { + "epoch": 2.8660839449163023, + "grad_norm": 1.8155728578567505, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7472423315048218, + "num_tokens": 934588181.0, + "step": 34757 + }, + { + "epoch": 2.866166405541354, + "grad_norm": 2.0459325313568115, + "learning_rate": 1e-06, + "loss": 0.6165, + "mean_token_accuracy": 0.7922695875167847, + "num_tokens": 934620541.0, + "step": 34758 + }, + { + "epoch": 2.8662488661664054, + "grad_norm": 2.0341784954071045, + "learning_rate": 1e-06, + "loss": 0.6922, + "mean_token_accuracy": 0.7857046127319336, + "num_tokens": 934652565.0, + "step": 34759 + }, + { + "epoch": 2.866331326791457, + "grad_norm": 2.284047842025757, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.7919513583183289, + "num_tokens": 934676573.0, + "step": 34760 + }, + { + "epoch": 2.8664137874165085, + "grad_norm": 1.8866316080093384, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7474259734153748, + "num_tokens": 934712997.0, + "step": 34761 + }, + { + "epoch": 2.8664962480415603, + "grad_norm": 1.9175009727478027, + "learning_rate": 1e-06, + "loss": 0.6739, + "mean_token_accuracy": 0.768700122833252, + "num_tokens": 934744531.0, + "step": 34762 + }, + { + "epoch": 2.8665787086666117, + "grad_norm": 2.313847541809082, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7799689769744873, + "num_tokens": 934769769.0, + "step": 34763 + }, + { + "epoch": 2.866661169291663, + "grad_norm": 2.1159403324127197, + "learning_rate": 1e-06, + "loss": 0.6049, + "mean_token_accuracy": 0.788326621055603, + "num_tokens": 934797796.0, + "step": 34764 + }, + { + "epoch": 2.866743629916715, + "grad_norm": 2.1200015544891357, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7274662256240845, + "num_tokens": 934830106.0, + "step": 34765 + }, + { + "epoch": 2.866826090541766, + "grad_norm": 1.8027987480163574, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7780340909957886, + "num_tokens": 934868562.0, + "step": 34766 + }, + { + "epoch": 2.866908551166818, + "grad_norm": 2.1932365894317627, + "learning_rate": 1e-06, + "loss": 0.7253, + "mean_token_accuracy": 0.75697922706604, + "num_tokens": 934896206.0, + "step": 34767 + }, + { + "epoch": 2.8669910117918693, + "grad_norm": 1.8021506071090698, + "learning_rate": 1e-06, + "loss": 0.5618, + "mean_token_accuracy": 0.8123682141304016, + "num_tokens": 934933013.0, + "step": 34768 + }, + { + "epoch": 2.867073472416921, + "grad_norm": 2.2172625064849854, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.742072582244873, + "num_tokens": 934960374.0, + "step": 34769 + }, + { + "epoch": 2.8671559330419725, + "grad_norm": 2.303156614303589, + "learning_rate": 1e-06, + "loss": 0.7008, + "mean_token_accuracy": 0.7711039781570435, + "num_tokens": 934986878.0, + "step": 34770 + }, + { + "epoch": 2.867238393667024, + "grad_norm": 2.0096683502197266, + "learning_rate": 1e-06, + "loss": 0.6136, + "mean_token_accuracy": 0.8042209148406982, + "num_tokens": 935020342.0, + "step": 34771 + }, + { + "epoch": 2.8673208542920756, + "grad_norm": 2.538296699523926, + "learning_rate": 1e-06, + "loss": 0.56, + "mean_token_accuracy": 0.7993736267089844, + "num_tokens": 935045952.0, + "step": 34772 + }, + { + "epoch": 2.867403314917127, + "grad_norm": 2.133599281311035, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7616465091705322, + "num_tokens": 935076459.0, + "step": 34773 + }, + { + "epoch": 2.8674857755421788, + "grad_norm": 2.033708095550537, + "learning_rate": 1e-06, + "loss": 0.6711, + "mean_token_accuracy": 0.7888633012771606, + "num_tokens": 935107896.0, + "step": 34774 + }, + { + "epoch": 2.86756823616723, + "grad_norm": 2.3209807872772217, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7766929864883423, + "num_tokens": 935131669.0, + "step": 34775 + }, + { + "epoch": 2.867650696792282, + "grad_norm": 2.1113052368164062, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.788080096244812, + "num_tokens": 935159566.0, + "step": 34776 + }, + { + "epoch": 2.8677331574173333, + "grad_norm": 2.03300142288208, + "learning_rate": 1e-06, + "loss": 0.6116, + "mean_token_accuracy": 0.7897775173187256, + "num_tokens": 935189556.0, + "step": 34777 + }, + { + "epoch": 2.8678156180423846, + "grad_norm": 1.8902668952941895, + "learning_rate": 1e-06, + "loss": 0.5969, + "mean_token_accuracy": 0.7914198040962219, + "num_tokens": 935224746.0, + "step": 34778 + }, + { + "epoch": 2.8678980786674364, + "grad_norm": 2.368638753890991, + "learning_rate": 1e-06, + "loss": 0.6659, + "mean_token_accuracy": 0.7806216478347778, + "num_tokens": 935247141.0, + "step": 34779 + }, + { + "epoch": 2.8679805392924878, + "grad_norm": 2.334984540939331, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7503490447998047, + "num_tokens": 935273217.0, + "step": 34780 + }, + { + "epoch": 2.868062999917539, + "grad_norm": 2.288329839706421, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7416782379150391, + "num_tokens": 935301168.0, + "step": 34781 + }, + { + "epoch": 2.868145460542591, + "grad_norm": 2.0619430541992188, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7799551486968994, + "num_tokens": 935331506.0, + "step": 34782 + }, + { + "epoch": 2.8682279211676427, + "grad_norm": 2.510180950164795, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7736964225769043, + "num_tokens": 935353196.0, + "step": 34783 + }, + { + "epoch": 2.868310381792694, + "grad_norm": 2.209070920944214, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.744481086730957, + "num_tokens": 935383438.0, + "step": 34784 + }, + { + "epoch": 2.8683928424177454, + "grad_norm": 2.3234214782714844, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7559424638748169, + "num_tokens": 935409558.0, + "step": 34785 + }, + { + "epoch": 2.868475303042797, + "grad_norm": 2.0980255603790283, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.7735636234283447, + "num_tokens": 935437845.0, + "step": 34786 + }, + { + "epoch": 2.8685577636678485, + "grad_norm": 1.9084820747375488, + "learning_rate": 1e-06, + "loss": 0.5366, + "mean_token_accuracy": 0.8171143531799316, + "num_tokens": 935472618.0, + "step": 34787 + }, + { + "epoch": 2.8686402242929, + "grad_norm": 2.287161111831665, + "learning_rate": 1e-06, + "loss": 0.6704, + "mean_token_accuracy": 0.7781535983085632, + "num_tokens": 935499663.0, + "step": 34788 + }, + { + "epoch": 2.8687226849179517, + "grad_norm": 1.8951510190963745, + "learning_rate": 1e-06, + "loss": 0.6, + "mean_token_accuracy": 0.7992485761642456, + "num_tokens": 935534868.0, + "step": 34789 + }, + { + "epoch": 2.8688051455430035, + "grad_norm": 2.140556573867798, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7393184304237366, + "num_tokens": 935565587.0, + "step": 34790 + }, + { + "epoch": 2.868887606168055, + "grad_norm": 2.1923065185546875, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.7698514461517334, + "num_tokens": 935593021.0, + "step": 34791 + }, + { + "epoch": 2.868970066793106, + "grad_norm": 2.410120725631714, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7372365593910217, + "num_tokens": 935615083.0, + "step": 34792 + }, + { + "epoch": 2.869052527418158, + "grad_norm": 2.0048606395721436, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7478574514389038, + "num_tokens": 935648476.0, + "step": 34793 + }, + { + "epoch": 2.8691349880432093, + "grad_norm": 2.047258138656616, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7455557584762573, + "num_tokens": 935681235.0, + "step": 34794 + }, + { + "epoch": 2.8692174486682607, + "grad_norm": 1.9172831773757935, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7587568759918213, + "num_tokens": 935714549.0, + "step": 34795 + }, + { + "epoch": 2.8692999092933125, + "grad_norm": 2.02864933013916, + "learning_rate": 1e-06, + "loss": 0.7071, + "mean_token_accuracy": 0.7679201364517212, + "num_tokens": 935747568.0, + "step": 34796 + }, + { + "epoch": 2.869382369918364, + "grad_norm": 1.8914750814437866, + "learning_rate": 1e-06, + "loss": 0.591, + "mean_token_accuracy": 0.7970661520957947, + "num_tokens": 935780330.0, + "step": 34797 + }, + { + "epoch": 2.8694648305434156, + "grad_norm": 2.041627883911133, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7357736229896545, + "num_tokens": 935810897.0, + "step": 34798 + }, + { + "epoch": 2.869547291168467, + "grad_norm": 2.037182569503784, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.7631031274795532, + "num_tokens": 935843576.0, + "step": 34799 + }, + { + "epoch": 2.8696297517935188, + "grad_norm": 1.9381910562515259, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.787391185760498, + "num_tokens": 935875605.0, + "step": 34800 + }, + { + "epoch": 2.86971221241857, + "grad_norm": 2.4056637287139893, + "learning_rate": 1e-06, + "loss": 0.7349, + "mean_token_accuracy": 0.7621129751205444, + "num_tokens": 935898525.0, + "step": 34801 + }, + { + "epoch": 2.8697946730436215, + "grad_norm": 2.085024833679199, + "learning_rate": 1e-06, + "loss": 0.532, + "mean_token_accuracy": 0.8165432214736938, + "num_tokens": 935925862.0, + "step": 34802 + }, + { + "epoch": 2.8698771336686733, + "grad_norm": 2.1740920543670654, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7609602808952332, + "num_tokens": 935956794.0, + "step": 34803 + }, + { + "epoch": 2.8699595942937246, + "grad_norm": 2.104815721511841, + "learning_rate": 1e-06, + "loss": 0.6026, + "mean_token_accuracy": 0.8008439540863037, + "num_tokens": 935985933.0, + "step": 34804 + }, + { + "epoch": 2.8700420549187764, + "grad_norm": 2.0836100578308105, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.7563270330429077, + "num_tokens": 936017914.0, + "step": 34805 + }, + { + "epoch": 2.8701245155438277, + "grad_norm": 2.113279342651367, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.768912672996521, + "num_tokens": 936047748.0, + "step": 34806 + }, + { + "epoch": 2.8702069761688795, + "grad_norm": 2.249795436859131, + "learning_rate": 1e-06, + "loss": 0.7901, + "mean_token_accuracy": 0.7530689239501953, + "num_tokens": 936078391.0, + "step": 34807 + }, + { + "epoch": 2.870289436793931, + "grad_norm": 2.2732043266296387, + "learning_rate": 1e-06, + "loss": 0.7118, + "mean_token_accuracy": 0.770680844783783, + "num_tokens": 936102287.0, + "step": 34808 + }, + { + "epoch": 2.8703718974189822, + "grad_norm": 2.0607070922851562, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7313696146011353, + "num_tokens": 936135371.0, + "step": 34809 + }, + { + "epoch": 2.870454358044034, + "grad_norm": 2.2528486251831055, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.7811460494995117, + "num_tokens": 936161884.0, + "step": 34810 + }, + { + "epoch": 2.8705368186690854, + "grad_norm": 2.5087339878082275, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.7830409407615662, + "num_tokens": 936185926.0, + "step": 34811 + }, + { + "epoch": 2.870619279294137, + "grad_norm": 2.1190028190612793, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7758853435516357, + "num_tokens": 936215933.0, + "step": 34812 + }, + { + "epoch": 2.8707017399191885, + "grad_norm": 1.9569147825241089, + "learning_rate": 1e-06, + "loss": 0.4375, + "mean_token_accuracy": 0.8262380361557007, + "num_tokens": 936243148.0, + "step": 34813 + }, + { + "epoch": 2.8707842005442403, + "grad_norm": 2.2193515300750732, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7316045165061951, + "num_tokens": 936272694.0, + "step": 34814 + }, + { + "epoch": 2.8708666611692917, + "grad_norm": 1.962923526763916, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7753794193267822, + "num_tokens": 936308362.0, + "step": 34815 + }, + { + "epoch": 2.870949121794343, + "grad_norm": 2.1287450790405273, + "learning_rate": 1e-06, + "loss": 0.6695, + "mean_token_accuracy": 0.7770673036575317, + "num_tokens": 936338182.0, + "step": 34816 + }, + { + "epoch": 2.871031582419395, + "grad_norm": 1.928691029548645, + "learning_rate": 1e-06, + "loss": 0.5795, + "mean_token_accuracy": 0.805823802947998, + "num_tokens": 936372556.0, + "step": 34817 + }, + { + "epoch": 2.871114043044446, + "grad_norm": 2.0891451835632324, + "learning_rate": 1e-06, + "loss": 0.6341, + "mean_token_accuracy": 0.7951645255088806, + "num_tokens": 936404432.0, + "step": 34818 + }, + { + "epoch": 2.871196503669498, + "grad_norm": 2.208010196685791, + "learning_rate": 1e-06, + "loss": 0.7224, + "mean_token_accuracy": 0.7627315521240234, + "num_tokens": 936430968.0, + "step": 34819 + }, + { + "epoch": 2.8712789642945493, + "grad_norm": 2.1346988677978516, + "learning_rate": 1e-06, + "loss": 0.6045, + "mean_token_accuracy": 0.7982068061828613, + "num_tokens": 936458482.0, + "step": 34820 + }, + { + "epoch": 2.871361424919601, + "grad_norm": 2.2901246547698975, + "learning_rate": 1e-06, + "loss": 0.7032, + "mean_token_accuracy": 0.7762537598609924, + "num_tokens": 936484160.0, + "step": 34821 + }, + { + "epoch": 2.8714438855446525, + "grad_norm": 2.362666606903076, + "learning_rate": 1e-06, + "loss": 0.6089, + "mean_token_accuracy": 0.7855740785598755, + "num_tokens": 936507762.0, + "step": 34822 + }, + { + "epoch": 2.871526346169704, + "grad_norm": 1.7952769994735718, + "learning_rate": 1e-06, + "loss": 0.6058, + "mean_token_accuracy": 0.8055536150932312, + "num_tokens": 936545426.0, + "step": 34823 + }, + { + "epoch": 2.8716088067947556, + "grad_norm": 2.4818739891052246, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7375131845474243, + "num_tokens": 936571339.0, + "step": 34824 + }, + { + "epoch": 2.871691267419807, + "grad_norm": 2.193859577178955, + "learning_rate": 1e-06, + "loss": 0.6428, + "mean_token_accuracy": 0.7838543653488159, + "num_tokens": 936599944.0, + "step": 34825 + }, + { + "epoch": 2.8717737280448583, + "grad_norm": 2.0824856758117676, + "learning_rate": 1e-06, + "loss": 0.5388, + "mean_token_accuracy": 0.8171334862709045, + "num_tokens": 936628385.0, + "step": 34826 + }, + { + "epoch": 2.87185618866991, + "grad_norm": 1.993053913116455, + "learning_rate": 1e-06, + "loss": 0.6032, + "mean_token_accuracy": 0.804304301738739, + "num_tokens": 936659416.0, + "step": 34827 + }, + { + "epoch": 2.871938649294962, + "grad_norm": 2.083768606185913, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7721905708312988, + "num_tokens": 936689882.0, + "step": 34828 + }, + { + "epoch": 2.8720211099200132, + "grad_norm": 2.3830604553222656, + "learning_rate": 1e-06, + "loss": 0.6412, + "mean_token_accuracy": 0.7843241691589355, + "num_tokens": 936713083.0, + "step": 34829 + }, + { + "epoch": 2.8721035705450646, + "grad_norm": 2.003526449203491, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7548956871032715, + "num_tokens": 936745498.0, + "step": 34830 + }, + { + "epoch": 2.8721860311701164, + "grad_norm": 2.2157187461853027, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.771643340587616, + "num_tokens": 936772083.0, + "step": 34831 + }, + { + "epoch": 2.8722684917951677, + "grad_norm": 2.055497407913208, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7502843737602234, + "num_tokens": 936805611.0, + "step": 34832 + }, + { + "epoch": 2.872350952420219, + "grad_norm": 1.924926519393921, + "learning_rate": 1e-06, + "loss": 0.5975, + "mean_token_accuracy": 0.8011304140090942, + "num_tokens": 936842131.0, + "step": 34833 + }, + { + "epoch": 2.872433413045271, + "grad_norm": 2.042245626449585, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7253408432006836, + "num_tokens": 936874213.0, + "step": 34834 + }, + { + "epoch": 2.8725158736703227, + "grad_norm": 1.9798327684402466, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7651095986366272, + "num_tokens": 936907850.0, + "step": 34835 + }, + { + "epoch": 2.872598334295374, + "grad_norm": 1.898333191871643, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7734063863754272, + "num_tokens": 936943067.0, + "step": 34836 + }, + { + "epoch": 2.8726807949204254, + "grad_norm": 2.181851387023926, + "learning_rate": 1e-06, + "loss": 0.6435, + "mean_token_accuracy": 0.7917976379394531, + "num_tokens": 936972298.0, + "step": 34837 + }, + { + "epoch": 2.872763255545477, + "grad_norm": 2.5649876594543457, + "learning_rate": 1e-06, + "loss": 0.6803, + "mean_token_accuracy": 0.7815215587615967, + "num_tokens": 936994496.0, + "step": 34838 + }, + { + "epoch": 2.8728457161705285, + "grad_norm": 2.0498595237731934, + "learning_rate": 1e-06, + "loss": 0.5203, + "mean_token_accuracy": 0.8234540224075317, + "num_tokens": 937024783.0, + "step": 34839 + }, + { + "epoch": 2.87292817679558, + "grad_norm": 1.8402479887008667, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7764701843261719, + "num_tokens": 937058895.0, + "step": 34840 + }, + { + "epoch": 2.8730106374206317, + "grad_norm": 2.371403932571411, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.779982328414917, + "num_tokens": 937083054.0, + "step": 34841 + }, + { + "epoch": 2.873093098045683, + "grad_norm": 2.1081881523132324, + "learning_rate": 1e-06, + "loss": 0.7543, + "mean_token_accuracy": 0.7586989998817444, + "num_tokens": 937111968.0, + "step": 34842 + }, + { + "epoch": 2.873175558670735, + "grad_norm": 2.365041732788086, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7717273831367493, + "num_tokens": 937136100.0, + "step": 34843 + }, + { + "epoch": 2.873258019295786, + "grad_norm": 2.1855502128601074, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7649853229522705, + "num_tokens": 937165014.0, + "step": 34844 + }, + { + "epoch": 2.873340479920838, + "grad_norm": 1.940774917602539, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7713475823402405, + "num_tokens": 937198528.0, + "step": 34845 + }, + { + "epoch": 2.8734229405458893, + "grad_norm": 2.1364359855651855, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7761077880859375, + "num_tokens": 937226616.0, + "step": 34846 + }, + { + "epoch": 2.8735054011709407, + "grad_norm": 2.254276752471924, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7360787391662598, + "num_tokens": 937254140.0, + "step": 34847 + }, + { + "epoch": 2.8735878617959925, + "grad_norm": 1.9523229598999023, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7622731924057007, + "num_tokens": 937287054.0, + "step": 34848 + }, + { + "epoch": 2.873670322421044, + "grad_norm": 2.1925530433654785, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.7925328612327576, + "num_tokens": 937315582.0, + "step": 34849 + }, + { + "epoch": 2.8737527830460956, + "grad_norm": 1.9108973741531372, + "learning_rate": 1e-06, + "loss": 0.6773, + "mean_token_accuracy": 0.7686421871185303, + "num_tokens": 937351986.0, + "step": 34850 + }, + { + "epoch": 2.873835243671147, + "grad_norm": 2.0223658084869385, + "learning_rate": 1e-06, + "loss": 0.6159, + "mean_token_accuracy": 0.7936289310455322, + "num_tokens": 937381030.0, + "step": 34851 + }, + { + "epoch": 2.8739177042961987, + "grad_norm": 1.968591570854187, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7792763710021973, + "num_tokens": 937414611.0, + "step": 34852 + }, + { + "epoch": 2.87400016492125, + "grad_norm": 2.3902323246002197, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7295130491256714, + "num_tokens": 937439670.0, + "step": 34853 + }, + { + "epoch": 2.8740826255463015, + "grad_norm": 2.12711501121521, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7588193416595459, + "num_tokens": 937469483.0, + "step": 34854 + }, + { + "epoch": 2.8741650861713532, + "grad_norm": 2.2421274185180664, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7622544765472412, + "num_tokens": 937496264.0, + "step": 34855 + }, + { + "epoch": 2.8742475467964046, + "grad_norm": 2.05558705329895, + "learning_rate": 1e-06, + "loss": 0.7467, + "mean_token_accuracy": 0.7642059326171875, + "num_tokens": 937527228.0, + "step": 34856 + }, + { + "epoch": 2.8743300074214564, + "grad_norm": 2.112879991531372, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.7533065676689148, + "num_tokens": 937558049.0, + "step": 34857 + }, + { + "epoch": 2.8744124680465077, + "grad_norm": 1.9987759590148926, + "learning_rate": 1e-06, + "loss": 0.789, + "mean_token_accuracy": 0.7482473254203796, + "num_tokens": 937590560.0, + "step": 34858 + }, + { + "epoch": 2.8744949286715595, + "grad_norm": 2.136436700820923, + "learning_rate": 1e-06, + "loss": 0.5959, + "mean_token_accuracy": 0.8038066625595093, + "num_tokens": 937618017.0, + "step": 34859 + }, + { + "epoch": 2.874577389296611, + "grad_norm": 2.1913363933563232, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7806028127670288, + "num_tokens": 937645210.0, + "step": 34860 + }, + { + "epoch": 2.8746598499216622, + "grad_norm": 2.0088775157928467, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.7606359720230103, + "num_tokens": 937677049.0, + "step": 34861 + }, + { + "epoch": 2.874742310546714, + "grad_norm": 2.017151355743408, + "learning_rate": 1e-06, + "loss": 0.5782, + "mean_token_accuracy": 0.7981417179107666, + "num_tokens": 937709313.0, + "step": 34862 + }, + { + "epoch": 2.8748247711717654, + "grad_norm": 2.064833402633667, + "learning_rate": 1e-06, + "loss": 0.7685, + "mean_token_accuracy": 0.7628787755966187, + "num_tokens": 937739318.0, + "step": 34863 + }, + { + "epoch": 2.874907231796817, + "grad_norm": 1.8445643186569214, + "learning_rate": 1e-06, + "loss": 0.6021, + "mean_token_accuracy": 0.805794358253479, + "num_tokens": 937772757.0, + "step": 34864 + }, + { + "epoch": 2.8749896924218685, + "grad_norm": 2.1592040061950684, + "learning_rate": 1e-06, + "loss": 0.6723, + "mean_token_accuracy": 0.7766088247299194, + "num_tokens": 937800566.0, + "step": 34865 + }, + { + "epoch": 2.8750721530469203, + "grad_norm": 2.0122017860412598, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7413915395736694, + "num_tokens": 937833677.0, + "step": 34866 + }, + { + "epoch": 2.8751546136719717, + "grad_norm": 2.0322861671447754, + "learning_rate": 1e-06, + "loss": 0.6492, + "mean_token_accuracy": 0.7690258622169495, + "num_tokens": 937862002.0, + "step": 34867 + }, + { + "epoch": 2.875237074297023, + "grad_norm": 2.0654313564300537, + "learning_rate": 1e-06, + "loss": 0.7083, + "mean_token_accuracy": 0.7701762914657593, + "num_tokens": 937896001.0, + "step": 34868 + }, + { + "epoch": 2.875319534922075, + "grad_norm": 1.9191638231277466, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.7955321669578552, + "num_tokens": 937930815.0, + "step": 34869 + }, + { + "epoch": 2.875401995547126, + "grad_norm": 1.8925213813781738, + "learning_rate": 1e-06, + "loss": 0.6968, + "mean_token_accuracy": 0.7654926776885986, + "num_tokens": 937967342.0, + "step": 34870 + }, + { + "epoch": 2.8754844561721775, + "grad_norm": 2.093024730682373, + "learning_rate": 1e-06, + "loss": 0.7206, + "mean_token_accuracy": 0.7824658751487732, + "num_tokens": 937997102.0, + "step": 34871 + }, + { + "epoch": 2.8755669167972293, + "grad_norm": 2.069263458251953, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.7867355942726135, + "num_tokens": 938025971.0, + "step": 34872 + }, + { + "epoch": 2.875649377422281, + "grad_norm": 2.0985453128814697, + "learning_rate": 1e-06, + "loss": 0.5793, + "mean_token_accuracy": 0.811493456363678, + "num_tokens": 938055660.0, + "step": 34873 + }, + { + "epoch": 2.8757318380473325, + "grad_norm": 2.084303855895996, + "learning_rate": 1e-06, + "loss": 0.5885, + "mean_token_accuracy": 0.8046695590019226, + "num_tokens": 938084357.0, + "step": 34874 + }, + { + "epoch": 2.875814298672384, + "grad_norm": 1.8906744718551636, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.7792474031448364, + "num_tokens": 938118047.0, + "step": 34875 + }, + { + "epoch": 2.8758967592974356, + "grad_norm": 1.9182944297790527, + "learning_rate": 1e-06, + "loss": 0.5859, + "mean_token_accuracy": 0.8004699945449829, + "num_tokens": 938150606.0, + "step": 34876 + }, + { + "epoch": 2.875979219922487, + "grad_norm": 2.299978494644165, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7335387468338013, + "num_tokens": 938175473.0, + "step": 34877 + }, + { + "epoch": 2.8760616805475383, + "grad_norm": 1.9862029552459717, + "learning_rate": 1e-06, + "loss": 0.5598, + "mean_token_accuracy": 0.8094122409820557, + "num_tokens": 938205751.0, + "step": 34878 + }, + { + "epoch": 2.87614414117259, + "grad_norm": 2.3040223121643066, + "learning_rate": 1e-06, + "loss": 0.7725, + "mean_token_accuracy": 0.7473611831665039, + "num_tokens": 938233177.0, + "step": 34879 + }, + { + "epoch": 2.876226601797642, + "grad_norm": 1.842721939086914, + "learning_rate": 1e-06, + "loss": 0.5436, + "mean_token_accuracy": 0.8144654631614685, + "num_tokens": 938267904.0, + "step": 34880 + }, + { + "epoch": 2.8763090624226932, + "grad_norm": 2.1286866664886475, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7640193700790405, + "num_tokens": 938298932.0, + "step": 34881 + }, + { + "epoch": 2.8763915230477446, + "grad_norm": 2.0213074684143066, + "learning_rate": 1e-06, + "loss": 0.5923, + "mean_token_accuracy": 0.7983514070510864, + "num_tokens": 938328673.0, + "step": 34882 + }, + { + "epoch": 2.8764739836727964, + "grad_norm": 1.9654539823532104, + "learning_rate": 1e-06, + "loss": 0.6066, + "mean_token_accuracy": 0.7954944372177124, + "num_tokens": 938360671.0, + "step": 34883 + }, + { + "epoch": 2.8765564442978477, + "grad_norm": 2.178335189819336, + "learning_rate": 1e-06, + "loss": 0.8028, + "mean_token_accuracy": 0.7427215576171875, + "num_tokens": 938391093.0, + "step": 34884 + }, + { + "epoch": 2.876638904922899, + "grad_norm": 2.061943531036377, + "learning_rate": 1e-06, + "loss": 0.6448, + "mean_token_accuracy": 0.7857440710067749, + "num_tokens": 938420916.0, + "step": 34885 + }, + { + "epoch": 2.876721365547951, + "grad_norm": 2.0390710830688477, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7198705673217773, + "num_tokens": 938453343.0, + "step": 34886 + }, + { + "epoch": 2.8768038261730022, + "grad_norm": 2.238401174545288, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.748793363571167, + "num_tokens": 938480913.0, + "step": 34887 + }, + { + "epoch": 2.876886286798054, + "grad_norm": 2.036709785461426, + "learning_rate": 1e-06, + "loss": 0.57, + "mean_token_accuracy": 0.8063139915466309, + "num_tokens": 938515531.0, + "step": 34888 + }, + { + "epoch": 2.8769687474231054, + "grad_norm": 1.899850606918335, + "learning_rate": 1e-06, + "loss": 0.6169, + "mean_token_accuracy": 0.8062073588371277, + "num_tokens": 938551966.0, + "step": 34889 + }, + { + "epoch": 2.877051208048157, + "grad_norm": 2.106661081314087, + "learning_rate": 1e-06, + "loss": 0.744, + "mean_token_accuracy": 0.759827733039856, + "num_tokens": 938579486.0, + "step": 34890 + }, + { + "epoch": 2.8771336686732085, + "grad_norm": 2.2897777557373047, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7702980041503906, + "num_tokens": 938604865.0, + "step": 34891 + }, + { + "epoch": 2.87721612929826, + "grad_norm": 2.39634370803833, + "learning_rate": 1e-06, + "loss": 0.665, + "mean_token_accuracy": 0.7839507460594177, + "num_tokens": 938630505.0, + "step": 34892 + }, + { + "epoch": 2.8772985899233117, + "grad_norm": 2.0810000896453857, + "learning_rate": 1e-06, + "loss": 0.7609, + "mean_token_accuracy": 0.7598996758460999, + "num_tokens": 938661192.0, + "step": 34893 + }, + { + "epoch": 2.877381050548363, + "grad_norm": 1.9912409782409668, + "learning_rate": 1e-06, + "loss": 0.7335, + "mean_token_accuracy": 0.7639180421829224, + "num_tokens": 938694511.0, + "step": 34894 + }, + { + "epoch": 2.877463511173415, + "grad_norm": 2.30476713180542, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.7568842172622681, + "num_tokens": 938721075.0, + "step": 34895 + }, + { + "epoch": 2.877545971798466, + "grad_norm": 2.0774192810058594, + "learning_rate": 1e-06, + "loss": 0.7069, + "mean_token_accuracy": 0.7752823829650879, + "num_tokens": 938751793.0, + "step": 34896 + }, + { + "epoch": 2.877628432423518, + "grad_norm": 2.0711164474487305, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.7854453325271606, + "num_tokens": 938779721.0, + "step": 34897 + }, + { + "epoch": 2.8777108930485693, + "grad_norm": 2.1717379093170166, + "learning_rate": 1e-06, + "loss": 0.7039, + "mean_token_accuracy": 0.7726600766181946, + "num_tokens": 938807305.0, + "step": 34898 + }, + { + "epoch": 2.8777933536736207, + "grad_norm": 2.3897006511688232, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7513540983200073, + "num_tokens": 938832345.0, + "step": 34899 + }, + { + "epoch": 2.8778758142986725, + "grad_norm": 2.5154919624328613, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7461515665054321, + "num_tokens": 938855235.0, + "step": 34900 + }, + { + "epoch": 2.877958274923724, + "grad_norm": 2.267474889755249, + "learning_rate": 1e-06, + "loss": 0.6092, + "mean_token_accuracy": 0.8041436672210693, + "num_tokens": 938879578.0, + "step": 34901 + }, + { + "epoch": 2.8780407355487756, + "grad_norm": 2.3130853176116943, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.783525824546814, + "num_tokens": 938905734.0, + "step": 34902 + }, + { + "epoch": 2.878123196173827, + "grad_norm": 2.0773415565490723, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7276409864425659, + "num_tokens": 938935695.0, + "step": 34903 + }, + { + "epoch": 2.8782056567988787, + "grad_norm": 2.220355749130249, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7399149537086487, + "num_tokens": 938963392.0, + "step": 34904 + }, + { + "epoch": 2.87828811742393, + "grad_norm": 2.5500168800354004, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7568434476852417, + "num_tokens": 938985603.0, + "step": 34905 + }, + { + "epoch": 2.8783705780489814, + "grad_norm": 2.2506561279296875, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7386825680732727, + "num_tokens": 939012342.0, + "step": 34906 + }, + { + "epoch": 2.8784530386740332, + "grad_norm": 1.8903565406799316, + "learning_rate": 1e-06, + "loss": 0.6987, + "mean_token_accuracy": 0.7778840661048889, + "num_tokens": 939048369.0, + "step": 34907 + }, + { + "epoch": 2.8785354992990846, + "grad_norm": 2.0081417560577393, + "learning_rate": 1e-06, + "loss": 0.6542, + "mean_token_accuracy": 0.784177303314209, + "num_tokens": 939078347.0, + "step": 34908 + }, + { + "epoch": 2.8786179599241364, + "grad_norm": 1.9704543352127075, + "learning_rate": 1e-06, + "loss": 0.7646, + "mean_token_accuracy": 0.7646629810333252, + "num_tokens": 939109605.0, + "step": 34909 + }, + { + "epoch": 2.8787004205491877, + "grad_norm": 2.161991596221924, + "learning_rate": 1e-06, + "loss": 0.6674, + "mean_token_accuracy": 0.7716741561889648, + "num_tokens": 939141674.0, + "step": 34910 + }, + { + "epoch": 2.8787828811742395, + "grad_norm": 1.9789912700653076, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.7881320118904114, + "num_tokens": 939174666.0, + "step": 34911 + }, + { + "epoch": 2.878865341799291, + "grad_norm": 2.0351169109344482, + "learning_rate": 1e-06, + "loss": 0.57, + "mean_token_accuracy": 0.8100696206092834, + "num_tokens": 939204933.0, + "step": 34912 + }, + { + "epoch": 2.8789478024243422, + "grad_norm": 2.056662082672119, + "learning_rate": 1e-06, + "loss": 0.6679, + "mean_token_accuracy": 0.7774098515510559, + "num_tokens": 939232989.0, + "step": 34913 + }, + { + "epoch": 2.879030263049394, + "grad_norm": 1.9756535291671753, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7669703364372253, + "num_tokens": 939268676.0, + "step": 34914 + }, + { + "epoch": 2.8791127236744454, + "grad_norm": 2.4380853176116943, + "learning_rate": 1e-06, + "loss": 0.8118, + "mean_token_accuracy": 0.7436458468437195, + "num_tokens": 939299756.0, + "step": 34915 + }, + { + "epoch": 2.8791951842994967, + "grad_norm": 2.462923049926758, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.7882338762283325, + "num_tokens": 939320311.0, + "step": 34916 + }, + { + "epoch": 2.8792776449245485, + "grad_norm": 2.0904152393341064, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7531039118766785, + "num_tokens": 939349113.0, + "step": 34917 + }, + { + "epoch": 2.8793601055496003, + "grad_norm": 2.0374536514282227, + "learning_rate": 1e-06, + "loss": 0.5104, + "mean_token_accuracy": 0.8130136132240295, + "num_tokens": 939376395.0, + "step": 34918 + }, + { + "epoch": 2.8794425661746517, + "grad_norm": 2.277204751968384, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7544090151786804, + "num_tokens": 939405002.0, + "step": 34919 + }, + { + "epoch": 2.879525026799703, + "grad_norm": 1.9994436502456665, + "learning_rate": 1e-06, + "loss": 0.5548, + "mean_token_accuracy": 0.817696750164032, + "num_tokens": 939436246.0, + "step": 34920 + }, + { + "epoch": 2.879607487424755, + "grad_norm": 2.1609866619110107, + "learning_rate": 1e-06, + "loss": 0.6209, + "mean_token_accuracy": 0.7932737469673157, + "num_tokens": 939465580.0, + "step": 34921 + }, + { + "epoch": 2.879689948049806, + "grad_norm": 2.203735113143921, + "learning_rate": 1e-06, + "loss": 0.6139, + "mean_token_accuracy": 0.791211724281311, + "num_tokens": 939491801.0, + "step": 34922 + }, + { + "epoch": 2.8797724086748575, + "grad_norm": 2.43137526512146, + "learning_rate": 1e-06, + "loss": 0.6999, + "mean_token_accuracy": 0.7725269794464111, + "num_tokens": 939518892.0, + "step": 34923 + }, + { + "epoch": 2.8798548692999093, + "grad_norm": 2.2445499897003174, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.7494477033615112, + "num_tokens": 939545948.0, + "step": 34924 + }, + { + "epoch": 2.879937329924961, + "grad_norm": 2.208495616912842, + "learning_rate": 1e-06, + "loss": 0.6076, + "mean_token_accuracy": 0.7905791997909546, + "num_tokens": 939569831.0, + "step": 34925 + }, + { + "epoch": 2.8800197905500124, + "grad_norm": 2.0723226070404053, + "learning_rate": 1e-06, + "loss": 0.7033, + "mean_token_accuracy": 0.7713837027549744, + "num_tokens": 939601169.0, + "step": 34926 + }, + { + "epoch": 2.880102251175064, + "grad_norm": 1.895667552947998, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7582086324691772, + "num_tokens": 939636928.0, + "step": 34927 + }, + { + "epoch": 2.8801847118001156, + "grad_norm": 2.040172815322876, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7531337738037109, + "num_tokens": 939668783.0, + "step": 34928 + }, + { + "epoch": 2.880267172425167, + "grad_norm": 2.006061553955078, + "learning_rate": 1e-06, + "loss": 0.7361, + "mean_token_accuracy": 0.7648879289627075, + "num_tokens": 939697912.0, + "step": 34929 + }, + { + "epoch": 2.8803496330502183, + "grad_norm": 1.9888088703155518, + "learning_rate": 1e-06, + "loss": 0.7589, + "mean_token_accuracy": 0.7560468316078186, + "num_tokens": 939731129.0, + "step": 34930 + }, + { + "epoch": 2.88043209367527, + "grad_norm": 1.8894314765930176, + "learning_rate": 1e-06, + "loss": 0.5979, + "mean_token_accuracy": 0.7905568480491638, + "num_tokens": 939764957.0, + "step": 34931 + }, + { + "epoch": 2.8805145543003214, + "grad_norm": 2.2369682788848877, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7669017314910889, + "num_tokens": 939792120.0, + "step": 34932 + }, + { + "epoch": 2.8805970149253732, + "grad_norm": 2.2684597969055176, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7469933032989502, + "num_tokens": 939820800.0, + "step": 34933 + }, + { + "epoch": 2.8806794755504246, + "grad_norm": 1.9846335649490356, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7460128664970398, + "num_tokens": 939853086.0, + "step": 34934 + }, + { + "epoch": 2.8807619361754764, + "grad_norm": 2.1873342990875244, + "learning_rate": 1e-06, + "loss": 0.6961, + "mean_token_accuracy": 0.7745883464813232, + "num_tokens": 939882450.0, + "step": 34935 + }, + { + "epoch": 2.8808443968005277, + "grad_norm": 2.2084271907806396, + "learning_rate": 1e-06, + "loss": 0.639, + "mean_token_accuracy": 0.7805629968643188, + "num_tokens": 939911466.0, + "step": 34936 + }, + { + "epoch": 2.880926857425579, + "grad_norm": 2.405534505844116, + "learning_rate": 1e-06, + "loss": 0.5876, + "mean_token_accuracy": 0.802260160446167, + "num_tokens": 939933887.0, + "step": 34937 + }, + { + "epoch": 2.881009318050631, + "grad_norm": 2.32336163520813, + "learning_rate": 1e-06, + "loss": 0.7188, + "mean_token_accuracy": 0.7663319110870361, + "num_tokens": 939961552.0, + "step": 34938 + }, + { + "epoch": 2.8810917786756822, + "grad_norm": 2.5385923385620117, + "learning_rate": 1e-06, + "loss": 0.7604, + "mean_token_accuracy": 0.7549712061882019, + "num_tokens": 939984424.0, + "step": 34939 + }, + { + "epoch": 2.881174239300734, + "grad_norm": 2.0539212226867676, + "learning_rate": 1e-06, + "loss": 0.5814, + "mean_token_accuracy": 0.7916301488876343, + "num_tokens": 940012262.0, + "step": 34940 + }, + { + "epoch": 2.8812566999257854, + "grad_norm": 1.9986011981964111, + "learning_rate": 1e-06, + "loss": 0.5733, + "mean_token_accuracy": 0.8099499344825745, + "num_tokens": 940043077.0, + "step": 34941 + }, + { + "epoch": 2.881339160550837, + "grad_norm": 2.045077323913574, + "learning_rate": 1e-06, + "loss": 0.5708, + "mean_token_accuracy": 0.8073006272315979, + "num_tokens": 940072296.0, + "step": 34942 + }, + { + "epoch": 2.8814216211758885, + "grad_norm": 2.209543466567993, + "learning_rate": 1e-06, + "loss": 0.6902, + "mean_token_accuracy": 0.7707495093345642, + "num_tokens": 940098818.0, + "step": 34943 + }, + { + "epoch": 2.88150408180094, + "grad_norm": 2.057018995285034, + "learning_rate": 1e-06, + "loss": 0.7827, + "mean_token_accuracy": 0.7498493194580078, + "num_tokens": 940129457.0, + "step": 34944 + }, + { + "epoch": 2.8815865424259917, + "grad_norm": 1.9670101404190063, + "learning_rate": 1e-06, + "loss": 0.7436, + "mean_token_accuracy": 0.763035774230957, + "num_tokens": 940162445.0, + "step": 34945 + }, + { + "epoch": 2.881669003051043, + "grad_norm": 2.139090061187744, + "learning_rate": 1e-06, + "loss": 0.6388, + "mean_token_accuracy": 0.7924110293388367, + "num_tokens": 940189950.0, + "step": 34946 + }, + { + "epoch": 2.881751463676095, + "grad_norm": 2.120119094848633, + "learning_rate": 1e-06, + "loss": 0.6213, + "mean_token_accuracy": 0.8041083812713623, + "num_tokens": 940217783.0, + "step": 34947 + }, + { + "epoch": 2.881833924301146, + "grad_norm": 2.019991397857666, + "learning_rate": 1e-06, + "loss": 0.7472, + "mean_token_accuracy": 0.7648756504058838, + "num_tokens": 940251524.0, + "step": 34948 + }, + { + "epoch": 2.881916384926198, + "grad_norm": 2.1373634338378906, + "learning_rate": 1e-06, + "loss": 0.5429, + "mean_token_accuracy": 0.8214808702468872, + "num_tokens": 940275877.0, + "step": 34949 + }, + { + "epoch": 2.8819988455512493, + "grad_norm": 2.417299747467041, + "learning_rate": 1e-06, + "loss": 0.7331, + "mean_token_accuracy": 0.7665351033210754, + "num_tokens": 940299798.0, + "step": 34950 + }, + { + "epoch": 2.8820813061763007, + "grad_norm": 2.1781516075134277, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7533998489379883, + "num_tokens": 940330748.0, + "step": 34951 + }, + { + "epoch": 2.8821637668013524, + "grad_norm": 1.962101936340332, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7618006467819214, + "num_tokens": 940366518.0, + "step": 34952 + }, + { + "epoch": 2.882246227426404, + "grad_norm": 1.8232977390289307, + "learning_rate": 1e-06, + "loss": 0.6531, + "mean_token_accuracy": 0.7978332042694092, + "num_tokens": 940402536.0, + "step": 34953 + }, + { + "epoch": 2.8823286880514556, + "grad_norm": 2.1314845085144043, + "learning_rate": 1e-06, + "loss": 0.726, + "mean_token_accuracy": 0.7654291391372681, + "num_tokens": 940434589.0, + "step": 34954 + }, + { + "epoch": 2.882411148676507, + "grad_norm": 1.753596305847168, + "learning_rate": 1e-06, + "loss": 0.6252, + "mean_token_accuracy": 0.8002086281776428, + "num_tokens": 940470490.0, + "step": 34955 + }, + { + "epoch": 2.8824936093015587, + "grad_norm": 2.3145763874053955, + "learning_rate": 1e-06, + "loss": 0.7258, + "mean_token_accuracy": 0.7704299092292786, + "num_tokens": 940496222.0, + "step": 34956 + }, + { + "epoch": 2.88257606992661, + "grad_norm": 1.8141810894012451, + "learning_rate": 1e-06, + "loss": 0.7878, + "mean_token_accuracy": 0.7599465250968933, + "num_tokens": 940534299.0, + "step": 34957 + }, + { + "epoch": 2.8826585305516614, + "grad_norm": 2.035982847213745, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7868670225143433, + "num_tokens": 940567089.0, + "step": 34958 + }, + { + "epoch": 2.8827409911767132, + "grad_norm": 2.2425975799560547, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.7826393842697144, + "num_tokens": 940591693.0, + "step": 34959 + }, + { + "epoch": 2.8828234518017646, + "grad_norm": 2.2764761447906494, + "learning_rate": 1e-06, + "loss": 0.7459, + "mean_token_accuracy": 0.7578240633010864, + "num_tokens": 940618846.0, + "step": 34960 + }, + { + "epoch": 2.882905912426816, + "grad_norm": 1.917321801185608, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7569018602371216, + "num_tokens": 940651581.0, + "step": 34961 + }, + { + "epoch": 2.8829883730518677, + "grad_norm": 1.9367730617523193, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.7778062224388123, + "num_tokens": 940682753.0, + "step": 34962 + }, + { + "epoch": 2.8830708336769195, + "grad_norm": 1.9844297170639038, + "learning_rate": 1e-06, + "loss": 0.5635, + "mean_token_accuracy": 0.806774377822876, + "num_tokens": 940715644.0, + "step": 34963 + }, + { + "epoch": 2.883153294301971, + "grad_norm": 2.3266379833221436, + "learning_rate": 1e-06, + "loss": 0.5667, + "mean_token_accuracy": 0.804833710193634, + "num_tokens": 940739248.0, + "step": 34964 + }, + { + "epoch": 2.883235754927022, + "grad_norm": 1.8264912366867065, + "learning_rate": 1e-06, + "loss": 0.6442, + "mean_token_accuracy": 0.792707085609436, + "num_tokens": 940775435.0, + "step": 34965 + }, + { + "epoch": 2.883318215552074, + "grad_norm": 1.9609529972076416, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7344968318939209, + "num_tokens": 940810500.0, + "step": 34966 + }, + { + "epoch": 2.8834006761771254, + "grad_norm": 2.2322394847869873, + "learning_rate": 1e-06, + "loss": 0.7195, + "mean_token_accuracy": 0.7695690989494324, + "num_tokens": 940837506.0, + "step": 34967 + }, + { + "epoch": 2.8834831368021767, + "grad_norm": 2.1802613735198975, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7577455043792725, + "num_tokens": 940869296.0, + "step": 34968 + }, + { + "epoch": 2.8835655974272285, + "grad_norm": 2.1107311248779297, + "learning_rate": 1e-06, + "loss": 0.7392, + "mean_token_accuracy": 0.7743167281150818, + "num_tokens": 940898282.0, + "step": 34969 + }, + { + "epoch": 2.8836480580522803, + "grad_norm": 1.9515529870986938, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7419309616088867, + "num_tokens": 940933722.0, + "step": 34970 + }, + { + "epoch": 2.8837305186773317, + "grad_norm": 2.241448402404785, + "learning_rate": 1e-06, + "loss": 0.5758, + "mean_token_accuracy": 0.800413966178894, + "num_tokens": 940959422.0, + "step": 34971 + }, + { + "epoch": 2.883812979302383, + "grad_norm": 2.1339142322540283, + "learning_rate": 1e-06, + "loss": 0.6088, + "mean_token_accuracy": 0.7900105118751526, + "num_tokens": 940987635.0, + "step": 34972 + }, + { + "epoch": 2.883895439927435, + "grad_norm": 1.9312376976013184, + "learning_rate": 1e-06, + "loss": 0.5846, + "mean_token_accuracy": 0.7999334335327148, + "num_tokens": 941018667.0, + "step": 34973 + }, + { + "epoch": 2.883977900552486, + "grad_norm": 2.194826364517212, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7478357553482056, + "num_tokens": 941045055.0, + "step": 34974 + }, + { + "epoch": 2.8840603611775375, + "grad_norm": 2.2004857063293457, + "learning_rate": 1e-06, + "loss": 0.7203, + "mean_token_accuracy": 0.7708439826965332, + "num_tokens": 941074981.0, + "step": 34975 + }, + { + "epoch": 2.8841428218025893, + "grad_norm": 2.027947187423706, + "learning_rate": 1e-06, + "loss": 0.5361, + "mean_token_accuracy": 0.8212388753890991, + "num_tokens": 941105324.0, + "step": 34976 + }, + { + "epoch": 2.8842252824276406, + "grad_norm": 1.932907223701477, + "learning_rate": 1e-06, + "loss": 0.7653, + "mean_token_accuracy": 0.7462852597236633, + "num_tokens": 941144249.0, + "step": 34977 + }, + { + "epoch": 2.8843077430526924, + "grad_norm": 1.9904017448425293, + "learning_rate": 1e-06, + "loss": 0.566, + "mean_token_accuracy": 0.78700852394104, + "num_tokens": 941176681.0, + "step": 34978 + }, + { + "epoch": 2.884390203677744, + "grad_norm": 1.9111689329147339, + "learning_rate": 1e-06, + "loss": 0.6071, + "mean_token_accuracy": 0.7974646091461182, + "num_tokens": 941210788.0, + "step": 34979 + }, + { + "epoch": 2.8844726643027956, + "grad_norm": 2.072223424911499, + "learning_rate": 1e-06, + "loss": 0.586, + "mean_token_accuracy": 0.8020730018615723, + "num_tokens": 941239673.0, + "step": 34980 + }, + { + "epoch": 2.884555124927847, + "grad_norm": 2.496666669845581, + "learning_rate": 1e-06, + "loss": 0.6581, + "mean_token_accuracy": 0.7750136256217957, + "num_tokens": 941262091.0, + "step": 34981 + }, + { + "epoch": 2.8846375855528983, + "grad_norm": 1.980542778968811, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7721350193023682, + "num_tokens": 941296560.0, + "step": 34982 + }, + { + "epoch": 2.88472004617795, + "grad_norm": 2.328071117401123, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.7926231622695923, + "num_tokens": 941321116.0, + "step": 34983 + }, + { + "epoch": 2.8848025068030014, + "grad_norm": 2.052327871322632, + "learning_rate": 1e-06, + "loss": 0.7951, + "mean_token_accuracy": 0.7494007349014282, + "num_tokens": 941352036.0, + "step": 34984 + }, + { + "epoch": 2.8848849674280532, + "grad_norm": 2.0319175720214844, + "learning_rate": 1e-06, + "loss": 0.6084, + "mean_token_accuracy": 0.7989720702171326, + "num_tokens": 941382753.0, + "step": 34985 + }, + { + "epoch": 2.8849674280531046, + "grad_norm": 2.21266770362854, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.773621678352356, + "num_tokens": 941411771.0, + "step": 34986 + }, + { + "epoch": 2.8850498886781564, + "grad_norm": 2.4181201457977295, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7525936365127563, + "num_tokens": 941438165.0, + "step": 34987 + }, + { + "epoch": 2.8851323493032077, + "grad_norm": 2.155442237854004, + "learning_rate": 1e-06, + "loss": 0.5907, + "mean_token_accuracy": 0.7993959188461304, + "num_tokens": 941464622.0, + "step": 34988 + }, + { + "epoch": 2.885214809928259, + "grad_norm": 1.9662208557128906, + "learning_rate": 1e-06, + "loss": 0.7503, + "mean_token_accuracy": 0.7636774182319641, + "num_tokens": 941496600.0, + "step": 34989 + }, + { + "epoch": 2.885297270553311, + "grad_norm": 2.102431297302246, + "learning_rate": 1e-06, + "loss": 0.5696, + "mean_token_accuracy": 0.8119511604309082, + "num_tokens": 941525514.0, + "step": 34990 + }, + { + "epoch": 2.885379731178362, + "grad_norm": 2.1088500022888184, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7445461750030518, + "num_tokens": 941555177.0, + "step": 34991 + }, + { + "epoch": 2.885462191803414, + "grad_norm": 1.999651551246643, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7867255806922913, + "num_tokens": 941584845.0, + "step": 34992 + }, + { + "epoch": 2.8855446524284654, + "grad_norm": 2.284740924835205, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7391309142112732, + "num_tokens": 941610530.0, + "step": 34993 + }, + { + "epoch": 2.885627113053517, + "grad_norm": 2.0956995487213135, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7680993676185608, + "num_tokens": 941638996.0, + "step": 34994 + }, + { + "epoch": 2.8857095736785685, + "grad_norm": 2.0266854763031006, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7858630418777466, + "num_tokens": 941670767.0, + "step": 34995 + }, + { + "epoch": 2.88579203430362, + "grad_norm": 2.566307783126831, + "learning_rate": 1e-06, + "loss": 0.6381, + "mean_token_accuracy": 0.7900720834732056, + "num_tokens": 941699020.0, + "step": 34996 + }, + { + "epoch": 2.8858744949286717, + "grad_norm": 2.169246196746826, + "learning_rate": 1e-06, + "loss": 0.7554, + "mean_token_accuracy": 0.7559287548065186, + "num_tokens": 941725545.0, + "step": 34997 + }, + { + "epoch": 2.885956955553723, + "grad_norm": 2.3379595279693604, + "learning_rate": 1e-06, + "loss": 0.6668, + "mean_token_accuracy": 0.7835259437561035, + "num_tokens": 941750813.0, + "step": 34998 + }, + { + "epoch": 2.886039416178775, + "grad_norm": 2.4035096168518066, + "learning_rate": 1e-06, + "loss": 0.7196, + "mean_token_accuracy": 0.766595721244812, + "num_tokens": 941773747.0, + "step": 34999 + }, + { + "epoch": 2.886121876803826, + "grad_norm": 2.1908018589019775, + "learning_rate": 1e-06, + "loss": 0.6281, + "mean_token_accuracy": 0.7914648652076721, + "num_tokens": 941802240.0, + "step": 35000 + }, + { + "epoch": 2.886204337428878, + "grad_norm": 2.285217523574829, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.7556946873664856, + "num_tokens": 941832741.0, + "step": 35001 + }, + { + "epoch": 2.8862867980539293, + "grad_norm": 2.0135421752929688, + "learning_rate": 1e-06, + "loss": 0.7311, + "mean_token_accuracy": 0.7598776817321777, + "num_tokens": 941862798.0, + "step": 35002 + }, + { + "epoch": 2.8863692586789806, + "grad_norm": 2.3568966388702393, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7344294786453247, + "num_tokens": 941887972.0, + "step": 35003 + }, + { + "epoch": 2.8864517193040324, + "grad_norm": 2.4141643047332764, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7437374591827393, + "num_tokens": 941916543.0, + "step": 35004 + }, + { + "epoch": 2.886534179929084, + "grad_norm": 2.6175642013549805, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7582315802574158, + "num_tokens": 941939978.0, + "step": 35005 + }, + { + "epoch": 2.886616640554135, + "grad_norm": 2.08913516998291, + "learning_rate": 1e-06, + "loss": 0.7323, + "mean_token_accuracy": 0.7639777660369873, + "num_tokens": 941969988.0, + "step": 35006 + }, + { + "epoch": 2.886699101179187, + "grad_norm": 2.293201446533203, + "learning_rate": 1e-06, + "loss": 0.6738, + "mean_token_accuracy": 0.7805818319320679, + "num_tokens": 941995409.0, + "step": 35007 + }, + { + "epoch": 2.8867815618042387, + "grad_norm": 2.09060001373291, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7790835499763489, + "num_tokens": 942025467.0, + "step": 35008 + }, + { + "epoch": 2.88686402242929, + "grad_norm": 2.07633638381958, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.778130054473877, + "num_tokens": 942055047.0, + "step": 35009 + }, + { + "epoch": 2.8869464830543414, + "grad_norm": 2.5505731105804443, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7547129988670349, + "num_tokens": 942077761.0, + "step": 35010 + }, + { + "epoch": 2.8870289436793932, + "grad_norm": 2.3020107746124268, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7736475467681885, + "num_tokens": 942103253.0, + "step": 35011 + }, + { + "epoch": 2.8871114043044446, + "grad_norm": 2.251802682876587, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7655361890792847, + "num_tokens": 942130103.0, + "step": 35012 + }, + { + "epoch": 2.887193864929496, + "grad_norm": 2.0818564891815186, + "learning_rate": 1e-06, + "loss": 0.6644, + "mean_token_accuracy": 0.7858982682228088, + "num_tokens": 942163193.0, + "step": 35013 + }, + { + "epoch": 2.8872763255545477, + "grad_norm": 1.9609376192092896, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7867838144302368, + "num_tokens": 942197594.0, + "step": 35014 + }, + { + "epoch": 2.8873587861795995, + "grad_norm": 2.0619726181030273, + "learning_rate": 1e-06, + "loss": 0.6208, + "mean_token_accuracy": 0.7941575050354004, + "num_tokens": 942228431.0, + "step": 35015 + }, + { + "epoch": 2.887441246804651, + "grad_norm": 1.8481552600860596, + "learning_rate": 1e-06, + "loss": 0.6168, + "mean_token_accuracy": 0.7977436780929565, + "num_tokens": 942262599.0, + "step": 35016 + }, + { + "epoch": 2.887523707429702, + "grad_norm": 1.989978551864624, + "learning_rate": 1e-06, + "loss": 0.5883, + "mean_token_accuracy": 0.7886342406272888, + "num_tokens": 942291218.0, + "step": 35017 + }, + { + "epoch": 2.887606168054754, + "grad_norm": 2.4235658645629883, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7526566386222839, + "num_tokens": 942313984.0, + "step": 35018 + }, + { + "epoch": 2.8876886286798054, + "grad_norm": 2.072984457015991, + "learning_rate": 1e-06, + "loss": 0.7843, + "mean_token_accuracy": 0.7490781545639038, + "num_tokens": 942344689.0, + "step": 35019 + }, + { + "epoch": 2.8877710893048567, + "grad_norm": 2.136829376220703, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7794721126556396, + "num_tokens": 942371544.0, + "step": 35020 + }, + { + "epoch": 2.8878535499299085, + "grad_norm": 2.3126792907714844, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.7900984883308411, + "num_tokens": 942395950.0, + "step": 35021 + }, + { + "epoch": 2.88793601055496, + "grad_norm": 2.088254690170288, + "learning_rate": 1e-06, + "loss": 0.5528, + "mean_token_accuracy": 0.8164416551589966, + "num_tokens": 942424011.0, + "step": 35022 + }, + { + "epoch": 2.8880184711800116, + "grad_norm": 2.037001371383667, + "learning_rate": 1e-06, + "loss": 0.6408, + "mean_token_accuracy": 0.7813882827758789, + "num_tokens": 942456872.0, + "step": 35023 + }, + { + "epoch": 2.888100931805063, + "grad_norm": 2.166353225708008, + "learning_rate": 1e-06, + "loss": 0.7099, + "mean_token_accuracy": 0.7607035040855408, + "num_tokens": 942490217.0, + "step": 35024 + }, + { + "epoch": 2.888183392430115, + "grad_norm": 1.9842888116836548, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7727551460266113, + "num_tokens": 942521608.0, + "step": 35025 + }, + { + "epoch": 2.888265853055166, + "grad_norm": 2.023362636566162, + "learning_rate": 1e-06, + "loss": 0.6845, + "mean_token_accuracy": 0.7737106680870056, + "num_tokens": 942557745.0, + "step": 35026 + }, + { + "epoch": 2.8883483136802175, + "grad_norm": 2.056485652923584, + "learning_rate": 1e-06, + "loss": 0.6862, + "mean_token_accuracy": 0.7768538594245911, + "num_tokens": 942587370.0, + "step": 35027 + }, + { + "epoch": 2.8884307743052693, + "grad_norm": 2.323458433151245, + "learning_rate": 1e-06, + "loss": 0.6965, + "mean_token_accuracy": 0.7715120911598206, + "num_tokens": 942611253.0, + "step": 35028 + }, + { + "epoch": 2.8885132349303206, + "grad_norm": 1.9619896411895752, + "learning_rate": 1e-06, + "loss": 0.6974, + "mean_token_accuracy": 0.7716047763824463, + "num_tokens": 942642238.0, + "step": 35029 + }, + { + "epoch": 2.8885956955553724, + "grad_norm": 2.140871524810791, + "learning_rate": 1e-06, + "loss": 0.7864, + "mean_token_accuracy": 0.7508091330528259, + "num_tokens": 942673515.0, + "step": 35030 + }, + { + "epoch": 2.888678156180424, + "grad_norm": 1.8989897966384888, + "learning_rate": 1e-06, + "loss": 0.5463, + "mean_token_accuracy": 0.8092268109321594, + "num_tokens": 942705621.0, + "step": 35031 + }, + { + "epoch": 2.8887606168054756, + "grad_norm": 2.113927125930786, + "learning_rate": 1e-06, + "loss": 0.543, + "mean_token_accuracy": 0.8167406320571899, + "num_tokens": 942732304.0, + "step": 35032 + }, + { + "epoch": 2.888843077430527, + "grad_norm": 2.108999490737915, + "learning_rate": 1e-06, + "loss": 0.7064, + "mean_token_accuracy": 0.7696833610534668, + "num_tokens": 942768063.0, + "step": 35033 + }, + { + "epoch": 2.8889255380555783, + "grad_norm": 1.8190752267837524, + "learning_rate": 1e-06, + "loss": 0.698, + "mean_token_accuracy": 0.7787442803382874, + "num_tokens": 942803799.0, + "step": 35034 + }, + { + "epoch": 2.88900799868063, + "grad_norm": 1.755037784576416, + "learning_rate": 1e-06, + "loss": 0.5471, + "mean_token_accuracy": 0.811618447303772, + "num_tokens": 942841925.0, + "step": 35035 + }, + { + "epoch": 2.8890904593056814, + "grad_norm": 2.122385025024414, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7838656902313232, + "num_tokens": 942872145.0, + "step": 35036 + }, + { + "epoch": 2.889172919930733, + "grad_norm": 2.318997621536255, + "learning_rate": 1e-06, + "loss": 0.7612, + "mean_token_accuracy": 0.749049186706543, + "num_tokens": 942897836.0, + "step": 35037 + }, + { + "epoch": 2.8892553805557846, + "grad_norm": 1.96101713180542, + "learning_rate": 1e-06, + "loss": 0.5549, + "mean_token_accuracy": 0.8184002637863159, + "num_tokens": 942931264.0, + "step": 35038 + }, + { + "epoch": 2.8893378411808364, + "grad_norm": 2.1029019355773926, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7672839760780334, + "num_tokens": 942960453.0, + "step": 35039 + }, + { + "epoch": 2.8894203018058877, + "grad_norm": 2.094874858856201, + "learning_rate": 1e-06, + "loss": 0.6914, + "mean_token_accuracy": 0.7743497490882874, + "num_tokens": 942993541.0, + "step": 35040 + }, + { + "epoch": 2.889502762430939, + "grad_norm": 2.0869710445404053, + "learning_rate": 1e-06, + "loss": 0.5873, + "mean_token_accuracy": 0.8034173846244812, + "num_tokens": 943022077.0, + "step": 35041 + }, + { + "epoch": 2.889585223055991, + "grad_norm": 2.054976463317871, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7821253538131714, + "num_tokens": 943052423.0, + "step": 35042 + }, + { + "epoch": 2.889667683681042, + "grad_norm": 2.1404244899749756, + "learning_rate": 1e-06, + "loss": 0.6848, + "mean_token_accuracy": 0.7786864638328552, + "num_tokens": 943081520.0, + "step": 35043 + }, + { + "epoch": 2.889750144306094, + "grad_norm": 2.2174837589263916, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7827160358428955, + "num_tokens": 943108409.0, + "step": 35044 + }, + { + "epoch": 2.8898326049311454, + "grad_norm": 2.3396871089935303, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7231634259223938, + "num_tokens": 943135527.0, + "step": 35045 + }, + { + "epoch": 2.889915065556197, + "grad_norm": 2.107107639312744, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7490932941436768, + "num_tokens": 943166093.0, + "step": 35046 + }, + { + "epoch": 2.8899975261812485, + "grad_norm": 1.9100573062896729, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.7917823195457458, + "num_tokens": 943202555.0, + "step": 35047 + }, + { + "epoch": 2.8900799868063, + "grad_norm": 1.921223759651184, + "learning_rate": 1e-06, + "loss": 0.5185, + "mean_token_accuracy": 0.8230730295181274, + "num_tokens": 943236279.0, + "step": 35048 + }, + { + "epoch": 2.8901624474313516, + "grad_norm": 2.087346076965332, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7412239909172058, + "num_tokens": 943267269.0, + "step": 35049 + }, + { + "epoch": 2.890244908056403, + "grad_norm": 2.1315245628356934, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.7852045893669128, + "num_tokens": 943301177.0, + "step": 35050 + }, + { + "epoch": 2.8903273686814543, + "grad_norm": 1.8693647384643555, + "learning_rate": 1e-06, + "loss": 0.6282, + "mean_token_accuracy": 0.7970101237297058, + "num_tokens": 943332625.0, + "step": 35051 + }, + { + "epoch": 2.890409829306506, + "grad_norm": 2.1309726238250732, + "learning_rate": 1e-06, + "loss": 0.6507, + "mean_token_accuracy": 0.7815977334976196, + "num_tokens": 943362277.0, + "step": 35052 + }, + { + "epoch": 2.890492289931558, + "grad_norm": 2.000340700149536, + "learning_rate": 1e-06, + "loss": 0.5661, + "mean_token_accuracy": 0.8141893148422241, + "num_tokens": 943392848.0, + "step": 35053 + }, + { + "epoch": 2.8905747505566093, + "grad_norm": 2.072596788406372, + "learning_rate": 1e-06, + "loss": 0.5449, + "mean_token_accuracy": 0.8066635131835938, + "num_tokens": 943422188.0, + "step": 35054 + }, + { + "epoch": 2.8906572111816606, + "grad_norm": 2.0758938789367676, + "learning_rate": 1e-06, + "loss": 0.6557, + "mean_token_accuracy": 0.7885544896125793, + "num_tokens": 943454608.0, + "step": 35055 + }, + { + "epoch": 2.8907396718067124, + "grad_norm": 2.224886655807495, + "learning_rate": 1e-06, + "loss": 0.5758, + "mean_token_accuracy": 0.8086200952529907, + "num_tokens": 943479254.0, + "step": 35056 + }, + { + "epoch": 2.890822132431764, + "grad_norm": 2.000148296356201, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7591887712478638, + "num_tokens": 943514690.0, + "step": 35057 + }, + { + "epoch": 2.890904593056815, + "grad_norm": 2.3100428581237793, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7642384767532349, + "num_tokens": 943541564.0, + "step": 35058 + }, + { + "epoch": 2.890987053681867, + "grad_norm": 2.3752360343933105, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7633378505706787, + "num_tokens": 943566484.0, + "step": 35059 + }, + { + "epoch": 2.8910695143069187, + "grad_norm": 2.1890523433685303, + "learning_rate": 1e-06, + "loss": 0.5531, + "mean_token_accuracy": 0.8023607134819031, + "num_tokens": 943596618.0, + "step": 35060 + }, + { + "epoch": 2.89115197493197, + "grad_norm": 2.454928159713745, + "learning_rate": 1e-06, + "loss": 0.7082, + "mean_token_accuracy": 0.7564641237258911, + "num_tokens": 943622531.0, + "step": 35061 + }, + { + "epoch": 2.8912344355570214, + "grad_norm": 1.9920916557312012, + "learning_rate": 1e-06, + "loss": 0.6308, + "mean_token_accuracy": 0.7926949262619019, + "num_tokens": 943654281.0, + "step": 35062 + }, + { + "epoch": 2.891316896182073, + "grad_norm": 1.9832006692886353, + "learning_rate": 1e-06, + "loss": 0.7695, + "mean_token_accuracy": 0.7515217661857605, + "num_tokens": 943689497.0, + "step": 35063 + }, + { + "epoch": 2.8913993568071246, + "grad_norm": 2.2788116931915283, + "learning_rate": 1e-06, + "loss": 0.6588, + "mean_token_accuracy": 0.7842690944671631, + "num_tokens": 943714399.0, + "step": 35064 + }, + { + "epoch": 2.891481817432176, + "grad_norm": 2.284149408340454, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7595157623291016, + "num_tokens": 943741134.0, + "step": 35065 + }, + { + "epoch": 2.8915642780572277, + "grad_norm": 2.234318256378174, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7384841442108154, + "num_tokens": 943768952.0, + "step": 35066 + }, + { + "epoch": 2.891646738682279, + "grad_norm": 1.693679690361023, + "learning_rate": 1e-06, + "loss": 0.5366, + "mean_token_accuracy": 0.8167963027954102, + "num_tokens": 943812091.0, + "step": 35067 + }, + { + "epoch": 2.891729199307331, + "grad_norm": 2.145512580871582, + "learning_rate": 1e-06, + "loss": 0.7592, + "mean_token_accuracy": 0.7479029297828674, + "num_tokens": 943843493.0, + "step": 35068 + }, + { + "epoch": 2.891811659932382, + "grad_norm": 2.062462091445923, + "learning_rate": 1e-06, + "loss": 0.6879, + "mean_token_accuracy": 0.7660163640975952, + "num_tokens": 943876108.0, + "step": 35069 + }, + { + "epoch": 2.891894120557434, + "grad_norm": 2.0218403339385986, + "learning_rate": 1e-06, + "loss": 0.6541, + "mean_token_accuracy": 0.7877224683761597, + "num_tokens": 943908464.0, + "step": 35070 + }, + { + "epoch": 2.8919765811824854, + "grad_norm": 1.8096994161605835, + "learning_rate": 1e-06, + "loss": 0.5391, + "mean_token_accuracy": 0.8200547695159912, + "num_tokens": 943942495.0, + "step": 35071 + }, + { + "epoch": 2.8920590418075367, + "grad_norm": 1.9777278900146484, + "learning_rate": 1e-06, + "loss": 0.6753, + "mean_token_accuracy": 0.7801305055618286, + "num_tokens": 943973270.0, + "step": 35072 + }, + { + "epoch": 2.8921415024325885, + "grad_norm": 2.2987823486328125, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.7771697640419006, + "num_tokens": 943997659.0, + "step": 35073 + }, + { + "epoch": 2.89222396305764, + "grad_norm": 2.075502395629883, + "learning_rate": 1e-06, + "loss": 0.5679, + "mean_token_accuracy": 0.8082771897315979, + "num_tokens": 944028002.0, + "step": 35074 + }, + { + "epoch": 2.8923064236826916, + "grad_norm": 1.979573130607605, + "learning_rate": 1e-06, + "loss": 0.7587, + "mean_token_accuracy": 0.7550208568572998, + "num_tokens": 944063743.0, + "step": 35075 + }, + { + "epoch": 2.892388884307743, + "grad_norm": 1.7796393632888794, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7781212329864502, + "num_tokens": 944104763.0, + "step": 35076 + }, + { + "epoch": 2.892471344932795, + "grad_norm": 2.0789897441864014, + "learning_rate": 1e-06, + "loss": 0.5549, + "mean_token_accuracy": 0.804190456867218, + "num_tokens": 944131613.0, + "step": 35077 + }, + { + "epoch": 2.892553805557846, + "grad_norm": 1.743966817855835, + "learning_rate": 1e-06, + "loss": 0.6472, + "mean_token_accuracy": 0.7904765605926514, + "num_tokens": 944170571.0, + "step": 35078 + }, + { + "epoch": 2.8926362661828975, + "grad_norm": 1.8851819038391113, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.762265145778656, + "num_tokens": 944208197.0, + "step": 35079 + }, + { + "epoch": 2.8927187268079493, + "grad_norm": 2.2526745796203613, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7497852444648743, + "num_tokens": 944236151.0, + "step": 35080 + }, + { + "epoch": 2.8928011874330006, + "grad_norm": 2.004584789276123, + "learning_rate": 1e-06, + "loss": 0.5719, + "mean_token_accuracy": 0.8140294551849365, + "num_tokens": 944265945.0, + "step": 35081 + }, + { + "epoch": 2.8928836480580524, + "grad_norm": 2.1497271060943604, + "learning_rate": 1e-06, + "loss": 0.6124, + "mean_token_accuracy": 0.7992841005325317, + "num_tokens": 944290706.0, + "step": 35082 + }, + { + "epoch": 2.8929661086831038, + "grad_norm": 2.1757664680480957, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7721453905105591, + "num_tokens": 944317347.0, + "step": 35083 + }, + { + "epoch": 2.8930485693081556, + "grad_norm": 2.3256895542144775, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.721116304397583, + "num_tokens": 944347576.0, + "step": 35084 + }, + { + "epoch": 2.893131029933207, + "grad_norm": 2.2814807891845703, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7244039177894592, + "num_tokens": 944375447.0, + "step": 35085 + }, + { + "epoch": 2.8932134905582583, + "grad_norm": 1.8430224657058716, + "learning_rate": 1e-06, + "loss": 0.7246, + "mean_token_accuracy": 0.7679548859596252, + "num_tokens": 944413648.0, + "step": 35086 + }, + { + "epoch": 2.89329595118331, + "grad_norm": 1.9604806900024414, + "learning_rate": 1e-06, + "loss": 0.7076, + "mean_token_accuracy": 0.7701020240783691, + "num_tokens": 944447497.0, + "step": 35087 + }, + { + "epoch": 2.8933784118083614, + "grad_norm": 1.9781068563461304, + "learning_rate": 1e-06, + "loss": 0.5956, + "mean_token_accuracy": 0.804577112197876, + "num_tokens": 944477658.0, + "step": 35088 + }, + { + "epoch": 2.893460872433413, + "grad_norm": 2.002560615539551, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7688095569610596, + "num_tokens": 944508304.0, + "step": 35089 + }, + { + "epoch": 2.8935433330584646, + "grad_norm": 2.2222094535827637, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7608792781829834, + "num_tokens": 944537288.0, + "step": 35090 + }, + { + "epoch": 2.8936257936835164, + "grad_norm": 2.339186191558838, + "learning_rate": 1e-06, + "loss": 0.7451, + "mean_token_accuracy": 0.7509796619415283, + "num_tokens": 944563872.0, + "step": 35091 + }, + { + "epoch": 2.8937082543085677, + "grad_norm": 2.195617198944092, + "learning_rate": 1e-06, + "loss": 0.6537, + "mean_token_accuracy": 0.7876052856445312, + "num_tokens": 944590264.0, + "step": 35092 + }, + { + "epoch": 2.893790714933619, + "grad_norm": 2.1375043392181396, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.731473445892334, + "num_tokens": 944622409.0, + "step": 35093 + }, + { + "epoch": 2.893873175558671, + "grad_norm": 2.1901373863220215, + "learning_rate": 1e-06, + "loss": 0.6841, + "mean_token_accuracy": 0.7760636806488037, + "num_tokens": 944652267.0, + "step": 35094 + }, + { + "epoch": 2.893955636183722, + "grad_norm": 2.038620948791504, + "learning_rate": 1e-06, + "loss": 0.5975, + "mean_token_accuracy": 0.8058997392654419, + "num_tokens": 944684034.0, + "step": 35095 + }, + { + "epoch": 2.8940380968087736, + "grad_norm": 1.8927339315414429, + "learning_rate": 1e-06, + "loss": 0.5434, + "mean_token_accuracy": 0.8134156465530396, + "num_tokens": 944718742.0, + "step": 35096 + }, + { + "epoch": 2.8941205574338253, + "grad_norm": 1.8343220949172974, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.7900758385658264, + "num_tokens": 944754571.0, + "step": 35097 + }, + { + "epoch": 2.894203018058877, + "grad_norm": 2.357508420944214, + "learning_rate": 1e-06, + "loss": 0.6018, + "mean_token_accuracy": 0.7918018102645874, + "num_tokens": 944780205.0, + "step": 35098 + }, + { + "epoch": 2.8942854786839285, + "grad_norm": 1.9534704685211182, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.764337420463562, + "num_tokens": 944819175.0, + "step": 35099 + }, + { + "epoch": 2.89436793930898, + "grad_norm": 1.8354555368423462, + "learning_rate": 1e-06, + "loss": 0.5445, + "mean_token_accuracy": 0.8138927221298218, + "num_tokens": 944854038.0, + "step": 35100 + }, + { + "epoch": 2.8944503999340316, + "grad_norm": 2.1317074298858643, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7296247482299805, + "num_tokens": 944883673.0, + "step": 35101 + }, + { + "epoch": 2.894532860559083, + "grad_norm": 1.986349105834961, + "learning_rate": 1e-06, + "loss": 0.694, + "mean_token_accuracy": 0.7796515822410583, + "num_tokens": 944914735.0, + "step": 35102 + }, + { + "epoch": 2.8946153211841343, + "grad_norm": 2.04319429397583, + "learning_rate": 1e-06, + "loss": 0.6729, + "mean_token_accuracy": 0.7896554470062256, + "num_tokens": 944945356.0, + "step": 35103 + }, + { + "epoch": 2.894697781809186, + "grad_norm": 1.8769127130508423, + "learning_rate": 1e-06, + "loss": 0.5307, + "mean_token_accuracy": 0.8200562596321106, + "num_tokens": 944980718.0, + "step": 35104 + }, + { + "epoch": 2.894780242434238, + "grad_norm": 2.2051844596862793, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7765052318572998, + "num_tokens": 945007902.0, + "step": 35105 + }, + { + "epoch": 2.8948627030592893, + "grad_norm": 2.0199530124664307, + "learning_rate": 1e-06, + "loss": 0.6547, + "mean_token_accuracy": 0.7829073667526245, + "num_tokens": 945039673.0, + "step": 35106 + }, + { + "epoch": 2.8949451636843406, + "grad_norm": 2.0968594551086426, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7642724514007568, + "num_tokens": 945069825.0, + "step": 35107 + }, + { + "epoch": 2.8950276243093924, + "grad_norm": 1.9001929759979248, + "learning_rate": 1e-06, + "loss": 0.648, + "mean_token_accuracy": 0.781737208366394, + "num_tokens": 945103763.0, + "step": 35108 + }, + { + "epoch": 2.8951100849344438, + "grad_norm": 1.850263237953186, + "learning_rate": 1e-06, + "loss": 0.5713, + "mean_token_accuracy": 0.810478687286377, + "num_tokens": 945140161.0, + "step": 35109 + }, + { + "epoch": 2.895192545559495, + "grad_norm": 2.2239251136779785, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7725613117218018, + "num_tokens": 945166366.0, + "step": 35110 + }, + { + "epoch": 2.895275006184547, + "grad_norm": 2.0844733715057373, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7609246969223022, + "num_tokens": 945195571.0, + "step": 35111 + }, + { + "epoch": 2.8953574668095983, + "grad_norm": 2.2317628860473633, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7396702170372009, + "num_tokens": 945223281.0, + "step": 35112 + }, + { + "epoch": 2.89543992743465, + "grad_norm": 2.4382731914520264, + "learning_rate": 1e-06, + "loss": 0.593, + "mean_token_accuracy": 0.7967752814292908, + "num_tokens": 945250469.0, + "step": 35113 + }, + { + "epoch": 2.8955223880597014, + "grad_norm": 2.027028799057007, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7521370053291321, + "num_tokens": 945282695.0, + "step": 35114 + }, + { + "epoch": 2.895604848684753, + "grad_norm": 2.3942859172821045, + "learning_rate": 1e-06, + "loss": 0.7217, + "mean_token_accuracy": 0.7623629570007324, + "num_tokens": 945306927.0, + "step": 35115 + }, + { + "epoch": 2.8956873093098046, + "grad_norm": 2.0871334075927734, + "learning_rate": 1e-06, + "loss": 0.5614, + "mean_token_accuracy": 0.8073574900627136, + "num_tokens": 945337130.0, + "step": 35116 + }, + { + "epoch": 2.895769769934856, + "grad_norm": 2.0467731952667236, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.777475893497467, + "num_tokens": 945368476.0, + "step": 35117 + }, + { + "epoch": 2.8958522305599077, + "grad_norm": 1.835494875907898, + "learning_rate": 1e-06, + "loss": 0.5321, + "mean_token_accuracy": 0.8218715786933899, + "num_tokens": 945401482.0, + "step": 35118 + }, + { + "epoch": 2.895934691184959, + "grad_norm": 2.0364527702331543, + "learning_rate": 1e-06, + "loss": 0.7031, + "mean_token_accuracy": 0.7719072103500366, + "num_tokens": 945434193.0, + "step": 35119 + }, + { + "epoch": 2.896017151810011, + "grad_norm": 2.0306646823883057, + "learning_rate": 1e-06, + "loss": 0.5734, + "mean_token_accuracy": 0.810894250869751, + "num_tokens": 945468354.0, + "step": 35120 + }, + { + "epoch": 2.896099612435062, + "grad_norm": 2.3225371837615967, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7513471841812134, + "num_tokens": 945494784.0, + "step": 35121 + }, + { + "epoch": 2.896182073060114, + "grad_norm": 2.175631523132324, + "learning_rate": 1e-06, + "loss": 0.6346, + "mean_token_accuracy": 0.7914910316467285, + "num_tokens": 945522372.0, + "step": 35122 + }, + { + "epoch": 2.8962645336851653, + "grad_norm": 2.0789902210235596, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7611085176467896, + "num_tokens": 945553012.0, + "step": 35123 + }, + { + "epoch": 2.8963469943102167, + "grad_norm": 2.2650346755981445, + "learning_rate": 1e-06, + "loss": 0.7368, + "mean_token_accuracy": 0.7624354958534241, + "num_tokens": 945582312.0, + "step": 35124 + }, + { + "epoch": 2.8964294549352685, + "grad_norm": 1.8470460176467896, + "learning_rate": 1e-06, + "loss": 0.4851, + "mean_token_accuracy": 0.8317257165908813, + "num_tokens": 945615313.0, + "step": 35125 + }, + { + "epoch": 2.89651191556032, + "grad_norm": 1.8703283071517944, + "learning_rate": 1e-06, + "loss": 0.7496, + "mean_token_accuracy": 0.7649847269058228, + "num_tokens": 945650999.0, + "step": 35126 + }, + { + "epoch": 2.8965943761853716, + "grad_norm": 2.1383397579193115, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7781676054000854, + "num_tokens": 945679602.0, + "step": 35127 + }, + { + "epoch": 2.896676836810423, + "grad_norm": 2.237055540084839, + "learning_rate": 1e-06, + "loss": 0.8032, + "mean_token_accuracy": 0.7362476587295532, + "num_tokens": 945706664.0, + "step": 35128 + }, + { + "epoch": 2.896759297435475, + "grad_norm": 2.3016607761383057, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.7463264465332031, + "num_tokens": 945734221.0, + "step": 35129 + }, + { + "epoch": 2.896841758060526, + "grad_norm": 2.002375841140747, + "learning_rate": 1e-06, + "loss": 0.525, + "mean_token_accuracy": 0.8216183185577393, + "num_tokens": 945763620.0, + "step": 35130 + }, + { + "epoch": 2.8969242186855775, + "grad_norm": 2.014918327331543, + "learning_rate": 1e-06, + "loss": 0.6801, + "mean_token_accuracy": 0.7747868895530701, + "num_tokens": 945795187.0, + "step": 35131 + }, + { + "epoch": 2.8970066793106293, + "grad_norm": 2.3903768062591553, + "learning_rate": 1e-06, + "loss": 0.6477, + "mean_token_accuracy": 0.7858824729919434, + "num_tokens": 945817804.0, + "step": 35132 + }, + { + "epoch": 2.8970891399356806, + "grad_norm": 2.1243903636932373, + "learning_rate": 1e-06, + "loss": 0.5925, + "mean_token_accuracy": 0.7982598543167114, + "num_tokens": 945845098.0, + "step": 35133 + }, + { + "epoch": 2.897171600560732, + "grad_norm": 2.107919454574585, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7673316597938538, + "num_tokens": 945876270.0, + "step": 35134 + }, + { + "epoch": 2.8972540611857838, + "grad_norm": 2.296574592590332, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7401226758956909, + "num_tokens": 945904423.0, + "step": 35135 + }, + { + "epoch": 2.8973365218108356, + "grad_norm": 2.0660362243652344, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7650488615036011, + "num_tokens": 945935480.0, + "step": 35136 + }, + { + "epoch": 2.897418982435887, + "grad_norm": 2.097805976867676, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7629101276397705, + "num_tokens": 945963471.0, + "step": 35137 + }, + { + "epoch": 2.8975014430609383, + "grad_norm": 1.8190025091171265, + "learning_rate": 1e-06, + "loss": 0.6035, + "mean_token_accuracy": 0.797143280506134, + "num_tokens": 946000582.0, + "step": 35138 + }, + { + "epoch": 2.89758390368599, + "grad_norm": 2.1299986839294434, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.75150465965271, + "num_tokens": 946031691.0, + "step": 35139 + }, + { + "epoch": 2.8976663643110414, + "grad_norm": 2.041743755340576, + "learning_rate": 1e-06, + "loss": 0.6457, + "mean_token_accuracy": 0.7887375354766846, + "num_tokens": 946060942.0, + "step": 35140 + }, + { + "epoch": 2.8977488249360928, + "grad_norm": 1.7170335054397583, + "learning_rate": 1e-06, + "loss": 0.5946, + "mean_token_accuracy": 0.8093778491020203, + "num_tokens": 946099442.0, + "step": 35141 + }, + { + "epoch": 2.8978312855611446, + "grad_norm": 2.0661420822143555, + "learning_rate": 1e-06, + "loss": 0.8155, + "mean_token_accuracy": 0.7393856644630432, + "num_tokens": 946133009.0, + "step": 35142 + }, + { + "epoch": 2.8979137461861963, + "grad_norm": 2.4654150009155273, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7448124885559082, + "num_tokens": 946158832.0, + "step": 35143 + }, + { + "epoch": 2.8979962068112477, + "grad_norm": 2.3749396800994873, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7369415760040283, + "num_tokens": 946184725.0, + "step": 35144 + }, + { + "epoch": 2.898078667436299, + "grad_norm": 1.9492624998092651, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.7832641005516052, + "num_tokens": 946219346.0, + "step": 35145 + }, + { + "epoch": 2.898161128061351, + "grad_norm": 1.9355378150939941, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7902494072914124, + "num_tokens": 946250511.0, + "step": 35146 + }, + { + "epoch": 2.898243588686402, + "grad_norm": 1.9390794038772583, + "learning_rate": 1e-06, + "loss": 0.5598, + "mean_token_accuracy": 0.8067364692687988, + "num_tokens": 946281595.0, + "step": 35147 + }, + { + "epoch": 2.8983260493114535, + "grad_norm": 1.9987901449203491, + "learning_rate": 1e-06, + "loss": 0.6853, + "mean_token_accuracy": 0.7847719788551331, + "num_tokens": 946316354.0, + "step": 35148 + }, + { + "epoch": 2.8984085099365053, + "grad_norm": 2.0930368900299072, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.7985104322433472, + "num_tokens": 946344296.0, + "step": 35149 + }, + { + "epoch": 2.898490970561557, + "grad_norm": 1.989872694015503, + "learning_rate": 1e-06, + "loss": 0.6279, + "mean_token_accuracy": 0.7892805337905884, + "num_tokens": 946376233.0, + "step": 35150 + }, + { + "epoch": 2.8985734311866085, + "grad_norm": 1.9999574422836304, + "learning_rate": 1e-06, + "loss": 0.611, + "mean_token_accuracy": 0.7839138507843018, + "num_tokens": 946407948.0, + "step": 35151 + }, + { + "epoch": 2.89865589181166, + "grad_norm": 2.0814898014068604, + "learning_rate": 1e-06, + "loss": 0.654, + "mean_token_accuracy": 0.782443642616272, + "num_tokens": 946439170.0, + "step": 35152 + }, + { + "epoch": 2.8987383524367116, + "grad_norm": 2.2215850353240967, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.7515056133270264, + "num_tokens": 946465213.0, + "step": 35153 + }, + { + "epoch": 2.898820813061763, + "grad_norm": 2.054802894592285, + "learning_rate": 1e-06, + "loss": 0.6568, + "mean_token_accuracy": 0.7859301567077637, + "num_tokens": 946495507.0, + "step": 35154 + }, + { + "epoch": 2.8989032736868143, + "grad_norm": 2.1558001041412354, + "learning_rate": 1e-06, + "loss": 0.7251, + "mean_token_accuracy": 0.7552975416183472, + "num_tokens": 946525317.0, + "step": 35155 + }, + { + "epoch": 2.898985734311866, + "grad_norm": 2.2500081062316895, + "learning_rate": 1e-06, + "loss": 0.6755, + "mean_token_accuracy": 0.7771407961845398, + "num_tokens": 946551834.0, + "step": 35156 + }, + { + "epoch": 2.8990681949369175, + "grad_norm": 2.299058437347412, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7664836049079895, + "num_tokens": 946579316.0, + "step": 35157 + }, + { + "epoch": 2.8991506555619693, + "grad_norm": 2.1316258907318115, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7790036201477051, + "num_tokens": 946607894.0, + "step": 35158 + }, + { + "epoch": 2.8992331161870206, + "grad_norm": 2.087733268737793, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7407285571098328, + "num_tokens": 946640346.0, + "step": 35159 + }, + { + "epoch": 2.8993155768120724, + "grad_norm": 1.9213414192199707, + "learning_rate": 1e-06, + "loss": 0.7298, + "mean_token_accuracy": 0.7626387476921082, + "num_tokens": 946676993.0, + "step": 35160 + }, + { + "epoch": 2.8993980374371238, + "grad_norm": 2.0362203121185303, + "learning_rate": 1e-06, + "loss": 0.6344, + "mean_token_accuracy": 0.7861930131912231, + "num_tokens": 946707082.0, + "step": 35161 + }, + { + "epoch": 2.899480498062175, + "grad_norm": 1.947265386581421, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.8000826835632324, + "num_tokens": 946740043.0, + "step": 35162 + }, + { + "epoch": 2.899562958687227, + "grad_norm": 2.4808688163757324, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7617648243904114, + "num_tokens": 946762020.0, + "step": 35163 + }, + { + "epoch": 2.8996454193122783, + "grad_norm": 2.073702335357666, + "learning_rate": 1e-06, + "loss": 0.7319, + "mean_token_accuracy": 0.7645289301872253, + "num_tokens": 946794600.0, + "step": 35164 + }, + { + "epoch": 2.89972787993733, + "grad_norm": 2.0060787200927734, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.7632800340652466, + "num_tokens": 946824492.0, + "step": 35165 + }, + { + "epoch": 2.8998103405623814, + "grad_norm": 2.2818965911865234, + "learning_rate": 1e-06, + "loss": 0.6791, + "mean_token_accuracy": 0.7671229243278503, + "num_tokens": 946852456.0, + "step": 35166 + }, + { + "epoch": 2.899892801187433, + "grad_norm": 2.321582555770874, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7720794677734375, + "num_tokens": 946879058.0, + "step": 35167 + }, + { + "epoch": 2.8999752618124846, + "grad_norm": 2.1689436435699463, + "learning_rate": 1e-06, + "loss": 0.5596, + "mean_token_accuracy": 0.8036642074584961, + "num_tokens": 946909384.0, + "step": 35168 + }, + { + "epoch": 2.900057722437536, + "grad_norm": 2.1743054389953613, + "learning_rate": 1e-06, + "loss": 0.6009, + "mean_token_accuracy": 0.8011065125465393, + "num_tokens": 946938530.0, + "step": 35169 + }, + { + "epoch": 2.9001401830625877, + "grad_norm": 2.3958089351654053, + "learning_rate": 1e-06, + "loss": 0.6586, + "mean_token_accuracy": 0.7817749381065369, + "num_tokens": 946963882.0, + "step": 35170 + }, + { + "epoch": 2.900222643687639, + "grad_norm": 2.1447598934173584, + "learning_rate": 1e-06, + "loss": 0.7543, + "mean_token_accuracy": 0.7572096586227417, + "num_tokens": 946995165.0, + "step": 35171 + }, + { + "epoch": 2.900305104312691, + "grad_norm": 2.048699378967285, + "learning_rate": 1e-06, + "loss": 0.6859, + "mean_token_accuracy": 0.7666524648666382, + "num_tokens": 947025954.0, + "step": 35172 + }, + { + "epoch": 2.900387564937742, + "grad_norm": 1.8144887685775757, + "learning_rate": 1e-06, + "loss": 0.5489, + "mean_token_accuracy": 0.8196463584899902, + "num_tokens": 947062588.0, + "step": 35173 + }, + { + "epoch": 2.900470025562794, + "grad_norm": 1.8249797821044922, + "learning_rate": 1e-06, + "loss": 0.6525, + "mean_token_accuracy": 0.7848213911056519, + "num_tokens": 947096435.0, + "step": 35174 + }, + { + "epoch": 2.9005524861878453, + "grad_norm": 2.1568617820739746, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7453939914703369, + "num_tokens": 947124928.0, + "step": 35175 + }, + { + "epoch": 2.9006349468128967, + "grad_norm": 2.2901771068573, + "learning_rate": 1e-06, + "loss": 0.6175, + "mean_token_accuracy": 0.7935218811035156, + "num_tokens": 947150943.0, + "step": 35176 + }, + { + "epoch": 2.9007174074379485, + "grad_norm": 2.299588441848755, + "learning_rate": 1e-06, + "loss": 0.657, + "mean_token_accuracy": 0.7791234254837036, + "num_tokens": 947174439.0, + "step": 35177 + }, + { + "epoch": 2.900799868063, + "grad_norm": 2.1604185104370117, + "learning_rate": 1e-06, + "loss": 0.636, + "mean_token_accuracy": 0.7885225415229797, + "num_tokens": 947202511.0, + "step": 35178 + }, + { + "epoch": 2.900882328688051, + "grad_norm": 2.2024049758911133, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7822128534317017, + "num_tokens": 947232139.0, + "step": 35179 + }, + { + "epoch": 2.900964789313103, + "grad_norm": 1.7709836959838867, + "learning_rate": 1e-06, + "loss": 0.6369, + "mean_token_accuracy": 0.7879215478897095, + "num_tokens": 947268095.0, + "step": 35180 + }, + { + "epoch": 2.9010472499381548, + "grad_norm": 2.007225275039673, + "learning_rate": 1e-06, + "loss": 0.7674, + "mean_token_accuracy": 0.7597213983535767, + "num_tokens": 947301739.0, + "step": 35181 + }, + { + "epoch": 2.901129710563206, + "grad_norm": 2.010758638381958, + "learning_rate": 1e-06, + "loss": 0.5654, + "mean_token_accuracy": 0.815669059753418, + "num_tokens": 947331562.0, + "step": 35182 + }, + { + "epoch": 2.9012121711882575, + "grad_norm": 2.102605104446411, + "learning_rate": 1e-06, + "loss": 0.6955, + "mean_token_accuracy": 0.7706807851791382, + "num_tokens": 947361418.0, + "step": 35183 + }, + { + "epoch": 2.9012946318133093, + "grad_norm": 1.9673798084259033, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7806621789932251, + "num_tokens": 947392521.0, + "step": 35184 + }, + { + "epoch": 2.9013770924383606, + "grad_norm": 2.2654078006744385, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7440491318702698, + "num_tokens": 947419149.0, + "step": 35185 + }, + { + "epoch": 2.901459553063412, + "grad_norm": 2.0173988342285156, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7576828598976135, + "num_tokens": 947453094.0, + "step": 35186 + }, + { + "epoch": 2.9015420136884638, + "grad_norm": 2.1559224128723145, + "learning_rate": 1e-06, + "loss": 0.7274, + "mean_token_accuracy": 0.7684881687164307, + "num_tokens": 947482515.0, + "step": 35187 + }, + { + "epoch": 2.9016244743135156, + "grad_norm": 1.9369597434997559, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.788758397102356, + "num_tokens": 947512915.0, + "step": 35188 + }, + { + "epoch": 2.901706934938567, + "grad_norm": 1.7463876008987427, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7654008865356445, + "num_tokens": 947552037.0, + "step": 35189 + }, + { + "epoch": 2.9017893955636183, + "grad_norm": 1.881730556488037, + "learning_rate": 1e-06, + "loss": 0.7177, + "mean_token_accuracy": 0.7647384405136108, + "num_tokens": 947588972.0, + "step": 35190 + }, + { + "epoch": 2.90187185618867, + "grad_norm": 2.0216760635375977, + "learning_rate": 1e-06, + "loss": 0.7619, + "mean_token_accuracy": 0.7520580291748047, + "num_tokens": 947624772.0, + "step": 35191 + }, + { + "epoch": 2.9019543168137214, + "grad_norm": 1.8627941608428955, + "learning_rate": 1e-06, + "loss": 0.692, + "mean_token_accuracy": 0.7815732359886169, + "num_tokens": 947660697.0, + "step": 35192 + }, + { + "epoch": 2.9020367774387728, + "grad_norm": 2.2906250953674316, + "learning_rate": 1e-06, + "loss": 0.7758, + "mean_token_accuracy": 0.7505570650100708, + "num_tokens": 947688980.0, + "step": 35193 + }, + { + "epoch": 2.9021192380638245, + "grad_norm": 1.8913410902023315, + "learning_rate": 1e-06, + "loss": 0.5407, + "mean_token_accuracy": 0.8118152618408203, + "num_tokens": 947722287.0, + "step": 35194 + }, + { + "epoch": 2.902201698688876, + "grad_norm": 2.2204599380493164, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7230452299118042, + "num_tokens": 947750551.0, + "step": 35195 + }, + { + "epoch": 2.9022841593139277, + "grad_norm": 2.1385200023651123, + "learning_rate": 1e-06, + "loss": 0.6768, + "mean_token_accuracy": 0.7822626233100891, + "num_tokens": 947777411.0, + "step": 35196 + }, + { + "epoch": 2.902366619938979, + "grad_norm": 2.1057002544403076, + "learning_rate": 1e-06, + "loss": 0.6627, + "mean_token_accuracy": 0.7776166200637817, + "num_tokens": 947806602.0, + "step": 35197 + }, + { + "epoch": 2.902449080564031, + "grad_norm": 1.9494197368621826, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7718808650970459, + "num_tokens": 947841735.0, + "step": 35198 + }, + { + "epoch": 2.902531541189082, + "grad_norm": 2.2183516025543213, + "learning_rate": 1e-06, + "loss": 0.7632, + "mean_token_accuracy": 0.7516292333602905, + "num_tokens": 947869355.0, + "step": 35199 + }, + { + "epoch": 2.9026140018141335, + "grad_norm": 2.109555244445801, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7774113416671753, + "num_tokens": 947898033.0, + "step": 35200 + }, + { + "epoch": 2.9026964624391853, + "grad_norm": 2.1021201610565186, + "learning_rate": 1e-06, + "loss": 0.7268, + "mean_token_accuracy": 0.7575763463973999, + "num_tokens": 947932948.0, + "step": 35201 + }, + { + "epoch": 2.9027789230642367, + "grad_norm": 2.060001850128174, + "learning_rate": 1e-06, + "loss": 0.5711, + "mean_token_accuracy": 0.8027189373970032, + "num_tokens": 947963624.0, + "step": 35202 + }, + { + "epoch": 2.9028613836892885, + "grad_norm": 1.900469183921814, + "learning_rate": 1e-06, + "loss": 0.6612, + "mean_token_accuracy": 0.7757505178451538, + "num_tokens": 948001420.0, + "step": 35203 + }, + { + "epoch": 2.90294384431434, + "grad_norm": 1.9033269882202148, + "learning_rate": 1e-06, + "loss": 0.6473, + "mean_token_accuracy": 0.7844914197921753, + "num_tokens": 948035063.0, + "step": 35204 + }, + { + "epoch": 2.9030263049393916, + "grad_norm": 2.0053839683532715, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7746009826660156, + "num_tokens": 948065276.0, + "step": 35205 + }, + { + "epoch": 2.903108765564443, + "grad_norm": 1.9614793062210083, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7784098386764526, + "num_tokens": 948099785.0, + "step": 35206 + }, + { + "epoch": 2.9031912261894943, + "grad_norm": 2.0745999813079834, + "learning_rate": 1e-06, + "loss": 0.627, + "mean_token_accuracy": 0.7934188842773438, + "num_tokens": 948130986.0, + "step": 35207 + }, + { + "epoch": 2.903273686814546, + "grad_norm": 2.0606768131256104, + "learning_rate": 1e-06, + "loss": 0.7379, + "mean_token_accuracy": 0.7596111297607422, + "num_tokens": 948163338.0, + "step": 35208 + }, + { + "epoch": 2.9033561474395975, + "grad_norm": 2.273167371749878, + "learning_rate": 1e-06, + "loss": 0.7041, + "mean_token_accuracy": 0.7738475799560547, + "num_tokens": 948187553.0, + "step": 35209 + }, + { + "epoch": 2.9034386080646493, + "grad_norm": 2.36845326423645, + "learning_rate": 1e-06, + "loss": 0.6243, + "mean_token_accuracy": 0.7939424514770508, + "num_tokens": 948212959.0, + "step": 35210 + }, + { + "epoch": 2.9035210686897006, + "grad_norm": 2.219594955444336, + "learning_rate": 1e-06, + "loss": 0.6734, + "mean_token_accuracy": 0.7727134823799133, + "num_tokens": 948241383.0, + "step": 35211 + }, + { + "epoch": 2.9036035293147524, + "grad_norm": 2.0511887073516846, + "learning_rate": 1e-06, + "loss": 0.6597, + "mean_token_accuracy": 0.7841641902923584, + "num_tokens": 948271023.0, + "step": 35212 + }, + { + "epoch": 2.9036859899398038, + "grad_norm": 2.1604907512664795, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7496011853218079, + "num_tokens": 948299476.0, + "step": 35213 + }, + { + "epoch": 2.903768450564855, + "grad_norm": 2.128835916519165, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.7929039597511292, + "num_tokens": 948329075.0, + "step": 35214 + }, + { + "epoch": 2.903850911189907, + "grad_norm": 2.1257283687591553, + "learning_rate": 1e-06, + "loss": 0.6733, + "mean_token_accuracy": 0.7778744101524353, + "num_tokens": 948359717.0, + "step": 35215 + }, + { + "epoch": 2.9039333718149583, + "grad_norm": 2.1958847045898438, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7564119696617126, + "num_tokens": 948388433.0, + "step": 35216 + }, + { + "epoch": 2.90401583244001, + "grad_norm": 2.198914051055908, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7430771589279175, + "num_tokens": 948416210.0, + "step": 35217 + }, + { + "epoch": 2.9040982930650614, + "grad_norm": 2.114560604095459, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7748162746429443, + "num_tokens": 948446279.0, + "step": 35218 + }, + { + "epoch": 2.904180753690113, + "grad_norm": 1.797919750213623, + "learning_rate": 1e-06, + "loss": 0.5854, + "mean_token_accuracy": 0.7979974150657654, + "num_tokens": 948480773.0, + "step": 35219 + }, + { + "epoch": 2.9042632143151645, + "grad_norm": 2.095252513885498, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.7747246623039246, + "num_tokens": 948514663.0, + "step": 35220 + }, + { + "epoch": 2.904345674940216, + "grad_norm": 1.8407975435256958, + "learning_rate": 1e-06, + "loss": 0.5928, + "mean_token_accuracy": 0.8029814958572388, + "num_tokens": 948549343.0, + "step": 35221 + }, + { + "epoch": 2.9044281355652677, + "grad_norm": 1.8709371089935303, + "learning_rate": 1e-06, + "loss": 0.6744, + "mean_token_accuracy": 0.7705403566360474, + "num_tokens": 948583913.0, + "step": 35222 + }, + { + "epoch": 2.904510596190319, + "grad_norm": 2.19533109664917, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7372087240219116, + "num_tokens": 948611830.0, + "step": 35223 + }, + { + "epoch": 2.9045930568153704, + "grad_norm": 2.1280956268310547, + "learning_rate": 1e-06, + "loss": 0.67, + "mean_token_accuracy": 0.776848316192627, + "num_tokens": 948640979.0, + "step": 35224 + }, + { + "epoch": 2.904675517440422, + "grad_norm": 1.8867943286895752, + "learning_rate": 1e-06, + "loss": 0.6998, + "mean_token_accuracy": 0.7676184177398682, + "num_tokens": 948676854.0, + "step": 35225 + }, + { + "epoch": 2.904757978065474, + "grad_norm": 2.132857322692871, + "learning_rate": 1e-06, + "loss": 0.5351, + "mean_token_accuracy": 0.8161928653717041, + "num_tokens": 948706264.0, + "step": 35226 + }, + { + "epoch": 2.9048404386905253, + "grad_norm": 2.037038564682007, + "learning_rate": 1e-06, + "loss": 0.6664, + "mean_token_accuracy": 0.773115873336792, + "num_tokens": 948735806.0, + "step": 35227 + }, + { + "epoch": 2.9049228993155767, + "grad_norm": 1.8888410329818726, + "learning_rate": 1e-06, + "loss": 0.6703, + "mean_token_accuracy": 0.7808178663253784, + "num_tokens": 948773817.0, + "step": 35228 + }, + { + "epoch": 2.9050053599406285, + "grad_norm": 2.2140188217163086, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.7881975769996643, + "num_tokens": 948799838.0, + "step": 35229 + }, + { + "epoch": 2.90508782056568, + "grad_norm": 2.033510684967041, + "learning_rate": 1e-06, + "loss": 0.6491, + "mean_token_accuracy": 0.7723957896232605, + "num_tokens": 948829960.0, + "step": 35230 + }, + { + "epoch": 2.905170281190731, + "grad_norm": 2.0495691299438477, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7445031404495239, + "num_tokens": 948861972.0, + "step": 35231 + }, + { + "epoch": 2.905252741815783, + "grad_norm": 2.081678867340088, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7424883842468262, + "num_tokens": 948893677.0, + "step": 35232 + }, + { + "epoch": 2.9053352024408348, + "grad_norm": 2.121596574783325, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7639211416244507, + "num_tokens": 948921383.0, + "step": 35233 + }, + { + "epoch": 2.905417663065886, + "grad_norm": 1.8291070461273193, + "learning_rate": 1e-06, + "loss": 0.528, + "mean_token_accuracy": 0.8184224367141724, + "num_tokens": 948960986.0, + "step": 35234 + }, + { + "epoch": 2.9055001236909375, + "grad_norm": 1.9967832565307617, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.7603174448013306, + "num_tokens": 948991781.0, + "step": 35235 + }, + { + "epoch": 2.9055825843159893, + "grad_norm": 2.1876628398895264, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7412171363830566, + "num_tokens": 949018489.0, + "step": 35236 + }, + { + "epoch": 2.9056650449410406, + "grad_norm": 2.1772453784942627, + "learning_rate": 1e-06, + "loss": 0.6919, + "mean_token_accuracy": 0.7755565643310547, + "num_tokens": 949046154.0, + "step": 35237 + }, + { + "epoch": 2.905747505566092, + "grad_norm": 2.0967743396759033, + "learning_rate": 1e-06, + "loss": 0.5967, + "mean_token_accuracy": 0.8033046722412109, + "num_tokens": 949074725.0, + "step": 35238 + }, + { + "epoch": 2.9058299661911438, + "grad_norm": 2.36053204536438, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7829480171203613, + "num_tokens": 949098609.0, + "step": 35239 + }, + { + "epoch": 2.905912426816195, + "grad_norm": 1.9514195919036865, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.7927048206329346, + "num_tokens": 949131979.0, + "step": 35240 + }, + { + "epoch": 2.905994887441247, + "grad_norm": 2.4870333671569824, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7385296821594238, + "num_tokens": 949154879.0, + "step": 35241 + }, + { + "epoch": 2.9060773480662982, + "grad_norm": 2.0331859588623047, + "learning_rate": 1e-06, + "loss": 0.605, + "mean_token_accuracy": 0.7924048900604248, + "num_tokens": 949184594.0, + "step": 35242 + }, + { + "epoch": 2.90615980869135, + "grad_norm": 1.7440671920776367, + "learning_rate": 1e-06, + "loss": 0.7297, + "mean_token_accuracy": 0.764833927154541, + "num_tokens": 949225438.0, + "step": 35243 + }, + { + "epoch": 2.9062422693164014, + "grad_norm": 1.9387524127960205, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7929816246032715, + "num_tokens": 949259095.0, + "step": 35244 + }, + { + "epoch": 2.9063247299414527, + "grad_norm": 2.2921786308288574, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7458888292312622, + "num_tokens": 949285403.0, + "step": 35245 + }, + { + "epoch": 2.9064071905665045, + "grad_norm": 1.9771928787231445, + "learning_rate": 1e-06, + "loss": 0.6534, + "mean_token_accuracy": 0.7912570238113403, + "num_tokens": 949319713.0, + "step": 35246 + }, + { + "epoch": 2.906489651191556, + "grad_norm": 1.8813916444778442, + "learning_rate": 1e-06, + "loss": 0.6608, + "mean_token_accuracy": 0.7840508222579956, + "num_tokens": 949356909.0, + "step": 35247 + }, + { + "epoch": 2.9065721118166077, + "grad_norm": 2.0343575477600098, + "learning_rate": 1e-06, + "loss": 0.6366, + "mean_token_accuracy": 0.7908865809440613, + "num_tokens": 949387473.0, + "step": 35248 + }, + { + "epoch": 2.906654572441659, + "grad_norm": 1.804713487625122, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.76728355884552, + "num_tokens": 949424525.0, + "step": 35249 + }, + { + "epoch": 2.906737033066711, + "grad_norm": 2.224716901779175, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.744488537311554, + "num_tokens": 949455061.0, + "step": 35250 + }, + { + "epoch": 2.906819493691762, + "grad_norm": 1.9114280939102173, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7673231959342957, + "num_tokens": 949488342.0, + "step": 35251 + }, + { + "epoch": 2.9069019543168135, + "grad_norm": 2.367035150527954, + "learning_rate": 1e-06, + "loss": 0.6759, + "mean_token_accuracy": 0.7822697758674622, + "num_tokens": 949513057.0, + "step": 35252 + }, + { + "epoch": 2.9069844149418653, + "grad_norm": 2.1207878589630127, + "learning_rate": 1e-06, + "loss": 0.6555, + "mean_token_accuracy": 0.7851695418357849, + "num_tokens": 949542452.0, + "step": 35253 + }, + { + "epoch": 2.9070668755669167, + "grad_norm": 1.894028663635254, + "learning_rate": 1e-06, + "loss": 0.7241, + "mean_token_accuracy": 0.7663890719413757, + "num_tokens": 949579768.0, + "step": 35254 + }, + { + "epoch": 2.9071493361919685, + "grad_norm": 1.8556950092315674, + "learning_rate": 1e-06, + "loss": 0.6544, + "mean_token_accuracy": 0.7820467948913574, + "num_tokens": 949616846.0, + "step": 35255 + }, + { + "epoch": 2.90723179681702, + "grad_norm": 1.8669840097427368, + "learning_rate": 1e-06, + "loss": 0.5953, + "mean_token_accuracy": 0.7987948656082153, + "num_tokens": 949650316.0, + "step": 35256 + }, + { + "epoch": 2.9073142574420716, + "grad_norm": 2.147886276245117, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.7856153249740601, + "num_tokens": 949677699.0, + "step": 35257 + }, + { + "epoch": 2.907396718067123, + "grad_norm": 2.424050807952881, + "learning_rate": 1e-06, + "loss": 0.5732, + "mean_token_accuracy": 0.8127380609512329, + "num_tokens": 949703125.0, + "step": 35258 + }, + { + "epoch": 2.9074791786921743, + "grad_norm": 2.2808732986450195, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7713208794593811, + "num_tokens": 949728564.0, + "step": 35259 + }, + { + "epoch": 2.907561639317226, + "grad_norm": 2.045093536376953, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.7650561332702637, + "num_tokens": 949757939.0, + "step": 35260 + }, + { + "epoch": 2.9076440999422775, + "grad_norm": 1.9805811643600464, + "learning_rate": 1e-06, + "loss": 0.6198, + "mean_token_accuracy": 0.7995572090148926, + "num_tokens": 949790989.0, + "step": 35261 + }, + { + "epoch": 2.9077265605673293, + "grad_norm": 1.781574010848999, + "learning_rate": 1e-06, + "loss": 0.5361, + "mean_token_accuracy": 0.8240997791290283, + "num_tokens": 949827407.0, + "step": 35262 + }, + { + "epoch": 2.9078090211923806, + "grad_norm": 1.9950438737869263, + "learning_rate": 1e-06, + "loss": 0.5342, + "mean_token_accuracy": 0.808932900428772, + "num_tokens": 949859902.0, + "step": 35263 + }, + { + "epoch": 2.9078914818174324, + "grad_norm": 2.0647873878479004, + "learning_rate": 1e-06, + "loss": 0.545, + "mean_token_accuracy": 0.8217018246650696, + "num_tokens": 949887245.0, + "step": 35264 + }, + { + "epoch": 2.9079739424424838, + "grad_norm": 2.3496005535125732, + "learning_rate": 1e-06, + "loss": 0.805, + "mean_token_accuracy": 0.7460150718688965, + "num_tokens": 949912894.0, + "step": 35265 + }, + { + "epoch": 2.908056403067535, + "grad_norm": 1.9566181898117065, + "learning_rate": 1e-06, + "loss": 0.5704, + "mean_token_accuracy": 0.812293529510498, + "num_tokens": 949942902.0, + "step": 35266 + }, + { + "epoch": 2.908138863692587, + "grad_norm": 2.085597038269043, + "learning_rate": 1e-06, + "loss": 0.5322, + "mean_token_accuracy": 0.8178733587265015, + "num_tokens": 949968664.0, + "step": 35267 + }, + { + "epoch": 2.9082213243176382, + "grad_norm": 1.7878836393356323, + "learning_rate": 1e-06, + "loss": 0.64, + "mean_token_accuracy": 0.7937301397323608, + "num_tokens": 950005215.0, + "step": 35268 + }, + { + "epoch": 2.9083037849426896, + "grad_norm": 2.0260512828826904, + "learning_rate": 1e-06, + "loss": 0.7764, + "mean_token_accuracy": 0.7482048273086548, + "num_tokens": 950038183.0, + "step": 35269 + }, + { + "epoch": 2.9083862455677414, + "grad_norm": 2.292646646499634, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7636332511901855, + "num_tokens": 950063457.0, + "step": 35270 + }, + { + "epoch": 2.908468706192793, + "grad_norm": 2.0127580165863037, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7593234777450562, + "num_tokens": 950097089.0, + "step": 35271 + }, + { + "epoch": 2.9085511668178445, + "grad_norm": 1.87245774269104, + "learning_rate": 1e-06, + "loss": 0.632, + "mean_token_accuracy": 0.7899910807609558, + "num_tokens": 950129873.0, + "step": 35272 + }, + { + "epoch": 2.908633627442896, + "grad_norm": 2.0668020248413086, + "learning_rate": 1e-06, + "loss": 0.6182, + "mean_token_accuracy": 0.7937566041946411, + "num_tokens": 950158972.0, + "step": 35273 + }, + { + "epoch": 2.9087160880679477, + "grad_norm": 2.3379290103912354, + "learning_rate": 1e-06, + "loss": 0.6662, + "mean_token_accuracy": 0.7782754898071289, + "num_tokens": 950183028.0, + "step": 35274 + }, + { + "epoch": 2.908798548692999, + "grad_norm": 1.9874913692474365, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7237942218780518, + "num_tokens": 950214472.0, + "step": 35275 + }, + { + "epoch": 2.9088810093180504, + "grad_norm": 1.9744130373001099, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7641309499740601, + "num_tokens": 950246381.0, + "step": 35276 + }, + { + "epoch": 2.908963469943102, + "grad_norm": 2.1842234134674072, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7729842662811279, + "num_tokens": 950273127.0, + "step": 35277 + }, + { + "epoch": 2.909045930568154, + "grad_norm": 2.1160848140716553, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7423839569091797, + "num_tokens": 950303818.0, + "step": 35278 + }, + { + "epoch": 2.9091283911932053, + "grad_norm": 1.866107702255249, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7728837728500366, + "num_tokens": 950337992.0, + "step": 35279 + }, + { + "epoch": 2.9092108518182567, + "grad_norm": 1.9007058143615723, + "learning_rate": 1e-06, + "loss": 0.5187, + "mean_token_accuracy": 0.8281723260879517, + "num_tokens": 950372437.0, + "step": 35280 + }, + { + "epoch": 2.9092933124433085, + "grad_norm": 1.9920766353607178, + "learning_rate": 1e-06, + "loss": 0.6969, + "mean_token_accuracy": 0.7867650389671326, + "num_tokens": 950404648.0, + "step": 35281 + }, + { + "epoch": 2.90937577306836, + "grad_norm": 2.0594329833984375, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.7934610247612, + "num_tokens": 950434995.0, + "step": 35282 + }, + { + "epoch": 2.909458233693411, + "grad_norm": 2.4064650535583496, + "learning_rate": 1e-06, + "loss": 0.6898, + "mean_token_accuracy": 0.7674797177314758, + "num_tokens": 950459366.0, + "step": 35283 + }, + { + "epoch": 2.909540694318463, + "grad_norm": 2.1131787300109863, + "learning_rate": 1e-06, + "loss": 0.6399, + "mean_token_accuracy": 0.7898125052452087, + "num_tokens": 950488269.0, + "step": 35284 + }, + { + "epoch": 2.9096231549435143, + "grad_norm": 2.2312352657318115, + "learning_rate": 1e-06, + "loss": 0.7212, + "mean_token_accuracy": 0.7636247873306274, + "num_tokens": 950514537.0, + "step": 35285 + }, + { + "epoch": 2.909705615568566, + "grad_norm": 2.0573978424072266, + "learning_rate": 1e-06, + "loss": 0.6822, + "mean_token_accuracy": 0.778684675693512, + "num_tokens": 950544676.0, + "step": 35286 + }, + { + "epoch": 2.9097880761936175, + "grad_norm": 2.197089195251465, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7450048923492432, + "num_tokens": 950573418.0, + "step": 35287 + }, + { + "epoch": 2.9098705368186693, + "grad_norm": 2.1905369758605957, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7788181304931641, + "num_tokens": 950601117.0, + "step": 35288 + }, + { + "epoch": 2.9099529974437206, + "grad_norm": 1.9717583656311035, + "learning_rate": 1e-06, + "loss": 0.6226, + "mean_token_accuracy": 0.8014914989471436, + "num_tokens": 950634861.0, + "step": 35289 + }, + { + "epoch": 2.910035458068772, + "grad_norm": 1.9325299263000488, + "learning_rate": 1e-06, + "loss": 0.6265, + "mean_token_accuracy": 0.8023169040679932, + "num_tokens": 950668297.0, + "step": 35290 + }, + { + "epoch": 2.9101179186938237, + "grad_norm": 2.3554847240448, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7524447441101074, + "num_tokens": 950693583.0, + "step": 35291 + }, + { + "epoch": 2.910200379318875, + "grad_norm": 1.8825888633728027, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.758742094039917, + "num_tokens": 950729640.0, + "step": 35292 + }, + { + "epoch": 2.910282839943927, + "grad_norm": 2.194770336151123, + "learning_rate": 1e-06, + "loss": 0.6426, + "mean_token_accuracy": 0.7902873158454895, + "num_tokens": 950757173.0, + "step": 35293 + }, + { + "epoch": 2.9103653005689782, + "grad_norm": 1.991412878036499, + "learning_rate": 1e-06, + "loss": 0.6048, + "mean_token_accuracy": 0.8098350763320923, + "num_tokens": 950788773.0, + "step": 35294 + }, + { + "epoch": 2.91044776119403, + "grad_norm": 1.8261336088180542, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.7980654835700989, + "num_tokens": 950828592.0, + "step": 35295 + }, + { + "epoch": 2.9105302218190814, + "grad_norm": 2.051168441772461, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.7506957650184631, + "num_tokens": 950860492.0, + "step": 35296 + }, + { + "epoch": 2.9106126824441327, + "grad_norm": 1.8299506902694702, + "learning_rate": 1e-06, + "loss": 0.7036, + "mean_token_accuracy": 0.7821674942970276, + "num_tokens": 950896942.0, + "step": 35297 + }, + { + "epoch": 2.9106951430691845, + "grad_norm": 2.2610981464385986, + "learning_rate": 1e-06, + "loss": 0.6607, + "mean_token_accuracy": 0.7738432884216309, + "num_tokens": 950924171.0, + "step": 35298 + }, + { + "epoch": 2.910777603694236, + "grad_norm": 1.95071542263031, + "learning_rate": 1e-06, + "loss": 0.7165, + "mean_token_accuracy": 0.7695000767707825, + "num_tokens": 950958182.0, + "step": 35299 + }, + { + "epoch": 2.9108600643192877, + "grad_norm": 1.9863582849502563, + "learning_rate": 1e-06, + "loss": 0.5751, + "mean_token_accuracy": 0.8015007376670837, + "num_tokens": 950990385.0, + "step": 35300 + }, + { + "epoch": 2.910942524944339, + "grad_norm": 1.8932510614395142, + "learning_rate": 1e-06, + "loss": 0.6869, + "mean_token_accuracy": 0.7727130055427551, + "num_tokens": 951022455.0, + "step": 35301 + }, + { + "epoch": 2.911024985569391, + "grad_norm": 2.2453017234802246, + "learning_rate": 1e-06, + "loss": 0.5897, + "mean_token_accuracy": 0.7935616970062256, + "num_tokens": 951048239.0, + "step": 35302 + }, + { + "epoch": 2.911107446194442, + "grad_norm": 2.0253336429595947, + "learning_rate": 1e-06, + "loss": 0.5219, + "mean_token_accuracy": 0.814490020275116, + "num_tokens": 951077337.0, + "step": 35303 + }, + { + "epoch": 2.9111899068194935, + "grad_norm": 1.968693494796753, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7538784742355347, + "num_tokens": 951113629.0, + "step": 35304 + }, + { + "epoch": 2.9112723674445453, + "grad_norm": 2.0714850425720215, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7481875419616699, + "num_tokens": 951146033.0, + "step": 35305 + }, + { + "epoch": 2.9113548280695967, + "grad_norm": 2.14347243309021, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7673671245574951, + "num_tokens": 951172913.0, + "step": 35306 + }, + { + "epoch": 2.9114372886946485, + "grad_norm": 2.033905267715454, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7703002691268921, + "num_tokens": 951204805.0, + "step": 35307 + }, + { + "epoch": 2.9115197493197, + "grad_norm": 1.900709629058838, + "learning_rate": 1e-06, + "loss": 0.5984, + "mean_token_accuracy": 0.794837236404419, + "num_tokens": 951240252.0, + "step": 35308 + }, + { + "epoch": 2.9116022099447516, + "grad_norm": 1.9996333122253418, + "learning_rate": 1e-06, + "loss": 0.6227, + "mean_token_accuracy": 0.7915970087051392, + "num_tokens": 951271859.0, + "step": 35309 + }, + { + "epoch": 2.911684670569803, + "grad_norm": 2.0711963176727295, + "learning_rate": 1e-06, + "loss": 0.6556, + "mean_token_accuracy": 0.791974663734436, + "num_tokens": 951303468.0, + "step": 35310 + }, + { + "epoch": 2.9117671311948543, + "grad_norm": 2.0506341457366943, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.746025562286377, + "num_tokens": 951335060.0, + "step": 35311 + }, + { + "epoch": 2.911849591819906, + "grad_norm": 2.045896291732788, + "learning_rate": 1e-06, + "loss": 0.5265, + "mean_token_accuracy": 0.8167427778244019, + "num_tokens": 951362797.0, + "step": 35312 + }, + { + "epoch": 2.9119320524449575, + "grad_norm": 2.3046085834503174, + "learning_rate": 1e-06, + "loss": 0.6594, + "mean_token_accuracy": 0.7772842645645142, + "num_tokens": 951388848.0, + "step": 35313 + }, + { + "epoch": 2.912014513070009, + "grad_norm": 1.8206868171691895, + "learning_rate": 1e-06, + "loss": 0.7813, + "mean_token_accuracy": 0.7499511241912842, + "num_tokens": 951424585.0, + "step": 35314 + }, + { + "epoch": 2.9120969736950606, + "grad_norm": 1.8743269443511963, + "learning_rate": 1e-06, + "loss": 0.7377, + "mean_token_accuracy": 0.7646387815475464, + "num_tokens": 951459381.0, + "step": 35315 + }, + { + "epoch": 2.9121794343201124, + "grad_norm": 2.011943817138672, + "learning_rate": 1e-06, + "loss": 0.6758, + "mean_token_accuracy": 0.7812650203704834, + "num_tokens": 951490935.0, + "step": 35316 + }, + { + "epoch": 2.9122618949451637, + "grad_norm": 2.052032232284546, + "learning_rate": 1e-06, + "loss": 0.7332, + "mean_token_accuracy": 0.7727792263031006, + "num_tokens": 951524291.0, + "step": 35317 + }, + { + "epoch": 2.912344355570215, + "grad_norm": 2.0761475563049316, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7360015511512756, + "num_tokens": 951555450.0, + "step": 35318 + }, + { + "epoch": 2.912426816195267, + "grad_norm": 2.180340051651001, + "learning_rate": 1e-06, + "loss": 0.6792, + "mean_token_accuracy": 0.7740076780319214, + "num_tokens": 951582314.0, + "step": 35319 + }, + { + "epoch": 2.9125092768203182, + "grad_norm": 2.100025177001953, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7623810768127441, + "num_tokens": 951614951.0, + "step": 35320 + }, + { + "epoch": 2.9125917374453696, + "grad_norm": 1.9216315746307373, + "learning_rate": 1e-06, + "loss": 0.6106, + "mean_token_accuracy": 0.7853118181228638, + "num_tokens": 951648807.0, + "step": 35321 + }, + { + "epoch": 2.9126741980704214, + "grad_norm": 1.795172929763794, + "learning_rate": 1e-06, + "loss": 0.535, + "mean_token_accuracy": 0.8173813223838806, + "num_tokens": 951687256.0, + "step": 35322 + }, + { + "epoch": 2.912756658695473, + "grad_norm": 1.9541646242141724, + "learning_rate": 1e-06, + "loss": 0.542, + "mean_token_accuracy": 0.8214187622070312, + "num_tokens": 951716434.0, + "step": 35323 + }, + { + "epoch": 2.9128391193205245, + "grad_norm": 2.0840444564819336, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7401319146156311, + "num_tokens": 951747265.0, + "step": 35324 + }, + { + "epoch": 2.912921579945576, + "grad_norm": 2.244324207305908, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.774899423122406, + "num_tokens": 951774685.0, + "step": 35325 + }, + { + "epoch": 2.9130040405706277, + "grad_norm": 2.1033830642700195, + "learning_rate": 1e-06, + "loss": 0.6834, + "mean_token_accuracy": 0.7710195779800415, + "num_tokens": 951803992.0, + "step": 35326 + }, + { + "epoch": 2.913086501195679, + "grad_norm": 2.0396485328674316, + "learning_rate": 1e-06, + "loss": 0.5425, + "mean_token_accuracy": 0.8198649287223816, + "num_tokens": 951831287.0, + "step": 35327 + }, + { + "epoch": 2.9131689618207304, + "grad_norm": 2.015079975128174, + "learning_rate": 1e-06, + "loss": 0.6688, + "mean_token_accuracy": 0.7866333723068237, + "num_tokens": 951862588.0, + "step": 35328 + }, + { + "epoch": 2.913251422445782, + "grad_norm": 1.9158202409744263, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7471518516540527, + "num_tokens": 951901546.0, + "step": 35329 + }, + { + "epoch": 2.9133338830708335, + "grad_norm": 1.8426518440246582, + "learning_rate": 1e-06, + "loss": 0.5783, + "mean_token_accuracy": 0.8055844902992249, + "num_tokens": 951940571.0, + "step": 35330 + }, + { + "epoch": 2.9134163436958853, + "grad_norm": 2.432370662689209, + "learning_rate": 1e-06, + "loss": 0.6582, + "mean_token_accuracy": 0.7902206182479858, + "num_tokens": 951962767.0, + "step": 35331 + }, + { + "epoch": 2.9134988043209367, + "grad_norm": 1.9137767553329468, + "learning_rate": 1e-06, + "loss": 0.7227, + "mean_token_accuracy": 0.7696476578712463, + "num_tokens": 951995488.0, + "step": 35332 + }, + { + "epoch": 2.9135812649459885, + "grad_norm": 1.873283863067627, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7902735471725464, + "num_tokens": 952026207.0, + "step": 35333 + }, + { + "epoch": 2.91366372557104, + "grad_norm": 2.13059401512146, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7408323287963867, + "num_tokens": 952057388.0, + "step": 35334 + }, + { + "epoch": 2.913746186196091, + "grad_norm": 2.118626832962036, + "learning_rate": 1e-06, + "loss": 0.7528, + "mean_token_accuracy": 0.751118540763855, + "num_tokens": 952086586.0, + "step": 35335 + }, + { + "epoch": 2.913828646821143, + "grad_norm": 2.0745205879211426, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.730393648147583, + "num_tokens": 952115809.0, + "step": 35336 + }, + { + "epoch": 2.9139111074461943, + "grad_norm": 2.2535741329193115, + "learning_rate": 1e-06, + "loss": 0.7314, + "mean_token_accuracy": 0.7574222683906555, + "num_tokens": 952141846.0, + "step": 35337 + }, + { + "epoch": 2.913993568071246, + "grad_norm": 2.226396083831787, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7513399720191956, + "num_tokens": 952170047.0, + "step": 35338 + }, + { + "epoch": 2.9140760286962974, + "grad_norm": 2.1839637756347656, + "learning_rate": 1e-06, + "loss": 0.6583, + "mean_token_accuracy": 0.770800769329071, + "num_tokens": 952198333.0, + "step": 35339 + }, + { + "epoch": 2.9141584893213492, + "grad_norm": 1.9712737798690796, + "learning_rate": 1e-06, + "loss": 0.6876, + "mean_token_accuracy": 0.7792730927467346, + "num_tokens": 952231752.0, + "step": 35340 + }, + { + "epoch": 2.9142409499464006, + "grad_norm": 2.109114408493042, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7576780915260315, + "num_tokens": 952261391.0, + "step": 35341 + }, + { + "epoch": 2.914323410571452, + "grad_norm": 1.962382435798645, + "learning_rate": 1e-06, + "loss": 0.5487, + "mean_token_accuracy": 0.8104247450828552, + "num_tokens": 952291030.0, + "step": 35342 + }, + { + "epoch": 2.9144058711965037, + "grad_norm": 2.1789004802703857, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7605651617050171, + "num_tokens": 952319443.0, + "step": 35343 + }, + { + "epoch": 2.914488331821555, + "grad_norm": 1.9295910596847534, + "learning_rate": 1e-06, + "loss": 0.5222, + "mean_token_accuracy": 0.8125098943710327, + "num_tokens": 952350615.0, + "step": 35344 + }, + { + "epoch": 2.914570792446607, + "grad_norm": 2.326970338821411, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7488267421722412, + "num_tokens": 952377934.0, + "step": 35345 + }, + { + "epoch": 2.9146532530716582, + "grad_norm": 2.153615713119507, + "learning_rate": 1e-06, + "loss": 0.6906, + "mean_token_accuracy": 0.7708717584609985, + "num_tokens": 952407610.0, + "step": 35346 + }, + { + "epoch": 2.91473571369671, + "grad_norm": 1.9245681762695312, + "learning_rate": 1e-06, + "loss": 0.5174, + "mean_token_accuracy": 0.8246197700500488, + "num_tokens": 952439330.0, + "step": 35347 + }, + { + "epoch": 2.9148181743217614, + "grad_norm": 1.9663746356964111, + "learning_rate": 1e-06, + "loss": 0.7507, + "mean_token_accuracy": 0.7623927593231201, + "num_tokens": 952470635.0, + "step": 35348 + }, + { + "epoch": 2.9149006349468127, + "grad_norm": 2.1921157836914062, + "learning_rate": 1e-06, + "loss": 0.6832, + "mean_token_accuracy": 0.7684043645858765, + "num_tokens": 952497002.0, + "step": 35349 + }, + { + "epoch": 2.9149830955718645, + "grad_norm": 2.06022310256958, + "learning_rate": 1e-06, + "loss": 0.6816, + "mean_token_accuracy": 0.782322108745575, + "num_tokens": 952528479.0, + "step": 35350 + }, + { + "epoch": 2.915065556196916, + "grad_norm": 2.400655508041382, + "learning_rate": 1e-06, + "loss": 0.8015, + "mean_token_accuracy": 0.7429128885269165, + "num_tokens": 952554348.0, + "step": 35351 + }, + { + "epoch": 2.9151480168219677, + "grad_norm": 1.8222874402999878, + "learning_rate": 1e-06, + "loss": 0.6102, + "mean_token_accuracy": 0.793622612953186, + "num_tokens": 952590055.0, + "step": 35352 + }, + { + "epoch": 2.915230477447019, + "grad_norm": 2.0702264308929443, + "learning_rate": 1e-06, + "loss": 0.7295, + "mean_token_accuracy": 0.7584935426712036, + "num_tokens": 952620659.0, + "step": 35353 + }, + { + "epoch": 2.915312938072071, + "grad_norm": 2.026573657989502, + "learning_rate": 1e-06, + "loss": 0.6025, + "mean_token_accuracy": 0.7900114059448242, + "num_tokens": 952649856.0, + "step": 35354 + }, + { + "epoch": 2.915395398697122, + "grad_norm": 1.8169081211090088, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7529723644256592, + "num_tokens": 952688117.0, + "step": 35355 + }, + { + "epoch": 2.9154778593221735, + "grad_norm": 2.0047521591186523, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.7749572992324829, + "num_tokens": 952720816.0, + "step": 35356 + }, + { + "epoch": 2.9155603199472253, + "grad_norm": 2.566723585128784, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7386229634284973, + "num_tokens": 952748049.0, + "step": 35357 + }, + { + "epoch": 2.9156427805722767, + "grad_norm": 1.8997557163238525, + "learning_rate": 1e-06, + "loss": 0.567, + "mean_token_accuracy": 0.8132002949714661, + "num_tokens": 952781999.0, + "step": 35358 + }, + { + "epoch": 2.915725241197328, + "grad_norm": 2.1051652431488037, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7531645894050598, + "num_tokens": 952817240.0, + "step": 35359 + }, + { + "epoch": 2.91580770182238, + "grad_norm": 2.0522639751434326, + "learning_rate": 1e-06, + "loss": 0.6245, + "mean_token_accuracy": 0.8011271953582764, + "num_tokens": 952845387.0, + "step": 35360 + }, + { + "epoch": 2.9158901624474316, + "grad_norm": 2.3883304595947266, + "learning_rate": 1e-06, + "loss": 0.7873, + "mean_token_accuracy": 0.7462140321731567, + "num_tokens": 952868720.0, + "step": 35361 + }, + { + "epoch": 2.915972623072483, + "grad_norm": 2.3503410816192627, + "learning_rate": 1e-06, + "loss": 0.5679, + "mean_token_accuracy": 0.8112555742263794, + "num_tokens": 952892242.0, + "step": 35362 + }, + { + "epoch": 2.9160550836975343, + "grad_norm": 1.9568750858306885, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.75101637840271, + "num_tokens": 952929813.0, + "step": 35363 + }, + { + "epoch": 2.916137544322586, + "grad_norm": 2.000032901763916, + "learning_rate": 1e-06, + "loss": 0.6569, + "mean_token_accuracy": 0.7926303148269653, + "num_tokens": 952960615.0, + "step": 35364 + }, + { + "epoch": 2.9162200049476374, + "grad_norm": 2.167958974838257, + "learning_rate": 1e-06, + "loss": 0.6654, + "mean_token_accuracy": 0.7893349528312683, + "num_tokens": 952988620.0, + "step": 35365 + }, + { + "epoch": 2.916302465572689, + "grad_norm": 2.3177285194396973, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7392225861549377, + "num_tokens": 953017817.0, + "step": 35366 + }, + { + "epoch": 2.9163849261977406, + "grad_norm": 1.8582851886749268, + "learning_rate": 1e-06, + "loss": 0.6089, + "mean_token_accuracy": 0.7902746200561523, + "num_tokens": 953056364.0, + "step": 35367 + }, + { + "epoch": 2.9164673868227924, + "grad_norm": 2.414670944213867, + "learning_rate": 1e-06, + "loss": 0.6595, + "mean_token_accuracy": 0.78864586353302, + "num_tokens": 953078812.0, + "step": 35368 + }, + { + "epoch": 2.9165498474478437, + "grad_norm": 2.25677752494812, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7736237049102783, + "num_tokens": 953108458.0, + "step": 35369 + }, + { + "epoch": 2.916632308072895, + "grad_norm": 2.295170545578003, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7387632131576538, + "num_tokens": 953135447.0, + "step": 35370 + }, + { + "epoch": 2.916714768697947, + "grad_norm": 2.190612554550171, + "learning_rate": 1e-06, + "loss": 0.5441, + "mean_token_accuracy": 0.824191689491272, + "num_tokens": 953163888.0, + "step": 35371 + }, + { + "epoch": 2.9167972293229982, + "grad_norm": 1.9078006744384766, + "learning_rate": 1e-06, + "loss": 0.5224, + "mean_token_accuracy": 0.817054033279419, + "num_tokens": 953196876.0, + "step": 35372 + }, + { + "epoch": 2.9168796899480496, + "grad_norm": 2.1963703632354736, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7464200258255005, + "num_tokens": 953223277.0, + "step": 35373 + }, + { + "epoch": 2.9169621505731014, + "grad_norm": 2.4199764728546143, + "learning_rate": 1e-06, + "loss": 0.7444, + "mean_token_accuracy": 0.7573273777961731, + "num_tokens": 953250581.0, + "step": 35374 + }, + { + "epoch": 2.9170446111981527, + "grad_norm": 2.0154170989990234, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.7887742519378662, + "num_tokens": 953284502.0, + "step": 35375 + }, + { + "epoch": 2.9171270718232045, + "grad_norm": 2.1070544719696045, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7689255475997925, + "num_tokens": 953311541.0, + "step": 35376 + }, + { + "epoch": 2.917209532448256, + "grad_norm": 2.0663681030273438, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.8019229769706726, + "num_tokens": 953338398.0, + "step": 35377 + }, + { + "epoch": 2.9172919930733077, + "grad_norm": 1.984210729598999, + "learning_rate": 1e-06, + "loss": 0.7294, + "mean_token_accuracy": 0.7622106075286865, + "num_tokens": 953370792.0, + "step": 35378 + }, + { + "epoch": 2.917374453698359, + "grad_norm": 2.1341278553009033, + "learning_rate": 1e-06, + "loss": 0.5398, + "mean_token_accuracy": 0.8150352239608765, + "num_tokens": 953397951.0, + "step": 35379 + }, + { + "epoch": 2.9174569143234104, + "grad_norm": 1.9349944591522217, + "learning_rate": 1e-06, + "loss": 0.6553, + "mean_token_accuracy": 0.7860872745513916, + "num_tokens": 953429491.0, + "step": 35380 + }, + { + "epoch": 2.917539374948462, + "grad_norm": 1.7941354513168335, + "learning_rate": 1e-06, + "loss": 0.6037, + "mean_token_accuracy": 0.801521897315979, + "num_tokens": 953467138.0, + "step": 35381 + }, + { + "epoch": 2.9176218355735135, + "grad_norm": 2.4477078914642334, + "learning_rate": 1e-06, + "loss": 0.4787, + "mean_token_accuracy": 0.8266998529434204, + "num_tokens": 953487215.0, + "step": 35382 + }, + { + "epoch": 2.9177042961985653, + "grad_norm": 2.4325106143951416, + "learning_rate": 1e-06, + "loss": 0.6244, + "mean_token_accuracy": 0.7862301468849182, + "num_tokens": 953511500.0, + "step": 35383 + }, + { + "epoch": 2.9177867568236167, + "grad_norm": 2.3572065830230713, + "learning_rate": 1e-06, + "loss": 0.6634, + "mean_token_accuracy": 0.7802041172981262, + "num_tokens": 953536243.0, + "step": 35384 + }, + { + "epoch": 2.9178692174486685, + "grad_norm": 2.3256633281707764, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7233556509017944, + "num_tokens": 953565721.0, + "step": 35385 + }, + { + "epoch": 2.91795167807372, + "grad_norm": 1.8307974338531494, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.7540967464447021, + "num_tokens": 953603328.0, + "step": 35386 + }, + { + "epoch": 2.918034138698771, + "grad_norm": 2.2980246543884277, + "learning_rate": 1e-06, + "loss": 0.7548, + "mean_token_accuracy": 0.7644057869911194, + "num_tokens": 953629404.0, + "step": 35387 + }, + { + "epoch": 2.918116599323823, + "grad_norm": 2.1555185317993164, + "learning_rate": 1e-06, + "loss": 0.5994, + "mean_token_accuracy": 0.8045593500137329, + "num_tokens": 953657576.0, + "step": 35388 + }, + { + "epoch": 2.9181990599488743, + "grad_norm": 1.966982126235962, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7822794914245605, + "num_tokens": 953692278.0, + "step": 35389 + }, + { + "epoch": 2.918281520573926, + "grad_norm": 2.1318254470825195, + "learning_rate": 1e-06, + "loss": 0.7713, + "mean_token_accuracy": 0.7498617172241211, + "num_tokens": 953725114.0, + "step": 35390 + }, + { + "epoch": 2.9183639811989774, + "grad_norm": 2.3746166229248047, + "learning_rate": 1e-06, + "loss": 0.6736, + "mean_token_accuracy": 0.7746289968490601, + "num_tokens": 953749021.0, + "step": 35391 + }, + { + "epoch": 2.9184464418240292, + "grad_norm": 2.103421449661255, + "learning_rate": 1e-06, + "loss": 0.7523, + "mean_token_accuracy": 0.754536509513855, + "num_tokens": 953782771.0, + "step": 35392 + }, + { + "epoch": 2.9185289024490806, + "grad_norm": 1.942989468574524, + "learning_rate": 1e-06, + "loss": 0.5924, + "mean_token_accuracy": 0.8063369989395142, + "num_tokens": 953815497.0, + "step": 35393 + }, + { + "epoch": 2.918611363074132, + "grad_norm": 1.95163893699646, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7840492725372314, + "num_tokens": 953846627.0, + "step": 35394 + }, + { + "epoch": 2.9186938236991837, + "grad_norm": 2.058255910873413, + "learning_rate": 1e-06, + "loss": 0.6886, + "mean_token_accuracy": 0.7798373699188232, + "num_tokens": 953875846.0, + "step": 35395 + }, + { + "epoch": 2.918776284324235, + "grad_norm": 2.0096631050109863, + "learning_rate": 1e-06, + "loss": 0.6235, + "mean_token_accuracy": 0.7908462285995483, + "num_tokens": 953906440.0, + "step": 35396 + }, + { + "epoch": 2.918858744949287, + "grad_norm": 2.061755418777466, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7508017420768738, + "num_tokens": 953938591.0, + "step": 35397 + }, + { + "epoch": 2.9189412055743382, + "grad_norm": 1.9207037687301636, + "learning_rate": 1e-06, + "loss": 0.5686, + "mean_token_accuracy": 0.7957817316055298, + "num_tokens": 953969680.0, + "step": 35398 + }, + { + "epoch": 2.91902366619939, + "grad_norm": 2.28192400932312, + "learning_rate": 1e-06, + "loss": 0.7329, + "mean_token_accuracy": 0.7763633131980896, + "num_tokens": 953997787.0, + "step": 35399 + }, + { + "epoch": 2.9191061268244414, + "grad_norm": 1.8650261163711548, + "learning_rate": 1e-06, + "loss": 0.746, + "mean_token_accuracy": 0.7610121965408325, + "num_tokens": 954032380.0, + "step": 35400 + }, + { + "epoch": 2.9191885874494927, + "grad_norm": 2.1616055965423584, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7514370679855347, + "num_tokens": 954061975.0, + "step": 35401 + }, + { + "epoch": 2.9192710480745445, + "grad_norm": 1.8888661861419678, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7486382722854614, + "num_tokens": 954098349.0, + "step": 35402 + }, + { + "epoch": 2.919353508699596, + "grad_norm": 2.3041465282440186, + "learning_rate": 1e-06, + "loss": 0.7591, + "mean_token_accuracy": 0.7565828561782837, + "num_tokens": 954123795.0, + "step": 35403 + }, + { + "epoch": 2.919435969324647, + "grad_norm": 2.287362575531006, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.7445189952850342, + "num_tokens": 954151409.0, + "step": 35404 + }, + { + "epoch": 2.919518429949699, + "grad_norm": 2.310457706451416, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7774485349655151, + "num_tokens": 954176642.0, + "step": 35405 + }, + { + "epoch": 2.919600890574751, + "grad_norm": 1.9625433683395386, + "learning_rate": 1e-06, + "loss": 0.6475, + "mean_token_accuracy": 0.795883059501648, + "num_tokens": 954210666.0, + "step": 35406 + }, + { + "epoch": 2.919683351199802, + "grad_norm": 2.3450512886047363, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7780231237411499, + "num_tokens": 954237319.0, + "step": 35407 + }, + { + "epoch": 2.9197658118248535, + "grad_norm": 1.9622044563293457, + "learning_rate": 1e-06, + "loss": 0.5957, + "mean_token_accuracy": 0.7971447110176086, + "num_tokens": 954272499.0, + "step": 35408 + }, + { + "epoch": 2.9198482724499053, + "grad_norm": 2.0381710529327393, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7320919036865234, + "num_tokens": 954305184.0, + "step": 35409 + }, + { + "epoch": 2.9199307330749567, + "grad_norm": 2.4994089603424072, + "learning_rate": 1e-06, + "loss": 0.6397, + "mean_token_accuracy": 0.7798656225204468, + "num_tokens": 954329680.0, + "step": 35410 + }, + { + "epoch": 2.920013193700008, + "grad_norm": 1.9613808393478394, + "learning_rate": 1e-06, + "loss": 0.7228, + "mean_token_accuracy": 0.7646297216415405, + "num_tokens": 954365292.0, + "step": 35411 + }, + { + "epoch": 2.92009565432506, + "grad_norm": 2.3835947513580322, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.7860085964202881, + "num_tokens": 954392225.0, + "step": 35412 + }, + { + "epoch": 2.9201781149501116, + "grad_norm": 2.0769829750061035, + "learning_rate": 1e-06, + "loss": 0.6011, + "mean_token_accuracy": 0.8023368120193481, + "num_tokens": 954419599.0, + "step": 35413 + }, + { + "epoch": 2.920260575575163, + "grad_norm": 2.3619589805603027, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7488830089569092, + "num_tokens": 954444619.0, + "step": 35414 + }, + { + "epoch": 2.9203430362002143, + "grad_norm": 2.0252974033355713, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.7652125358581543, + "num_tokens": 954476982.0, + "step": 35415 + }, + { + "epoch": 2.920425496825266, + "grad_norm": 1.998400092124939, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7657238841056824, + "num_tokens": 954510078.0, + "step": 35416 + }, + { + "epoch": 2.9205079574503174, + "grad_norm": 2.213472604751587, + "learning_rate": 1e-06, + "loss": 0.549, + "mean_token_accuracy": 0.8162485361099243, + "num_tokens": 954534033.0, + "step": 35417 + }, + { + "epoch": 2.920590418075369, + "grad_norm": 1.9848929643630981, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7568966150283813, + "num_tokens": 954565269.0, + "step": 35418 + }, + { + "epoch": 2.9206728787004206, + "grad_norm": 2.098390817642212, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7589218020439148, + "num_tokens": 954594491.0, + "step": 35419 + }, + { + "epoch": 2.920755339325472, + "grad_norm": 2.070711851119995, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.777324914932251, + "num_tokens": 954626539.0, + "step": 35420 + }, + { + "epoch": 2.9208377999505237, + "grad_norm": 2.4447896480560303, + "learning_rate": 1e-06, + "loss": 0.7197, + "mean_token_accuracy": 0.7710025310516357, + "num_tokens": 954649490.0, + "step": 35421 + }, + { + "epoch": 2.920920260575575, + "grad_norm": 1.8680006265640259, + "learning_rate": 1e-06, + "loss": 0.6158, + "mean_token_accuracy": 0.8061875104904175, + "num_tokens": 954683433.0, + "step": 35422 + }, + { + "epoch": 2.921002721200627, + "grad_norm": 1.8536721467971802, + "learning_rate": 1e-06, + "loss": 0.5294, + "mean_token_accuracy": 0.8201041221618652, + "num_tokens": 954713407.0, + "step": 35423 + }, + { + "epoch": 2.9210851818256782, + "grad_norm": 1.9313102960586548, + "learning_rate": 1e-06, + "loss": 0.5164, + "mean_token_accuracy": 0.8221731185913086, + "num_tokens": 954749496.0, + "step": 35424 + }, + { + "epoch": 2.9211676424507296, + "grad_norm": 2.214987277984619, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7548601627349854, + "num_tokens": 954776406.0, + "step": 35425 + }, + { + "epoch": 2.9212501030757814, + "grad_norm": 1.9043314456939697, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7650665044784546, + "num_tokens": 954809971.0, + "step": 35426 + }, + { + "epoch": 2.9213325637008327, + "grad_norm": 2.1319894790649414, + "learning_rate": 1e-06, + "loss": 0.708, + "mean_token_accuracy": 0.762633204460144, + "num_tokens": 954839986.0, + "step": 35427 + }, + { + "epoch": 2.9214150243258845, + "grad_norm": 2.30823016166687, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7526850700378418, + "num_tokens": 954865052.0, + "step": 35428 + }, + { + "epoch": 2.921497484950936, + "grad_norm": 2.060758352279663, + "learning_rate": 1e-06, + "loss": 0.6655, + "mean_token_accuracy": 0.7747553586959839, + "num_tokens": 954896853.0, + "step": 35429 + }, + { + "epoch": 2.9215799455759877, + "grad_norm": 1.9617669582366943, + "learning_rate": 1e-06, + "loss": 0.6276, + "mean_token_accuracy": 0.7840267419815063, + "num_tokens": 954928086.0, + "step": 35430 + }, + { + "epoch": 2.921662406201039, + "grad_norm": 2.2184183597564697, + "learning_rate": 1e-06, + "loss": 0.685, + "mean_token_accuracy": 0.7731301784515381, + "num_tokens": 954960737.0, + "step": 35431 + }, + { + "epoch": 2.9217448668260904, + "grad_norm": 2.282228469848633, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7620593309402466, + "num_tokens": 954986550.0, + "step": 35432 + }, + { + "epoch": 2.921827327451142, + "grad_norm": 2.1046109199523926, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7736933827400208, + "num_tokens": 955016174.0, + "step": 35433 + }, + { + "epoch": 2.9219097880761935, + "grad_norm": 2.1154754161834717, + "learning_rate": 1e-06, + "loss": 0.6707, + "mean_token_accuracy": 0.7789680361747742, + "num_tokens": 955046977.0, + "step": 35434 + }, + { + "epoch": 2.9219922487012453, + "grad_norm": 2.196019411087036, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7443720102310181, + "num_tokens": 955076636.0, + "step": 35435 + }, + { + "epoch": 2.9220747093262966, + "grad_norm": 1.8985124826431274, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.7912657260894775, + "num_tokens": 955110422.0, + "step": 35436 + }, + { + "epoch": 2.9221571699513484, + "grad_norm": 2.2869067192077637, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7612429261207581, + "num_tokens": 955138117.0, + "step": 35437 + }, + { + "epoch": 2.9222396305764, + "grad_norm": 1.9611995220184326, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7489858269691467, + "num_tokens": 955172562.0, + "step": 35438 + }, + { + "epoch": 2.922322091201451, + "grad_norm": 2.3966448307037354, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7559444904327393, + "num_tokens": 955198852.0, + "step": 35439 + }, + { + "epoch": 2.922404551826503, + "grad_norm": 1.9354597330093384, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.7978551387786865, + "num_tokens": 955233632.0, + "step": 35440 + }, + { + "epoch": 2.9224870124515543, + "grad_norm": 2.1417033672332764, + "learning_rate": 1e-06, + "loss": 0.7476, + "mean_token_accuracy": 0.7542616128921509, + "num_tokens": 955263777.0, + "step": 35441 + }, + { + "epoch": 2.922569473076606, + "grad_norm": 2.166908025741577, + "learning_rate": 1e-06, + "loss": 0.6699, + "mean_token_accuracy": 0.777806282043457, + "num_tokens": 955289663.0, + "step": 35442 + }, + { + "epoch": 2.9226519337016574, + "grad_norm": 2.2023465633392334, + "learning_rate": 1e-06, + "loss": 0.747, + "mean_token_accuracy": 0.7566648721694946, + "num_tokens": 955317089.0, + "step": 35443 + }, + { + "epoch": 2.9227343943267092, + "grad_norm": 2.101539373397827, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.760998547077179, + "num_tokens": 955347506.0, + "step": 35444 + }, + { + "epoch": 2.9228168549517606, + "grad_norm": 2.187905788421631, + "learning_rate": 1e-06, + "loss": 0.7487, + "mean_token_accuracy": 0.7712790369987488, + "num_tokens": 955374766.0, + "step": 35445 + }, + { + "epoch": 2.922899315576812, + "grad_norm": 2.2077224254608154, + "learning_rate": 1e-06, + "loss": 0.6763, + "mean_token_accuracy": 0.774293065071106, + "num_tokens": 955403753.0, + "step": 35446 + }, + { + "epoch": 2.9229817762018637, + "grad_norm": 2.1793646812438965, + "learning_rate": 1e-06, + "loss": 0.6982, + "mean_token_accuracy": 0.7720247507095337, + "num_tokens": 955432105.0, + "step": 35447 + }, + { + "epoch": 2.923064236826915, + "grad_norm": 1.9858366250991821, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7470561265945435, + "num_tokens": 955465036.0, + "step": 35448 + }, + { + "epoch": 2.9231466974519664, + "grad_norm": 1.9967085123062134, + "learning_rate": 1e-06, + "loss": 0.6562, + "mean_token_accuracy": 0.7866631746292114, + "num_tokens": 955501780.0, + "step": 35449 + }, + { + "epoch": 2.923229158077018, + "grad_norm": 1.9112143516540527, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7670238018035889, + "num_tokens": 955536632.0, + "step": 35450 + }, + { + "epoch": 2.92331161870207, + "grad_norm": 2.1376354694366455, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.748895525932312, + "num_tokens": 955566546.0, + "step": 35451 + }, + { + "epoch": 2.9233940793271214, + "grad_norm": 1.9515972137451172, + "learning_rate": 1e-06, + "loss": 0.5717, + "mean_token_accuracy": 0.8069824576377869, + "num_tokens": 955597442.0, + "step": 35452 + }, + { + "epoch": 2.9234765399521727, + "grad_norm": 1.8112146854400635, + "learning_rate": 1e-06, + "loss": 0.5575, + "mean_token_accuracy": 0.8123159408569336, + "num_tokens": 955630369.0, + "step": 35453 + }, + { + "epoch": 2.9235590005772245, + "grad_norm": 2.386458396911621, + "learning_rate": 1e-06, + "loss": 0.5717, + "mean_token_accuracy": 0.800402820110321, + "num_tokens": 955654704.0, + "step": 35454 + }, + { + "epoch": 2.923641461202276, + "grad_norm": 2.159414768218994, + "learning_rate": 1e-06, + "loss": 0.6748, + "mean_token_accuracy": 0.7698958516120911, + "num_tokens": 955684918.0, + "step": 35455 + }, + { + "epoch": 2.923723921827327, + "grad_norm": 2.088752031326294, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7334483861923218, + "num_tokens": 955718564.0, + "step": 35456 + }, + { + "epoch": 2.923806382452379, + "grad_norm": 2.052701711654663, + "learning_rate": 1e-06, + "loss": 0.5607, + "mean_token_accuracy": 0.8117340803146362, + "num_tokens": 955747136.0, + "step": 35457 + }, + { + "epoch": 2.923888843077431, + "grad_norm": 2.0895769596099854, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7746568322181702, + "num_tokens": 955780087.0, + "step": 35458 + }, + { + "epoch": 2.923971303702482, + "grad_norm": 2.4038808345794678, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.7025482654571533, + "num_tokens": 955808647.0, + "step": 35459 + }, + { + "epoch": 2.9240537643275335, + "grad_norm": 2.3026721477508545, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7576674222946167, + "num_tokens": 955834008.0, + "step": 35460 + }, + { + "epoch": 2.9241362249525853, + "grad_norm": 2.2732596397399902, + "learning_rate": 1e-06, + "loss": 0.7682, + "mean_token_accuracy": 0.7625264525413513, + "num_tokens": 955862344.0, + "step": 35461 + }, + { + "epoch": 2.9242186855776366, + "grad_norm": 1.932332158088684, + "learning_rate": 1e-06, + "loss": 0.7711, + "mean_token_accuracy": 0.7556906342506409, + "num_tokens": 955896047.0, + "step": 35462 + }, + { + "epoch": 2.924301146202688, + "grad_norm": 2.1028401851654053, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.72333163022995, + "num_tokens": 955924298.0, + "step": 35463 + }, + { + "epoch": 2.92438360682774, + "grad_norm": 1.9445545673370361, + "learning_rate": 1e-06, + "loss": 0.6831, + "mean_token_accuracy": 0.7766551971435547, + "num_tokens": 955956983.0, + "step": 35464 + }, + { + "epoch": 2.924466067452791, + "grad_norm": 2.0360782146453857, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.7946542501449585, + "num_tokens": 955987796.0, + "step": 35465 + }, + { + "epoch": 2.924548528077843, + "grad_norm": 2.272042751312256, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7651153802871704, + "num_tokens": 956016325.0, + "step": 35466 + }, + { + "epoch": 2.9246309887028943, + "grad_norm": 2.089996814727783, + "learning_rate": 1e-06, + "loss": 0.7638, + "mean_token_accuracy": 0.7498785257339478, + "num_tokens": 956046015.0, + "step": 35467 + }, + { + "epoch": 2.924713449327946, + "grad_norm": 2.0650064945220947, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7693184614181519, + "num_tokens": 956075549.0, + "step": 35468 + }, + { + "epoch": 2.9247959099529974, + "grad_norm": 1.9307020902633667, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7797768115997314, + "num_tokens": 956106089.0, + "step": 35469 + }, + { + "epoch": 2.924878370578049, + "grad_norm": 2.248239040374756, + "learning_rate": 1e-06, + "loss": 0.6774, + "mean_token_accuracy": 0.7803101539611816, + "num_tokens": 956132866.0, + "step": 35470 + }, + { + "epoch": 2.9249608312031006, + "grad_norm": 2.093064069747925, + "learning_rate": 1e-06, + "loss": 0.6204, + "mean_token_accuracy": 0.7899239659309387, + "num_tokens": 956163018.0, + "step": 35471 + }, + { + "epoch": 2.925043291828152, + "grad_norm": 2.143177032470703, + "learning_rate": 1e-06, + "loss": 0.5877, + "mean_token_accuracy": 0.8028771877288818, + "num_tokens": 956190586.0, + "step": 35472 + }, + { + "epoch": 2.9251257524532037, + "grad_norm": 2.162912368774414, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7594813108444214, + "num_tokens": 956219261.0, + "step": 35473 + }, + { + "epoch": 2.925208213078255, + "grad_norm": 2.1662683486938477, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.7966126799583435, + "num_tokens": 956247625.0, + "step": 35474 + }, + { + "epoch": 2.925290673703307, + "grad_norm": 2.0601460933685303, + "learning_rate": 1e-06, + "loss": 0.6455, + "mean_token_accuracy": 0.7836154699325562, + "num_tokens": 956279214.0, + "step": 35475 + }, + { + "epoch": 2.925373134328358, + "grad_norm": 2.1543920040130615, + "learning_rate": 1e-06, + "loss": 0.6459, + "mean_token_accuracy": 0.7767408490180969, + "num_tokens": 956306779.0, + "step": 35476 + }, + { + "epoch": 2.9254555949534096, + "grad_norm": 1.9077558517456055, + "learning_rate": 1e-06, + "loss": 0.5161, + "mean_token_accuracy": 0.8327710032463074, + "num_tokens": 956335765.0, + "step": 35477 + }, + { + "epoch": 2.9255380555784614, + "grad_norm": 1.9054481983184814, + "learning_rate": 1e-06, + "loss": 0.5843, + "mean_token_accuracy": 0.8141760230064392, + "num_tokens": 956367833.0, + "step": 35478 + }, + { + "epoch": 2.9256205162035127, + "grad_norm": 1.7817126512527466, + "learning_rate": 1e-06, + "loss": 0.6342, + "mean_token_accuracy": 0.7939222455024719, + "num_tokens": 956403129.0, + "step": 35479 + }, + { + "epoch": 2.9257029768285645, + "grad_norm": 2.6483609676361084, + "learning_rate": 1e-06, + "loss": 0.7934, + "mean_token_accuracy": 0.7438801527023315, + "num_tokens": 956426821.0, + "step": 35480 + }, + { + "epoch": 2.925785437453616, + "grad_norm": 2.4256205558776855, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7566297054290771, + "num_tokens": 956451305.0, + "step": 35481 + }, + { + "epoch": 2.9258678980786677, + "grad_norm": 2.1085739135742188, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7341512441635132, + "num_tokens": 956482938.0, + "step": 35482 + }, + { + "epoch": 2.925950358703719, + "grad_norm": 2.2269577980041504, + "learning_rate": 1e-06, + "loss": 0.7759, + "mean_token_accuracy": 0.7558289766311646, + "num_tokens": 956511162.0, + "step": 35483 + }, + { + "epoch": 2.9260328193287704, + "grad_norm": 2.273669481277466, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7624104022979736, + "num_tokens": 956540715.0, + "step": 35484 + }, + { + "epoch": 2.926115279953822, + "grad_norm": 2.1250367164611816, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.723423421382904, + "num_tokens": 956571237.0, + "step": 35485 + }, + { + "epoch": 2.9261977405788735, + "grad_norm": 1.9986076354980469, + "learning_rate": 1e-06, + "loss": 0.7226, + "mean_token_accuracy": 0.774250328540802, + "num_tokens": 956603378.0, + "step": 35486 + }, + { + "epoch": 2.9262802012039253, + "grad_norm": 2.163529396057129, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7622729539871216, + "num_tokens": 956632882.0, + "step": 35487 + }, + { + "epoch": 2.9263626618289766, + "grad_norm": 2.055274248123169, + "learning_rate": 1e-06, + "loss": 0.7199, + "mean_token_accuracy": 0.7633733153343201, + "num_tokens": 956662131.0, + "step": 35488 + }, + { + "epoch": 2.9264451224540284, + "grad_norm": 2.361894369125366, + "learning_rate": 1e-06, + "loss": 0.667, + "mean_token_accuracy": 0.77898108959198, + "num_tokens": 956687481.0, + "step": 35489 + }, + { + "epoch": 2.92652758307908, + "grad_norm": 1.98366379737854, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.7898043394088745, + "num_tokens": 956718683.0, + "step": 35490 + }, + { + "epoch": 2.926610043704131, + "grad_norm": 1.9904478788375854, + "learning_rate": 1e-06, + "loss": 0.6075, + "mean_token_accuracy": 0.8093008995056152, + "num_tokens": 956749724.0, + "step": 35491 + }, + { + "epoch": 2.926692504329183, + "grad_norm": 2.276197910308838, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7667821645736694, + "num_tokens": 956778540.0, + "step": 35492 + }, + { + "epoch": 2.9267749649542343, + "grad_norm": 2.421262502670288, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7511224150657654, + "num_tokens": 956803534.0, + "step": 35493 + }, + { + "epoch": 2.9268574255792856, + "grad_norm": 2.226724147796631, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7408750653266907, + "num_tokens": 956832091.0, + "step": 35494 + }, + { + "epoch": 2.9269398862043374, + "grad_norm": 1.8756895065307617, + "learning_rate": 1e-06, + "loss": 0.6647, + "mean_token_accuracy": 0.7909537553787231, + "num_tokens": 956869236.0, + "step": 35495 + }, + { + "epoch": 2.927022346829389, + "grad_norm": 2.07047176361084, + "learning_rate": 1e-06, + "loss": 0.5617, + "mean_token_accuracy": 0.8177483081817627, + "num_tokens": 956897885.0, + "step": 35496 + }, + { + "epoch": 2.9271048074544406, + "grad_norm": 2.094663619995117, + "learning_rate": 1e-06, + "loss": 0.6004, + "mean_token_accuracy": 0.7908944487571716, + "num_tokens": 956927049.0, + "step": 35497 + }, + { + "epoch": 2.927187268079492, + "grad_norm": 2.174328327178955, + "learning_rate": 1e-06, + "loss": 0.6916, + "mean_token_accuracy": 0.7724275588989258, + "num_tokens": 956953806.0, + "step": 35498 + }, + { + "epoch": 2.9272697287045437, + "grad_norm": 2.1962735652923584, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.735030472278595, + "num_tokens": 956981203.0, + "step": 35499 + }, + { + "epoch": 2.927352189329595, + "grad_norm": 2.078141927719116, + "learning_rate": 1e-06, + "loss": 0.7001, + "mean_token_accuracy": 0.7743687629699707, + "num_tokens": 957009383.0, + "step": 35500 + }, + { + "epoch": 2.9274346499546464, + "grad_norm": 1.9203312397003174, + "learning_rate": 1e-06, + "loss": 0.7572, + "mean_token_accuracy": 0.757184624671936, + "num_tokens": 957046111.0, + "step": 35501 + }, + { + "epoch": 2.927517110579698, + "grad_norm": 1.7679418325424194, + "learning_rate": 1e-06, + "loss": 0.621, + "mean_token_accuracy": 0.7934767007827759, + "num_tokens": 957082776.0, + "step": 35502 + }, + { + "epoch": 2.92759957120475, + "grad_norm": 2.0198943614959717, + "learning_rate": 1e-06, + "loss": 0.6599, + "mean_token_accuracy": 0.7850592136383057, + "num_tokens": 957113986.0, + "step": 35503 + }, + { + "epoch": 2.9276820318298014, + "grad_norm": 2.076751470565796, + "learning_rate": 1e-06, + "loss": 0.6844, + "mean_token_accuracy": 0.7805113196372986, + "num_tokens": 957142082.0, + "step": 35504 + }, + { + "epoch": 2.9277644924548527, + "grad_norm": 1.8561389446258545, + "learning_rate": 1e-06, + "loss": 0.7211, + "mean_token_accuracy": 0.7639328837394714, + "num_tokens": 957181438.0, + "step": 35505 + }, + { + "epoch": 2.9278469530799045, + "grad_norm": 2.0100364685058594, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7539995908737183, + "num_tokens": 957213247.0, + "step": 35506 + }, + { + "epoch": 2.927929413704956, + "grad_norm": 2.0052127838134766, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7619696855545044, + "num_tokens": 957246452.0, + "step": 35507 + }, + { + "epoch": 2.928011874330007, + "grad_norm": 2.0157506465911865, + "learning_rate": 1e-06, + "loss": 0.7235, + "mean_token_accuracy": 0.764686107635498, + "num_tokens": 957278851.0, + "step": 35508 + }, + { + "epoch": 2.928094334955059, + "grad_norm": 1.778623104095459, + "learning_rate": 1e-06, + "loss": 0.7229, + "mean_token_accuracy": 0.7652127742767334, + "num_tokens": 957316955.0, + "step": 35509 + }, + { + "epoch": 2.9281767955801103, + "grad_norm": 2.238391399383545, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7569714784622192, + "num_tokens": 957345924.0, + "step": 35510 + }, + { + "epoch": 2.928259256205162, + "grad_norm": 2.294719696044922, + "learning_rate": 1e-06, + "loss": 0.5926, + "mean_token_accuracy": 0.797968864440918, + "num_tokens": 957369971.0, + "step": 35511 + }, + { + "epoch": 2.9283417168302135, + "grad_norm": 2.10713791847229, + "learning_rate": 1e-06, + "loss": 0.6451, + "mean_token_accuracy": 0.7785199880599976, + "num_tokens": 957398924.0, + "step": 35512 + }, + { + "epoch": 2.9284241774552653, + "grad_norm": 2.0728538036346436, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.783319890499115, + "num_tokens": 957430602.0, + "step": 35513 + }, + { + "epoch": 2.9285066380803166, + "grad_norm": 2.353111505508423, + "learning_rate": 1e-06, + "loss": 0.7998, + "mean_token_accuracy": 0.7420566082000732, + "num_tokens": 957455602.0, + "step": 35514 + }, + { + "epoch": 2.928589098705368, + "grad_norm": 2.060234546661377, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.753131091594696, + "num_tokens": 957487155.0, + "step": 35515 + }, + { + "epoch": 2.92867155933042, + "grad_norm": 2.0982916355133057, + "learning_rate": 1e-06, + "loss": 0.5951, + "mean_token_accuracy": 0.7995948195457458, + "num_tokens": 957517085.0, + "step": 35516 + }, + { + "epoch": 2.928754019955471, + "grad_norm": 2.143537998199463, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7407318949699402, + "num_tokens": 957547338.0, + "step": 35517 + }, + { + "epoch": 2.928836480580523, + "grad_norm": 2.0612757205963135, + "learning_rate": 1e-06, + "loss": 0.6481, + "mean_token_accuracy": 0.7831317186355591, + "num_tokens": 957579142.0, + "step": 35518 + }, + { + "epoch": 2.9289189412055743, + "grad_norm": 1.9629393815994263, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7635002136230469, + "num_tokens": 957612797.0, + "step": 35519 + }, + { + "epoch": 2.929001401830626, + "grad_norm": 2.092146396636963, + "learning_rate": 1e-06, + "loss": 0.6762, + "mean_token_accuracy": 0.7805625796318054, + "num_tokens": 957644199.0, + "step": 35520 + }, + { + "epoch": 2.9290838624556774, + "grad_norm": 2.221926212310791, + "learning_rate": 1e-06, + "loss": 0.7541, + "mean_token_accuracy": 0.7557435035705566, + "num_tokens": 957671608.0, + "step": 35521 + }, + { + "epoch": 2.9291663230807288, + "grad_norm": 2.0752761363983154, + "learning_rate": 1e-06, + "loss": 0.618, + "mean_token_accuracy": 0.7927049398422241, + "num_tokens": 957702700.0, + "step": 35522 + }, + { + "epoch": 2.9292487837057806, + "grad_norm": 1.83815598487854, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7814750671386719, + "num_tokens": 957740085.0, + "step": 35523 + }, + { + "epoch": 2.929331244330832, + "grad_norm": 1.992206335067749, + "learning_rate": 1e-06, + "loss": 0.6552, + "mean_token_accuracy": 0.7907984256744385, + "num_tokens": 957774711.0, + "step": 35524 + }, + { + "epoch": 2.9294137049558837, + "grad_norm": 2.065516710281372, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7438486218452454, + "num_tokens": 957805393.0, + "step": 35525 + }, + { + "epoch": 2.929496165580935, + "grad_norm": 2.6173043251037598, + "learning_rate": 1e-06, + "loss": 0.7717, + "mean_token_accuracy": 0.7500366568565369, + "num_tokens": 957825539.0, + "step": 35526 + }, + { + "epoch": 2.929578626205987, + "grad_norm": 1.9296363592147827, + "learning_rate": 1e-06, + "loss": 0.6636, + "mean_token_accuracy": 0.7838906049728394, + "num_tokens": 957860651.0, + "step": 35527 + }, + { + "epoch": 2.929661086831038, + "grad_norm": 1.9919219017028809, + "learning_rate": 1e-06, + "loss": 0.5695, + "mean_token_accuracy": 0.8129019737243652, + "num_tokens": 957890829.0, + "step": 35528 + }, + { + "epoch": 2.9297435474560896, + "grad_norm": 2.008727550506592, + "learning_rate": 1e-06, + "loss": 0.5325, + "mean_token_accuracy": 0.8242976665496826, + "num_tokens": 957925659.0, + "step": 35529 + }, + { + "epoch": 2.9298260080811414, + "grad_norm": 2.1682097911834717, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7441790699958801, + "num_tokens": 957955099.0, + "step": 35530 + }, + { + "epoch": 2.9299084687061927, + "grad_norm": 1.8353655338287354, + "learning_rate": 1e-06, + "loss": 0.5838, + "mean_token_accuracy": 0.8138633966445923, + "num_tokens": 957990266.0, + "step": 35531 + }, + { + "epoch": 2.929990929331244, + "grad_norm": 2.007906436920166, + "learning_rate": 1e-06, + "loss": 0.5666, + "mean_token_accuracy": 0.8061248660087585, + "num_tokens": 958021176.0, + "step": 35532 + }, + { + "epoch": 2.930073389956296, + "grad_norm": 2.165724992752075, + "learning_rate": 1e-06, + "loss": 0.672, + "mean_token_accuracy": 0.7779816389083862, + "num_tokens": 958051751.0, + "step": 35533 + }, + { + "epoch": 2.9301558505813476, + "grad_norm": 2.114804983139038, + "learning_rate": 1e-06, + "loss": 0.711, + "mean_token_accuracy": 0.77276211977005, + "num_tokens": 958080642.0, + "step": 35534 + }, + { + "epoch": 2.930238311206399, + "grad_norm": 1.9784165620803833, + "learning_rate": 1e-06, + "loss": 0.5041, + "mean_token_accuracy": 0.8127508759498596, + "num_tokens": 958111005.0, + "step": 35535 + }, + { + "epoch": 2.9303207718314503, + "grad_norm": 1.9639759063720703, + "learning_rate": 1e-06, + "loss": 0.7504, + "mean_token_accuracy": 0.7701696157455444, + "num_tokens": 958147697.0, + "step": 35536 + }, + { + "epoch": 2.930403232456502, + "grad_norm": 2.150270700454712, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7481579184532166, + "num_tokens": 958174762.0, + "step": 35537 + }, + { + "epoch": 2.9304856930815535, + "grad_norm": 2.2549550533294678, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.7868176698684692, + "num_tokens": 958201181.0, + "step": 35538 + }, + { + "epoch": 2.930568153706605, + "grad_norm": 2.1043756008148193, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7575682997703552, + "num_tokens": 958232093.0, + "step": 35539 + }, + { + "epoch": 2.9306506143316566, + "grad_norm": 1.7383226156234741, + "learning_rate": 1e-06, + "loss": 0.549, + "mean_token_accuracy": 0.8221334218978882, + "num_tokens": 958267363.0, + "step": 35540 + }, + { + "epoch": 2.9307330749567084, + "grad_norm": 1.8951482772827148, + "learning_rate": 1e-06, + "loss": 0.7232, + "mean_token_accuracy": 0.7774640321731567, + "num_tokens": 958302603.0, + "step": 35541 + }, + { + "epoch": 2.93081553558176, + "grad_norm": 1.781267762184143, + "learning_rate": 1e-06, + "loss": 0.7531, + "mean_token_accuracy": 0.7653179168701172, + "num_tokens": 958342222.0, + "step": 35542 + }, + { + "epoch": 2.930897996206811, + "grad_norm": 2.005617380142212, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.7512222528457642, + "num_tokens": 958373804.0, + "step": 35543 + }, + { + "epoch": 2.930980456831863, + "grad_norm": 2.1233983039855957, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7492749691009521, + "num_tokens": 958404152.0, + "step": 35544 + }, + { + "epoch": 2.9310629174569143, + "grad_norm": 2.098189353942871, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7453802227973938, + "num_tokens": 958436509.0, + "step": 35545 + }, + { + "epoch": 2.9311453780819656, + "grad_norm": 2.042651653289795, + "learning_rate": 1e-06, + "loss": 0.6812, + "mean_token_accuracy": 0.7797180414199829, + "num_tokens": 958465520.0, + "step": 35546 + }, + { + "epoch": 2.9312278387070174, + "grad_norm": 1.9291009902954102, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7780596017837524, + "num_tokens": 958497990.0, + "step": 35547 + }, + { + "epoch": 2.931310299332069, + "grad_norm": 2.2358171939849854, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7743057012557983, + "num_tokens": 958524968.0, + "step": 35548 + }, + { + "epoch": 2.9313927599571206, + "grad_norm": 1.9282091856002808, + "learning_rate": 1e-06, + "loss": 0.7262, + "mean_token_accuracy": 0.7646478414535522, + "num_tokens": 958557190.0, + "step": 35549 + }, + { + "epoch": 2.931475220582172, + "grad_norm": 1.9389324188232422, + "learning_rate": 1e-06, + "loss": 0.5915, + "mean_token_accuracy": 0.8064745664596558, + "num_tokens": 958588810.0, + "step": 35550 + }, + { + "epoch": 2.9315576812072237, + "grad_norm": 2.125166893005371, + "learning_rate": 1e-06, + "loss": 0.6321, + "mean_token_accuracy": 0.79270339012146, + "num_tokens": 958617012.0, + "step": 35551 + }, + { + "epoch": 2.931640141832275, + "grad_norm": 1.7234638929367065, + "learning_rate": 1e-06, + "loss": 0.5298, + "mean_token_accuracy": 0.8251560926437378, + "num_tokens": 958654183.0, + "step": 35552 + }, + { + "epoch": 2.9317226024573264, + "grad_norm": 2.094660520553589, + "learning_rate": 1e-06, + "loss": 0.6468, + "mean_token_accuracy": 0.7963860034942627, + "num_tokens": 958685753.0, + "step": 35553 + }, + { + "epoch": 2.931805063082378, + "grad_norm": 2.3110547065734863, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7524502277374268, + "num_tokens": 958712565.0, + "step": 35554 + }, + { + "epoch": 2.9318875237074296, + "grad_norm": 1.8334389925003052, + "learning_rate": 1e-06, + "loss": 0.5175, + "mean_token_accuracy": 0.8253679275512695, + "num_tokens": 958747456.0, + "step": 35555 + }, + { + "epoch": 2.9319699843324813, + "grad_norm": 1.9434524774551392, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7604363560676575, + "num_tokens": 958780227.0, + "step": 35556 + }, + { + "epoch": 2.9320524449575327, + "grad_norm": 2.5911121368408203, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7690644264221191, + "num_tokens": 958799229.0, + "step": 35557 + }, + { + "epoch": 2.9321349055825845, + "grad_norm": 2.0886518955230713, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7563333511352539, + "num_tokens": 958830584.0, + "step": 35558 + }, + { + "epoch": 2.932217366207636, + "grad_norm": 2.6076114177703857, + "learning_rate": 1e-06, + "loss": 0.7289, + "mean_token_accuracy": 0.7610673904418945, + "num_tokens": 958852934.0, + "step": 35559 + }, + { + "epoch": 2.932299826832687, + "grad_norm": 1.8039087057113647, + "learning_rate": 1e-06, + "loss": 0.6006, + "mean_token_accuracy": 0.7931410074234009, + "num_tokens": 958886425.0, + "step": 35560 + }, + { + "epoch": 2.932382287457739, + "grad_norm": 2.025164842605591, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7456705570220947, + "num_tokens": 958917273.0, + "step": 35561 + }, + { + "epoch": 2.9324647480827903, + "grad_norm": 2.6216113567352295, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7392358779907227, + "num_tokens": 958938687.0, + "step": 35562 + }, + { + "epoch": 2.932547208707842, + "grad_norm": 1.96371328830719, + "learning_rate": 1e-06, + "loss": 0.7847, + "mean_token_accuracy": 0.7537183165550232, + "num_tokens": 958975764.0, + "step": 35563 + }, + { + "epoch": 2.9326296693328935, + "grad_norm": 1.8976866006851196, + "learning_rate": 1e-06, + "loss": 0.6453, + "mean_token_accuracy": 0.7819600701332092, + "num_tokens": 959011883.0, + "step": 35564 + }, + { + "epoch": 2.9327121299579453, + "grad_norm": 1.9936318397521973, + "learning_rate": 1e-06, + "loss": 0.6953, + "mean_token_accuracy": 0.7676621675491333, + "num_tokens": 959044246.0, + "step": 35565 + }, + { + "epoch": 2.9327945905829966, + "grad_norm": 2.008220911026001, + "learning_rate": 1e-06, + "loss": 0.7555, + "mean_token_accuracy": 0.7543996572494507, + "num_tokens": 959075309.0, + "step": 35566 + }, + { + "epoch": 2.932877051208048, + "grad_norm": 2.0067460536956787, + "learning_rate": 1e-06, + "loss": 0.7383, + "mean_token_accuracy": 0.758142352104187, + "num_tokens": 959107392.0, + "step": 35567 + }, + { + "epoch": 2.9329595118330998, + "grad_norm": 2.070953845977783, + "learning_rate": 1e-06, + "loss": 0.6615, + "mean_token_accuracy": 0.788004457950592, + "num_tokens": 959138823.0, + "step": 35568 + }, + { + "epoch": 2.933041972458151, + "grad_norm": 2.106205463409424, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7801626324653625, + "num_tokens": 959169268.0, + "step": 35569 + }, + { + "epoch": 2.933124433083203, + "grad_norm": 2.2918541431427, + "learning_rate": 1e-06, + "loss": 0.7731, + "mean_token_accuracy": 0.7475218176841736, + "num_tokens": 959193744.0, + "step": 35570 + }, + { + "epoch": 2.9332068937082543, + "grad_norm": 2.228703737258911, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.7382976412773132, + "num_tokens": 959221150.0, + "step": 35571 + }, + { + "epoch": 2.933289354333306, + "grad_norm": 2.3219592571258545, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7602574229240417, + "num_tokens": 959246185.0, + "step": 35572 + }, + { + "epoch": 2.9333718149583574, + "grad_norm": 2.121037721633911, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7571122646331787, + "num_tokens": 959274823.0, + "step": 35573 + }, + { + "epoch": 2.9334542755834088, + "grad_norm": 2.1505606174468994, + "learning_rate": 1e-06, + "loss": 0.6164, + "mean_token_accuracy": 0.7961865663528442, + "num_tokens": 959301463.0, + "step": 35574 + }, + { + "epoch": 2.9335367362084606, + "grad_norm": 2.11604905128479, + "learning_rate": 1e-06, + "loss": 0.679, + "mean_token_accuracy": 0.7749291658401489, + "num_tokens": 959329877.0, + "step": 35575 + }, + { + "epoch": 2.933619196833512, + "grad_norm": 2.100579023361206, + "learning_rate": 1e-06, + "loss": 0.7337, + "mean_token_accuracy": 0.7530006766319275, + "num_tokens": 959357901.0, + "step": 35576 + }, + { + "epoch": 2.9337016574585633, + "grad_norm": 2.1728386878967285, + "learning_rate": 1e-06, + "loss": 0.7305, + "mean_token_accuracy": 0.7527124881744385, + "num_tokens": 959387171.0, + "step": 35577 + }, + { + "epoch": 2.933784118083615, + "grad_norm": 2.2186343669891357, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7448086738586426, + "num_tokens": 959418706.0, + "step": 35578 + }, + { + "epoch": 2.933866578708667, + "grad_norm": 1.9023373126983643, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7561982870101929, + "num_tokens": 959452007.0, + "step": 35579 + }, + { + "epoch": 2.933949039333718, + "grad_norm": 2.3112781047821045, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.7615598440170288, + "num_tokens": 959476588.0, + "step": 35580 + }, + { + "epoch": 2.9340314999587696, + "grad_norm": 2.1418683528900146, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.746406078338623, + "num_tokens": 959507837.0, + "step": 35581 + }, + { + "epoch": 2.9341139605838213, + "grad_norm": 2.118527412414551, + "learning_rate": 1e-06, + "loss": 0.7676, + "mean_token_accuracy": 0.748725414276123, + "num_tokens": 959537331.0, + "step": 35582 + }, + { + "epoch": 2.9341964212088727, + "grad_norm": 2.2521347999572754, + "learning_rate": 1e-06, + "loss": 0.5994, + "mean_token_accuracy": 0.805851936340332, + "num_tokens": 959563726.0, + "step": 35583 + }, + { + "epoch": 2.934278881833924, + "grad_norm": 2.173334836959839, + "learning_rate": 1e-06, + "loss": 0.6945, + "mean_token_accuracy": 0.7813756465911865, + "num_tokens": 959593860.0, + "step": 35584 + }, + { + "epoch": 2.934361342458976, + "grad_norm": 2.2567896842956543, + "learning_rate": 1e-06, + "loss": 0.6715, + "mean_token_accuracy": 0.7825244069099426, + "num_tokens": 959620163.0, + "step": 35585 + }, + { + "epoch": 2.9344438030840276, + "grad_norm": 2.224713087081909, + "learning_rate": 1e-06, + "loss": 0.6911, + "mean_token_accuracy": 0.7787261605262756, + "num_tokens": 959647333.0, + "step": 35586 + }, + { + "epoch": 2.934526263709079, + "grad_norm": 2.2625677585601807, + "learning_rate": 1e-06, + "loss": 0.7586, + "mean_token_accuracy": 0.7654085159301758, + "num_tokens": 959673207.0, + "step": 35587 + }, + { + "epoch": 2.9346087243341303, + "grad_norm": 1.9902697801589966, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7813146114349365, + "num_tokens": 959704990.0, + "step": 35588 + }, + { + "epoch": 2.934691184959182, + "grad_norm": 1.9988104104995728, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.7952871918678284, + "num_tokens": 959738893.0, + "step": 35589 + }, + { + "epoch": 2.9347736455842335, + "grad_norm": 2.138646125793457, + "learning_rate": 1e-06, + "loss": 0.6402, + "mean_token_accuracy": 0.7955096364021301, + "num_tokens": 959767588.0, + "step": 35590 + }, + { + "epoch": 2.934856106209285, + "grad_norm": 2.2241408824920654, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7680993676185608, + "num_tokens": 959795044.0, + "step": 35591 + }, + { + "epoch": 2.9349385668343366, + "grad_norm": 2.285738945007324, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7617716789245605, + "num_tokens": 959823515.0, + "step": 35592 + }, + { + "epoch": 2.935021027459388, + "grad_norm": 2.0834226608276367, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7758946418762207, + "num_tokens": 959858627.0, + "step": 35593 + }, + { + "epoch": 2.9351034880844398, + "grad_norm": 2.416822910308838, + "learning_rate": 1e-06, + "loss": 0.6517, + "mean_token_accuracy": 0.7873160243034363, + "num_tokens": 959884530.0, + "step": 35594 + }, + { + "epoch": 2.935185948709491, + "grad_norm": 2.2626757621765137, + "learning_rate": 1e-06, + "loss": 0.6287, + "mean_token_accuracy": 0.7851971387863159, + "num_tokens": 959910822.0, + "step": 35595 + }, + { + "epoch": 2.935268409334543, + "grad_norm": 2.061267375946045, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7596614956855774, + "num_tokens": 959941779.0, + "step": 35596 + }, + { + "epoch": 2.9353508699595943, + "grad_norm": 1.9414284229278564, + "learning_rate": 1e-06, + "loss": 0.6157, + "mean_token_accuracy": 0.7949826717376709, + "num_tokens": 959974460.0, + "step": 35597 + }, + { + "epoch": 2.9354333305846456, + "grad_norm": 2.245105028152466, + "learning_rate": 1e-06, + "loss": 0.7573, + "mean_token_accuracy": 0.7543671131134033, + "num_tokens": 960003769.0, + "step": 35598 + }, + { + "epoch": 2.9355157912096974, + "grad_norm": 2.142697811126709, + "learning_rate": 1e-06, + "loss": 0.6993, + "mean_token_accuracy": 0.7706804275512695, + "num_tokens": 960032970.0, + "step": 35599 + }, + { + "epoch": 2.9355982518347488, + "grad_norm": 1.8073352575302124, + "learning_rate": 1e-06, + "loss": 0.5688, + "mean_token_accuracy": 0.8147069215774536, + "num_tokens": 960068126.0, + "step": 35600 + }, + { + "epoch": 2.9356807124598006, + "grad_norm": 2.113135814666748, + "learning_rate": 1e-06, + "loss": 0.7522, + "mean_token_accuracy": 0.7559929490089417, + "num_tokens": 960097686.0, + "step": 35601 + }, + { + "epoch": 2.935763173084852, + "grad_norm": 2.580728769302368, + "learning_rate": 1e-06, + "loss": 0.626, + "mean_token_accuracy": 0.7868643999099731, + "num_tokens": 960116868.0, + "step": 35602 + }, + { + "epoch": 2.9358456337099037, + "grad_norm": 1.9119611978530884, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7674921154975891, + "num_tokens": 960151068.0, + "step": 35603 + }, + { + "epoch": 2.935928094334955, + "grad_norm": 2.1505701541900635, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7628323435783386, + "num_tokens": 960179270.0, + "step": 35604 + }, + { + "epoch": 2.9360105549600064, + "grad_norm": 1.9393773078918457, + "learning_rate": 1e-06, + "loss": 0.7605, + "mean_token_accuracy": 0.7635300159454346, + "num_tokens": 960212194.0, + "step": 35605 + }, + { + "epoch": 2.936093015585058, + "grad_norm": 1.8217450380325317, + "learning_rate": 1e-06, + "loss": 0.4983, + "mean_token_accuracy": 0.820438027381897, + "num_tokens": 960245422.0, + "step": 35606 + }, + { + "epoch": 2.9361754762101095, + "grad_norm": 2.04111647605896, + "learning_rate": 1e-06, + "loss": 0.7077, + "mean_token_accuracy": 0.7720820307731628, + "num_tokens": 960278045.0, + "step": 35607 + }, + { + "epoch": 2.9362579368351613, + "grad_norm": 1.821883201599121, + "learning_rate": 1e-06, + "loss": 0.5613, + "mean_token_accuracy": 0.8168800473213196, + "num_tokens": 960315781.0, + "step": 35608 + }, + { + "epoch": 2.9363403974602127, + "grad_norm": 1.9346153736114502, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7669738531112671, + "num_tokens": 960347551.0, + "step": 35609 + }, + { + "epoch": 2.9364228580852645, + "grad_norm": 1.903210163116455, + "learning_rate": 1e-06, + "loss": 0.6754, + "mean_token_accuracy": 0.7674592733383179, + "num_tokens": 960379498.0, + "step": 35610 + }, + { + "epoch": 2.936505318710316, + "grad_norm": 1.970671534538269, + "learning_rate": 1e-06, + "loss": 0.7726, + "mean_token_accuracy": 0.749536395072937, + "num_tokens": 960411570.0, + "step": 35611 + }, + { + "epoch": 2.936587779335367, + "grad_norm": 2.1260077953338623, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.7753326892852783, + "num_tokens": 960441584.0, + "step": 35612 + }, + { + "epoch": 2.936670239960419, + "grad_norm": 2.173356294631958, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7482144832611084, + "num_tokens": 960472738.0, + "step": 35613 + }, + { + "epoch": 2.9367527005854703, + "grad_norm": 1.9946521520614624, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7673726081848145, + "num_tokens": 960506519.0, + "step": 35614 + }, + { + "epoch": 2.936835161210522, + "grad_norm": 1.9003406763076782, + "learning_rate": 1e-06, + "loss": 0.4565, + "mean_token_accuracy": 0.839958667755127, + "num_tokens": 960539631.0, + "step": 35615 + }, + { + "epoch": 2.9369176218355735, + "grad_norm": 2.1619558334350586, + "learning_rate": 1e-06, + "loss": 0.6296, + "mean_token_accuracy": 0.7956691384315491, + "num_tokens": 960567770.0, + "step": 35616 + }, + { + "epoch": 2.9370000824606253, + "grad_norm": 2.097285509109497, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7742855548858643, + "num_tokens": 960596756.0, + "step": 35617 + }, + { + "epoch": 2.9370825430856766, + "grad_norm": 2.225393533706665, + "learning_rate": 1e-06, + "loss": 0.7648, + "mean_token_accuracy": 0.7546786069869995, + "num_tokens": 960626867.0, + "step": 35618 + }, + { + "epoch": 2.937165003710728, + "grad_norm": 2.030520439147949, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7372775077819824, + "num_tokens": 960660185.0, + "step": 35619 + }, + { + "epoch": 2.9372474643357798, + "grad_norm": 2.1893460750579834, + "learning_rate": 1e-06, + "loss": 0.7216, + "mean_token_accuracy": 0.7706286311149597, + "num_tokens": 960688204.0, + "step": 35620 + }, + { + "epoch": 2.937329924960831, + "grad_norm": 1.8729232549667358, + "learning_rate": 1e-06, + "loss": 0.66, + "mean_token_accuracy": 0.7822238802909851, + "num_tokens": 960726947.0, + "step": 35621 + }, + { + "epoch": 2.9374123855858825, + "grad_norm": 2.1414642333984375, + "learning_rate": 1e-06, + "loss": 0.7497, + "mean_token_accuracy": 0.7592432498931885, + "num_tokens": 960756554.0, + "step": 35622 + }, + { + "epoch": 2.9374948462109343, + "grad_norm": 2.090322494506836, + "learning_rate": 1e-06, + "loss": 0.7345, + "mean_token_accuracy": 0.7716652154922485, + "num_tokens": 960789256.0, + "step": 35623 + }, + { + "epoch": 2.937577306835986, + "grad_norm": 1.7732969522476196, + "learning_rate": 1e-06, + "loss": 0.6815, + "mean_token_accuracy": 0.7829744815826416, + "num_tokens": 960826958.0, + "step": 35624 + }, + { + "epoch": 2.9376597674610374, + "grad_norm": 1.9655578136444092, + "learning_rate": 1e-06, + "loss": 0.5876, + "mean_token_accuracy": 0.8003565073013306, + "num_tokens": 960860463.0, + "step": 35625 + }, + { + "epoch": 2.9377422280860888, + "grad_norm": 2.3545875549316406, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7794644236564636, + "num_tokens": 960886076.0, + "step": 35626 + }, + { + "epoch": 2.9378246887111406, + "grad_norm": 2.1554176807403564, + "learning_rate": 1e-06, + "loss": 0.6894, + "mean_token_accuracy": 0.7683181762695312, + "num_tokens": 960915862.0, + "step": 35627 + }, + { + "epoch": 2.937907149336192, + "grad_norm": 2.102055072784424, + "learning_rate": 1e-06, + "loss": 0.7202, + "mean_token_accuracy": 0.7660536766052246, + "num_tokens": 960947961.0, + "step": 35628 + }, + { + "epoch": 2.9379896099612433, + "grad_norm": 2.0137054920196533, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7562862038612366, + "num_tokens": 960986254.0, + "step": 35629 + }, + { + "epoch": 2.938072070586295, + "grad_norm": 2.301943302154541, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7636966109275818, + "num_tokens": 961012497.0, + "step": 35630 + }, + { + "epoch": 2.938154531211347, + "grad_norm": 1.9612936973571777, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7730215787887573, + "num_tokens": 961045942.0, + "step": 35631 + }, + { + "epoch": 2.938236991836398, + "grad_norm": 2.064742088317871, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.7805755734443665, + "num_tokens": 961078839.0, + "step": 35632 + }, + { + "epoch": 2.9383194524614495, + "grad_norm": 1.8256171941757202, + "learning_rate": 1e-06, + "loss": 0.6944, + "mean_token_accuracy": 0.7694339752197266, + "num_tokens": 961118106.0, + "step": 35633 + }, + { + "epoch": 2.9384019130865013, + "grad_norm": 1.9761347770690918, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.7551512718200684, + "num_tokens": 961151900.0, + "step": 35634 + }, + { + "epoch": 2.9384843737115527, + "grad_norm": 1.8465449810028076, + "learning_rate": 1e-06, + "loss": 0.5736, + "mean_token_accuracy": 0.7989866733551025, + "num_tokens": 961186279.0, + "step": 35635 + }, + { + "epoch": 2.938566834336604, + "grad_norm": 2.3100907802581787, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7746968269348145, + "num_tokens": 961214931.0, + "step": 35636 + }, + { + "epoch": 2.938649294961656, + "grad_norm": 2.1957809925079346, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7652010917663574, + "num_tokens": 961243880.0, + "step": 35637 + }, + { + "epoch": 2.938731755586707, + "grad_norm": 1.8408784866333008, + "learning_rate": 1e-06, + "loss": 0.5682, + "mean_token_accuracy": 0.8175707459449768, + "num_tokens": 961277729.0, + "step": 35638 + }, + { + "epoch": 2.938814216211759, + "grad_norm": 1.873408555984497, + "learning_rate": 1e-06, + "loss": 0.6266, + "mean_token_accuracy": 0.785043478012085, + "num_tokens": 961312540.0, + "step": 35639 + }, + { + "epoch": 2.9388966768368103, + "grad_norm": 1.9916445016860962, + "learning_rate": 1e-06, + "loss": 0.5796, + "mean_token_accuracy": 0.8083134889602661, + "num_tokens": 961343653.0, + "step": 35640 + }, + { + "epoch": 2.938979137461862, + "grad_norm": 2.4459359645843506, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7558327913284302, + "num_tokens": 961367167.0, + "step": 35641 + }, + { + "epoch": 2.9390615980869135, + "grad_norm": 2.1761560440063477, + "learning_rate": 1e-06, + "loss": 0.74, + "mean_token_accuracy": 0.7616778612136841, + "num_tokens": 961395274.0, + "step": 35642 + }, + { + "epoch": 2.939144058711965, + "grad_norm": 2.0502402782440186, + "learning_rate": 1e-06, + "loss": 0.4982, + "mean_token_accuracy": 0.826594889163971, + "num_tokens": 961424943.0, + "step": 35643 + }, + { + "epoch": 2.9392265193370166, + "grad_norm": 2.1725010871887207, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.755621075630188, + "num_tokens": 961453587.0, + "step": 35644 + }, + { + "epoch": 2.939308979962068, + "grad_norm": 1.936939001083374, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7774808406829834, + "num_tokens": 961489574.0, + "step": 35645 + }, + { + "epoch": 2.9393914405871198, + "grad_norm": 2.15549635887146, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.775080680847168, + "num_tokens": 961516017.0, + "step": 35646 + }, + { + "epoch": 2.939473901212171, + "grad_norm": 2.2134978771209717, + "learning_rate": 1e-06, + "loss": 0.7842, + "mean_token_accuracy": 0.7510949969291687, + "num_tokens": 961543027.0, + "step": 35647 + }, + { + "epoch": 2.939556361837223, + "grad_norm": 2.2388699054718018, + "learning_rate": 1e-06, + "loss": 0.7678, + "mean_token_accuracy": 0.7483624219894409, + "num_tokens": 961570138.0, + "step": 35648 + }, + { + "epoch": 2.9396388224622743, + "grad_norm": 2.324842691421509, + "learning_rate": 1e-06, + "loss": 0.6868, + "mean_token_accuracy": 0.7757972478866577, + "num_tokens": 961597814.0, + "step": 35649 + }, + { + "epoch": 2.9397212830873256, + "grad_norm": 1.9188662767410278, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7394638061523438, + "num_tokens": 961630829.0, + "step": 35650 + }, + { + "epoch": 2.9398037437123774, + "grad_norm": 1.875507116317749, + "learning_rate": 1e-06, + "loss": 0.7397, + "mean_token_accuracy": 0.7500871419906616, + "num_tokens": 961668684.0, + "step": 35651 + }, + { + "epoch": 2.9398862043374288, + "grad_norm": 2.5081143379211426, + "learning_rate": 1e-06, + "loss": 0.7015, + "mean_token_accuracy": 0.7696051001548767, + "num_tokens": 961693046.0, + "step": 35652 + }, + { + "epoch": 2.9399686649624805, + "grad_norm": 1.9247156381607056, + "learning_rate": 1e-06, + "loss": 0.5803, + "mean_token_accuracy": 0.8110895156860352, + "num_tokens": 961727673.0, + "step": 35653 + }, + { + "epoch": 2.940051125587532, + "grad_norm": 2.2052910327911377, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7463759779930115, + "num_tokens": 961756840.0, + "step": 35654 + }, + { + "epoch": 2.9401335862125837, + "grad_norm": 2.1628596782684326, + "learning_rate": 1e-06, + "loss": 0.6042, + "mean_token_accuracy": 0.7939437031745911, + "num_tokens": 961785921.0, + "step": 35655 + }, + { + "epoch": 2.940216046837635, + "grad_norm": 1.9744895696640015, + "learning_rate": 1e-06, + "loss": 0.6952, + "mean_token_accuracy": 0.770471453666687, + "num_tokens": 961821010.0, + "step": 35656 + }, + { + "epoch": 2.9402985074626864, + "grad_norm": 2.2921693325042725, + "learning_rate": 1e-06, + "loss": 0.7204, + "mean_token_accuracy": 0.769740104675293, + "num_tokens": 961846683.0, + "step": 35657 + }, + { + "epoch": 2.940380968087738, + "grad_norm": 2.2719812393188477, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7406230568885803, + "num_tokens": 961876847.0, + "step": 35658 + }, + { + "epoch": 2.9404634287127895, + "grad_norm": 2.2324330806732178, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7671369910240173, + "num_tokens": 961907372.0, + "step": 35659 + }, + { + "epoch": 2.9405458893378413, + "grad_norm": 2.267066240310669, + "learning_rate": 1e-06, + "loss": 0.7276, + "mean_token_accuracy": 0.7615060806274414, + "num_tokens": 961933478.0, + "step": 35660 + }, + { + "epoch": 2.9406283499628927, + "grad_norm": 2.139256000518799, + "learning_rate": 1e-06, + "loss": 0.5957, + "mean_token_accuracy": 0.8005954027175903, + "num_tokens": 961960932.0, + "step": 35661 + }, + { + "epoch": 2.9407108105879445, + "grad_norm": 2.143543004989624, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.7522678971290588, + "num_tokens": 961990320.0, + "step": 35662 + }, + { + "epoch": 2.940793271212996, + "grad_norm": 1.8235907554626465, + "learning_rate": 1e-06, + "loss": 0.6838, + "mean_token_accuracy": 0.7855885028839111, + "num_tokens": 962027459.0, + "step": 35663 + }, + { + "epoch": 2.940875731838047, + "grad_norm": 2.060730218887329, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7671980857849121, + "num_tokens": 962059659.0, + "step": 35664 + }, + { + "epoch": 2.940958192463099, + "grad_norm": 1.8197846412658691, + "learning_rate": 1e-06, + "loss": 0.6833, + "mean_token_accuracy": 0.7692459225654602, + "num_tokens": 962095262.0, + "step": 35665 + }, + { + "epoch": 2.9410406530881503, + "grad_norm": 2.082387924194336, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.7846397757530212, + "num_tokens": 962125078.0, + "step": 35666 + }, + { + "epoch": 2.9411231137132017, + "grad_norm": 2.110832929611206, + "learning_rate": 1e-06, + "loss": 0.7581, + "mean_token_accuracy": 0.753839373588562, + "num_tokens": 962159170.0, + "step": 35667 + }, + { + "epoch": 2.9412055743382535, + "grad_norm": 2.1739556789398193, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7435232996940613, + "num_tokens": 962191043.0, + "step": 35668 + }, + { + "epoch": 2.9412880349633053, + "grad_norm": 2.06048321723938, + "learning_rate": 1e-06, + "loss": 0.5719, + "mean_token_accuracy": 0.8039158582687378, + "num_tokens": 962219875.0, + "step": 35669 + }, + { + "epoch": 2.9413704955883566, + "grad_norm": 2.4664394855499268, + "learning_rate": 1e-06, + "loss": 0.6657, + "mean_token_accuracy": 0.7821055054664612, + "num_tokens": 962243425.0, + "step": 35670 + }, + { + "epoch": 2.941452956213408, + "grad_norm": 2.342954397201538, + "learning_rate": 1e-06, + "loss": 0.714, + "mean_token_accuracy": 0.7617158889770508, + "num_tokens": 962270216.0, + "step": 35671 + }, + { + "epoch": 2.9415354168384598, + "grad_norm": 2.0666592121124268, + "learning_rate": 1e-06, + "loss": 0.7473, + "mean_token_accuracy": 0.7659190893173218, + "num_tokens": 962300187.0, + "step": 35672 + }, + { + "epoch": 2.941617877463511, + "grad_norm": 2.2440571784973145, + "learning_rate": 1e-06, + "loss": 0.713, + "mean_token_accuracy": 0.7618513107299805, + "num_tokens": 962328242.0, + "step": 35673 + }, + { + "epoch": 2.9417003380885625, + "grad_norm": 2.1962051391601562, + "learning_rate": 1e-06, + "loss": 0.6667, + "mean_token_accuracy": 0.7856374382972717, + "num_tokens": 962358218.0, + "step": 35674 + }, + { + "epoch": 2.9417827987136143, + "grad_norm": 1.8198062181472778, + "learning_rate": 1e-06, + "loss": 0.6901, + "mean_token_accuracy": 0.7758228778839111, + "num_tokens": 962393582.0, + "step": 35675 + }, + { + "epoch": 2.941865259338666, + "grad_norm": 2.2196154594421387, + "learning_rate": 1e-06, + "loss": 0.6809, + "mean_token_accuracy": 0.7800161242485046, + "num_tokens": 962419726.0, + "step": 35676 + }, + { + "epoch": 2.9419477199637174, + "grad_norm": 1.977451205253601, + "learning_rate": 1e-06, + "loss": 0.7003, + "mean_token_accuracy": 0.7664860486984253, + "num_tokens": 962455286.0, + "step": 35677 + }, + { + "epoch": 2.9420301805887688, + "grad_norm": 2.0885655879974365, + "learning_rate": 1e-06, + "loss": 0.6317, + "mean_token_accuracy": 0.7914802432060242, + "num_tokens": 962482654.0, + "step": 35678 + }, + { + "epoch": 2.9421126412138205, + "grad_norm": 2.1198153495788574, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7810455560684204, + "num_tokens": 962513001.0, + "step": 35679 + }, + { + "epoch": 2.942195101838872, + "grad_norm": 2.1560916900634766, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7512861490249634, + "num_tokens": 962542761.0, + "step": 35680 + }, + { + "epoch": 2.9422775624639232, + "grad_norm": 2.1765389442443848, + "learning_rate": 1e-06, + "loss": 0.7842, + "mean_token_accuracy": 0.7626205682754517, + "num_tokens": 962571498.0, + "step": 35681 + }, + { + "epoch": 2.942360023088975, + "grad_norm": 2.017488479614258, + "learning_rate": 1e-06, + "loss": 0.7157, + "mean_token_accuracy": 0.7702063918113708, + "num_tokens": 962603656.0, + "step": 35682 + }, + { + "epoch": 2.9424424837140264, + "grad_norm": 2.302835464477539, + "learning_rate": 1e-06, + "loss": 0.7799, + "mean_token_accuracy": 0.7521476149559021, + "num_tokens": 962629181.0, + "step": 35683 + }, + { + "epoch": 2.942524944339078, + "grad_norm": 1.869971752166748, + "learning_rate": 1e-06, + "loss": 0.5239, + "mean_token_accuracy": 0.8115097284317017, + "num_tokens": 962662526.0, + "step": 35684 + }, + { + "epoch": 2.9426074049641295, + "grad_norm": 2.089372158050537, + "learning_rate": 1e-06, + "loss": 0.7304, + "mean_token_accuracy": 0.7604897618293762, + "num_tokens": 962688840.0, + "step": 35685 + }, + { + "epoch": 2.9426898655891813, + "grad_norm": 2.4037926197052, + "learning_rate": 1e-06, + "loss": 0.7821, + "mean_token_accuracy": 0.7448854446411133, + "num_tokens": 962714037.0, + "step": 35686 + }, + { + "epoch": 2.9427723262142327, + "grad_norm": 1.9160300493240356, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7812540531158447, + "num_tokens": 962749337.0, + "step": 35687 + }, + { + "epoch": 2.942854786839284, + "grad_norm": 1.9063199758529663, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.7644277215003967, + "num_tokens": 962787241.0, + "step": 35688 + }, + { + "epoch": 2.942937247464336, + "grad_norm": 1.9878358840942383, + "learning_rate": 1e-06, + "loss": 0.5763, + "mean_token_accuracy": 0.8055988550186157, + "num_tokens": 962817516.0, + "step": 35689 + }, + { + "epoch": 2.943019708089387, + "grad_norm": 2.020622968673706, + "learning_rate": 1e-06, + "loss": 0.678, + "mean_token_accuracy": 0.7738369703292847, + "num_tokens": 962850513.0, + "step": 35690 + }, + { + "epoch": 2.943102168714439, + "grad_norm": 2.2478535175323486, + "learning_rate": 1e-06, + "loss": 0.7608, + "mean_token_accuracy": 0.7518066763877869, + "num_tokens": 962878115.0, + "step": 35691 + }, + { + "epoch": 2.9431846293394903, + "grad_norm": 2.4501664638519287, + "learning_rate": 1e-06, + "loss": 0.7095, + "mean_token_accuracy": 0.7608913779258728, + "num_tokens": 962900561.0, + "step": 35692 + }, + { + "epoch": 2.943267089964542, + "grad_norm": 2.501171827316284, + "learning_rate": 1e-06, + "loss": 0.7174, + "mean_token_accuracy": 0.766741156578064, + "num_tokens": 962922952.0, + "step": 35693 + }, + { + "epoch": 2.9433495505895935, + "grad_norm": 2.01120924949646, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7677386999130249, + "num_tokens": 962956985.0, + "step": 35694 + }, + { + "epoch": 2.943432011214645, + "grad_norm": 1.8120380640029907, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7753703594207764, + "num_tokens": 962996113.0, + "step": 35695 + }, + { + "epoch": 2.9435144718396966, + "grad_norm": 2.3031561374664307, + "learning_rate": 1e-06, + "loss": 0.7421, + "mean_token_accuracy": 0.7580348253250122, + "num_tokens": 963021839.0, + "step": 35696 + }, + { + "epoch": 2.943596932464748, + "grad_norm": 2.057706832885742, + "learning_rate": 1e-06, + "loss": 0.6629, + "mean_token_accuracy": 0.785824179649353, + "num_tokens": 963053262.0, + "step": 35697 + }, + { + "epoch": 2.9436793930897998, + "grad_norm": 2.016615152359009, + "learning_rate": 1e-06, + "loss": 0.5654, + "mean_token_accuracy": 0.8077864646911621, + "num_tokens": 963085330.0, + "step": 35698 + }, + { + "epoch": 2.943761853714851, + "grad_norm": 1.863173484802246, + "learning_rate": 1e-06, + "loss": 0.5692, + "mean_token_accuracy": 0.8003073334693909, + "num_tokens": 963116568.0, + "step": 35699 + }, + { + "epoch": 2.943844314339903, + "grad_norm": 2.2151458263397217, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7693479061126709, + "num_tokens": 963144550.0, + "step": 35700 + }, + { + "epoch": 2.9439267749649543, + "grad_norm": 2.180689811706543, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7244665026664734, + "num_tokens": 963171034.0, + "step": 35701 + }, + { + "epoch": 2.9440092355900056, + "grad_norm": 2.2157187461853027, + "learning_rate": 1e-06, + "loss": 0.6288, + "mean_token_accuracy": 0.7886478304862976, + "num_tokens": 963198638.0, + "step": 35702 + }, + { + "epoch": 2.9440916962150574, + "grad_norm": 2.134587526321411, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7435294985771179, + "num_tokens": 963227652.0, + "step": 35703 + }, + { + "epoch": 2.9441741568401087, + "grad_norm": 2.045201063156128, + "learning_rate": 1e-06, + "loss": 0.546, + "mean_token_accuracy": 0.8000421524047852, + "num_tokens": 963255696.0, + "step": 35704 + }, + { + "epoch": 2.9442566174651605, + "grad_norm": 2.3895976543426514, + "learning_rate": 1e-06, + "loss": 0.7494, + "mean_token_accuracy": 0.7529708743095398, + "num_tokens": 963282233.0, + "step": 35705 + }, + { + "epoch": 2.944339078090212, + "grad_norm": 2.306133985519409, + "learning_rate": 1e-06, + "loss": 0.7535, + "mean_token_accuracy": 0.7621487379074097, + "num_tokens": 963307793.0, + "step": 35706 + }, + { + "epoch": 2.9444215387152637, + "grad_norm": 2.3276007175445557, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7420388460159302, + "num_tokens": 963334758.0, + "step": 35707 + }, + { + "epoch": 2.944503999340315, + "grad_norm": 2.182769536972046, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7886667847633362, + "num_tokens": 963361979.0, + "step": 35708 + }, + { + "epoch": 2.9445864599653664, + "grad_norm": 2.0173392295837402, + "learning_rate": 1e-06, + "loss": 0.6548, + "mean_token_accuracy": 0.7832666635513306, + "num_tokens": 963394150.0, + "step": 35709 + }, + { + "epoch": 2.944668920590418, + "grad_norm": 2.0398149490356445, + "learning_rate": 1e-06, + "loss": 0.5262, + "mean_token_accuracy": 0.8207709193229675, + "num_tokens": 963426121.0, + "step": 35710 + }, + { + "epoch": 2.9447513812154695, + "grad_norm": 2.124985456466675, + "learning_rate": 1e-06, + "loss": 0.7692, + "mean_token_accuracy": 0.7456955909729004, + "num_tokens": 963460198.0, + "step": 35711 + }, + { + "epoch": 2.944833841840521, + "grad_norm": 2.02909517288208, + "learning_rate": 1e-06, + "loss": 0.6854, + "mean_token_accuracy": 0.7778327465057373, + "num_tokens": 963490935.0, + "step": 35712 + }, + { + "epoch": 2.9449163024655727, + "grad_norm": 1.8969688415527344, + "learning_rate": 1e-06, + "loss": 0.7114, + "mean_token_accuracy": 0.7626578211784363, + "num_tokens": 963523869.0, + "step": 35713 + }, + { + "epoch": 2.9449987630906245, + "grad_norm": 1.9059133529663086, + "learning_rate": 1e-06, + "loss": 0.6686, + "mean_token_accuracy": 0.7804079055786133, + "num_tokens": 963558408.0, + "step": 35714 + }, + { + "epoch": 2.945081223715676, + "grad_norm": 1.9524328708648682, + "learning_rate": 1e-06, + "loss": 0.7187, + "mean_token_accuracy": 0.7625678777694702, + "num_tokens": 963590287.0, + "step": 35715 + }, + { + "epoch": 2.945163684340727, + "grad_norm": 2.3862757682800293, + "learning_rate": 1e-06, + "loss": 0.6116, + "mean_token_accuracy": 0.7980377674102783, + "num_tokens": 963614708.0, + "step": 35716 + }, + { + "epoch": 2.945246144965779, + "grad_norm": 2.206610679626465, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7716996669769287, + "num_tokens": 963643623.0, + "step": 35717 + }, + { + "epoch": 2.9453286055908303, + "grad_norm": 2.030625104904175, + "learning_rate": 1e-06, + "loss": 0.7084, + "mean_token_accuracy": 0.7761945724487305, + "num_tokens": 963675349.0, + "step": 35718 + }, + { + "epoch": 2.9454110662158817, + "grad_norm": 2.1748855113983154, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7785366773605347, + "num_tokens": 963704821.0, + "step": 35719 + }, + { + "epoch": 2.9454935268409335, + "grad_norm": 2.2039523124694824, + "learning_rate": 1e-06, + "loss": 0.6893, + "mean_token_accuracy": 0.7728841304779053, + "num_tokens": 963733660.0, + "step": 35720 + }, + { + "epoch": 2.9455759874659853, + "grad_norm": 2.170259952545166, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7731547355651855, + "num_tokens": 963765168.0, + "step": 35721 + }, + { + "epoch": 2.9456584480910366, + "grad_norm": 2.280236005783081, + "learning_rate": 1e-06, + "loss": 0.727, + "mean_token_accuracy": 0.7568998336791992, + "num_tokens": 963791577.0, + "step": 35722 + }, + { + "epoch": 2.945740908716088, + "grad_norm": 2.8193559646606445, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7482158541679382, + "num_tokens": 963811745.0, + "step": 35723 + }, + { + "epoch": 2.9458233693411398, + "grad_norm": 2.3306996822357178, + "learning_rate": 1e-06, + "loss": 0.6943, + "mean_token_accuracy": 0.7760619521141052, + "num_tokens": 963837697.0, + "step": 35724 + }, + { + "epoch": 2.945905829966191, + "grad_norm": 2.0972206592559814, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.739689290523529, + "num_tokens": 963869354.0, + "step": 35725 + }, + { + "epoch": 2.9459882905912425, + "grad_norm": 2.2332985401153564, + "learning_rate": 1e-06, + "loss": 0.6559, + "mean_token_accuracy": 0.7862166166305542, + "num_tokens": 963894953.0, + "step": 35726 + }, + { + "epoch": 2.9460707512162942, + "grad_norm": 1.896192193031311, + "learning_rate": 1e-06, + "loss": 0.6628, + "mean_token_accuracy": 0.7962942123413086, + "num_tokens": 963929722.0, + "step": 35727 + }, + { + "epoch": 2.9461532118413456, + "grad_norm": 2.1292169094085693, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7693448662757874, + "num_tokens": 963958379.0, + "step": 35728 + }, + { + "epoch": 2.9462356724663974, + "grad_norm": 2.247922897338867, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.785672128200531, + "num_tokens": 963985863.0, + "step": 35729 + }, + { + "epoch": 2.9463181330914487, + "grad_norm": 2.463620662689209, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.7836646437644958, + "num_tokens": 964009234.0, + "step": 35730 + }, + { + "epoch": 2.9464005937165005, + "grad_norm": 1.9884793758392334, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7448556423187256, + "num_tokens": 964045289.0, + "step": 35731 + }, + { + "epoch": 2.946483054341552, + "grad_norm": 2.02128529548645, + "learning_rate": 1e-06, + "loss": 0.6883, + "mean_token_accuracy": 0.7788659930229187, + "num_tokens": 964078142.0, + "step": 35732 + }, + { + "epoch": 2.9465655149666032, + "grad_norm": 2.0989532470703125, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.7683703899383545, + "num_tokens": 964108704.0, + "step": 35733 + }, + { + "epoch": 2.946647975591655, + "grad_norm": 1.9499551057815552, + "learning_rate": 1e-06, + "loss": 0.6063, + "mean_token_accuracy": 0.8038467764854431, + "num_tokens": 964142019.0, + "step": 35734 + }, + { + "epoch": 2.9467304362167064, + "grad_norm": 2.551478862762451, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7317203283309937, + "num_tokens": 964164010.0, + "step": 35735 + }, + { + "epoch": 2.946812896841758, + "grad_norm": 2.173238515853882, + "learning_rate": 1e-06, + "loss": 0.4919, + "mean_token_accuracy": 0.8336439728736877, + "num_tokens": 964197371.0, + "step": 35736 + }, + { + "epoch": 2.9468953574668095, + "grad_norm": 2.0224077701568604, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7550604343414307, + "num_tokens": 964232234.0, + "step": 35737 + }, + { + "epoch": 2.9469778180918613, + "grad_norm": 2.3258981704711914, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7688614130020142, + "num_tokens": 964255464.0, + "step": 35738 + }, + { + "epoch": 2.9470602787169127, + "grad_norm": 2.049851417541504, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7482597827911377, + "num_tokens": 964286819.0, + "step": 35739 + }, + { + "epoch": 2.947142739341964, + "grad_norm": 2.5614895820617676, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7568610906600952, + "num_tokens": 964309805.0, + "step": 35740 + }, + { + "epoch": 2.947225199967016, + "grad_norm": 2.0561940670013428, + "learning_rate": 1e-06, + "loss": 0.6663, + "mean_token_accuracy": 0.7853068113327026, + "num_tokens": 964339626.0, + "step": 35741 + }, + { + "epoch": 2.947307660592067, + "grad_norm": 2.3802237510681152, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7509391903877258, + "num_tokens": 964362598.0, + "step": 35742 + }, + { + "epoch": 2.947390121217119, + "grad_norm": 2.168823719024658, + "learning_rate": 1e-06, + "loss": 0.6852, + "mean_token_accuracy": 0.7843353748321533, + "num_tokens": 964391264.0, + "step": 35743 + }, + { + "epoch": 2.9474725818421703, + "grad_norm": 2.2047767639160156, + "learning_rate": 1e-06, + "loss": 0.6054, + "mean_token_accuracy": 0.7859193086624146, + "num_tokens": 964420815.0, + "step": 35744 + }, + { + "epoch": 2.947555042467222, + "grad_norm": 2.198974370956421, + "learning_rate": 1e-06, + "loss": 0.6637, + "mean_token_accuracy": 0.7721124291419983, + "num_tokens": 964449392.0, + "step": 35745 + }, + { + "epoch": 2.9476375030922735, + "grad_norm": 2.23439884185791, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.778590202331543, + "num_tokens": 964472861.0, + "step": 35746 + }, + { + "epoch": 2.947719963717325, + "grad_norm": 2.3106393814086914, + "learning_rate": 1e-06, + "loss": 0.7785, + "mean_token_accuracy": 0.7525955438613892, + "num_tokens": 964500013.0, + "step": 35747 + }, + { + "epoch": 2.9478024243423766, + "grad_norm": 2.0542855262756348, + "learning_rate": 1e-06, + "loss": 0.6884, + "mean_token_accuracy": 0.7709981799125671, + "num_tokens": 964531027.0, + "step": 35748 + }, + { + "epoch": 2.947884884967428, + "grad_norm": 2.1137921810150146, + "learning_rate": 1e-06, + "loss": 0.7762, + "mean_token_accuracy": 0.7513260841369629, + "num_tokens": 964561301.0, + "step": 35749 + }, + { + "epoch": 2.9479673455924797, + "grad_norm": 2.1218361854553223, + "learning_rate": 1e-06, + "loss": 0.674, + "mean_token_accuracy": 0.7735432982444763, + "num_tokens": 964590197.0, + "step": 35750 + }, + { + "epoch": 2.948049806217531, + "grad_norm": 2.0916342735290527, + "learning_rate": 1e-06, + "loss": 0.7928, + "mean_token_accuracy": 0.7471991777420044, + "num_tokens": 964621380.0, + "step": 35751 + }, + { + "epoch": 2.948132266842583, + "grad_norm": 1.9385136365890503, + "learning_rate": 1e-06, + "loss": 0.613, + "mean_token_accuracy": 0.8019385933876038, + "num_tokens": 964653950.0, + "step": 35752 + }, + { + "epoch": 2.9482147274676342, + "grad_norm": 1.9384676218032837, + "learning_rate": 1e-06, + "loss": 0.6978, + "mean_token_accuracy": 0.7796342372894287, + "num_tokens": 964687094.0, + "step": 35753 + }, + { + "epoch": 2.9482971880926856, + "grad_norm": 1.9257596731185913, + "learning_rate": 1e-06, + "loss": 0.5823, + "mean_token_accuracy": 0.8023176193237305, + "num_tokens": 964718991.0, + "step": 35754 + }, + { + "epoch": 2.9483796487177374, + "grad_norm": 2.0336670875549316, + "learning_rate": 1e-06, + "loss": 0.6054, + "mean_token_accuracy": 0.8008031845092773, + "num_tokens": 964750337.0, + "step": 35755 + }, + { + "epoch": 2.9484621093427887, + "grad_norm": 2.2496728897094727, + "learning_rate": 1e-06, + "loss": 0.7088, + "mean_token_accuracy": 0.7711572051048279, + "num_tokens": 964777088.0, + "step": 35756 + }, + { + "epoch": 2.94854456996784, + "grad_norm": 2.079216241836548, + "learning_rate": 1e-06, + "loss": 0.5672, + "mean_token_accuracy": 0.8015490770339966, + "num_tokens": 964805806.0, + "step": 35757 + }, + { + "epoch": 2.948627030592892, + "grad_norm": 1.902984380722046, + "learning_rate": 1e-06, + "loss": 0.5889, + "mean_token_accuracy": 0.8018561601638794, + "num_tokens": 964840351.0, + "step": 35758 + }, + { + "epoch": 2.9487094912179437, + "grad_norm": 2.0054237842559814, + "learning_rate": 1e-06, + "loss": 0.7427, + "mean_token_accuracy": 0.759738564491272, + "num_tokens": 964875751.0, + "step": 35759 + }, + { + "epoch": 2.948791951842995, + "grad_norm": 1.9415448904037476, + "learning_rate": 1e-06, + "loss": 0.571, + "mean_token_accuracy": 0.8085839748382568, + "num_tokens": 964907325.0, + "step": 35760 + }, + { + "epoch": 2.9488744124680464, + "grad_norm": 1.8587286472320557, + "learning_rate": 1e-06, + "loss": 0.7062, + "mean_token_accuracy": 0.7769623398780823, + "num_tokens": 964945046.0, + "step": 35761 + }, + { + "epoch": 2.948956873093098, + "grad_norm": 2.3395321369171143, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7279801368713379, + "num_tokens": 964973116.0, + "step": 35762 + }, + { + "epoch": 2.9490393337181495, + "grad_norm": 1.9263513088226318, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7390470504760742, + "num_tokens": 965008438.0, + "step": 35763 + }, + { + "epoch": 2.949121794343201, + "grad_norm": 2.1646416187286377, + "learning_rate": 1e-06, + "loss": 0.6365, + "mean_token_accuracy": 0.7756298780441284, + "num_tokens": 965038372.0, + "step": 35764 + }, + { + "epoch": 2.9492042549682527, + "grad_norm": 2.37666392326355, + "learning_rate": 1e-06, + "loss": 0.7549, + "mean_token_accuracy": 0.7614758014678955, + "num_tokens": 965062529.0, + "step": 35765 + }, + { + "epoch": 2.9492867155933045, + "grad_norm": 1.9813990592956543, + "learning_rate": 1e-06, + "loss": 0.6697, + "mean_token_accuracy": 0.7714253664016724, + "num_tokens": 965096595.0, + "step": 35766 + }, + { + "epoch": 2.949369176218356, + "grad_norm": 2.082805633544922, + "learning_rate": 1e-06, + "loss": 0.6771, + "mean_token_accuracy": 0.7784119844436646, + "num_tokens": 965126177.0, + "step": 35767 + }, + { + "epoch": 2.949451636843407, + "grad_norm": 2.0464987754821777, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.7593554258346558, + "num_tokens": 965157423.0, + "step": 35768 + }, + { + "epoch": 2.949534097468459, + "grad_norm": 2.3720901012420654, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7422984838485718, + "num_tokens": 965182083.0, + "step": 35769 + }, + { + "epoch": 2.9496165580935103, + "grad_norm": 1.989180326461792, + "learning_rate": 1e-06, + "loss": 0.6312, + "mean_token_accuracy": 0.7906208038330078, + "num_tokens": 965213637.0, + "step": 35770 + }, + { + "epoch": 2.9496990187185617, + "grad_norm": 1.9870532751083374, + "learning_rate": 1e-06, + "loss": 0.7788, + "mean_token_accuracy": 0.7626028060913086, + "num_tokens": 965247543.0, + "step": 35771 + }, + { + "epoch": 2.9497814793436135, + "grad_norm": 1.9092109203338623, + "learning_rate": 1e-06, + "loss": 0.6383, + "mean_token_accuracy": 0.7942635416984558, + "num_tokens": 965283368.0, + "step": 35772 + }, + { + "epoch": 2.949863939968665, + "grad_norm": 2.335899591445923, + "learning_rate": 1e-06, + "loss": 0.6203, + "mean_token_accuracy": 0.7897505760192871, + "num_tokens": 965310474.0, + "step": 35773 + }, + { + "epoch": 2.9499464005937166, + "grad_norm": 1.9571961164474487, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.8012737035751343, + "num_tokens": 965345617.0, + "step": 35774 + }, + { + "epoch": 2.950028861218768, + "grad_norm": 2.2515482902526855, + "learning_rate": 1e-06, + "loss": 0.8014, + "mean_token_accuracy": 0.7491722106933594, + "num_tokens": 965372057.0, + "step": 35775 + }, + { + "epoch": 2.9501113218438197, + "grad_norm": 2.1075291633605957, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7331197261810303, + "num_tokens": 965401342.0, + "step": 35776 + }, + { + "epoch": 2.950193782468871, + "grad_norm": 2.0924501419067383, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7359760403633118, + "num_tokens": 965432425.0, + "step": 35777 + }, + { + "epoch": 2.9502762430939224, + "grad_norm": 2.109225273132324, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7493535280227661, + "num_tokens": 965462546.0, + "step": 35778 + }, + { + "epoch": 2.9503587037189742, + "grad_norm": 1.9869730472564697, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7786200046539307, + "num_tokens": 965495497.0, + "step": 35779 + }, + { + "epoch": 2.9504411643440256, + "grad_norm": 2.234846353530884, + "learning_rate": 1e-06, + "loss": 0.7405, + "mean_token_accuracy": 0.7548142671585083, + "num_tokens": 965524395.0, + "step": 35780 + }, + { + "epoch": 2.9505236249690774, + "grad_norm": 1.9213743209838867, + "learning_rate": 1e-06, + "loss": 0.6331, + "mean_token_accuracy": 0.7909598350524902, + "num_tokens": 965558691.0, + "step": 35781 + }, + { + "epoch": 2.9506060855941287, + "grad_norm": 2.1296818256378174, + "learning_rate": 1e-06, + "loss": 0.7068, + "mean_token_accuracy": 0.7698337435722351, + "num_tokens": 965590505.0, + "step": 35782 + }, + { + "epoch": 2.9506885462191805, + "grad_norm": 2.0527706146240234, + "learning_rate": 1e-06, + "loss": 0.7299, + "mean_token_accuracy": 0.7557555437088013, + "num_tokens": 965623822.0, + "step": 35783 + }, + { + "epoch": 2.950771006844232, + "grad_norm": 1.9775819778442383, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7613562941551208, + "num_tokens": 965658850.0, + "step": 35784 + }, + { + "epoch": 2.9508534674692832, + "grad_norm": 2.152672290802002, + "learning_rate": 1e-06, + "loss": 0.6622, + "mean_token_accuracy": 0.7857906222343445, + "num_tokens": 965685946.0, + "step": 35785 + }, + { + "epoch": 2.950935928094335, + "grad_norm": 1.7632031440734863, + "learning_rate": 1e-06, + "loss": 0.6316, + "mean_token_accuracy": 0.7942006587982178, + "num_tokens": 965722883.0, + "step": 35786 + }, + { + "epoch": 2.9510183887193864, + "grad_norm": 1.964140772819519, + "learning_rate": 1e-06, + "loss": 0.6882, + "mean_token_accuracy": 0.7796398997306824, + "num_tokens": 965754268.0, + "step": 35787 + }, + { + "epoch": 2.951100849344438, + "grad_norm": 2.0215437412261963, + "learning_rate": 1e-06, + "loss": 0.6621, + "mean_token_accuracy": 0.7773693203926086, + "num_tokens": 965788479.0, + "step": 35788 + }, + { + "epoch": 2.9511833099694895, + "grad_norm": 2.3498916625976562, + "learning_rate": 1e-06, + "loss": 0.7067, + "mean_token_accuracy": 0.7593483328819275, + "num_tokens": 965813755.0, + "step": 35789 + }, + { + "epoch": 2.9512657705945413, + "grad_norm": 1.9218838214874268, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.7869449853897095, + "num_tokens": 965847548.0, + "step": 35790 + }, + { + "epoch": 2.9513482312195927, + "grad_norm": 2.1936748027801514, + "learning_rate": 1e-06, + "loss": 0.7261, + "mean_token_accuracy": 0.7611784338951111, + "num_tokens": 965876346.0, + "step": 35791 + }, + { + "epoch": 2.951430691844644, + "grad_norm": 2.554776430130005, + "learning_rate": 1e-06, + "loss": 0.6725, + "mean_token_accuracy": 0.7786917686462402, + "num_tokens": 965897531.0, + "step": 35792 + }, + { + "epoch": 2.951513152469696, + "grad_norm": 2.5034372806549072, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7217273712158203, + "num_tokens": 965924368.0, + "step": 35793 + }, + { + "epoch": 2.951595613094747, + "grad_norm": 1.8726600408554077, + "learning_rate": 1e-06, + "loss": 0.6147, + "mean_token_accuracy": 0.7995374798774719, + "num_tokens": 965961596.0, + "step": 35794 + }, + { + "epoch": 2.951678073719799, + "grad_norm": 2.0202157497406006, + "learning_rate": 1e-06, + "loss": 0.6494, + "mean_token_accuracy": 0.7867869138717651, + "num_tokens": 965991597.0, + "step": 35795 + }, + { + "epoch": 2.9517605343448503, + "grad_norm": 2.2365362644195557, + "learning_rate": 1e-06, + "loss": 0.5895, + "mean_token_accuracy": 0.7968759536743164, + "num_tokens": 966016115.0, + "step": 35796 + }, + { + "epoch": 2.951842994969902, + "grad_norm": 2.051487922668457, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7798348069190979, + "num_tokens": 966051882.0, + "step": 35797 + }, + { + "epoch": 2.9519254555949535, + "grad_norm": 2.4919564723968506, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7629141211509705, + "num_tokens": 966075396.0, + "step": 35798 + }, + { + "epoch": 2.952007916220005, + "grad_norm": 2.0241808891296387, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7698599696159363, + "num_tokens": 966105105.0, + "step": 35799 + }, + { + "epoch": 2.9520903768450566, + "grad_norm": 2.0294315814971924, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7456552982330322, + "num_tokens": 966137567.0, + "step": 35800 + }, + { + "epoch": 2.952172837470108, + "grad_norm": 2.24835205078125, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7364674806594849, + "num_tokens": 966168187.0, + "step": 35801 + }, + { + "epoch": 2.9522552980951593, + "grad_norm": 2.144313097000122, + "learning_rate": 1e-06, + "loss": 0.6836, + "mean_token_accuracy": 0.7733964920043945, + "num_tokens": 966197516.0, + "step": 35802 + }, + { + "epoch": 2.952337758720211, + "grad_norm": 2.104463815689087, + "learning_rate": 1e-06, + "loss": 0.7183, + "mean_token_accuracy": 0.762961208820343, + "num_tokens": 966226408.0, + "step": 35803 + }, + { + "epoch": 2.952420219345263, + "grad_norm": 2.119508981704712, + "learning_rate": 1e-06, + "loss": 0.7362, + "mean_token_accuracy": 0.7643739581108093, + "num_tokens": 966253775.0, + "step": 35804 + }, + { + "epoch": 2.9525026799703142, + "grad_norm": 1.9382765293121338, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7878785729408264, + "num_tokens": 966284527.0, + "step": 35805 + }, + { + "epoch": 2.9525851405953656, + "grad_norm": 2.2664854526519775, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7310550212860107, + "num_tokens": 966309749.0, + "step": 35806 + }, + { + "epoch": 2.9526676012204174, + "grad_norm": 2.1432690620422363, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7259874939918518, + "num_tokens": 966338087.0, + "step": 35807 + }, + { + "epoch": 2.9527500618454687, + "grad_norm": 2.2056665420532227, + "learning_rate": 1e-06, + "loss": 0.5279, + "mean_token_accuracy": 0.8213306665420532, + "num_tokens": 966364220.0, + "step": 35808 + }, + { + "epoch": 2.95283252247052, + "grad_norm": 2.009398937225342, + "learning_rate": 1e-06, + "loss": 0.8158, + "mean_token_accuracy": 0.7384480237960815, + "num_tokens": 966399413.0, + "step": 35809 + }, + { + "epoch": 2.952914983095572, + "grad_norm": 1.8712793588638306, + "learning_rate": 1e-06, + "loss": 0.6641, + "mean_token_accuracy": 0.7817784547805786, + "num_tokens": 966432327.0, + "step": 35810 + }, + { + "epoch": 2.9529974437206237, + "grad_norm": 2.0748374462127686, + "learning_rate": 1e-06, + "loss": 0.5986, + "mean_token_accuracy": 0.8035535216331482, + "num_tokens": 966461668.0, + "step": 35811 + }, + { + "epoch": 2.953079904345675, + "grad_norm": 2.024907112121582, + "learning_rate": 1e-06, + "loss": 0.6687, + "mean_token_accuracy": 0.7952322959899902, + "num_tokens": 966492296.0, + "step": 35812 + }, + { + "epoch": 2.9531623649707264, + "grad_norm": 2.408980369567871, + "learning_rate": 1e-06, + "loss": 0.634, + "mean_token_accuracy": 0.7872892022132874, + "num_tokens": 966517380.0, + "step": 35813 + }, + { + "epoch": 2.953244825595778, + "grad_norm": 2.125221014022827, + "learning_rate": 1e-06, + "loss": 0.7139, + "mean_token_accuracy": 0.767670750617981, + "num_tokens": 966545940.0, + "step": 35814 + }, + { + "epoch": 2.9533272862208295, + "grad_norm": 2.144430637359619, + "learning_rate": 1e-06, + "loss": 0.6171, + "mean_token_accuracy": 0.7862175703048706, + "num_tokens": 966574531.0, + "step": 35815 + }, + { + "epoch": 2.953409746845881, + "grad_norm": 2.050279378890991, + "learning_rate": 1e-06, + "loss": 0.5752, + "mean_token_accuracy": 0.802130401134491, + "num_tokens": 966612146.0, + "step": 35816 + }, + { + "epoch": 2.9534922074709327, + "grad_norm": 2.544625997543335, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7482771873474121, + "num_tokens": 966634423.0, + "step": 35817 + }, + { + "epoch": 2.953574668095984, + "grad_norm": 1.9421743154525757, + "learning_rate": 1e-06, + "loss": 0.6034, + "mean_token_accuracy": 0.7959340214729309, + "num_tokens": 966666532.0, + "step": 35818 + }, + { + "epoch": 2.953657128721036, + "grad_norm": 2.2258288860321045, + "learning_rate": 1e-06, + "loss": 0.6718, + "mean_token_accuracy": 0.7647274732589722, + "num_tokens": 966692212.0, + "step": 35819 + }, + { + "epoch": 2.953739589346087, + "grad_norm": 2.021658182144165, + "learning_rate": 1e-06, + "loss": 0.6002, + "mean_token_accuracy": 0.7947037220001221, + "num_tokens": 966725961.0, + "step": 35820 + }, + { + "epoch": 2.953822049971139, + "grad_norm": 2.042290687561035, + "learning_rate": 1e-06, + "loss": 0.754, + "mean_token_accuracy": 0.7614883184432983, + "num_tokens": 966758841.0, + "step": 35821 + }, + { + "epoch": 2.9539045105961903, + "grad_norm": 2.022158622741699, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.8037266731262207, + "num_tokens": 966792438.0, + "step": 35822 + }, + { + "epoch": 2.9539869712212417, + "grad_norm": 1.9277386665344238, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7600787281990051, + "num_tokens": 966827171.0, + "step": 35823 + }, + { + "epoch": 2.9540694318462934, + "grad_norm": 2.356315851211548, + "learning_rate": 1e-06, + "loss": 0.718, + "mean_token_accuracy": 0.765240490436554, + "num_tokens": 966850223.0, + "step": 35824 + }, + { + "epoch": 2.954151892471345, + "grad_norm": 2.2624990940093994, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7444767355918884, + "num_tokens": 966875222.0, + "step": 35825 + }, + { + "epoch": 2.9542343530963966, + "grad_norm": 1.9241554737091064, + "learning_rate": 1e-06, + "loss": 0.662, + "mean_token_accuracy": 0.7838646173477173, + "num_tokens": 966908708.0, + "step": 35826 + }, + { + "epoch": 2.954316813721448, + "grad_norm": 2.0623652935028076, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7383068799972534, + "num_tokens": 966942864.0, + "step": 35827 + }, + { + "epoch": 2.9543992743464997, + "grad_norm": 1.919788122177124, + "learning_rate": 1e-06, + "loss": 0.7037, + "mean_token_accuracy": 0.7792408466339111, + "num_tokens": 966977932.0, + "step": 35828 + }, + { + "epoch": 2.954481734971551, + "grad_norm": 2.0323667526245117, + "learning_rate": 1e-06, + "loss": 0.7078, + "mean_token_accuracy": 0.7692471146583557, + "num_tokens": 967008811.0, + "step": 35829 + }, + { + "epoch": 2.9545641955966024, + "grad_norm": 1.9292924404144287, + "learning_rate": 1e-06, + "loss": 0.6746, + "mean_token_accuracy": 0.7800145149230957, + "num_tokens": 967041150.0, + "step": 35830 + }, + { + "epoch": 2.9546466562216542, + "grad_norm": 2.4325149059295654, + "learning_rate": 1e-06, + "loss": 0.7173, + "mean_token_accuracy": 0.7647421360015869, + "num_tokens": 967066328.0, + "step": 35831 + }, + { + "epoch": 2.9547291168467056, + "grad_norm": 2.1593570709228516, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.746481716632843, + "num_tokens": 967094872.0, + "step": 35832 + }, + { + "epoch": 2.9548115774717574, + "grad_norm": 2.0298821926116943, + "learning_rate": 1e-06, + "loss": 0.638, + "mean_token_accuracy": 0.7834662795066833, + "num_tokens": 967127623.0, + "step": 35833 + }, + { + "epoch": 2.9548940380968087, + "grad_norm": 1.9269198179244995, + "learning_rate": 1e-06, + "loss": 0.5489, + "mean_token_accuracy": 0.8127475380897522, + "num_tokens": 967161890.0, + "step": 35834 + }, + { + "epoch": 2.9549764987218605, + "grad_norm": 2.1193182468414307, + "learning_rate": 1e-06, + "loss": 0.6478, + "mean_token_accuracy": 0.7860864996910095, + "num_tokens": 967193301.0, + "step": 35835 + }, + { + "epoch": 2.955058959346912, + "grad_norm": 2.224701166152954, + "learning_rate": 1e-06, + "loss": 0.6934, + "mean_token_accuracy": 0.771139919757843, + "num_tokens": 967220056.0, + "step": 35836 + }, + { + "epoch": 2.9551414199719632, + "grad_norm": 2.0240495204925537, + "learning_rate": 1e-06, + "loss": 0.6931, + "mean_token_accuracy": 0.7795174717903137, + "num_tokens": 967253605.0, + "step": 35837 + }, + { + "epoch": 2.955223880597015, + "grad_norm": 2.2367734909057617, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.770377516746521, + "num_tokens": 967282625.0, + "step": 35838 + }, + { + "epoch": 2.9553063412220664, + "grad_norm": 2.022648811340332, + "learning_rate": 1e-06, + "loss": 0.7501, + "mean_token_accuracy": 0.7588323354721069, + "num_tokens": 967313366.0, + "step": 35839 + }, + { + "epoch": 2.955388801847118, + "grad_norm": 2.2204761505126953, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7530845999717712, + "num_tokens": 967342044.0, + "step": 35840 + }, + { + "epoch": 2.9554712624721695, + "grad_norm": 2.7689473628997803, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7627506256103516, + "num_tokens": 967364119.0, + "step": 35841 + }, + { + "epoch": 2.9555537230972213, + "grad_norm": 1.9734119176864624, + "learning_rate": 1e-06, + "loss": 0.7053, + "mean_token_accuracy": 0.7792375087738037, + "num_tokens": 967399126.0, + "step": 35842 + }, + { + "epoch": 2.9556361837222727, + "grad_norm": 2.461256980895996, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7341657876968384, + "num_tokens": 967423219.0, + "step": 35843 + }, + { + "epoch": 2.955718644347324, + "grad_norm": 2.0133702754974365, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7542434930801392, + "num_tokens": 967455094.0, + "step": 35844 + }, + { + "epoch": 2.955801104972376, + "grad_norm": 1.9546422958374023, + "learning_rate": 1e-06, + "loss": 0.5969, + "mean_token_accuracy": 0.8019632697105408, + "num_tokens": 967484918.0, + "step": 35845 + }, + { + "epoch": 2.955883565597427, + "grad_norm": 1.9351142644882202, + "learning_rate": 1e-06, + "loss": 0.521, + "mean_token_accuracy": 0.8279147148132324, + "num_tokens": 967516823.0, + "step": 35846 + }, + { + "epoch": 2.9559660262224785, + "grad_norm": 2.3953354358673096, + "learning_rate": 1e-06, + "loss": 0.6827, + "mean_token_accuracy": 0.7773910760879517, + "num_tokens": 967541069.0, + "step": 35847 + }, + { + "epoch": 2.9560484868475303, + "grad_norm": 2.4196712970733643, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7609649896621704, + "num_tokens": 967564975.0, + "step": 35848 + }, + { + "epoch": 2.956130947472582, + "grad_norm": 1.9890562295913696, + "learning_rate": 1e-06, + "loss": 0.7842, + "mean_token_accuracy": 0.7522906064987183, + "num_tokens": 967599690.0, + "step": 35849 + }, + { + "epoch": 2.9562134080976334, + "grad_norm": 2.2850141525268555, + "learning_rate": 1e-06, + "loss": 0.6977, + "mean_token_accuracy": 0.7676005363464355, + "num_tokens": 967623932.0, + "step": 35850 + }, + { + "epoch": 2.956295868722685, + "grad_norm": 1.9906580448150635, + "learning_rate": 1e-06, + "loss": 0.5988, + "mean_token_accuracy": 0.7970715165138245, + "num_tokens": 967656726.0, + "step": 35851 + }, + { + "epoch": 2.9563783293477366, + "grad_norm": 2.3771352767944336, + "learning_rate": 1e-06, + "loss": 0.5922, + "mean_token_accuracy": 0.8033548593521118, + "num_tokens": 967678841.0, + "step": 35852 + }, + { + "epoch": 2.956460789972788, + "grad_norm": 2.2745797634124756, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7751503586769104, + "num_tokens": 967707764.0, + "step": 35853 + }, + { + "epoch": 2.9565432505978393, + "grad_norm": 2.0400004386901855, + "learning_rate": 1e-06, + "loss": 0.5308, + "mean_token_accuracy": 0.8182461261749268, + "num_tokens": 967741197.0, + "step": 35854 + }, + { + "epoch": 2.956625711222891, + "grad_norm": 2.0785322189331055, + "learning_rate": 1e-06, + "loss": 0.762, + "mean_token_accuracy": 0.7668219804763794, + "num_tokens": 967773792.0, + "step": 35855 + }, + { + "epoch": 2.956708171847943, + "grad_norm": 2.033282995223999, + "learning_rate": 1e-06, + "loss": 0.6808, + "mean_token_accuracy": 0.7720049619674683, + "num_tokens": 967805122.0, + "step": 35856 + }, + { + "epoch": 2.9567906324729942, + "grad_norm": 2.488776922225952, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7376838326454163, + "num_tokens": 967831934.0, + "step": 35857 + }, + { + "epoch": 2.9568730930980456, + "grad_norm": 2.066509962081909, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.742978572845459, + "num_tokens": 967863488.0, + "step": 35858 + }, + { + "epoch": 2.9569555537230974, + "grad_norm": 2.1804146766662598, + "learning_rate": 1e-06, + "loss": 0.721, + "mean_token_accuracy": 0.7575743198394775, + "num_tokens": 967892263.0, + "step": 35859 + }, + { + "epoch": 2.9570380143481487, + "grad_norm": 2.0050199031829834, + "learning_rate": 1e-06, + "loss": 0.6814, + "mean_token_accuracy": 0.7762611508369446, + "num_tokens": 967924848.0, + "step": 35860 + }, + { + "epoch": 2.9571204749732, + "grad_norm": 1.9329736232757568, + "learning_rate": 1e-06, + "loss": 0.6489, + "mean_token_accuracy": 0.7901919484138489, + "num_tokens": 967959726.0, + "step": 35861 + }, + { + "epoch": 2.957202935598252, + "grad_norm": 2.263375997543335, + "learning_rate": 1e-06, + "loss": 0.7209, + "mean_token_accuracy": 0.754919171333313, + "num_tokens": 967987834.0, + "step": 35862 + }, + { + "epoch": 2.957285396223303, + "grad_norm": 2.108815908432007, + "learning_rate": 1e-06, + "loss": 0.7468, + "mean_token_accuracy": 0.7542343139648438, + "num_tokens": 968015465.0, + "step": 35863 + }, + { + "epoch": 2.957367856848355, + "grad_norm": 2.1791346073150635, + "learning_rate": 1e-06, + "loss": 0.6017, + "mean_token_accuracy": 0.7982780933380127, + "num_tokens": 968042825.0, + "step": 35864 + }, + { + "epoch": 2.9574503174734064, + "grad_norm": 2.0931482315063477, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7791965007781982, + "num_tokens": 968072243.0, + "step": 35865 + }, + { + "epoch": 2.957532778098458, + "grad_norm": 2.267233371734619, + "learning_rate": 1e-06, + "loss": 0.6617, + "mean_token_accuracy": 0.7874186635017395, + "num_tokens": 968098403.0, + "step": 35866 + }, + { + "epoch": 2.9576152387235095, + "grad_norm": 1.7772127389907837, + "learning_rate": 1e-06, + "loss": 0.4797, + "mean_token_accuracy": 0.8326687812805176, + "num_tokens": 968132464.0, + "step": 35867 + }, + { + "epoch": 2.957697699348561, + "grad_norm": 2.1676199436187744, + "learning_rate": 1e-06, + "loss": 0.5186, + "mean_token_accuracy": 0.8253786563873291, + "num_tokens": 968156624.0, + "step": 35868 + }, + { + "epoch": 2.9577801599736127, + "grad_norm": 2.150578498840332, + "learning_rate": 1e-06, + "loss": 0.7103, + "mean_token_accuracy": 0.7802164554595947, + "num_tokens": 968185450.0, + "step": 35869 + }, + { + "epoch": 2.957862620598664, + "grad_norm": 2.037764310836792, + "learning_rate": 1e-06, + "loss": 0.755, + "mean_token_accuracy": 0.7643343210220337, + "num_tokens": 968218958.0, + "step": 35870 + }, + { + "epoch": 2.957945081223716, + "grad_norm": 2.29728102684021, + "learning_rate": 1e-06, + "loss": 0.7399, + "mean_token_accuracy": 0.7607511878013611, + "num_tokens": 968245463.0, + "step": 35871 + }, + { + "epoch": 2.958027541848767, + "grad_norm": 1.7428300380706787, + "learning_rate": 1e-06, + "loss": 0.601, + "mean_token_accuracy": 0.8043836355209351, + "num_tokens": 968284997.0, + "step": 35872 + }, + { + "epoch": 2.958110002473819, + "grad_norm": 2.254631996154785, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.7644109129905701, + "num_tokens": 968310603.0, + "step": 35873 + }, + { + "epoch": 2.9581924630988703, + "grad_norm": 2.14833402633667, + "learning_rate": 1e-06, + "loss": 0.7536, + "mean_token_accuracy": 0.7612534761428833, + "num_tokens": 968337378.0, + "step": 35874 + }, + { + "epoch": 2.9582749237239216, + "grad_norm": 1.9918887615203857, + "learning_rate": 1e-06, + "loss": 0.7409, + "mean_token_accuracy": 0.7582610845565796, + "num_tokens": 968370899.0, + "step": 35875 + }, + { + "epoch": 2.9583573843489734, + "grad_norm": 2.180271863937378, + "learning_rate": 1e-06, + "loss": 0.6923, + "mean_token_accuracy": 0.766236424446106, + "num_tokens": 968398675.0, + "step": 35876 + }, + { + "epoch": 2.958439844974025, + "grad_norm": 2.0023152828216553, + "learning_rate": 1e-06, + "loss": 0.5778, + "mean_token_accuracy": 0.7956591844558716, + "num_tokens": 968430513.0, + "step": 35877 + }, + { + "epoch": 2.9585223055990766, + "grad_norm": 2.174811601638794, + "learning_rate": 1e-06, + "loss": 0.7662, + "mean_token_accuracy": 0.7532569169998169, + "num_tokens": 968459644.0, + "step": 35878 + }, + { + "epoch": 2.958604766224128, + "grad_norm": 2.0672571659088135, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.775061309337616, + "num_tokens": 968491116.0, + "step": 35879 + }, + { + "epoch": 2.9586872268491797, + "grad_norm": 2.047879219055176, + "learning_rate": 1e-06, + "loss": 0.6384, + "mean_token_accuracy": 0.7852206230163574, + "num_tokens": 968523310.0, + "step": 35880 + }, + { + "epoch": 2.958769687474231, + "grad_norm": 2.3010973930358887, + "learning_rate": 1e-06, + "loss": 0.6779, + "mean_token_accuracy": 0.7730386257171631, + "num_tokens": 968547149.0, + "step": 35881 + }, + { + "epoch": 2.9588521480992824, + "grad_norm": 2.1167120933532715, + "learning_rate": 1e-06, + "loss": 0.7005, + "mean_token_accuracy": 0.7747958898544312, + "num_tokens": 968577382.0, + "step": 35882 + }, + { + "epoch": 2.9589346087243342, + "grad_norm": 2.133582830429077, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.7838969230651855, + "num_tokens": 968606377.0, + "step": 35883 + }, + { + "epoch": 2.9590170693493856, + "grad_norm": 2.1874637603759766, + "learning_rate": 1e-06, + "loss": 0.7112, + "mean_token_accuracy": 0.762750506401062, + "num_tokens": 968631845.0, + "step": 35884 + }, + { + "epoch": 2.9590995299744374, + "grad_norm": 1.9876350164413452, + "learning_rate": 1e-06, + "loss": 0.6574, + "mean_token_accuracy": 0.7778737545013428, + "num_tokens": 968664098.0, + "step": 35885 + }, + { + "epoch": 2.9591819905994887, + "grad_norm": 2.068650007247925, + "learning_rate": 1e-06, + "loss": 0.6638, + "mean_token_accuracy": 0.7802420258522034, + "num_tokens": 968692193.0, + "step": 35886 + }, + { + "epoch": 2.9592644512245405, + "grad_norm": 2.2016348838806152, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7630487680435181, + "num_tokens": 968720084.0, + "step": 35887 + }, + { + "epoch": 2.959346911849592, + "grad_norm": 2.0962414741516113, + "learning_rate": 1e-06, + "loss": 0.7286, + "mean_token_accuracy": 0.7547576427459717, + "num_tokens": 968749010.0, + "step": 35888 + }, + { + "epoch": 2.959429372474643, + "grad_norm": 2.2373125553131104, + "learning_rate": 1e-06, + "loss": 0.586, + "mean_token_accuracy": 0.8031777143478394, + "num_tokens": 968774922.0, + "step": 35889 + }, + { + "epoch": 2.959511833099695, + "grad_norm": 2.1625983715057373, + "learning_rate": 1e-06, + "loss": 0.606, + "mean_token_accuracy": 0.7884102463722229, + "num_tokens": 968803993.0, + "step": 35890 + }, + { + "epoch": 2.9595942937247464, + "grad_norm": 1.828756332397461, + "learning_rate": 1e-06, + "loss": 0.5214, + "mean_token_accuracy": 0.8299661874771118, + "num_tokens": 968838791.0, + "step": 35891 + }, + { + "epoch": 2.9596767543497977, + "grad_norm": 2.315828323364258, + "learning_rate": 1e-06, + "loss": 0.547, + "mean_token_accuracy": 0.8140180110931396, + "num_tokens": 968862936.0, + "step": 35892 + }, + { + "epoch": 2.9597592149748495, + "grad_norm": 2.1225602626800537, + "learning_rate": 1e-06, + "loss": 0.5261, + "mean_token_accuracy": 0.8204487562179565, + "num_tokens": 968889645.0, + "step": 35893 + }, + { + "epoch": 2.9598416755999013, + "grad_norm": 2.2130491733551025, + "learning_rate": 1e-06, + "loss": 0.7279, + "mean_token_accuracy": 0.7633032202720642, + "num_tokens": 968915833.0, + "step": 35894 + }, + { + "epoch": 2.9599241362249527, + "grad_norm": 1.9503995180130005, + "learning_rate": 1e-06, + "loss": 0.7018, + "mean_token_accuracy": 0.7669574618339539, + "num_tokens": 968951904.0, + "step": 35895 + }, + { + "epoch": 2.960006596850004, + "grad_norm": 1.9951859712600708, + "learning_rate": 1e-06, + "loss": 0.5563, + "mean_token_accuracy": 0.8046126365661621, + "num_tokens": 968980492.0, + "step": 35896 + }, + { + "epoch": 2.960089057475056, + "grad_norm": 2.1903209686279297, + "learning_rate": 1e-06, + "loss": 0.7491, + "mean_token_accuracy": 0.7579655647277832, + "num_tokens": 969010268.0, + "step": 35897 + }, + { + "epoch": 2.960171518100107, + "grad_norm": 2.0547008514404297, + "learning_rate": 1e-06, + "loss": 0.6129, + "mean_token_accuracy": 0.79851233959198, + "num_tokens": 969038913.0, + "step": 35898 + }, + { + "epoch": 2.9602539787251585, + "grad_norm": 2.1277341842651367, + "learning_rate": 1e-06, + "loss": 0.5879, + "mean_token_accuracy": 0.8093295097351074, + "num_tokens": 969066998.0, + "step": 35899 + }, + { + "epoch": 2.9603364393502103, + "grad_norm": 1.8236963748931885, + "learning_rate": 1e-06, + "loss": 0.6799, + "mean_token_accuracy": 0.7793217897415161, + "num_tokens": 969103023.0, + "step": 35900 + }, + { + "epoch": 2.960418899975262, + "grad_norm": 2.127274751663208, + "learning_rate": 1e-06, + "loss": 0.7437, + "mean_token_accuracy": 0.761215090751648, + "num_tokens": 969131699.0, + "step": 35901 + }, + { + "epoch": 2.9605013606003134, + "grad_norm": 2.212416410446167, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7607858180999756, + "num_tokens": 969161233.0, + "step": 35902 + }, + { + "epoch": 2.960583821225365, + "grad_norm": 2.316318988800049, + "learning_rate": 1e-06, + "loss": 0.7419, + "mean_token_accuracy": 0.7571362257003784, + "num_tokens": 969184849.0, + "step": 35903 + }, + { + "epoch": 2.9606662818504166, + "grad_norm": 2.128674030303955, + "learning_rate": 1e-06, + "loss": 0.751, + "mean_token_accuracy": 0.7599892616271973, + "num_tokens": 969214022.0, + "step": 35904 + }, + { + "epoch": 2.960748742475468, + "grad_norm": 2.415440559387207, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.737903356552124, + "num_tokens": 969238311.0, + "step": 35905 + }, + { + "epoch": 2.9608312031005193, + "grad_norm": 2.259152889251709, + "learning_rate": 1e-06, + "loss": 0.699, + "mean_token_accuracy": 0.7698801755905151, + "num_tokens": 969267045.0, + "step": 35906 + }, + { + "epoch": 2.960913663725571, + "grad_norm": 2.4726250171661377, + "learning_rate": 1e-06, + "loss": 0.5846, + "mean_token_accuracy": 0.8059526681900024, + "num_tokens": 969288749.0, + "step": 35907 + }, + { + "epoch": 2.9609961243506224, + "grad_norm": 2.2814900875091553, + "learning_rate": 1e-06, + "loss": 0.6787, + "mean_token_accuracy": 0.7747710347175598, + "num_tokens": 969316422.0, + "step": 35908 + }, + { + "epoch": 2.961078584975674, + "grad_norm": 2.1070566177368164, + "learning_rate": 1e-06, + "loss": 0.6721, + "mean_token_accuracy": 0.7777497172355652, + "num_tokens": 969345910.0, + "step": 35909 + }, + { + "epoch": 2.9611610456007256, + "grad_norm": 2.01282000541687, + "learning_rate": 1e-06, + "loss": 0.7296, + "mean_token_accuracy": 0.765954852104187, + "num_tokens": 969378907.0, + "step": 35910 + }, + { + "epoch": 2.9612435062257774, + "grad_norm": 2.2851803302764893, + "learning_rate": 1e-06, + "loss": 0.7643, + "mean_token_accuracy": 0.7557230591773987, + "num_tokens": 969406417.0, + "step": 35911 + }, + { + "epoch": 2.9613259668508287, + "grad_norm": 2.414435863494873, + "learning_rate": 1e-06, + "loss": 0.7502, + "mean_token_accuracy": 0.7539482116699219, + "num_tokens": 969431060.0, + "step": 35912 + }, + { + "epoch": 2.96140842747588, + "grad_norm": 2.1823432445526123, + "learning_rate": 1e-06, + "loss": 0.5569, + "mean_token_accuracy": 0.8063747882843018, + "num_tokens": 969457504.0, + "step": 35913 + }, + { + "epoch": 2.961490888100932, + "grad_norm": 2.1953020095825195, + "learning_rate": 1e-06, + "loss": 0.5954, + "mean_token_accuracy": 0.7928814888000488, + "num_tokens": 969483182.0, + "step": 35914 + }, + { + "epoch": 2.961573348725983, + "grad_norm": 2.0329182147979736, + "learning_rate": 1e-06, + "loss": 0.5316, + "mean_token_accuracy": 0.8254650831222534, + "num_tokens": 969512235.0, + "step": 35915 + }, + { + "epoch": 2.961655809351035, + "grad_norm": 2.1649036407470703, + "learning_rate": 1e-06, + "loss": 0.6924, + "mean_token_accuracy": 0.770002543926239, + "num_tokens": 969540999.0, + "step": 35916 + }, + { + "epoch": 2.9617382699760864, + "grad_norm": 2.1409833431243896, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.7420651912689209, + "num_tokens": 969570647.0, + "step": 35917 + }, + { + "epoch": 2.961820730601138, + "grad_norm": 2.064332962036133, + "learning_rate": 1e-06, + "loss": 0.5259, + "mean_token_accuracy": 0.8114542961120605, + "num_tokens": 969598776.0, + "step": 35918 + }, + { + "epoch": 2.9619031912261895, + "grad_norm": 2.2772912979125977, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.752498209476471, + "num_tokens": 969625316.0, + "step": 35919 + }, + { + "epoch": 2.961985651851241, + "grad_norm": 2.1507046222686768, + "learning_rate": 1e-06, + "loss": 0.6577, + "mean_token_accuracy": 0.7762401700019836, + "num_tokens": 969651594.0, + "step": 35920 + }, + { + "epoch": 2.9620681124762926, + "grad_norm": 2.7475991249084473, + "learning_rate": 1e-06, + "loss": 0.8039, + "mean_token_accuracy": 0.7489824891090393, + "num_tokens": 969672593.0, + "step": 35921 + }, + { + "epoch": 2.962150573101344, + "grad_norm": 2.122633218765259, + "learning_rate": 1e-06, + "loss": 0.5555, + "mean_token_accuracy": 0.8151344060897827, + "num_tokens": 969702161.0, + "step": 35922 + }, + { + "epoch": 2.962233033726396, + "grad_norm": 1.9438194036483765, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7132982611656189, + "num_tokens": 969738251.0, + "step": 35923 + }, + { + "epoch": 2.962315494351447, + "grad_norm": 1.992665410041809, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.7880961298942566, + "num_tokens": 969769187.0, + "step": 35924 + }, + { + "epoch": 2.962397954976499, + "grad_norm": 2.402637004852295, + "learning_rate": 1e-06, + "loss": 0.7904, + "mean_token_accuracy": 0.7424502372741699, + "num_tokens": 969795259.0, + "step": 35925 + }, + { + "epoch": 2.9624804156015503, + "grad_norm": 1.906495213508606, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7548479437828064, + "num_tokens": 969833661.0, + "step": 35926 + }, + { + "epoch": 2.9625628762266016, + "grad_norm": 2.0341360569000244, + "learning_rate": 1e-06, + "loss": 0.5538, + "mean_token_accuracy": 0.8039360046386719, + "num_tokens": 969864327.0, + "step": 35927 + }, + { + "epoch": 2.9626453368516534, + "grad_norm": 2.1201963424682617, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7679768800735474, + "num_tokens": 969892764.0, + "step": 35928 + }, + { + "epoch": 2.962727797476705, + "grad_norm": 1.9682836532592773, + "learning_rate": 1e-06, + "loss": 0.6939, + "mean_token_accuracy": 0.7724225521087646, + "num_tokens": 969927852.0, + "step": 35929 + }, + { + "epoch": 2.9628102581017566, + "grad_norm": 2.6921045780181885, + "learning_rate": 1e-06, + "loss": 0.6328, + "mean_token_accuracy": 0.7993497848510742, + "num_tokens": 969946815.0, + "step": 35930 + }, + { + "epoch": 2.962892718726808, + "grad_norm": 2.242637872695923, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7438452243804932, + "num_tokens": 969974411.0, + "step": 35931 + }, + { + "epoch": 2.9629751793518597, + "grad_norm": 2.2875022888183594, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7421194314956665, + "num_tokens": 970000080.0, + "step": 35932 + }, + { + "epoch": 2.963057639976911, + "grad_norm": 2.020657539367676, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7595295310020447, + "num_tokens": 970031113.0, + "step": 35933 + }, + { + "epoch": 2.9631401006019624, + "grad_norm": 2.2975995540618896, + "learning_rate": 1e-06, + "loss": 0.587, + "mean_token_accuracy": 0.8022621273994446, + "num_tokens": 970055582.0, + "step": 35934 + }, + { + "epoch": 2.963222561227014, + "grad_norm": 2.3585362434387207, + "learning_rate": 1e-06, + "loss": 0.6096, + "mean_token_accuracy": 0.7927871346473694, + "num_tokens": 970081398.0, + "step": 35935 + }, + { + "epoch": 2.9633050218520656, + "grad_norm": 2.1518335342407227, + "learning_rate": 1e-06, + "loss": 0.6461, + "mean_token_accuracy": 0.7854105234146118, + "num_tokens": 970112348.0, + "step": 35936 + }, + { + "epoch": 2.963387482477117, + "grad_norm": 2.0818302631378174, + "learning_rate": 1e-06, + "loss": 0.6782, + "mean_token_accuracy": 0.7730874419212341, + "num_tokens": 970144433.0, + "step": 35937 + }, + { + "epoch": 2.9634699431021687, + "grad_norm": 2.0640156269073486, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7656872272491455, + "num_tokens": 970174531.0, + "step": 35938 + }, + { + "epoch": 2.9635524037272205, + "grad_norm": 2.0682287216186523, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7589319348335266, + "num_tokens": 970205424.0, + "step": 35939 + }, + { + "epoch": 2.963634864352272, + "grad_norm": 1.9432032108306885, + "learning_rate": 1e-06, + "loss": 0.5141, + "mean_token_accuracy": 0.8272089958190918, + "num_tokens": 970235707.0, + "step": 35940 + }, + { + "epoch": 2.963717324977323, + "grad_norm": 2.0904276371002197, + "learning_rate": 1e-06, + "loss": 0.6396, + "mean_token_accuracy": 0.789434015750885, + "num_tokens": 970265181.0, + "step": 35941 + }, + { + "epoch": 2.963799785602375, + "grad_norm": 1.881982684135437, + "learning_rate": 1e-06, + "loss": 0.6414, + "mean_token_accuracy": 0.7907524108886719, + "num_tokens": 970298707.0, + "step": 35942 + }, + { + "epoch": 2.9638822462274264, + "grad_norm": 2.1597838401794434, + "learning_rate": 1e-06, + "loss": 0.6431, + "mean_token_accuracy": 0.7862030267715454, + "num_tokens": 970329873.0, + "step": 35943 + }, + { + "epoch": 2.9639647068524777, + "grad_norm": 1.8958758115768433, + "learning_rate": 1e-06, + "loss": 0.7214, + "mean_token_accuracy": 0.7619135975837708, + "num_tokens": 970368149.0, + "step": 35944 + }, + { + "epoch": 2.9640471674775295, + "grad_norm": 2.0774426460266113, + "learning_rate": 1e-06, + "loss": 0.6216, + "mean_token_accuracy": 0.7922208309173584, + "num_tokens": 970400241.0, + "step": 35945 + }, + { + "epoch": 2.9641296281025813, + "grad_norm": 1.9931820631027222, + "learning_rate": 1e-06, + "loss": 0.6807, + "mean_token_accuracy": 0.7769964933395386, + "num_tokens": 970433676.0, + "step": 35946 + }, + { + "epoch": 2.9642120887276326, + "grad_norm": 2.363751173019409, + "learning_rate": 1e-06, + "loss": 0.5811, + "mean_token_accuracy": 0.814162015914917, + "num_tokens": 970458282.0, + "step": 35947 + }, + { + "epoch": 2.964294549352684, + "grad_norm": 2.0614140033721924, + "learning_rate": 1e-06, + "loss": 0.4912, + "mean_token_accuracy": 0.8183138966560364, + "num_tokens": 970486629.0, + "step": 35948 + }, + { + "epoch": 2.964377009977736, + "grad_norm": 1.9433636665344238, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.7488515377044678, + "num_tokens": 970520217.0, + "step": 35949 + }, + { + "epoch": 2.964459470602787, + "grad_norm": 2.111459732055664, + "learning_rate": 1e-06, + "loss": 0.7641, + "mean_token_accuracy": 0.7572543025016785, + "num_tokens": 970550053.0, + "step": 35950 + }, + { + "epoch": 2.9645419312278385, + "grad_norm": 1.9930012226104736, + "learning_rate": 1e-06, + "loss": 0.6903, + "mean_token_accuracy": 0.7774001359939575, + "num_tokens": 970580876.0, + "step": 35951 + }, + { + "epoch": 2.9646243918528903, + "grad_norm": 1.985393762588501, + "learning_rate": 1e-06, + "loss": 0.6469, + "mean_token_accuracy": 0.7835525274276733, + "num_tokens": 970614306.0, + "step": 35952 + }, + { + "epoch": 2.9647068524779416, + "grad_norm": 2.194819688796997, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7808781862258911, + "num_tokens": 970641789.0, + "step": 35953 + }, + { + "epoch": 2.9647893131029934, + "grad_norm": 2.1548044681549072, + "learning_rate": 1e-06, + "loss": 0.6788, + "mean_token_accuracy": 0.7709544897079468, + "num_tokens": 970673717.0, + "step": 35954 + }, + { + "epoch": 2.964871773728045, + "grad_norm": 2.1086673736572266, + "learning_rate": 1e-06, + "loss": 0.5429, + "mean_token_accuracy": 0.8018051385879517, + "num_tokens": 970702111.0, + "step": 35955 + }, + { + "epoch": 2.9649542343530966, + "grad_norm": 2.141892194747925, + "learning_rate": 1e-06, + "loss": 0.6214, + "mean_token_accuracy": 0.7893311977386475, + "num_tokens": 970728193.0, + "step": 35956 + }, + { + "epoch": 2.965036694978148, + "grad_norm": 2.1366922855377197, + "learning_rate": 1e-06, + "loss": 0.7527, + "mean_token_accuracy": 0.7602015137672424, + "num_tokens": 970757693.0, + "step": 35957 + }, + { + "epoch": 2.9651191556031993, + "grad_norm": 1.8918110132217407, + "learning_rate": 1e-06, + "loss": 0.6151, + "mean_token_accuracy": 0.7924876809120178, + "num_tokens": 970794995.0, + "step": 35958 + }, + { + "epoch": 2.965201616228251, + "grad_norm": 2.11057710647583, + "learning_rate": 1e-06, + "loss": 0.6091, + "mean_token_accuracy": 0.7905228734016418, + "num_tokens": 970822247.0, + "step": 35959 + }, + { + "epoch": 2.9652840768533024, + "grad_norm": 1.8025596141815186, + "learning_rate": 1e-06, + "loss": 0.6318, + "mean_token_accuracy": 0.7932392358779907, + "num_tokens": 970858781.0, + "step": 35960 + }, + { + "epoch": 2.965366537478354, + "grad_norm": 2.100738763809204, + "learning_rate": 1e-06, + "loss": 0.5584, + "mean_token_accuracy": 0.8174198269844055, + "num_tokens": 970889539.0, + "step": 35961 + }, + { + "epoch": 2.9654489981034056, + "grad_norm": 2.0561623573303223, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.7526158094406128, + "num_tokens": 970922380.0, + "step": 35962 + }, + { + "epoch": 2.9655314587284574, + "grad_norm": 2.2191474437713623, + "learning_rate": 1e-06, + "loss": 0.735, + "mean_token_accuracy": 0.7651373147964478, + "num_tokens": 970950510.0, + "step": 35963 + }, + { + "epoch": 2.9656139193535087, + "grad_norm": 2.0046820640563965, + "learning_rate": 1e-06, + "loss": 0.6195, + "mean_token_accuracy": 0.7946573495864868, + "num_tokens": 970981802.0, + "step": 35964 + }, + { + "epoch": 2.96569637997856, + "grad_norm": 1.9456614255905151, + "learning_rate": 1e-06, + "loss": 0.7117, + "mean_token_accuracy": 0.7746723890304565, + "num_tokens": 971015418.0, + "step": 35965 + }, + { + "epoch": 2.965778840603612, + "grad_norm": 2.206407308578491, + "learning_rate": 1e-06, + "loss": 0.5212, + "mean_token_accuracy": 0.8155672550201416, + "num_tokens": 971040898.0, + "step": 35966 + }, + { + "epoch": 2.965861301228663, + "grad_norm": 1.9657938480377197, + "learning_rate": 1e-06, + "loss": 0.7667, + "mean_token_accuracy": 0.7626104354858398, + "num_tokens": 971076413.0, + "step": 35967 + }, + { + "epoch": 2.965943761853715, + "grad_norm": 2.057968854904175, + "learning_rate": 1e-06, + "loss": 0.513, + "mean_token_accuracy": 0.8184254765510559, + "num_tokens": 971104220.0, + "step": 35968 + }, + { + "epoch": 2.9660262224787663, + "grad_norm": 2.1995689868927, + "learning_rate": 1e-06, + "loss": 0.7059, + "mean_token_accuracy": 0.7690733671188354, + "num_tokens": 971130036.0, + "step": 35969 + }, + { + "epoch": 2.966108683103818, + "grad_norm": 2.2711169719696045, + "learning_rate": 1e-06, + "loss": 0.7414, + "mean_token_accuracy": 0.7712110280990601, + "num_tokens": 971157218.0, + "step": 35970 + }, + { + "epoch": 2.9661911437288695, + "grad_norm": 2.0643460750579834, + "learning_rate": 1e-06, + "loss": 0.7106, + "mean_token_accuracy": 0.7693889141082764, + "num_tokens": 971187575.0, + "step": 35971 + }, + { + "epoch": 2.966273604353921, + "grad_norm": 2.0821707248687744, + "learning_rate": 1e-06, + "loss": 0.653, + "mean_token_accuracy": 0.7896521091461182, + "num_tokens": 971216051.0, + "step": 35972 + }, + { + "epoch": 2.9663560649789726, + "grad_norm": 2.1003811359405518, + "learning_rate": 1e-06, + "loss": 0.7107, + "mean_token_accuracy": 0.7708081007003784, + "num_tokens": 971246358.0, + "step": 35973 + }, + { + "epoch": 2.966438525604024, + "grad_norm": 2.378009080886841, + "learning_rate": 1e-06, + "loss": 0.7351, + "mean_token_accuracy": 0.7652952671051025, + "num_tokens": 971268882.0, + "step": 35974 + }, + { + "epoch": 2.9665209862290753, + "grad_norm": 2.3277080059051514, + "learning_rate": 1e-06, + "loss": 0.6423, + "mean_token_accuracy": 0.7877892851829529, + "num_tokens": 971290150.0, + "step": 35975 + }, + { + "epoch": 2.966603446854127, + "grad_norm": 2.1128647327423096, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7455100417137146, + "num_tokens": 971318742.0, + "step": 35976 + }, + { + "epoch": 2.966685907479179, + "grad_norm": 2.4689507484436035, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7557130455970764, + "num_tokens": 971340910.0, + "step": 35977 + }, + { + "epoch": 2.9667683681042303, + "grad_norm": 2.2319650650024414, + "learning_rate": 1e-06, + "loss": 0.8031, + "mean_token_accuracy": 0.7387717962265015, + "num_tokens": 971369384.0, + "step": 35978 + }, + { + "epoch": 2.9668508287292816, + "grad_norm": 1.8641209602355957, + "learning_rate": 1e-06, + "loss": 0.783, + "mean_token_accuracy": 0.7526733875274658, + "num_tokens": 971408192.0, + "step": 35979 + }, + { + "epoch": 2.9669332893543334, + "grad_norm": 2.026165008544922, + "learning_rate": 1e-06, + "loss": 0.7633, + "mean_token_accuracy": 0.7500774264335632, + "num_tokens": 971442207.0, + "step": 35980 + }, + { + "epoch": 2.9670157499793848, + "grad_norm": 2.298748731613159, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7205749154090881, + "num_tokens": 971470147.0, + "step": 35981 + }, + { + "epoch": 2.967098210604436, + "grad_norm": 2.1067004203796387, + "learning_rate": 1e-06, + "loss": 0.6865, + "mean_token_accuracy": 0.7734715938568115, + "num_tokens": 971499149.0, + "step": 35982 + }, + { + "epoch": 2.967180671229488, + "grad_norm": 2.131614923477173, + "learning_rate": 1e-06, + "loss": 0.6127, + "mean_token_accuracy": 0.79482102394104, + "num_tokens": 971529386.0, + "step": 35983 + }, + { + "epoch": 2.9672631318545397, + "grad_norm": 2.2177159786224365, + "learning_rate": 1e-06, + "loss": 0.7492, + "mean_token_accuracy": 0.7604355216026306, + "num_tokens": 971559197.0, + "step": 35984 + }, + { + "epoch": 2.967345592479591, + "grad_norm": 2.023820638656616, + "learning_rate": 1e-06, + "loss": 0.6043, + "mean_token_accuracy": 0.797255277633667, + "num_tokens": 971590348.0, + "step": 35985 + }, + { + "epoch": 2.9674280531046424, + "grad_norm": 1.951162338256836, + "learning_rate": 1e-06, + "loss": 0.631, + "mean_token_accuracy": 0.7887462377548218, + "num_tokens": 971624117.0, + "step": 35986 + }, + { + "epoch": 2.967510513729694, + "grad_norm": 2.1741554737091064, + "learning_rate": 1e-06, + "loss": 0.6983, + "mean_token_accuracy": 0.7722361087799072, + "num_tokens": 971653607.0, + "step": 35987 + }, + { + "epoch": 2.9675929743547456, + "grad_norm": 2.1135668754577637, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7591313123703003, + "num_tokens": 971684254.0, + "step": 35988 + }, + { + "epoch": 2.967675434979797, + "grad_norm": 1.9006285667419434, + "learning_rate": 1e-06, + "loss": 0.5222, + "mean_token_accuracy": 0.8255918025970459, + "num_tokens": 971715199.0, + "step": 35989 + }, + { + "epoch": 2.9677578956048487, + "grad_norm": 2.1512153148651123, + "learning_rate": 1e-06, + "loss": 0.5903, + "mean_token_accuracy": 0.8069684505462646, + "num_tokens": 971742275.0, + "step": 35990 + }, + { + "epoch": 2.9678403562299, + "grad_norm": 1.8635233640670776, + "learning_rate": 1e-06, + "loss": 0.6722, + "mean_token_accuracy": 0.7946305274963379, + "num_tokens": 971774198.0, + "step": 35991 + }, + { + "epoch": 2.967922816854952, + "grad_norm": 2.274519681930542, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.744396448135376, + "num_tokens": 971800055.0, + "step": 35992 + }, + { + "epoch": 2.968005277480003, + "grad_norm": 1.990110993385315, + "learning_rate": 1e-06, + "loss": 0.6979, + "mean_token_accuracy": 0.7749342918395996, + "num_tokens": 971832367.0, + "step": 35993 + }, + { + "epoch": 2.968087738105055, + "grad_norm": 2.240020990371704, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7384366393089294, + "num_tokens": 971860214.0, + "step": 35994 + }, + { + "epoch": 2.9681701987301063, + "grad_norm": 2.5321273803710938, + "learning_rate": 1e-06, + "loss": 0.6731, + "mean_token_accuracy": 0.7703828811645508, + "num_tokens": 971882502.0, + "step": 35995 + }, + { + "epoch": 2.9682526593551577, + "grad_norm": 1.9965163469314575, + "learning_rate": 1e-06, + "loss": 0.696, + "mean_token_accuracy": 0.7701891660690308, + "num_tokens": 971918295.0, + "step": 35996 + }, + { + "epoch": 2.9683351199802095, + "grad_norm": 2.299912929534912, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7316312789916992, + "num_tokens": 971945533.0, + "step": 35997 + }, + { + "epoch": 2.968417580605261, + "grad_norm": 2.220998525619507, + "learning_rate": 1e-06, + "loss": 0.6887, + "mean_token_accuracy": 0.7675576210021973, + "num_tokens": 971973970.0, + "step": 35998 + }, + { + "epoch": 2.9685000412303126, + "grad_norm": 1.7509734630584717, + "learning_rate": 1e-06, + "loss": 0.6176, + "mean_token_accuracy": 0.798241138458252, + "num_tokens": 972013965.0, + "step": 35999 + }, + { + "epoch": 2.968582501855364, + "grad_norm": 2.060943841934204, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7571970820426941, + "num_tokens": 972044590.0, + "step": 36000 + }, + { + "epoch": 2.968664962480416, + "grad_norm": 2.6426265239715576, + "learning_rate": 1e-06, + "loss": 0.7234, + "mean_token_accuracy": 0.7580348253250122, + "num_tokens": 972064972.0, + "step": 36001 + }, + { + "epoch": 2.968747423105467, + "grad_norm": 2.0146985054016113, + "learning_rate": 1e-06, + "loss": 0.5975, + "mean_token_accuracy": 0.7918457984924316, + "num_tokens": 972094981.0, + "step": 36002 + }, + { + "epoch": 2.9688298837305185, + "grad_norm": 2.4090638160705566, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7559748291969299, + "num_tokens": 972120589.0, + "step": 36003 + }, + { + "epoch": 2.9689123443555703, + "grad_norm": 1.8222030401229858, + "learning_rate": 1e-06, + "loss": 0.7448, + "mean_token_accuracy": 0.7585722804069519, + "num_tokens": 972156171.0, + "step": 36004 + }, + { + "epoch": 2.9689948049806216, + "grad_norm": 2.0503084659576416, + "learning_rate": 1e-06, + "loss": 0.7138, + "mean_token_accuracy": 0.7701236605644226, + "num_tokens": 972187736.0, + "step": 36005 + }, + { + "epoch": 2.9690772656056734, + "grad_norm": 1.7804076671600342, + "learning_rate": 1e-06, + "loss": 0.743, + "mean_token_accuracy": 0.7617363929748535, + "num_tokens": 972226976.0, + "step": 36006 + }, + { + "epoch": 2.9691597262307248, + "grad_norm": 2.0032100677490234, + "learning_rate": 1e-06, + "loss": 0.645, + "mean_token_accuracy": 0.7990496158599854, + "num_tokens": 972256812.0, + "step": 36007 + }, + { + "epoch": 2.9692421868557766, + "grad_norm": 1.9930274486541748, + "learning_rate": 1e-06, + "loss": 0.7131, + "mean_token_accuracy": 0.7718733549118042, + "num_tokens": 972286558.0, + "step": 36008 + }, + { + "epoch": 2.969324647480828, + "grad_norm": 2.0795395374298096, + "learning_rate": 1e-06, + "loss": 0.619, + "mean_token_accuracy": 0.7892951369285583, + "num_tokens": 972315118.0, + "step": 36009 + }, + { + "epoch": 2.9694071081058793, + "grad_norm": 2.081808567047119, + "learning_rate": 1e-06, + "loss": 0.5119, + "mean_token_accuracy": 0.8165337443351746, + "num_tokens": 972341590.0, + "step": 36010 + }, + { + "epoch": 2.969489568730931, + "grad_norm": 2.0335662364959717, + "learning_rate": 1e-06, + "loss": 0.715, + "mean_token_accuracy": 0.7658920884132385, + "num_tokens": 972376726.0, + "step": 36011 + }, + { + "epoch": 2.9695720293559824, + "grad_norm": 1.8749736547470093, + "learning_rate": 1e-06, + "loss": 0.6772, + "mean_token_accuracy": 0.7791272401809692, + "num_tokens": 972411481.0, + "step": 36012 + }, + { + "epoch": 2.969654489981034, + "grad_norm": 1.9940592050552368, + "learning_rate": 1e-06, + "loss": 0.6255, + "mean_token_accuracy": 0.7865842580795288, + "num_tokens": 972442415.0, + "step": 36013 + }, + { + "epoch": 2.9697369506060856, + "grad_norm": 2.3821768760681152, + "learning_rate": 1e-06, + "loss": 0.7093, + "mean_token_accuracy": 0.7734686136245728, + "num_tokens": 972466852.0, + "step": 36014 + }, + { + "epoch": 2.9698194112311374, + "grad_norm": 2.032256603240967, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.7830280065536499, + "num_tokens": 972498423.0, + "step": 36015 + }, + { + "epoch": 2.9699018718561887, + "grad_norm": 2.199080228805542, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7288889288902283, + "num_tokens": 972526516.0, + "step": 36016 + }, + { + "epoch": 2.96998433248124, + "grad_norm": 1.9020739793777466, + "learning_rate": 1e-06, + "loss": 0.6619, + "mean_token_accuracy": 0.7837774753570557, + "num_tokens": 972561256.0, + "step": 36017 + }, + { + "epoch": 2.970066793106292, + "grad_norm": 2.192575216293335, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.7626855373382568, + "num_tokens": 972589916.0, + "step": 36018 + }, + { + "epoch": 2.970149253731343, + "grad_norm": 2.179370164871216, + "learning_rate": 1e-06, + "loss": 0.6856, + "mean_token_accuracy": 0.7855738997459412, + "num_tokens": 972617248.0, + "step": 36019 + }, + { + "epoch": 2.9702317143563945, + "grad_norm": 1.8655922412872314, + "learning_rate": 1e-06, + "loss": 0.7579, + "mean_token_accuracy": 0.7565370202064514, + "num_tokens": 972651568.0, + "step": 36020 + }, + { + "epoch": 2.9703141749814463, + "grad_norm": 1.741742730140686, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.7988455295562744, + "num_tokens": 972693212.0, + "step": 36021 + }, + { + "epoch": 2.970396635606498, + "grad_norm": 2.063749313354492, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7711877822875977, + "num_tokens": 972725310.0, + "step": 36022 + }, + { + "epoch": 2.9704790962315495, + "grad_norm": 2.1962499618530273, + "learning_rate": 1e-06, + "loss": 0.6899, + "mean_token_accuracy": 0.7762393355369568, + "num_tokens": 972754568.0, + "step": 36023 + }, + { + "epoch": 2.970561556856601, + "grad_norm": 2.0112032890319824, + "learning_rate": 1e-06, + "loss": 0.6052, + "mean_token_accuracy": 0.8057315349578857, + "num_tokens": 972785104.0, + "step": 36024 + }, + { + "epoch": 2.9706440174816526, + "grad_norm": 2.2135281562805176, + "learning_rate": 1e-06, + "loss": 0.7009, + "mean_token_accuracy": 0.7679979801177979, + "num_tokens": 972812564.0, + "step": 36025 + }, + { + "epoch": 2.970726478106704, + "grad_norm": 2.5138027667999268, + "learning_rate": 1e-06, + "loss": 0.7287, + "mean_token_accuracy": 0.7633081674575806, + "num_tokens": 972834222.0, + "step": 36026 + }, + { + "epoch": 2.9708089387317553, + "grad_norm": 2.2130532264709473, + "learning_rate": 1e-06, + "loss": 0.6364, + "mean_token_accuracy": 0.7967045903205872, + "num_tokens": 972860699.0, + "step": 36027 + }, + { + "epoch": 2.970891399356807, + "grad_norm": 2.0051796436309814, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7706875205039978, + "num_tokens": 972896135.0, + "step": 36028 + }, + { + "epoch": 2.970973859981859, + "grad_norm": 1.8133710622787476, + "learning_rate": 1e-06, + "loss": 0.7191, + "mean_token_accuracy": 0.7636535167694092, + "num_tokens": 972936465.0, + "step": 36029 + }, + { + "epoch": 2.9710563206069103, + "grad_norm": 1.9330027103424072, + "learning_rate": 1e-06, + "loss": 0.738, + "mean_token_accuracy": 0.7576131820678711, + "num_tokens": 972972640.0, + "step": 36030 + }, + { + "epoch": 2.9711387812319616, + "grad_norm": 2.256800413131714, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.768121600151062, + "num_tokens": 972999817.0, + "step": 36031 + }, + { + "epoch": 2.9712212418570134, + "grad_norm": 2.040572166442871, + "learning_rate": 1e-06, + "loss": 0.623, + "mean_token_accuracy": 0.790351152420044, + "num_tokens": 973030249.0, + "step": 36032 + }, + { + "epoch": 2.9713037024820648, + "grad_norm": 2.092886209487915, + "learning_rate": 1e-06, + "loss": 0.5481, + "mean_token_accuracy": 0.8194732069969177, + "num_tokens": 973057905.0, + "step": 36033 + }, + { + "epoch": 2.971386163107116, + "grad_norm": 1.974770426750183, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7322823405265808, + "num_tokens": 973092639.0, + "step": 36034 + }, + { + "epoch": 2.971468623732168, + "grad_norm": 1.8931920528411865, + "learning_rate": 1e-06, + "loss": 0.7534, + "mean_token_accuracy": 0.7582630515098572, + "num_tokens": 973130092.0, + "step": 36035 + }, + { + "epoch": 2.9715510843572193, + "grad_norm": 2.315808057785034, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7219424247741699, + "num_tokens": 973157071.0, + "step": 36036 + }, + { + "epoch": 2.971633544982271, + "grad_norm": 2.0142974853515625, + "learning_rate": 1e-06, + "loss": 0.72, + "mean_token_accuracy": 0.7733030915260315, + "num_tokens": 973189012.0, + "step": 36037 + }, + { + "epoch": 2.9717160056073224, + "grad_norm": 1.9251514673233032, + "learning_rate": 1e-06, + "loss": 0.6437, + "mean_token_accuracy": 0.7889291644096375, + "num_tokens": 973222897.0, + "step": 36038 + }, + { + "epoch": 2.971798466232374, + "grad_norm": 2.024186849594116, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7421309351921082, + "num_tokens": 973253895.0, + "step": 36039 + }, + { + "epoch": 2.9718809268574256, + "grad_norm": 2.3434524536132812, + "learning_rate": 1e-06, + "loss": 0.7432, + "mean_token_accuracy": 0.7674922943115234, + "num_tokens": 973280755.0, + "step": 36040 + }, + { + "epoch": 2.971963387482477, + "grad_norm": 2.3879764080047607, + "learning_rate": 1e-06, + "loss": 0.6675, + "mean_token_accuracy": 0.7755141854286194, + "num_tokens": 973305677.0, + "step": 36041 + }, + { + "epoch": 2.9720458481075287, + "grad_norm": 1.9858759641647339, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7444683313369751, + "num_tokens": 973340187.0, + "step": 36042 + }, + { + "epoch": 2.97212830873258, + "grad_norm": 1.9184545278549194, + "learning_rate": 1e-06, + "loss": 0.5626, + "mean_token_accuracy": 0.8066105246543884, + "num_tokens": 973372412.0, + "step": 36043 + }, + { + "epoch": 2.972210769357632, + "grad_norm": 2.1679558753967285, + "learning_rate": 1e-06, + "loss": 0.7122, + "mean_token_accuracy": 0.7710269689559937, + "num_tokens": 973401325.0, + "step": 36044 + }, + { + "epoch": 2.972293229982683, + "grad_norm": 2.146958351135254, + "learning_rate": 1e-06, + "loss": 0.6825, + "mean_token_accuracy": 0.7851094603538513, + "num_tokens": 973430010.0, + "step": 36045 + }, + { + "epoch": 2.972375690607735, + "grad_norm": 2.079756736755371, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7421197891235352, + "num_tokens": 973460471.0, + "step": 36046 + }, + { + "epoch": 2.9724581512327863, + "grad_norm": 2.3368892669677734, + "learning_rate": 1e-06, + "loss": 0.6407, + "mean_token_accuracy": 0.7911133766174316, + "num_tokens": 973487146.0, + "step": 36047 + }, + { + "epoch": 2.9725406118578377, + "grad_norm": 2.5179147720336914, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7344634532928467, + "num_tokens": 973512259.0, + "step": 36048 + }, + { + "epoch": 2.9726230724828895, + "grad_norm": 1.9955291748046875, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.7884229421615601, + "num_tokens": 973544838.0, + "step": 36049 + }, + { + "epoch": 2.972705533107941, + "grad_norm": 2.165515422821045, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.7519224882125854, + "num_tokens": 973574696.0, + "step": 36050 + }, + { + "epoch": 2.9727879937329926, + "grad_norm": 1.742134690284729, + "learning_rate": 1e-06, + "loss": 0.6083, + "mean_token_accuracy": 0.8032081723213196, + "num_tokens": 973611520.0, + "step": 36051 + }, + { + "epoch": 2.972870454358044, + "grad_norm": 1.9853514432907104, + "learning_rate": 1e-06, + "loss": 0.6064, + "mean_token_accuracy": 0.7998629808425903, + "num_tokens": 973643274.0, + "step": 36052 + }, + { + "epoch": 2.9729529149830958, + "grad_norm": 1.9632140398025513, + "learning_rate": 1e-06, + "loss": 0.7201, + "mean_token_accuracy": 0.7713391780853271, + "num_tokens": 973679270.0, + "step": 36053 + }, + { + "epoch": 2.973035375608147, + "grad_norm": 1.8355894088745117, + "learning_rate": 1e-06, + "loss": 0.5467, + "mean_token_accuracy": 0.8219594955444336, + "num_tokens": 973713290.0, + "step": 36054 + }, + { + "epoch": 2.9731178362331985, + "grad_norm": 2.0798890590667725, + "learning_rate": 1e-06, + "loss": 0.616, + "mean_token_accuracy": 0.7880859971046448, + "num_tokens": 973741794.0, + "step": 36055 + }, + { + "epoch": 2.9732002968582503, + "grad_norm": 2.221416473388672, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.7905645966529846, + "num_tokens": 973770028.0, + "step": 36056 + }, + { + "epoch": 2.9732827574833016, + "grad_norm": 1.8575592041015625, + "learning_rate": 1e-06, + "loss": 0.6516, + "mean_token_accuracy": 0.7829781770706177, + "num_tokens": 973806384.0, + "step": 36057 + }, + { + "epoch": 2.9733652181083534, + "grad_norm": 2.063589096069336, + "learning_rate": 1e-06, + "loss": 0.5563, + "mean_token_accuracy": 0.8159351348876953, + "num_tokens": 973834426.0, + "step": 36058 + }, + { + "epoch": 2.9734476787334048, + "grad_norm": 2.014704465866089, + "learning_rate": 1e-06, + "loss": 0.7645, + "mean_token_accuracy": 0.7596101760864258, + "num_tokens": 973867395.0, + "step": 36059 + }, + { + "epoch": 2.9735301393584566, + "grad_norm": 2.2086093425750732, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7664185762405396, + "num_tokens": 973894950.0, + "step": 36060 + }, + { + "epoch": 2.973612599983508, + "grad_norm": 2.1845953464508057, + "learning_rate": 1e-06, + "loss": 0.603, + "mean_token_accuracy": 0.8008230924606323, + "num_tokens": 973922773.0, + "step": 36061 + }, + { + "epoch": 2.9736950606085593, + "grad_norm": 1.9288057088851929, + "learning_rate": 1e-06, + "loss": 0.6249, + "mean_token_accuracy": 0.7864368557929993, + "num_tokens": 973960925.0, + "step": 36062 + }, + { + "epoch": 2.973777521233611, + "grad_norm": 2.0128281116485596, + "learning_rate": 1e-06, + "loss": 0.5784, + "mean_token_accuracy": 0.8018662929534912, + "num_tokens": 973993817.0, + "step": 36063 + }, + { + "epoch": 2.9738599818586624, + "grad_norm": 1.9786261320114136, + "learning_rate": 1e-06, + "loss": 0.6575, + "mean_token_accuracy": 0.787018358707428, + "num_tokens": 974023862.0, + "step": 36064 + }, + { + "epoch": 2.9739424424837138, + "grad_norm": 2.1152048110961914, + "learning_rate": 1e-06, + "loss": 0.555, + "mean_token_accuracy": 0.8092523813247681, + "num_tokens": 974052005.0, + "step": 36065 + }, + { + "epoch": 2.9740249031087655, + "grad_norm": 2.209602117538452, + "learning_rate": 1e-06, + "loss": 0.6049, + "mean_token_accuracy": 0.7886562943458557, + "num_tokens": 974076676.0, + "step": 36066 + }, + { + "epoch": 2.9741073637338173, + "grad_norm": 1.9605815410614014, + "learning_rate": 1e-06, + "loss": 0.6173, + "mean_token_accuracy": 0.7917014360427856, + "num_tokens": 974107320.0, + "step": 36067 + }, + { + "epoch": 2.9741898243588687, + "grad_norm": 2.0961103439331055, + "learning_rate": 1e-06, + "loss": 0.6324, + "mean_token_accuracy": 0.7805545926094055, + "num_tokens": 974135611.0, + "step": 36068 + }, + { + "epoch": 2.97427228498392, + "grad_norm": 2.400390863418579, + "learning_rate": 1e-06, + "loss": 0.7271, + "mean_token_accuracy": 0.7589244246482849, + "num_tokens": 974161622.0, + "step": 36069 + }, + { + "epoch": 2.974354745608972, + "grad_norm": 2.209218978881836, + "learning_rate": 1e-06, + "loss": 0.6179, + "mean_token_accuracy": 0.795921802520752, + "num_tokens": 974186319.0, + "step": 36070 + }, + { + "epoch": 2.974437206234023, + "grad_norm": 2.348294973373413, + "learning_rate": 1e-06, + "loss": 0.7111, + "mean_token_accuracy": 0.7632045745849609, + "num_tokens": 974212600.0, + "step": 36071 + }, + { + "epoch": 2.9745196668590745, + "grad_norm": 1.8624343872070312, + "learning_rate": 1e-06, + "loss": 0.6077, + "mean_token_accuracy": 0.8124693632125854, + "num_tokens": 974246730.0, + "step": 36072 + }, + { + "epoch": 2.9746021274841263, + "grad_norm": 1.9970102310180664, + "learning_rate": 1e-06, + "loss": 0.7837, + "mean_token_accuracy": 0.7530359625816345, + "num_tokens": 974278703.0, + "step": 36073 + }, + { + "epoch": 2.974684588109178, + "grad_norm": 2.4323129653930664, + "learning_rate": 1e-06, + "loss": 0.7423, + "mean_token_accuracy": 0.7498178482055664, + "num_tokens": 974304746.0, + "step": 36074 + }, + { + "epoch": 2.9747670487342295, + "grad_norm": 1.9036204814910889, + "learning_rate": 1e-06, + "loss": 0.6942, + "mean_token_accuracy": 0.7659769058227539, + "num_tokens": 974339588.0, + "step": 36075 + }, + { + "epoch": 2.974849509359281, + "grad_norm": 1.9335100650787354, + "learning_rate": 1e-06, + "loss": 0.6066, + "mean_token_accuracy": 0.7811181545257568, + "num_tokens": 974372494.0, + "step": 36076 + }, + { + "epoch": 2.9749319699843326, + "grad_norm": 2.0512640476226807, + "learning_rate": 1e-06, + "loss": 0.6444, + "mean_token_accuracy": 0.7831683158874512, + "num_tokens": 974403566.0, + "step": 36077 + }, + { + "epoch": 2.975014430609384, + "grad_norm": 1.8494452238082886, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7494799494743347, + "num_tokens": 974442322.0, + "step": 36078 + }, + { + "epoch": 2.9750968912344353, + "grad_norm": 2.1134345531463623, + "learning_rate": 1e-06, + "loss": 0.6684, + "mean_token_accuracy": 0.7869359254837036, + "num_tokens": 974472195.0, + "step": 36079 + }, + { + "epoch": 2.975179351859487, + "grad_norm": 1.9619828462600708, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7788386940956116, + "num_tokens": 974507691.0, + "step": 36080 + }, + { + "epoch": 2.9752618124845385, + "grad_norm": 2.0532240867614746, + "learning_rate": 1e-06, + "loss": 0.68, + "mean_token_accuracy": 0.7859079241752625, + "num_tokens": 974536882.0, + "step": 36081 + }, + { + "epoch": 2.9753442731095903, + "grad_norm": 2.1217763423919678, + "learning_rate": 1e-06, + "loss": 0.6115, + "mean_token_accuracy": 0.7905793190002441, + "num_tokens": 974568131.0, + "step": 36082 + }, + { + "epoch": 2.9754267337346416, + "grad_norm": 2.308638095855713, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7315231561660767, + "num_tokens": 974592819.0, + "step": 36083 + }, + { + "epoch": 2.9755091943596934, + "grad_norm": 2.191906452178955, + "learning_rate": 1e-06, + "loss": 0.6082, + "mean_token_accuracy": 0.7994377017021179, + "num_tokens": 974619620.0, + "step": 36084 + }, + { + "epoch": 2.9755916549847448, + "grad_norm": 2.0047945976257324, + "learning_rate": 1e-06, + "loss": 0.5575, + "mean_token_accuracy": 0.8040916919708252, + "num_tokens": 974651508.0, + "step": 36085 + }, + { + "epoch": 2.975674115609796, + "grad_norm": 2.2664430141448975, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.762097954750061, + "num_tokens": 974679087.0, + "step": 36086 + }, + { + "epoch": 2.975756576234848, + "grad_norm": 2.35212779045105, + "learning_rate": 1e-06, + "loss": 0.6624, + "mean_token_accuracy": 0.7812532186508179, + "num_tokens": 974704238.0, + "step": 36087 + }, + { + "epoch": 2.9758390368598993, + "grad_norm": 2.0953049659729004, + "learning_rate": 1e-06, + "loss": 0.7264, + "mean_token_accuracy": 0.7624610662460327, + "num_tokens": 974736507.0, + "step": 36088 + }, + { + "epoch": 2.975921497484951, + "grad_norm": 2.0462520122528076, + "learning_rate": 1e-06, + "loss": 0.6677, + "mean_token_accuracy": 0.7809109687805176, + "num_tokens": 974767252.0, + "step": 36089 + }, + { + "epoch": 2.9760039581100024, + "grad_norm": 1.971693754196167, + "learning_rate": 1e-06, + "loss": 0.6197, + "mean_token_accuracy": 0.7919464111328125, + "num_tokens": 974799595.0, + "step": 36090 + }, + { + "epoch": 2.976086418735054, + "grad_norm": 2.0876235961914062, + "learning_rate": 1e-06, + "loss": 0.6611, + "mean_token_accuracy": 0.785122811794281, + "num_tokens": 974829930.0, + "step": 36091 + }, + { + "epoch": 2.9761688793601055, + "grad_norm": 2.0036797523498535, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.7561297416687012, + "num_tokens": 974863562.0, + "step": 36092 + }, + { + "epoch": 2.976251339985157, + "grad_norm": 2.111816167831421, + "learning_rate": 1e-06, + "loss": 0.688, + "mean_token_accuracy": 0.7772294878959656, + "num_tokens": 974892553.0, + "step": 36093 + }, + { + "epoch": 2.9763338006102087, + "grad_norm": 2.2071404457092285, + "learning_rate": 1e-06, + "loss": 0.7955, + "mean_token_accuracy": 0.7449067831039429, + "num_tokens": 974921463.0, + "step": 36094 + }, + { + "epoch": 2.97641626123526, + "grad_norm": 1.985450029373169, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7455669641494751, + "num_tokens": 974952902.0, + "step": 36095 + }, + { + "epoch": 2.976498721860312, + "grad_norm": 2.3587024211883545, + "learning_rate": 1e-06, + "loss": 0.6967, + "mean_token_accuracy": 0.7640366554260254, + "num_tokens": 974978633.0, + "step": 36096 + }, + { + "epoch": 2.976581182485363, + "grad_norm": 2.439603090286255, + "learning_rate": 1e-06, + "loss": 0.7192, + "mean_token_accuracy": 0.7582154870033264, + "num_tokens": 975002666.0, + "step": 36097 + }, + { + "epoch": 2.976663643110415, + "grad_norm": 2.098583936691284, + "learning_rate": 1e-06, + "loss": 0.6508, + "mean_token_accuracy": 0.7931933999061584, + "num_tokens": 975029804.0, + "step": 36098 + }, + { + "epoch": 2.9767461037354663, + "grad_norm": 2.399266004562378, + "learning_rate": 1e-06, + "loss": 0.7594, + "mean_token_accuracy": 0.757829487323761, + "num_tokens": 975056213.0, + "step": 36099 + }, + { + "epoch": 2.9768285643605177, + "grad_norm": 1.8289145231246948, + "learning_rate": 1e-06, + "loss": 0.5888, + "mean_token_accuracy": 0.8030740022659302, + "num_tokens": 975091108.0, + "step": 36100 + }, + { + "epoch": 2.9769110249855695, + "grad_norm": 2.1187896728515625, + "learning_rate": 1e-06, + "loss": 0.5469, + "mean_token_accuracy": 0.8174553513526917, + "num_tokens": 975119420.0, + "step": 36101 + }, + { + "epoch": 2.976993485610621, + "grad_norm": 2.000149726867676, + "learning_rate": 1e-06, + "loss": 0.6352, + "mean_token_accuracy": 0.7911404371261597, + "num_tokens": 975150625.0, + "step": 36102 + }, + { + "epoch": 2.9770759462356726, + "grad_norm": 2.1164278984069824, + "learning_rate": 1e-06, + "loss": 0.6847, + "mean_token_accuracy": 0.7671425342559814, + "num_tokens": 975178668.0, + "step": 36103 + }, + { + "epoch": 2.977158406860724, + "grad_norm": 2.345848560333252, + "learning_rate": 1e-06, + "loss": 0.7386, + "mean_token_accuracy": 0.7628799676895142, + "num_tokens": 975203915.0, + "step": 36104 + }, + { + "epoch": 2.9772408674857758, + "grad_norm": 2.2179367542266846, + "learning_rate": 1e-06, + "loss": 0.658, + "mean_token_accuracy": 0.7803471684455872, + "num_tokens": 975229651.0, + "step": 36105 + }, + { + "epoch": 2.977323328110827, + "grad_norm": 1.8346236944198608, + "learning_rate": 1e-06, + "loss": 0.6626, + "mean_token_accuracy": 0.782060980796814, + "num_tokens": 975263964.0, + "step": 36106 + }, + { + "epoch": 2.9774057887358785, + "grad_norm": 2.035335063934326, + "learning_rate": 1e-06, + "loss": 0.5298, + "mean_token_accuracy": 0.813422441482544, + "num_tokens": 975291719.0, + "step": 36107 + }, + { + "epoch": 2.9774882493609303, + "grad_norm": 2.474350929260254, + "learning_rate": 1e-06, + "loss": 0.7458, + "mean_token_accuracy": 0.760332465171814, + "num_tokens": 975314385.0, + "step": 36108 + }, + { + "epoch": 2.9775707099859816, + "grad_norm": 2.258368968963623, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.7481170296669006, + "num_tokens": 975343024.0, + "step": 36109 + }, + { + "epoch": 2.977653170611033, + "grad_norm": 2.2431447505950928, + "learning_rate": 1e-06, + "loss": 0.7434, + "mean_token_accuracy": 0.7557331919670105, + "num_tokens": 975369948.0, + "step": 36110 + }, + { + "epoch": 2.9777356312360848, + "grad_norm": 1.986965537071228, + "learning_rate": 1e-06, + "loss": 0.6614, + "mean_token_accuracy": 0.7780354619026184, + "num_tokens": 975402500.0, + "step": 36111 + }, + { + "epoch": 2.9778180918611366, + "grad_norm": 2.119791269302368, + "learning_rate": 1e-06, + "loss": 0.5772, + "mean_token_accuracy": 0.8114458322525024, + "num_tokens": 975428622.0, + "step": 36112 + }, + { + "epoch": 2.977900552486188, + "grad_norm": 2.1994640827178955, + "learning_rate": 1e-06, + "loss": 0.7346, + "mean_token_accuracy": 0.7628713846206665, + "num_tokens": 975455986.0, + "step": 36113 + }, + { + "epoch": 2.9779830131112393, + "grad_norm": 2.178603410720825, + "learning_rate": 1e-06, + "loss": 0.691, + "mean_token_accuracy": 0.7585548758506775, + "num_tokens": 975486077.0, + "step": 36114 + }, + { + "epoch": 2.978065473736291, + "grad_norm": 1.9591342210769653, + "learning_rate": 1e-06, + "loss": 0.6623, + "mean_token_accuracy": 0.786316990852356, + "num_tokens": 975519980.0, + "step": 36115 + }, + { + "epoch": 2.9781479343613424, + "grad_norm": 2.065962791442871, + "learning_rate": 1e-06, + "loss": 0.6806, + "mean_token_accuracy": 0.7721613049507141, + "num_tokens": 975549610.0, + "step": 36116 + }, + { + "epoch": 2.9782303949863937, + "grad_norm": 1.9372094869613647, + "learning_rate": 1e-06, + "loss": 0.6054, + "mean_token_accuracy": 0.7935030460357666, + "num_tokens": 975584603.0, + "step": 36117 + }, + { + "epoch": 2.9783128556114455, + "grad_norm": 2.2270774841308594, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7810469269752502, + "num_tokens": 975612613.0, + "step": 36118 + }, + { + "epoch": 2.9783953162364973, + "grad_norm": 1.9503825902938843, + "learning_rate": 1e-06, + "loss": 0.6735, + "mean_token_accuracy": 0.7797804474830627, + "num_tokens": 975648018.0, + "step": 36119 + }, + { + "epoch": 2.9784777768615487, + "grad_norm": 2.3886187076568604, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7302477359771729, + "num_tokens": 975673097.0, + "step": 36120 + }, + { + "epoch": 2.9785602374866, + "grad_norm": 2.173027276992798, + "learning_rate": 1e-06, + "loss": 0.6078, + "mean_token_accuracy": 0.7869611978530884, + "num_tokens": 975699477.0, + "step": 36121 + }, + { + "epoch": 2.978642698111652, + "grad_norm": 2.1977028846740723, + "learning_rate": 1e-06, + "loss": 0.6596, + "mean_token_accuracy": 0.7862353920936584, + "num_tokens": 975726303.0, + "step": 36122 + }, + { + "epoch": 2.978725158736703, + "grad_norm": 2.1401586532592773, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7511382102966309, + "num_tokens": 975759198.0, + "step": 36123 + }, + { + "epoch": 2.9788076193617545, + "grad_norm": 2.0062575340270996, + "learning_rate": 1e-06, + "loss": 0.583, + "mean_token_accuracy": 0.807096004486084, + "num_tokens": 975788668.0, + "step": 36124 + }, + { + "epoch": 2.9788900799868063, + "grad_norm": 2.567704916000366, + "learning_rate": 1e-06, + "loss": 0.6403, + "mean_token_accuracy": 0.7812995910644531, + "num_tokens": 975809798.0, + "step": 36125 + }, + { + "epoch": 2.9789725406118577, + "grad_norm": 1.9603383541107178, + "learning_rate": 1e-06, + "loss": 0.6798, + "mean_token_accuracy": 0.7806653380393982, + "num_tokens": 975843631.0, + "step": 36126 + }, + { + "epoch": 2.9790550012369095, + "grad_norm": 2.0489330291748047, + "learning_rate": 1e-06, + "loss": 0.4276, + "mean_token_accuracy": 0.8545670509338379, + "num_tokens": 975869504.0, + "step": 36127 + }, + { + "epoch": 2.979137461861961, + "grad_norm": 2.1385982036590576, + "learning_rate": 1e-06, + "loss": 0.7029, + "mean_token_accuracy": 0.7726725339889526, + "num_tokens": 975899494.0, + "step": 36128 + }, + { + "epoch": 2.9792199224870126, + "grad_norm": 2.1540751457214355, + "learning_rate": 1e-06, + "loss": 0.6681, + "mean_token_accuracy": 0.7766464352607727, + "num_tokens": 975926545.0, + "step": 36129 + }, + { + "epoch": 2.979302383112064, + "grad_norm": 2.073113441467285, + "learning_rate": 1e-06, + "loss": 0.6652, + "mean_token_accuracy": 0.7780216336250305, + "num_tokens": 975955581.0, + "step": 36130 + }, + { + "epoch": 2.9793848437371153, + "grad_norm": 1.9295953512191772, + "learning_rate": 1e-06, + "loss": 0.7101, + "mean_token_accuracy": 0.7688448429107666, + "num_tokens": 975993273.0, + "step": 36131 + }, + { + "epoch": 2.979467304362167, + "grad_norm": 1.8840388059616089, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7309734225273132, + "num_tokens": 976030967.0, + "step": 36132 + }, + { + "epoch": 2.9795497649872185, + "grad_norm": 2.0169458389282227, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.7847241759300232, + "num_tokens": 976061078.0, + "step": 36133 + }, + { + "epoch": 2.9796322256122703, + "grad_norm": 2.075986862182617, + "learning_rate": 1e-06, + "loss": 0.5998, + "mean_token_accuracy": 0.8031638860702515, + "num_tokens": 976088408.0, + "step": 36134 + }, + { + "epoch": 2.9797146862373216, + "grad_norm": 2.2905163764953613, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7305516600608826, + "num_tokens": 976114047.0, + "step": 36135 + }, + { + "epoch": 2.9797971468623734, + "grad_norm": 2.2212157249450684, + "learning_rate": 1e-06, + "loss": 0.7164, + "mean_token_accuracy": 0.7705146074295044, + "num_tokens": 976142541.0, + "step": 36136 + }, + { + "epoch": 2.9798796074874248, + "grad_norm": 2.555501937866211, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.7912658452987671, + "num_tokens": 976167295.0, + "step": 36137 + }, + { + "epoch": 2.979962068112476, + "grad_norm": 1.8318792581558228, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.7427586317062378, + "num_tokens": 976204596.0, + "step": 36138 + }, + { + "epoch": 2.980044528737528, + "grad_norm": 1.9039709568023682, + "learning_rate": 1e-06, + "loss": 0.7021, + "mean_token_accuracy": 0.7769489288330078, + "num_tokens": 976242590.0, + "step": 36139 + }, + { + "epoch": 2.9801269893625792, + "grad_norm": 2.1910572052001953, + "learning_rate": 1e-06, + "loss": 0.6932, + "mean_token_accuracy": 0.7784156799316406, + "num_tokens": 976270541.0, + "step": 36140 + }, + { + "epoch": 2.980209449987631, + "grad_norm": 2.2772409915924072, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7682514190673828, + "num_tokens": 976297728.0, + "step": 36141 + }, + { + "epoch": 2.9802919106126824, + "grad_norm": 1.9693571329116821, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7396703958511353, + "num_tokens": 976330715.0, + "step": 36142 + }, + { + "epoch": 2.980374371237734, + "grad_norm": 2.3090808391571045, + "learning_rate": 1e-06, + "loss": 0.7381, + "mean_token_accuracy": 0.7632580399513245, + "num_tokens": 976357128.0, + "step": 36143 + }, + { + "epoch": 2.9804568318627855, + "grad_norm": 2.1119325160980225, + "learning_rate": 1e-06, + "loss": 0.7181, + "mean_token_accuracy": 0.7667520046234131, + "num_tokens": 976386295.0, + "step": 36144 + }, + { + "epoch": 2.980539292487837, + "grad_norm": 2.1548995971679688, + "learning_rate": 1e-06, + "loss": 0.5909, + "mean_token_accuracy": 0.7974636554718018, + "num_tokens": 976414199.0, + "step": 36145 + }, + { + "epoch": 2.9806217531128887, + "grad_norm": 1.9891451597213745, + "learning_rate": 1e-06, + "loss": 0.6162, + "mean_token_accuracy": 0.7948458194732666, + "num_tokens": 976443371.0, + "step": 36146 + }, + { + "epoch": 2.98070421373794, + "grad_norm": 2.3616247177124023, + "learning_rate": 1e-06, + "loss": 0.7408, + "mean_token_accuracy": 0.7681566476821899, + "num_tokens": 976469661.0, + "step": 36147 + }, + { + "epoch": 2.980786674362992, + "grad_norm": 1.8207770586013794, + "learning_rate": 1e-06, + "loss": 0.6241, + "mean_token_accuracy": 0.7980597019195557, + "num_tokens": 976508056.0, + "step": 36148 + }, + { + "epoch": 2.980869134988043, + "grad_norm": 1.7933709621429443, + "learning_rate": 1e-06, + "loss": 0.578, + "mean_token_accuracy": 0.8108323812484741, + "num_tokens": 976543601.0, + "step": 36149 + }, + { + "epoch": 2.980951595613095, + "grad_norm": 1.862987756729126, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.7601673007011414, + "num_tokens": 976580765.0, + "step": 36150 + }, + { + "epoch": 2.9810340562381463, + "grad_norm": 1.9835575819015503, + "learning_rate": 1e-06, + "loss": 0.7061, + "mean_token_accuracy": 0.772088885307312, + "num_tokens": 976616660.0, + "step": 36151 + }, + { + "epoch": 2.9811165168631977, + "grad_norm": 1.8778181076049805, + "learning_rate": 1e-06, + "loss": 0.7798, + "mean_token_accuracy": 0.7492429614067078, + "num_tokens": 976655461.0, + "step": 36152 + }, + { + "epoch": 2.9811989774882495, + "grad_norm": 1.9624509811401367, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.780052900314331, + "num_tokens": 976687657.0, + "step": 36153 + }, + { + "epoch": 2.981281438113301, + "grad_norm": 2.128844738006592, + "learning_rate": 1e-06, + "loss": 0.7708, + "mean_token_accuracy": 0.7541599869728088, + "num_tokens": 976716589.0, + "step": 36154 + }, + { + "epoch": 2.981363898738352, + "grad_norm": 1.8889752626419067, + "learning_rate": 1e-06, + "loss": 0.7623, + "mean_token_accuracy": 0.7530550360679626, + "num_tokens": 976751034.0, + "step": 36155 + }, + { + "epoch": 2.981446359363404, + "grad_norm": 1.962449312210083, + "learning_rate": 1e-06, + "loss": 0.7, + "mean_token_accuracy": 0.7678275108337402, + "num_tokens": 976782655.0, + "step": 36156 + }, + { + "epoch": 2.9815288199884558, + "grad_norm": 2.4173192977905273, + "learning_rate": 1e-06, + "loss": 0.7389, + "mean_token_accuracy": 0.7580324411392212, + "num_tokens": 976806409.0, + "step": 36157 + }, + { + "epoch": 2.981611280613507, + "grad_norm": 2.220280408859253, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7417991161346436, + "num_tokens": 976834072.0, + "step": 36158 + }, + { + "epoch": 2.9816937412385585, + "grad_norm": 2.1436057090759277, + "learning_rate": 1e-06, + "loss": 0.6829, + "mean_token_accuracy": 0.776369571685791, + "num_tokens": 976861587.0, + "step": 36159 + }, + { + "epoch": 2.9817762018636103, + "grad_norm": 2.08454966545105, + "learning_rate": 1e-06, + "loss": 0.6819, + "mean_token_accuracy": 0.7730529308319092, + "num_tokens": 976891616.0, + "step": 36160 + }, + { + "epoch": 2.9818586624886616, + "grad_norm": 2.4015562534332275, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.749526858329773, + "num_tokens": 976916736.0, + "step": 36161 + }, + { + "epoch": 2.981941123113713, + "grad_norm": 2.3641698360443115, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7381425499916077, + "num_tokens": 976942380.0, + "step": 36162 + }, + { + "epoch": 2.9820235837387647, + "grad_norm": 1.9166934490203857, + "learning_rate": 1e-06, + "loss": 0.6603, + "mean_token_accuracy": 0.7772626876831055, + "num_tokens": 976979101.0, + "step": 36163 + }, + { + "epoch": 2.9821060443638165, + "grad_norm": 2.1021103858947754, + "learning_rate": 1e-06, + "loss": 0.6927, + "mean_token_accuracy": 0.7716928124427795, + "num_tokens": 977007008.0, + "step": 36164 + }, + { + "epoch": 2.982188504988868, + "grad_norm": 1.9696012735366821, + "learning_rate": 1e-06, + "loss": 0.5778, + "mean_token_accuracy": 0.807856559753418, + "num_tokens": 977036487.0, + "step": 36165 + }, + { + "epoch": 2.9822709656139192, + "grad_norm": 1.9009987115859985, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7256817817687988, + "num_tokens": 977072346.0, + "step": 36166 + }, + { + "epoch": 2.982353426238971, + "grad_norm": 2.114769697189331, + "learning_rate": 1e-06, + "loss": 0.7775, + "mean_token_accuracy": 0.7517259120941162, + "num_tokens": 977104479.0, + "step": 36167 + }, + { + "epoch": 2.9824358868640224, + "grad_norm": 2.1673591136932373, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7571290731430054, + "num_tokens": 977140060.0, + "step": 36168 + }, + { + "epoch": 2.9825183474890737, + "grad_norm": 1.994145154953003, + "learning_rate": 1e-06, + "loss": 0.7644, + "mean_token_accuracy": 0.7544544339179993, + "num_tokens": 977171852.0, + "step": 36169 + }, + { + "epoch": 2.9826008081141255, + "grad_norm": 2.487924814224243, + "learning_rate": 1e-06, + "loss": 0.6973, + "mean_token_accuracy": 0.7755118608474731, + "num_tokens": 977195689.0, + "step": 36170 + }, + { + "epoch": 2.982683268739177, + "grad_norm": 2.53653883934021, + "learning_rate": 1e-06, + "loss": 0.7219, + "mean_token_accuracy": 0.7646707892417908, + "num_tokens": 977218304.0, + "step": 36171 + }, + { + "epoch": 2.9827657293642287, + "grad_norm": 1.9671825170516968, + "learning_rate": 1e-06, + "loss": 0.6724, + "mean_token_accuracy": 0.7826204895973206, + "num_tokens": 977254333.0, + "step": 36172 + }, + { + "epoch": 2.98284818998928, + "grad_norm": 1.850137710571289, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7748003005981445, + "num_tokens": 977290216.0, + "step": 36173 + }, + { + "epoch": 2.982930650614332, + "grad_norm": 1.913097858428955, + "learning_rate": 1e-06, + "loss": 0.6271, + "mean_token_accuracy": 0.7956769466400146, + "num_tokens": 977323629.0, + "step": 36174 + }, + { + "epoch": 2.983013111239383, + "grad_norm": 2.472623825073242, + "learning_rate": 1e-06, + "loss": 0.6672, + "mean_token_accuracy": 0.7797881364822388, + "num_tokens": 977346102.0, + "step": 36175 + }, + { + "epoch": 2.9830955718644345, + "grad_norm": 2.1226322650909424, + "learning_rate": 1e-06, + "loss": 0.7042, + "mean_token_accuracy": 0.7611274123191833, + "num_tokens": 977376378.0, + "step": 36176 + }, + { + "epoch": 2.9831780324894863, + "grad_norm": 2.1839842796325684, + "learning_rate": 1e-06, + "loss": 0.6962, + "mean_token_accuracy": 0.7760181427001953, + "num_tokens": 977403592.0, + "step": 36177 + }, + { + "epoch": 2.9832604931145377, + "grad_norm": 1.998356580734253, + "learning_rate": 1e-06, + "loss": 0.5921, + "mean_token_accuracy": 0.7966539859771729, + "num_tokens": 977435054.0, + "step": 36178 + }, + { + "epoch": 2.9833429537395895, + "grad_norm": 2.134320020675659, + "learning_rate": 1e-06, + "loss": 0.5587, + "mean_token_accuracy": 0.8138341903686523, + "num_tokens": 977462186.0, + "step": 36179 + }, + { + "epoch": 2.983425414364641, + "grad_norm": 2.2096939086914062, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7602143883705139, + "num_tokens": 977493636.0, + "step": 36180 + }, + { + "epoch": 2.9835078749896926, + "grad_norm": 2.038266897201538, + "learning_rate": 1e-06, + "loss": 0.748, + "mean_token_accuracy": 0.7619014978408813, + "num_tokens": 977527572.0, + "step": 36181 + }, + { + "epoch": 2.983590335614744, + "grad_norm": 2.160454750061035, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7393741607666016, + "num_tokens": 977555309.0, + "step": 36182 + }, + { + "epoch": 2.9836727962397953, + "grad_norm": 1.94526207447052, + "learning_rate": 1e-06, + "loss": 0.5648, + "mean_token_accuracy": 0.8046859502792358, + "num_tokens": 977585516.0, + "step": 36183 + }, + { + "epoch": 2.983755256864847, + "grad_norm": 2.202755928039551, + "learning_rate": 1e-06, + "loss": 0.6936, + "mean_token_accuracy": 0.7667072415351868, + "num_tokens": 977614412.0, + "step": 36184 + }, + { + "epoch": 2.9838377174898985, + "grad_norm": 2.150780200958252, + "learning_rate": 1e-06, + "loss": 0.6947, + "mean_token_accuracy": 0.7679699659347534, + "num_tokens": 977642311.0, + "step": 36185 + }, + { + "epoch": 2.9839201781149503, + "grad_norm": 2.082866907119751, + "learning_rate": 1e-06, + "loss": 0.6765, + "mean_token_accuracy": 0.7758867144584656, + "num_tokens": 977673554.0, + "step": 36186 + }, + { + "epoch": 2.9840026387400016, + "grad_norm": 2.283740997314453, + "learning_rate": 1e-06, + "loss": 0.729, + "mean_token_accuracy": 0.7531958222389221, + "num_tokens": 977701710.0, + "step": 36187 + }, + { + "epoch": 2.9840850993650534, + "grad_norm": 2.1712636947631836, + "learning_rate": 1e-06, + "loss": 0.709, + "mean_token_accuracy": 0.7761576175689697, + "num_tokens": 977732217.0, + "step": 36188 + }, + { + "epoch": 2.9841675599901047, + "grad_norm": 1.9309537410736084, + "learning_rate": 1e-06, + "loss": 0.6395, + "mean_token_accuracy": 0.7975742220878601, + "num_tokens": 977765782.0, + "step": 36189 + }, + { + "epoch": 2.984250020615156, + "grad_norm": 2.30252742767334, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.7463339567184448, + "num_tokens": 977791269.0, + "step": 36190 + }, + { + "epoch": 2.984332481240208, + "grad_norm": 1.8003637790679932, + "learning_rate": 1e-06, + "loss": 0.5723, + "mean_token_accuracy": 0.8032933473587036, + "num_tokens": 977825553.0, + "step": 36191 + }, + { + "epoch": 2.9844149418652592, + "grad_norm": 2.330315351486206, + "learning_rate": 1e-06, + "loss": 0.6824, + "mean_token_accuracy": 0.7729123830795288, + "num_tokens": 977851940.0, + "step": 36192 + }, + { + "epoch": 2.984497402490311, + "grad_norm": 2.382620096206665, + "learning_rate": 1e-06, + "loss": 0.7184, + "mean_token_accuracy": 0.7605960369110107, + "num_tokens": 977876177.0, + "step": 36193 + }, + { + "epoch": 2.9845798631153624, + "grad_norm": 1.9167717695236206, + "learning_rate": 1e-06, + "loss": 0.6107, + "mean_token_accuracy": 0.8014199733734131, + "num_tokens": 977909506.0, + "step": 36194 + }, + { + "epoch": 2.984662323740414, + "grad_norm": 2.3386013507843018, + "learning_rate": 1e-06, + "loss": 0.5713, + "mean_token_accuracy": 0.8032053112983704, + "num_tokens": 977931786.0, + "step": 36195 + }, + { + "epoch": 2.9847447843654655, + "grad_norm": 2.0527522563934326, + "learning_rate": 1e-06, + "loss": 0.7158, + "mean_token_accuracy": 0.7704676389694214, + "num_tokens": 977964665.0, + "step": 36196 + }, + { + "epoch": 2.984827244990517, + "grad_norm": 1.9186762571334839, + "learning_rate": 1e-06, + "loss": 0.7086, + "mean_token_accuracy": 0.7685272693634033, + "num_tokens": 977997848.0, + "step": 36197 + }, + { + "epoch": 2.9849097056155687, + "grad_norm": 2.1918578147888184, + "learning_rate": 1e-06, + "loss": 0.6246, + "mean_token_accuracy": 0.7904415130615234, + "num_tokens": 978023164.0, + "step": 36198 + }, + { + "epoch": 2.98499216624062, + "grad_norm": 1.934765100479126, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.7948319911956787, + "num_tokens": 978057130.0, + "step": 36199 + }, + { + "epoch": 2.9850746268656714, + "grad_norm": 2.3032166957855225, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7750840187072754, + "num_tokens": 978082327.0, + "step": 36200 + }, + { + "epoch": 2.985157087490723, + "grad_norm": 2.330660581588745, + "learning_rate": 1e-06, + "loss": 0.7485, + "mean_token_accuracy": 0.7528796792030334, + "num_tokens": 978108897.0, + "step": 36201 + }, + { + "epoch": 2.985239548115775, + "grad_norm": 2.184415340423584, + "learning_rate": 1e-06, + "loss": 0.6327, + "mean_token_accuracy": 0.7910116910934448, + "num_tokens": 978135039.0, + "step": 36202 + }, + { + "epoch": 2.9853220087408263, + "grad_norm": 1.881372332572937, + "learning_rate": 1e-06, + "loss": 0.6305, + "mean_token_accuracy": 0.7932670712471008, + "num_tokens": 978167409.0, + "step": 36203 + }, + { + "epoch": 2.9854044693658777, + "grad_norm": 2.075263261795044, + "learning_rate": 1e-06, + "loss": 0.6975, + "mean_token_accuracy": 0.7809314727783203, + "num_tokens": 978196913.0, + "step": 36204 + }, + { + "epoch": 2.9854869299909295, + "grad_norm": 1.8795510530471802, + "learning_rate": 1e-06, + "loss": 0.6379, + "mean_token_accuracy": 0.7868824005126953, + "num_tokens": 978233859.0, + "step": 36205 + }, + { + "epoch": 2.985569390615981, + "grad_norm": 2.161231279373169, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7391445636749268, + "num_tokens": 978264588.0, + "step": 36206 + }, + { + "epoch": 2.985651851241032, + "grad_norm": 1.9852452278137207, + "learning_rate": 1e-06, + "loss": 0.6325, + "mean_token_accuracy": 0.802222490310669, + "num_tokens": 978296640.0, + "step": 36207 + }, + { + "epoch": 2.985734311866084, + "grad_norm": 2.390648126602173, + "learning_rate": 1e-06, + "loss": 0.7288, + "mean_token_accuracy": 0.7736201286315918, + "num_tokens": 978322573.0, + "step": 36208 + }, + { + "epoch": 2.9858167724911358, + "grad_norm": 2.0108420848846436, + "learning_rate": 1e-06, + "loss": 0.7324, + "mean_token_accuracy": 0.761744499206543, + "num_tokens": 978356018.0, + "step": 36209 + }, + { + "epoch": 2.985899233116187, + "grad_norm": 1.9753271341323853, + "learning_rate": 1e-06, + "loss": 0.5747, + "mean_token_accuracy": 0.8014131784439087, + "num_tokens": 978387364.0, + "step": 36210 + }, + { + "epoch": 2.9859816937412385, + "grad_norm": 1.989768147468567, + "learning_rate": 1e-06, + "loss": 0.6143, + "mean_token_accuracy": 0.7955780029296875, + "num_tokens": 978417596.0, + "step": 36211 + }, + { + "epoch": 2.9860641543662902, + "grad_norm": 1.853000283241272, + "learning_rate": 1e-06, + "loss": 0.5451, + "mean_token_accuracy": 0.8153766989707947, + "num_tokens": 978449425.0, + "step": 36212 + }, + { + "epoch": 2.9861466149913416, + "grad_norm": 1.9951049089431763, + "learning_rate": 1e-06, + "loss": 0.6823, + "mean_token_accuracy": 0.7763513326644897, + "num_tokens": 978479302.0, + "step": 36213 + }, + { + "epoch": 2.986229075616393, + "grad_norm": 2.1433193683624268, + "learning_rate": 1e-06, + "loss": 0.4852, + "mean_token_accuracy": 0.8325068354606628, + "num_tokens": 978505309.0, + "step": 36214 + }, + { + "epoch": 2.9863115362414447, + "grad_norm": 2.2329328060150146, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7365441918373108, + "num_tokens": 978532167.0, + "step": 36215 + }, + { + "epoch": 2.986393996866496, + "grad_norm": 2.0530781745910645, + "learning_rate": 1e-06, + "loss": 0.5731, + "mean_token_accuracy": 0.8094241619110107, + "num_tokens": 978567896.0, + "step": 36216 + }, + { + "epoch": 2.986476457491548, + "grad_norm": 2.1551742553710938, + "learning_rate": 1e-06, + "loss": 0.701, + "mean_token_accuracy": 0.7752019762992859, + "num_tokens": 978595300.0, + "step": 36217 + }, + { + "epoch": 2.9865589181165992, + "grad_norm": 1.8904401063919067, + "learning_rate": 1e-06, + "loss": 0.6726, + "mean_token_accuracy": 0.7792949676513672, + "num_tokens": 978629268.0, + "step": 36218 + }, + { + "epoch": 2.986641378741651, + "grad_norm": 2.1414902210235596, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.733344316482544, + "num_tokens": 978663286.0, + "step": 36219 + }, + { + "epoch": 2.9867238393667024, + "grad_norm": 2.456791877746582, + "learning_rate": 1e-06, + "loss": 0.6702, + "mean_token_accuracy": 0.7867141962051392, + "num_tokens": 978688964.0, + "step": 36220 + }, + { + "epoch": 2.9868062999917537, + "grad_norm": 1.845044493675232, + "learning_rate": 1e-06, + "loss": 0.5496, + "mean_token_accuracy": 0.8109906911849976, + "num_tokens": 978719177.0, + "step": 36221 + }, + { + "epoch": 2.9868887606168055, + "grad_norm": 2.084947109222412, + "learning_rate": 1e-06, + "loss": 0.7034, + "mean_token_accuracy": 0.7626110911369324, + "num_tokens": 978749138.0, + "step": 36222 + }, + { + "epoch": 2.986971221241857, + "grad_norm": 2.0521929264068604, + "learning_rate": 1e-06, + "loss": 0.6926, + "mean_token_accuracy": 0.7816271781921387, + "num_tokens": 978781659.0, + "step": 36223 + }, + { + "epoch": 2.9870536818669087, + "grad_norm": 2.2874703407287598, + "learning_rate": 1e-06, + "loss": 0.7616, + "mean_token_accuracy": 0.7533540725708008, + "num_tokens": 978809024.0, + "step": 36224 + }, + { + "epoch": 2.98713614249196, + "grad_norm": 2.4041953086853027, + "learning_rate": 1e-06, + "loss": 0.7651, + "mean_token_accuracy": 0.7518638372421265, + "num_tokens": 978834869.0, + "step": 36225 + }, + { + "epoch": 2.987218603117012, + "grad_norm": 2.573930263519287, + "learning_rate": 1e-06, + "loss": 0.7946, + "mean_token_accuracy": 0.7472382187843323, + "num_tokens": 978857095.0, + "step": 36226 + }, + { + "epoch": 2.987301063742063, + "grad_norm": 2.115105390548706, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7492114901542664, + "num_tokens": 978893270.0, + "step": 36227 + }, + { + "epoch": 2.9873835243671145, + "grad_norm": 2.3416216373443604, + "learning_rate": 1e-06, + "loss": 0.6605, + "mean_token_accuracy": 0.7783299684524536, + "num_tokens": 978919730.0, + "step": 36228 + }, + { + "epoch": 2.9874659849921663, + "grad_norm": 2.220536947250366, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7383255958557129, + "num_tokens": 978950069.0, + "step": 36229 + }, + { + "epoch": 2.9875484456172177, + "grad_norm": 1.9354674816131592, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7522825598716736, + "num_tokens": 978984921.0, + "step": 36230 + }, + { + "epoch": 2.9876309062422695, + "grad_norm": 1.9234992265701294, + "learning_rate": 1e-06, + "loss": 0.5568, + "mean_token_accuracy": 0.804305911064148, + "num_tokens": 979016733.0, + "step": 36231 + }, + { + "epoch": 2.987713366867321, + "grad_norm": 2.0854358673095703, + "learning_rate": 1e-06, + "loss": 0.7792, + "mean_token_accuracy": 0.7543971538543701, + "num_tokens": 979045190.0, + "step": 36232 + }, + { + "epoch": 2.9877958274923726, + "grad_norm": 2.1270911693573, + "learning_rate": 1e-06, + "loss": 0.7143, + "mean_token_accuracy": 0.7586950063705444, + "num_tokens": 979075359.0, + "step": 36233 + }, + { + "epoch": 2.987878288117424, + "grad_norm": 2.106670379638672, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7822304964065552, + "num_tokens": 979104324.0, + "step": 36234 + }, + { + "epoch": 2.9879607487424753, + "grad_norm": 1.9710321426391602, + "learning_rate": 1e-06, + "loss": 0.6861, + "mean_token_accuracy": 0.7763183116912842, + "num_tokens": 979136269.0, + "step": 36235 + }, + { + "epoch": 2.988043209367527, + "grad_norm": 1.9482311010360718, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.7890982627868652, + "num_tokens": 979169372.0, + "step": 36236 + }, + { + "epoch": 2.9881256699925784, + "grad_norm": 1.8868883848190308, + "learning_rate": 1e-06, + "loss": 0.712, + "mean_token_accuracy": 0.7712889313697815, + "num_tokens": 979205447.0, + "step": 36237 + }, + { + "epoch": 2.9882081306176302, + "grad_norm": 1.9999667406082153, + "learning_rate": 1e-06, + "loss": 0.6835, + "mean_token_accuracy": 0.7834662199020386, + "num_tokens": 979234728.0, + "step": 36238 + }, + { + "epoch": 2.9882905912426816, + "grad_norm": 2.1490609645843506, + "learning_rate": 1e-06, + "loss": 0.6237, + "mean_token_accuracy": 0.7901102304458618, + "num_tokens": 979263926.0, + "step": 36239 + }, + { + "epoch": 2.9883730518677334, + "grad_norm": 2.1471009254455566, + "learning_rate": 1e-06, + "loss": 0.6354, + "mean_token_accuracy": 0.7880213856697083, + "num_tokens": 979296803.0, + "step": 36240 + }, + { + "epoch": 2.9884555124927847, + "grad_norm": 2.2103121280670166, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7361936569213867, + "num_tokens": 979326112.0, + "step": 36241 + }, + { + "epoch": 2.988537973117836, + "grad_norm": 2.1017696857452393, + "learning_rate": 1e-06, + "loss": 0.6573, + "mean_token_accuracy": 0.7870789170265198, + "num_tokens": 979356012.0, + "step": 36242 + }, + { + "epoch": 2.988620433742888, + "grad_norm": 2.2537643909454346, + "learning_rate": 1e-06, + "loss": 0.7126, + "mean_token_accuracy": 0.7671275734901428, + "num_tokens": 979382535.0, + "step": 36243 + }, + { + "epoch": 2.9887028943679392, + "grad_norm": 2.069526433944702, + "learning_rate": 1e-06, + "loss": 0.6471, + "mean_token_accuracy": 0.7800233960151672, + "num_tokens": 979410603.0, + "step": 36244 + }, + { + "epoch": 2.9887853549929906, + "grad_norm": 1.915332317352295, + "learning_rate": 1e-06, + "loss": 0.7194, + "mean_token_accuracy": 0.7792401909828186, + "num_tokens": 979445916.0, + "step": 36245 + }, + { + "epoch": 2.9888678156180424, + "grad_norm": 2.587202787399292, + "learning_rate": 1e-06, + "loss": 0.7222, + "mean_token_accuracy": 0.7706541419029236, + "num_tokens": 979466043.0, + "step": 36246 + }, + { + "epoch": 2.988950276243094, + "grad_norm": 2.3864734172821045, + "learning_rate": 1e-06, + "loss": 0.7124, + "mean_token_accuracy": 0.7687525749206543, + "num_tokens": 979489516.0, + "step": 36247 + }, + { + "epoch": 2.9890327368681455, + "grad_norm": 2.269178628921509, + "learning_rate": 1e-06, + "loss": 0.6323, + "mean_token_accuracy": 0.7930496335029602, + "num_tokens": 979514294.0, + "step": 36248 + }, + { + "epoch": 2.989115197493197, + "grad_norm": 1.9769995212554932, + "learning_rate": 1e-06, + "loss": 0.463, + "mean_token_accuracy": 0.8348878622055054, + "num_tokens": 979546273.0, + "step": 36249 + }, + { + "epoch": 2.9891976581182487, + "grad_norm": 2.1742358207702637, + "learning_rate": 1e-06, + "loss": 0.5666, + "mean_token_accuracy": 0.8142865896224976, + "num_tokens": 979571645.0, + "step": 36250 + }, + { + "epoch": 2.9892801187433, + "grad_norm": 2.3395960330963135, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7463594675064087, + "num_tokens": 979597357.0, + "step": 36251 + }, + { + "epoch": 2.9893625793683514, + "grad_norm": 2.0914905071258545, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7434110641479492, + "num_tokens": 979629834.0, + "step": 36252 + }, + { + "epoch": 2.989445039993403, + "grad_norm": 2.098253011703491, + "learning_rate": 1e-06, + "loss": 0.7338, + "mean_token_accuracy": 0.7647871375083923, + "num_tokens": 979659642.0, + "step": 36253 + }, + { + "epoch": 2.989527500618455, + "grad_norm": 2.0399112701416016, + "learning_rate": 1e-06, + "loss": 0.6378, + "mean_token_accuracy": 0.7956525087356567, + "num_tokens": 979687803.0, + "step": 36254 + }, + { + "epoch": 2.9896099612435063, + "grad_norm": 2.1805055141448975, + "learning_rate": 1e-06, + "loss": 0.6421, + "mean_token_accuracy": 0.7741549015045166, + "num_tokens": 979713445.0, + "step": 36255 + }, + { + "epoch": 2.9896924218685577, + "grad_norm": 2.0092082023620605, + "learning_rate": 1e-06, + "loss": 0.583, + "mean_token_accuracy": 0.8079043626785278, + "num_tokens": 979745584.0, + "step": 36256 + }, + { + "epoch": 2.9897748824936095, + "grad_norm": 2.0913846492767334, + "learning_rate": 1e-06, + "loss": 0.682, + "mean_token_accuracy": 0.7749276161193848, + "num_tokens": 979774461.0, + "step": 36257 + }, + { + "epoch": 2.989857343118661, + "grad_norm": 2.0153234004974365, + "learning_rate": 1e-06, + "loss": 0.6706, + "mean_token_accuracy": 0.7859790921211243, + "num_tokens": 979806741.0, + "step": 36258 + }, + { + "epoch": 2.989939803743712, + "grad_norm": 1.9879353046417236, + "learning_rate": 1e-06, + "loss": 0.7876, + "mean_token_accuracy": 0.7453204393386841, + "num_tokens": 979840374.0, + "step": 36259 + }, + { + "epoch": 2.990022264368764, + "grad_norm": 2.244314670562744, + "learning_rate": 1e-06, + "loss": 0.7539, + "mean_token_accuracy": 0.7614327073097229, + "num_tokens": 979870386.0, + "step": 36260 + }, + { + "epoch": 2.9901047249938153, + "grad_norm": 1.928520917892456, + "learning_rate": 1e-06, + "loss": 0.7326, + "mean_token_accuracy": 0.7657194137573242, + "num_tokens": 979905377.0, + "step": 36261 + }, + { + "epoch": 2.990187185618867, + "grad_norm": 2.0070481300354004, + "learning_rate": 1e-06, + "loss": 0.635, + "mean_token_accuracy": 0.7924156188964844, + "num_tokens": 979936595.0, + "step": 36262 + }, + { + "epoch": 2.9902696462439184, + "grad_norm": 2.2165215015411377, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7615677118301392, + "num_tokens": 979965200.0, + "step": 36263 + }, + { + "epoch": 2.9903521068689702, + "grad_norm": 2.11513614654541, + "learning_rate": 1e-06, + "loss": 0.6314, + "mean_token_accuracy": 0.7815021872520447, + "num_tokens": 979994476.0, + "step": 36264 + }, + { + "epoch": 2.9904345674940216, + "grad_norm": 2.193676471710205, + "learning_rate": 1e-06, + "loss": 0.5731, + "mean_token_accuracy": 0.793976366519928, + "num_tokens": 980020032.0, + "step": 36265 + }, + { + "epoch": 2.990517028119073, + "grad_norm": 2.168983221054077, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7478219270706177, + "num_tokens": 980050220.0, + "step": 36266 + }, + { + "epoch": 2.9905994887441247, + "grad_norm": 2.3082473278045654, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.751113772392273, + "num_tokens": 980077599.0, + "step": 36267 + }, + { + "epoch": 2.990681949369176, + "grad_norm": 2.3512654304504395, + "learning_rate": 1e-06, + "loss": 0.7013, + "mean_token_accuracy": 0.7680612206459045, + "num_tokens": 980103864.0, + "step": 36268 + }, + { + "epoch": 2.990764409994228, + "grad_norm": 1.9725066423416138, + "learning_rate": 1e-06, + "loss": 0.7515, + "mean_token_accuracy": 0.7616060972213745, + "num_tokens": 980139971.0, + "step": 36269 + }, + { + "epoch": 2.9908468706192792, + "grad_norm": 2.299504041671753, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7300214767456055, + "num_tokens": 980167474.0, + "step": 36270 + }, + { + "epoch": 2.990929331244331, + "grad_norm": 2.0580179691314697, + "learning_rate": 1e-06, + "loss": 0.6761, + "mean_token_accuracy": 0.7782033681869507, + "num_tokens": 980200023.0, + "step": 36271 + }, + { + "epoch": 2.9910117918693824, + "grad_norm": 2.2554452419281006, + "learning_rate": 1e-06, + "loss": 0.6908, + "mean_token_accuracy": 0.7739375829696655, + "num_tokens": 980227665.0, + "step": 36272 + }, + { + "epoch": 2.9910942524944337, + "grad_norm": 1.964038372039795, + "learning_rate": 1e-06, + "loss": 0.6137, + "mean_token_accuracy": 0.7958303689956665, + "num_tokens": 980258612.0, + "step": 36273 + }, + { + "epoch": 2.9911767131194855, + "grad_norm": 1.8682104349136353, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7559033036231995, + "num_tokens": 980295199.0, + "step": 36274 + }, + { + "epoch": 2.991259173744537, + "grad_norm": 2.0965116024017334, + "learning_rate": 1e-06, + "loss": 0.655, + "mean_token_accuracy": 0.7807449698448181, + "num_tokens": 980325527.0, + "step": 36275 + }, + { + "epoch": 2.9913416343695887, + "grad_norm": 1.7762653827667236, + "learning_rate": 1e-06, + "loss": 0.485, + "mean_token_accuracy": 0.835195004940033, + "num_tokens": 980362093.0, + "step": 36276 + }, + { + "epoch": 2.99142409499464, + "grad_norm": 2.1902079582214355, + "learning_rate": 1e-06, + "loss": 0.6864, + "mean_token_accuracy": 0.7811335325241089, + "num_tokens": 980389984.0, + "step": 36277 + }, + { + "epoch": 2.991506555619692, + "grad_norm": 2.069270610809326, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7610691785812378, + "num_tokens": 980420497.0, + "step": 36278 + }, + { + "epoch": 2.991589016244743, + "grad_norm": 1.776016354560852, + "learning_rate": 1e-06, + "loss": 0.5847, + "mean_token_accuracy": 0.8011682033538818, + "num_tokens": 980459141.0, + "step": 36279 + }, + { + "epoch": 2.9916714768697945, + "grad_norm": 2.500206470489502, + "learning_rate": 1e-06, + "loss": 0.6994, + "mean_token_accuracy": 0.7777256369590759, + "num_tokens": 980483506.0, + "step": 36280 + }, + { + "epoch": 2.9917539374948463, + "grad_norm": 2.044651985168457, + "learning_rate": 1e-06, + "loss": 0.6188, + "mean_token_accuracy": 0.7948293685913086, + "num_tokens": 980512561.0, + "step": 36281 + }, + { + "epoch": 2.9918363981198977, + "grad_norm": 2.1667487621307373, + "learning_rate": 1e-06, + "loss": 0.6651, + "mean_token_accuracy": 0.7776498794555664, + "num_tokens": 980540848.0, + "step": 36282 + }, + { + "epoch": 2.9919188587449494, + "grad_norm": 2.1406145095825195, + "learning_rate": 1e-06, + "loss": 0.6669, + "mean_token_accuracy": 0.7847576141357422, + "num_tokens": 980570749.0, + "step": 36283 + }, + { + "epoch": 2.992001319370001, + "grad_norm": 1.863020420074463, + "learning_rate": 1e-06, + "loss": 0.586, + "mean_token_accuracy": 0.806596040725708, + "num_tokens": 980603153.0, + "step": 36284 + }, + { + "epoch": 2.9920837799950526, + "grad_norm": 2.2800612449645996, + "learning_rate": 1e-06, + "loss": 0.7065, + "mean_token_accuracy": 0.7589256167411804, + "num_tokens": 980627169.0, + "step": 36285 + }, + { + "epoch": 2.992166240620104, + "grad_norm": 1.8882356882095337, + "learning_rate": 1e-06, + "loss": 0.6571, + "mean_token_accuracy": 0.7876019477844238, + "num_tokens": 980665860.0, + "step": 36286 + }, + { + "epoch": 2.9922487012451553, + "grad_norm": 2.373300790786743, + "learning_rate": 1e-06, + "loss": 0.6671, + "mean_token_accuracy": 0.7828565835952759, + "num_tokens": 980687725.0, + "step": 36287 + }, + { + "epoch": 2.992331161870207, + "grad_norm": 2.1425774097442627, + "learning_rate": 1e-06, + "loss": 0.6311, + "mean_token_accuracy": 0.7854759693145752, + "num_tokens": 980718274.0, + "step": 36288 + }, + { + "epoch": 2.9924136224952584, + "grad_norm": 2.2876996994018555, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7409639358520508, + "num_tokens": 980745668.0, + "step": 36289 + }, + { + "epoch": 2.99249608312031, + "grad_norm": 1.9354385137557983, + "learning_rate": 1e-06, + "loss": 0.7028, + "mean_token_accuracy": 0.7710661292076111, + "num_tokens": 980783284.0, + "step": 36290 + }, + { + "epoch": 2.9925785437453616, + "grad_norm": 2.1825053691864014, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7334871292114258, + "num_tokens": 980814158.0, + "step": 36291 + }, + { + "epoch": 2.9926610043704134, + "grad_norm": 1.9964109659194946, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7406598329544067, + "num_tokens": 980847747.0, + "step": 36292 + }, + { + "epoch": 2.9927434649954647, + "grad_norm": 2.2065985202789307, + "learning_rate": 1e-06, + "loss": 0.6191, + "mean_token_accuracy": 0.7941507697105408, + "num_tokens": 980874582.0, + "step": 36293 + }, + { + "epoch": 2.992825925620516, + "grad_norm": 1.9608324766159058, + "learning_rate": 1e-06, + "loss": 0.7818, + "mean_token_accuracy": 0.7524400353431702, + "num_tokens": 980908721.0, + "step": 36294 + }, + { + "epoch": 2.992908386245568, + "grad_norm": 2.070265293121338, + "learning_rate": 1e-06, + "loss": 0.7358, + "mean_token_accuracy": 0.759407103061676, + "num_tokens": 980938928.0, + "step": 36295 + }, + { + "epoch": 2.9929908468706192, + "grad_norm": 2.564192533493042, + "learning_rate": 1e-06, + "loss": 0.7168, + "mean_token_accuracy": 0.7656341791152954, + "num_tokens": 980960843.0, + "step": 36296 + }, + { + "epoch": 2.9930733074956706, + "grad_norm": 2.0130434036254883, + "learning_rate": 1e-06, + "loss": 0.5941, + "mean_token_accuracy": 0.8008305430412292, + "num_tokens": 980992525.0, + "step": 36297 + }, + { + "epoch": 2.9931557681207224, + "grad_norm": 2.176858901977539, + "learning_rate": 1e-06, + "loss": 0.4957, + "mean_token_accuracy": 0.8309873342514038, + "num_tokens": 981020143.0, + "step": 36298 + }, + { + "epoch": 2.993238228745774, + "grad_norm": 2.0698108673095703, + "learning_rate": 1e-06, + "loss": 0.7334, + "mean_token_accuracy": 0.7616450786590576, + "num_tokens": 981050331.0, + "step": 36299 + }, + { + "epoch": 2.9933206893708255, + "grad_norm": 2.253101348876953, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7379542589187622, + "num_tokens": 981077579.0, + "step": 36300 + }, + { + "epoch": 2.993403149995877, + "grad_norm": 2.159862518310547, + "learning_rate": 1e-06, + "loss": 0.7524, + "mean_token_accuracy": 0.7557528018951416, + "num_tokens": 981105588.0, + "step": 36301 + }, + { + "epoch": 2.9934856106209287, + "grad_norm": 1.985586166381836, + "learning_rate": 1e-06, + "loss": 0.6892, + "mean_token_accuracy": 0.7743236422538757, + "num_tokens": 981139232.0, + "step": 36302 + }, + { + "epoch": 2.99356807124598, + "grad_norm": 2.016157627105713, + "learning_rate": 1e-06, + "loss": 0.6535, + "mean_token_accuracy": 0.7999708652496338, + "num_tokens": 981170688.0, + "step": 36303 + }, + { + "epoch": 2.9936505318710314, + "grad_norm": 2.306540012359619, + "learning_rate": 1e-06, + "loss": 0.7395, + "mean_token_accuracy": 0.7554856538772583, + "num_tokens": 981200016.0, + "step": 36304 + }, + { + "epoch": 2.993732992496083, + "grad_norm": 2.065337657928467, + "learning_rate": 1e-06, + "loss": 0.5843, + "mean_token_accuracy": 0.8001085519790649, + "num_tokens": 981231995.0, + "step": 36305 + }, + { + "epoch": 2.9938154531211345, + "grad_norm": 2.1333374977111816, + "learning_rate": 1e-06, + "loss": 0.6339, + "mean_token_accuracy": 0.7827591896057129, + "num_tokens": 981259402.0, + "step": 36306 + }, + { + "epoch": 2.9938979137461863, + "grad_norm": 1.9409376382827759, + "learning_rate": 1e-06, + "loss": 0.8012, + "mean_token_accuracy": 0.7549780011177063, + "num_tokens": 981293972.0, + "step": 36307 + }, + { + "epoch": 2.9939803743712377, + "grad_norm": 1.7753239870071411, + "learning_rate": 1e-06, + "loss": 0.6935, + "mean_token_accuracy": 0.7553561925888062, + "num_tokens": 981333204.0, + "step": 36308 + }, + { + "epoch": 2.9940628349962894, + "grad_norm": 2.035020112991333, + "learning_rate": 1e-06, + "loss": 0.6132, + "mean_token_accuracy": 0.803091287612915, + "num_tokens": 981365709.0, + "step": 36309 + }, + { + "epoch": 2.994145295621341, + "grad_norm": 1.7795377969741821, + "learning_rate": 1e-06, + "loss": 0.5152, + "mean_token_accuracy": 0.8332524299621582, + "num_tokens": 981399281.0, + "step": 36310 + }, + { + "epoch": 2.994227756246392, + "grad_norm": 2.0860090255737305, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.7576332092285156, + "num_tokens": 981428839.0, + "step": 36311 + }, + { + "epoch": 2.994310216871444, + "grad_norm": 2.2795727252960205, + "learning_rate": 1e-06, + "loss": 0.752, + "mean_token_accuracy": 0.7583019733428955, + "num_tokens": 981455086.0, + "step": 36312 + }, + { + "epoch": 2.9943926774964953, + "grad_norm": 2.1927928924560547, + "learning_rate": 1e-06, + "loss": 0.731, + "mean_token_accuracy": 0.7677632570266724, + "num_tokens": 981482460.0, + "step": 36313 + }, + { + "epoch": 2.994475138121547, + "grad_norm": 2.2014636993408203, + "learning_rate": 1e-06, + "loss": 0.6682, + "mean_token_accuracy": 0.7794963121414185, + "num_tokens": 981511786.0, + "step": 36314 + }, + { + "epoch": 2.9945575987465984, + "grad_norm": 1.868045449256897, + "learning_rate": 1e-06, + "loss": 0.6867, + "mean_token_accuracy": 0.7782204151153564, + "num_tokens": 981547328.0, + "step": 36315 + }, + { + "epoch": 2.9946400593716502, + "grad_norm": 2.3223795890808105, + "learning_rate": 1e-06, + "loss": 0.7081, + "mean_token_accuracy": 0.7615618705749512, + "num_tokens": 981575495.0, + "step": 36316 + }, + { + "epoch": 2.9947225199967016, + "grad_norm": 2.2317590713500977, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.757704496383667, + "num_tokens": 981605941.0, + "step": 36317 + }, + { + "epoch": 2.994804980621753, + "grad_norm": 2.238176107406616, + "learning_rate": 1e-06, + "loss": 0.7252, + "mean_token_accuracy": 0.762019157409668, + "num_tokens": 981629790.0, + "step": 36318 + }, + { + "epoch": 2.9948874412468047, + "grad_norm": 2.0372815132141113, + "learning_rate": 1e-06, + "loss": 0.7092, + "mean_token_accuracy": 0.7774708867073059, + "num_tokens": 981661198.0, + "step": 36319 + }, + { + "epoch": 2.994969901871856, + "grad_norm": 1.8945001363754272, + "learning_rate": 1e-06, + "loss": 0.5941, + "mean_token_accuracy": 0.8039984107017517, + "num_tokens": 981693570.0, + "step": 36320 + }, + { + "epoch": 2.995052362496908, + "grad_norm": 2.319420337677002, + "learning_rate": 1e-06, + "loss": 0.6456, + "mean_token_accuracy": 0.7901655435562134, + "num_tokens": 981718977.0, + "step": 36321 + }, + { + "epoch": 2.9951348231219592, + "grad_norm": 2.1723718643188477, + "learning_rate": 1e-06, + "loss": 0.5483, + "mean_token_accuracy": 0.8163719177246094, + "num_tokens": 981744780.0, + "step": 36322 + }, + { + "epoch": 2.995217283747011, + "grad_norm": 2.1565001010894775, + "learning_rate": 1e-06, + "loss": 0.5443, + "mean_token_accuracy": 0.8013246059417725, + "num_tokens": 981772227.0, + "step": 36323 + }, + { + "epoch": 2.9952997443720624, + "grad_norm": 2.2276463508605957, + "learning_rate": 1e-06, + "loss": 0.753, + "mean_token_accuracy": 0.7543935179710388, + "num_tokens": 981803125.0, + "step": 36324 + }, + { + "epoch": 2.9953822049971137, + "grad_norm": 1.7326472997665405, + "learning_rate": 1e-06, + "loss": 0.7275, + "mean_token_accuracy": 0.7580181360244751, + "num_tokens": 981840397.0, + "step": 36325 + }, + { + "epoch": 2.9954646656221655, + "grad_norm": 2.211430072784424, + "learning_rate": 1e-06, + "loss": 0.6692, + "mean_token_accuracy": 0.7770752310752869, + "num_tokens": 981867062.0, + "step": 36326 + }, + { + "epoch": 2.995547126247217, + "grad_norm": 1.8782845735549927, + "learning_rate": 1e-06, + "loss": 0.745, + "mean_token_accuracy": 0.7679084539413452, + "num_tokens": 981905344.0, + "step": 36327 + }, + { + "epoch": 2.9956295868722687, + "grad_norm": 2.238852024078369, + "learning_rate": 1e-06, + "loss": 0.6404, + "mean_token_accuracy": 0.7858399748802185, + "num_tokens": 981933138.0, + "step": 36328 + }, + { + "epoch": 2.99571204749732, + "grad_norm": 2.3651692867279053, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7362319231033325, + "num_tokens": 981957394.0, + "step": 36329 + }, + { + "epoch": 2.995794508122372, + "grad_norm": 2.0070087909698486, + "learning_rate": 1e-06, + "loss": 0.7611, + "mean_token_accuracy": 0.7651165723800659, + "num_tokens": 981989371.0, + "step": 36330 + }, + { + "epoch": 2.995876968747423, + "grad_norm": 1.977372646331787, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7653297781944275, + "num_tokens": 982021765.0, + "step": 36331 + }, + { + "epoch": 2.9959594293724745, + "grad_norm": 1.979193091392517, + "learning_rate": 1e-06, + "loss": 0.4647, + "mean_token_accuracy": 0.8407889008522034, + "num_tokens": 982053634.0, + "step": 36332 + }, + { + "epoch": 2.9960418899975263, + "grad_norm": 2.171919584274292, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7401057481765747, + "num_tokens": 982086624.0, + "step": 36333 + }, + { + "epoch": 2.9961243506225776, + "grad_norm": 2.195568799972534, + "learning_rate": 1e-06, + "loss": 0.7822, + "mean_token_accuracy": 0.7441421747207642, + "num_tokens": 982116816.0, + "step": 36334 + }, + { + "epoch": 2.996206811247629, + "grad_norm": 2.3226914405822754, + "learning_rate": 1e-06, + "loss": 0.6283, + "mean_token_accuracy": 0.7897452712059021, + "num_tokens": 982140633.0, + "step": 36335 + }, + { + "epoch": 2.996289271872681, + "grad_norm": 1.8615792989730835, + "learning_rate": 1e-06, + "loss": 0.7577, + "mean_token_accuracy": 0.7533518671989441, + "num_tokens": 982177562.0, + "step": 36336 + }, + { + "epoch": 2.9963717324977326, + "grad_norm": 2.142132520675659, + "learning_rate": 1e-06, + "loss": 0.7301, + "mean_token_accuracy": 0.7614244222640991, + "num_tokens": 982204550.0, + "step": 36337 + }, + { + "epoch": 2.996454193122784, + "grad_norm": 1.9788604974746704, + "learning_rate": 1e-06, + "loss": 0.6338, + "mean_token_accuracy": 0.792611300945282, + "num_tokens": 982235477.0, + "step": 36338 + }, + { + "epoch": 2.9965366537478353, + "grad_norm": 1.8715083599090576, + "learning_rate": 1e-06, + "loss": 0.7997, + "mean_token_accuracy": 0.7493022084236145, + "num_tokens": 982272673.0, + "step": 36339 + }, + { + "epoch": 2.996619114372887, + "grad_norm": 1.98967707157135, + "learning_rate": 1e-06, + "loss": 0.6937, + "mean_token_accuracy": 0.7695568799972534, + "num_tokens": 982303930.0, + "step": 36340 + }, + { + "epoch": 2.9967015749979384, + "grad_norm": 1.965749979019165, + "learning_rate": 1e-06, + "loss": 0.7057, + "mean_token_accuracy": 0.7697814106941223, + "num_tokens": 982335496.0, + "step": 36341 + }, + { + "epoch": 2.99678403562299, + "grad_norm": 2.101952314376831, + "learning_rate": 1e-06, + "loss": 0.614, + "mean_token_accuracy": 0.8000958561897278, + "num_tokens": 982367018.0, + "step": 36342 + }, + { + "epoch": 2.9968664962480416, + "grad_norm": 1.7499163150787354, + "learning_rate": 1e-06, + "loss": 0.6067, + "mean_token_accuracy": 0.8080642223358154, + "num_tokens": 982405564.0, + "step": 36343 + }, + { + "epoch": 2.9969489568730934, + "grad_norm": 2.1632742881774902, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7676998376846313, + "num_tokens": 982436688.0, + "step": 36344 + }, + { + "epoch": 2.9970314174981447, + "grad_norm": 1.9388277530670166, + "learning_rate": 1e-06, + "loss": 0.7244, + "mean_token_accuracy": 0.7720769643783569, + "num_tokens": 982474178.0, + "step": 36345 + }, + { + "epoch": 2.997113878123196, + "grad_norm": 1.9426252841949463, + "learning_rate": 1e-06, + "loss": 0.6336, + "mean_token_accuracy": 0.7925112247467041, + "num_tokens": 982508129.0, + "step": 36346 + }, + { + "epoch": 2.997196338748248, + "grad_norm": 2.4355945587158203, + "learning_rate": 1e-06, + "loss": 0.7699, + "mean_token_accuracy": 0.7495040893554688, + "num_tokens": 982532912.0, + "step": 36347 + }, + { + "epoch": 2.997278799373299, + "grad_norm": 1.9812849760055542, + "learning_rate": 1e-06, + "loss": 0.6025, + "mean_token_accuracy": 0.8023433089256287, + "num_tokens": 982565563.0, + "step": 36348 + }, + { + "epoch": 2.9973612599983506, + "grad_norm": 2.160983085632324, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.785045325756073, + "num_tokens": 982593448.0, + "step": 36349 + }, + { + "epoch": 2.9974437206234024, + "grad_norm": 1.9999597072601318, + "learning_rate": 1e-06, + "loss": 0.6767, + "mean_token_accuracy": 0.7677283883094788, + "num_tokens": 982625468.0, + "step": 36350 + }, + { + "epoch": 2.9975261812484537, + "grad_norm": 1.7075034379959106, + "learning_rate": 1e-06, + "loss": 0.6234, + "mean_token_accuracy": 0.796854555606842, + "num_tokens": 982662910.0, + "step": 36351 + }, + { + "epoch": 2.9976086418735055, + "grad_norm": 2.007269859313965, + "learning_rate": 1e-06, + "loss": 0.7019, + "mean_token_accuracy": 0.7682563066482544, + "num_tokens": 982696127.0, + "step": 36352 + }, + { + "epoch": 2.997691102498557, + "grad_norm": 1.9931124448776245, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7282046675682068, + "num_tokens": 982730614.0, + "step": 36353 + }, + { + "epoch": 2.9977735631236087, + "grad_norm": 1.892863392829895, + "learning_rate": 1e-06, + "loss": 0.6181, + "mean_token_accuracy": 0.8069778680801392, + "num_tokens": 982763237.0, + "step": 36354 + }, + { + "epoch": 2.99785602374866, + "grad_norm": 2.216191053390503, + "learning_rate": 1e-06, + "loss": 0.7091, + "mean_token_accuracy": 0.7703822255134583, + "num_tokens": 982788041.0, + "step": 36355 + }, + { + "epoch": 2.9979384843737114, + "grad_norm": 1.7634588479995728, + "learning_rate": 1e-06, + "loss": 0.5496, + "mean_token_accuracy": 0.8221221566200256, + "num_tokens": 982822587.0, + "step": 36356 + }, + { + "epoch": 2.998020944998763, + "grad_norm": 2.1652145385742188, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7451499104499817, + "num_tokens": 982853121.0, + "step": 36357 + }, + { + "epoch": 2.9981034056238145, + "grad_norm": 2.328165054321289, + "learning_rate": 1e-06, + "loss": 0.7341, + "mean_token_accuracy": 0.7612166404724121, + "num_tokens": 982877451.0, + "step": 36358 + }, + { + "epoch": 2.9981858662488663, + "grad_norm": 2.3506476879119873, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7493080496788025, + "num_tokens": 982901856.0, + "step": 36359 + }, + { + "epoch": 2.9982683268739176, + "grad_norm": 2.154304265975952, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7681586742401123, + "num_tokens": 982930760.0, + "step": 36360 + }, + { + "epoch": 2.9983507874989694, + "grad_norm": 2.4555792808532715, + "learning_rate": 1e-06, + "loss": 0.647, + "mean_token_accuracy": 0.7938910722732544, + "num_tokens": 982957479.0, + "step": 36361 + }, + { + "epoch": 2.998433248124021, + "grad_norm": 2.1059558391571045, + "learning_rate": 1e-06, + "loss": 0.774, + "mean_token_accuracy": 0.7477751970291138, + "num_tokens": 982986815.0, + "step": 36362 + }, + { + "epoch": 2.998515708749072, + "grad_norm": 2.327908754348755, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7226628661155701, + "num_tokens": 983015330.0, + "step": 36363 + }, + { + "epoch": 2.998598169374124, + "grad_norm": 2.160466432571411, + "learning_rate": 1e-06, + "loss": 0.6532, + "mean_token_accuracy": 0.7820870876312256, + "num_tokens": 983043520.0, + "step": 36364 + }, + { + "epoch": 2.9986806299991753, + "grad_norm": 2.2511544227600098, + "learning_rate": 1e-06, + "loss": 0.7102, + "mean_token_accuracy": 0.7673799991607666, + "num_tokens": 983069224.0, + "step": 36365 + }, + { + "epoch": 2.998763090624227, + "grad_norm": 1.9732675552368164, + "learning_rate": 1e-06, + "loss": 0.5203, + "mean_token_accuracy": 0.8274717926979065, + "num_tokens": 983099825.0, + "step": 36366 + }, + { + "epoch": 2.9988455512492784, + "grad_norm": 2.3429267406463623, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7382116317749023, + "num_tokens": 983125370.0, + "step": 36367 + }, + { + "epoch": 2.9989280118743302, + "grad_norm": 1.901803731918335, + "learning_rate": 1e-06, + "loss": 0.5099, + "mean_token_accuracy": 0.8218302726745605, + "num_tokens": 983158104.0, + "step": 36368 + }, + { + "epoch": 2.9990104724993816, + "grad_norm": 1.9157259464263916, + "learning_rate": 1e-06, + "loss": 0.7055, + "mean_token_accuracy": 0.7725241184234619, + "num_tokens": 983190926.0, + "step": 36369 + }, + { + "epoch": 2.999092933124433, + "grad_norm": 2.387542724609375, + "learning_rate": 1e-06, + "loss": 0.734, + "mean_token_accuracy": 0.7626774907112122, + "num_tokens": 983218008.0, + "step": 36370 + }, + { + "epoch": 2.9991753937494847, + "grad_norm": 2.1468822956085205, + "learning_rate": 1e-06, + "loss": 0.7521, + "mean_token_accuracy": 0.7464923858642578, + "num_tokens": 983248021.0, + "step": 36371 + }, + { + "epoch": 2.999257854374536, + "grad_norm": 2.0833425521850586, + "learning_rate": 1e-06, + "loss": 0.664, + "mean_token_accuracy": 0.7785981297492981, + "num_tokens": 983279580.0, + "step": 36372 + }, + { + "epoch": 2.9993403149995874, + "grad_norm": 1.930299162864685, + "learning_rate": 1e-06, + "loss": 0.6766, + "mean_token_accuracy": 0.7713443040847778, + "num_tokens": 983312709.0, + "step": 36373 + }, + { + "epoch": 2.999422775624639, + "grad_norm": 1.9189645051956177, + "learning_rate": 1e-06, + "loss": 0.5845, + "mean_token_accuracy": 0.8097238540649414, + "num_tokens": 983345051.0, + "step": 36374 + }, + { + "epoch": 2.999505236249691, + "grad_norm": 2.0305848121643066, + "learning_rate": 1e-06, + "loss": 0.7142, + "mean_token_accuracy": 0.773680567741394, + "num_tokens": 983379389.0, + "step": 36375 + }, + { + "epoch": 2.9995876968747424, + "grad_norm": 2.3793294429779053, + "learning_rate": 1e-06, + "loss": 0.6752, + "mean_token_accuracy": 0.7742100954055786, + "num_tokens": 983401831.0, + "step": 36376 + }, + { + "epoch": 2.9996701574997937, + "grad_norm": 2.3930490016937256, + "learning_rate": 1e-06, + "loss": 0.6976, + "mean_token_accuracy": 0.7729455828666687, + "num_tokens": 983424190.0, + "step": 36377 + }, + { + "epoch": 2.9997526181248455, + "grad_norm": 2.0772483348846436, + "learning_rate": 1e-06, + "loss": 0.6709, + "mean_token_accuracy": 0.7719948291778564, + "num_tokens": 983452655.0, + "step": 36378 + }, + { + "epoch": 2.999835078749897, + "grad_norm": 2.0058834552764893, + "learning_rate": 1e-06, + "loss": 0.6989, + "mean_token_accuracy": 0.7839986085891724, + "num_tokens": 983486315.0, + "step": 36379 + }, + { + "epoch": 2.999917539374948, + "grad_norm": 2.219658851623535, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7244356870651245, + "num_tokens": 983514300.0, + "step": 36380 + }, + { + "epoch": 3.0, + "grad_norm": 2.0398213863372803, + "learning_rate": 1e-06, + "loss": 0.7307, + "mean_token_accuracy": 0.7667967081069946, + "num_tokens": 983546528.0, + "step": 36381 + }, + { + "epoch": 3.0, + "step": 36381, + "total_flos": 4.923132181932828e+19, + "train_loss": 0.6614166504737585, + "train_runtime": 43606.5901, + "train_samples_per_second": 13.349, + "train_steps_per_second": 0.834 + } + ], + "logging_steps": 1, + "max_steps": 36381, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 3639, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 4.923132181932828e+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..190ab8c --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1df390480886dc4f26de9736beeba3884b0b06a18652de0abcdf992224a53046 +size 13265